diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,35335 +10,35335 @@
     {
       "epoch": 0.0,
       "learning_rate": 2.4242424242424243e-08,
-      "loss": 1.6987,
+      "loss": 1.514,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 4.8484848484848486e-08,
-      "loss": 1.5854,
+      "loss": 1.5784,
       "step": 2
     },
     {
       "epoch": 0.0,
       "learning_rate": 7.272727272727274e-08,
-      "loss": 1.2092,
+      "loss": 1.6176,
       "step": 3
     },
     {
       "epoch": 0.0,
       "learning_rate": 9.696969696969697e-08,
-      "loss": 1.6627,
+      "loss": 1.5808,
       "step": 4
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.2121212121212122e-07,
-      "loss": 1.6635,
+      "loss": 1.5679,
       "step": 5
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.4545454545454548e-07,
-      "loss": 1.2817,
+      "loss": 1.6865,
       "step": 6
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6969696969696974e-07,
-      "loss": 1.3649,
+      "loss": 1.5656,
       "step": 7
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9393939393939395e-07,
-      "loss": 1.5404,
+      "loss": 1.5635,
       "step": 8
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.181818181818182e-07,
-      "loss": 1.4598,
+      "loss": 1.5613,
       "step": 9
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.4242424242424244e-07,
-      "loss": 1.55,
+      "loss": 1.5896,
       "step": 10
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.666666666666667e-07,
-      "loss": 1.4717,
+      "loss": 1.4943,
       "step": 11
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.9090909090909096e-07,
-      "loss": 1.5116,
+      "loss": 1.563,
       "step": 12
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.151515151515152e-07,
-      "loss": 1.437,
+      "loss": 1.6144,
       "step": 13
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.393939393939395e-07,
-      "loss": 1.3459,
+      "loss": 1.4873,
       "step": 14
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.6363636363636366e-07,
-      "loss": 1.4357,
+      "loss": 1.5507,
       "step": 15
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.878787878787879e-07,
-      "loss": 1.3477,
+      "loss": 1.4913,
       "step": 16
     },
     {
       "epoch": 0.0,
       "learning_rate": 4.121212121212122e-07,
-      "loss": 1.52,
+      "loss": 1.5588,
       "step": 17
     },
     {
       "epoch": 0.0,
       "learning_rate": 4.363636363636364e-07,
-      "loss": 1.2056,
+      "loss": 1.5239,
       "step": 18
     },
     {
       "epoch": 0.0,
       "learning_rate": 4.6060606060606064e-07,
-      "loss": 1.3439,
+      "loss": 1.5721,
       "step": 19
     },
     {
       "epoch": 0.0,
       "learning_rate": 4.848484848484849e-07,
-      "loss": 1.3791,
+      "loss": 1.4437,
       "step": 20
     },
     {
       "epoch": 0.0,
       "learning_rate": 5.090909090909092e-07,
-      "loss": 1.558,
+      "loss": 1.5623,
       "step": 21
     },
     {
       "epoch": 0.0,
       "learning_rate": 5.333333333333335e-07,
-      "loss": 1.3298,
+      "loss": 1.5258,
       "step": 22
     },
     {
       "epoch": 0.0,
       "learning_rate": 5.575757575757576e-07,
-      "loss": 1.3212,
+      "loss": 1.4556,
       "step": 23
     },
     {
       "epoch": 0.0,
       "learning_rate": 5.818181818181819e-07,
-      "loss": 1.4628,
+      "loss": 1.4836,
       "step": 24
     },
     {
       "epoch": 0.0,
       "learning_rate": 6.060606060606061e-07,
-      "loss": 1.4097,
+      "loss": 1.4175,
       "step": 25
     },
     {
       "epoch": 0.0,
       "learning_rate": 6.303030303030304e-07,
-      "loss": 1.354,
+      "loss": 1.4844,
       "step": 26
     },
     {
       "epoch": 0.0,
       "learning_rate": 6.545454545454547e-07,
-      "loss": 1.1559,
+      "loss": 1.3837,
       "step": 27
     },
     {
       "epoch": 0.0,
       "learning_rate": 6.78787878787879e-07,
-      "loss": 1.4767,
+      "loss": 1.4982,
       "step": 28
     },
     {
       "epoch": 0.0,
       "learning_rate": 7.03030303030303e-07,
-      "loss": 1.4767,
+      "loss": 1.3137,
       "step": 29
     },
     {
       "epoch": 0.0,
       "learning_rate": 7.272727272727273e-07,
-      "loss": 1.3789,
+      "loss": 1.3314,
       "step": 30
     },
     {
       "epoch": 0.0,
       "learning_rate": 7.515151515151516e-07,
-      "loss": 1.3076,
+      "loss": 1.3929,
       "step": 31
     },
     {
       "epoch": 0.0,
       "learning_rate": 7.757575757575758e-07,
-      "loss": 1.3638,
+      "loss": 1.2963,
       "step": 32
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 1.3488,
+      "loss": 1.318,
       "step": 33
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.242424242424244e-07,
-      "loss": 1.2502,
+      "loss": 1.3569,
       "step": 34
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.484848484848486e-07,
-      "loss": 1.2429,
+      "loss": 1.4148,
       "step": 35
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.727272727272728e-07,
-      "loss": 1.3299,
+      "loss": 1.2521,
       "step": 36
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.96969696969697e-07,
-      "loss": 1.322,
+      "loss": 1.3124,
       "step": 37
     },
     {
       "epoch": 0.0,
       "learning_rate": 9.212121212121213e-07,
-      "loss": 1.3719,
+      "loss": 1.3291,
       "step": 38
     },
     {
       "epoch": 0.0,
       "learning_rate": 9.454545454545455e-07,
-      "loss": 1.2424,
+      "loss": 1.2855,
       "step": 39
     },
     {
       "epoch": 0.0,
       "learning_rate": 9.696969696969698e-07,
-      "loss": 1.2517,
+      "loss": 1.1934,
       "step": 40
     },
     {
       "epoch": 0.0,
       "learning_rate": 9.93939393939394e-07,
-      "loss": 1.472,
+      "loss": 1.3961,
       "step": 41
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.0181818181818183e-06,
-      "loss": 1.1496,
+      "loss": 1.2598,
       "step": 42
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.0424242424242426e-06,
-      "loss": 1.3052,
+      "loss": 1.2781,
       "step": 43
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.066666666666667e-06,
-      "loss": 1.3087,
+      "loss": 1.2413,
       "step": 44
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.090909090909091e-06,
-      "loss": 1.2195,
+      "loss": 1.187,
       "step": 45
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.1151515151515153e-06,
-      "loss": 1.2886,
+      "loss": 1.2688,
       "step": 46
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.1393939393939395e-06,
-      "loss": 1.033,
+      "loss": 1.0652,
       "step": 47
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.1636363636363638e-06,
-      "loss": 1.2632,
+      "loss": 1.3285,
       "step": 48
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.187878787878788e-06,
-      "loss": 1.2044,
+      "loss": 1.2953,
       "step": 49
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.2121212121212122e-06,
-      "loss": 1.2284,
+      "loss": 1.2557,
       "step": 50
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.2363636363636365e-06,
-      "loss": 1.2469,
+      "loss": 1.2089,
       "step": 51
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.2606060606060608e-06,
-      "loss": 1.2243,
+      "loss": 1.1736,
       "step": 52
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.2848484848484848e-06,
-      "loss": 1.2397,
+      "loss": 1.2454,
       "step": 53
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.3090909090909093e-06,
-      "loss": 1.1756,
+      "loss": 1.2783,
       "step": 54
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 1.3632,
+      "loss": 1.2751,
       "step": 55
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.357575757575758e-06,
-      "loss": 1.3694,
+      "loss": 1.2265,
       "step": 56
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.381818181818182e-06,
-      "loss": 1.3304,
+      "loss": 1.2119,
       "step": 57
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.406060606060606e-06,
-      "loss": 1.2651,
+      "loss": 1.1777,
       "step": 58
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.4303030303030306e-06,
-      "loss": 1.4877,
+      "loss": 1.3282,
       "step": 59
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.4545454545454546e-06,
-      "loss": 1.4033,
+      "loss": 1.2714,
       "step": 60
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.478787878787879e-06,
-      "loss": 1.2885,
+      "loss": 1.2488,
       "step": 61
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.5030303030303032e-06,
-      "loss": 1.19,
+      "loss": 1.1635,
       "step": 62
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.5272727272727275e-06,
-      "loss": 1.2359,
+      "loss": 1.1826,
       "step": 63
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.5515151515151516e-06,
-      "loss": 1.3363,
+      "loss": 1.2318,
       "step": 64
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.5757575757575759e-06,
-      "loss": 1.2093,
+      "loss": 1.1616,
       "step": 65
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6000000000000001e-06,
-      "loss": 1.2051,
+      "loss": 1.2142,
       "step": 66
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6242424242424242e-06,
-      "loss": 1.3679,
+      "loss": 1.266,
       "step": 67
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6484848484848487e-06,
-      "loss": 1.326,
+      "loss": 1.1119,
       "step": 68
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6727272727272728e-06,
-      "loss": 1.2382,
+      "loss": 1.2154,
       "step": 69
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.6969696969696973e-06,
-      "loss": 1.2124,
+      "loss": 1.2303,
       "step": 70
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.7212121212121214e-06,
-      "loss": 1.1504,
+      "loss": 1.119,
       "step": 71
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.7454545454545456e-06,
-      "loss": 1.3028,
+      "loss": 1.1658,
       "step": 72
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.76969696969697e-06,
-      "loss": 1.1479,
+      "loss": 1.1497,
       "step": 73
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.793939393939394e-06,
-      "loss": 1.2022,
+      "loss": 1.23,
       "step": 74
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.8181818181818183e-06,
-      "loss": 1.2469,
+      "loss": 1.0383,
       "step": 75
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.8424242424242426e-06,
-      "loss": 1.2317,
+      "loss": 1.1545,
       "step": 76
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.8666666666666669e-06,
-      "loss": 1.1981,
+      "loss": 1.1726,
       "step": 77
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.890909090909091e-06,
-      "loss": 1.2163,
+      "loss": 1.2009,
       "step": 78
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9151515151515154e-06,
-      "loss": 1.2478,
+      "loss": 1.1579,
       "step": 79
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9393939393939395e-06,
-      "loss": 1.2389,
+      "loss": 1.2468,
       "step": 80
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9636363636363636e-06,
-      "loss": 1.2238,
+      "loss": 1.1769,
       "step": 81
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.987878787878788e-06,
-      "loss": 1.2056,
+      "loss": 1.2183,
       "step": 82
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.012121212121212e-06,
-      "loss": 1.2347,
+      "loss": 1.17,
       "step": 83
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.0363636363636367e-06,
-      "loss": 1.1615,
+      "loss": 1.2103,
       "step": 84
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.0606060606060607e-06,
-      "loss": 1.2059,
+      "loss": 1.2373,
       "step": 85
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.0848484848484852e-06,
-      "loss": 1.2269,
+      "loss": 1.197,
       "step": 86
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.1090909090909093e-06,
-      "loss": 0.9801,
+      "loss": 1.0681,
       "step": 87
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.133333333333334e-06,
-      "loss": 1.0384,
+      "loss": 1.097,
       "step": 88
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.157575757575758e-06,
-      "loss": 1.2206,
+      "loss": 1.2188,
       "step": 89
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.181818181818182e-06,
-      "loss": 1.0367,
+      "loss": 1.1121,
       "step": 90
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.2060606060606064e-06,
-      "loss": 1.2977,
+      "loss": 1.2227,
       "step": 91
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.2303030303030305e-06,
-      "loss": 1.2971,
+      "loss": 1.1478,
       "step": 92
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.254545454545455e-06,
-      "loss": 1.2061,
+      "loss": 1.1385,
       "step": 93
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.278787878787879e-06,
-      "loss": 1.2646,
+      "loss": 1.1803,
       "step": 94
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.303030303030303e-06,
-      "loss": 1.3765,
+      "loss": 1.2562,
       "step": 95
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.3272727272727277e-06,
-      "loss": 1.1031,
+      "loss": 1.1153,
       "step": 96
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.3515151515151517e-06,
-      "loss": 1.1585,
+      "loss": 1.1145,
       "step": 97
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.375757575757576e-06,
-      "loss": 1.1874,
+      "loss": 1.1396,
       "step": 98
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 1.2159,
+      "loss": 1.1476,
       "step": 99
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.4242424242424244e-06,
-      "loss": 1.2643,
+      "loss": 1.202,
       "step": 100
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.4484848484848485e-06,
-      "loss": 1.1393,
+      "loss": 1.1792,
       "step": 101
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.472727272727273e-06,
-      "loss": 1.3131,
+      "loss": 1.1799,
       "step": 102
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.496969696969697e-06,
-      "loss": 1.3467,
+      "loss": 1.2919,
       "step": 103
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.5212121212121215e-06,
-      "loss": 1.0082,
+      "loss": 1.094,
       "step": 104
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.5454545454545456e-06,
-      "loss": 1.1285,
+      "loss": 1.093,
       "step": 105
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.5696969696969697e-06,
-      "loss": 1.0621,
+      "loss": 1.0606,
       "step": 106
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.593939393939394e-06,
-      "loss": 1.1825,
+      "loss": 1.1605,
       "step": 107
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.6181818181818187e-06,
-      "loss": 1.0834,
+      "loss": 1.091,
       "step": 108
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.6424242424242423e-06,
-      "loss": 1.1602,
+      "loss": 1.0985,
       "step": 109
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 1.2201,
+      "loss": 1.2567,
       "step": 110
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.6909090909090913e-06,
-      "loss": 1.3038,
+      "loss": 1.2498,
       "step": 111
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.715151515151516e-06,
-      "loss": 1.0039,
+      "loss": 1.0693,
       "step": 112
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.7393939393939395e-06,
-      "loss": 1.1817,
+      "loss": 1.1226,
       "step": 113
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.763636363636364e-06,
-      "loss": 1.087,
+      "loss": 1.1238,
       "step": 114
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.7878787878787885e-06,
-      "loss": 1.1532,
+      "loss": 1.0915,
       "step": 115
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.812121212121212e-06,
-      "loss": 1.1703,
+      "loss": 1.1547,
       "step": 116
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.8363636363636366e-06,
-      "loss": 1.1945,
+      "loss": 1.2103,
       "step": 117
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.860606060606061e-06,
-      "loss": 1.1067,
+      "loss": 1.2206,
       "step": 118
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.884848484848485e-06,
-      "loss": 1.2303,
+      "loss": 1.17,
       "step": 119
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.9090909090909093e-06,
-      "loss": 1.0272,
+      "loss": 1.1274,
       "step": 120
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.9333333333333338e-06,
-      "loss": 1.1311,
+      "loss": 1.1317,
       "step": 121
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.957575757575758e-06,
-      "loss": 1.1606,
+      "loss": 1.1342,
       "step": 122
     },
     {
       "epoch": 0.0,
       "learning_rate": 2.981818181818182e-06,
-      "loss": 1.0471,
+      "loss": 1.066,
       "step": 123
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.0060606060606064e-06,
-      "loss": 1.1225,
+      "loss": 1.1248,
       "step": 124
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.0303030303030305e-06,
-      "loss": 1.0823,
+      "loss": 1.0703,
       "step": 125
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.054545454545455e-06,
-      "loss": 1.2246,
+      "loss": 1.0766,
       "step": 126
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.078787878787879e-06,
-      "loss": 1.1876,
+      "loss": 1.1253,
       "step": 127
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.103030303030303e-06,
-      "loss": 1.1311,
+      "loss": 1.1418,
       "step": 128
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.1272727272727276e-06,
-      "loss": 1.2069,
+      "loss": 1.1267,
       "step": 129
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.1515151515151517e-06,
-      "loss": 1.1306,
+      "loss": 1.1363,
       "step": 130
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.1757575757575758e-06,
-      "loss": 1.1174,
+      "loss": 1.1235,
       "step": 131
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.2000000000000003e-06,
-      "loss": 1.0453,
+      "loss": 1.1107,
       "step": 132
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.2242424242424248e-06,
-      "loss": 1.1398,
+      "loss": 1.1159,
       "step": 133
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.2484848484848484e-06,
-      "loss": 1.2033,
+      "loss": 1.1593,
       "step": 134
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.272727272727273e-06,
-      "loss": 1.0988,
+      "loss": 1.0771,
       "step": 135
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.2969696969696974e-06,
-      "loss": 1.0732,
+      "loss": 1.1276,
       "step": 136
     },
     {
       "epoch": 0.0,
       "learning_rate": 3.321212121212121e-06,
-      "loss": 1.1435,
+      "loss": 1.0914,
       "step": 137
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.3454545454545456e-06,
-      "loss": 1.1708,
+      "loss": 1.1921,
       "step": 138
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.36969696969697e-06,
-      "loss": 0.9957,
+      "loss": 1.042,
       "step": 139
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.3939393939393946e-06,
-      "loss": 0.9716,
+      "loss": 1.1063,
       "step": 140
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.4181818181818182e-06,
-      "loss": 1.08,
+      "loss": 1.0434,
       "step": 141
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.4424242424242427e-06,
-      "loss": 1.18,
+      "loss": 1.1905,
       "step": 142
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.4666666666666672e-06,
-      "loss": 1.0322,
+      "loss": 1.0328,
       "step": 143
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.4909090909090913e-06,
-      "loss": 1.0327,
+      "loss": 1.0885,
       "step": 144
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.5151515151515154e-06,
-      "loss": 1.1354,
+      "loss": 1.1307,
       "step": 145
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.53939393939394e-06,
-      "loss": 0.9842,
+      "loss": 1.0995,
       "step": 146
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.563636363636364e-06,
-      "loss": 1.1077,
+      "loss": 1.1297,
       "step": 147
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.587878787878788e-06,
-      "loss": 1.0487,
+      "loss": 1.1387,
       "step": 148
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.6121212121212125e-06,
-      "loss": 1.0375,
+      "loss": 1.0713,
       "step": 149
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.6363636363636366e-06,
-      "loss": 1.096,
+      "loss": 1.0695,
       "step": 150
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.660606060606061e-06,
-      "loss": 1.0926,
+      "loss": 1.1583,
       "step": 151
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.684848484848485e-06,
-      "loss": 1.0386,
+      "loss": 0.9832,
       "step": 152
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.7090909090909092e-06,
-      "loss": 1.1586,
+      "loss": 1.1375,
       "step": 153
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.7333333333333337e-06,
-      "loss": 1.119,
+      "loss": 1.0883,
       "step": 154
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.757575757575758e-06,
-      "loss": 1.1678,
+      "loss": 1.1682,
       "step": 155
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.781818181818182e-06,
-      "loss": 1.0794,
+      "loss": 1.1226,
       "step": 156
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.8060606060606064e-06,
-      "loss": 1.0662,
+      "loss": 1.1194,
       "step": 157
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.830303030303031e-06,
-      "loss": 1.1257,
+      "loss": 1.1473,
       "step": 158
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.8545454545454545e-06,
-      "loss": 1.096,
+      "loss": 1.1827,
       "step": 159
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.878787878787879e-06,
-      "loss": 0.9469,
+      "loss": 1.0785,
       "step": 160
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.9030303030303035e-06,
-      "loss": 1.038,
+      "loss": 1.065,
       "step": 161
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.927272727272727e-06,
-      "loss": 1.0524,
+      "loss": 1.0722,
       "step": 162
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.951515151515152e-06,
-      "loss": 0.9794,
+      "loss": 1.0886,
       "step": 163
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.975757575757576e-06,
-      "loss": 1.0481,
+      "loss": 1.1071,
       "step": 164
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.9634,
+      "loss": 0.9835,
       "step": 165
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.024242424242424e-06,
-      "loss": 1.0881,
+      "loss": 1.0638,
       "step": 166
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.048484848484849e-06,
-      "loss": 1.1821,
+      "loss": 1.2014,
       "step": 167
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.072727272727273e-06,
-      "loss": 1.0453,
+      "loss": 1.041,
       "step": 168
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.096969696969697e-06,
-      "loss": 1.1285,
+      "loss": 1.1228,
       "step": 169
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.1212121212121215e-06,
-      "loss": 1.0477,
+      "loss": 1.0525,
       "step": 170
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.145454545454546e-06,
-      "loss": 0.9348,
+      "loss": 0.9763,
       "step": 171
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.1696969696969705e-06,
-      "loss": 0.9867,
+      "loss": 1.1371,
       "step": 172
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.193939393939394e-06,
-      "loss": 1.1132,
+      "loss": 1.0968,
       "step": 173
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.218181818181819e-06,
-      "loss": 1.1333,
+      "loss": 1.0452,
       "step": 174
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.242424242424243e-06,
-      "loss": 1.0848,
+      "loss": 1.1182,
       "step": 175
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.266666666666668e-06,
-      "loss": 1.1526,
+      "loss": 1.1224,
       "step": 176
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.290909090909091e-06,
-      "loss": 1.0865,
+      "loss": 1.0941,
       "step": 177
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.315151515151516e-06,
-      "loss": 0.966,
+      "loss": 0.9844,
       "step": 178
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.33939393939394e-06,
-      "loss": 1.0426,
+      "loss": 1.1006,
       "step": 179
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.363636363636364e-06,
-      "loss": 1.0334,
+      "loss": 1.0974,
       "step": 180
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.387878787878788e-06,
-      "loss": 1.1146,
+      "loss": 1.1107,
       "step": 181
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.412121212121213e-06,
-      "loss": 1.1295,
+      "loss": 1.1357,
       "step": 182
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.436363636363637e-06,
-      "loss": 1.1383,
+      "loss": 1.1684,
       "step": 183
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.460606060606061e-06,
-      "loss": 1.0954,
+      "loss": 1.0655,
       "step": 184
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.4848484848484855e-06,
-      "loss": 1.069,
+      "loss": 1.0405,
       "step": 185
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.50909090909091e-06,
-      "loss": 0.9932,
+      "loss": 1.0665,
       "step": 186
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.533333333333334e-06,
-      "loss": 1.0863,
+      "loss": 1.1149,
       "step": 187
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.557575757575758e-06,
-      "loss": 1.0724,
+      "loss": 1.1586,
       "step": 188
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.581818181818183e-06,
-      "loss": 1.0044,
+      "loss": 1.0295,
       "step": 189
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.606060606060606e-06,
-      "loss": 1.0146,
+      "loss": 1.0724,
       "step": 190
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.630303030303031e-06,
-      "loss": 1.0823,
+      "loss": 1.1061,
       "step": 191
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.654545454545455e-06,
-      "loss": 1.1749,
+      "loss": 1.1981,
       "step": 192
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.678787878787879e-06,
-      "loss": 0.9935,
+      "loss": 1.086,
       "step": 193
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.7030303030303035e-06,
-      "loss": 1.021,
+      "loss": 1.0677,
       "step": 194
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.727272727272728e-06,
-      "loss": 1.0765,
+      "loss": 1.2049,
       "step": 195
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.751515151515152e-06,
-      "loss": 1.0656,
+      "loss": 1.1306,
       "step": 196
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.775757575757576e-06,
-      "loss": 1.0904,
+      "loss": 1.157,
       "step": 197
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.800000000000001e-06,
-      "loss": 1.0089,
+      "loss": 1.1083,
       "step": 198
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.824242424242424e-06,
-      "loss": 1.0253,
+      "loss": 1.1212,
       "step": 199
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.848484848484849e-06,
-      "loss": 1.1081,
+      "loss": 1.0711,
       "step": 200
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.872727272727273e-06,
-      "loss": 1.1135,
+      "loss": 1.1459,
       "step": 201
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.896969696969697e-06,
-      "loss": 1.0468,
+      "loss": 1.0533,
       "step": 202
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.9212121212121214e-06,
-      "loss": 1.0285,
+      "loss": 1.1046,
       "step": 203
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.945454545454546e-06,
-      "loss": 1.0454,
+      "loss": 1.0908,
       "step": 204
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.9696969696969696e-06,
-      "loss": 1.0176,
+      "loss": 1.0947,
       "step": 205
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.993939393939394e-06,
-      "loss": 1.0514,
+      "loss": 1.0977,
       "step": 206
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.0181818181818186e-06,
-      "loss": 0.9556,
+      "loss": 1.0518,
       "step": 207
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.042424242424243e-06,
-      "loss": 1.1005,
+      "loss": 1.1041,
       "step": 208
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.0666666666666676e-06,
-      "loss": 0.9556,
+      "loss": 1.0479,
       "step": 209
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.090909090909091e-06,
-      "loss": 0.905,
+      "loss": 0.9499,
       "step": 210
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.115151515151515e-06,
-      "loss": 1.042,
+      "loss": 1.0862,
       "step": 211
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.139393939393939e-06,
-      "loss": 1.0291,
+      "loss": 1.0558,
       "step": 212
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.163636363636364e-06,
-      "loss": 0.9359,
+      "loss": 1.0154,
       "step": 213
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.187878787878788e-06,
-      "loss": 1.1368,
+      "loss": 1.0869,
       "step": 214
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.212121212121213e-06,
-      "loss": 1.0538,
+      "loss": 1.1,
       "step": 215
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.236363636363637e-06,
-      "loss": 0.9745,
+      "loss": 0.9034,
       "step": 216
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.26060606060606e-06,
-      "loss": 0.9639,
+      "loss": 1.0953,
       "step": 217
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.284848484848485e-06,
-      "loss": 1.0652,
+      "loss": 1.0956,
       "step": 218
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.309090909090909e-06,
-      "loss": 0.9904,
+      "loss": 1.0356,
       "step": 219
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 1.065,
+      "loss": 1.1084,
       "step": 220
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.357575757575758e-06,
-      "loss": 1.0168,
+      "loss": 0.9804,
       "step": 221
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.381818181818183e-06,
-      "loss": 1.0897,
+      "loss": 1.109,
       "step": 222
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.406060606060607e-06,
-      "loss": 0.9181,
+      "loss": 1.0414,
       "step": 223
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.430303030303032e-06,
-      "loss": 1.165,
+      "loss": 1.203,
       "step": 224
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.4545454545454545e-06,
-      "loss": 1.0044,
+      "loss": 1.0573,
       "step": 225
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.478787878787879e-06,
-      "loss": 1.1221,
+      "loss": 1.0781,
       "step": 226
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.5030303030303034e-06,
-      "loss": 0.9194,
+      "loss": 1.0334,
       "step": 227
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.527272727272728e-06,
-      "loss": 0.9858,
+      "loss": 1.0524,
       "step": 228
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.5515151515151524e-06,
-      "loss": 0.9315,
+      "loss": 1.038,
       "step": 229
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.575757575757577e-06,
-      "loss": 1.0234,
+      "loss": 1.132,
       "step": 230
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.600000000000001e-06,
-      "loss": 0.9855,
+      "loss": 1.1669,
       "step": 231
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.624242424242424e-06,
-      "loss": 1.0211,
+      "loss": 1.0591,
       "step": 232
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.648484848484849e-06,
-      "loss": 1.0399,
+      "loss": 1.0587,
       "step": 233
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.672727272727273e-06,
-      "loss": 0.9492,
+      "loss": 1.0459,
       "step": 234
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.696969696969698e-06,
-      "loss": 1.0857,
+      "loss": 1.0753,
       "step": 235
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.721212121212122e-06,
-      "loss": 0.9768,
+      "loss": 1.0093,
       "step": 236
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.745454545454546e-06,
-      "loss": 1.0233,
+      "loss": 1.0505,
       "step": 237
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.76969696969697e-06,
-      "loss": 1.0466,
+      "loss": 1.0091,
       "step": 238
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.793939393939394e-06,
-      "loss": 1.1139,
+      "loss": 1.1056,
       "step": 239
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.8181818181818185e-06,
-      "loss": 0.9979,
+      "loss": 1.0089,
       "step": 240
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.842424242424243e-06,
-      "loss": 0.9316,
+      "loss": 1.0138,
       "step": 241
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.8666666666666675e-06,
-      "loss": 1.0461,
+      "loss": 1.1113,
       "step": 242
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.890909090909091e-06,
-      "loss": 0.9526,
+      "loss": 0.9985,
       "step": 243
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.915151515151516e-06,
-      "loss": 0.9982,
+      "loss": 1.1292,
       "step": 244
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.93939393939394e-06,
-      "loss": 1.0647,
+      "loss": 1.0314,
       "step": 245
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.963636363636364e-06,
-      "loss": 1.033,
+      "loss": 1.0565,
       "step": 246
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.987878787878788e-06,
-      "loss": 1.077,
+      "loss": 1.0848,
       "step": 247
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.012121212121213e-06,
-      "loss": 1.1399,
+      "loss": 1.1049,
       "step": 248
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.0363636363636365e-06,
-      "loss": 1.0799,
+      "loss": 1.1427,
       "step": 249
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.060606060606061e-06,
-      "loss": 1.1109,
+      "loss": 1.1436,
       "step": 250
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.0848484848484855e-06,
-      "loss": 0.9975,
+      "loss": 1.044,
       "step": 251
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.10909090909091e-06,
-      "loss": 0.9468,
+      "loss": 1.0294,
       "step": 252
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.133333333333334e-06,
-      "loss": 1.0463,
+      "loss": 1.0599,
       "step": 253
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.157575757575758e-06,
-      "loss": 1.0077,
+      "loss": 1.0865,
       "step": 254
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.181818181818182e-06,
-      "loss": 1.0552,
+      "loss": 1.1107,
       "step": 255
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.206060606060606e-06,
-      "loss": 0.8938,
+      "loss": 1.0261,
       "step": 256
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.230303030303031e-06,
-      "loss": 1.1541,
+      "loss": 1.1296,
       "step": 257
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.254545454545455e-06,
-      "loss": 1.0948,
+      "loss": 1.1581,
       "step": 258
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.27878787878788e-06,
-      "loss": 1.1237,
+      "loss": 1.1534,
       "step": 259
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.303030303030303e-06,
-      "loss": 0.9331,
+      "loss": 1.0556,
       "step": 260
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.327272727272727e-06,
-      "loss": 0.9571,
+      "loss": 0.9806,
       "step": 261
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.3515151515151516e-06,
-      "loss": 0.9982,
+      "loss": 1.0757,
       "step": 262
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.375757575757576e-06,
-      "loss": 1.0045,
+      "loss": 1.1261,
       "step": 263
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.9911,
+      "loss": 1.0646,
       "step": 264
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.424242424242425e-06,
-      "loss": 1.0405,
+      "loss": 1.0701,
       "step": 265
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.4484848484848496e-06,
-      "loss": 1.0207,
+      "loss": 1.1102,
       "step": 266
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.472727272727272e-06,
-      "loss": 1.0751,
+      "loss": 1.0975,
       "step": 267
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.496969696969697e-06,
-      "loss": 1.0027,
+      "loss": 1.0226,
       "step": 268
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.521212121212121e-06,
-      "loss": 1.1473,
+      "loss": 1.1131,
       "step": 269
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.545454545454546e-06,
-      "loss": 0.9017,
+      "loss": 0.991,
       "step": 270
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.56969696969697e-06,
-      "loss": 1.0608,
+      "loss": 1.1003,
       "step": 271
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.593939393939395e-06,
-      "loss": 1.0498,
+      "loss": 1.1168,
       "step": 272
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.618181818181819e-06,
-      "loss": 1.0672,
+      "loss": 1.1181,
       "step": 273
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.642424242424242e-06,
-      "loss": 0.9366,
+      "loss": 1.0035,
       "step": 274
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 1.0125,
+      "loss": 1.0383,
       "step": 275
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.690909090909091e-06,
-      "loss": 1.0408,
+      "loss": 1.1263,
       "step": 276
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.715151515151516e-06,
-      "loss": 0.9255,
+      "loss": 0.9999,
       "step": 277
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.73939393939394e-06,
-      "loss": 1.1105,
+      "loss": 1.0663,
       "step": 278
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.763636363636365e-06,
-      "loss": 1.0049,
+      "loss": 1.0709,
       "step": 279
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.787878787878789e-06,
-      "loss": 1.1372,
+      "loss": 1.1867,
       "step": 280
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.812121212121212e-06,
-      "loss": 1.0199,
+      "loss": 1.0711,
       "step": 281
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.8363636363636364e-06,
-      "loss": 1.0291,
+      "loss": 1.1375,
       "step": 282
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.860606060606061e-06,
-      "loss": 0.9768,
+      "loss": 1.0707,
       "step": 283
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.8848484848484854e-06,
-      "loss": 0.9889,
+      "loss": 1.013,
       "step": 284
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.90909090909091e-06,
-      "loss": 1.0565,
+      "loss": 1.0702,
       "step": 285
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.9333333333333344e-06,
-      "loss": 0.8777,
+      "loss": 1.0337,
       "step": 286
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.957575757575759e-06,
-      "loss": 0.955,
+      "loss": 1.0544,
       "step": 287
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.981818181818183e-06,
-      "loss": 1.038,
+      "loss": 1.0174,
       "step": 288
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.006060606060606e-06,
-      "loss": 0.9429,
+      "loss": 1.153,
       "step": 289
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.030303030303031e-06,
-      "loss": 1.0228,
+      "loss": 1.0279,
       "step": 290
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.054545454545455e-06,
-      "loss": 0.8573,
+      "loss": 1.0078,
       "step": 291
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.07878787878788e-06,
-      "loss": 0.9639,
+      "loss": 1.0629,
       "step": 292
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.103030303030304e-06,
-      "loss": 1.0018,
+      "loss": 1.0512,
       "step": 293
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.127272727272728e-06,
-      "loss": 0.9725,
+      "loss": 1.05,
       "step": 294
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.151515151515152e-06,
-      "loss": 1.0997,
+      "loss": 1.1177,
       "step": 295
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.175757575757576e-06,
-      "loss": 1.0896,
+      "loss": 1.1666,
       "step": 296
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.2000000000000005e-06,
-      "loss": 1.235,
+      "loss": 1.1831,
       "step": 297
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.224242424242425e-06,
-      "loss": 0.9864,
+      "loss": 1.106,
       "step": 298
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.2484848484848495e-06,
-      "loss": 1.0224,
+      "loss": 1.0895,
       "step": 299
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.272727272727273e-06,
-      "loss": 1.1011,
+      "loss": 1.1064,
       "step": 300
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.296969696969698e-06,
-      "loss": 1.0754,
+      "loss": 1.0397,
       "step": 301
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.321212121212122e-06,
-      "loss": 0.9422,
+      "loss": 1.0782,
       "step": 302
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.345454545454546e-06,
-      "loss": 0.9945,
+      "loss": 0.9962,
       "step": 303
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.36969696969697e-06,
-      "loss": 0.9337,
+      "loss": 1.1317,
       "step": 304
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.393939393939395e-06,
-      "loss": 0.9758,
+      "loss": 1.0349,
       "step": 305
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.4181818181818185e-06,
-      "loss": 1.0701,
+      "loss": 1.0931,
       "step": 306
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.442424242424243e-06,
-      "loss": 0.9496,
+      "loss": 1.0385,
       "step": 307
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.4666666666666675e-06,
-      "loss": 0.9258,
+      "loss": 0.9968,
       "step": 308
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.490909090909092e-06,
-      "loss": 0.9463,
+      "loss": 1.0752,
       "step": 309
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.515151515151516e-06,
-      "loss": 1.0198,
+      "loss": 1.1242,
       "step": 310
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.53939393939394e-06,
-      "loss": 1.0195,
+      "loss": 1.0366,
       "step": 311
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.563636363636364e-06,
-      "loss": 0.8819,
+      "loss": 1.1515,
       "step": 312
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.587878787878788e-06,
-      "loss": 0.9759,
+      "loss": 1.0287,
       "step": 313
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.612121212121213e-06,
-      "loss": 0.9143,
+      "loss": 1.0817,
       "step": 314
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.636363636363638e-06,
-      "loss": 1.0636,
+      "loss": 1.0756,
       "step": 315
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.660606060606062e-06,
-      "loss": 1.004,
+      "loss": 1.0315,
       "step": 316
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.684848484848485e-06,
-      "loss": 1.0773,
+      "loss": 1.1512,
       "step": 317
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.709090909090909e-06,
-      "loss": 0.9843,
+      "loss": 0.9959,
       "step": 318
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.733333333333334e-06,
-      "loss": 0.955,
+      "loss": 1.0194,
       "step": 319
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.757575757575758e-06,
-      "loss": 1.0221,
+      "loss": 1.0662,
       "step": 320
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.781818181818183e-06,
-      "loss": 0.9839,
+      "loss": 1.0668,
       "step": 321
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.806060606060607e-06,
-      "loss": 0.9375,
+      "loss": 1.0179,
       "step": 322
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.83030303030303e-06,
-      "loss": 0.9941,
+      "loss": 1.0269,
       "step": 323
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.854545454545454e-06,
-      "loss": 0.9088,
+      "loss": 0.9997,
       "step": 324
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.87878787878788e-06,
-      "loss": 1.029,
+      "loss": 1.0753,
       "step": 325
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.903030303030303e-06,
-      "loss": 0.9056,
+      "loss": 0.9996,
       "step": 326
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.927272727272729e-06,
-      "loss": 1.0311,
+      "loss": 1.1403,
       "step": 327
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.951515151515152e-06,
-      "loss": 1.1028,
+      "loss": 1.111,
       "step": 328
     },
     {
       "epoch": 0.01,
       "learning_rate": 7.975757575757576e-06,
-      "loss": 0.9458,
+      "loss": 0.9708,
       "step": 329
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.9936,
+      "loss": 1.1031,
       "step": 330
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.024242424242425e-06,
-      "loss": 0.8806,
+      "loss": 1.0875,
       "step": 331
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.048484848484849e-06,
-      "loss": 0.9288,
+      "loss": 1.0337,
       "step": 332
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.072727272727274e-06,
-      "loss": 1.0559,
+      "loss": 1.0985,
       "step": 333
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.096969696969698e-06,
-      "loss": 1.0497,
+      "loss": 1.0861,
       "step": 334
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.121212121212121e-06,
-      "loss": 1.0129,
+      "loss": 1.0889,
       "step": 335
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.145454545454547e-06,
-      "loss": 0.9451,
+      "loss": 1.095,
       "step": 336
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.16969696969697e-06,
-      "loss": 1.0288,
+      "loss": 0.9848,
       "step": 337
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.193939393939394e-06,
-      "loss": 1.0009,
+      "loss": 1.0594,
       "step": 338
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.21818181818182e-06,
-      "loss": 1.1037,
+      "loss": 1.1282,
       "step": 339
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.242424242424243e-06,
-      "loss": 0.9644,
+      "loss": 0.983,
       "step": 340
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.266666666666667e-06,
-      "loss": 0.9142,
+      "loss": 1.0569,
       "step": 341
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.290909090909092e-06,
-      "loss": 0.99,
+      "loss": 1.0588,
       "step": 342
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.315151515151516e-06,
-      "loss": 0.9293,
+      "loss": 0.9938,
       "step": 343
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.339393939393941e-06,
-      "loss": 1.0214,
+      "loss": 1.0618,
       "step": 344
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.363636363636365e-06,
-      "loss": 1.0297,
+      "loss": 1.0962,
       "step": 345
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.387878787878788e-06,
-      "loss": 0.9505,
+      "loss": 1.0031,
       "step": 346
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.412121212121212e-06,
-      "loss": 0.9303,
+      "loss": 1.0215,
       "step": 347
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.436363636363637e-06,
-      "loss": 1.1558,
+      "loss": 1.1493,
       "step": 348
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.460606060606061e-06,
-      "loss": 0.974,
+      "loss": 1.0095,
       "step": 349
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.484848484848486e-06,
-      "loss": 1.05,
+      "loss": 1.0985,
       "step": 350
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.50909090909091e-06,
-      "loss": 1.1152,
+      "loss": 1.141,
       "step": 351
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.533333333333335e-06,
-      "loss": 0.9243,
+      "loss": 1.0034,
       "step": 352
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.557575757575757e-06,
-      "loss": 0.9436,
+      "loss": 1.0451,
       "step": 353
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.581818181818183e-06,
-      "loss": 0.9432,
+      "loss": 0.9537,
       "step": 354
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.606060606060606e-06,
-      "loss": 1.0048,
+      "loss": 1.1069,
       "step": 355
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.630303030303032e-06,
-      "loss": 1.1177,
+      "loss": 1.0908,
       "step": 356
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.654545454545455e-06,
-      "loss": 1.093,
+      "loss": 1.1108,
       "step": 357
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.67878787878788e-06,
-      "loss": 1.136,
+      "loss": 1.131,
       "step": 358
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.703030303030304e-06,
-      "loss": 1.1038,
+      "loss": 1.1711,
       "step": 359
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.727272727272728e-06,
-      "loss": 1.0908,
+      "loss": 1.1362,
       "step": 360
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.751515151515151e-06,
-      "loss": 1.0506,
+      "loss": 1.0908,
       "step": 361
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.775757575757577e-06,
-      "loss": 1.052,
+      "loss": 1.1183,
       "step": 362
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.8e-06,
-      "loss": 1.0807,
+      "loss": 1.0957,
       "step": 363
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.824242424242426e-06,
-      "loss": 0.933,
+      "loss": 1.0223,
       "step": 364
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.84848484848485e-06,
-      "loss": 0.9734,
+      "loss": 1.0486,
       "step": 365
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.872727272727275e-06,
-      "loss": 0.9887,
+      "loss": 1.0941,
       "step": 366
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.896969696969697e-06,
-      "loss": 0.9566,
+      "loss": 1.0478,
       "step": 367
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.921212121212122e-06,
-      "loss": 1.0405,
+      "loss": 1.0969,
       "step": 368
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.945454545454546e-06,
-      "loss": 0.9836,
+      "loss": 1.0972,
       "step": 369
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.969696969696971e-06,
-      "loss": 0.9774,
+      "loss": 1.0372,
       "step": 370
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.993939393939395e-06,
-      "loss": 1.0739,
+      "loss": 1.0903,
       "step": 371
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.01818181818182e-06,
-      "loss": 0.9805,
+      "loss": 0.9333,
       "step": 372
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.042424242424244e-06,
-      "loss": 0.9164,
+      "loss": 1.0056,
       "step": 373
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.066666666666667e-06,
-      "loss": 1.0022,
+      "loss": 1.1202,
       "step": 374
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.090909090909091e-06,
-      "loss": 1.0072,
+      "loss": 1.1383,
       "step": 375
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.115151515151516e-06,
-      "loss": 0.8852,
+      "loss": 1.0388,
       "step": 376
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.13939393939394e-06,
-      "loss": 0.8659,
+      "loss": 0.9407,
       "step": 377
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.163636363636365e-06,
-      "loss": 1.0882,
+      "loss": 1.1451,
       "step": 378
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.187878787878789e-06,
-      "loss": 0.9186,
+      "loss": 1.0555,
       "step": 379
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.212121212121213e-06,
-      "loss": 1.0127,
+      "loss": 1.1437,
       "step": 380
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.236363636363636e-06,
-      "loss": 1.1046,
+      "loss": 1.1337,
       "step": 381
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.260606060606062e-06,
-      "loss": 0.9043,
+      "loss": 1.0406,
       "step": 382
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.284848484848485e-06,
-      "loss": 1.03,
+      "loss": 1.084,
       "step": 383
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.30909090909091e-06,
-      "loss": 0.9753,
+      "loss": 1.0709,
       "step": 384
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 1.0912,
+      "loss": 1.1787,
       "step": 385
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.357575757575758e-06,
-      "loss": 1.0542,
+      "loss": 1.0919,
       "step": 386
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.381818181818183e-06,
-      "loss": 1.0486,
+      "loss": 1.0518,
       "step": 387
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.406060606060607e-06,
-      "loss": 0.952,
+      "loss": 1.0172,
       "step": 388
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.43030303030303e-06,
-      "loss": 1.013,
+      "loss": 1.062,
       "step": 389
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.454545454545456e-06,
-      "loss": 0.9559,
+      "loss": 1.0476,
       "step": 390
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.47878787878788e-06,
-      "loss": 0.968,
+      "loss": 1.0332,
       "step": 391
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.503030303030303e-06,
-      "loss": 0.9348,
+      "loss": 1.0282,
       "step": 392
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.527272727272729e-06,
-      "loss": 1.0862,
+      "loss": 1.0781,
       "step": 393
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.551515151515152e-06,
-      "loss": 0.9709,
+      "loss": 1.0834,
       "step": 394
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.575757575757576e-06,
-      "loss": 1.0186,
+      "loss": 1.0756,
       "step": 395
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.600000000000001e-06,
-      "loss": 0.9822,
+      "loss": 0.9843,
       "step": 396
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.624242424242425e-06,
-      "loss": 0.9563,
+      "loss": 1.0076,
       "step": 397
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.648484848484849e-06,
-      "loss": 1.011,
+      "loss": 1.0884,
       "step": 398
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.672727272727274e-06,
-      "loss": 0.9319,
+      "loss": 1.0281,
       "step": 399
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.696969696969698e-06,
-      "loss": 0.9077,
+      "loss": 0.9926,
       "step": 400
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.721212121212123e-06,
-      "loss": 1.025,
+      "loss": 1.1191,
       "step": 401
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.745454545454547e-06,
-      "loss": 0.9751,
+      "loss": 1.0696,
       "step": 402
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.76969696969697e-06,
-      "loss": 0.9677,
+      "loss": 1.0085,
       "step": 403
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.793939393939394e-06,
-      "loss": 1.1356,
+      "loss": 1.1302,
       "step": 404
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.81818181818182e-06,
-      "loss": 1.0264,
+      "loss": 1.0569,
       "step": 405
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.842424242424243e-06,
-      "loss": 1.0539,
+      "loss": 1.0756,
       "step": 406
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.866666666666668e-06,
-      "loss": 1.0885,
+      "loss": 1.1017,
       "step": 407
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.890909090909092e-06,
-      "loss": 0.9339,
+      "loss": 1.0167,
       "step": 408
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.915151515151515e-06,
-      "loss": 1.0198,
+      "loss": 1.0087,
       "step": 409
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.939393939393939e-06,
-      "loss": 1.081,
+      "loss": 1.1479,
       "step": 410
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.963636363636364e-06,
-      "loss": 1.0036,
+      "loss": 1.0819,
       "step": 411
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.987878787878788e-06,
-      "loss": 0.8641,
+      "loss": 1.0363,
       "step": 412
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0012121212121213e-05,
-      "loss": 0.9871,
+      "loss": 1.1445,
       "step": 413
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0036363636363637e-05,
-      "loss": 0.8251,
+      "loss": 0.9869,
       "step": 414
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0060606060606062e-05,
-      "loss": 0.8833,
+      "loss": 1.0858,
       "step": 415
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0084848484848486e-05,
-      "loss": 1.0146,
+      "loss": 1.0154,
       "step": 416
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.010909090909091e-05,
-      "loss": 1.0311,
+      "loss": 1.1058,
       "step": 417
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0133333333333335e-05,
-      "loss": 1.0182,
+      "loss": 1.112,
       "step": 418
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0157575757575759e-05,
-      "loss": 1.0547,
+      "loss": 1.0912,
       "step": 419
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0181818181818182e-05,
-      "loss": 0.9456,
+      "loss": 0.97,
       "step": 420
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0206060606060606e-05,
-      "loss": 0.8888,
+      "loss": 1.0381,
       "step": 421
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.023030303030303e-05,
-      "loss": 0.8695,
+      "loss": 1.0171,
       "step": 422
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0254545454545455e-05,
-      "loss": 1.0685,
+      "loss": 1.0971,
       "step": 423
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0278787878787879e-05,
-      "loss": 0.8322,
+      "loss": 1.0432,
       "step": 424
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0303030303030304e-05,
-      "loss": 0.939,
+      "loss": 0.9634,
       "step": 425
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0327272727272728e-05,
-      "loss": 1.047,
+      "loss": 1.0842,
       "step": 426
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0351515151515153e-05,
-      "loss": 0.9616,
+      "loss": 0.9891,
       "step": 427
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0375757575757577e-05,
-      "loss": 1.0043,
+      "loss": 1.0869,
       "step": 428
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.04e-05,
-      "loss": 1.0298,
+      "loss": 1.0929,
       "step": 429
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0424242424242426e-05,
-      "loss": 0.9956,
+      "loss": 1.1176,
       "step": 430
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.044848484848485e-05,
-      "loss": 1.014,
+      "loss": 1.0727,
       "step": 431
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0472727272727275e-05,
-      "loss": 0.9547,
+      "loss": 0.9889,
       "step": 432
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0496969696969698e-05,
-      "loss": 1.0705,
+      "loss": 1.0784,
       "step": 433
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.052121212121212e-05,
-      "loss": 1.0069,
+      "loss": 1.0189,
       "step": 434
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0545454545454546e-05,
-      "loss": 1.0002,
+      "loss": 1.0396,
       "step": 435
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.056969696969697e-05,
-      "loss": 1.1284,
+      "loss": 1.1684,
       "step": 436
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0593939393939395e-05,
-      "loss": 0.9493,
+      "loss": 1.0258,
       "step": 437
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0618181818181818e-05,
-      "loss": 0.8569,
+      "loss": 0.9882,
       "step": 438
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0642424242424244e-05,
-      "loss": 1.0222,
+      "loss": 1.1191,
       "step": 439
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0666666666666667e-05,
-      "loss": 0.9901,
+      "loss": 1.1473,
       "step": 440
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0690909090909091e-05,
-      "loss": 0.9697,
+      "loss": 1.0214,
       "step": 441
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0715151515151516e-05,
-      "loss": 0.8931,
+      "loss": 0.9852,
       "step": 442
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.073939393939394e-05,
-      "loss": 1.0044,
+      "loss": 1.0284,
       "step": 443
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0763636363636365e-05,
-      "loss": 0.9779,
+      "loss": 1.049,
       "step": 444
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0787878787878789e-05,
-      "loss": 1.0327,
+      "loss": 1.075,
       "step": 445
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0812121212121214e-05,
-      "loss": 0.9652,
+      "loss": 1.0023,
       "step": 446
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0836363636363638e-05,
-      "loss": 1.0812,
+      "loss": 1.1312,
       "step": 447
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0860606060606063e-05,
-      "loss": 0.8544,
+      "loss": 1.0391,
       "step": 448
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0884848484848485e-05,
-      "loss": 1.0157,
+      "loss": 1.0466,
       "step": 449
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0909090909090909e-05,
-      "loss": 0.9223,
+      "loss": 1.0685,
       "step": 450
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0933333333333334e-05,
-      "loss": 1.0549,
+      "loss": 1.1218,
       "step": 451
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0957575757575758e-05,
-      "loss": 1.0783,
+      "loss": 1.0737,
       "step": 452
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.0981818181818182e-05,
-      "loss": 1.0176,
+      "loss": 1.064,
       "step": 453
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1006060606060607e-05,
-      "loss": 0.968,
+      "loss": 1.0843,
       "step": 454
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.103030303030303e-05,
-      "loss": 0.9172,
+      "loss": 1.0411,
       "step": 455
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1054545454545456e-05,
-      "loss": 0.9946,
+      "loss": 0.9615,
       "step": 456
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.107878787878788e-05,
-      "loss": 1.0086,
+      "loss": 1.0452,
       "step": 457
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1103030303030305e-05,
-      "loss": 0.8899,
+      "loss": 0.9723,
       "step": 458
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1127272727272729e-05,
-      "loss": 1.0049,
+      "loss": 1.0718,
       "step": 459
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1151515151515154e-05,
-      "loss": 1.068,
+      "loss": 1.1296,
       "step": 460
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1175757575757578e-05,
-      "loss": 1.0269,
+      "loss": 1.0886,
       "step": 461
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1200000000000001e-05,
-      "loss": 1.0768,
+      "loss": 1.1447,
       "step": 462
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1224242424242425e-05,
-      "loss": 0.9727,
+      "loss": 1.0323,
       "step": 463
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1248484848484848e-05,
-      "loss": 1.0385,
+      "loss": 1.1717,
       "step": 464
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1272727272727272e-05,
-      "loss": 0.9641,
+      "loss": 1.0784,
       "step": 465
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1296969696969697e-05,
-      "loss": 0.9472,
+      "loss": 1.0706,
       "step": 466
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1321212121212121e-05,
-      "loss": 1.0099,
+      "loss": 1.0966,
       "step": 467
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1345454545454546e-05,
-      "loss": 0.9694,
+      "loss": 1.0104,
       "step": 468
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.136969696969697e-05,
-      "loss": 0.9368,
+      "loss": 1.0489,
       "step": 469
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1393939393939395e-05,
-      "loss": 0.981,
+      "loss": 1.1174,
       "step": 470
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1418181818181819e-05,
-      "loss": 1.0309,
+      "loss": 1.0908,
       "step": 471
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1442424242424244e-05,
-      "loss": 0.9067,
+      "loss": 1.0419,
       "step": 472
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1466666666666668e-05,
-      "loss": 1.0327,
+      "loss": 1.0299,
       "step": 473
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1490909090909092e-05,
-      "loss": 0.9264,
+      "loss": 1.0361,
       "step": 474
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1515151515151517e-05,
-      "loss": 1.038,
+      "loss": 1.068,
       "step": 475
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.153939393939394e-05,
-      "loss": 1.0631,
+      "loss": 1.1443,
       "step": 476
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1563636363636363e-05,
-      "loss": 1.023,
+      "loss": 1.0095,
       "step": 477
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1587878787878788e-05,
-      "loss": 0.9725,
+      "loss": 0.9879,
       "step": 478
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1612121212121212e-05,
-      "loss": 1.0072,
+      "loss": 0.9979,
       "step": 479
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1636363636363637e-05,
-      "loss": 0.9189,
+      "loss": 1.0462,
       "step": 480
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.166060606060606e-05,
-      "loss": 0.9744,
+      "loss": 1.0579,
       "step": 481
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1684848484848486e-05,
-      "loss": 0.975,
+      "loss": 1.0844,
       "step": 482
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.170909090909091e-05,
-      "loss": 0.8733,
+      "loss": 1.0277,
       "step": 483
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1733333333333335e-05,
-      "loss": 1.1056,
+      "loss": 1.0796,
       "step": 484
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1757575757575759e-05,
-      "loss": 1.1273,
+      "loss": 1.1377,
       "step": 485
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1781818181818182e-05,
-      "loss": 0.9313,
+      "loss": 1.0529,
       "step": 486
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1806060606060608e-05,
-      "loss": 1.0004,
+      "loss": 1.0316,
       "step": 487
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1830303030303031e-05,
-      "loss": 1.0353,
+      "loss": 1.1453,
       "step": 488
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1854545454545457e-05,
-      "loss": 1.0148,
+      "loss": 1.0503,
       "step": 489
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.187878787878788e-05,
-      "loss": 0.8915,
+      "loss": 1.0572,
       "step": 490
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1903030303030302e-05,
-      "loss": 1.0799,
+      "loss": 1.1099,
       "step": 491
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1927272727272728e-05,
-      "loss": 0.8135,
+      "loss": 0.8896,
       "step": 492
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1951515151515151e-05,
-      "loss": 1.0939,
+      "loss": 1.0934,
       "step": 493
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.1975757575757577e-05,
-      "loss": 0.9223,
+      "loss": 1.0428,
       "step": 494
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2e-05,
-      "loss": 1.0322,
+      "loss": 1.0079,
       "step": 495
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2024242424242426e-05,
-      "loss": 0.9416,
+      "loss": 1.0134,
       "step": 496
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.204848484848485e-05,
-      "loss": 1.113,
+      "loss": 1.0882,
       "step": 497
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2072727272727273e-05,
-      "loss": 0.8567,
+      "loss": 1.0088,
       "step": 498
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2096969696969698e-05,
-      "loss": 1.191,
+      "loss": 1.1289,
       "step": 499
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2121212121212122e-05,
-      "loss": 1.0619,
+      "loss": 1.0567,
       "step": 500
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2145454545454547e-05,
-      "loss": 1.0834,
+      "loss": 1.1013,
       "step": 501
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2169696969696971e-05,
-      "loss": 1.0083,
+      "loss": 1.0789,
       "step": 502
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2193939393939396e-05,
-      "loss": 1.0486,
+      "loss": 1.141,
       "step": 503
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.221818181818182e-05,
-      "loss": 1.0092,
+      "loss": 1.0219,
       "step": 504
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2242424242424242e-05,
-      "loss": 0.8847,
+      "loss": 1.0247,
       "step": 505
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2266666666666667e-05,
-      "loss": 0.9261,
+      "loss": 1.0378,
       "step": 506
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2290909090909091e-05,
-      "loss": 1.1131,
+      "loss": 1.1577,
       "step": 507
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2315151515151516e-05,
-      "loss": 0.8491,
+      "loss": 0.9851,
       "step": 508
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.233939393939394e-05,
-      "loss": 1.0694,
+      "loss": 1.1039,
       "step": 509
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2363636363636364e-05,
-      "loss": 0.8828,
+      "loss": 1.0101,
       "step": 510
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2387878787878789e-05,
-      "loss": 0.9609,
+      "loss": 1.1105,
       "step": 511
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2412121212121213e-05,
-      "loss": 1.0473,
+      "loss": 1.1359,
       "step": 512
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2436363636363638e-05,
-      "loss": 0.9523,
+      "loss": 1.0542,
       "step": 513
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2460606060606062e-05,
-      "loss": 0.9319,
+      "loss": 1.059,
       "step": 514
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2484848484848487e-05,
-      "loss": 0.988,
+      "loss": 1.0781,
       "step": 515
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.250909090909091e-05,
-      "loss": 1.0002,
+      "loss": 1.0451,
       "step": 516
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2533333333333336e-05,
-      "loss": 1.0516,
+      "loss": 1.1246,
       "step": 517
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.255757575757576e-05,
-      "loss": 0.8629,
+      "loss": 1.0107,
       "step": 518
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2581818181818183e-05,
-      "loss": 0.8519,
+      "loss": 0.9545,
       "step": 519
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2606060606060607e-05,
-      "loss": 1.019,
+      "loss": 1.0973,
       "step": 520
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.263030303030303e-05,
-      "loss": 0.9752,
+      "loss": 1.0915,
       "step": 521
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2654545454545454e-05,
-      "loss": 0.9368,
+      "loss": 1.0549,
       "step": 522
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.267878787878788e-05,
-      "loss": 0.9732,
+      "loss": 0.9928,
       "step": 523
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2703030303030303e-05,
-      "loss": 1.02,
+      "loss": 1.0019,
       "step": 524
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2727272727272728e-05,
-      "loss": 0.9222,
+      "loss": 1.0242,
       "step": 525
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2751515151515152e-05,
-      "loss": 0.9656,
+      "loss": 1.066,
       "step": 526
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2775757575757577e-05,
-      "loss": 0.8636,
+      "loss": 0.9562,
       "step": 527
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2800000000000001e-05,
-      "loss": 0.9193,
+      "loss": 1.038,
       "step": 528
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2824242424242426e-05,
-      "loss": 0.8901,
+      "loss": 1.063,
       "step": 529
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.284848484848485e-05,
-      "loss": 0.9619,
+      "loss": 0.9942,
       "step": 530
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2872727272727274e-05,
-      "loss": 1.0383,
+      "loss": 1.0444,
       "step": 531
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2896969696969699e-05,
-      "loss": 1.1451,
+      "loss": 1.0988,
       "step": 532
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2921212121212123e-05,
-      "loss": 0.9832,
+      "loss": 1.0337,
       "step": 533
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2945454545454545e-05,
-      "loss": 0.9171,
+      "loss": 1.0144,
       "step": 534
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.296969696969697e-05,
-      "loss": 1.056,
+      "loss": 1.0998,
       "step": 535
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.2993939393939394e-05,
-      "loss": 1.0854,
+      "loss": 1.0763,
       "step": 536
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3018181818181819e-05,
-      "loss": 0.9858,
+      "loss": 1.0493,
       "step": 537
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3042424242424243e-05,
-      "loss": 0.8495,
+      "loss": 1.0513,
       "step": 538
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3066666666666668e-05,
-      "loss": 0.9823,
+      "loss": 1.0863,
       "step": 539
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3090909090909092e-05,
-      "loss": 0.8875,
+      "loss": 0.9913,
       "step": 540
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3115151515151517e-05,
-      "loss": 0.9793,
+      "loss": 1.103,
       "step": 541
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.313939393939394e-05,
-      "loss": 0.7912,
+      "loss": 0.9059,
       "step": 542
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3163636363636364e-05,
-      "loss": 0.9403,
+      "loss": 1.0531,
       "step": 543
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.318787878787879e-05,
-      "loss": 0.9013,
+      "loss": 0.9603,
       "step": 544
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3212121212121213e-05,
-      "loss": 0.9833,
+      "loss": 1.0276,
       "step": 545
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3236363636363639e-05,
-      "loss": 0.9442,
+      "loss": 1.0542,
       "step": 546
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3260606060606062e-05,
-      "loss": 0.9821,
+      "loss": 1.0941,
       "step": 547
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3284848484848484e-05,
-      "loss": 1.0125,
+      "loss": 1.0295,
       "step": 548
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.330909090909091e-05,
-      "loss": 0.95,
+      "loss": 1.0279,
       "step": 549
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 1.0859,
+      "loss": 1.1381,
       "step": 550
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3357575757575759e-05,
-      "loss": 1.0045,
+      "loss": 1.072,
       "step": 551
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3381818181818182e-05,
-      "loss": 1.0347,
+      "loss": 1.0972,
       "step": 552
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3406060606060608e-05,
-      "loss": 1.0616,
+      "loss": 1.0638,
       "step": 553
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3430303030303031e-05,
-      "loss": 1.0905,
+      "loss": 1.1154,
       "step": 554
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3454545454545455e-05,
-      "loss": 0.9241,
+      "loss": 1.0634,
       "step": 555
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.347878787878788e-05,
-      "loss": 0.9914,
+      "loss": 1.0367,
       "step": 556
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3503030303030304e-05,
-      "loss": 1.0598,
+      "loss": 1.1603,
       "step": 557
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.352727272727273e-05,
-      "loss": 1.0872,
+      "loss": 1.1311,
       "step": 558
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3551515151515153e-05,
-      "loss": 0.9897,
+      "loss": 1.0533,
       "step": 559
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3575757575757578e-05,
-      "loss": 0.9791,
+      "loss": 1.0389,
       "step": 560
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3600000000000002e-05,
-      "loss": 1.0258,
+      "loss": 1.0137,
       "step": 561
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3624242424242424e-05,
-      "loss": 1.0731,
+      "loss": 1.135,
       "step": 562
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.364848484848485e-05,
-      "loss": 1.0327,
+      "loss": 1.0206,
       "step": 563
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3672727272727273e-05,
-      "loss": 1.0211,
+      "loss": 1.1205,
       "step": 564
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3696969696969698e-05,
-      "loss": 0.8882,
+      "loss": 1.0496,
       "step": 565
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3721212121212122e-05,
-      "loss": 1.0481,
+      "loss": 1.0591,
       "step": 566
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3745454545454546e-05,
-      "loss": 0.9914,
+      "loss": 1.079,
       "step": 567
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3769696969696971e-05,
-      "loss": 0.9851,
+      "loss": 1.1118,
       "step": 568
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3793939393939395e-05,
-      "loss": 0.9847,
+      "loss": 0.9971,
       "step": 569
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.381818181818182e-05,
-      "loss": 1.0181,
+      "loss": 1.0388,
       "step": 570
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3842424242424244e-05,
-      "loss": 0.8444,
+      "loss": 0.9811,
       "step": 571
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3866666666666669e-05,
-      "loss": 1.0413,
+      "loss": 1.0959,
       "step": 572
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3890909090909093e-05,
-      "loss": 1.0215,
+      "loss": 1.0942,
       "step": 573
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3915151515151518e-05,
-      "loss": 0.9097,
+      "loss": 1.0127,
       "step": 574
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3939393939393942e-05,
-      "loss": 0.9082,
+      "loss": 0.9933,
       "step": 575
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3963636363636365e-05,
-      "loss": 0.9972,
+      "loss": 1.0132,
       "step": 576
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.3987878787878789e-05,
-      "loss": 0.9569,
+      "loss": 1.0194,
       "step": 577
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4012121212121212e-05,
-      "loss": 0.9753,
+      "loss": 1.061,
       "step": 578
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4036363636363636e-05,
-      "loss": 0.9083,
+      "loss": 1.0018,
       "step": 579
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4060606060606061e-05,
-      "loss": 0.9359,
+      "loss": 0.9986,
       "step": 580
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4084848484848485e-05,
-      "loss": 0.956,
+      "loss": 1.0283,
       "step": 581
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.410909090909091e-05,
-      "loss": 1.0011,
+      "loss": 1.108,
       "step": 582
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4133333333333334e-05,
-      "loss": 1.0084,
+      "loss": 1.0632,
       "step": 583
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.415757575757576e-05,
-      "loss": 0.9568,
+      "loss": 1.0261,
       "step": 584
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4181818181818183e-05,
-      "loss": 1.1362,
+      "loss": 1.2462,
       "step": 585
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4206060606060608e-05,
-      "loss": 0.9944,
+      "loss": 1.0516,
       "step": 586
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4230303030303032e-05,
-      "loss": 1.0917,
+      "loss": 1.0841,
       "step": 587
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4254545454545456e-05,
-      "loss": 0.9279,
+      "loss": 1.0542,
       "step": 588
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4278787878787881e-05,
-      "loss": 0.9725,
+      "loss": 1.0695,
       "step": 589
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4303030303030305e-05,
-      "loss": 0.957,
+      "loss": 1.037,
       "step": 590
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4327272727272727e-05,
-      "loss": 0.9619,
+      "loss": 1.0173,
       "step": 591
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4351515151515152e-05,
-      "loss": 0.9181,
+      "loss": 0.9724,
       "step": 592
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4375757575757576e-05,
-      "loss": 0.9846,
+      "loss": 1.1323,
       "step": 593
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4400000000000001e-05,
-      "loss": 1.0157,
+      "loss": 1.1108,
       "step": 594
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4424242424242425e-05,
-      "loss": 0.9289,
+      "loss": 1.0025,
       "step": 595
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.444848484848485e-05,
-      "loss": 0.8975,
+      "loss": 1.0844,
       "step": 596
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4472727272727274e-05,
-      "loss": 1.0571,
+      "loss": 1.0995,
       "step": 597
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4496969696969699e-05,
-      "loss": 0.962,
+      "loss": 1.0646,
       "step": 598
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4521212121212123e-05,
-      "loss": 0.9789,
+      "loss": 1.0448,
       "step": 599
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4545454545454546e-05,
-      "loss": 0.9005,
+      "loss": 1.0118,
       "step": 600
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4569696969696972e-05,
-      "loss": 1.0008,
+      "loss": 0.9843,
       "step": 601
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4593939393939395e-05,
-      "loss": 1.0435,
+      "loss": 1.1102,
       "step": 602
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.461818181818182e-05,
-      "loss": 0.9572,
+      "loss": 1.1296,
       "step": 603
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4642424242424244e-05,
-      "loss": 0.9261,
+      "loss": 1.0051,
       "step": 604
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4666666666666666e-05,
-      "loss": 1.0231,
+      "loss": 1.1396,
       "step": 605
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4690909090909092e-05,
-      "loss": 1.107,
+      "loss": 1.1175,
       "step": 606
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4715151515151515e-05,
-      "loss": 0.9901,
+      "loss": 1.0105,
       "step": 607
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.473939393939394e-05,
-      "loss": 0.9757,
+      "loss": 1.1188,
       "step": 608
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4763636363636364e-05,
-      "loss": 0.8766,
+      "loss": 0.9895,
       "step": 609
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.478787878787879e-05,
-      "loss": 1.0567,
+      "loss": 1.0925,
       "step": 610
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4812121212121213e-05,
-      "loss": 0.826,
+      "loss": 0.9293,
       "step": 611
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4836363636363637e-05,
-      "loss": 0.9584,
+      "loss": 1.0118,
       "step": 612
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4860606060606062e-05,
-      "loss": 1.1051,
+      "loss": 1.1557,
       "step": 613
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4884848484848486e-05,
-      "loss": 1.0836,
+      "loss": 1.0852,
       "step": 614
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4909090909090911e-05,
-      "loss": 0.9754,
+      "loss": 1.0577,
       "step": 615
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4933333333333335e-05,
-      "loss": 0.9102,
+      "loss": 1.0917,
       "step": 616
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.495757575757576e-05,
-      "loss": 0.918,
+      "loss": 1.0173,
       "step": 617
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.4981818181818184e-05,
-      "loss": 0.9744,
+      "loss": 1.0276,
       "step": 618
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5006060606060606e-05,
-      "loss": 1.055,
+      "loss": 1.1153,
       "step": 619
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5030303030303031e-05,
-      "loss": 1.0228,
+      "loss": 1.0283,
       "step": 620
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5054545454545455e-05,
-      "loss": 1.0556,
+      "loss": 1.129,
       "step": 621
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.507878787878788e-05,
-      "loss": 0.8741,
+      "loss": 0.9767,
       "step": 622
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5103030303030304e-05,
-      "loss": 0.9457,
+      "loss": 1.0195,
       "step": 623
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5127272727272728e-05,
-      "loss": 1.0231,
+      "loss": 1.0868,
       "step": 624
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5151515151515153e-05,
-      "loss": 0.8835,
+      "loss": 1.011,
       "step": 625
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5175757575757577e-05,
-      "loss": 0.9531,
+      "loss": 0.9949,
       "step": 626
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5200000000000002e-05,
-      "loss": 1.0057,
+      "loss": 1.146,
       "step": 627
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5224242424242426e-05,
-      "loss": 0.9568,
+      "loss": 1.0413,
       "step": 628
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5248484848484851e-05,
-      "loss": 1.0152,
+      "loss": 1.0937,
       "step": 629
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5272727272727276e-05,
-      "loss": 1.0332,
+      "loss": 1.0901,
       "step": 630
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5296969696969698e-05,
-      "loss": 1.0295,
+      "loss": 1.0497,
       "step": 631
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5321212121212123e-05,
-      "loss": 1.0125,
+      "loss": 1.1093,
       "step": 632
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5345454545454545e-05,
-      "loss": 0.9171,
+      "loss": 0.9405,
       "step": 633
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.536969696969697e-05,
-      "loss": 0.9839,
+      "loss": 1.0788,
       "step": 634
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5393939393939393e-05,
-      "loss": 0.952,
+      "loss": 1.0066,
       "step": 635
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5418181818181818e-05,
-      "loss": 0.8418,
+      "loss": 1.067,
       "step": 636
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5442424242424243e-05,
-      "loss": 0.9028,
+      "loss": 1.0146,
       "step": 637
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.546666666666667e-05,
-      "loss": 1.0102,
+      "loss": 1.08,
       "step": 638
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.549090909090909e-05,
-      "loss": 1.0538,
+      "loss": 1.0784,
       "step": 639
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5515151515151516e-05,
-      "loss": 0.7842,
+      "loss": 0.9784,
       "step": 640
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.553939393939394e-05,
-      "loss": 1.1426,
+      "loss": 1.1594,
       "step": 641
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5563636363636367e-05,
-      "loss": 0.894,
+      "loss": 1.0221,
       "step": 642
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.558787878787879e-05,
-      "loss": 0.9069,
+      "loss": 1.0088,
       "step": 643
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5612121212121214e-05,
-      "loss": 0.9755,
+      "loss": 1.0565,
       "step": 644
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.563636363636364e-05,
-      "loss": 0.9317,
+      "loss": 1.0161,
       "step": 645
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.566060606060606e-05,
-      "loss": 1.1376,
+      "loss": 1.148,
       "step": 646
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5684848484848487e-05,
-      "loss": 0.9953,
+      "loss": 1.1106,
       "step": 647
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.570909090909091e-05,
-      "loss": 0.9614,
+      "loss": 1.048,
       "step": 648
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5733333333333334e-05,
-      "loss": 0.9209,
+      "loss": 1.0685,
       "step": 649
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.575757575757576e-05,
-      "loss": 0.8784,
+      "loss": 0.9701,
       "step": 650
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.578181818181818e-05,
-      "loss": 1.0142,
+      "loss": 0.9781,
       "step": 651
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5806060606060607e-05,
-      "loss": 0.9909,
+      "loss": 1.0145,
       "step": 652
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5830303030303032e-05,
-      "loss": 1.0881,
+      "loss": 1.1639,
       "step": 653
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5854545454545457e-05,
-      "loss": 1.0133,
+      "loss": 1.1082,
       "step": 654
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.587878787878788e-05,
-      "loss": 0.9289,
+      "loss": 0.9752,
       "step": 655
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5903030303030305e-05,
-      "loss": 0.9013,
+      "loss": 0.9469,
       "step": 656
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.592727272727273e-05,
-      "loss": 0.8373,
+      "loss": 0.942,
       "step": 657
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5951515151515152e-05,
-      "loss": 0.7811,
+      "loss": 0.9494,
       "step": 658
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.5975757575757577e-05,
-      "loss": 0.8992,
+      "loss": 0.9746,
       "step": 659
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.9581,
+      "loss": 1.0843,
       "step": 660
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6024242424242428e-05,
-      "loss": 1.0722,
+      "loss": 1.1459,
       "step": 661
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.604848484848485e-05,
-      "loss": 1.0634,
+      "loss": 1.0779,
       "step": 662
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6072727272727272e-05,
-      "loss": 0.9036,
+      "loss": 1.015,
       "step": 663
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6096969696969697e-05,
-      "loss": 0.8986,
+      "loss": 0.9786,
       "step": 664
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6121212121212123e-05,
-      "loss": 0.9232,
+      "loss": 1.0502,
       "step": 665
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6145454545454548e-05,
-      "loss": 0.9969,
+      "loss": 1.0781,
       "step": 666
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.616969696969697e-05,
-      "loss": 0.9898,
+      "loss": 1.0151,
       "step": 667
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6193939393939395e-05,
-      "loss": 0.9309,
+      "loss": 1.1122,
       "step": 668
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.621818181818182e-05,
-      "loss": 0.956,
+      "loss": 1.043,
       "step": 669
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6242424242424243e-05,
-      "loss": 0.9805,
+      "loss": 1.0934,
       "step": 670
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6266666666666668e-05,
-      "loss": 0.9619,
+      "loss": 1.064,
       "step": 671
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6290909090909093e-05,
-      "loss": 1.0583,
+      "loss": 1.1047,
       "step": 672
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.631515151515152e-05,
-      "loss": 0.9552,
+      "loss": 1.036,
       "step": 673
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.633939393939394e-05,
-      "loss": 0.972,
+      "loss": 1.1083,
       "step": 674
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6363636363636366e-05,
-      "loss": 0.953,
+      "loss": 1.0599,
       "step": 675
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6387878787878788e-05,
-      "loss": 1.0192,
+      "loss": 1.126,
       "step": 676
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6412121212121213e-05,
-      "loss": 1.0088,
+      "loss": 1.0807,
       "step": 677
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.643636363636364e-05,
-      "loss": 0.9689,
+      "loss": 0.9948,
       "step": 678
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.646060606060606e-05,
-      "loss": 0.9942,
+      "loss": 1.087,
       "step": 679
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6484848484848486e-05,
-      "loss": 0.8764,
+      "loss": 1.0257,
       "step": 680
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.650909090909091e-05,
-      "loss": 0.925,
+      "loss": 1.0425,
       "step": 681
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6533333333333333e-05,
-      "loss": 0.9956,
+      "loss": 1.068,
       "step": 682
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.655757575757576e-05,
-      "loss": 0.9248,
+      "loss": 1.0609,
       "step": 683
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6581818181818184e-05,
-      "loss": 1.0151,
+      "loss": 1.0708,
       "step": 684
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.660606060606061e-05,
-      "loss": 1.0272,
+      "loss": 1.0738,
       "step": 685
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.663030303030303e-05,
-      "loss": 1.0319,
+      "loss": 1.0603,
       "step": 686
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.6654545454545456e-05,
-      "loss": 0.9955,
+      "loss": 1.0701,
       "step": 687
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6678787878787882e-05,
-      "loss": 0.9018,
+      "loss": 0.9855,
       "step": 688
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6703030303030304e-05,
-      "loss": 0.9634,
+      "loss": 1.0517,
       "step": 689
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.672727272727273e-05,
-      "loss": 0.9531,
+      "loss": 1.008,
       "step": 690
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.675151515151515e-05,
-      "loss": 1.0959,
+      "loss": 1.1395,
       "step": 691
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6775757575757576e-05,
-      "loss": 0.9236,
+      "loss": 1.0521,
       "step": 692
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6800000000000002e-05,
-      "loss": 1.0477,
+      "loss": 1.0774,
       "step": 693
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6824242424242424e-05,
-      "loss": 1.041,
+      "loss": 1.0734,
       "step": 694
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.684848484848485e-05,
-      "loss": 0.8647,
+      "loss": 0.9955,
       "step": 695
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6872727272727274e-05,
-      "loss": 1.0141,
+      "loss": 1.1395,
       "step": 696
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.68969696969697e-05,
-      "loss": 0.9065,
+      "loss": 1.0361,
       "step": 697
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6921212121212122e-05,
-      "loss": 0.8793,
+      "loss": 0.9775,
       "step": 698
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6945454545454547e-05,
-      "loss": 1.0113,
+      "loss": 1.1015,
       "step": 699
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6969696969696972e-05,
-      "loss": 0.8582,
+      "loss": 0.999,
       "step": 700
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.6993939393939394e-05,
-      "loss": 1.013,
+      "loss": 1.0663,
       "step": 701
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.701818181818182e-05,
-      "loss": 1.1261,
+      "loss": 1.0933,
       "step": 702
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7042424242424245e-05,
-      "loss": 1.033,
+      "loss": 1.0552,
       "step": 703
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.706666666666667e-05,
-      "loss": 0.9991,
+      "loss": 1.0753,
       "step": 704
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7090909090909092e-05,
-      "loss": 0.8682,
+      "loss": 0.9709,
       "step": 705
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7115151515151514e-05,
-      "loss": 0.9728,
+      "loss": 1.1252,
       "step": 706
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.713939393939394e-05,
-      "loss": 0.9592,
+      "loss": 1.0238,
       "step": 707
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7163636363636365e-05,
-      "loss": 0.838,
+      "loss": 1.0189,
       "step": 708
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.718787878787879e-05,
-      "loss": 1.056,
+      "loss": 1.1003,
       "step": 709
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7212121212121212e-05,
-      "loss": 1.074,
+      "loss": 1.0559,
       "step": 710
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7236363636363638e-05,
-      "loss": 0.9141,
+      "loss": 1.0078,
       "step": 711
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7260606060606063e-05,
-      "loss": 0.9914,
+      "loss": 1.1169,
       "step": 712
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7284848484848485e-05,
-      "loss": 0.9562,
+      "loss": 1.0822,
       "step": 713
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.730909090909091e-05,
-      "loss": 0.8734,
+      "loss": 1.023,
       "step": 714
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7333333333333336e-05,
-      "loss": 0.8505,
+      "loss": 0.9726,
       "step": 715
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.735757575757576e-05,
-      "loss": 0.9588,
+      "loss": 1.0283,
       "step": 716
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7381818181818183e-05,
-      "loss": 1.0804,
+      "loss": 1.0367,
       "step": 717
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7406060606060608e-05,
-      "loss": 0.9118,
+      "loss": 0.9965,
       "step": 718
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.743030303030303e-05,
-      "loss": 0.951,
+      "loss": 1.0283,
       "step": 719
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7454545454545456e-05,
-      "loss": 0.8858,
+      "loss": 1.0125,
       "step": 720
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.747878787878788e-05,
-      "loss": 0.9343,
+      "loss": 1.0063,
       "step": 721
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7503030303030303e-05,
-      "loss": 1.0188,
+      "loss": 1.0626,
       "step": 722
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7527272727272728e-05,
-      "loss": 0.9626,
+      "loss": 1.0081,
       "step": 723
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7551515151515154e-05,
-      "loss": 1.0452,
+      "loss": 1.0616,
       "step": 724
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7575757575757576e-05,
-      "loss": 0.9116,
+      "loss": 1.0149,
       "step": 725
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.76e-05,
-      "loss": 0.9724,
+      "loss": 1.0779,
       "step": 726
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7624242424242426e-05,
-      "loss": 0.9854,
+      "loss": 1.01,
       "step": 727
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.764848484848485e-05,
-      "loss": 0.909,
+      "loss": 0.9168,
       "step": 728
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7672727272727274e-05,
-      "loss": 1.0143,
+      "loss": 1.1046,
       "step": 729
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.76969696969697e-05,
-      "loss": 0.9732,
+      "loss": 1.0905,
       "step": 730
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7721212121212124e-05,
-      "loss": 0.9978,
+      "loss": 1.084,
       "step": 731
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.774545454545455e-05,
-      "loss": 0.8226,
+      "loss": 0.9565,
       "step": 732
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.776969696969697e-05,
-      "loss": 1.0103,
+      "loss": 1.1307,
       "step": 733
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7793939393939393e-05,
-      "loss": 0.9009,
+      "loss": 1.0143,
       "step": 734
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.781818181818182e-05,
-      "loss": 1.0126,
+      "loss": 1.0889,
       "step": 735
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7842424242424244e-05,
-      "loss": 1.0517,
+      "loss": 1.0579,
       "step": 736
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7866666666666666e-05,
-      "loss": 0.9897,
+      "loss": 1.0881,
       "step": 737
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.789090909090909e-05,
-      "loss": 0.8305,
+      "loss": 0.9915,
       "step": 738
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7915151515151517e-05,
-      "loss": 0.9112,
+      "loss": 1.0093,
       "step": 739
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7939393939393942e-05,
-      "loss": 0.92,
+      "loss": 1.0629,
       "step": 740
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.7963636363636364e-05,
-      "loss": 0.9519,
+      "loss": 1.062,
       "step": 741
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.798787878787879e-05,
-      "loss": 1.0342,
+      "loss": 1.0116,
       "step": 742
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8012121212121215e-05,
-      "loss": 0.8543,
+      "loss": 0.9752,
       "step": 743
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.803636363636364e-05,
-      "loss": 0.8812,
+      "loss": 1.0138,
       "step": 744
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8060606060606062e-05,
-      "loss": 1.0093,
+      "loss": 1.0671,
       "step": 745
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8084848484848487e-05,
-      "loss": 0.9857,
+      "loss": 1.0901,
       "step": 746
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.810909090909091e-05,
-      "loss": 1.0139,
+      "loss": 1.0874,
       "step": 747
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8133333333333335e-05,
-      "loss": 0.8121,
+      "loss": 0.9506,
       "step": 748
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8157575757575757e-05,
-      "loss": 0.981,
+      "loss": 1.0797,
       "step": 749
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8181818181818182e-05,
-      "loss": 1.0502,
+      "loss": 1.0862,
       "step": 750
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8206060606060607e-05,
-      "loss": 0.9649,
+      "loss": 0.9783,
       "step": 751
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8230303030303033e-05,
-      "loss": 1.0551,
+      "loss": 1.0987,
       "step": 752
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8254545454545455e-05,
-      "loss": 0.8858,
+      "loss": 1.0103,
       "step": 753
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.827878787878788e-05,
-      "loss": 1.083,
+      "loss": 1.1352,
       "step": 754
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8303030303030305e-05,
-      "loss": 0.917,
+      "loss": 1.0145,
       "step": 755
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.832727272727273e-05,
-      "loss": 0.9982,
+      "loss": 1.1152,
       "step": 756
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8351515151515153e-05,
-      "loss": 1.032,
+      "loss": 1.1345,
       "step": 757
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8375757575757578e-05,
-      "loss": 0.8538,
+      "loss": 0.9539,
       "step": 758
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8400000000000003e-05,
-      "loss": 0.9556,
+      "loss": 0.9581,
       "step": 759
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8424242424242425e-05,
-      "loss": 0.9448,
+      "loss": 0.9607,
       "step": 760
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8448484848484847e-05,
-      "loss": 0.9531,
+      "loss": 1.0474,
       "step": 761
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8472727272727273e-05,
-      "loss": 0.9376,
+      "loss": 1.0098,
       "step": 762
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8496969696969698e-05,
-      "loss": 0.8279,
+      "loss": 0.9423,
       "step": 763
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8521212121212123e-05,
-      "loss": 0.9463,
+      "loss": 1.0048,
       "step": 764
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8545454545454545e-05,
-      "loss": 0.9706,
+      "loss": 1.1053,
       "step": 765
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.856969696969697e-05,
-      "loss": 0.9128,
+      "loss": 1.0446,
       "step": 766
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8593939393939396e-05,
-      "loss": 1.0854,
+      "loss": 1.0579,
       "step": 767
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.861818181818182e-05,
-      "loss": 1.0629,
+      "loss": 1.1847,
       "step": 768
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8642424242424243e-05,
-      "loss": 0.9473,
+      "loss": 0.993,
       "step": 769
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.866666666666667e-05,
-      "loss": 1.109,
+      "loss": 1.1023,
       "step": 770
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8690909090909094e-05,
-      "loss": 0.994,
+      "loss": 1.0441,
       "step": 771
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8715151515151516e-05,
-      "loss": 0.9925,
+      "loss": 1.0193,
       "step": 772
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.873939393939394e-05,
-      "loss": 0.8597,
+      "loss": 0.978,
       "step": 773
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8763636363636367e-05,
-      "loss": 1.0354,
+      "loss": 1.0784,
       "step": 774
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8787878787878792e-05,
-      "loss": 0.9381,
+      "loss": 1.1028,
       "step": 775
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8812121212121214e-05,
-      "loss": 0.9846,
+      "loss": 1.099,
       "step": 776
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8836363636363636e-05,
-      "loss": 0.9796,
+      "loss": 1.0368,
       "step": 777
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.886060606060606e-05,
-      "loss": 1.0134,
+      "loss": 1.0657,
       "step": 778
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8884848484848487e-05,
-      "loss": 0.8531,
+      "loss": 1.0082,
       "step": 779
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8909090909090912e-05,
-      "loss": 0.9282,
+      "loss": 1.0417,
       "step": 780
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8933333333333334e-05,
-      "loss": 0.9873,
+      "loss": 1.0655,
       "step": 781
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.895757575757576e-05,
-      "loss": 0.8835,
+      "loss": 0.9698,
       "step": 782
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.8981818181818185e-05,
-      "loss": 0.992,
+      "loss": 1.0277,
       "step": 783
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9006060606060607e-05,
-      "loss": 0.9819,
+      "loss": 1.0984,
       "step": 784
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9030303030303032e-05,
-      "loss": 0.9211,
+      "loss": 0.9461,
       "step": 785
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9054545454545457e-05,
-      "loss": 0.8553,
+      "loss": 0.9641,
       "step": 786
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9078787878787883e-05,
-      "loss": 0.9893,
+      "loss": 1.0027,
       "step": 787
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9103030303030305e-05,
-      "loss": 0.9997,
+      "loss": 1.0471,
       "step": 788
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.912727272727273e-05,
-      "loss": 0.9842,
+      "loss": 1.0028,
       "step": 789
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9151515151515152e-05,
-      "loss": 0.9267,
+      "loss": 1.0045,
       "step": 790
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9175757575757577e-05,
-      "loss": 0.9419,
+      "loss": 0.9506,
       "step": 791
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9200000000000003e-05,
-      "loss": 0.8737,
+      "loss": 1.0059,
       "step": 792
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9224242424242424e-05,
-      "loss": 1.0056,
+      "loss": 1.0005,
       "step": 793
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.924848484848485e-05,
-      "loss": 0.9407,
+      "loss": 1.0828,
       "step": 794
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9272727272727275e-05,
-      "loss": 1.1013,
+      "loss": 1.1978,
       "step": 795
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9296969696969697e-05,
-      "loss": 1.1171,
+      "loss": 1.1371,
       "step": 796
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9321212121212122e-05,
-      "loss": 1.0151,
+      "loss": 1.0837,
       "step": 797
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9345454545454548e-05,
-      "loss": 0.9159,
+      "loss": 1.0543,
       "step": 798
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9369696969696973e-05,
-      "loss": 1.0078,
+      "loss": 1.0727,
       "step": 799
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9393939393939395e-05,
-      "loss": 0.742,
+      "loss": 0.9201,
       "step": 800
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.941818181818182e-05,
-      "loss": 0.9283,
+      "loss": 1.0473,
       "step": 801
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9442424242424246e-05,
-      "loss": 1.0595,
+      "loss": 1.1221,
       "step": 802
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9466666666666668e-05,
-      "loss": 1.0052,
+      "loss": 1.074,
       "step": 803
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9490909090909093e-05,
-      "loss": 0.8898,
+      "loss": 1.0324,
       "step": 804
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9515151515151515e-05,
-      "loss": 0.8003,
+      "loss": 0.8905,
       "step": 805
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.953939393939394e-05,
-      "loss": 1.074,
+      "loss": 1.0972,
       "step": 806
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9563636363636366e-05,
-      "loss": 1.0077,
+      "loss": 1.0523,
       "step": 807
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9587878787878788e-05,
-      "loss": 0.9571,
+      "loss": 1.012,
       "step": 808
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9612121212121213e-05,
-      "loss": 0.9781,
+      "loss": 1.0509,
       "step": 809
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.963636363636364e-05,
-      "loss": 0.9349,
+      "loss": 1.0717,
       "step": 810
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9660606060606064e-05,
-      "loss": 0.8989,
+      "loss": 1.0185,
       "step": 811
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9684848484848486e-05,
-      "loss": 0.9979,
+      "loss": 1.0695,
       "step": 812
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.970909090909091e-05,
-      "loss": 1.0399,
+      "loss": 1.0895,
       "step": 813
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9733333333333336e-05,
-      "loss": 0.9387,
+      "loss": 1.0595,
       "step": 814
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.975757575757576e-05,
-      "loss": 0.915,
+      "loss": 1.0033,
       "step": 815
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9781818181818184e-05,
-      "loss": 0.9926,
+      "loss": 1.0945,
       "step": 816
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.980606060606061e-05,
-      "loss": 1.0643,
+      "loss": 1.0807,
       "step": 817
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.983030303030303e-05,
-      "loss": 1.0085,
+      "loss": 1.0342,
       "step": 818
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9854545454545456e-05,
-      "loss": 0.8381,
+      "loss": 1.0051,
       "step": 819
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9878787878787878e-05,
-      "loss": 0.8631,
+      "loss": 0.9932,
       "step": 820
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9903030303030304e-05,
-      "loss": 0.7532,
+      "loss": 0.9186,
       "step": 821
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.992727272727273e-05,
-      "loss": 0.9017,
+      "loss": 1.0485,
       "step": 822
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9951515151515154e-05,
-      "loss": 1.0543,
+      "loss": 1.0554,
       "step": 823
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9975757575757576e-05,
-      "loss": 0.9777,
+      "loss": 1.062,
       "step": 824
     },
     {
       "epoch": 0.03,
       "learning_rate": 2e-05,
-      "loss": 0.9622,
+      "loss": 1.0649,
       "step": 825
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999999930632096e-05,
-      "loss": 0.9676,
+      "loss": 1.012,
       "step": 826
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999999972252838e-05,
-      "loss": 1.0205,
+      "loss": 1.0414,
       "step": 827
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999999375688855e-05,
-      "loss": 0.8802,
+      "loss": 0.9581,
       "step": 828
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999998890113527e-05,
-      "loss": 0.915,
+      "loss": 0.9701,
       "step": 829
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999998265802404e-05,
-      "loss": 0.9953,
+      "loss": 1.0462,
       "step": 830
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999997502755488e-05,
-      "loss": 0.7952,
+      "loss": 0.9345,
       "step": 831
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.99999966009728e-05,
-      "loss": 0.9802,
+      "loss": 1.0384,
       "step": 832
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999995560454347e-05,
-      "loss": 1.0666,
+      "loss": 1.0959,
       "step": 833
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999999438120014e-05,
-      "loss": 1.0338,
+      "loss": 1.0365,
       "step": 834
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999993063210202e-05,
-      "loss": 0.9806,
+      "loss": 1.1049,
       "step": 835
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999999160648455e-05,
-      "loss": 0.9449,
+      "loss": 1.0496,
       "step": 836
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.99999900110232e-05,
-      "loss": 0.9942,
+      "loss": 1.0482,
       "step": 837
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999988276826175e-05,
-      "loss": 0.9357,
+      "loss": 1.0926,
       "step": 838
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999986403893505e-05,
-      "loss": 1.0482,
+      "loss": 1.1577,
       "step": 839
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999984392225208e-05,
-      "loss": 0.8453,
+      "loss": 0.9662,
       "step": 840
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999998224182132e-05,
-      "loss": 1.2324,
+      "loss": 1.1905,
       "step": 841
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999979952681863e-05,
-      "loss": 0.9109,
+      "loss": 0.973,
       "step": 842
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999977524806874e-05,
-      "loss": 0.9202,
+      "loss": 0.9208,
       "step": 843
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999974958196383e-05,
-      "loss": 0.9294,
+      "loss": 1.0362,
       "step": 844
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999997225285043e-05,
-      "loss": 0.9181,
+      "loss": 1.0615,
       "step": 845
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999969408769047e-05,
-      "loss": 0.9508,
+      "loss": 1.0458,
       "step": 846
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999996642595228e-05,
-      "loss": 1.1392,
+      "loss": 1.1122,
       "step": 847
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999996330440017e-05,
-      "loss": 0.9822,
+      "loss": 1.0162,
       "step": 848
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999996004411275e-05,
-      "loss": 1.041,
+      "loss": 1.0835,
       "step": 849
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999956645090072e-05,
-      "loss": 0.9596,
+      "loss": 1.1135,
       "step": 850
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999995310733219e-05,
-      "loss": 0.9312,
+      "loss": 0.991,
       "step": 851
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999949430839143e-05,
-      "loss": 0.9783,
+      "loss": 1.1287,
       "step": 852
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999945615610987e-05,
-      "loss": 1.0068,
+      "loss": 1.0638,
       "step": 853
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999941661647773e-05,
-      "loss": 0.9218,
+      "loss": 0.9828,
       "step": 854
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999937568949557e-05,
-      "loss": 0.9971,
+      "loss": 1.0204,
       "step": 855
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999933337516393e-05,
-      "loss": 0.9344,
+      "loss": 1.0346,
       "step": 856
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999928967348344e-05,
-      "loss": 1.0381,
+      "loss": 1.0961,
       "step": 857
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999924458445467e-05,
-      "loss": 1.1544,
+      "loss": 1.1828,
       "step": 858
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999919810807828e-05,
-      "loss": 1.1791,
+      "loss": 1.1873,
       "step": 859
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999991502443549e-05,
-      "loss": 0.9453,
+      "loss": 1.0325,
       "step": 860
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999910099328517e-05,
-      "loss": 0.8801,
+      "loss": 1.0562,
       "step": 861
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999905035486982e-05,
-      "loss": 1.0616,
+      "loss": 1.1251,
       "step": 862
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999899832910952e-05,
-      "loss": 1.1443,
+      "loss": 1.096,
       "step": 863
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.99998944916005e-05,
-      "loss": 0.9171,
+      "loss": 1.0485,
       "step": 864
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999889011555696e-05,
-      "loss": 1.0229,
+      "loss": 1.1384,
       "step": 865
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999883392776623e-05,
-      "loss": 0.9356,
+      "loss": 1.0328,
       "step": 866
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999877635263355e-05,
-      "loss": 1.0284,
+      "loss": 1.0886,
       "step": 867
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999871739015974e-05,
-      "loss": 0.8089,
+      "loss": 0.9491,
       "step": 868
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999986570403456e-05,
-      "loss": 0.8608,
+      "loss": 0.9978,
       "step": 869
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.99998595303192e-05,
-      "loss": 1.0421,
+      "loss": 1.1514,
       "step": 870
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999853217869972e-05,
-      "loss": 0.9909,
+      "loss": 0.9667,
       "step": 871
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999846766686973e-05,
-      "loss": 0.7941,
+      "loss": 0.9787,
       "step": 872
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999840176770285e-05,
-      "loss": 0.9116,
+      "loss": 1.0944,
       "step": 873
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999833448120004e-05,
-      "loss": 1.0099,
+      "loss": 1.1,
       "step": 874
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999982658073622e-05,
-      "loss": 0.9473,
+      "loss": 1.0984,
       "step": 875
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999819574619033e-05,
-      "loss": 1.0096,
+      "loss": 1.1213,
       "step": 876
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999812429768536e-05,
-      "loss": 1.0288,
+      "loss": 1.0931,
       "step": 877
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999980514618483e-05,
-      "loss": 0.9459,
+      "loss": 1.0013,
       "step": 878
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999797723868014e-05,
-      "loss": 0.901,
+      "loss": 0.9597,
       "step": 879
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999790162818195e-05,
-      "loss": 0.8879,
+      "loss": 0.9816,
       "step": 880
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999782463035472e-05,
-      "loss": 0.9424,
+      "loss": 1.1107,
       "step": 881
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999774624519953e-05,
-      "loss": 0.9727,
+      "loss": 1.0756,
       "step": 882
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999766647271755e-05,
-      "loss": 1.0693,
+      "loss": 1.1081,
       "step": 883
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999758531290977e-05,
-      "loss": 0.927,
+      "loss": 0.9592,
       "step": 884
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999750276577743e-05,
-      "loss": 1.14,
+      "loss": 1.109,
       "step": 885
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999741883132156e-05,
-      "loss": 1.0009,
+      "loss": 1.0573,
       "step": 886
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999733350954344e-05,
-      "loss": 0.9629,
+      "loss": 1.0346,
       "step": 887
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999724680044417e-05,
-      "loss": 0.9664,
+      "loss": 1.0876,
       "step": 888
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999715870402497e-05,
-      "loss": 1.0059,
+      "loss": 1.0698,
       "step": 889
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999706922028706e-05,
-      "loss": 0.9697,
+      "loss": 1.0679,
       "step": 890
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999969783492317e-05,
-      "loss": 1.0623,
+      "loss": 1.0776,
       "step": 891
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999688609086015e-05,
-      "loss": 1.0733,
+      "loss": 1.1891,
       "step": 892
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999967924451737e-05,
-      "loss": 0.9986,
+      "loss": 1.0988,
       "step": 893
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999966974121736e-05,
-      "loss": 0.9993,
+      "loss": 1.0958,
       "step": 894
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999660099186123e-05,
-      "loss": 0.9581,
+      "loss": 1.0558,
       "step": 895
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999650318423786e-05,
-      "loss": 0.8575,
+      "loss": 0.9477,
       "step": 896
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999640398930492e-05,
-      "loss": 1.0208,
+      "loss": 1.1362,
       "step": 897
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999963034070637e-05,
-      "loss": 0.9358,
+      "loss": 1.0675,
       "step": 898
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999620143751573e-05,
-      "loss": 1.0633,
+      "loss": 1.1277,
       "step": 899
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999960980806623e-05,
-      "loss": 0.9922,
+      "loss": 1.0934,
       "step": 900
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999599333650487e-05,
-      "loss": 0.952,
+      "loss": 1.0135,
       "step": 901
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999958872050449e-05,
-      "loss": 0.9233,
+      "loss": 1.0327,
       "step": 902
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999957796862839e-05,
-      "loss": 0.9748,
+      "loss": 0.9923,
       "step": 903
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999956707802233e-05,
-      "loss": 0.8926,
+      "loss": 1.0159,
       "step": 904
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999556048686468e-05,
-      "loss": 1.0424,
+      "loss": 1.0997,
       "step": 905
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999544880620952e-05,
-      "loss": 1.0672,
+      "loss": 1.1556,
       "step": 906
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999533573825937e-05,
-      "loss": 0.8997,
+      "loss": 1.0623,
       "step": 907
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999522128301586e-05,
-      "loss": 0.9984,
+      "loss": 1.0576,
       "step": 908
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999510544048047e-05,
-      "loss": 0.9921,
+      "loss": 1.0421,
       "step": 909
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999498821065487e-05,
-      "loss": 0.9584,
+      "loss": 1.0157,
       "step": 910
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999486959354072e-05,
-      "loss": 1.0246,
+      "loss": 1.0648,
       "step": 911
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999947495891396e-05,
-      "loss": 0.9361,
+      "loss": 1.0687,
       "step": 912
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999946281974532e-05,
-      "loss": 1.0224,
+      "loss": 1.0746,
       "step": 913
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999450541848322e-05,
-      "loss": 0.8394,
+      "loss": 0.8962,
       "step": 914
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999438125223136e-05,
-      "loss": 0.8324,
+      "loss": 1.0316,
       "step": 915
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999425569869932e-05,
-      "loss": 0.9505,
+      "loss": 1.0222,
       "step": 916
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999412875788886e-05,
-      "loss": 0.8156,
+      "loss": 0.9812,
       "step": 917
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999400042980173e-05,
-      "loss": 0.973,
+      "loss": 1.0561,
       "step": 918
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999387071443975e-05,
-      "loss": 0.9618,
+      "loss": 1.039,
       "step": 919
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999373961180466e-05,
-      "loss": 0.809,
+      "loss": 0.9503,
       "step": 920
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999936071218983e-05,
-      "loss": 1.118,
+      "loss": 1.1033,
       "step": 921
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999347324472252e-05,
-      "loss": 0.9678,
+      "loss": 1.0434,
       "step": 922
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999333798027914e-05,
-      "loss": 0.8091,
+      "loss": 0.9935,
       "step": 923
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999932013285701e-05,
-      "loss": 1.0357,
+      "loss": 1.1183,
       "step": 924
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999306328959726e-05,
-      "loss": 0.9666,
+      "loss": 1.0456,
       "step": 925
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999929238633626e-05,
-      "loss": 0.8594,
+      "loss": 1.0265,
       "step": 926
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999927830498679e-05,
-      "loss": 0.9312,
+      "loss": 1.0146,
       "step": 927
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999264084911524e-05,
-      "loss": 1.0502,
+      "loss": 1.1596,
       "step": 928
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999249726110653e-05,
-      "loss": 0.8513,
+      "loss": 0.9837,
       "step": 929
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999235228584383e-05,
-      "loss": 0.9655,
+      "loss": 1.0465,
       "step": 930
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999922059233291e-05,
-      "loss": 0.9836,
+      "loss": 1.0653,
       "step": 931
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999920581735644e-05,
-      "loss": 0.9571,
+      "loss": 1.1025,
       "step": 932
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999190903655175e-05,
-      "loss": 0.9756,
+      "loss": 1.0599,
       "step": 933
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999175851229323e-05,
-      "loss": 0.9756,
+      "loss": 1.0856,
       "step": 934
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999916066007909e-05,
-      "loss": 0.8692,
+      "loss": 0.9945,
       "step": 935
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999145330204694e-05,
-      "loss": 0.9247,
+      "loss": 1.036,
       "step": 936
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999129861606345e-05,
-      "loss": 0.9391,
+      "loss": 1.0666,
       "step": 937
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999114254284252e-05,
-      "loss": 0.9334,
+      "loss": 1.0016,
       "step": 938
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999098508238638e-05,
-      "loss": 1.0118,
+      "loss": 1.0963,
       "step": 939
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999082623469718e-05,
-      "loss": 1.0253,
+      "loss": 1.1152,
       "step": 940
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999066599977712e-05,
-      "loss": 1.0746,
+      "loss": 1.1207,
       "step": 941
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999050437762846e-05,
-      "loss": 1.0069,
+      "loss": 1.0797,
       "step": 942
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999903413682534e-05,
-      "loss": 0.988,
+      "loss": 1.0564,
       "step": 943
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999017697165425e-05,
-      "loss": 0.9673,
+      "loss": 1.0106,
       "step": 944
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9999001118783322e-05,
-      "loss": 0.8248,
+      "loss": 0.9075,
       "step": 945
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999898440167927e-05,
-      "loss": 0.9822,
+      "loss": 1.0378,
       "step": 946
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998967545853492e-05,
-      "loss": 0.8274,
+      "loss": 0.966,
       "step": 947
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999895055130623e-05,
-      "loss": 0.8457,
+      "loss": 1.0612,
       "step": 948
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998933418037715e-05,
-      "loss": 1.0398,
+      "loss": 1.063,
       "step": 949
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998916146048182e-05,
-      "loss": 0.8676,
+      "loss": 1.0452,
       "step": 950
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998898735337877e-05,
-      "loss": 0.9698,
+      "loss": 1.0094,
       "step": 951
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998881185907036e-05,
-      "loss": 0.9871,
+      "loss": 1.0101,
       "step": 952
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998863497755908e-05,
-      "loss": 0.906,
+      "loss": 0.9679,
       "step": 953
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998845670884733e-05,
-      "loss": 0.9214,
+      "loss": 0.9639,
       "step": 954
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998827705293764e-05,
-      "loss": 0.959,
+      "loss": 1.0743,
       "step": 955
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998809600983246e-05,
-      "loss": 0.9954,
+      "loss": 1.1143,
       "step": 956
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998791357953428e-05,
-      "loss": 1.0962,
+      "loss": 1.1478,
       "step": 957
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999877297620457e-05,
-      "loss": 0.9453,
+      "loss": 0.9899,
       "step": 958
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998754455736922e-05,
-      "loss": 0.9565,
+      "loss": 1.0714,
       "step": 959
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9998735796550744e-05,
-      "loss": 0.9368,
+      "loss": 1.0397,
       "step": 960
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999871699864629e-05,
-      "loss": 0.8917,
+      "loss": 1.0564,
       "step": 961
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.999869806202383e-05,
-      "loss": 1.0145,
+      "loss": 1.0438,
       "step": 962
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998678986683613e-05,
-      "loss": 0.9181,
+      "loss": 1.0729,
       "step": 963
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998659772625914e-05,
-      "loss": 0.7929,
+      "loss": 0.941,
       "step": 964
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998640419850998e-05,
-      "loss": 0.8815,
+      "loss": 1.0805,
       "step": 965
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998620928359132e-05,
-      "loss": 0.8866,
+      "loss": 1.039,
       "step": 966
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998601298150588e-05,
-      "loss": 0.9435,
+      "loss": 1.0526,
       "step": 967
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998581529225633e-05,
-      "loss": 0.951,
+      "loss": 1.1083,
       "step": 968
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999856162158455e-05,
-      "loss": 1.0041,
+      "loss": 1.0617,
       "step": 969
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998541575227607e-05,
-      "loss": 0.9012,
+      "loss": 1.0597,
       "step": 970
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998521390155087e-05,
-      "loss": 1.0289,
+      "loss": 1.1061,
       "step": 971
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998501066367272e-05,
-      "loss": 0.8828,
+      "loss": 0.9415,
       "step": 972
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998480603864435e-05,
-      "loss": 1.0721,
+      "loss": 1.1678,
       "step": 973
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998460002646872e-05,
-      "loss": 0.8336,
+      "loss": 0.9424,
       "step": 974
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998439262714856e-05,
-      "loss": 1.0695,
+      "loss": 1.0977,
       "step": 975
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998418384068686e-05,
-      "loss": 0.9778,
+      "loss": 1.0205,
       "step": 976
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998397366708647e-05,
-      "loss": 1.0055,
+      "loss": 1.075,
       "step": 977
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999837621063503e-05,
-      "loss": 0.9616,
+      "loss": 1.022,
       "step": 978
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998354915848125e-05,
-      "loss": 0.9417,
+      "loss": 1.0784,
       "step": 979
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998333482348233e-05,
-      "loss": 0.9856,
+      "loss": 1.0222,
       "step": 980
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999831191013565e-05,
-      "loss": 0.7556,
+      "loss": 0.9232,
       "step": 981
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998290199210678e-05,
-      "loss": 0.9863,
+      "loss": 1.0801,
       "step": 982
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999826834957361e-05,
-      "loss": 0.9425,
+      "loss": 1.0615,
       "step": 983
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999824636122476e-05,
-      "loss": 0.9165,
+      "loss": 0.9979,
       "step": 984
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998224234164423e-05,
-      "loss": 0.7871,
+      "loss": 0.9771,
       "step": 985
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999820196839291e-05,
-      "loss": 0.972,
+      "loss": 1.024,
       "step": 986
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998179563910535e-05,
-      "loss": 0.8509,
+      "loss": 1.063,
       "step": 987
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99981570207176e-05,
-      "loss": 0.9677,
+      "loss": 1.0656,
       "step": 988
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998134338814423e-05,
-      "loss": 0.9403,
+      "loss": 1.0685,
       "step": 989
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998111518201317e-05,
-      "loss": 1.0009,
+      "loss": 0.9991,
       "step": 990
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99980885588786e-05,
-      "loss": 0.976,
+      "loss": 1.0988,
       "step": 991
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999806546084659e-05,
-      "loss": 0.9925,
+      "loss": 1.0255,
       "step": 992
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998042224105608e-05,
-      "loss": 0.9921,
+      "loss": 1.0571,
       "step": 993
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9998018848655974e-05,
-      "loss": 0.8564,
+      "loss": 1.0265,
       "step": 994
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997995334498014e-05,
-      "loss": 1.0476,
+      "loss": 1.1512,
       "step": 995
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997971681632052e-05,
-      "loss": 0.9513,
+      "loss": 0.9966,
       "step": 996
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997947890058422e-05,
-      "loss": 0.9298,
+      "loss": 1.0106,
       "step": 997
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997923959777447e-05,
-      "loss": 0.9713,
+      "loss": 1.0531,
       "step": 998
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997899890789464e-05,
-      "loss": 0.909,
+      "loss": 1.0645,
       "step": 999
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997875683094805e-05,
-      "loss": 1.0227,
+      "loss": 1.112,
       "step": 1000
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997851336693804e-05,
-      "loss": 1.0485,
+      "loss": 1.0537,
       "step": 1001
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997826851586804e-05,
-      "loss": 0.8982,
+      "loss": 0.9653,
       "step": 1002
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997802227774138e-05,
-      "loss": 0.999,
+      "loss": 1.1215,
       "step": 1003
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997777465256157e-05,
-      "loss": 0.9431,
+      "loss": 1.053,
       "step": 1004
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997752564033194e-05,
-      "loss": 0.8805,
+      "loss": 1.0127,
       "step": 1005
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997727524105597e-05,
-      "loss": 1.175,
+      "loss": 1.2076,
       "step": 1006
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997702345473723e-05,
-      "loss": 1.0586,
+      "loss": 1.0596,
       "step": 1007
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997677028137906e-05,
-      "loss": 1.0898,
+      "loss": 1.2019,
       "step": 1008
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999765157209851e-05,
-      "loss": 0.9724,
+      "loss": 1.0844,
       "step": 1009
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997625977355883e-05,
-      "loss": 0.9076,
+      "loss": 0.9871,
       "step": 1010
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997600243910377e-05,
-      "loss": 1.0394,
+      "loss": 1.098,
       "step": 1011
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997574371762356e-05,
-      "loss": 1.0334,
+      "loss": 1.103,
       "step": 1012
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997548360912174e-05,
-      "loss": 0.9345,
+      "loss": 1.0207,
       "step": 1013
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997522211360192e-05,
-      "loss": 0.944,
+      "loss": 1.0291,
       "step": 1014
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997495923106775e-05,
-      "loss": 0.9782,
+      "loss": 1.0294,
       "step": 1015
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997469496152288e-05,
-      "loss": 0.9444,
+      "loss": 1.0005,
       "step": 1016
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999744293049709e-05,
-      "loss": 0.9296,
+      "loss": 0.8979,
       "step": 1017
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997416226141564e-05,
-      "loss": 0.9676,
+      "loss": 1.0916,
       "step": 1018
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999738938308607e-05,
-      "loss": 0.9731,
+      "loss": 1.0516,
       "step": 1019
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999736240133098e-05,
-      "loss": 1.0269,
+      "loss": 1.0723,
       "step": 1020
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997335280876676e-05,
-      "loss": 0.9241,
+      "loss": 1.0237,
       "step": 1021
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997308021723523e-05,
-      "loss": 0.9032,
+      "loss": 0.9469,
       "step": 1022
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999728062387191e-05,
-      "loss": 0.9609,
+      "loss": 1.0571,
       "step": 1023
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999725308732221e-05,
-      "loss": 1.0004,
+      "loss": 1.0526,
       "step": 1024
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999722541207481e-05,
-      "loss": 1.1022,
+      "loss": 1.1924,
       "step": 1025
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999719759813009e-05,
-      "loss": 0.8914,
+      "loss": 1.0268,
       "step": 1026
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997169645488438e-05,
-      "loss": 0.7836,
+      "loss": 0.9362,
       "step": 1027
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997141554150243e-05,
-      "loss": 0.9624,
+      "loss": 0.9999,
       "step": 1028
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999711332411589e-05,
-      "loss": 0.9018,
+      "loss": 1.0106,
       "step": 1029
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997084955385776e-05,
-      "loss": 1.0101,
+      "loss": 1.1172,
       "step": 1030
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999705644796029e-05,
-      "loss": 1.0213,
+      "loss": 1.0826,
       "step": 1031
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9997027801839832e-05,
-      "loss": 1.0202,
+      "loss": 1.1577,
       "step": 1032
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996999017024795e-05,
-      "loss": 0.9624,
+      "loss": 1.1097,
       "step": 1033
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999697009351558e-05,
-      "loss": 0.9811,
+      "loss": 1.1038,
       "step": 1034
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996941031312593e-05,
-      "loss": 0.8844,
+      "loss": 1.0362,
       "step": 1035
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996911830416233e-05,
-      "loss": 0.9324,
+      "loss": 1.0055,
       "step": 1036
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99968824908269e-05,
-      "loss": 0.9839,
+      "loss": 1.076,
       "step": 1037
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999685301254501e-05,
-      "loss": 0.9477,
+      "loss": 1.0612,
       "step": 1038
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996823395570967e-05,
-      "loss": 1.0973,
+      "loss": 1.0901,
       "step": 1039
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996793639905184e-05,
-      "loss": 0.9461,
+      "loss": 1.0565,
       "step": 1040
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999676374554807e-05,
-      "loss": 0.9395,
+      "loss": 1.0319,
       "step": 1041
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996733712500044e-05,
-      "loss": 0.9147,
+      "loss": 1.0013,
       "step": 1042
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996703540761523e-05,
-      "loss": 0.8115,
+      "loss": 1.0322,
       "step": 1043
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996673230332925e-05,
-      "loss": 0.959,
+      "loss": 1.1118,
       "step": 1044
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996642781214665e-05,
-      "loss": 1.0306,
+      "loss": 1.1011,
       "step": 1045
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999661219340717e-05,
-      "loss": 0.7998,
+      "loss": 0.9539,
       "step": 1046
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999658146691087e-05,
-      "loss": 0.8795,
+      "loss": 0.9445,
       "step": 1047
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999655060172618e-05,
-      "loss": 0.8928,
+      "loss": 1.0172,
       "step": 1048
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996519597853533e-05,
-      "loss": 0.8999,
+      "loss": 1.0508,
       "step": 1049
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999648845529336e-05,
-      "loss": 1.1316,
+      "loss": 1.2047,
       "step": 1050
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996457174046097e-05,
-      "loss": 0.9556,
+      "loss": 1.0738,
       "step": 1051
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999642575411217e-05,
-      "loss": 0.9754,
+      "loss": 1.0514,
       "step": 1052
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999639419549202e-05,
-      "loss": 0.8634,
+      "loss": 1.0026,
       "step": 1053
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996362498186084e-05,
-      "loss": 0.9785,
+      "loss": 1.0146,
       "step": 1054
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996330662194802e-05,
-      "loss": 0.9736,
+      "loss": 1.0496,
       "step": 1055
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996298687518613e-05,
-      "loss": 0.8524,
+      "loss": 0.9535,
       "step": 1056
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996266574157964e-05,
-      "loss": 0.8932,
+      "loss": 0.9597,
       "step": 1057
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99962343221133e-05,
-      "loss": 1.0308,
+      "loss": 1.0809,
       "step": 1058
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996201931385065e-05,
-      "loss": 1.0516,
+      "loss": 1.1536,
       "step": 1059
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996169401973714e-05,
-      "loss": 0.9246,
+      "loss": 1.0125,
       "step": 1060
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999613673387969e-05,
-      "loss": 0.7956,
+      "loss": 1.0043,
       "step": 1061
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996103927103458e-05,
-      "loss": 1.0824,
+      "loss": 1.1075,
       "step": 1062
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996070981645464e-05,
-      "loss": 1.0336,
+      "loss": 1.0385,
       "step": 1063
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996037897506168e-05,
-      "loss": 0.917,
+      "loss": 0.9946,
       "step": 1064
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9996004674686026e-05,
-      "loss": 0.9673,
+      "loss": 1.107,
       "step": 1065
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995971313185506e-05,
-      "loss": 0.9204,
+      "loss": 1.0743,
       "step": 1066
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995937813005063e-05,
-      "loss": 1.0095,
+      "loss": 1.074,
       "step": 1067
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995904174145163e-05,
-      "loss": 0.9922,
+      "loss": 1.0525,
       "step": 1068
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999587039660628e-05,
-      "loss": 0.9393,
+      "loss": 1.0352,
       "step": 1069
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995836480388876e-05,
-      "loss": 0.9225,
+      "loss": 1.0706,
       "step": 1070
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995802425493426e-05,
-      "loss": 0.8724,
+      "loss": 0.9941,
       "step": 1071
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995768231920394e-05,
-      "loss": 0.8734,
+      "loss": 1.048,
       "step": 1072
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999573389967026e-05,
-      "loss": 0.8717,
+      "loss": 1.0027,
       "step": 1073
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995699428743503e-05,
-      "loss": 0.7822,
+      "loss": 0.8928,
       "step": 1074
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99956648191406e-05,
-      "loss": 0.8545,
+      "loss": 0.9649,
       "step": 1075
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995630070862025e-05,
-      "loss": 0.9392,
+      "loss": 1.0969,
       "step": 1076
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999559518390827e-05,
-      "loss": 0.9656,
+      "loss": 1.0435,
       "step": 1077
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995560158279808e-05,
-      "loss": 1.0243,
+      "loss": 1.0434,
       "step": 1078
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995524993977135e-05,
-      "loss": 0.9572,
+      "loss": 1.0156,
       "step": 1079
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995489691000735e-05,
-      "loss": 0.8372,
+      "loss": 1.01,
       "step": 1080
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995454249351093e-05,
-      "loss": 0.945,
+      "loss": 1.0018,
       "step": 1081
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999541866902871e-05,
-      "loss": 0.8972,
+      "loss": 1.0224,
       "step": 1082
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995382950034074e-05,
-      "loss": 1.0126,
+      "loss": 1.0666,
       "step": 1083
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999534709236768e-05,
-      "loss": 1.0352,
+      "loss": 1.1182,
       "step": 1084
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995311096030024e-05,
-      "loss": 1.0117,
+      "loss": 1.011,
       "step": 1085
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999527496102161e-05,
-      "loss": 0.9902,
+      "loss": 1.0951,
       "step": 1086
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995238687342936e-05,
-      "loss": 1.0219,
+      "loss": 1.07,
       "step": 1087
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995202274994508e-05,
-      "loss": 0.9218,
+      "loss": 1.0103,
       "step": 1088
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999516572397683e-05,
-      "loss": 0.9604,
+      "loss": 1.0488,
       "step": 1089
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995129034290408e-05,
-      "loss": 0.8366,
+      "loss": 0.9601,
       "step": 1090
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995092205935752e-05,
-      "loss": 0.9876,
+      "loss": 1.0225,
       "step": 1091
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995055238913376e-05,
-      "loss": 0.8273,
+      "loss": 0.9632,
       "step": 1092
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9995018133223785e-05,
-      "loss": 0.8907,
+      "loss": 1.0771,
       "step": 1093
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99949808888675e-05,
-      "loss": 0.9309,
+      "loss": 1.0283,
       "step": 1094
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999494350584504e-05,
-      "loss": 0.9174,
+      "loss": 1.0096,
       "step": 1095
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994905984156913e-05,
-      "loss": 0.8348,
+      "loss": 0.9156,
       "step": 1096
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999486832380365e-05,
-      "loss": 1.0034,
+      "loss": 1.0181,
       "step": 1097
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999483052478577e-05,
-      "loss": 0.9269,
+      "loss": 1.0521,
       "step": 1098
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994792587103796e-05,
-      "loss": 0.9314,
+      "loss": 1.0101,
       "step": 1099
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994754510758256e-05,
-      "loss": 0.8091,
+      "loss": 0.9795,
       "step": 1100
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994716295749682e-05,
-      "loss": 0.9975,
+      "loss": 1.0631,
       "step": 1101
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994677942078593e-05,
-      "loss": 0.8898,
+      "loss": 1.0242,
       "step": 1102
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994639449745533e-05,
-      "loss": 1.0843,
+      "loss": 1.0844,
       "step": 1103
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999460081875103e-05,
-      "loss": 1.0072,
+      "loss": 1.1075,
       "step": 1104
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999456204909562e-05,
-      "loss": 0.879,
+      "loss": 0.9558,
       "step": 1105
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994523140779843e-05,
-      "loss": 1.0983,
+      "loss": 1.1611,
       "step": 1106
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999448409380424e-05,
-      "loss": 0.9633,
+      "loss": 1.0978,
       "step": 1107
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999444490816935e-05,
-      "loss": 0.8145,
+      "loss": 0.9698,
       "step": 1108
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994405583875714e-05,
-      "loss": 0.8648,
+      "loss": 0.9309,
       "step": 1109
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999436612092388e-05,
-      "loss": 1.0424,
+      "loss": 1.1146,
       "step": 1110
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994326519314396e-05,
-      "loss": 0.9277,
+      "loss": 0.9963,
       "step": 1111
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994286779047817e-05,
-      "loss": 1.0077,
+      "loss": 1.0792,
       "step": 1112
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994246900124686e-05,
-      "loss": 0.9036,
+      "loss": 1.0207,
       "step": 1113
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994206882545558e-05,
-      "loss": 0.7493,
+      "loss": 1.0077,
       "step": 1114
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994166726310992e-05,
-      "loss": 0.8912,
+      "loss": 1.0032,
       "step": 1115
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999412643142154e-05,
-      "loss": 1.0003,
+      "loss": 1.0612,
       "step": 1116
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994085997877766e-05,
-      "loss": 0.9987,
+      "loss": 1.0757,
       "step": 1117
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994045425680228e-05,
-      "loss": 1.0658,
+      "loss": 1.0721,
       "step": 1118
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9994004714829488e-05,
-      "loss": 1.0521,
+      "loss": 1.0951,
       "step": 1119
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993963865326115e-05,
-      "loss": 0.8618,
+      "loss": 1.0229,
       "step": 1120
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999392287717067e-05,
-      "loss": 0.9617,
+      "loss": 1.059,
       "step": 1121
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993881750363723e-05,
-      "loss": 0.8473,
+      "loss": 1.046,
       "step": 1122
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999384048490585e-05,
-      "loss": 0.951,
+      "loss": 1.0005,
       "step": 1123
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993799080797618e-05,
-      "loss": 0.8793,
+      "loss": 0.9284,
       "step": 1124
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993757538039605e-05,
-      "loss": 0.881,
+      "loss": 0.9761,
       "step": 1125
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999371585663238e-05,
-      "loss": 0.899,
+      "loss": 0.9605,
       "step": 1126
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993674036576533e-05,
-      "loss": 0.8404,
+      "loss": 0.9486,
       "step": 1127
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993632077872637e-05,
-      "loss": 0.9107,
+      "loss": 1.0385,
       "step": 1128
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999358998052127e-05,
-      "loss": 0.839,
+      "loss": 0.9894,
       "step": 1129
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993547744523027e-05,
-      "loss": 0.975,
+      "loss": 1.0091,
       "step": 1130
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993505369878486e-05,
-      "loss": 0.8952,
+      "loss": 1.0069,
       "step": 1131
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993462856588235e-05,
-      "loss": 0.907,
+      "loss": 1.0494,
       "step": 1132
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999342020465287e-05,
-      "loss": 0.8542,
+      "loss": 1.0314,
       "step": 1133
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993377414072976e-05,
-      "loss": 0.7593,
+      "loss": 0.9508,
       "step": 1134
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993334484849147e-05,
-      "loss": 0.9605,
+      "loss": 1.0398,
       "step": 1135
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993291416981983e-05,
-      "loss": 0.9634,
+      "loss": 1.1385,
       "step": 1136
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999324821047208e-05,
-      "loss": 0.9772,
+      "loss": 1.0934,
       "step": 1137
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9993204865320034e-05,
-      "loss": 0.9281,
+      "loss": 1.0168,
       "step": 1138
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999316138152645e-05,
-      "loss": 0.7891,
+      "loss": 0.934,
       "step": 1139
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999311775909193e-05,
-      "loss": 0.9601,
+      "loss": 1.0298,
       "step": 1140
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999307399801708e-05,
-      "loss": 0.8159,
+      "loss": 0.942,
       "step": 1141
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999303009830251e-05,
-      "loss": 0.9125,
+      "loss": 0.9863,
       "step": 1142
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999298605994882e-05,
-      "loss": 0.9955,
+      "loss": 1.0746,
       "step": 1143
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999294188295663e-05,
-      "loss": 0.8765,
+      "loss": 0.9792,
       "step": 1144
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999289756732655e-05,
-      "loss": 0.9415,
+      "loss": 1.0419,
       "step": 1145
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999285311305919e-05,
-      "loss": 0.9539,
+      "loss": 1.0224,
       "step": 1146
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992808520155175e-05,
-      "loss": 0.9918,
+      "loss": 1.0773,
       "step": 1147
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992763788615116e-05,
-      "loss": 1.0703,
+      "loss": 1.0913,
       "step": 1148
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992718918439643e-05,
-      "loss": 0.8758,
+      "loss": 1.0315,
       "step": 1149
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999267390962937e-05,
-      "loss": 0.9333,
+      "loss": 0.93,
       "step": 1150
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992628762184924e-05,
-      "loss": 0.9143,
+      "loss": 1.0047,
       "step": 1151
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999258347610693e-05,
-      "loss": 1.0954,
+      "loss": 1.1421,
       "step": 1152
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999253805139602e-05,
-      "loss": 0.8985,
+      "loss": 0.9856,
       "step": 1153
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992492488052822e-05,
-      "loss": 0.8547,
+      "loss": 0.9386,
       "step": 1154
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992446786077967e-05,
-      "loss": 0.9085,
+      "loss": 0.9706,
       "step": 1155
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992400945472093e-05,
-      "loss": 1.0199,
+      "loss": 1.1268,
       "step": 1156
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999235496623583e-05,
-      "loss": 1.0663,
+      "loss": 1.0975,
       "step": 1157
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992308848369823e-05,
-      "loss": 1.0231,
+      "loss": 1.0378,
       "step": 1158
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992262591874705e-05,
-      "loss": 1.0643,
+      "loss": 1.1147,
       "step": 1159
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999221619675112e-05,
-      "loss": 0.892,
+      "loss": 0.9852,
       "step": 1160
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992169662999715e-05,
-      "loss": 0.9158,
+      "loss": 1.0564,
       "step": 1161
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999212299062113e-05,
-      "loss": 0.978,
+      "loss": 1.0576,
       "step": 1162
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999207617961602e-05,
-      "loss": 0.8818,
+      "loss": 1.0107,
       "step": 1163
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9992029229985024e-05,
-      "loss": 0.9467,
+      "loss": 1.0759,
       "step": 1164
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991982141728804e-05,
-      "loss": 0.9615,
+      "loss": 0.9954,
       "step": 1165
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991934914848005e-05,
-      "loss": 0.8716,
+      "loss": 1.0475,
       "step": 1166
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991887549343288e-05,
-      "loss": 1.004,
+      "loss": 1.0986,
       "step": 1167
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991840045215307e-05,
-      "loss": 0.8839,
+      "loss": 1.0537,
       "step": 1168
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999179240246472e-05,
-      "loss": 1.0065,
+      "loss": 1.0173,
       "step": 1169
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991744621092195e-05,
-      "loss": 0.9249,
+      "loss": 1.0012,
       "step": 1170
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991696701098386e-05,
-      "loss": 1.0292,
+      "loss": 1.0489,
       "step": 1171
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991648642483962e-05,
-      "loss": 0.7466,
+      "loss": 0.9428,
       "step": 1172
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999160044524959e-05,
-      "loss": 0.8455,
+      "loss": 1.0037,
       "step": 1173
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999155210939594e-05,
-      "loss": 1.0439,
+      "loss": 1.0408,
       "step": 1174
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991503634923678e-05,
-      "loss": 1.0721,
+      "loss": 1.1187,
       "step": 1175
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999145502183348e-05,
-      "loss": 0.9109,
+      "loss": 1.0264,
       "step": 1176
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999140627012602e-05,
-      "loss": 0.9613,
+      "loss": 1.015,
       "step": 1177
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991357379801973e-05,
-      "loss": 0.9979,
+      "loss": 1.1089,
       "step": 1178
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991308350862017e-05,
-      "loss": 0.9582,
+      "loss": 1.0519,
       "step": 1179
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991259183306834e-05,
-      "loss": 0.8483,
+      "loss": 1.028,
       "step": 1180
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999120987713711e-05,
-      "loss": 1.014,
+      "loss": 1.0757,
       "step": 1181
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999116043235352e-05,
-      "loss": 0.9535,
+      "loss": 1.0916,
       "step": 1182
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991110848956755e-05,
-      "loss": 0.8672,
+      "loss": 0.983,
       "step": 1183
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991061126947503e-05,
-      "loss": 0.9613,
+      "loss": 1.0947,
       "step": 1184
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9991011266326453e-05,
-      "loss": 0.8799,
+      "loss": 0.974,
       "step": 1185
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99909612670943e-05,
-      "loss": 0.8275,
+      "loss": 1.0177,
       "step": 1186
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990911129251734e-05,
-      "loss": 0.9881,
+      "loss": 1.1572,
       "step": 1187
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999086085279945e-05,
-      "loss": 0.9826,
+      "loss": 1.0836,
       "step": 1188
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990810437738142e-05,
-      "loss": 0.9249,
+      "loss": 1.0671,
       "step": 1189
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990759884068524e-05,
-      "loss": 1.0157,
+      "loss": 1.1241,
       "step": 1190
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999070919179128e-05,
-      "loss": 0.9548,
+      "loss": 1.096,
       "step": 1191
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990658360907125e-05,
-      "loss": 0.8859,
+      "loss": 1.0037,
       "step": 1192
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990607391416758e-05,
-      "loss": 0.8619,
+      "loss": 0.9514,
       "step": 1193
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999055628332089e-05,
-      "loss": 0.8785,
+      "loss": 0.9876,
       "step": 1194
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990505036620228e-05,
-      "loss": 0.96,
+      "loss": 1.0084,
       "step": 1195
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999045365131548e-05,
-      "loss": 0.9058,
+      "loss": 1.0341,
       "step": 1196
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990402127407365e-05,
-      "loss": 0.956,
+      "loss": 1.0452,
       "step": 1197
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990350464896595e-05,
-      "loss": 1.0069,
+      "loss": 1.114,
       "step": 1198
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990298663783886e-05,
-      "loss": 0.8825,
+      "loss": 1.045,
       "step": 1199
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999024672406996e-05,
-      "loss": 0.92,
+      "loss": 1.0479,
       "step": 1200
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.999019464575553e-05,
-      "loss": 0.9206,
+      "loss": 1.0586,
       "step": 1201
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990142428841328e-05,
-      "loss": 1.0055,
+      "loss": 1.0509,
       "step": 1202
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990090073328075e-05,
-      "loss": 0.8715,
+      "loss": 0.9772,
       "step": 1203
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9990037579216492e-05,
-      "loss": 0.8706,
+      "loss": 1.071,
       "step": 1204
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989984946507314e-05,
-      "loss": 0.9583,
+      "loss": 1.0972,
       "step": 1205
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989932175201268e-05,
-      "loss": 1.1109,
+      "loss": 1.2057,
       "step": 1206
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989879265299087e-05,
-      "loss": 0.8338,
+      "loss": 1.0275,
       "step": 1207
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989826216801506e-05,
-      "loss": 0.8692,
+      "loss": 1.0394,
       "step": 1208
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998977302970926e-05,
-      "loss": 0.8696,
+      "loss": 0.9678,
       "step": 1209
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989719704023084e-05,
-      "loss": 0.9951,
+      "loss": 1.0165,
       "step": 1210
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998966623974372e-05,
-      "loss": 0.8997,
+      "loss": 1.0163,
       "step": 1211
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989612636871915e-05,
-      "loss": 0.9007,
+      "loss": 1.0275,
       "step": 1212
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989558895408406e-05,
-      "loss": 0.8952,
+      "loss": 1.0089,
       "step": 1213
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989505015353942e-05,
-      "loss": 0.9302,
+      "loss": 0.9928,
       "step": 1214
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998945099670927e-05,
-      "loss": 0.8756,
+      "loss": 1.0134,
       "step": 1215
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989396839475134e-05,
-      "loss": 1.0264,
+      "loss": 1.0892,
       "step": 1216
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998934254365229e-05,
-      "loss": 0.884,
+      "loss": 0.9989,
       "step": 1217
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989288109241496e-05,
-      "loss": 1.0212,
+      "loss": 1.0385,
       "step": 1218
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.99892335362435e-05,
-      "loss": 0.9087,
+      "loss": 0.9961,
       "step": 1219
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998917882465906e-05,
-      "loss": 0.9375,
+      "loss": 0.9618,
       "step": 1220
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989123974488937e-05,
-      "loss": 0.8984,
+      "loss": 1.0654,
       "step": 1221
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9989068985733895e-05,
-      "loss": 0.775,
+      "loss": 0.9686,
       "step": 1222
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998901385839469e-05,
-      "loss": 1.0638,
+      "loss": 1.1155,
       "step": 1223
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988958592472092e-05,
-      "loss": 0.9034,
+      "loss": 1.0233,
       "step": 1224
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988903187966866e-05,
-      "loss": 0.9243,
+      "loss": 1.0776,
       "step": 1225
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998884764487978e-05,
-      "loss": 0.9885,
+      "loss": 0.9551,
       "step": 1226
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988791963211603e-05,
-      "loss": 0.8987,
+      "loss": 1.027,
       "step": 1227
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988736142963114e-05,
-      "loss": 1.0825,
+      "loss": 1.1125,
       "step": 1228
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988680184135082e-05,
-      "loss": 0.9171,
+      "loss": 0.9874,
       "step": 1229
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988624086728286e-05,
-      "loss": 0.8671,
+      "loss": 0.9822,
       "step": 1230
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988567850743497e-05,
-      "loss": 0.9345,
+      "loss": 1.0251,
       "step": 1231
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988511476181506e-05,
-      "loss": 0.892,
+      "loss": 1.0519,
       "step": 1232
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998845496304309e-05,
-      "loss": 0.8616,
+      "loss": 1.0401,
       "step": 1233
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998839831132903e-05,
-      "loss": 0.9678,
+      "loss": 1.0465,
       "step": 1234
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.998834152104012e-05,
-      "loss": 0.9071,
+      "loss": 0.9567,
       "step": 1235
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988284592177138e-05,
-      "loss": 0.8767,
+      "loss": 1.0182,
       "step": 1236
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9988227524740882e-05,
-      "loss": 0.8097,
+      "loss": 1.0235,
       "step": 1237
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998817031873214e-05,
-      "loss": 0.9178,
+      "loss": 1.0193,
       "step": 1238
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9988112974151704e-05,
-      "loss": 0.7979,
+      "loss": 1.0137,
       "step": 1239
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9988055491000374e-05,
-      "loss": 0.9025,
+      "loss": 1.0125,
       "step": 1240
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987997869278947e-05,
-      "loss": 0.8899,
+      "loss": 1.0366,
       "step": 1241
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987940108988215e-05,
-      "loss": 0.9205,
+      "loss": 0.9826,
       "step": 1242
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998788221012899e-05,
-      "loss": 0.9086,
+      "loss": 1.0088,
       "step": 1243
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987824172702067e-05,
-      "loss": 0.9677,
+      "loss": 1.0116,
       "step": 1244
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987765996708256e-05,
-      "loss": 0.9682,
+      "loss": 1.1375,
       "step": 1245
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987707682148363e-05,
-      "loss": 1.0212,
+      "loss": 1.0715,
       "step": 1246
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987649229023196e-05,
-      "loss": 0.8866,
+      "loss": 1.0242,
       "step": 1247
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987590637333567e-05,
-      "loss": 0.9079,
+      "loss": 1.0421,
       "step": 1248
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987531907080288e-05,
-      "loss": 0.8811,
+      "loss": 0.9823,
       "step": 1249
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987473038264172e-05,
-      "loss": 1.0197,
+      "loss": 1.0366,
       "step": 1250
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987414030886043e-05,
-      "loss": 0.9384,
+      "loss": 1.0765,
       "step": 1251
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987354884946713e-05,
-      "loss": 0.9557,
+      "loss": 1.0573,
       "step": 1252
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987295600447003e-05,
-      "loss": 0.8839,
+      "loss": 1.0779,
       "step": 1253
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987236177387736e-05,
-      "loss": 0.7509,
+      "loss": 0.9867,
       "step": 1254
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998717661576974e-05,
-      "loss": 1.0076,
+      "loss": 1.0667,
       "step": 1255
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987116915593835e-05,
-      "loss": 0.8927,
+      "loss": 1.0403,
       "step": 1256
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9987057076860853e-05,
-      "loss": 0.9091,
+      "loss": 1.0078,
       "step": 1257
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986997099571625e-05,
-      "loss": 0.9117,
+      "loss": 0.9198,
       "step": 1258
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986936983726984e-05,
-      "loss": 0.9838,
+      "loss": 1.0143,
       "step": 1259
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998687672932776e-05,
-      "loss": 1.0162,
+      "loss": 1.0606,
       "step": 1260
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998681633637479e-05,
-      "loss": 0.8987,
+      "loss": 1.0124,
       "step": 1261
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998675580486891e-05,
-      "loss": 0.8552,
+      "loss": 0.9246,
       "step": 1262
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986695134810968e-05,
-      "loss": 0.9872,
+      "loss": 1.115,
       "step": 1263
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99866343262018e-05,
-      "loss": 1.0945,
+      "loss": 1.2056,
       "step": 1264
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998657337904225e-05,
-      "loss": 0.8583,
+      "loss": 1.0431,
       "step": 1265
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986512293333156e-05,
-      "loss": 0.7827,
+      "loss": 0.8521,
       "step": 1266
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986451069075383e-05,
-      "loss": 1.0024,
+      "loss": 1.1635,
       "step": 1267
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986389706269765e-05,
-      "loss": 0.7902,
+      "loss": 0.9158,
       "step": 1268
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986328204917157e-05,
-      "loss": 1.0209,
+      "loss": 1.0128,
       "step": 1269
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998626656501842e-05,
-      "loss": 0.9813,
+      "loss": 1.0946,
       "step": 1270
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986204786574398e-05,
-      "loss": 0.8917,
+      "loss": 0.9803,
       "step": 1271
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986142869585955e-05,
-      "loss": 1.0256,
+      "loss": 1.0255,
       "step": 1272
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998608081405395e-05,
-      "loss": 0.9575,
+      "loss": 1.0566,
       "step": 1273
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9986018619979237e-05,
-      "loss": 0.9214,
+      "loss": 0.9362,
       "step": 1274
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985956287362688e-05,
-      "loss": 0.8836,
+      "loss": 0.9714,
       "step": 1275
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998589381620516e-05,
-      "loss": 0.8799,
+      "loss": 1.019,
       "step": 1276
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998583120650753e-05,
-      "loss": 0.9964,
+      "loss": 1.1252,
       "step": 1277
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985768458270656e-05,
-      "loss": 0.8421,
+      "loss": 1.0115,
       "step": 1278
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985705571495414e-05,
-      "loss": 0.9511,
+      "loss": 1.0373,
       "step": 1279
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985642546182672e-05,
-      "loss": 1.0762,
+      "loss": 1.1504,
       "step": 1280
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985579382333308e-05,
-      "loss": 0.9789,
+      "loss": 1.042,
       "step": 1281
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985516079948198e-05,
-      "loss": 1.0042,
+      "loss": 1.1176,
       "step": 1282
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985452639028224e-05,
-      "loss": 0.8366,
+      "loss": 0.9417,
       "step": 1283
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998538905957426e-05,
-      "loss": 1.0021,
+      "loss": 1.0847,
       "step": 1284
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985325341587186e-05,
-      "loss": 0.8399,
+      "loss": 0.9788,
       "step": 1285
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985261485067895e-05,
-      "loss": 0.9063,
+      "loss": 1.0122,
       "step": 1286
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998519749001727e-05,
-      "loss": 0.8877,
+      "loss": 1.0261,
       "step": 1287
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998513335643619e-05,
-      "loss": 0.8969,
+      "loss": 0.9305,
       "step": 1288
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985069084325557e-05,
-      "loss": 0.8897,
+      "loss": 0.9287,
       "step": 1289
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9985004673686255e-05,
-      "loss": 1.025,
+      "loss": 1.1059,
       "step": 1290
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998494012451918e-05,
-      "loss": 0.8257,
+      "loss": 1.0016,
       "step": 1291
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984875436825232e-05,
-      "loss": 0.8269,
+      "loss": 1.0262,
       "step": 1292
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984810610605295e-05,
-      "loss": 0.8512,
+      "loss": 0.9231,
       "step": 1293
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984745645860286e-05,
-      "loss": 0.9224,
+      "loss": 1.0247,
       "step": 1294
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984680542591092e-05,
-      "loss": 0.8992,
+      "loss": 1.0952,
       "step": 1295
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984615300798628e-05,
-      "loss": 1.0499,
+      "loss": 1.0582,
       "step": 1296
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984549920483788e-05,
-      "loss": 1.0088,
+      "loss": 1.0497,
       "step": 1297
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984484401647487e-05,
-      "loss": 0.8696,
+      "loss": 0.9404,
       "step": 1298
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998441874429063e-05,
-      "loss": 0.9034,
+      "loss": 1.1048,
       "step": 1299
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984352948414124e-05,
-      "loss": 0.8369,
+      "loss": 1.0319,
       "step": 1300
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984287014018892e-05,
-      "loss": 0.8702,
+      "loss": 1.0022,
       "step": 1301
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984220941105845e-05,
-      "loss": 0.8703,
+      "loss": 1.0116,
       "step": 1302
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984154729675894e-05,
-      "loss": 1.0349,
+      "loss": 1.1475,
       "step": 1303
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984088379729964e-05,
-      "loss": 0.9499,
+      "loss": 1.0013,
       "step": 1304
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9984021891268973e-05,
-      "loss": 0.985,
+      "loss": 1.1088,
       "step": 1305
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983955264293843e-05,
-      "loss": 0.9719,
+      "loss": 1.0646,
       "step": 1306
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983888498805503e-05,
-      "loss": 1.0598,
+      "loss": 1.0673,
       "step": 1307
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998382159480487e-05,
-      "loss": 0.9533,
+      "loss": 0.9772,
       "step": 1308
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998375455229288e-05,
-      "loss": 1.1099,
+      "loss": 1.1606,
       "step": 1309
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998368737127046e-05,
-      "loss": 0.9489,
+      "loss": 1.0153,
       "step": 1310
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983620051738545e-05,
-      "loss": 0.8948,
+      "loss": 0.9982,
       "step": 1311
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983552593698063e-05,
-      "loss": 0.9151,
+      "loss": 1.0049,
       "step": 1312
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998348499714996e-05,
-      "loss": 0.9215,
+      "loss": 1.0062,
       "step": 1313
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998341726209516e-05,
-      "loss": 0.9953,
+      "loss": 1.0389,
       "step": 1314
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983349388534612e-05,
-      "loss": 0.8747,
+      "loss": 1.0425,
       "step": 1315
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983281376469255e-05,
-      "loss": 0.9461,
+      "loss": 1.0579,
       "step": 1316
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983213225900036e-05,
-      "loss": 0.8401,
+      "loss": 0.9153,
       "step": 1317
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983144936827895e-05,
-      "loss": 1.0506,
+      "loss": 1.0237,
       "step": 1318
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998307650925379e-05,
-      "loss": 0.9537,
+      "loss": 1.0277,
       "step": 1319
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9983007943178655e-05,
-      "loss": 0.8371,
+      "loss": 0.986,
       "step": 1320
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982939238603447e-05,
-      "loss": 0.9744,
+      "loss": 1.0815,
       "step": 1321
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982870395529125e-05,
-      "loss": 0.8852,
+      "loss": 0.9763,
       "step": 1322
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998280141395664e-05,
-      "loss": 0.907,
+      "loss": 0.9717,
       "step": 1323
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998273229388695e-05,
-      "loss": 0.9163,
+      "loss": 1.001,
       "step": 1324
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998266303532101e-05,
-      "loss": 0.8461,
+      "loss": 0.8956,
       "step": 1325
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982593638259785e-05,
-      "loss": 0.8482,
+      "loss": 0.9156,
       "step": 1326
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998252410270424e-05,
-      "loss": 1.0674,
+      "loss": 1.1143,
       "step": 1327
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982454428655332e-05,
-      "loss": 0.9764,
+      "loss": 1.0801,
       "step": 1328
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982384616114036e-05,
-      "loss": 0.9237,
+      "loss": 1.0109,
       "step": 1329
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982314665081314e-05,
-      "loss": 0.7976,
+      "loss": 0.9395,
       "step": 1330
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998224457555814e-05,
-      "loss": 1.0125,
+      "loss": 1.071,
       "step": 1331
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9982174347545484e-05,
-      "loss": 0.9803,
+      "loss": 1.1248,
       "step": 1332
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998210398104432e-05,
-      "loss": 1.0457,
+      "loss": 1.1169,
       "step": 1333
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998203347605563e-05,
-      "loss": 0.7851,
+      "loss": 0.8956,
       "step": 1334
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998196283258039e-05,
-      "loss": 0.9122,
+      "loss": 1.001,
       "step": 1335
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981892050619574e-05,
-      "loss": 0.9618,
+      "loss": 1.0474,
       "step": 1336
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981821130174168e-05,
-      "loss": 1.0688,
+      "loss": 1.1219,
       "step": 1337
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998175007124516e-05,
-      "loss": 0.8475,
+      "loss": 1.0547,
       "step": 1338
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998167887383353e-05,
-      "loss": 0.967,
+      "loss": 1.0788,
       "step": 1339
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981607537940265e-05,
-      "loss": 1.0,
+      "loss": 1.0428,
       "step": 1340
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998153606356636e-05,
-      "loss": 0.8525,
+      "loss": 0.9312,
       "step": 1341
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981464450712804e-05,
-      "loss": 1.044,
+      "loss": 1.1134,
       "step": 1342
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981392699380593e-05,
-      "loss": 0.8897,
+      "loss": 1.0323,
       "step": 1343
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981320809570716e-05,
-      "loss": 0.8036,
+      "loss": 0.9745,
       "step": 1344
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981248781284175e-05,
-      "loss": 0.9461,
+      "loss": 0.9964,
       "step": 1345
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998117661452197e-05,
-      "loss": 0.9977,
+      "loss": 1.0325,
       "step": 1346
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99811043092851e-05,
-      "loss": 0.95,
+      "loss": 1.0184,
       "step": 1347
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9981031865574572e-05,
-      "loss": 0.933,
+      "loss": 0.9703,
       "step": 1348
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980959283391385e-05,
-      "loss": 0.8407,
+      "loss": 1.0662,
       "step": 1349
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998088656273655e-05,
-      "loss": 0.9766,
+      "loss": 1.017,
       "step": 1350
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998081370361107e-05,
-      "loss": 0.9451,
+      "loss": 1.0879,
       "step": 1351
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998074070601597e-05,
-      "loss": 0.9707,
+      "loss": 1.1027,
       "step": 1352
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998066756995225e-05,
-      "loss": 1.0107,
+      "loss": 1.0005,
       "step": 1353
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980594295420928e-05,
-      "loss": 0.8614,
+      "loss": 0.9648,
       "step": 1354
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980520882423018e-05,
-      "loss": 1.0363,
+      "loss": 1.0524,
       "step": 1355
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980447330959545e-05,
-      "loss": 0.9041,
+      "loss": 1.0479,
       "step": 1356
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980373641031525e-05,
-      "loss": 0.9385,
+      "loss": 1.068,
       "step": 1357
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980299812639977e-05,
-      "loss": 0.9224,
+      "loss": 1.0208,
       "step": 1358
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980225845785936e-05,
-      "loss": 1.0067,
+      "loss": 1.1129,
       "step": 1359
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9980151740470414e-05,
-      "loss": 0.8195,
+      "loss": 0.9796,
       "step": 1360
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998007749669445e-05,
-      "loss": 0.9637,
+      "loss": 1.0836,
       "step": 1361
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.998000311445907e-05,
-      "loss": 0.8668,
+      "loss": 0.9543,
       "step": 1362
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997992859376531e-05,
-      "loss": 1.051,
+      "loss": 1.0619,
       "step": 1363
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99798539346142e-05,
-      "loss": 0.8309,
+      "loss": 0.9781,
       "step": 1364
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997977913700677e-05,
-      "loss": 0.9641,
+      "loss": 1.0302,
       "step": 1365
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997970420094407e-05,
-      "loss": 0.8953,
+      "loss": 0.9754,
       "step": 1366
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979629126427132e-05,
-      "loss": 0.9339,
+      "loss": 1.0387,
       "step": 1367
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979553913456996e-05,
-      "loss": 0.899,
+      "loss": 1.017,
       "step": 1368
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979478562034713e-05,
-      "loss": 0.9216,
+      "loss": 0.9771,
       "step": 1369
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979403072161324e-05,
-      "loss": 0.8584,
+      "loss": 0.9799,
       "step": 1370
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979327443837872e-05,
-      "loss": 0.9681,
+      "loss": 1.0143,
       "step": 1371
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979251677065415e-05,
-      "loss": 0.922,
+      "loss": 0.9766,
       "step": 1372
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979175771844995e-05,
-      "loss": 1.0003,
+      "loss": 1.0016,
       "step": 1373
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979099728177674e-05,
-      "loss": 0.9263,
+      "loss": 1.0202,
       "step": 1374
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9979023546064505e-05,
-      "loss": 0.8768,
+      "loss": 0.9014,
       "step": 1375
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997894722550654e-05,
-      "loss": 0.9427,
+      "loss": 1.0881,
       "step": 1376
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978870766504838e-05,
-      "loss": 0.8168,
+      "loss": 1.0175,
       "step": 1377
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978794169060464e-05,
-      "loss": 0.9257,
+      "loss": 0.9673,
       "step": 1378
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978717433174478e-05,
-      "loss": 0.8897,
+      "loss": 1.0566,
       "step": 1379
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978640558847946e-05,
-      "loss": 0.9726,
+      "loss": 1.0139,
       "step": 1380
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978563546081932e-05,
-      "loss": 0.8219,
+      "loss": 0.9138,
       "step": 1381
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997848639487751e-05,
-      "loss": 0.9612,
+      "loss": 1.0393,
       "step": 1382
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997840910523575e-05,
-      "loss": 0.8201,
+      "loss": 0.9444,
       "step": 1383
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978331677157714e-05,
-      "loss": 0.9319,
+      "loss": 0.9912,
       "step": 1384
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978254110644487e-05,
-      "loss": 0.9753,
+      "loss": 1.054,
       "step": 1385
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997817640569714e-05,
-      "loss": 0.9062,
+      "loss": 0.9877,
       "step": 1386
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978098562316753e-05,
-      "loss": 0.9308,
+      "loss": 1.0378,
       "step": 1387
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9978020580504407e-05,
-      "loss": 0.9027,
+      "loss": 1.0119,
       "step": 1388
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977942460261182e-05,
-      "loss": 0.9199,
+      "loss": 0.9995,
       "step": 1389
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977864201588162e-05,
-      "loss": 0.8807,
+      "loss": 1.0084,
       "step": 1390
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977785804486434e-05,
-      "loss": 0.8887,
+      "loss": 0.9662,
       "step": 1391
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977707268957084e-05,
-      "loss": 1.0316,
+      "loss": 1.0695,
       "step": 1392
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99776285950012e-05,
-      "loss": 0.8319,
+      "loss": 1.0236,
       "step": 1393
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977549782619877e-05,
-      "loss": 1.1089,
+      "loss": 1.1158,
       "step": 1394
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977470831814206e-05,
-      "loss": 0.9719,
+      "loss": 1.0291,
       "step": 1395
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977391742585282e-05,
-      "loss": 0.8914,
+      "loss": 1.0138,
       "step": 1396
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977312514934207e-05,
-      "loss": 0.9197,
+      "loss": 1.0246,
       "step": 1397
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977233148862076e-05,
-      "loss": 0.866,
+      "loss": 1.0085,
       "step": 1398
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997715364436999e-05,
-      "loss": 1.012,
+      "loss": 1.0546,
       "step": 1399
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9977074001459056e-05,
-      "loss": 0.7981,
+      "loss": 1.0338,
       "step": 1400
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976994220130374e-05,
-      "loss": 0.9193,
+      "loss": 0.9663,
       "step": 1401
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997691430038505e-05,
-      "loss": 0.9674,
+      "loss": 1.0616,
       "step": 1402
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99768342422242e-05,
-      "loss": 0.9715,
+      "loss": 1.0298,
       "step": 1403
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997675404564893e-05,
-      "loss": 0.9997,
+      "loss": 1.1254,
       "step": 1404
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997667371066035e-05,
-      "loss": 0.9251,
+      "loss": 1.026,
       "step": 1405
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976593237259575e-05,
-      "loss": 0.9744,
+      "loss": 1.0764,
       "step": 1406
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976512625447726e-05,
-      "loss": 1.0457,
+      "loss": 1.0891,
       "step": 1407
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976431875225923e-05,
-      "loss": 0.9868,
+      "loss": 1.0319,
       "step": 1408
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976350986595277e-05,
-      "loss": 0.9459,
+      "loss": 0.9965,
       "step": 1409
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976269959556916e-05,
-      "loss": 0.9133,
+      "loss": 0.9509,
       "step": 1410
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997618879411197e-05,
-      "loss": 1.0893,
+      "loss": 1.0203,
       "step": 1411
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9976107490261554e-05,
-      "loss": 0.9718,
+      "loss": 1.0844,
       "step": 1412
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99760260480068e-05,
-      "loss": 0.9563,
+      "loss": 1.1241,
       "step": 1413
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975944467348843e-05,
-      "loss": 0.9064,
+      "loss": 0.9895,
       "step": 1414
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975862748288806e-05,
-      "loss": 0.9202,
+      "loss": 1.0386,
       "step": 1415
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997578089082783e-05,
-      "loss": 0.7781,
+      "loss": 1.0212,
       "step": 1416
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975698894967044e-05,
-      "loss": 0.8908,
+      "loss": 1.0128,
       "step": 1417
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975616760707594e-05,
-      "loss": 0.8374,
+      "loss": 0.9324,
       "step": 1418
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997553448805061e-05,
-      "loss": 0.8363,
+      "loss": 1.0199,
       "step": 1419
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975452076997244e-05,
-      "loss": 0.8049,
+      "loss": 0.9758,
       "step": 1420
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975369527548634e-05,
-      "loss": 0.9805,
+      "loss": 1.0568,
       "step": 1421
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997528683970592e-05,
-      "loss": 0.8883,
+      "loss": 0.9636,
       "step": 1422
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997520401347026e-05,
-      "loss": 0.9447,
+      "loss": 1.0763,
       "step": 1423
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997512104884279e-05,
-      "loss": 0.8575,
+      "loss": 1.079,
       "step": 1424
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9975037945824672e-05,
-      "loss": 0.9224,
+      "loss": 0.9883,
       "step": 1425
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974954704417056e-05,
-      "loss": 0.9775,
+      "loss": 1.0803,
       "step": 1426
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974871324621096e-05,
-      "loss": 0.9951,
+      "loss": 1.0285,
       "step": 1427
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997478780643795e-05,
-      "loss": 0.7686,
+      "loss": 0.9933,
       "step": 1428
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974704149868773e-05,
-      "loss": 0.8867,
+      "loss": 0.9902,
       "step": 1429
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997462035491473e-05,
-      "loss": 1.0082,
+      "loss": 1.042,
       "step": 1430
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997453642157698e-05,
-      "loss": 0.8914,
+      "loss": 1.0521,
       "step": 1431
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997445234985669e-05,
-      "loss": 0.868,
+      "loss": 1.0425,
       "step": 1432
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974368139755028e-05,
-      "loss": 1.0115,
+      "loss": 1.0749,
       "step": 1433
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974283791273158e-05,
-      "loss": 0.9318,
+      "loss": 1.0082,
       "step": 1434
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997419930441225e-05,
-      "loss": 0.8862,
+      "loss": 1.026,
       "step": 1435
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974114679173478e-05,
-      "loss": 0.8989,
+      "loss": 0.9924,
       "step": 1436
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9974029915558016e-05,
-      "loss": 1.0279,
+      "loss": 1.0346,
       "step": 1437
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973945013567044e-05,
-      "loss": 0.8991,
+      "loss": 0.9856,
       "step": 1438
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973859973201733e-05,
-      "loss": 1.0716,
+      "loss": 1.1036,
       "step": 1439
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997377479446327e-05,
-      "loss": 0.8439,
+      "loss": 0.9858,
       "step": 1440
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997368947735283e-05,
-      "loss": 0.8628,
+      "loss": 1.041,
       "step": 1441
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99736040218716e-05,
-      "loss": 1.043,
+      "loss": 1.0932,
       "step": 1442
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973518428020765e-05,
-      "loss": 0.8477,
+      "loss": 1.0543,
       "step": 1443
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973432695801508e-05,
-      "loss": 0.9794,
+      "loss": 1.0396,
       "step": 1444
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973346825215025e-05,
-      "loss": 0.8343,
+      "loss": 0.9965,
       "step": 1445
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973260816262508e-05,
-      "loss": 1.0168,
+      "loss": 1.1041,
       "step": 1446
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973174668945147e-05,
-      "loss": 0.9418,
+      "loss": 1.056,
       "step": 1447
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9973088383264137e-05,
-      "loss": 0.9855,
+      "loss": 1.0656,
       "step": 1448
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997300195922067e-05,
-      "loss": 0.864,
+      "loss": 1.0489,
       "step": 1449
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972915396815956e-05,
-      "loss": 0.9168,
+      "loss": 0.9662,
       "step": 1450
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972828696051194e-05,
-      "loss": 0.8843,
+      "loss": 0.9442,
       "step": 1451
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972741856927578e-05,
-      "loss": 0.9608,
+      "loss": 1.0997,
       "step": 1452
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997265487944632e-05,
-      "loss": 1.0685,
+      "loss": 1.1587,
       "step": 1453
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972567763608627e-05,
-      "loss": 1.0719,
+      "loss": 1.0952,
       "step": 1454
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972480509415702e-05,
-      "loss": 0.9299,
+      "loss": 0.9988,
       "step": 1455
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997239311686876e-05,
-      "loss": 1.0465,
+      "loss": 1.1304,
       "step": 1456
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972305585969013e-05,
-      "loss": 0.8666,
+      "loss": 0.9168,
       "step": 1457
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9972217916717675e-05,
-      "loss": 0.8428,
+      "loss": 0.9786,
       "step": 1458
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997213010911596e-05,
-      "loss": 0.8877,
+      "loss": 0.9714,
       "step": 1459
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997204216316509e-05,
-      "loss": 0.9556,
+      "loss": 1.0791,
       "step": 1460
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971954078866284e-05,
-      "loss": 1.0166,
+      "loss": 0.9966,
       "step": 1461
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971865856220763e-05,
-      "loss": 0.891,
+      "loss": 0.9931,
       "step": 1462
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997177749522975e-05,
-      "loss": 0.9406,
+      "loss": 0.942,
       "step": 1463
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971688995894476e-05,
-      "loss": 0.9847,
+      "loss": 1.0968,
       "step": 1464
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997160035821616e-05,
-      "loss": 1.0657,
+      "loss": 1.0654,
       "step": 1465
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971511582196045e-05,
-      "loss": 0.862,
+      "loss": 0.9655,
       "step": 1466
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971422667835348e-05,
-      "loss": 0.9077,
+      "loss": 1.0357,
       "step": 1467
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997133361513531e-05,
-      "loss": 0.9245,
+      "loss": 1.0125,
       "step": 1468
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971244424097165e-05,
-      "loss": 0.8746,
+      "loss": 0.9827,
       "step": 1469
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9971155094722153e-05,
-      "loss": 1.0043,
+      "loss": 1.0355,
       "step": 1470
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997106562701151e-05,
-      "loss": 0.9088,
+      "loss": 0.9911,
       "step": 1471
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970976020966478e-05,
-      "loss": 0.877,
+      "loss": 0.9797,
       "step": 1472
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970886276588303e-05,
-      "loss": 0.9474,
+      "loss": 1.0311,
       "step": 1473
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970796393878226e-05,
-      "loss": 1.0363,
+      "loss": 1.0876,
       "step": 1474
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970706372837494e-05,
-      "loss": 0.891,
+      "loss": 0.9988,
       "step": 1475
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970616213467358e-05,
-      "loss": 0.825,
+      "loss": 1.0129,
       "step": 1476
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997052591576907e-05,
-      "loss": 0.7684,
+      "loss": 0.9168,
       "step": 1477
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970435479743883e-05,
-      "loss": 0.7708,
+      "loss": 0.931,
       "step": 1478
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997034490539305e-05,
-      "loss": 0.9016,
+      "loss": 1.0094,
       "step": 1479
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970254192717826e-05,
-      "loss": 0.8961,
+      "loss": 0.9415,
       "step": 1480
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.997016334171947e-05,
-      "loss": 0.9155,
+      "loss": 0.9844,
       "step": 1481
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9970072352399246e-05,
-      "loss": 0.9485,
+      "loss": 1.0037,
       "step": 1482
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996998122475841e-05,
-      "loss": 0.7876,
+      "loss": 0.9264,
       "step": 1483
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9969889958798233e-05,
-      "loss": 0.7991,
+      "loss": 1.029,
       "step": 1484
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9969798554519976e-05,
-      "loss": 0.8601,
+      "loss": 0.9784,
       "step": 1485
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996970701192491e-05,
-      "loss": 0.8817,
+      "loss": 1.0392,
       "step": 1486
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9969615331014306e-05,
-      "loss": 0.972,
+      "loss": 1.0657,
       "step": 1487
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996952351178943e-05,
-      "loss": 0.9849,
+      "loss": 1.0355,
       "step": 1488
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9969431554251568e-05,
-      "loss": 0.9405,
+      "loss": 1.0125,
       "step": 1489
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996933945840198e-05,
-      "loss": 0.8509,
+      "loss": 0.9083,
       "step": 1490
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9969247224241958e-05,
-      "loss": 0.725,
+      "loss": 0.8915,
       "step": 1491
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996915485177277e-05,
-      "loss": 0.7391,
+      "loss": 0.9343,
       "step": 1492
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996906234099571e-05,
-      "loss": 1.0159,
+      "loss": 1.0407,
       "step": 1493
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996896969191205e-05,
-      "loss": 0.934,
+      "loss": 0.9978,
       "step": 1494
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968876904523075e-05,
-      "loss": 0.7912,
+      "loss": 0.921,
       "step": 1495
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968783978830082e-05,
-      "loss": 0.8167,
+      "loss": 0.8825,
       "step": 1496
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968690914834353e-05,
-      "loss": 0.9454,
+      "loss": 1.05,
       "step": 1497
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968597712537186e-05,
-      "loss": 0.8988,
+      "loss": 1.0455,
       "step": 1498
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968504371939866e-05,
-      "loss": 0.851,
+      "loss": 0.9462,
       "step": 1499
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.996841089304369e-05,
-      "loss": 0.8597,
+      "loss": 0.9999,
       "step": 1500
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968317275849957e-05,
-      "loss": 0.8954,
+      "loss": 0.9688,
       "step": 1501
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968223520359966e-05,
-      "loss": 1.131,
+      "loss": 1.1392,
       "step": 1502
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968129626575017e-05,
-      "loss": 0.9323,
+      "loss": 1.0483,
       "step": 1503
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9968035594496408e-05,
-      "loss": 0.9379,
+      "loss": 1.0102,
       "step": 1504
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967941424125454e-05,
-      "loss": 0.9273,
+      "loss": 1.0242,
       "step": 1505
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967847115463453e-05,
-      "loss": 0.8784,
+      "loss": 0.8953,
       "step": 1506
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967752668511715e-05,
-      "loss": 1.0278,
+      "loss": 1.0626,
       "step": 1507
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967658083271552e-05,
-      "loss": 0.717,
+      "loss": 0.9007,
       "step": 1508
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967563359744275e-05,
-      "loss": 0.9323,
+      "loss": 1.0295,
       "step": 1509
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.99674684979312e-05,
-      "loss": 0.9074,
+      "loss": 1.0649,
       "step": 1510
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967373497833637e-05,
-      "loss": 0.8296,
+      "loss": 0.9545,
       "step": 1511
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9967278359452913e-05,
-      "loss": 0.9728,
+      "loss": 1.0396,
       "step": 1512
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9967183082790343e-05,
-      "loss": 0.8799,
+      "loss": 0.9986,
       "step": 1513
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9967087667847246e-05,
-      "loss": 0.8948,
+      "loss": 1.0201,
       "step": 1514
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966992114624952e-05,
-      "loss": 0.9021,
+      "loss": 0.9163,
       "step": 1515
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966896423124785e-05,
-      "loss": 0.7834,
+      "loss": 1.0342,
       "step": 1516
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996680059334807e-05,
-      "loss": 0.9069,
+      "loss": 0.9659,
       "step": 1517
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966704625296135e-05,
-      "loss": 0.9169,
+      "loss": 0.9975,
       "step": 1518
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966608518970315e-05,
-      "loss": 0.8267,
+      "loss": 0.9501,
       "step": 1519
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966512274371944e-05,
-      "loss": 0.9419,
+      "loss": 1.0345,
       "step": 1520
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966415891502357e-05,
-      "loss": 1.012,
+      "loss": 1.0238,
       "step": 1521
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996631937036289e-05,
-      "loss": 0.9789,
+      "loss": 1.1107,
       "step": 1522
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996622271095488e-05,
-      "loss": 0.9781,
+      "loss": 1.0269,
       "step": 1523
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966125913279672e-05,
-      "loss": 0.7862,
+      "loss": 0.9504,
       "step": 1524
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9966028977338607e-05,
-      "loss": 1.0318,
+      "loss": 1.1383,
       "step": 1525
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996593190313303e-05,
-      "loss": 0.8126,
+      "loss": 0.922,
       "step": 1526
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965834690664286e-05,
-      "loss": 0.9598,
+      "loss": 1.024,
       "step": 1527
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965737339933728e-05,
-      "loss": 0.9681,
+      "loss": 1.0409,
       "step": 1528
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99656398509427e-05,
-      "loss": 0.8422,
+      "loss": 0.9473,
       "step": 1529
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996554222369256e-05,
-      "loss": 0.9109,
+      "loss": 0.9292,
       "step": 1530
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965444458184662e-05,
-      "loss": 0.7585,
+      "loss": 0.9102,
       "step": 1531
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996534655442036e-05,
-      "loss": 0.8452,
+      "loss": 0.9697,
       "step": 1532
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965248512401016e-05,
-      "loss": 0.8698,
+      "loss": 0.9148,
       "step": 1533
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965150332127985e-05,
-      "loss": 0.9112,
+      "loss": 1.0692,
       "step": 1534
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9965052013602633e-05,
-      "loss": 0.9739,
+      "loss": 1.0919,
       "step": 1535
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964953556826325e-05,
-      "loss": 0.7978,
+      "loss": 0.9364,
       "step": 1536
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996485496180042e-05,
-      "loss": 1.0874,
+      "loss": 1.0613,
       "step": 1537
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964756228526293e-05,
-      "loss": 0.7559,
+      "loss": 0.9598,
       "step": 1538
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996465735700531e-05,
-      "loss": 0.968,
+      "loss": 1.018,
       "step": 1539
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964558347238846e-05,
-      "loss": 0.8604,
+      "loss": 1.0164,
       "step": 1540
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964459199228274e-05,
-      "loss": 0.9673,
+      "loss": 1.041,
       "step": 1541
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964359912974966e-05,
-      "loss": 0.8011,
+      "loss": 0.954,
       "step": 1542
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99642604884803e-05,
-      "loss": 0.9309,
+      "loss": 1.0416,
       "step": 1543
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964160925745664e-05,
-      "loss": 1.0199,
+      "loss": 1.0435,
       "step": 1544
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9964061224772428e-05,
-      "loss": 1.0429,
+      "loss": 1.0552,
       "step": 1545
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963961385561978e-05,
-      "loss": 0.9887,
+      "loss": 0.988,
       "step": 1546
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963861408115704e-05,
-      "loss": 0.9159,
+      "loss": 1.052,
       "step": 1547
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963761292434988e-05,
-      "loss": 0.907,
+      "loss": 1.021,
       "step": 1548
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963661038521222e-05,
-      "loss": 0.899,
+      "loss": 1.041,
       "step": 1549
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963560646375796e-05,
-      "loss": 1.0241,
+      "loss": 1.0602,
       "step": 1550
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99634601160001e-05,
-      "loss": 0.8831,
+      "loss": 0.9498,
       "step": 1551
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963359447395532e-05,
-      "loss": 0.7004,
+      "loss": 0.9943,
       "step": 1552
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996325864056349e-05,
-      "loss": 0.9152,
+      "loss": 0.9633,
       "step": 1553
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9963157695505368e-05,
-      "loss": 0.9601,
+      "loss": 1.0119,
       "step": 1554
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996305661222257e-05,
-      "loss": 1.0024,
+      "loss": 1.0585,
       "step": 1555
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962955390716497e-05,
-      "loss": 0.8168,
+      "loss": 0.9875,
       "step": 1556
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962854030988554e-05,
-      "loss": 0.9586,
+      "loss": 1.0659,
       "step": 1557
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962752533040144e-05,
-      "loss": 0.9049,
+      "loss": 0.9818,
       "step": 1558
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962650896872682e-05,
-      "loss": 0.9036,
+      "loss": 1.1152,
       "step": 1559
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962549122487572e-05,
-      "loss": 0.9491,
+      "loss": 1.014,
       "step": 1560
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962447209886225e-05,
-      "loss": 1.137,
+      "loss": 1.1346,
       "step": 1561
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962345159070063e-05,
-      "loss": 0.9518,
+      "loss": 1.0236,
       "step": 1562
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9962242970040497e-05,
-      "loss": 0.8148,
+      "loss": 0.8929,
       "step": 1563
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996214064279894e-05,
-      "loss": 0.9184,
+      "loss": 1.0241,
       "step": 1564
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996203817734682e-05,
-      "loss": 0.8927,
+      "loss": 0.9707,
       "step": 1565
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996193557368555e-05,
-      "loss": 0.8955,
+      "loss": 0.9848,
       "step": 1566
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996183283181656e-05,
-      "loss": 0.7781,
+      "loss": 0.8283,
       "step": 1567
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961729951741275e-05,
-      "loss": 0.9869,
+      "loss": 1.0942,
       "step": 1568
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996162693346112e-05,
-      "loss": 1.1365,
+      "loss": 1.1973,
       "step": 1569
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961523776977524e-05,
-      "loss": 0.8623,
+      "loss": 1.0064,
       "step": 1570
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961420482291924e-05,
-      "loss": 1.0131,
+      "loss": 1.0925,
       "step": 1571
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961317049405742e-05,
-      "loss": 0.8688,
+      "loss": 0.9851,
       "step": 1572
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996121347832042e-05,
-      "loss": 0.768,
+      "loss": 0.8742,
       "step": 1573
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961109769037397e-05,
-      "loss": 0.9471,
+      "loss": 1.019,
       "step": 1574
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9961005921558106e-05,
-      "loss": 0.9061,
+      "loss": 0.9853,
       "step": 1575
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9960901935883993e-05,
-      "loss": 0.7639,
+      "loss": 0.9419,
       "step": 1576
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9960797812016496e-05,
-      "loss": 0.8619,
+      "loss": 1.0238,
       "step": 1577
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996069354995706e-05,
-      "loss": 0.9978,
+      "loss": 1.0154,
       "step": 1578
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996058914970714e-05,
-      "loss": 0.9862,
+      "loss": 1.0527,
       "step": 1579
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9960484611268173e-05,
-      "loss": 0.8048,
+      "loss": 0.9559,
       "step": 1580
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9960379934641615e-05,
-      "loss": 1.0604,
+      "loss": 1.0669,
       "step": 1581
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996027511982892e-05,
-      "loss": 0.9108,
+      "loss": 1.0497,
       "step": 1582
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9960170166831535e-05,
-      "loss": 0.9583,
+      "loss": 1.0618,
       "step": 1583
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.996006507565092e-05,
-      "loss": 0.9578,
+      "loss": 1.0529,
       "step": 1584
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995995984628854e-05,
-      "loss": 0.8454,
+      "loss": 0.9213,
       "step": 1585
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9959854478745847e-05,
-      "loss": 0.8973,
+      "loss": 0.9088,
       "step": 1586
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9959748973024303e-05,
-      "loss": 0.8571,
+      "loss": 0.9807,
       "step": 1587
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995964332912537e-05,
-      "loss": 0.9048,
+      "loss": 1.0363,
       "step": 1588
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995953754705052e-05,
-      "loss": 1.0065,
+      "loss": 1.0583,
       "step": 1589
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9959431626801215e-05,
-      "loss": 0.9317,
+      "loss": 1.0482,
       "step": 1590
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9959325568378927e-05,
-      "loss": 0.8911,
+      "loss": 0.9988,
       "step": 1591
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9959219371785128e-05,
-      "loss": 0.9235,
+      "loss": 1.045,
       "step": 1592
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995911303702129e-05,
-      "loss": 0.9534,
+      "loss": 1.047,
       "step": 1593
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995900656408889e-05,
-      "loss": 0.9484,
+      "loss": 1.0491,
       "step": 1594
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9958899952989403e-05,
-      "loss": 1.0616,
+      "loss": 1.0938,
       "step": 1595
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995879320372431e-05,
-      "loss": 0.9065,
+      "loss": 1.0078,
       "step": 1596
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995868631629509e-05,
-      "loss": 0.7729,
+      "loss": 0.9352,
       "step": 1597
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9958579290703227e-05,
-      "loss": 0.9901,
+      "loss": 1.0515,
       "step": 1598
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9958472126950207e-05,
-      "loss": 0.9904,
+      "loss": 1.0327,
       "step": 1599
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9958364825037514e-05,
-      "loss": 0.9404,
+      "loss": 0.9722,
       "step": 1600
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995825738496664e-05,
-      "loss": 0.8372,
+      "loss": 0.9274,
       "step": 1601
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995814980673907e-05,
-      "loss": 0.7298,
+      "loss": 0.931,
       "step": 1602
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9958042090356304e-05,
-      "loss": 0.8642,
+      "loss": 0.9758,
       "step": 1603
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9957934235819828e-05,
-      "loss": 0.9171,
+      "loss": 0.9658,
       "step": 1604
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995782624313115e-05,
-      "loss": 0.8627,
+      "loss": 0.9585,
       "step": 1605
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9957718112291754e-05,
-      "loss": 0.9438,
+      "loss": 0.9492,
       "step": 1606
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995760984330315e-05,
-      "loss": 0.7986,
+      "loss": 0.9256,
       "step": 1607
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995750143616684e-05,
-      "loss": 1.0889,
+      "loss": 1.008,
       "step": 1608
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9957392890884322e-05,
-      "loss": 0.9362,
+      "loss": 1.02,
       "step": 1609
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9957284207457108e-05,
-      "loss": 0.9922,
+      "loss": 1.0347,
       "step": 1610
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99571753858867e-05,
-      "loss": 0.8714,
+      "loss": 1.0038,
       "step": 1611
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9957066426174615e-05,
-      "loss": 1.0278,
+      "loss": 1.1345,
       "step": 1612
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956957328322355e-05,
-      "loss": 0.8165,
+      "loss": 1.0008,
       "step": 1613
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956848092331446e-05,
-      "loss": 1.0146,
+      "loss": 1.0348,
       "step": 1614
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995673871820339e-05,
-      "loss": 0.9642,
+      "loss": 1.0399,
       "step": 1615
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956629205939714e-05,
-      "loss": 0.8492,
+      "loss": 1.0986,
       "step": 1616
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956519555541934e-05,
-      "loss": 0.9659,
+      "loss": 1.0208,
       "step": 1617
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956409767011572e-05,
-      "loss": 0.8867,
+      "loss": 0.9333,
       "step": 1618
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995629984035015e-05,
-      "loss": 0.9941,
+      "loss": 1.0345,
       "step": 1619
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956189775559196e-05,
-      "loss": 0.8447,
+      "loss": 0.9953,
       "step": 1620
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9956079572640235e-05,
-      "loss": 0.8085,
+      "loss": 0.8832,
       "step": 1621
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995596923159479e-05,
-      "loss": 1.0015,
+      "loss": 1.0513,
       "step": 1622
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9955858752424403e-05,
-      "loss": 0.8869,
+      "loss": 1.0741,
       "step": 1623
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9955748135130602e-05,
-      "loss": 1.0614,
+      "loss": 1.1353,
       "step": 1624
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995563737971492e-05,
-      "loss": 0.8289,
+      "loss": 0.9404,
       "step": 1625
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9955526486178894e-05,
-      "loss": 0.8602,
+      "loss": 0.9594,
       "step": 1626
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995541545452406e-05,
-      "loss": 0.913,
+      "loss": 0.9909,
       "step": 1627
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9955304284751965e-05,
-      "loss": 0.9924,
+      "loss": 1.1204,
       "step": 1628
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9955192976864145e-05,
-      "loss": 0.9728,
+      "loss": 1.0303,
       "step": 1629
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995508153086215e-05,
-      "loss": 0.922,
+      "loss": 1.0139,
       "step": 1630
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954969946747522e-05,
-      "loss": 0.8393,
+      "loss": 1.0028,
       "step": 1631
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995485822452181e-05,
-      "loss": 0.9141,
+      "loss": 1.0197,
       "step": 1632
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995474636418657e-05,
-      "loss": 0.9677,
+      "loss": 1.1047,
       "step": 1633
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954634365743338e-05,
-      "loss": 0.9123,
+      "loss": 1.0443,
       "step": 1634
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954522229193685e-05,
-      "loss": 0.785,
+      "loss": 0.9509,
       "step": 1635
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954409954539156e-05,
-      "loss": 0.8305,
+      "loss": 1.0199,
       "step": 1636
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954297541781312e-05,
-      "loss": 1.0179,
+      "loss": 1.1205,
       "step": 1637
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954184990921716e-05,
-      "loss": 0.8694,
+      "loss": 0.9854,
       "step": 1638
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9954072301961926e-05,
-      "loss": 0.7063,
+      "loss": 0.9624,
       "step": 1639
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953959474903506e-05,
-      "loss": 0.8632,
+      "loss": 1.0151,
       "step": 1640
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953846509748017e-05,
-      "loss": 0.8737,
+      "loss": 1.0069,
       "step": 1641
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953733406497036e-05,
-      "loss": 0.882,
+      "loss": 1.0211,
       "step": 1642
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953620165152122e-05,
-      "loss": 0.7313,
+      "loss": 0.8728,
       "step": 1643
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953506785714854e-05,
-      "loss": 1.0152,
+      "loss": 1.0984,
       "step": 1644
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953393268186798e-05,
-      "loss": 0.9405,
+      "loss": 0.9199,
       "step": 1645
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953279612569535e-05,
-      "loss": 0.856,
+      "loss": 1.0135,
       "step": 1646
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9953165818864638e-05,
-      "loss": 0.9308,
+      "loss": 1.0243,
       "step": 1647
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995305188707369e-05,
-      "loss": 0.9387,
+      "loss": 1.0538,
       "step": 1648
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9952937817198266e-05,
-      "loss": 0.7989,
+      "loss": 0.9366,
       "step": 1649
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995282360923995e-05,
-      "loss": 0.9304,
+      "loss": 1.0021,
       "step": 1650
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995270926320033e-05,
-      "loss": 0.8723,
+      "loss": 0.8954,
       "step": 1651
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995259477908099e-05,
-      "loss": 0.8654,
+      "loss": 0.9913,
       "step": 1652
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9952480156883513e-05,
-      "loss": 0.9993,
+      "loss": 1.0546,
       "step": 1653
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99523653966095e-05,
-      "loss": 0.7859,
+      "loss": 0.9905,
       "step": 1654
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9952250498260536e-05,
-      "loss": 0.8894,
+      "loss": 0.9787,
       "step": 1655
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995213546183822e-05,
-      "loss": 0.9523,
+      "loss": 0.9745,
       "step": 1656
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995202028734414e-05,
-      "loss": 0.8257,
+      "loss": 0.9739,
       "step": 1657
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99519049747799e-05,
-      "loss": 0.9098,
+      "loss": 1.0153,
       "step": 1658
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99517895241471e-05,
-      "loss": 0.8859,
+      "loss": 1.0981,
       "step": 1659
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9951673935447337e-05,
-      "loss": 0.8526,
+      "loss": 1.0059,
       "step": 1660
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995155820868222e-05,
-      "loss": 0.8751,
+      "loss": 0.9874,
       "step": 1661
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9951442343853352e-05,
-      "loss": 0.9411,
+      "loss": 1.0356,
       "step": 1662
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9951326340962343e-05,
-      "loss": 1.0322,
+      "loss": 1.0649,
       "step": 1663
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9951210200010798e-05,
-      "loss": 0.8545,
+      "loss": 1.0162,
       "step": 1664
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995109392100033e-05,
-      "loss": 0.7903,
+      "loss": 0.9008,
       "step": 1665
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995097750393255e-05,
-      "loss": 0.7944,
+      "loss": 0.9129,
       "step": 1666
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995086094880908e-05,
-      "loss": 0.774,
+      "loss": 0.9287,
       "step": 1667
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995074425563153e-05,
-      "loss": 0.9524,
+      "loss": 1.1063,
       "step": 1668
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9950627424401524e-05,
-      "loss": 0.8392,
+      "loss": 0.9619,
       "step": 1669
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995051045512068e-05,
-      "loss": 0.9785,
+      "loss": 0.9468,
       "step": 1670
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9950393347790622e-05,
-      "loss": 0.8036,
+      "loss": 0.9234,
       "step": 1671
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995027610241297e-05,
-      "loss": 0.7961,
+      "loss": 0.9713,
       "step": 1672
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9950158718989358e-05,
-      "loss": 0.8763,
+      "loss": 0.9808,
       "step": 1673
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.995004119752141e-05,
-      "loss": 0.944,
+      "loss": 1.061,
       "step": 1674
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994992353801076e-05,
-      "loss": 1.0024,
+      "loss": 1.0544,
       "step": 1675
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994980574045904e-05,
-      "loss": 1.046,
+      "loss": 1.0282,
       "step": 1676
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9949687804867875e-05,
-      "loss": 0.8622,
+      "loss": 0.9525,
       "step": 1677
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9949569731238913e-05,
-      "loss": 0.9621,
+      "loss": 1.0354,
       "step": 1678
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994945151957379e-05,
-      "loss": 0.9258,
+      "loss": 0.9344,
       "step": 1679
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9949333169874137e-05,
-      "loss": 0.9957,
+      "loss": 1.0416,
       "step": 1680
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9949214682141606e-05,
-      "loss": 0.9566,
+      "loss": 0.96,
       "step": 1681
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994909605637784e-05,
-      "loss": 0.9619,
+      "loss": 1.0871,
       "step": 1682
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9948977292584477e-05,
-      "loss": 0.9761,
+      "loss": 1.1322,
       "step": 1683
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9948858390763173e-05,
-      "loss": 0.9223,
+      "loss": 0.9084,
       "step": 1684
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9948739350915573e-05,
-      "loss": 0.912,
+      "loss": 0.9432,
       "step": 1685
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994862017304333e-05,
-      "loss": 0.8424,
+      "loss": 1.0144,
       "step": 1686
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99485008571481e-05,
-      "loss": 0.8596,
+      "loss": 1.0026,
       "step": 1687
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9948381403231534e-05,
-      "loss": 0.8358,
+      "loss": 1.0199,
       "step": 1688
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994826181129529e-05,
-      "loss": 0.8408,
+      "loss": 1.0229,
       "step": 1689
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994814208134103e-05,
-      "loss": 0.8545,
+      "loss": 1.0161,
       "step": 1690
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9948022213370408e-05,
-      "loss": 0.8551,
+      "loss": 1.0,
       "step": 1691
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99479022073851e-05,
-      "loss": 0.9976,
+      "loss": 1.0905,
       "step": 1692
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9947782063386758e-05,
-      "loss": 0.8976,
+      "loss": 0.9757,
       "step": 1693
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9947661781377054e-05,
-      "loss": 0.8874,
+      "loss": 1.0237,
       "step": 1694
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994754136135766e-05,
-      "loss": 0.9255,
+      "loss": 1.0642,
       "step": 1695
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994742080333024e-05,
-      "loss": 0.779,
+      "loss": 0.916,
       "step": 1696
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994730010729647e-05,
-      "loss": 0.9681,
+      "loss": 1.0352,
       "step": 1697
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9947179273258026e-05,
-      "loss": 0.7941,
+      "loss": 1.0319,
       "step": 1698
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994705830121658e-05,
-      "loss": 0.9185,
+      "loss": 1.04,
       "step": 1699
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9946937191173817e-05,
-      "loss": 0.8743,
+      "loss": 1.0394,
       "step": 1700
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994681594313141e-05,
-      "loss": 0.9856,
+      "loss": 1.095,
       "step": 1701
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9946694557091045e-05,
-      "loss": 0.8977,
+      "loss": 1.0503,
       "step": 1702
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994657303305441e-05,
-      "loss": 0.8784,
+      "loss": 0.9928,
       "step": 1703
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994645137102318e-05,
-      "loss": 0.8444,
+      "loss": 0.9308,
       "step": 1704
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9946329570999054e-05,
-      "loss": 0.8402,
+      "loss": 0.9129,
       "step": 1705
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9946207632983715e-05,
-      "loss": 0.9073,
+      "loss": 1.0348,
       "step": 1706
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994608555697886e-05,
-      "loss": 0.8926,
+      "loss": 1.0046,
       "step": 1707
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9945963342986175e-05,
-      "loss": 0.9484,
+      "loss": 0.9726,
       "step": 1708
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9945840991007363e-05,
-      "loss": 0.9237,
+      "loss": 0.8997,
       "step": 1709
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994571850104412e-05,
-      "loss": 0.9512,
+      "loss": 1.0353,
       "step": 1710
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994559587309814e-05,
-      "loss": 0.8721,
+      "loss": 0.9548,
       "step": 1711
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994547310717113e-05,
-      "loss": 0.9731,
+      "loss": 1.0616,
       "step": 1712
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994535020326479e-05,
-      "loss": 1.0555,
+      "loss": 1.045,
       "step": 1713
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9945227161380828e-05,
-      "loss": 0.7336,
+      "loss": 0.9547,
       "step": 1714
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994510398152095e-05,
-      "loss": 0.9884,
+      "loss": 1.0183,
       "step": 1715
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9944980663686864e-05,
-      "loss": 0.9221,
+      "loss": 0.9985,
       "step": 1716
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994485720788028e-05,
-      "loss": 0.7963,
+      "loss": 0.9478,
       "step": 1717
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9944733614102914e-05,
-      "loss": 0.9025,
+      "loss": 1.0638,
       "step": 1718
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9944609882356478e-05,
-      "loss": 0.9204,
+      "loss": 0.9483,
       "step": 1719
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994448601264269e-05,
-      "loss": 0.9531,
+      "loss": 1.1164,
       "step": 1720
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994436200496327e-05,
-      "loss": 0.8266,
+      "loss": 0.9515,
       "step": 1721
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9944237859319932e-05,
-      "loss": 1.0084,
+      "loss": 1.0827,
       "step": 1722
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9944113575714406e-05,
-      "loss": 0.9706,
+      "loss": 1.0502,
       "step": 1723
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994398915414841e-05,
-      "loss": 0.8964,
+      "loss": 0.9728,
       "step": 1724
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9943864594623677e-05,
-      "loss": 0.8384,
+      "loss": 0.9838,
       "step": 1725
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994373989714193e-05,
-      "loss": 0.6963,
+      "loss": 0.8415,
       "step": 1726
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9943615061704898e-05,
-      "loss": 0.9331,
+      "loss": 1.0725,
       "step": 1727
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9943490088314317e-05,
-      "loss": 0.9399,
+      "loss": 1.023,
       "step": 1728
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994336497697192e-05,
-      "loss": 1.0307,
+      "loss": 1.058,
       "step": 1729
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9943239727679438e-05,
-      "loss": 0.9325,
+      "loss": 0.9572,
       "step": 1730
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9943114340438615e-05,
-      "loss": 0.8871,
+      "loss": 1.0146,
       "step": 1731
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994298881525119e-05,
-      "loss": 1.0019,
+      "loss": 1.0649,
       "step": 1732
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9942863152118898e-05,
-      "loss": 0.9913,
+      "loss": 1.0671,
       "step": 1733
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9942737351043487e-05,
-      "loss": 0.7751,
+      "loss": 0.901,
       "step": 1734
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9942611412026707e-05,
-      "loss": 0.8792,
+      "loss": 0.9552,
       "step": 1735
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9942485335070296e-05,
-      "loss": 0.8649,
+      "loss": 0.9521,
       "step": 1736
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994235912017601e-05,
-      "loss": 0.9678,
+      "loss": 1.0057,
       "step": 1737
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9942232767345596e-05,
-      "loss": 0.7864,
+      "loss": 0.9138,
       "step": 1738
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994210627658081e-05,
-      "loss": 0.9521,
+      "loss": 1.0269,
       "step": 1739
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99419796478834e-05,
-      "loss": 0.8886,
+      "loss": 1.0199,
       "step": 1740
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9941852881255137e-05,
-      "loss": 0.9133,
+      "loss": 0.9855,
       "step": 1741
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9941725976697768e-05,
-      "loss": 1.0225,
+      "loss": 1.0677,
       "step": 1742
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9941598934213055e-05,
-      "loss": 0.8863,
+      "loss": 0.965,
       "step": 1743
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994147175380276e-05,
-      "loss": 0.8777,
+      "loss": 0.9158,
       "step": 1744
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9941344435468656e-05,
-      "loss": 0.9364,
+      "loss": 1.0169,
       "step": 1745
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9941216979212498e-05,
-      "loss": 0.877,
+      "loss": 0.9873,
       "step": 1746
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994108938503606e-05,
-      "loss": 0.7823,
+      "loss": 0.95,
       "step": 1747
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940961652941108e-05,
-      "loss": 0.9796,
+      "loss": 1.0731,
       "step": 1748
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940833782929424e-05,
-      "loss": 0.8878,
+      "loss": 0.9691,
       "step": 1749
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.994070577500277e-05,
-      "loss": 0.8366,
+      "loss": 0.9734,
       "step": 1750
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940577629162928e-05,
-      "loss": 0.8974,
+      "loss": 1.0392,
       "step": 1751
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940449345411678e-05,
-      "loss": 0.8691,
+      "loss": 0.9392,
       "step": 1752
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940320923750796e-05,
-      "loss": 0.8516,
+      "loss": 0.9282,
       "step": 1753
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940192364182064e-05,
-      "loss": 0.9165,
+      "loss": 0.9975,
       "step": 1754
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9940063666707264e-05,
-      "loss": 0.7528,
+      "loss": 0.9119,
       "step": 1755
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993993483132819e-05,
-      "loss": 0.8314,
+      "loss": 0.964,
       "step": 1756
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993980585804662e-05,
-      "loss": 0.9377,
+      "loss": 1.0072,
       "step": 1757
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939676746864345e-05,
-      "loss": 0.7383,
+      "loss": 0.8124,
       "step": 1758
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939547497783157e-05,
-      "loss": 1.0593,
+      "loss": 1.0652,
       "step": 1759
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939418110804852e-05,
-      "loss": 0.9005,
+      "loss": 1.0901,
       "step": 1760
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939288585931224e-05,
-      "loss": 0.8687,
+      "loss": 1.0081,
       "step": 1761
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939158923164068e-05,
-      "loss": 0.9288,
+      "loss": 1.0751,
       "step": 1762
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9939029122505182e-05,
-      "loss": 0.9415,
+      "loss": 1.0751,
       "step": 1763
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993889918395637e-05,
-      "loss": 0.9236,
+      "loss": 1.0121,
       "step": 1764
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993876910751943e-05,
-      "loss": 0.8204,
+      "loss": 0.9402,
       "step": 1765
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9938638893196175e-05,
-      "loss": 0.9717,
+      "loss": 1.0164,
       "step": 1766
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9938508540988404e-05,
-      "loss": 0.9992,
+      "loss": 1.0117,
       "step": 1767
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9938378050897927e-05,
-      "loss": 0.8186,
+      "loss": 0.9747,
       "step": 1768
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9938247422926556e-05,
-      "loss": 0.9903,
+      "loss": 1.1245,
       "step": 1769
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9938116657076104e-05,
-      "loss": 0.8227,
+      "loss": 0.9543,
       "step": 1770
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937985753348383e-05,
-      "loss": 0.9551,
+      "loss": 0.9425,
       "step": 1771
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993785471174521e-05,
-      "loss": 0.8843,
+      "loss": 1.0187,
       "step": 1772
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937723532268403e-05,
-      "loss": 0.9579,
+      "loss": 1.0892,
       "step": 1773
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937592214919782e-05,
-      "loss": 0.8499,
+      "loss": 0.9408,
       "step": 1774
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937460759701164e-05,
-      "loss": 0.9945,
+      "loss": 1.0341,
       "step": 1775
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993732916661438e-05,
-      "loss": 0.863,
+      "loss": 0.9691,
       "step": 1776
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937197435661257e-05,
-      "loss": 1.1715,
+      "loss": 1.2139,
       "step": 1777
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9937065566843613e-05,
-      "loss": 0.8983,
+      "loss": 0.971,
       "step": 1778
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9936933560163287e-05,
-      "loss": 0.8817,
+      "loss": 0.9675,
       "step": 1779
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9936801415622104e-05,
-      "loss": 0.9052,
+      "loss": 1.0038,
       "step": 1780
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.99366691332219e-05,
-      "loss": 0.9149,
+      "loss": 1.0201,
       "step": 1781
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9936536712964515e-05,
-      "loss": 0.8429,
+      "loss": 0.9445,
       "step": 1782
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9936404154851777e-05,
-      "loss": 0.7938,
+      "loss": 0.9522,
       "step": 1783
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993627145888553e-05,
-      "loss": 0.8162,
+      "loss": 0.9575,
       "step": 1784
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9936138625067612e-05,
-      "loss": 0.9277,
+      "loss": 0.9733,
       "step": 1785
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.993600565339987e-05,
-      "loss": 1.0302,
+      "loss": 1.0904,
       "step": 1786
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9935872543884148e-05,
-      "loss": 0.9558,
+      "loss": 1.0126,
       "step": 1787
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993573929652229e-05,
-      "loss": 0.8375,
+      "loss": 0.9471,
       "step": 1788
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993560591131615e-05,
-      "loss": 0.8235,
+      "loss": 0.9858,
       "step": 1789
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993547238826757e-05,
-      "loss": 0.8931,
+      "loss": 0.9769,
       "step": 1790
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993533872737841e-05,
-      "loss": 0.9904,
+      "loss": 1.0717,
       "step": 1791
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9935204928650517e-05,
-      "loss": 0.8352,
+      "loss": 1.0206,
       "step": 1792
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9935070992085757e-05,
-      "loss": 0.8504,
+      "loss": 1.0351,
       "step": 1793
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993493691768598e-05,
-      "loss": 0.961,
+      "loss": 0.9907,
       "step": 1794
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993480270545305e-05,
-      "loss": 0.8495,
+      "loss": 0.9194,
       "step": 1795
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9934668355388828e-05,
-      "loss": 0.9421,
+      "loss": 1.0002,
       "step": 1796
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993453386749518e-05,
-      "loss": 0.9014,
+      "loss": 0.9983,
       "step": 1797
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9934399241773968e-05,
-      "loss": 0.9099,
+      "loss": 1.0015,
       "step": 1798
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993426447822706e-05,
-      "loss": 0.8656,
+      "loss": 0.9794,
       "step": 1799
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993412957685633e-05,
-      "loss": 0.7428,
+      "loss": 0.9041,
       "step": 1800
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993399453766365e-05,
-      "loss": 0.9067,
+      "loss": 0.998,
       "step": 1801
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993385936065088e-05,
-      "loss": 0.8456,
+      "loss": 0.9878,
       "step": 1802
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9933724045819916e-05,
-      "loss": 0.7942,
+      "loss": 0.929,
       "step": 1803
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993358859317262e-05,
-      "loss": 1.0427,
+      "loss": 1.0103,
       "step": 1804
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9933453002710878e-05,
-      "loss": 1.0029,
+      "loss": 1.0348,
       "step": 1805
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993331727443657e-05,
-      "loss": 0.9011,
+      "loss": 1.0071,
       "step": 1806
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9933181408351574e-05,
-      "loss": 0.7607,
+      "loss": 0.8646,
       "step": 1807
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9933045404457782e-05,
-      "loss": 1.015,
+      "loss": 1.0413,
       "step": 1808
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993290926275708e-05,
-      "loss": 0.8454,
+      "loss": 0.9807,
       "step": 1809
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9932772983251353e-05,
-      "loss": 0.8191,
+      "loss": 1.0044,
       "step": 1810
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993263656594249e-05,
-      "loss": 0.7483,
+      "loss": 0.8397,
       "step": 1811
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9932500010832392e-05,
-      "loss": 0.8999,
+      "loss": 1.0032,
       "step": 1812
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9932363317922945e-05,
-      "loss": 0.8687,
+      "loss": 1.0375,
       "step": 1813
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9932226487216055e-05,
-      "loss": 0.8568,
+      "loss": 1.0502,
       "step": 1814
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993208951871361e-05,
-      "loss": 1.0591,
+      "loss": 1.096,
       "step": 1815
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931952412417517e-05,
-      "loss": 0.9189,
+      "loss": 0.9227,
       "step": 1816
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931815168329674e-05,
-      "loss": 0.746,
+      "loss": 0.8968,
       "step": 1817
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931677786451987e-05,
-      "loss": 0.9386,
+      "loss": 0.9963,
       "step": 1818
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931540266786366e-05,
-      "loss": 0.865,
+      "loss": 1.0118,
       "step": 1819
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993140260933471e-05,
-      "loss": 1.0291,
+      "loss": 1.092,
       "step": 1820
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931264814098936e-05,
-      "loss": 0.8001,
+      "loss": 0.8106,
       "step": 1821
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9931126881080952e-05,
-      "loss": 0.8469,
+      "loss": 0.9687,
       "step": 1822
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993098881028267e-05,
-      "loss": 0.862,
+      "loss": 0.9336,
       "step": 1823
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930850601706015e-05,
-      "loss": 0.8888,
+      "loss": 0.9891,
       "step": 1824
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930712255352896e-05,
-      "loss": 0.9285,
+      "loss": 1.0848,
       "step": 1825
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.993057377122523e-05,
-      "loss": 1.0378,
+      "loss": 1.0438,
       "step": 1826
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930435149324947e-05,
-      "loss": 1.0041,
+      "loss": 1.1077,
       "step": 1827
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930296389653967e-05,
-      "loss": 0.8205,
+      "loss": 0.9048,
       "step": 1828
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930157492214207e-05,
-      "loss": 0.9987,
+      "loss": 1.009,
       "step": 1829
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9930018457007607e-05,
-      "loss": 0.843,
+      "loss": 1.0293,
       "step": 1830
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992987928403609e-05,
-      "loss": 0.9102,
+      "loss": 0.9864,
       "step": 1831
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9929739973301582e-05,
-      "loss": 0.8707,
+      "loss": 0.9568,
       "step": 1832
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992960052480602e-05,
-      "loss": 0.94,
+      "loss": 1.0587,
       "step": 1833
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9929460938551347e-05,
-      "loss": 0.85,
+      "loss": 0.8342,
       "step": 1834
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9929321214539486e-05,
-      "loss": 0.904,
+      "loss": 0.9794,
       "step": 1835
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992918135277238e-05,
-      "loss": 0.9482,
+      "loss": 1.0422,
       "step": 1836
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992904135325197e-05,
-      "loss": 0.7707,
+      "loss": 0.977,
       "step": 1837
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928901215980202e-05,
-      "loss": 0.8959,
+      "loss": 0.9457,
       "step": 1838
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928760940959016e-05,
-      "loss": 0.7011,
+      "loss": 0.8685,
       "step": 1839
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928620528190358e-05,
-      "loss": 0.947,
+      "loss": 1.0116,
       "step": 1840
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992847997767618e-05,
-      "loss": 0.7885,
+      "loss": 0.9524,
       "step": 1841
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928339289418422e-05,
-      "loss": 0.9366,
+      "loss": 0.9877,
       "step": 1842
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928198463419052e-05,
-      "loss": 0.9882,
+      "loss": 1.0423,
       "step": 1843
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9928057499680008e-05,
-      "loss": 0.8126,
+      "loss": 0.9228,
       "step": 1844
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9927916398203253e-05,
-      "loss": 1.0172,
+      "loss": 0.9612,
       "step": 1845
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9927775158990745e-05,
-      "loss": 0.971,
+      "loss": 0.9684,
       "step": 1846
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992763378204444e-05,
-      "loss": 0.999,
+      "loss": 1.0682,
       "step": 1847
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9927492267366306e-05,
-      "loss": 0.8971,
+      "loss": 0.9888,
       "step": 1848
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99273506149583e-05,
-      "loss": 0.9451,
+      "loss": 0.9588,
       "step": 1849
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9927208824822388e-05,
-      "loss": 0.8878,
+      "loss": 1.0616,
       "step": 1850
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992706689696054e-05,
-      "loss": 0.8588,
+      "loss": 0.9941,
       "step": 1851
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9926924831374723e-05,
-      "loss": 0.8137,
+      "loss": 0.9525,
       "step": 1852
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992678262806691e-05,
-      "loss": 1.0498,
+      "loss": 1.096,
       "step": 1853
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992664028703907e-05,
-      "loss": 0.8557,
+      "loss": 0.9621,
       "step": 1854
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9926497808293182e-05,
-      "loss": 0.9305,
+      "loss": 0.9915,
       "step": 1855
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992635519183122e-05,
-      "loss": 0.7761,
+      "loss": 0.9445,
       "step": 1856
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9926212437655163e-05,
-      "loss": 0.9706,
+      "loss": 1.1215,
       "step": 1857
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992606954576699e-05,
-      "loss": 0.8662,
+      "loss": 0.9746,
       "step": 1858
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992592651616869e-05,
-      "loss": 0.7971,
+      "loss": 1.0012,
       "step": 1859
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992578334886224e-05,
-      "loss": 0.9293,
+      "loss": 0.9825,
       "step": 1860
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992564004384963e-05,
-      "loss": 0.9303,
+      "loss": 1.0117,
       "step": 1861
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9925496601132845e-05,
-      "loss": 0.8888,
+      "loss": 0.9295,
       "step": 1862
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9925353020713878e-05,
-      "loss": 1.0561,
+      "loss": 1.1083,
       "step": 1863
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992520930259472e-05,
-      "loss": 0.9753,
+      "loss": 1.0926,
       "step": 1864
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9925065446777365e-05,
-      "loss": 0.8083,
+      "loss": 0.822,
       "step": 1865
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9924921453263807e-05,
-      "loss": 0.9352,
+      "loss": 0.9812,
       "step": 1866
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9924777322056048e-05,
-      "loss": 0.7108,
+      "loss": 0.9292,
       "step": 1867
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9924633053156084e-05,
-      "loss": 0.9223,
+      "loss": 0.9895,
       "step": 1868
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992448864656592e-05,
-      "loss": 1.0734,
+      "loss": 1.0582,
       "step": 1869
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9924344102287555e-05,
-      "loss": 0.8505,
+      "loss": 1.0651,
       "step": 1870
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9924199420322995e-05,
-      "loss": 0.9726,
+      "loss": 1.0411,
       "step": 1871
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992405460067425e-05,
-      "loss": 0.8748,
+      "loss": 0.961,
       "step": 1872
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9923909643343328e-05,
-      "loss": 0.8409,
+      "loss": 1.0038,
       "step": 1873
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992376454833224e-05,
-      "loss": 0.9098,
+      "loss": 0.9525,
       "step": 1874
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9923619315642998e-05,
-      "loss": 1.1771,
+      "loss": 1.2051,
       "step": 1875
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9923473945277618e-05,
-      "loss": 0.9314,
+      "loss": 1.0297,
       "step": 1876
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9923328437238115e-05,
-      "loss": 0.9575,
+      "loss": 1.1254,
       "step": 1877
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9923182791526516e-05,
-      "loss": 0.9146,
+      "loss": 1.0179,
       "step": 1878
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992303700814483e-05,
-      "loss": 0.9032,
+      "loss": 0.9871,
       "step": 1879
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9922891087095083e-05,
-      "loss": 0.9117,
+      "loss": 0.9722,
       "step": 1880
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99227450283793e-05,
-      "loss": 0.9388,
+      "loss": 1.0139,
       "step": 1881
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992259883199951e-05,
-      "loss": 0.9374,
+      "loss": 1.0013,
       "step": 1882
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992245249795774e-05,
-      "loss": 0.7838,
+      "loss": 0.933,
       "step": 1883
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9922306026256023e-05,
-      "loss": 0.8697,
+      "loss": 0.9499,
       "step": 1884
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9922159416896384e-05,
-      "loss": 0.8628,
+      "loss": 0.9264,
       "step": 1885
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992201266988086e-05,
-      "loss": 0.808,
+      "loss": 0.9466,
       "step": 1886
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9921865785211493e-05,
-      "loss": 0.8958,
+      "loss": 0.9556,
       "step": 1887
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992171876289031e-05,
-      "loss": 0.9002,
+      "loss": 1.0244,
       "step": 1888
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9921571602919356e-05,
-      "loss": 0.8112,
+      "loss": 0.8462,
       "step": 1889
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9921424305300677e-05,
-      "loss": 0.9882,
+      "loss": 1.0634,
       "step": 1890
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9921276870036312e-05,
-      "loss": 0.9321,
+      "loss": 1.0462,
       "step": 1891
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99211292971283e-05,
-      "loss": 0.9735,
+      "loss": 1.1083,
       "step": 1892
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9920981586578704e-05,
-      "loss": 0.7934,
+      "loss": 1.0137,
       "step": 1893
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992083373838956e-05,
-      "loss": 0.9415,
+      "loss": 1.0278,
       "step": 1894
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9920685752562925e-05,
-      "loss": 0.8871,
+      "loss": 0.9192,
       "step": 1895
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9920537629100848e-05,
-      "loss": 1.0023,
+      "loss": 1.0608,
       "step": 1896
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992038936800539e-05,
-      "loss": 0.7758,
+      "loss": 0.8986,
       "step": 1897
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9920240969278605e-05,
-      "loss": 0.9403,
+      "loss": 1.0115,
       "step": 1898
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.992009243292255e-05,
-      "loss": 0.8952,
+      "loss": 1.0131,
       "step": 1899
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9919943758939288e-05,
-      "loss": 0.7809,
+      "loss": 0.8762,
       "step": 1900
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9919794947330883e-05,
-      "loss": 0.8899,
+      "loss": 1.092,
       "step": 1901
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9919645998099394e-05,
-      "loss": 0.9684,
+      "loss": 1.0776,
       "step": 1902
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991949691124689e-05,
-      "loss": 0.8732,
+      "loss": 0.9711,
       "step": 1903
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9919347686775445e-05,
-      "loss": 0.9837,
+      "loss": 0.9994,
       "step": 1904
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991919832468712e-05,
-      "loss": 0.7852,
+      "loss": 0.8994,
       "step": 1905
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9919048824983993e-05,
-      "loss": 1.0322,
+      "loss": 1.0729,
       "step": 1906
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991889918766814e-05,
-      "loss": 0.8716,
+      "loss": 1.0358,
       "step": 1907
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991874941274163e-05,
-      "loss": 0.9556,
+      "loss": 1.0724,
       "step": 1908
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9918599500206545e-05,
-      "loss": 0.8885,
+      "loss": 0.9411,
       "step": 1909
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9918449450064966e-05,
-      "loss": 0.9376,
+      "loss": 0.9972,
       "step": 1910
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9918299262318974e-05,
-      "loss": 0.9829,
+      "loss": 1.0204,
       "step": 1911
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991814893697065e-05,
-      "loss": 0.9246,
+      "loss": 1.0108,
       "step": 1912
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917998474022083e-05,
-      "loss": 0.7704,
+      "loss": 0.9391,
       "step": 1913
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917847873475357e-05,
-      "loss": 0.9432,
+      "loss": 1.0215,
       "step": 1914
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917697135332567e-05,
-      "loss": 0.8651,
+      "loss": 0.9034,
       "step": 1915
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917546259595798e-05,
-      "loss": 0.9624,
+      "loss": 1.093,
       "step": 1916
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917395246267145e-05,
-      "loss": 0.9617,
+      "loss": 1.0029,
       "step": 1917
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917244095348707e-05,
-      "loss": 0.8689,
+      "loss": 0.9844,
       "step": 1918
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9917092806842577e-05,
-      "loss": 0.9215,
+      "loss": 1.0443,
       "step": 1919
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916941380750852e-05,
-      "loss": 0.9189,
+      "loss": 0.9822,
       "step": 1920
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916789817075637e-05,
-      "loss": 0.965,
+      "loss": 1.0399,
       "step": 1921
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916638115819032e-05,
-      "loss": 0.8616,
+      "loss": 0.9622,
       "step": 1922
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916486276983148e-05,
-      "loss": 0.8802,
+      "loss": 0.9897,
       "step": 1923
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916334300570083e-05,
-      "loss": 0.7704,
+      "loss": 0.9451,
       "step": 1924
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916182186581946e-05,
-      "loss": 0.7711,
+      "loss": 0.971,
       "step": 1925
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9916029935020856e-05,
-      "loss": 0.957,
+      "loss": 1.026,
       "step": 1926
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9915877545888917e-05,
-      "loss": 0.8916,
+      "loss": 0.9735,
       "step": 1927
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9915725019188245e-05,
-      "loss": 0.8577,
+      "loss": 0.9446,
       "step": 1928
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991557235492096e-05,
-      "loss": 0.8653,
+      "loss": 1.0439,
       "step": 1929
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9915419553089178e-05,
-      "loss": 0.7715,
+      "loss": 0.8596,
       "step": 1930
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9915266613695014e-05,
-      "loss": 0.8079,
+      "loss": 0.9706,
       "step": 1931
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9915113536740595e-05,
-      "loss": 0.9615,
+      "loss": 1.1196,
       "step": 1932
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914960322228042e-05,
-      "loss": 0.9323,
+      "loss": 0.9845,
       "step": 1933
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914806970159484e-05,
-      "loss": 0.9484,
+      "loss": 0.9921,
       "step": 1934
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914653480537047e-05,
-      "loss": 0.9608,
+      "loss": 1.0274,
       "step": 1935
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914499853362853e-05,
-      "loss": 0.8129,
+      "loss": 0.9901,
       "step": 1936
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914346088639046e-05,
-      "loss": 0.8738,
+      "loss": 1.0285,
       "step": 1937
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914192186367755e-05,
-      "loss": 0.9233,
+      "loss": 1.0543,
       "step": 1938
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9914038146551113e-05,
-      "loss": 0.882,
+      "loss": 1.0425,
       "step": 1939
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991388396919126e-05,
-      "loss": 0.7999,
+      "loss": 0.9732,
       "step": 1940
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9913729654290327e-05,
-      "loss": 0.8296,
+      "loss": 0.9727,
       "step": 1941
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9913575201850465e-05,
-      "loss": 0.904,
+      "loss": 1.0383,
       "step": 1942
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9913420611873812e-05,
-      "loss": 1.0568,
+      "loss": 1.0552,
       "step": 1943
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9913265884362515e-05,
-      "loss": 0.7841,
+      "loss": 0.9238,
       "step": 1944
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9913111019318715e-05,
-      "loss": 1.061,
+      "loss": 1.0408,
       "step": 1945
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9912956016744568e-05,
-      "loss": 0.9944,
+      "loss": 1.0472,
       "step": 1946
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9912800876642217e-05,
-      "loss": 0.9578,
+      "loss": 0.998,
       "step": 1947
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991264559901382e-05,
-      "loss": 0.896,
+      "loss": 1.0276,
       "step": 1948
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991249018386153e-05,
-      "loss": 0.8065,
+      "loss": 1.0244,
       "step": 1949
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9912334631187504e-05,
-      "loss": 0.8171,
+      "loss": 0.94,
       "step": 1950
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9912178940993897e-05,
-      "loss": 0.9097,
+      "loss": 1.0712,
       "step": 1951
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9912023113282872e-05,
-      "loss": 0.9872,
+      "loss": 1.0742,
       "step": 1952
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911867148056587e-05,
-      "loss": 0.9305,
+      "loss": 1.0089,
       "step": 1953
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911711045317213e-05,
-      "loss": 0.8695,
+      "loss": 0.9794,
       "step": 1954
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911554805066905e-05,
-      "loss": 1.002,
+      "loss": 1.0257,
       "step": 1955
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911398427307842e-05,
-      "loss": 0.9712,
+      "loss": 1.0797,
       "step": 1956
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911241912042186e-05,
-      "loss": 0.7925,
+      "loss": 0.9019,
       "step": 1957
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9911085259272112e-05,
-      "loss": 0.8431,
+      "loss": 0.9631,
       "step": 1958
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991092846899979e-05,
-      "loss": 1.0583,
+      "loss": 1.0787,
       "step": 1959
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99107715412274e-05,
-      "loss": 0.8695,
+      "loss": 1.0132,
       "step": 1960
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9910614475957114e-05,
-      "loss": 0.9918,
+      "loss": 1.0879,
       "step": 1961
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9910457273191115e-05,
-      "loss": 0.807,
+      "loss": 0.9006,
       "step": 1962
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.991029993293158e-05,
-      "loss": 0.8756,
+      "loss": 0.9837,
       "step": 1963
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9910142455180696e-05,
-      "loss": 0.9014,
+      "loss": 1.0746,
       "step": 1964
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990998483994065e-05,
-      "loss": 0.9205,
+      "loss": 1.0673,
       "step": 1965
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9909827087213617e-05,
-      "loss": 0.847,
+      "loss": 0.9842,
       "step": 1966
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99096691970018e-05,
-      "loss": 0.9331,
+      "loss": 1.043,
       "step": 1967
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9909511169307378e-05,
-      "loss": 0.8219,
+      "loss": 0.9848,
       "step": 1968
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990935300413255e-05,
-      "loss": 0.8151,
+      "loss": 0.9204,
       "step": 1969
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990919470147951e-05,
-      "loss": 0.9886,
+      "loss": 1.042,
       "step": 1970
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990903626135045e-05,
-      "loss": 0.7815,
+      "loss": 0.989,
       "step": 1971
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9908877683747572e-05,
-      "loss": 0.9416,
+      "loss": 1.0581,
       "step": 1972
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9908718968673076e-05,
-      "loss": 0.8954,
+      "loss": 0.962,
       "step": 1973
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9908560116129164e-05,
-      "loss": 0.8534,
+      "loss": 0.925,
       "step": 1974
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990840112611804e-05,
-      "loss": 0.9803,
+      "loss": 1.0425,
       "step": 1975
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9908241998641903e-05,
-      "loss": 0.9153,
+      "loss": 0.9514,
       "step": 1976
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9908082733702972e-05,
-      "loss": 1.0276,
+      "loss": 1.1094,
       "step": 1977
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990792333130345e-05,
-      "loss": 0.9643,
+      "loss": 1.0738,
       "step": 1978
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9907763791445544e-05,
-      "loss": 1.0051,
+      "loss": 1.033,
       "step": 1979
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9907604114131476e-05,
-      "loss": 0.8176,
+      "loss": 0.9404,
       "step": 1980
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9907444299363455e-05,
-      "loss": 0.901,
+      "loss": 1.0768,
       "step": 1981
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990728434714371e-05,
-      "loss": 0.8126,
+      "loss": 0.9682,
       "step": 1982
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9907124257474445e-05,
-      "loss": 0.8886,
+      "loss": 0.9948,
       "step": 1983
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9906964030357884e-05,
-      "loss": 0.9041,
+      "loss": 1.0115,
       "step": 1984
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9906803665796257e-05,
-      "loss": 0.8381,
+      "loss": 0.9599,
       "step": 1985
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9906643163791785e-05,
-      "loss": 0.7684,
+      "loss": 0.9542,
       "step": 1986
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9906482524346694e-05,
-      "loss": 0.9368,
+      "loss": 1.024,
       "step": 1987
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9906321747463218e-05,
-      "loss": 0.9066,
+      "loss": 1.0859,
       "step": 1988
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990616083314358e-05,
-      "loss": 0.7816,
+      "loss": 0.9403,
       "step": 1989
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990599978139001e-05,
-      "loss": 0.7684,
+      "loss": 0.9858,
       "step": 1990
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9905838592204754e-05,
-      "loss": 0.9457,
+      "loss": 0.9929,
       "step": 1991
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9905677265590042e-05,
-      "loss": 0.9394,
+      "loss": 1.0164,
       "step": 1992
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990551580154811e-05,
-      "loss": 1.0792,
+      "loss": 1.0833,
       "step": 1993
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9905354200081198e-05,
-      "loss": 0.9665,
+      "loss": 1.0834,
       "step": 1994
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990519246119156e-05,
-      "loss": 0.8326,
+      "loss": 0.9976,
       "step": 1995
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990503058488142e-05,
-      "loss": 0.9587,
+      "loss": 1.0051,
       "step": 1996
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9904868571153038e-05,
-      "loss": 0.7696,
+      "loss": 0.9515,
       "step": 1997
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990470642000866e-05,
-      "loss": 0.8916,
+      "loss": 1.0411,
       "step": 1998
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9904544131450527e-05,
-      "loss": 1.0815,
+      "loss": 1.1053,
       "step": 1999
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99043817054809e-05,
-      "loss": 0.8859,
+      "loss": 1.0025,
       "step": 2000
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9904219142102032e-05,
-      "loss": 0.756,
+      "loss": 1.039,
       "step": 2001
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9904056441316174e-05,
-      "loss": 0.8029,
+      "loss": 0.941,
       "step": 2002
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9903893603125587e-05,
-      "loss": 0.8591,
+      "loss": 1.0524,
       "step": 2003
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9903730627532524e-05,
-      "loss": 0.8598,
+      "loss": 1.0163,
       "step": 2004
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9903567514539254e-05,
-      "loss": 0.7795,
+      "loss": 0.9662,
       "step": 2005
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990340426414803e-05,
-      "loss": 0.9853,
+      "loss": 1.0184,
       "step": 2006
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9903240876361126e-05,
-      "loss": 0.9599,
+      "loss": 1.0307,
       "step": 2007
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9903077351180808e-05,
-      "loss": 0.964,
+      "loss": 1.1084,
       "step": 2008
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990291368860934e-05,
-      "loss": 0.867,
+      "loss": 0.9684,
       "step": 2009
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9902749888648996e-05,
-      "loss": 0.7978,
+      "loss": 0.8842,
       "step": 2010
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9902585951302046e-05,
-      "loss": 0.9093,
+      "loss": 0.937,
       "step": 2011
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990242187657077e-05,
-      "loss": 0.8399,
+      "loss": 0.9712,
       "step": 2012
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9902257664457435e-05,
-      "loss": 0.93,
+      "loss": 1.0134,
       "step": 2013
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9902093314964327e-05,
-      "loss": 0.7782,
+      "loss": 0.8859,
       "step": 2014
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9901928828093724e-05,
-      "loss": 0.9589,
+      "loss": 1.0621,
       "step": 2015
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9901764203847906e-05,
-      "loss": 0.8701,
+      "loss": 1.0484,
       "step": 2016
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990159944222916e-05,
-      "loss": 0.8935,
+      "loss": 0.959,
       "step": 2017
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990143454323977e-05,
-      "loss": 0.879,
+      "loss": 0.965,
       "step": 2018
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9901269506882023e-05,
-      "loss": 0.9213,
+      "loss": 0.9951,
       "step": 2019
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990110433315821e-05,
-      "loss": 0.9083,
+      "loss": 0.996,
       "step": 2020
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9900939022070624e-05,
-      "loss": 1.1019,
+      "loss": 1.0951,
       "step": 2021
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990077357362155e-05,
-      "loss": 0.8999,
+      "loss": 0.9477,
       "step": 2022
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99006079878133e-05,
-      "loss": 0.9251,
+      "loss": 0.9178,
       "step": 2023
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9900442264648155e-05,
-      "loss": 0.7969,
+      "loss": 0.968,
       "step": 2024
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.990027640412842e-05,
-      "loss": 0.8011,
+      "loss": 0.9779,
       "step": 2025
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99001104062564e-05,
-      "loss": 0.9121,
+      "loss": 1.0485,
       "step": 2026
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989994427103439e-05,
-      "loss": 0.936,
+      "loss": 1.0456,
       "step": 2027
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.98997779984647e-05,
-      "loss": 0.8527,
+      "loss": 0.8848,
       "step": 2028
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989961158854964e-05,
-      "loss": 0.7367,
+      "loss": 0.8676,
       "step": 2029
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9899445041291516e-05,
-      "loss": 0.9649,
+      "loss": 1.0025,
       "step": 2030
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9899278356692636e-05,
-      "loss": 0.825,
+      "loss": 0.9464,
       "step": 2031
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9899111534755312e-05,
-      "loss": 0.7958,
+      "loss": 1.0032,
       "step": 2032
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9898944575481862e-05,
-      "loss": 0.808,
+      "loss": 0.9554,
       "step": 2033
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.98987774788746e-05,
-      "loss": 0.8636,
+      "loss": 1.0543,
       "step": 2034
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9898610244935847e-05,
-      "loss": 0.8084,
+      "loss": 0.9228,
       "step": 2035
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9898442873667916e-05,
-      "loss": 0.9143,
+      "loss": 0.9703,
       "step": 2036
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989827536507314e-05,
-      "loss": 0.8368,
+      "loss": 0.9757,
       "step": 2037
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9898107719153833e-05,
-      "loss": 0.8798,
+      "loss": 0.9529,
       "step": 2038
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897939935912326e-05,
-      "loss": 1.0312,
+      "loss": 1.0244,
       "step": 2039
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897772015350945e-05,
-      "loss": 1.0038,
+      "loss": 1.0656,
       "step": 2040
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897603957472022e-05,
-      "loss": 0.9247,
+      "loss": 1.0027,
       "step": 2041
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897435762277888e-05,
-      "loss": 0.975,
+      "loss": 1.0237,
       "step": 2042
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897267429770873e-05,
-      "loss": 1.0372,
+      "loss": 1.0662,
       "step": 2043
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9897098959953316e-05,
-      "loss": 0.7395,
+      "loss": 0.9586,
       "step": 2044
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9896930352827554e-05,
-      "loss": 0.8217,
+      "loss": 0.8845,
       "step": 2045
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989676160839592e-05,
-      "loss": 0.7793,
+      "loss": 0.9257,
       "step": 2046
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989659272666077e-05,
-      "loss": 0.8398,
+      "loss": 1.0091,
       "step": 2047
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989642370762443e-05,
-      "loss": 0.9416,
+      "loss": 0.9238,
       "step": 2048
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9896254551289255e-05,
-      "loss": 0.81,
+      "loss": 0.9538,
       "step": 2049
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9896085257657587e-05,
-      "loss": 0.9498,
+      "loss": 1.0368,
       "step": 2050
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9895915826731778e-05,
-      "loss": 0.9378,
+      "loss": 0.9902,
       "step": 2051
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9895746258514177e-05,
-      "loss": 1.0408,
+      "loss": 1.0792,
       "step": 2052
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9895576553007138e-05,
-      "loss": 0.9805,
+      "loss": 1.0407,
       "step": 2053
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9895406710213012e-05,
-      "loss": 0.9476,
+      "loss": 0.9986,
       "step": 2054
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989523673013416e-05,
-      "loss": 0.9232,
+      "loss": 0.9865,
       "step": 2055
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9895066612772937e-05,
-      "loss": 0.9288,
+      "loss": 0.9616,
       "step": 2056
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9894896358131707e-05,
-      "loss": 0.7604,
+      "loss": 0.9059,
       "step": 2057
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9894725966212824e-05,
-      "loss": 0.7447,
+      "loss": 0.9361,
       "step": 2058
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9894555437018658e-05,
-      "loss": 0.8481,
+      "loss": 0.9149,
       "step": 2059
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9894384770551574e-05,
-      "loss": 0.8963,
+      "loss": 0.984,
       "step": 2060
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9894213966813942e-05,
-      "loss": 0.7953,
+      "loss": 0.9478,
       "step": 2061
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.989404302580813e-05,
-      "loss": 0.8673,
+      "loss": 0.9052,
       "step": 2062
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9893871947536507e-05,
-      "loss": 0.937,
+      "loss": 0.9929,
       "step": 2063
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9893700732001447e-05,
-      "loss": 0.8711,
+      "loss": 0.9988,
       "step": 2064
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.989352937920533e-05,
-      "loss": 0.9143,
+      "loss": 1.0217,
       "step": 2065
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9893357889150523e-05,
-      "loss": 0.8755,
+      "loss": 1.0071,
       "step": 2066
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9893186261839415e-05,
-      "loss": 1.0051,
+      "loss": 1.0005,
       "step": 2067
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9893014497274387e-05,
-      "loss": 0.9309,
+      "loss": 1.0235,
       "step": 2068
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.989284259545782e-05,
-      "loss": 0.7927,
+      "loss": 0.9872,
       "step": 2069
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98926705563921e-05,
-      "loss": 0.9101,
+      "loss": 0.9907,
       "step": 2070
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9892498380079605e-05,
-      "loss": 1.0224,
+      "loss": 1.0612,
       "step": 2071
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9892326066522732e-05,
-      "loss": 0.8866,
+      "loss": 1.0308,
       "step": 2072
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9892153615723873e-05,
-      "loss": 0.901,
+      "loss": 0.9461,
       "step": 2073
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9891981027685418e-05,
-      "loss": 0.8398,
+      "loss": 1.0218,
       "step": 2074
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.989180830240976e-05,
-      "loss": 1.0269,
+      "loss": 1.1562,
       "step": 2075
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9891635439899294e-05,
-      "loss": 0.7643,
+      "loss": 1.0093,
       "step": 2076
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9891462440156423e-05,
-      "loss": 0.9062,
+      "loss": 0.9234,
       "step": 2077
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9891289303183545e-05,
-      "loss": 0.9291,
+      "loss": 0.9774,
       "step": 2078
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.989111602898306e-05,
-      "loss": 0.843,
+      "loss": 0.9343,
       "step": 2079
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890942617557378e-05,
-      "loss": 0.9669,
+      "loss": 1.0652,
       "step": 2080
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890769068908896e-05,
-      "loss": 0.8135,
+      "loss": 0.8582,
       "step": 2081
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890595383040027e-05,
-      "loss": 0.8349,
+      "loss": 0.9182,
       "step": 2082
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890421559953182e-05,
-      "loss": 0.9849,
+      "loss": 0.9941,
       "step": 2083
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890247599650768e-05,
-      "loss": 0.9548,
+      "loss": 1.0176,
       "step": 2084
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9890073502135206e-05,
-      "loss": 0.9171,
+      "loss": 1.0204,
       "step": 2085
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9889899267408903e-05,
-      "loss": 0.9597,
+      "loss": 1.002,
       "step": 2086
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988972489547428e-05,
-      "loss": 0.9149,
+      "loss": 1.0218,
       "step": 2087
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9889550386333753e-05,
-      "loss": 0.9801,
+      "loss": 1.0403,
       "step": 2088
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988937573998975e-05,
-      "loss": 0.8482,
+      "loss": 1.009,
       "step": 2089
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9889200956444687e-05,
-      "loss": 0.9344,
+      "loss": 1.0747,
       "step": 2090
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9889026035700992e-05,
-      "loss": 0.8638,
+      "loss": 0.9445,
       "step": 2091
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9888850977761094e-05,
-      "loss": 0.9637,
+      "loss": 1.0558,
       "step": 2092
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9888675782627415e-05,
-      "loss": 1.0337,
+      "loss": 1.0393,
       "step": 2093
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9888500450302394e-05,
-      "loss": 0.8336,
+      "loss": 0.9936,
       "step": 2094
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9888324980788455e-05,
-      "loss": 0.997,
+      "loss": 1.1106,
       "step": 2095
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988814937408804e-05,
-      "loss": 0.8207,
+      "loss": 0.9395,
       "step": 2096
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887973630203578e-05,
-      "loss": 0.927,
+      "loss": 0.9569,
       "step": 2097
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887797749137516e-05,
-      "loss": 1.0243,
+      "loss": 0.9829,
       "step": 2098
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887621730892285e-05,
-      "loss": 0.9004,
+      "loss": 0.9711,
       "step": 2099
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887445575470334e-05,
-      "loss": 0.9583,
+      "loss": 0.999,
       "step": 2100
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887269282874104e-05,
-      "loss": 0.8188,
+      "loss": 0.9672,
       "step": 2101
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9887092853106037e-05,
-      "loss": 0.9628,
+      "loss": 1.0407,
       "step": 2102
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988691628616859e-05,
-      "loss": 0.8776,
+      "loss": 1.0077,
       "step": 2103
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9886739582064206e-05,
-      "loss": 0.8662,
+      "loss": 0.9827,
       "step": 2104
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9886562740795337e-05,
-      "loss": 0.8713,
+      "loss": 0.9716,
       "step": 2105
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9886385762364434e-05,
-      "loss": 0.8409,
+      "loss": 0.8961,
       "step": 2106
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988620864677396e-05,
-      "loss": 0.9199,
+      "loss": 0.9845,
       "step": 2107
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9886031394026365e-05,
-      "loss": 0.788,
+      "loss": 0.9117,
       "step": 2108
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988585400412411e-05,
-      "loss": 0.9027,
+      "loss": 1.0241,
       "step": 2109
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9885676477069664e-05,
-      "loss": 0.9523,
+      "loss": 1.0078,
       "step": 2110
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9885498812865474e-05,
-      "loss": 0.8718,
+      "loss": 0.9507,
       "step": 2111
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988532101151402e-05,
-      "loss": 1.0319,
+      "loss": 1.0989,
       "step": 2112
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9885143073017755e-05,
-      "loss": 0.796,
+      "loss": 0.8833,
       "step": 2113
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9884964997379163e-05,
-      "loss": 0.8437,
+      "loss": 0.9947,
       "step": 2114
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98847867846007e-05,
-      "loss": 0.8911,
+      "loss": 0.9409,
       "step": 2115
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988460843468485e-05,
-      "loss": 0.7871,
+      "loss": 0.9166,
       "step": 2116
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9884429947634083e-05,
-      "loss": 0.8,
+      "loss": 0.8891,
       "step": 2117
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988425132345087e-05,
-      "loss": 0.9292,
+      "loss": 1.0511,
       "step": 2118
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9884072562137696e-05,
-      "loss": 0.9737,
+      "loss": 1.0026,
       "step": 2119
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9883893663697042e-05,
-      "loss": 0.951,
+      "loss": 1.0585,
       "step": 2120
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9883714628131382e-05,
-      "loss": 0.8702,
+      "loss": 0.9792,
       "step": 2121
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9883535455443205e-05,
-      "loss": 0.8799,
+      "loss": 0.9934,
       "step": 2122
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9883356145635e-05,
-      "loss": 0.9583,
+      "loss": 1.0467,
       "step": 2123
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988317669870925e-05,
-      "loss": 0.8455,
+      "loss": 0.9319,
       "step": 2124
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988299711466844e-05,
-      "loss": 0.8598,
+      "loss": 0.9777,
       "step": 2125
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9882817393515073e-05,
-      "loss": 0.9926,
+      "loss": 1.0774,
       "step": 2126
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9882637535251635e-05,
-      "loss": 0.8123,
+      "loss": 0.8596,
       "step": 2127
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9882457539880627e-05,
-      "loss": 1.0811,
+      "loss": 1.0957,
       "step": 2128
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9882277407404536e-05,
-      "loss": 0.7883,
+      "loss": 0.9573,
       "step": 2129
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988209713782587e-05,
-      "loss": 0.9747,
+      "loss": 1.0268,
       "step": 2130
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9881916731147125e-05,
-      "loss": 0.7837,
+      "loss": 0.917,
       "step": 2131
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988173618737081e-05,
-      "loss": 0.8815,
+      "loss": 0.9949,
       "step": 2132
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988155550649942e-05,
-      "loss": 0.958,
+      "loss": 1.0059,
       "step": 2133
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9881374688535473e-05,
-      "loss": 0.7895,
+      "loss": 0.9491,
       "step": 2134
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9881193733481468e-05,
-      "loss": 0.8565,
+      "loss": 0.9603,
       "step": 2135
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9881012641339925e-05,
-      "loss": 0.9141,
+      "loss": 0.9922,
       "step": 2136
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9880831412113346e-05,
-      "loss": 0.9955,
+      "loss": 1.0038,
       "step": 2137
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9880650045804253e-05,
-      "loss": 0.9382,
+      "loss": 1.1061,
       "step": 2138
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9880468542415156e-05,
-      "loss": 0.856,
+      "loss": 1.0186,
       "step": 2139
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988028690194858e-05,
-      "loss": 0.8697,
+      "loss": 0.9379,
       "step": 2140
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.988010512440704e-05,
-      "loss": 0.8521,
+      "loss": 0.9546,
       "step": 2141
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9879923209793058e-05,
-      "loss": 0.9045,
+      "loss": 0.9216,
       "step": 2142
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987974115810916e-05,
-      "loss": 0.9619,
+      "loss": 1.0508,
       "step": 2143
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987955896935787e-05,
-      "loss": 0.9641,
+      "loss": 1.0941,
       "step": 2144
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987937664354172e-05,
-      "loss": 0.8139,
+      "loss": 0.9416,
       "step": 2145
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9879194180663232e-05,
-      "loss": 0.9788,
+      "loss": 1.042,
       "step": 2146
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9879011580724946e-05,
-      "loss": 0.8952,
+      "loss": 0.9086,
       "step": 2147
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9878828843729387e-05,
-      "loss": 0.9565,
+      "loss": 1.0317,
       "step": 2148
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9878645969679096e-05,
-      "loss": 0.9258,
+      "loss": 1.0191,
       "step": 2149
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9878462958576608e-05,
-      "loss": 0.847,
+      "loss": 0.9638,
       "step": 2150
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9878279810424463e-05,
-      "loss": 0.9218,
+      "loss": 1.0127,
       "step": 2151
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9878096525225196e-05,
-      "loss": 0.9182,
+      "loss": 0.9447,
       "step": 2152
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987791310298136e-05,
-      "loss": 0.9615,
+      "loss": 1.0661,
       "step": 2153
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9877729543695494e-05,
-      "loss": 0.9284,
+      "loss": 1.0308,
       "step": 2154
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9877545847370145e-05,
-      "loss": 0.8542,
+      "loss": 0.9794,
       "step": 2155
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987736201400786e-05,
-      "loss": 0.9592,
+      "loss": 1.0215,
       "step": 2156
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9877178043611195e-05,
-      "loss": 0.8318,
+      "loss": 1.0193,
       "step": 2157
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9876993936182694e-05,
-      "loss": 0.8944,
+      "loss": 1.0383,
       "step": 2158
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987680969172492e-05,
-      "loss": 0.8001,
+      "loss": 1.0489,
       "step": 2159
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987662531024042e-05,
-      "loss": 0.857,
+      "loss": 0.9523,
       "step": 2160
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9876440791731763e-05,
-      "loss": 0.8781,
+      "loss": 0.9899,
       "step": 2161
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9876256136201496e-05,
-      "loss": 0.9278,
+      "loss": 1.0469,
       "step": 2162
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9876071343652192e-05,
-      "loss": 0.9554,
+      "loss": 1.0506,
       "step": 2163
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987588641408641e-05,
-      "loss": 0.721,
+      "loss": 0.8733,
       "step": 2164
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9875701347506715e-05,
-      "loss": 0.854,
+      "loss": 0.9466,
       "step": 2165
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987551614391568e-05,
-      "loss": 0.9743,
+      "loss": 1.048,
       "step": 2166
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9875330803315864e-05,
-      "loss": 0.8701,
+      "loss": 0.923,
       "step": 2167
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9875145325709845e-05,
-      "loss": 0.7975,
+      "loss": 0.9289,
       "step": 2168
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98749597111002e-05,
-      "loss": 0.9442,
+      "loss": 1.0101,
       "step": 2169
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9874773959489494e-05,
-      "loss": 0.809,
+      "loss": 0.9596,
       "step": 2170
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9874588070880315e-05,
-      "loss": 0.9059,
+      "loss": 0.9971,
       "step": 2171
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9874402045275235e-05,
-      "loss": 0.8361,
+      "loss": 0.9324,
       "step": 2172
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9874215882676834e-05,
-      "loss": 0.9202,
+      "loss": 1.0111,
       "step": 2173
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98740295830877e-05,
-      "loss": 0.9745,
+      "loss": 0.988,
       "step": 2174
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9873843146510415e-05,
-      "loss": 0.8035,
+      "loss": 1.0079,
       "step": 2175
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9873656572947565e-05,
-      "loss": 0.9051,
+      "loss": 1.0103,
       "step": 2176
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9873469862401737e-05,
-      "loss": 0.9156,
+      "loss": 1.0531,
       "step": 2177
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9873283014875527e-05,
-      "loss": 0.9006,
+      "loss": 0.9523,
       "step": 2178
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9873096030371522e-05,
-      "loss": 0.9287,
+      "loss": 1.0208,
       "step": 2179
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9872908908892315e-05,
-      "loss": 0.8891,
+      "loss": 0.9971,
       "step": 2180
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9872721650440508e-05,
-      "loss": 0.8918,
+      "loss": 0.9891,
       "step": 2181
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9872534255018692e-05,
-      "loss": 0.8043,
+      "loss": 0.9492,
       "step": 2182
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9872346722629478e-05,
-      "loss": 0.8029,
+      "loss": 0.878,
       "step": 2183
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9872159053275452e-05,
-      "loss": 0.9363,
+      "loss": 0.9659,
       "step": 2184
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9871971246959228e-05,
-      "loss": 0.8644,
+      "loss": 1.0214,
       "step": 2185
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9871783303683414e-05,
-      "loss": 1.0176,
+      "loss": 1.0957,
       "step": 2186
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987159522345061e-05,
-      "loss": 0.8999,
+      "loss": 1.0506,
       "step": 2187
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9871407006263426e-05,
-      "loss": 0.7168,
+      "loss": 0.8422,
       "step": 2188
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987121865212448e-05,
-      "loss": 0.6797,
+      "loss": 0.8605,
       "step": 2189
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9871030161036375e-05,
-      "loss": 0.9497,
+      "loss": 0.997,
       "step": 2190
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9870841533001735e-05,
-      "loss": 0.9729,
+      "loss": 1.0596,
       "step": 2191
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9870652768023172e-05,
-      "loss": 0.9397,
+      "loss": 0.9933,
       "step": 2192
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987046386610331e-05,
-      "loss": 0.8987,
+      "loss": 0.9861,
       "step": 2193
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.987027482724476e-05,
-      "loss": 0.8595,
+      "loss": 1.0742,
       "step": 2194
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9870085651450157e-05,
-      "loss": 0.7258,
+      "loss": 0.9028,
       "step": 2195
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9869896338722116e-05,
-      "loss": 0.9086,
+      "loss": 1.0027,
       "step": 2196
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9869706889063268e-05,
-      "loss": 0.8216,
+      "loss": 1.0059,
       "step": 2197
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986951730247624e-05,
-      "loss": 0.914,
+      "loss": 0.9984,
       "step": 2198
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9869327578963666e-05,
-      "loss": 0.962,
+      "loss": 1.0753,
       "step": 2199
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986913771852817e-05,
-      "loss": 0.9941,
+      "loss": 1.0854,
       "step": 2200
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986894772117239e-05,
-      "loss": 0.8347,
+      "loss": 0.9087,
       "step": 2201
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9868757586898966e-05,
-      "loss": 0.9259,
+      "loss": 1.0026,
       "step": 2202
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9868567315710533e-05,
-      "loss": 1.0009,
+      "loss": 1.0466,
       "step": 2203
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9868376907609727e-05,
-      "loss": 0.7461,
+      "loss": 0.9296,
       "step": 2204
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986818636259919e-05,
-      "loss": 0.7513,
+      "loss": 0.9237,
       "step": 2205
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9867995680681575e-05,
-      "loss": 0.8736,
+      "loss": 0.9423,
       "step": 2206
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9867804861859518e-05,
-      "loss": 0.8123,
+      "loss": 1.0085,
       "step": 2207
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986761390613567e-05,
-      "loss": 0.9567,
+      "loss": 1.0472,
       "step": 2208
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9867422813512678e-05,
-      "loss": 0.8778,
+      "loss": 0.9949,
       "step": 2209
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9867231583993194e-05,
-      "loss": 0.9024,
+      "loss": 0.8902,
       "step": 2210
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986704021757987e-05,
-      "loss": 0.9457,
+      "loss": 0.9735,
       "step": 2211
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9866848714275368e-05,
-      "loss": 1.0334,
+      "loss": 1.0391,
       "step": 2212
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9866657074082335e-05,
-      "loss": 0.9188,
+      "loss": 1.0052,
       "step": 2213
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9866465297003437e-05,
-      "loss": 0.8694,
+      "loss": 0.967,
       "step": 2214
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986627338304133e-05,
-      "loss": 0.9154,
+      "loss": 0.9737,
       "step": 2215
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9866081332198676e-05,
-      "loss": 0.8877,
+      "loss": 1.01,
       "step": 2216
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9865889144478144e-05,
-      "loss": 0.8029,
+      "loss": 0.9314,
       "step": 2217
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98656968198824e-05,
-      "loss": 0.8698,
+      "loss": 1.002,
       "step": 2218
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9865504358414107e-05,
-      "loss": 1.0045,
+      "loss": 1.0253,
       "step": 2219
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986531176007594e-05,
-      "loss": 0.8112,
+      "loss": 1.0301,
       "step": 2220
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986511902487057e-05,
-      "loss": 0.8116,
+      "loss": 0.905,
       "step": 2221
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986492615280067e-05,
-      "loss": 0.8345,
+      "loss": 0.9084,
       "step": 2222
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9864733143868912e-05,
-      "loss": 0.8748,
+      "loss": 1.038,
       "step": 2223
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9864539998077983e-05,
-      "loss": 0.9871,
+      "loss": 1.0476,
       "step": 2224
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9864346715430554e-05,
-      "loss": 0.7504,
+      "loss": 0.8556,
       "step": 2225
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9864153295929314e-05,
-      "loss": 0.8207,
+      "loss": 0.9942,
       "step": 2226
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9863959739576937e-05,
-      "loss": 0.8186,
+      "loss": 0.9725,
       "step": 2227
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986376604637612e-05,
-      "loss": 0.9229,
+      "loss": 1.0476,
       "step": 2228
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986357221632954e-05,
-      "loss": 1.1085,
+      "loss": 1.2073,
       "step": 2229
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986337824943989e-05,
-      "loss": 1.0238,
+      "loss": 1.0776,
       "step": 2230
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9863184145709863e-05,
-      "loss": 0.7888,
+      "loss": 0.9167,
       "step": 2231
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9862989905142148e-05,
-      "loss": 1.0094,
+      "loss": 1.043,
       "step": 2232
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986279552773944e-05,
-      "loss": 0.8826,
+      "loss": 0.9178,
       "step": 2233
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9862601013504444e-05,
-      "loss": 0.8482,
+      "loss": 0.972,
       "step": 2234
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9862406362439846e-05,
-      "loss": 0.7426,
+      "loss": 0.9034,
       "step": 2235
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9862211574548355e-05,
-      "loss": 0.8699,
+      "loss": 1.0057,
       "step": 2236
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986201664983267e-05,
-      "loss": 0.8148,
+      "loss": 0.9209,
       "step": 2237
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98618215882955e-05,
-      "loss": 0.9881,
+      "loss": 1.1102,
       "step": 2238
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9861626389939546e-05,
-      "loss": 0.8879,
+      "loss": 0.9847,
       "step": 2239
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9861431054767518e-05,
-      "loss": 0.966,
+      "loss": 0.9386,
       "step": 2240
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986123558278212e-05,
-      "loss": 0.7115,
+      "loss": 0.8025,
       "step": 2241
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9861039973986074e-05,
-      "loss": 0.902,
+      "loss": 0.9502,
       "step": 2242
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986084422838209e-05,
-      "loss": 1.0078,
+      "loss": 1.0782,
       "step": 2243
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9860648345972883e-05,
-      "loss": 0.8632,
+      "loss": 0.9712,
       "step": 2244
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986045232676117e-05,
-      "loss": 0.9724,
+      "loss": 1.0213,
       "step": 2245
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.986025617074967e-05,
-      "loss": 0.8676,
+      "loss": 0.9297,
       "step": 2246
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9860059877941106e-05,
-      "loss": 0.9136,
+      "loss": 1.1049,
       "step": 2247
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9859863448338203e-05,
-      "loss": 0.8918,
+      "loss": 1.0162,
       "step": 2248
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9859666881943684e-05,
-      "loss": 0.9481,
+      "loss": 1.0186,
       "step": 2249
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9859470178760272e-05,
-      "loss": 0.8341,
+      "loss": 0.8907,
       "step": 2250
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9859273338790703e-05,
-      "loss": 0.8478,
+      "loss": 0.9866,
       "step": 2251
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9859076362037703e-05,
-      "loss": 0.9295,
+      "loss": 1.0463,
       "step": 2252
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985887924850401e-05,
-      "loss": 0.8305,
+      "loss": 0.946,
       "step": 2253
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985868199819235e-05,
-      "loss": 0.8557,
+      "loss": 0.8965,
       "step": 2254
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985848461110547e-05,
-      "loss": 1.1018,
+      "loss": 1.1488,
       "step": 2255
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98582870872461e-05,
-      "loss": 0.9375,
+      "loss": 0.9916,
       "step": 2256
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9858089426616986e-05,
-      "loss": 0.9418,
+      "loss": 1.0203,
       "step": 2257
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9857891629220868e-05,
-      "loss": 0.8564,
+      "loss": 1.0006,
       "step": 2258
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985769369506049e-05,
-      "loss": 0.8781,
+      "loss": 0.9648,
       "step": 2259
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98574956241386e-05,
-      "loss": 0.8886,
+      "loss": 1.0233,
       "step": 2260
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985729741645794e-05,
-      "loss": 0.9254,
+      "loss": 0.9983,
       "step": 2261
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9857099072021264e-05,
-      "loss": 0.9121,
+      "loss": 1.0257,
       "step": 2262
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985690059083133e-05,
-      "loss": 0.9194,
+      "loss": 1.0118,
       "step": 2263
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9856701972890878e-05,
-      "loss": 0.844,
+      "loss": 0.9689,
       "step": 2264
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9856503218202674e-05,
-      "loss": 0.8688,
+      "loss": 0.968,
       "step": 2265
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9856304326769474e-05,
-      "loss": 0.95,
+      "loss": 1.0534,
       "step": 2266
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9856105298594033e-05,
-      "loss": 1.0212,
+      "loss": 1.0195,
       "step": 2267
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9855906133679116e-05,
-      "loss": 0.9553,
+      "loss": 1.0315,
       "step": 2268
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9855706832027484e-05,
-      "loss": 0.8202,
+      "loss": 1.0005,
       "step": 2269
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9855507393641905e-05,
-      "loss": 0.9306,
+      "loss": 1.0691,
       "step": 2270
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9855307818525144e-05,
-      "loss": 1.1019,
+      "loss": 1.1006,
       "step": 2271
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985510810667997e-05,
-      "loss": 0.9939,
+      "loss": 1.0497,
       "step": 2272
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9854908258109154e-05,
-      "loss": 0.7909,
+      "loss": 0.9651,
       "step": 2273
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9854708272815465e-05,
-      "loss": 0.8118,
+      "loss": 1.0002,
       "step": 2274
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9854508150801683e-05,
-      "loss": 0.8986,
+      "loss": 1.0114,
       "step": 2275
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985430789207058e-05,
-      "loss": 0.9862,
+      "loss": 0.9633,
       "step": 2276
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985410749662494e-05,
-      "loss": 0.7268,
+      "loss": 0.8462,
       "step": 2277
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985390696446754e-05,
-      "loss": 0.8863,
+      "loss": 1.0233,
       "step": 2278
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985370629560116e-05,
-      "loss": 0.9714,
+      "loss": 1.0493,
       "step": 2279
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9853505490028583e-05,
-      "loss": 0.7772,
+      "loss": 0.9871,
       "step": 2280
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98533045477526e-05,
-      "loss": 0.843,
+      "loss": 0.9409,
       "step": 2281
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9853103468775998e-05,
-      "loss": 0.8455,
+      "loss": 1.0031,
       "step": 2282
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9852902253101562e-05,
-      "loss": 1.0244,
+      "loss": 1.0927,
       "step": 2283
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985270090073209e-05,
-      "loss": 0.8549,
+      "loss": 0.9751,
       "step": 2284
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985249941167037e-05,
-      "loss": 0.861,
+      "loss": 0.9928,
       "step": 2285
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98522977859192e-05,
-      "loss": 0.807,
+      "loss": 0.9612,
       "step": 2286
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9852096023481374e-05,
-      "loss": 0.8046,
+      "loss": 0.9349,
       "step": 2287
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98518941243597e-05,
-      "loss": 0.9581,
+      "loss": 1.0278,
       "step": 2288
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985169208855697e-05,
-      "loss": 0.9428,
+      "loss": 1.0054,
       "step": 2289
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9851489916075986e-05,
-      "loss": 1.0233,
+      "loss": 1.0505,
       "step": 2290
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9851287606919563e-05,
-      "loss": 0.9863,
+      "loss": 1.0687,
       "step": 2291
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9851085161090497e-05,
-      "loss": 0.9522,
+      "loss": 1.0472,
       "step": 2292
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9850882578591605e-05,
-      "loss": 0.9243,
+      "loss": 0.995,
       "step": 2293
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9850679859425692e-05,
-      "loss": 0.9139,
+      "loss": 1.0179,
       "step": 2294
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9850477003595573e-05,
-      "loss": 0.6947,
+      "loss": 0.9222,
       "step": 2295
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.985027401110406e-05,
-      "loss": 0.8389,
+      "loss": 0.9439,
       "step": 2296
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9850070881953974e-05,
-      "loss": 0.8175,
+      "loss": 0.9666,
       "step": 2297
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9849867616148125e-05,
-      "loss": 0.9111,
+      "loss": 1.0217,
       "step": 2298
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9849664213689342e-05,
-      "loss": 0.9232,
+      "loss": 1.0481,
       "step": 2299
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9849460674580442e-05,
-      "loss": 0.9549,
+      "loss": 1.0015,
       "step": 2300
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984925699882425e-05,
-      "loss": 0.7793,
+      "loss": 0.9377,
       "step": 2301
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984905318642359e-05,
-      "loss": 0.917,
+      "loss": 1.0165,
       "step": 2302
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9848849237381295e-05,
-      "loss": 0.836,
+      "loss": 1.0368,
       "step": 2303
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9848645151700184e-05,
-      "loss": 1.0679,
+      "loss": 1.0956,
       "step": 2304
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.98484409293831e-05,
-      "loss": 0.8814,
+      "loss": 1.0071,
       "step": 2305
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9848236570432867e-05,
-      "loss": 0.9479,
+      "loss": 1.0019,
       "step": 2306
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9848032074852327e-05,
-      "loss": 0.8478,
+      "loss": 1.071,
       "step": 2307
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9847827442644318e-05,
-      "loss": 0.8223,
+      "loss": 0.908,
       "step": 2308
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984762267381167e-05,
-      "loss": 0.8373,
+      "loss": 1.0027,
       "step": 2309
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9847417768357233e-05,
-      "loss": 0.872,
+      "loss": 1.0088,
       "step": 2310
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984721272628384e-05,
-      "loss": 0.8784,
+      "loss": 1.0121,
       "step": 2311
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9847007547594346e-05,
-      "loss": 0.8944,
+      "loss": 0.9952,
       "step": 2312
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9846802232291593e-05,
-      "loss": 0.8606,
+      "loss": 0.9369,
       "step": 2313
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984659678037843e-05,
-      "loss": 0.9246,
+      "loss": 1.0934,
       "step": 2314
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9846391191857708e-05,
-      "loss": 0.9069,
+      "loss": 1.0474,
       "step": 2315
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984618546673228e-05,
-      "loss": 0.8667,
+      "loss": 0.9492,
       "step": 2316
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9845979605004997e-05,
-      "loss": 0.8583,
+      "loss": 0.9831,
       "step": 2317
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9845773606678714e-05,
-      "loss": 0.904,
+      "loss": 0.9683,
       "step": 2318
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9845567471756296e-05,
-      "loss": 0.8896,
+      "loss": 0.9782,
       "step": 2319
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9845361200240593e-05,
-      "loss": 0.8763,
+      "loss": 0.9181,
       "step": 2320
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9845154792134477e-05,
-      "loss": 0.8627,
+      "loss": 0.9663,
       "step": 2321
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9844948247440802e-05,
-      "loss": 0.8077,
+      "loss": 0.9738,
       "step": 2322
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984474156616244e-05,
-      "loss": 1.0076,
+      "loss": 1.0425,
       "step": 2323
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984453474830226e-05,
-      "loss": 0.847,
+      "loss": 1.0257,
       "step": 2324
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9844327793863127e-05,
-      "loss": 0.7639,
+      "loss": 0.8125,
       "step": 2325
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984412070284791e-05,
-      "loss": 0.8458,
+      "loss": 0.9803,
       "step": 2326
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9843913475259487e-05,
-      "loss": 0.8439,
+      "loss": 0.9908,
       "step": 2327
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9843706111100732e-05,
-      "loss": 0.8612,
+      "loss": 1.0012,
       "step": 2328
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984349861037452e-05,
-      "loss": 0.8495,
+      "loss": 0.9269,
       "step": 2329
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9843290973083728e-05,
-      "loss": 0.8671,
+      "loss": 0.9746,
       "step": 2330
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9843083199231243e-05,
-      "loss": 0.9575,
+      "loss": 1.0191,
       "step": 2331
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9842875288819943e-05,
-      "loss": 0.8641,
+      "loss": 0.9574,
       "step": 2332
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9842667241852716e-05,
-      "loss": 0.8563,
+      "loss": 0.8914,
       "step": 2333
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9842459058332445e-05,
-      "loss": 0.8074,
+      "loss": 0.931,
       "step": 2334
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9842250738262016e-05,
-      "loss": 0.8843,
+      "loss": 1.0034,
       "step": 2335
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9842042281644326e-05,
-      "loss": 0.8203,
+      "loss": 1.0026,
       "step": 2336
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.984183368848226e-05,
-      "loss": 0.9824,
+      "loss": 1.0074,
       "step": 2337
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9841624958778716e-05,
-      "loss": 0.9045,
+      "loss": 1.039,
       "step": 2338
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9841416092536594e-05,
-      "loss": 0.9577,
+      "loss": 1.0642,
       "step": 2339
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9841207089758785e-05,
-      "loss": 0.8182,
+      "loss": 0.9577,
       "step": 2340
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9840997950448186e-05,
-      "loss": 0.8348,
+      "loss": 0.913,
       "step": 2341
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9840788674607707e-05,
-      "loss": 0.8778,
+      "loss": 0.9572,
       "step": 2342
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9840579262240245e-05,
-      "loss": 0.9263,
+      "loss": 1.0193,
       "step": 2343
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.984036971334871e-05,
-      "loss": 0.7895,
+      "loss": 0.9587,
       "step": 2344
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9840160027936005e-05,
-      "loss": 0.953,
+      "loss": 0.9919,
       "step": 2345
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9839950206005043e-05,
-      "loss": 0.9075,
+      "loss": 1.0037,
       "step": 2346
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9839740247558733e-05,
-      "loss": 0.8577,
+      "loss": 1.0033,
       "step": 2347
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9839530152599983e-05,
-      "loss": 0.8855,
+      "loss": 1.0073,
       "step": 2348
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983931992113172e-05,
-      "loss": 0.8462,
+      "loss": 0.9787,
       "step": 2349
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9839109553156847e-05,
-      "loss": 0.7596,
+      "loss": 0.8892,
       "step": 2350
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9838899048678294e-05,
-      "loss": 0.796,
+      "loss": 0.9134,
       "step": 2351
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983868840769897e-05,
-      "loss": 0.9876,
+      "loss": 1.0596,
       "step": 2352
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9838477630221807e-05,
-      "loss": 0.973,
+      "loss": 1.0076,
       "step": 2353
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983826671624973e-05,
-      "loss": 0.9256,
+      "loss": 0.9593,
       "step": 2354
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9838055665785653e-05,
-      "loss": 1.0338,
+      "loss": 1.047,
       "step": 2355
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9837844478832518e-05,
-      "loss": 0.9098,
+      "loss": 1.0027,
       "step": 2356
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9837633155393247e-05,
-      "loss": 0.9397,
+      "loss": 1.0024,
       "step": 2357
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9837421695470772e-05,
-      "loss": 0.9516,
+      "loss": 1.0117,
       "step": 2358
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9837210099068033e-05,
-      "loss": 0.8185,
+      "loss": 0.9583,
       "step": 2359
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9836998366187957e-05,
-      "loss": 0.9036,
+      "loss": 0.9679,
       "step": 2360
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9836786496833485e-05,
-      "loss": 1.0009,
+      "loss": 1.0648,
       "step": 2361
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983657449100756e-05,
-      "loss": 0.863,
+      "loss": 0.9256,
       "step": 2362
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983636234871312e-05,
-      "loss": 0.8052,
+      "loss": 0.9573,
       "step": 2363
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9836150069953105e-05,
-      "loss": 0.77,
+      "loss": 0.9038,
       "step": 2364
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983593765473047e-05,
-      "loss": 0.844,
+      "loss": 1.0073,
       "step": 2365
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9835725103048147e-05,
-      "loss": 0.8253,
+      "loss": 1.0097,
       "step": 2366
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98355124149091e-05,
-      "loss": 1.0154,
+      "loss": 1.0775,
       "step": 2367
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983529959031627e-05,
-      "loss": 0.8268,
+      "loss": 1.0193,
       "step": 2368
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9835086629272612e-05,
-      "loss": 0.8151,
+      "loss": 0.9132,
       "step": 2369
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983487353178108e-05,
-      "loss": 0.975,
+      "loss": 1.0347,
       "step": 2370
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9834660297844634e-05,
-      "loss": 0.9564,
+      "loss": 1.084,
       "step": 2371
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983444692746623e-05,
-      "loss": 1.036,
+      "loss": 1.1229,
       "step": 2372
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983423342064883e-05,
-      "loss": 0.7227,
+      "loss": 0.8894,
       "step": 2373
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983401977739539e-05,
-      "loss": 0.8499,
+      "loss": 0.9722,
       "step": 2374
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983380599770888e-05,
-      "loss": 0.8501,
+      "loss": 0.9117,
       "step": 2375
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9833592081592265e-05,
-      "loss": 0.9125,
+      "loss": 1.0044,
       "step": 2376
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9833378029048512e-05,
-      "loss": 0.8392,
+      "loss": 0.9762,
       "step": 2377
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9833163840080593e-05,
-      "loss": 0.8354,
+      "loss": 0.9686,
       "step": 2378
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983294951469147e-05,
-      "loss": 0.7623,
+      "loss": 0.9708,
       "step": 2379
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9832735052884133e-05,
-      "loss": 0.8666,
+      "loss": 0.9684,
       "step": 2380
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9832520454661538e-05,
-      "loss": 0.8203,
+      "loss": 1.0069,
       "step": 2381
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983230572002668e-05,
-      "loss": 0.8706,
+      "loss": 0.9791,
       "step": 2382
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9832090848982524e-05,
-      "loss": 0.7934,
+      "loss": 0.9087,
       "step": 2383
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9831875841532062e-05,
-      "loss": 0.8216,
+      "loss": 0.9459,
       "step": 2384
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983166069767827e-05,
-      "loss": 0.8683,
+      "loss": 0.9949,
       "step": 2385
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9831445417424136e-05,
-      "loss": 0.7597,
+      "loss": 0.905,
       "step": 2386
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9831230000772646e-05,
-      "loss": 0.93,
+      "loss": 1.0249,
       "step": 2387
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9831014447726788e-05,
-      "loss": 0.8839,
+      "loss": 1.0385,
       "step": 2388
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9830798758289554e-05,
-      "loss": 0.7722,
+      "loss": 0.8465,
       "step": 2389
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9830582932463935e-05,
-      "loss": 0.7598,
+      "loss": 0.9444,
       "step": 2390
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9830366970252927e-05,
-      "loss": 0.8797,
+      "loss": 0.9568,
       "step": 2391
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.983015087165952e-05,
-      "loss": 0.8358,
+      "loss": 0.9806,
       "step": 2392
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9829934636686722e-05,
-      "loss": 0.9815,
+      "loss": 1.035,
       "step": 2393
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9829718265337523e-05,
-      "loss": 0.8775,
+      "loss": 0.9869,
       "step": 2394
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982950175761493e-05,
-      "loss": 0.8464,
+      "loss": 0.9535,
       "step": 2395
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9829285113521953e-05,
-      "loss": 0.9044,
+      "loss": 0.9633,
       "step": 2396
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9829068333061587e-05,
-      "loss": 0.8081,
+      "loss": 0.9279,
       "step": 2397
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982885141623684e-05,
-      "loss": 0.7842,
+      "loss": 0.9734,
       "step": 2398
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982863436305073e-05,
-      "loss": 0.792,
+      "loss": 0.9076,
       "step": 2399
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982841717350626e-05,
-      "loss": 0.7755,
+      "loss": 0.9766,
       "step": 2400
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9828199847606448e-05,
-      "loss": 0.8272,
+      "loss": 1.0238,
       "step": 2401
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9827982385354307e-05,
-      "loss": 0.9006,
+      "loss": 0.9522,
       "step": 2402
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9827764786752853e-05,
-      "loss": 0.7363,
+      "loss": 0.8443,
       "step": 2403
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982754705180511e-05,
-      "loss": 0.7851,
+      "loss": 0.949,
       "step": 2404
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9827329180514093e-05,
-      "loss": 0.8548,
+      "loss": 1.0233,
       "step": 2405
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9827111172882828e-05,
-      "loss": 0.9072,
+      "loss": 1.0382,
       "step": 2406
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9826893028914336e-05,
-      "loss": 0.8424,
+      "loss": 0.9604,
       "step": 2407
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9826674748611646e-05,
-      "loss": 0.9132,
+      "loss": 0.9987,
       "step": 2408
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9826456331977786e-05,
-      "loss": 1.0301,
+      "loss": 1.1356,
       "step": 2409
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982623777901579e-05,
-      "loss": 0.7788,
+      "loss": 0.9194,
       "step": 2410
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9826019089728682e-05,
-      "loss": 0.9865,
+      "loss": 1.0316,
       "step": 2411
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9825800264119502e-05,
-      "loss": 0.7994,
+      "loss": 0.9663,
       "step": 2412
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9825581302191286e-05,
-      "loss": 0.7942,
+      "loss": 0.9753,
       "step": 2413
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9825362203947066e-05,
-      "loss": 0.7163,
+      "loss": 0.9095,
       "step": 2414
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982514296938989e-05,
-      "loss": 0.8842,
+      "loss": 0.9755,
       "step": 2415
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9824923598522795e-05,
-      "loss": 0.8176,
+      "loss": 0.9972,
       "step": 2416
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982470409134882e-05,
-      "loss": 1.0054,
+      "loss": 0.9831,
       "step": 2417
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982448444787102e-05,
-      "loss": 1.0133,
+      "loss": 1.0077,
       "step": 2418
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9824264668092437e-05,
-      "loss": 0.8494,
+      "loss": 1.0199,
       "step": 2419
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9824044752016117e-05,
-      "loss": 0.8567,
+      "loss": 0.9062,
       "step": 2420
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982382469964512e-05,
-      "loss": 0.8632,
+      "loss": 0.9828,
       "step": 2421
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982360451098249e-05,
-      "loss": 0.854,
+      "loss": 1.0075,
       "step": 2422
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982338418603129e-05,
-      "loss": 0.9616,
+      "loss": 1.0497,
       "step": 2423
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9823163724794562e-05,
-      "loss": 0.824,
+      "loss": 0.8495,
       "step": 2424
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982294312727538e-05,
-      "loss": 0.9422,
+      "loss": 1.0521,
       "step": 2425
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9822722393476802e-05,
-      "loss": 0.8314,
+      "loss": 0.9648,
       "step": 2426
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9822501523401883e-05,
-      "loss": 0.8032,
+      "loss": 0.935,
       "step": 2427
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9822280517053695e-05,
-      "loss": 0.7801,
+      "loss": 0.9177,
       "step": 2428
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98220593744353e-05,
-      "loss": 0.8185,
+      "loss": 1.0036,
       "step": 2429
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9821838095549766e-05,
-      "loss": 0.9548,
+      "loss": 1.0344,
       "step": 2430
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9821616680400164e-05,
-      "loss": 0.9622,
+      "loss": 1.0353,
       "step": 2431
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9821395128989565e-05,
-      "loss": 0.9132,
+      "loss": 1.0134,
       "step": 2432
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9821173441321044e-05,
-      "loss": 0.8322,
+      "loss": 1.0167,
       "step": 2433
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982095161739768e-05,
-      "loss": 0.9136,
+      "loss": 0.9402,
       "step": 2434
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9820729657222538e-05,
-      "loss": 0.798,
+      "loss": 0.9428,
       "step": 2435
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982050756079871e-05,
-      "loss": 0.9043,
+      "loss": 0.9792,
       "step": 2436
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9820285328129274e-05,
-      "loss": 0.8285,
+      "loss": 0.916,
       "step": 2437
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.982006295921731e-05,
-      "loss": 0.9307,
+      "loss": 1.004,
       "step": 2438
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981984045406591e-05,
-      "loss": 0.8955,
+      "loss": 1.0319,
       "step": 2439
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981961781267815e-05,
-      "loss": 1.0129,
+      "loss": 1.0409,
       "step": 2440
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981939503505713e-05,
-      "loss": 0.7992,
+      "loss": 0.9839,
       "step": 2441
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9819172121205934e-05,
-      "loss": 0.7821,
+      "loss": 0.9075,
       "step": 2442
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9818949071127655e-05,
-      "loss": 0.8135,
+      "loss": 0.9064,
       "step": 2443
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9818725884825388e-05,
-      "loss": 0.7795,
+      "loss": 0.9153,
       "step": 2444
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9818502562302235e-05,
-      "loss": 0.7763,
+      "loss": 0.9781,
       "step": 2445
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9818279103561286e-05,
-      "loss": 0.8557,
+      "loss": 0.9505,
       "step": 2446
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9818055508605646e-05,
-      "loss": 0.7401,
+      "loss": 0.8944,
       "step": 2447
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9817831777438416e-05,
-      "loss": 0.7829,
+      "loss": 0.9196,
       "step": 2448
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98176079100627e-05,
-      "loss": 0.9319,
+      "loss": 0.9862,
       "step": 2449
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9817383906481604e-05,
-      "loss": 0.8939,
+      "loss": 1.0508,
       "step": 2450
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9817159766698234e-05,
-      "loss": 0.9657,
+      "loss": 1.0343,
       "step": 2451
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98169354907157e-05,
-      "loss": 0.9553,
+      "loss": 1.0092,
       "step": 2452
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9816711078537115e-05,
-      "loss": 0.9737,
+      "loss": 0.952,
       "step": 2453
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9816486530165593e-05,
-      "loss": 0.8064,
+      "loss": 0.9986,
       "step": 2454
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981626184560425e-05,
-      "loss": 0.8367,
+      "loss": 0.9266,
       "step": 2455
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9816037024856198e-05,
-      "loss": 0.8501,
+      "loss": 0.9641,
       "step": 2456
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9815812067924562e-05,
-      "loss": 0.8518,
+      "loss": 0.9938,
       "step": 2457
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9815586974812454e-05,
-      "loss": 0.8128,
+      "loss": 1.0389,
       "step": 2458
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981536174552301e-05,
-      "loss": 0.8477,
+      "loss": 0.9688,
       "step": 2459
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9815136380059344e-05,
-      "loss": 0.9278,
+      "loss": 0.9634,
       "step": 2460
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981491087842459e-05,
-      "loss": 0.773,
+      "loss": 0.9191,
       "step": 2461
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9814685240621875e-05,
-      "loss": 0.8031,
+      "loss": 0.9224,
       "step": 2462
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9814459466654323e-05,
-      "loss": 0.83,
+      "loss": 0.9662,
       "step": 2463
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9814233556525072e-05,
-      "loss": 0.8272,
+      "loss": 0.9866,
       "step": 2464
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9814007510237255e-05,
-      "loss": 0.7668,
+      "loss": 0.866,
       "step": 2465
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9813781327794007e-05,
-      "loss": 0.8415,
+      "loss": 0.9438,
       "step": 2466
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981355500919847e-05,
-      "loss": 0.8438,
+      "loss": 0.9833,
       "step": 2467
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981332855445378e-05,
-      "loss": 1.0029,
+      "loss": 1.0001,
       "step": 2468
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9813101963563076e-05,
-      "loss": 0.8815,
+      "loss": 1.001,
       "step": 2469
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9812875236529508e-05,
-      "loss": 0.8846,
+      "loss": 0.9375,
       "step": 2470
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9812648373356218e-05,
-      "loss": 0.8271,
+      "loss": 0.9407,
       "step": 2471
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9812421374046356e-05,
-      "loss": 0.8383,
+      "loss": 0.9331,
       "step": 2472
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981219423860307e-05,
-      "loss": 0.9692,
+      "loss": 1.0062,
       "step": 2473
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981196696702951e-05,
-      "loss": 0.9534,
+      "loss": 1.0526,
       "step": 2474
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9811739559328824e-05,
-      "loss": 1.0014,
+      "loss": 1.0722,
       "step": 2475
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9811512015504177e-05,
-      "loss": 0.9029,
+      "loss": 1.0181,
       "step": 2476
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9811284335558723e-05,
-      "loss": 0.9352,
+      "loss": 1.001,
       "step": 2477
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9811056519495617e-05,
-      "loss": 1.0355,
+      "loss": 1.0741,
       "step": 2478
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9810828567318023e-05,
-      "loss": 0.8842,
+      "loss": 1.002,
       "step": 2479
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9810600479029102e-05,
-      "loss": 0.8573,
+      "loss": 1.0316,
       "step": 2480
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981037225463202e-05,
-      "loss": 0.9502,
+      "loss": 1.0198,
       "step": 2481
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.981014389412994e-05,
-      "loss": 0.8658,
+      "loss": 1.0018,
       "step": 2482
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9809915397526033e-05,
-      "loss": 0.8874,
+      "loss": 1.0151,
       "step": 2483
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980968676482347e-05,
-      "loss": 0.854,
+      "loss": 0.916,
       "step": 2484
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980945799602542e-05,
-      "loss": 0.8631,
+      "loss": 0.9905,
       "step": 2485
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9809229091135056e-05,
-      "loss": 0.7945,
+      "loss": 0.9784,
       "step": 2486
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9809000050155558e-05,
-      "loss": 0.8294,
+      "loss": 0.9893,
       "step": 2487
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98087708730901e-05,
-      "loss": 0.7189,
+      "loss": 0.8877,
       "step": 2488
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9808541559941864e-05,
-      "loss": 0.7989,
+      "loss": 0.9632,
       "step": 2489
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980831211071403e-05,
-      "loss": 0.9396,
+      "loss": 1.0042,
       "step": 2490
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9808082525409782e-05,
-      "loss": 0.8502,
+      "loss": 1.0363,
       "step": 2491
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9807852804032306e-05,
-      "loss": 0.7989,
+      "loss": 0.9034,
       "step": 2492
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9807622946584787e-05,
-      "loss": 0.8221,
+      "loss": 0.9527,
       "step": 2493
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9807392953070414e-05,
-      "loss": 0.8272,
+      "loss": 0.9771,
       "step": 2494
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9807162823492382e-05,
-      "loss": 0.8402,
+      "loss": 0.9867,
       "step": 2495
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9806932557853876e-05,
-      "loss": 0.9344,
+      "loss": 0.9442,
       "step": 2496
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9806702156158096e-05,
-      "loss": 0.8468,
+      "loss": 1.0342,
       "step": 2497
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980647161840824e-05,
-      "loss": 0.8858,
+      "loss": 1.0151,
       "step": 2498
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.98062409446075e-05,
-      "loss": 0.86,
+      "loss": 1.0061,
       "step": 2499
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9806010134759083e-05,
-      "loss": 0.917,
+      "loss": 1.0141,
       "step": 2500
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9805779188866187e-05,
-      "loss": 0.9386,
+      "loss": 1.021,
       "step": 2501
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9805548106932018e-05,
-      "loss": 0.835,
+      "loss": 0.9457,
       "step": 2502
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980531688895978e-05,
-      "loss": 0.8264,
+      "loss": 0.9515,
       "step": 2503
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9805085534952684e-05,
-      "loss": 1.0554,
+      "loss": 1.087,
       "step": 2504
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9804854044913934e-05,
-      "loss": 0.8914,
+      "loss": 1.0165,
       "step": 2505
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9804622418846747e-05,
-      "loss": 0.8507,
+      "loss": 1.0075,
       "step": 2506
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9804390656754337e-05,
-      "loss": 0.7761,
+      "loss": 0.9318,
       "step": 2507
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9804158758639917e-05,
-      "loss": 0.8603,
+      "loss": 0.9545,
       "step": 2508
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9803926724506707e-05,
-      "loss": 0.8836,
+      "loss": 0.9327,
       "step": 2509
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980369455435792e-05,
-      "loss": 0.9532,
+      "loss": 1.0011,
       "step": 2510
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980346224819678e-05,
-      "loss": 0.9021,
+      "loss": 1.0466,
       "step": 2511
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980322980602651e-05,
-      "loss": 0.9518,
+      "loss": 1.0583,
       "step": 2512
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9802997227850337e-05,
-      "loss": 0.8215,
+      "loss": 0.9978,
       "step": 2513
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9802764513671487e-05,
-      "loss": 0.8266,
+      "loss": 0.9116,
       "step": 2514
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9802531663493187e-05,
-      "loss": 0.8177,
+      "loss": 0.9906,
       "step": 2515
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980229867731867e-05,
-      "loss": 0.7878,
+      "loss": 1.0101,
       "step": 2516
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9802065555151163e-05,
-      "loss": 0.8277,
+      "loss": 0.9616,
       "step": 2517
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9801832296993907e-05,
-      "loss": 0.8217,
+      "loss": 0.9387,
       "step": 2518
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9801598902850133e-05,
-      "loss": 0.7923,
+      "loss": 0.8441,
       "step": 2519
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980136537272308e-05,
-      "loss": 0.9255,
+      "loss": 1.012,
       "step": 2520
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980113170661599e-05,
-      "loss": 0.9493,
+      "loss": 0.968,
       "step": 2521
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9800897904532102e-05,
-      "loss": 0.8721,
+      "loss": 1.0649,
       "step": 2522
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9800663966474666e-05,
-      "loss": 0.8103,
+      "loss": 0.8997,
       "step": 2523
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9800429892446917e-05,
-      "loss": 0.8743,
+      "loss": 1.0323,
       "step": 2524
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.980019568245211e-05,
-      "loss": 0.7935,
+      "loss": 0.8414,
       "step": 2525
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9799961336493495e-05,
-      "loss": 0.8271,
+      "loss": 0.9386,
       "step": 2526
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9799726854574316e-05,
-      "loss": 0.8547,
+      "loss": 0.9811,
       "step": 2527
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979949223669784e-05,
-      "loss": 0.8523,
+      "loss": 1.0214,
       "step": 2528
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97992574828673e-05,
-      "loss": 1.0018,
+      "loss": 1.1013,
       "step": 2529
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9799022593085976e-05,
-      "loss": 0.7501,
+      "loss": 0.9542,
       "step": 2530
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979878756735711e-05,
-      "loss": 0.9198,
+      "loss": 1.0008,
       "step": 2531
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9798552405683976e-05,
-      "loss": 0.8057,
+      "loss": 0.9651,
       "step": 2532
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9798317108069823e-05,
-      "loss": 0.9111,
+      "loss": 0.9677,
       "step": 2533
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9798081674517927e-05,
-      "loss": 0.8953,
+      "loss": 0.9852,
       "step": 2534
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9797846105031545e-05,
-      "loss": 0.7818,
+      "loss": 0.9089,
       "step": 2535
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9797610399613955e-05,
-      "loss": 0.7631,
+      "loss": 0.9568,
       "step": 2536
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9797374558268418e-05,
-      "loss": 0.8372,
+      "loss": 0.9769,
       "step": 2537
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979713858099821e-05,
-      "loss": 0.9285,
+      "loss": 1.0202,
       "step": 2538
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9796902467806606e-05,
-      "loss": 0.7908,
+      "loss": 0.9363,
       "step": 2539
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9796666218696877e-05,
-      "loss": 0.8553,
+      "loss": 0.9719,
       "step": 2540
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9796429833672305e-05,
-      "loss": 0.8768,
+      "loss": 0.984,
       "step": 2541
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979619331273617e-05,
-      "loss": 0.8249,
+      "loss": 0.9173,
       "step": 2542
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979595665589175e-05,
-      "loss": 0.8515,
+      "loss": 0.9014,
       "step": 2543
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9795719863142334e-05,
-      "loss": 0.9301,
+      "loss": 1.0199,
       "step": 2544
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9795482934491198e-05,
-      "loss": 0.9711,
+      "loss": 1.0331,
       "step": 2545
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9795245869941637e-05,
-      "loss": 0.8487,
+      "loss": 0.9995,
       "step": 2546
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9795008669496937e-05,
-      "loss": 0.9536,
+      "loss": 1.0002,
       "step": 2547
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979477133316039e-05,
-      "loss": 0.8443,
+      "loss": 1.0154,
       "step": 2548
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9794533860935287e-05,
-      "loss": 1.0406,
+      "loss": 0.9764,
       "step": 2549
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9794296252824922e-05,
-      "loss": 0.8658,
+      "loss": 1.0282,
       "step": 2550
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9794058508832596e-05,
-      "loss": 0.8292,
+      "loss": 0.9291,
       "step": 2551
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97938206289616e-05,
-      "loss": 0.8802,
+      "loss": 0.9616,
       "step": 2552
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9793582613215242e-05,
-      "loss": 0.8594,
+      "loss": 1.0125,
       "step": 2553
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9793344461596814e-05,
-      "loss": 0.6977,
+      "loss": 0.8835,
       "step": 2554
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9793106174109633e-05,
-      "loss": 0.8936,
+      "loss": 0.9456,
       "step": 2555
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9792867750757e-05,
-      "loss": 1.0174,
+      "loss": 1.0567,
       "step": 2556
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9792629191542217e-05,
-      "loss": 0.9338,
+      "loss": 1.0518,
       "step": 2557
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97923904964686e-05,
-      "loss": 0.9199,
+      "loss": 0.9673,
       "step": 2558
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9792151665539454e-05,
-      "loss": 0.8584,
+      "loss": 0.9164,
       "step": 2559
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97919126987581e-05,
-      "loss": 0.9729,
+      "loss": 1.0219,
       "step": 2560
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9791673596127848e-05,
-      "loss": 0.9308,
+      "loss": 1.0306,
       "step": 2561
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9791434357652023e-05,
-      "loss": 0.8772,
+      "loss": 0.9589,
       "step": 2562
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9791194983333933e-05,
-      "loss": 0.9334,
+      "loss": 0.9659,
       "step": 2563
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979095547317691e-05,
-      "loss": 0.8826,
+      "loss": 0.9775,
       "step": 2564
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.979071582718427e-05,
-      "loss": 0.9832,
+      "loss": 1.063,
       "step": 2565
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9790476045359333e-05,
-      "loss": 0.8969,
+      "loss": 1.015,
       "step": 2566
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9790236127705437e-05,
-      "loss": 1.0199,
+      "loss": 1.007,
       "step": 2567
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9789996074225905e-05,
-      "loss": 0.8436,
+      "loss": 0.9522,
       "step": 2568
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9789755884924067e-05,
-      "loss": 0.9773,
+      "loss": 1.0319,
       "step": 2569
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9789515559803253e-05,
-      "loss": 0.8732,
+      "loss": 1.0073,
       "step": 2570
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9789275098866806e-05,
-      "loss": 0.9238,
+      "loss": 1.0411,
       "step": 2571
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978903450211805e-05,
-      "loss": 0.8045,
+      "loss": 0.9976,
       "step": 2572
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9788793769560333e-05,
-      "loss": 0.781,
+      "loss": 0.9173,
       "step": 2573
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9788552901196992e-05,
-      "loss": 0.8769,
+      "loss": 1.0028,
       "step": 2574
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9788311897031366e-05,
-      "loss": 0.9063,
+      "loss": 1.007,
       "step": 2575
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97880707570668e-05,
-      "loss": 0.9396,
+      "loss": 1.0311,
       "step": 2576
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9787829481306638e-05,
-      "loss": 0.845,
+      "loss": 0.9519,
       "step": 2577
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978758806975423e-05,
-      "loss": 0.7396,
+      "loss": 0.8174,
       "step": 2578
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9787346522412925e-05,
-      "loss": 1.0196,
+      "loss": 1.0656,
       "step": 2579
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9787104839286075e-05,
-      "loss": 0.8539,
+      "loss": 0.9386,
       "step": 2580
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978686302037703e-05,
-      "loss": 0.7624,
+      "loss": 0.9061,
       "step": 2581
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9786621065689145e-05,
-      "loss": 0.916,
+      "loss": 1.0464,
       "step": 2582
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978637897522578e-05,
-      "loss": 0.934,
+      "loss": 1.0442,
       "step": 2583
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9786136748990292e-05,
-      "loss": 0.9162,
+      "loss": 1.002,
       "step": 2584
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9785894386986043e-05,
-      "loss": 0.9596,
+      "loss": 1.0253,
       "step": 2585
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978565188921639e-05,
-      "loss": 0.9152,
+      "loss": 0.9954,
       "step": 2586
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.97854092556847e-05,
-      "loss": 0.996,
+      "loss": 1.0569,
       "step": 2587
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9785166486394343e-05,
-      "loss": 0.9133,
+      "loss": 1.0136,
       "step": 2588
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9784923581348684e-05,
-      "loss": 0.8727,
+      "loss": 0.9598,
       "step": 2589
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9784680540551093e-05,
-      "loss": 0.8994,
+      "loss": 0.9974,
       "step": 2590
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9784437364004946e-05,
-      "loss": 0.9122,
+      "loss": 1.0803,
       "step": 2591
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9784194051713607e-05,
-      "loss": 0.8003,
+      "loss": 0.9037,
       "step": 2592
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9783950603680458e-05,
-      "loss": 0.9415,
+      "loss": 0.9786,
       "step": 2593
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9783707019908877e-05,
-      "loss": 0.8072,
+      "loss": 0.9438,
       "step": 2594
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9783463300402244e-05,
-      "loss": 0.8099,
+      "loss": 0.9376,
       "step": 2595
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9783219445163938e-05,
-      "loss": 0.8586,
+      "loss": 0.9762,
       "step": 2596
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9782975454197344e-05,
-      "loss": 0.823,
+      "loss": 0.9766,
       "step": 2597
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9782731327505842e-05,
-      "loss": 0.876,
+      "loss": 0.9419,
       "step": 2598
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9782487065092825e-05,
-      "loss": 0.944,
+      "loss": 1.0132,
       "step": 2599
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978224266696168e-05,
-      "loss": 0.9209,
+      "loss": 1.0278,
       "step": 2600
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9781998133115797e-05,
-      "loss": 0.8405,
+      "loss": 0.937,
       "step": 2601
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978175346355857e-05,
-      "loss": 0.9013,
+      "loss": 0.9905,
       "step": 2602
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978150865829339e-05,
-      "loss": 0.853,
+      "loss": 0.9352,
       "step": 2603
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9781263717323657e-05,
-      "loss": 0.9153,
+      "loss": 0.9855,
       "step": 2604
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9781018640652765e-05,
-      "loss": 0.8324,
+      "loss": 0.9435,
       "step": 2605
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978077342828412e-05,
-      "loss": 0.8891,
+      "loss": 0.8877,
       "step": 2606
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9780528080221116e-05,
-      "loss": 0.8917,
+      "loss": 1.0478,
       "step": 2607
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.978028259646717e-05,
-      "loss": 0.8182,
+      "loss": 0.9197,
       "step": 2608
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9780036977025673e-05,
-      "loss": 0.8863,
+      "loss": 1.0151,
       "step": 2609
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.977979122190004e-05,
-      "loss": 0.8889,
+      "loss": 0.9976,
       "step": 2610
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.977954533109368e-05,
-      "loss": 0.8767,
+      "loss": 0.9088,
       "step": 2611
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.977929930461e-05,
-      "loss": 0.8923,
+      "loss": 0.9566,
       "step": 2612
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9779053142452422e-05,
-      "loss": 0.8728,
+      "loss": 0.967,
       "step": 2613
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9778806844624356e-05,
-      "loss": 0.8554,
+      "loss": 0.9282,
       "step": 2614
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977856041112922e-05,
-      "loss": 0.9337,
+      "loss": 1.0258,
       "step": 2615
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977831384197043e-05,
-      "loss": 0.8982,
+      "loss": 0.9835,
       "step": 2616
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977806713715141e-05,
-      "loss": 0.9522,
+      "loss": 1.0916,
       "step": 2617
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9777820296675584e-05,
-      "loss": 0.7245,
+      "loss": 0.8893,
       "step": 2618
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977757332054637e-05,
-      "loss": 0.879,
+      "loss": 1.0078,
       "step": 2619
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97773262087672e-05,
-      "loss": 0.8949,
+      "loss": 0.9744,
       "step": 2620
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97770789613415e-05,
-      "loss": 0.8105,
+      "loss": 0.9312,
       "step": 2621
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9776831578272705e-05,
-      "loss": 0.8044,
+      "loss": 1.0175,
       "step": 2622
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977658405956424e-05,
-      "loss": 0.8532,
+      "loss": 0.9846,
       "step": 2623
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9776336405219546e-05,
-      "loss": 0.8706,
+      "loss": 1.0198,
       "step": 2624
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977608861524205e-05,
-      "loss": 0.8611,
+      "loss": 0.9319,
       "step": 2625
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9775840689635202e-05,
-      "loss": 0.9031,
+      "loss": 0.9826,
       "step": 2626
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977559262840243e-05,
-      "loss": 0.9677,
+      "loss": 0.9666,
       "step": 2627
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977534443154718e-05,
-      "loss": 0.9247,
+      "loss": 1.0148,
       "step": 2628
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9775096099072896e-05,
-      "loss": 0.8074,
+      "loss": 0.9993,
       "step": 2629
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977484763098302e-05,
-      "loss": 0.9913,
+      "loss": 1.028,
       "step": 2630
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9774599027281006e-05,
-      "loss": 1.0289,
+      "loss": 1.0555,
       "step": 2631
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9774350287970295e-05,
-      "loss": 0.9562,
+      "loss": 1.0926,
       "step": 2632
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9774101413054347e-05,
-      "loss": 0.8885,
+      "loss": 1.0109,
       "step": 2633
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9773852402536605e-05,
-      "loss": 0.9202,
+      "loss": 0.9827,
       "step": 2634
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977360325642053e-05,
-      "loss": 0.8844,
+      "loss": 0.9948,
       "step": 2635
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977335397470958e-05,
-      "loss": 0.8565,
+      "loss": 0.9888,
       "step": 2636
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9773104557407202e-05,
-      "loss": 0.9657,
+      "loss": 1.0544,
       "step": 2637
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977285500451687e-05,
-      "loss": 0.9223,
+      "loss": 0.9559,
       "step": 2638
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9772605316042038e-05,
-      "loss": 0.8698,
+      "loss": 0.9723,
       "step": 2639
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9772355491986176e-05,
-      "loss": 0.8219,
+      "loss": 0.9823,
       "step": 2640
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9772105532352747e-05,
-      "loss": 0.9637,
+      "loss": 1.035,
       "step": 2641
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9771855437145217e-05,
-      "loss": 0.7772,
+      "loss": 0.9202,
       "step": 2642
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9771605206367054e-05,
-      "loss": 0.8725,
+      "loss": 0.938,
       "step": 2643
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9771354840021736e-05,
-      "loss": 1.1625,
+      "loss": 1.1786,
       "step": 2644
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9771104338112736e-05,
-      "loss": 0.8803,
+      "loss": 1.0153,
       "step": 2645
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977085370064352e-05,
-      "loss": 0.9273,
+      "loss": 1.0161,
       "step": 2646
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.977060292761758e-05,
-      "loss": 0.8457,
+      "loss": 0.9366,
       "step": 2647
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9770352019038382e-05,
-      "loss": 0.8633,
+      "loss": 0.9951,
       "step": 2648
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9770100974909412e-05,
-      "loss": 0.9291,
+      "loss": 1.0674,
       "step": 2649
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9769849795234154e-05,
-      "loss": 0.8121,
+      "loss": 0.914,
       "step": 2650
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976959848001609e-05,
-      "loss": 0.8243,
+      "loss": 0.9563,
       "step": 2651
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976934702925871e-05,
-      "loss": 0.945,
+      "loss": 0.9472,
       "step": 2652
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9769095442965502e-05,
-      "loss": 0.8848,
+      "loss": 0.9726,
       "step": 2653
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9768843721139955e-05,
-      "loss": 0.9548,
+      "loss": 1.0404,
       "step": 2654
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976859186378556e-05,
-      "loss": 0.7716,
+      "loss": 0.9128,
       "step": 2655
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9768339870905814e-05,
-      "loss": 0.8953,
+      "loss": 1.0126,
       "step": 2656
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976808774250421e-05,
-      "loss": 0.9926,
+      "loss": 1.0558,
       "step": 2657
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9767835478584247e-05,
-      "loss": 0.9216,
+      "loss": 0.9742,
       "step": 2658
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976758307914943e-05,
-      "loss": 0.7192,
+      "loss": 0.9114,
       "step": 2659
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976733054420325e-05,
-      "loss": 0.8344,
+      "loss": 1.0396,
       "step": 2660
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976707787374922e-05,
-      "loss": 0.8027,
+      "loss": 1.0118,
       "step": 2661
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9766825067790846e-05,
-      "loss": 0.9546,
+      "loss": 1.0295,
       "step": 2662
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976657212633163e-05,
-      "loss": 0.9669,
+      "loss": 1.0173,
       "step": 2663
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9766319049375075e-05,
-      "loss": 0.8864,
+      "loss": 0.9563,
       "step": 2664
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9766065836924705e-05,
-      "loss": 1.056,
+      "loss": 1.0815,
       "step": 2665
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976581248898403e-05,
-      "loss": 0.9,
+      "loss": 1.0012,
       "step": 2666
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9765559005556557e-05,
-      "loss": 0.9023,
+      "loss": 1.0026,
       "step": 2667
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9765305386645813e-05,
-      "loss": 0.9796,
+      "loss": 1.1226,
       "step": 2668
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9765051632255313e-05,
-      "loss": 0.9467,
+      "loss": 0.9923,
       "step": 2669
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9764797742388572e-05,
-      "loss": 0.7306,
+      "loss": 0.92,
       "step": 2670
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976454371704912e-05,
-      "loss": 0.8452,
+      "loss": 0.9754,
       "step": 2671
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9764289556240478e-05,
-      "loss": 0.8548,
+      "loss": 0.9991,
       "step": 2672
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976403525996617e-05,
-      "loss": 0.7505,
+      "loss": 0.8996,
       "step": 2673
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9763780828229725e-05,
-      "loss": 0.9544,
+      "loss": 1.0464,
       "step": 2674
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976352626103468e-05,
-      "loss": 0.8671,
+      "loss": 1.0442,
       "step": 2675
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9763271558384554e-05,
-      "loss": 0.8566,
+      "loss": 1.0306,
       "step": 2676
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976301672028289e-05,
-      "loss": 0.8529,
+      "loss": 0.884,
       "step": 2677
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9762761746733225e-05,
-      "loss": 0.8944,
+      "loss": 0.951,
       "step": 2678
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9762506637739087e-05,
-      "loss": 0.8393,
+      "loss": 0.9885,
       "step": 2679
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9762251393304023e-05,
-      "loss": 0.8482,
+      "loss": 0.9877,
       "step": 2680
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976199601343157e-05,
-      "loss": 0.8476,
+      "loss": 0.9561,
       "step": 2681
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9761740498125274e-05,
-      "loss": 0.8497,
+      "loss": 0.9847,
       "step": 2682
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9761484847388677e-05,
-      "loss": 0.9202,
+      "loss": 0.9807,
       "step": 2683
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.976122906122533e-05,
-      "loss": 0.9181,
+      "loss": 0.9888,
       "step": 2684
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9760973139638777e-05,
-      "loss": 0.9307,
+      "loss": 1.0509,
       "step": 2685
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9760717082632572e-05,
-      "loss": 0.7874,
+      "loss": 0.8965,
       "step": 2686
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9760460890210264e-05,
-      "loss": 0.9024,
+      "loss": 1.0334,
       "step": 2687
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9760204562375415e-05,
-      "loss": 0.8487,
+      "loss": 0.9047,
       "step": 2688
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975994809913157e-05,
-      "loss": 0.7692,
+      "loss": 0.9104,
       "step": 2689
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9759691500482294e-05,
-      "loss": 0.9371,
+      "loss": 1.0676,
       "step": 2690
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9759434766431148e-05,
-      "loss": 0.8652,
+      "loss": 0.9422,
       "step": 2691
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9759177896981687e-05,
-      "loss": 0.7146,
+      "loss": 0.8692,
       "step": 2692
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9758920892137482e-05,
-      "loss": 0.9399,
+      "loss": 1.0253,
       "step": 2693
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9758663751902097e-05,
-      "loss": 0.8875,
+      "loss": 0.9634,
       "step": 2694
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9758406476279095e-05,
-      "loss": 0.9492,
+      "loss": 1.0112,
       "step": 2695
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975814906527205e-05,
-      "loss": 0.8127,
+      "loss": 1.0052,
       "step": 2696
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9757891518884535e-05,
-      "loss": 0.8933,
+      "loss": 1.0288,
       "step": 2697
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9757633837120113e-05,
-      "loss": 0.8906,
+      "loss": 1.0106,
       "step": 2698
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9757376019982374e-05,
-      "loss": 1.0042,
+      "loss": 1.07,
       "step": 2699
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9757118067474882e-05,
-      "loss": 0.9805,
+      "loss": 1.0507,
       "step": 2700
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975685997960122e-05,
-      "loss": 0.9391,
+      "loss": 1.0106,
       "step": 2701
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9756601756364964e-05,
-      "loss": 0.8379,
+      "loss": 0.967,
       "step": 2702
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975634339776971e-05,
-      "loss": 0.8317,
+      "loss": 0.935,
       "step": 2703
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975608490381903e-05,
-      "loss": 0.9201,
+      "loss": 1.064,
       "step": 2704
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9755826274516513e-05,
-      "loss": 0.8045,
+      "loss": 0.9601,
       "step": 2705
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975556750986575e-05,
-      "loss": 0.8807,
+      "loss": 1.0304,
       "step": 2706
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9755308609870328e-05,
-      "loss": 0.8141,
+      "loss": 0.9347,
       "step": 2707
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9755049574533843e-05,
-      "loss": 0.7913,
+      "loss": 0.9048,
       "step": 2708
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9754790403859883e-05,
-      "loss": 0.6787,
+      "loss": 0.8347,
       "step": 2709
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975453109785205e-05,
-      "loss": 0.797,
+      "loss": 0.931,
       "step": 2710
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9754271656513935e-05,
-      "loss": 0.9142,
+      "loss": 1.0196,
       "step": 2711
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9754012079849142e-05,
-      "loss": 0.9214,
+      "loss": 1.0153,
       "step": 2712
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9753752367861265e-05,
-      "loss": 0.9723,
+      "loss": 0.9933,
       "step": 2713
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9753492520553918e-05,
-      "loss": 0.8331,
+      "loss": 0.9672,
       "step": 2714
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97532325379307e-05,
-      "loss": 0.958,
+      "loss": 1.0474,
       "step": 2715
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9752972419995218e-05,
-      "loss": 0.8027,
+      "loss": 0.9282,
       "step": 2716
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975271216675108e-05,
-      "loss": 0.8228,
+      "loss": 0.9382,
       "step": 2717
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97524517782019e-05,
-      "loss": 0.9697,
+      "loss": 1.017,
       "step": 2718
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9752191254351288e-05,
-      "loss": 0.9091,
+      "loss": 0.9988,
       "step": 2719
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9751930595202857e-05,
-      "loss": 0.8168,
+      "loss": 0.9223,
       "step": 2720
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9751669800760226e-05,
-      "loss": 0.8677,
+      "loss": 0.9914,
       "step": 2721
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9751408871027013e-05,
-      "loss": 0.7462,
+      "loss": 0.9036,
       "step": 2722
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9751147806006835e-05,
-      "loss": 1.0505,
+      "loss": 1.014,
       "step": 2723
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.975088660570332e-05,
-      "loss": 0.9378,
+      "loss": 1.0269,
       "step": 2724
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9750625270120086e-05,
-      "loss": 0.8432,
+      "loss": 1.0454,
       "step": 2725
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9750363799260762e-05,
-      "loss": 0.9416,
+      "loss": 1.0932,
       "step": 2726
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9750102193128975e-05,
-      "loss": 0.798,
+      "loss": 0.9317,
       "step": 2727
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974984045172835e-05,
-      "loss": 0.9462,
+      "loss": 0.9426,
       "step": 2728
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9749578575062524e-05,
-      "loss": 0.9157,
+      "loss": 1.0003,
       "step": 2729
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9749316563135128e-05,
-      "loss": 0.7569,
+      "loss": 0.9605,
       "step": 2730
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9749054415949797e-05,
-      "loss": 0.8851,
+      "loss": 0.9853,
       "step": 2731
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974879213351017e-05,
-      "loss": 0.8036,
+      "loss": 0.939,
       "step": 2732
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974852971581988e-05,
-      "loss": 0.8891,
+      "loss": 0.9396,
       "step": 2733
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9748267162882574e-05,
-      "loss": 0.8683,
+      "loss": 0.9496,
       "step": 2734
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9748004474701894e-05,
-      "loss": 0.9534,
+      "loss": 1.0125,
       "step": 2735
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974774165128148e-05,
-      "loss": 0.8837,
+      "loss": 0.9418,
       "step": 2736
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974747869262498e-05,
-      "loss": 0.8186,
+      "loss": 1.0007,
       "step": 2737
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9747215598736048e-05,
-      "loss": 0.8727,
+      "loss": 0.921,
       "step": 2738
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974695236961832e-05,
-      "loss": 0.9241,
+      "loss": 0.9945,
       "step": 2739
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9746689005275465e-05,
-      "loss": 0.8575,
+      "loss": 0.8404,
       "step": 2740
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9746425505711127e-05,
-      "loss": 0.8694,
+      "loss": 0.9353,
       "step": 2741
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974616187092896e-05,
-      "loss": 0.8225,
+      "loss": 0.9517,
       "step": 2742
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974589810093263e-05,
-      "loss": 0.9284,
+      "loss": 1.0162,
       "step": 2743
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974563419572579e-05,
-      "loss": 0.8579,
+      "loss": 0.9532,
       "step": 2744
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97453701553121e-05,
-      "loss": 0.8528,
+      "loss": 0.9426,
       "step": 2745
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974510597969523e-05,
-      "loss": 0.868,
+      "loss": 0.9693,
       "step": 2746
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974484166887884e-05,
-      "loss": 0.8928,
+      "loss": 0.9526,
       "step": 2747
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9744577222866594e-05,
-      "loss": 0.947,
+      "loss": 1.071,
       "step": 2748
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9744312641662165e-05,
-      "loss": 0.8193,
+      "loss": 0.9573,
       "step": 2749
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9744047925269228e-05,
-      "loss": 0.922,
+      "loss": 1.051,
       "step": 2750
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9743783073691452e-05,
-      "loss": 0.856,
+      "loss": 1.0257,
       "step": 2751
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9743518086932507e-05,
-      "loss": 1.0287,
+      "loss": 1.0453,
       "step": 2752
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974325296499607e-05,
-      "loss": 0.8889,
+      "loss": 0.9923,
       "step": 2753
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974298770788583e-05,
-      "loss": 0.7969,
+      "loss": 0.9242,
       "step": 2754
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9742722315605452e-05,
-      "loss": 0.7345,
+      "loss": 0.7859,
       "step": 2755
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9742456788158627e-05,
-      "loss": 1.0216,
+      "loss": 1.0293,
       "step": 2756
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974219112554904e-05,
-      "loss": 1.0099,
+      "loss": 1.0591,
       "step": 2757
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9741925327780368e-05,
-      "loss": 0.7411,
+      "loss": 0.859,
       "step": 2758
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9741659394856306e-05,
-      "loss": 0.8495,
+      "loss": 0.9499,
       "step": 2759
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9741393326780542e-05,
-      "loss": 0.9981,
+      "loss": 1.0064,
       "step": 2760
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974112712355677e-05,
-      "loss": 0.9402,
+      "loss": 1.0429,
       "step": 2761
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9740860785188676e-05,
-      "loss": 0.9359,
+      "loss": 0.9988,
       "step": 2762
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974059431167996e-05,
-      "loss": 0.8645,
+      "loss": 1.0297,
       "step": 2763
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974032770303432e-05,
-      "loss": 0.9253,
+      "loss": 1.0303,
       "step": 2764
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.974006095925545e-05,
-      "loss": 0.7625,
+      "loss": 0.9567,
       "step": 2765
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9739794080347056e-05,
-      "loss": 0.8694,
+      "loss": 0.9385,
       "step": 2766
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9739527066312837e-05,
-      "loss": 0.8027,
+      "loss": 0.9484,
       "step": 2767
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9739259917156498e-05,
-      "loss": 0.9225,
+      "loss": 0.986,
       "step": 2768
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9738992632881744e-05,
-      "loss": 0.8175,
+      "loss": 0.9228,
       "step": 2769
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9738725213492286e-05,
-      "loss": 0.9067,
+      "loss": 1.0215,
       "step": 2770
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9738457658991836e-05,
-      "loss": 0.8132,
+      "loss": 0.9686,
       "step": 2771
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9738189969384102e-05,
-      "loss": 0.9382,
+      "loss": 1.0251,
       "step": 2772
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9737922144672797e-05,
-      "loss": 0.8196,
+      "loss": 1.027,
       "step": 2773
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9737654184861637e-05,
-      "loss": 0.8667,
+      "loss": 0.9814,
       "step": 2774
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9737386089954347e-05,
-      "loss": 0.7802,
+      "loss": 0.9432,
       "step": 2775
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973711785995464e-05,
-      "loss": 0.8553,
+      "loss": 1.0432,
       "step": 2776
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9736849494866234e-05,
-      "loss": 0.915,
+      "loss": 1.0773,
       "step": 2777
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9736580994692852e-05,
-      "loss": 0.8218,
+      "loss": 0.9348,
       "step": 2778
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973631235943823e-05,
-      "loss": 0.9904,
+      "loss": 1.0555,
       "step": 2779
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973604358910609e-05,
-      "loss": 0.8614,
+      "loss": 0.9909,
       "step": 2780
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9735774683700154e-05,
-      "loss": 0.8339,
+      "loss": 0.9052,
       "step": 2781
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9735505643224157e-05,
-      "loss": 0.9536,
+      "loss": 1.0266,
       "step": 2782
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9735236467681834e-05,
-      "loss": 0.8932,
+      "loss": 0.9579,
       "step": 2783
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973496715707692e-05,
-      "loss": 0.9017,
+      "loss": 0.9593,
       "step": 2784
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973469771141314e-05,
-      "loss": 0.9375,
+      "loss": 0.9782,
       "step": 2785
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973442813069425e-05,
-      "loss": 0.9987,
+      "loss": 1.0288,
       "step": 2786
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9734158414923978e-05,
-      "loss": 0.8486,
+      "loss": 0.9292,
       "step": 2787
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973388856410607e-05,
-      "loss": 0.8049,
+      "loss": 1.0162,
       "step": 2788
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973361857824427e-05,
-      "loss": 0.8352,
+      "loss": 0.9419,
       "step": 2789
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973334845734232e-05,
-      "loss": 0.9397,
+      "loss": 1.0058,
       "step": 2790
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973307820140397e-05,
-      "loss": 0.8129,
+      "loss": 0.9404,
       "step": 2791
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9732807810432966e-05,
-      "loss": 0.7846,
+      "loss": 0.8171,
       "step": 2792
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973253728443307e-05,
-      "loss": 1.0108,
+      "loss": 1.0232,
       "step": 2793
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9732266623408024e-05,
-      "loss": 0.815,
+      "loss": 0.92,
       "step": 2794
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9731995827361586e-05,
-      "loss": 0.8128,
+      "loss": 1.0217,
       "step": 2795
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9731724896297514e-05,
-      "loss": 0.94,
+      "loss": 1.045,
       "step": 2796
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9731453830219567e-05,
-      "loss": 0.8335,
+      "loss": 0.9267,
       "step": 2797
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9731182629131505e-05,
-      "loss": 0.8042,
+      "loss": 0.9734,
       "step": 2798
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9730911293037094e-05,
-      "loss": 0.76,
+      "loss": 0.9534,
       "step": 2799
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973063982194009e-05,
-      "loss": 1.0147,
+      "loss": 1.0381,
       "step": 2800
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973036821584427e-05,
-      "loss": 1.0158,
+      "loss": 1.093,
       "step": 2801
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.973009647475339e-05,
-      "loss": 0.9674,
+      "loss": 1.0392,
       "step": 2802
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9729824598671233e-05,
-      "loss": 0.8833,
+      "loss": 0.959,
       "step": 2803
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972955258760156e-05,
-      "loss": 1.0451,
+      "loss": 1.0933,
       "step": 2804
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972928044154815e-05,
-      "loss": 0.8774,
+      "loss": 0.9597,
       "step": 2805
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9729008160514782e-05,
-      "loss": 0.7995,
+      "loss": 0.9523,
       "step": 2806
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9728735744505225e-05,
-      "loss": 0.8674,
+      "loss": 0.9976,
       "step": 2807
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9728463193523265e-05,
-      "loss": 1.0016,
+      "loss": 1.0543,
       "step": 2808
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9728190507572677e-05,
-      "loss": 0.8025,
+      "loss": 0.9086,
       "step": 2809
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9727917686657255e-05,
-      "loss": 0.9017,
+      "loss": 0.9589,
       "step": 2810
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972764473078077e-05,
-      "loss": 0.8477,
+      "loss": 0.9673,
       "step": 2811
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9727371639947023e-05,
-      "loss": 0.7699,
+      "loss": 0.8449,
       "step": 2812
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972709841415979e-05,
-      "loss": 0.7691,
+      "loss": 0.9772,
       "step": 2813
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972682505342287e-05,
-      "loss": 0.9431,
+      "loss": 0.9904,
       "step": 2814
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9726551557740053e-05,
-      "loss": 0.9248,
+      "loss": 0.9894,
       "step": 2815
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9726277927115137e-05,
-      "loss": 0.9103,
+      "loss": 1.0426,
       "step": 2816
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972600416155191e-05,
-      "loss": 0.981,
+      "loss": 1.0783,
       "step": 2817
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9725730261054174e-05,
-      "loss": 0.9899,
+      "loss": 1.1613,
       "step": 2818
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972545622562573e-05,
-      "loss": 0.8874,
+      "loss": 1.0007,
       "step": 2819
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9725182055270383e-05,
-      "loss": 0.917,
+      "loss": 1.1145,
       "step": 2820
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972490774999193e-05,
-      "loss": 0.7957,
+      "loss": 0.9333,
       "step": 2821
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972463330979418e-05,
-      "loss": 0.8707,
+      "loss": 0.9846,
       "step": 2822
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972435873468094e-05,
-      "loss": 0.9122,
+      "loss": 1.0281,
       "step": 2823
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972408402465602e-05,
-      "loss": 0.9989,
+      "loss": 1.0571,
       "step": 2824
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972380917972323e-05,
-      "loss": 0.9712,
+      "loss": 0.978,
       "step": 2825
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9723534199886387e-05,
-      "loss": 0.9763,
+      "loss": 0.9948,
       "step": 2826
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97232590851493e-05,
-      "loss": 0.7936,
+      "loss": 0.9222,
       "step": 2827
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972298383551579e-05,
-      "loss": 0.9604,
+      "loss": 0.9705,
       "step": 2828
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9722708450989678e-05,
-      "loss": 0.8923,
+      "loss": 0.9877,
       "step": 2829
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9722432931574774e-05,
-      "loss": 0.7079,
+      "loss": 0.843,
       "step": 2830
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9722157277274912e-05,
-      "loss": 0.9598,
+      "loss": 1.0278,
       "step": 2831
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972188148809391e-05,
-      "loss": 0.886,
+      "loss": 0.975,
       "step": 2832
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9721605564035595e-05,
-      "loss": 0.8807,
+      "loss": 1.008,
       "step": 2833
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9721329505103798e-05,
-      "loss": 0.8556,
+      "loss": 0.9532,
       "step": 2834
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9721053311302346e-05,
-      "loss": 0.8462,
+      "loss": 0.961,
       "step": 2835
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9720776982635072e-05,
-      "loss": 1.0484,
+      "loss": 1.1102,
       "step": 2836
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9720500519105807e-05,
-      "loss": 0.8981,
+      "loss": 0.9919,
       "step": 2837
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.972022392071839e-05,
-      "loss": 0.8333,
+      "loss": 0.8967,
       "step": 2838
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971994718747666e-05,
-      "loss": 1.0715,
+      "loss": 1.1332,
       "step": 2839
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971967031938445e-05,
-      "loss": 0.9774,
+      "loss": 1.0565,
       "step": 2840
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9719393316445607e-05,
-      "loss": 0.845,
+      "loss": 0.9795,
       "step": 2841
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9719116178663975e-05,
-      "loss": 0.9292,
+      "loss": 0.9932,
       "step": 2842
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971883890604339e-05,
-      "loss": 0.9192,
+      "loss": 0.9998,
       "step": 2843
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9718561498587707e-05,
-      "loss": 0.8467,
+      "loss": 0.936,
       "step": 2844
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971828395630077e-05,
-      "loss": 0.9241,
+      "loss": 1.0248,
       "step": 2845
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9718006279186433e-05,
-      "loss": 0.8168,
+      "loss": 0.9708,
       "step": 2846
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9717728467248547e-05,
-      "loss": 0.8724,
+      "loss": 1.0378,
       "step": 2847
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9717450520490967e-05,
-      "loss": 0.806,
+      "loss": 0.9059,
       "step": 2848
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9717172438917547e-05,
-      "loss": 1.0345,
+      "loss": 1.0727,
       "step": 2849
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9716894222532145e-05,
-      "loss": 0.8806,
+      "loss": 0.8572,
       "step": 2850
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9716615871338623e-05,
-      "loss": 0.8157,
+      "loss": 1.0086,
       "step": 2851
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9716337385340845e-05,
-      "loss": 0.8644,
+      "loss": 1.0644,
       "step": 2852
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971605876454267e-05,
-      "loss": 0.9141,
+      "loss": 1.0736,
       "step": 2853
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971578000894796e-05,
-      "loss": 0.8312,
+      "loss": 0.9551,
       "step": 2854
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9715501118560595e-05,
-      "loss": 0.9191,
+      "loss": 0.9263,
       "step": 2855
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971522209338443e-05,
-      "loss": 0.8663,
+      "loss": 0.9525,
       "step": 2856
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9714942933423343e-05,
-      "loss": 0.7553,
+      "loss": 0.9065,
       "step": 2857
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971466363868121e-05,
-      "loss": 0.8532,
+      "loss": 0.9298,
       "step": 2858
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9714384209161903e-05,
-      "loss": 0.9794,
+      "loss": 1.0524,
       "step": 2859
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9714104644869296e-05,
-      "loss": 0.8637,
+      "loss": 1.0433,
       "step": 2860
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971382494580727e-05,
-      "loss": 0.8449,
+      "loss": 0.9778,
       "step": 2861
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9713545111979704e-05,
-      "loss": 0.8593,
+      "loss": 0.9104,
       "step": 2862
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9713265143390485e-05,
-      "loss": 0.8001,
+      "loss": 0.9127,
       "step": 2863
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9712985040043487e-05,
-      "loss": 0.8353,
+      "loss": 0.9106,
       "step": 2864
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9712704801942605e-05,
-      "loss": 0.9124,
+      "loss": 1.0323,
       "step": 2865
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.971242442909173e-05,
-      "loss": 0.8945,
+      "loss": 0.9552,
       "step": 2866
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9712143921494738e-05,
-      "loss": 0.8167,
+      "loss": 0.8972,
       "step": 2867
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9711863279155533e-05,
-      "loss": 1.0069,
+      "loss": 1.0854,
       "step": 2868
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9711582502078004e-05,
-      "loss": 0.8281,
+      "loss": 0.9363,
       "step": 2869
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9711301590266044e-05,
-      "loss": 0.8517,
+      "loss": 0.8603,
       "step": 2870
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9711020543723557e-05,
-      "loss": 0.9968,
+      "loss": 0.9962,
       "step": 2871
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9710739362454435e-05,
-      "loss": 0.9535,
+      "loss": 1.0285,
       "step": 2872
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9710458046462585e-05,
-      "loss": 0.9039,
+      "loss": 1.0107,
       "step": 2873
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9710176595751903e-05,
-      "loss": 0.8432,
+      "loss": 0.9713,
       "step": 2874
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.97098950103263e-05,
-      "loss": 1.021,
+      "loss": 1.0066,
       "step": 2875
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9709613290189684e-05,
-      "loss": 0.825,
+      "loss": 0.9482,
       "step": 2876
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9709331435345952e-05,
-      "loss": 0.8573,
+      "loss": 0.8973,
       "step": 2877
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.970904944579903e-05,
-      "loss": 0.9717,
+      "loss": 1.0464,
       "step": 2878
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9708767321552816e-05,
-      "loss": 0.818,
+      "loss": 0.9393,
       "step": 2879
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.970848506261123e-05,
-      "loss": 1.0117,
+      "loss": 1.0687,
       "step": 2880
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9708202668978193e-05,
-      "loss": 0.8507,
+      "loss": 0.9635,
       "step": 2881
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9707920140657616e-05,
-      "loss": 0.8672,
+      "loss": 0.9445,
       "step": 2882
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.970763747765342e-05,
-      "loss": 0.9122,
+      "loss": 1.0018,
       "step": 2883
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.970735467996953e-05,
-      "loss": 0.8099,
+      "loss": 0.902,
       "step": 2884
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9707071747609864e-05,
-      "loss": 0.938,
+      "loss": 1.0259,
       "step": 2885
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9706788680578352e-05,
-      "loss": 0.8946,
+      "loss": 1.0196,
       "step": 2886
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9706505478878915e-05,
-      "loss": 0.7444,
+      "loss": 0.8919,
       "step": 2887
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9706222142515494e-05,
-      "loss": 0.9631,
+      "loss": 1.0837,
       "step": 2888
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9705938671492007e-05,
-      "loss": 0.8618,
+      "loss": 0.965,
       "step": 2889
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.970565506581239e-05,
-      "loss": 0.7296,
+      "loss": 0.8568,
       "step": 2890
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9705371325480582e-05,
-      "loss": 0.9282,
+      "loss": 0.9544,
       "step": 2891
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9705087450500517e-05,
-      "loss": 1.0504,
+      "loss": 1.1145,
       "step": 2892
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9704803440876136e-05,
-      "loss": 0.7722,
+      "loss": 0.9404,
       "step": 2893
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9704519296611374e-05,
-      "loss": 0.9255,
+      "loss": 1.0817,
       "step": 2894
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9704235017710174e-05,
-      "loss": 0.915,
+      "loss": 0.9747,
       "step": 2895
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.970395060417648e-05,
-      "loss": 0.831,
+      "loss": 0.9232,
       "step": 2896
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9703666056014244e-05,
-      "loss": 0.9022,
+      "loss": 0.9794,
       "step": 2897
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.970338137322741e-05,
-      "loss": 0.853,
+      "loss": 0.9577,
       "step": 2898
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9703096555819923e-05,
-      "loss": 1.0494,
+      "loss": 1.0723,
       "step": 2899
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.970281160379574e-05,
-      "loss": 0.8354,
+      "loss": 0.9699,
       "step": 2900
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9702526517158813e-05,
-      "loss": 0.9974,
+      "loss": 1.1019,
       "step": 2901
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9702241295913093e-05,
-      "loss": 0.9579,
+      "loss": 1.0661,
       "step": 2902
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9701955940062544e-05,
-      "loss": 0.8436,
+      "loss": 0.9183,
       "step": 2903
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9701670449611122e-05,
-      "loss": 0.8251,
+      "loss": 0.922,
       "step": 2904
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9701384824562786e-05,
-      "loss": 0.8848,
+      "loss": 0.9514,
       "step": 2905
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.97010990649215e-05,
-      "loss": 0.9465,
+      "loss": 0.9764,
       "step": 2906
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9700813170691232e-05,
-      "loss": 0.8952,
+      "loss": 0.9877,
       "step": 2907
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.970052714187594e-05,
-      "loss": 0.8997,
+      "loss": 0.9717,
       "step": 2908
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.97002409784796e-05,
-      "loss": 0.8231,
+      "loss": 0.8525,
       "step": 2909
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969995468050618e-05,
-      "loss": 0.9895,
+      "loss": 1.0217,
       "step": 2910
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969966824795965e-05,
-      "loss": 0.9767,
+      "loss": 1.0344,
       "step": 2911
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9699381680843983e-05,
-      "loss": 0.8512,
+      "loss": 0.987,
       "step": 2912
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9699094979163158e-05,
-      "loss": 0.8899,
+      "loss": 1.0099,
       "step": 2913
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969880814292115e-05,
-      "loss": 0.822,
+      "loss": 0.9681,
       "step": 2914
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9698521172121942e-05,
-      "loss": 0.8756,
+      "loss": 0.9962,
       "step": 2915
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9698234066769514e-05,
-      "loss": 0.9324,
+      "loss": 0.963,
       "step": 2916
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969794682686785e-05,
-      "loss": 0.8062,
+      "loss": 0.8955,
       "step": 2917
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9697659452420928e-05,
-      "loss": 0.8691,
+      "loss": 0.9282,
       "step": 2918
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9697371943432743e-05,
-      "loss": 0.8667,
+      "loss": 1.0199,
       "step": 2919
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969708429990728e-05,
-      "loss": 0.9071,
+      "loss": 0.9581,
       "step": 2920
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9696796521848532e-05,
-      "loss": 0.9336,
+      "loss": 0.9775,
       "step": 2921
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969650860926049e-05,
-      "loss": 0.9665,
+      "loss": 1.0355,
       "step": 2922
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9696220562147152e-05,
-      "loss": 0.8715,
+      "loss": 1.0069,
       "step": 2923
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9695932380512502e-05,
-      "loss": 0.923,
+      "loss": 1.0014,
       "step": 2924
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9695644064360554e-05,
-      "loss": 0.9732,
+      "loss": 1.0337,
       "step": 2925
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96953556136953e-05,
-      "loss": 0.849,
+      "loss": 0.9835,
       "step": 2926
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969506702852074e-05,
-      "loss": 0.9013,
+      "loss": 1.0942,
       "step": 2927
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9694778308840884e-05,
-      "loss": 0.874,
+      "loss": 0.9364,
       "step": 2928
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969448945465973e-05,
-      "loss": 0.7808,
+      "loss": 0.8914,
       "step": 2929
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9694200465981293e-05,
-      "loss": 1.0927,
+      "loss": 1.1286,
       "step": 2930
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969391134280958e-05,
-      "loss": 0.912,
+      "loss": 1.0268,
       "step": 2931
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9693622085148594e-05,
-      "loss": 0.927,
+      "loss": 1.01,
       "step": 2932
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969333269300236e-05,
-      "loss": 0.8361,
+      "loss": 1.003,
       "step": 2933
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969304316637489e-05,
-      "loss": 1.0194,
+      "loss": 1.1145,
       "step": 2934
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9692753505270197e-05,
-      "loss": 1.0549,
+      "loss": 1.0876,
       "step": 2935
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96924637096923e-05,
-      "loss": 0.7837,
+      "loss": 0.9868,
       "step": 2936
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9692173779645222e-05,
-      "loss": 0.7868,
+      "loss": 0.9204,
       "step": 2937
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9691883715132984e-05,
-      "loss": 1.0014,
+      "loss": 0.9899,
       "step": 2938
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969159351615961e-05,
-      "loss": 1.0256,
+      "loss": 1.0308,
       "step": 2939
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9691303182729127e-05,
-      "loss": 0.9518,
+      "loss": 1.0046,
       "step": 2940
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969101271484556e-05,
-      "loss": 0.9624,
+      "loss": 1.0352,
       "step": 2941
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9690722112512946e-05,
-      "loss": 0.8377,
+      "loss": 0.9501,
       "step": 2942
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.969043137573531e-05,
-      "loss": 0.899,
+      "loss": 0.9668,
       "step": 2943
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9690140504516685e-05,
-      "loss": 0.7861,
+      "loss": 0.9266,
       "step": 2944
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968984949886111e-05,
-      "loss": 0.8668,
+      "loss": 1.0062,
       "step": 2945
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9689558358772623e-05,
-      "loss": 0.8448,
+      "loss": 0.8697,
       "step": 2946
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9689267084255263e-05,
-      "loss": 0.9703,
+      "loss": 1.0057,
       "step": 2947
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9688975675313066e-05,
-      "loss": 0.8794,
+      "loss": 0.9434,
       "step": 2948
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968868413195008e-05,
-      "loss": 0.7046,
+      "loss": 0.8517,
       "step": 2949
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968839245417035e-05,
-      "loss": 0.8721,
+      "loss": 1.1019,
       "step": 2950
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9688100641977918e-05,
-      "loss": 1.0033,
+      "loss": 1.031,
       "step": 2951
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9687808695376834e-05,
-      "loss": 0.9386,
+      "loss": 1.0053,
       "step": 2952
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968751661437115e-05,
-      "loss": 0.9011,
+      "loss": 1.0517,
       "step": 2953
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968722439896492e-05,
-      "loss": 0.864,
+      "loss": 0.9172,
       "step": 2954
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9686932049162196e-05,
-      "loss": 0.8931,
+      "loss": 0.9287,
       "step": 2955
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968663956496703e-05,
-      "loss": 0.7947,
+      "loss": 0.9231,
       "step": 2956
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968634694638349e-05,
-      "loss": 0.9025,
+      "loss": 1.0015,
       "step": 2957
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9686054193415625e-05,
-      "loss": 0.8358,
+      "loss": 0.9467,
       "step": 2958
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9685761306067502e-05,
-      "loss": 0.8427,
+      "loss": 0.9733,
       "step": 2959
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9685468284343183e-05,
-      "loss": 0.8189,
+      "loss": 0.977,
       "step": 2960
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9685175128246732e-05,
-      "loss": 0.7935,
+      "loss": 0.9084,
       "step": 2961
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968488183778222e-05,
-      "loss": 0.862,
+      "loss": 0.9402,
       "step": 2962
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968458841295371e-05,
-      "loss": 0.7915,
+      "loss": 0.9539,
       "step": 2963
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9684294853765278e-05,
-      "loss": 0.8325,
+      "loss": 1.0229,
       "step": 2964
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9684001160220994e-05,
-      "loss": 0.7803,
+      "loss": 0.9425,
       "step": 2965
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968370733232494e-05,
-      "loss": 0.8529,
+      "loss": 0.9789,
       "step": 2966
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968341337008118e-05,
-      "loss": 0.9385,
+      "loss": 0.9926,
       "step": 2967
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96831192734938e-05,
-      "loss": 0.9786,
+      "loss": 1.0977,
       "step": 2968
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9682825042566874e-05,
-      "loss": 0.8953,
+      "loss": 0.9623,
       "step": 2969
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9682530677304498e-05,
-      "loss": 0.7714,
+      "loss": 0.9144,
       "step": 2970
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968223617771074e-05,
-      "loss": 0.9908,
+      "loss": 1.0334,
       "step": 2971
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9681941543789695e-05,
-      "loss": 0.7754,
+      "loss": 0.8291,
       "step": 2972
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9681646775545444e-05,
-      "loss": 0.9043,
+      "loss": 1.0108,
       "step": 2973
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9681351872982083e-05,
-      "loss": 0.8025,
+      "loss": 0.9725,
       "step": 2974
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96810568361037e-05,
-      "loss": 0.9106,
+      "loss": 0.9968,
       "step": 2975
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9680761664914395e-05,
-      "loss": 0.7678,
+      "loss": 0.9254,
       "step": 2976
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968046635941825e-05,
-      "loss": 0.9287,
+      "loss": 0.994,
       "step": 2977
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.968017091961937e-05,
-      "loss": 0.7561,
+      "loss": 0.8767,
       "step": 2978
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967987534552186e-05,
-      "loss": 0.9735,
+      "loss": 1.0326,
       "step": 2979
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9679579637129804e-05,
-      "loss": 0.8421,
+      "loss": 0.996,
       "step": 2980
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967928379444732e-05,
-      "loss": 0.8993,
+      "loss": 0.9891,
       "step": 2981
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9678987817478504e-05,
-      "loss": 0.8062,
+      "loss": 0.9792,
       "step": 2982
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9678691706227463e-05,
-      "loss": 0.6969,
+      "loss": 0.9141,
       "step": 2983
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967839546069831e-05,
-      "loss": 0.9882,
+      "loss": 1.0148,
       "step": 2984
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967809908089515e-05,
-      "loss": 0.8882,
+      "loss": 1.0034,
       "step": 2985
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9677802566822097e-05,
-      "loss": 0.7833,
+      "loss": 0.9696,
       "step": 2986
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9677505918483266e-05,
-      "loss": 0.9503,
+      "loss": 0.9609,
       "step": 2987
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967720913588277e-05,
-      "loss": 0.8563,
+      "loss": 0.9277,
       "step": 2988
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967691221902473e-05,
-      "loss": 0.9729,
+      "loss": 1.0281,
       "step": 2989
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967661516791326e-05,
-      "loss": 0.8167,
+      "loss": 0.946,
       "step": 2990
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9676317982552483e-05,
-      "loss": 0.9371,
+      "loss": 1.0179,
       "step": 2991
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967602066294652e-05,
-      "loss": 0.8152,
+      "loss": 0.9496,
       "step": 2992
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9675723209099504e-05,
-      "loss": 0.9262,
+      "loss": 1.0077,
       "step": 2993
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9675425621015554e-05,
-      "loss": 0.9319,
+      "loss": 1.0222,
       "step": 2994
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9675127898698804e-05,
-      "loss": 0.9422,
+      "loss": 0.9923,
       "step": 2995
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967483004215338e-05,
-      "loss": 0.9554,
+      "loss": 1.0245,
       "step": 2996
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9674532051383414e-05,
-      "loss": 0.9567,
+      "loss": 1.0619,
       "step": 2997
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9674233926393045e-05,
-      "loss": 0.8384,
+      "loss": 0.9821,
       "step": 2998
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9673935667186406e-05,
-      "loss": 0.8868,
+      "loss": 0.9692,
       "step": 2999
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967363727376763e-05,
-      "loss": 0.8494,
+      "loss": 0.9156,
       "step": 3000
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9673338746140868e-05,
-      "loss": 0.9101,
+      "loss": 0.9838,
       "step": 3001
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9673040084310253e-05,
-      "loss": 0.875,
+      "loss": 0.9103,
       "step": 3002
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9672741288279933e-05,
-      "loss": 0.9429,
+      "loss": 1.0107,
       "step": 3003
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967244235805405e-05,
-      "loss": 0.7841,
+      "loss": 1.0067,
       "step": 3004
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9672143293636753e-05,
-      "loss": 0.8769,
+      "loss": 0.9556,
       "step": 3005
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967184409503219e-05,
-      "loss": 0.8327,
+      "loss": 0.9944,
       "step": 3006
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9671544762244516e-05,
-      "loss": 0.8968,
+      "loss": 0.9699,
       "step": 3007
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9671245295277876e-05,
-      "loss": 0.8963,
+      "loss": 0.9887,
       "step": 3008
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967094569413643e-05,
-      "loss": 0.8245,
+      "loss": 0.9301,
       "step": 3009
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9670645958824336e-05,
-      "loss": 0.9594,
+      "loss": 0.8998,
       "step": 3010
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.967034608934575e-05,
-      "loss": 0.8221,
+      "loss": 0.9379,
       "step": 3011
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9670046085704832e-05,
-      "loss": 0.8238,
+      "loss": 1.0127,
       "step": 3012
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9669745947905744e-05,
-      "loss": 0.8395,
+      "loss": 0.9908,
       "step": 3013
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966944567595265e-05,
-      "loss": 0.7978,
+      "loss": 0.9242,
       "step": 3014
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966914526984972e-05,
-      "loss": 0.8946,
+      "loss": 0.9845,
       "step": 3015
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966884472960111e-05,
-      "loss": 0.9681,
+      "loss": 0.9927,
       "step": 3016
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9668544055211004e-05,
-      "loss": 0.7547,
+      "loss": 0.9473,
       "step": 3017
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9668243246683567e-05,
-      "loss": 0.9196,
+      "loss": 0.9489,
       "step": 3018
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966794230402297e-05,
-      "loss": 0.8274,
+      "loss": 1.0115,
       "step": 3019
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9667641227233393e-05,
-      "loss": 0.7763,
+      "loss": 0.8719,
       "step": 3020
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966734001631901e-05,
-      "loss": 0.9262,
+      "loss": 0.9876,
       "step": 3021
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9667038671284e-05,
-      "loss": 0.9448,
+      "loss": 1.0495,
       "step": 3022
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9666737192132545e-05,
-      "loss": 0.9076,
+      "loss": 0.9914,
       "step": 3023
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9666435578868826e-05,
-      "loss": 0.9326,
+      "loss": 0.9865,
       "step": 3024
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9666133831497026e-05,
-      "loss": 0.9447,
+      "loss": 1.0126,
       "step": 3025
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9665831950021337e-05,
-      "loss": 0.8911,
+      "loss": 0.962,
       "step": 3026
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966552993444594e-05,
-      "loss": 0.892,
+      "loss": 1.0053,
       "step": 3027
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9665227784775033e-05,
-      "loss": 0.7003,
+      "loss": 0.9015,
       "step": 3028
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9664925501012803e-05,
-      "loss": 0.9441,
+      "loss": 0.9673,
       "step": 3029
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966462308316344e-05,
-      "loss": 0.8107,
+      "loss": 0.8755,
       "step": 3030
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966432053123115e-05,
-      "loss": 0.9319,
+      "loss": 0.956,
       "step": 3031
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966401784522012e-05,
-      "loss": 0.8941,
+      "loss": 0.9094,
       "step": 3032
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9663715025134555e-05,
-      "loss": 1.0432,
+      "loss": 1.1149,
       "step": 3033
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9663412070978656e-05,
-      "loss": 0.8942,
+      "loss": 1.0488,
       "step": 3034
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9663108982756624e-05,
-      "loss": 0.8833,
+      "loss": 1.0474,
       "step": 3035
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9662805760472666e-05,
-      "loss": 0.8021,
+      "loss": 1.0143,
       "step": 3036
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966250240413099e-05,
-      "loss": 0.8585,
+      "loss": 0.9554,
       "step": 3037
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9662198913735797e-05,
-      "loss": 0.8617,
+      "loss": 1.0154,
       "step": 3038
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9661895289291305e-05,
-      "loss": 0.9135,
+      "loss": 1.0206,
       "step": 3039
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9661591530801728e-05,
-      "loss": 0.8273,
+      "loss": 0.9368,
       "step": 3040
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9661287638271275e-05,
-      "loss": 0.8549,
+      "loss": 0.958,
       "step": 3041
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9660983611704162e-05,
-      "loss": 0.9216,
+      "loss": 1.0523,
       "step": 3042
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9660679451104607e-05,
-      "loss": 0.8735,
+      "loss": 0.9936,
       "step": 3043
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9660375156476835e-05,
-      "loss": 0.8896,
+      "loss": 0.9629,
       "step": 3044
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.966007072782506e-05,
-      "loss": 0.9854,
+      "loss": 1.0547,
       "step": 3045
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965976616515351e-05,
-      "loss": 0.7774,
+      "loss": 0.9182,
       "step": 3046
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9659461468466415e-05,
-      "loss": 0.8972,
+      "loss": 1.0372,
       "step": 3047
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9659156637767992e-05,
-      "loss": 0.7579,
+      "loss": 0.9301,
       "step": 3048
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9658851673062475e-05,
-      "loss": 0.7584,
+      "loss": 0.9414,
       "step": 3049
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96585465743541e-05,
-      "loss": 0.8918,
+      "loss": 0.9764,
       "step": 3050
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965824134164709e-05,
-      "loss": 0.9998,
+      "loss": 1.0944,
       "step": 3051
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965793597494569e-05,
-      "loss": 0.9293,
+      "loss": 1.0214,
       "step": 3052
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9657630474254125e-05,
-      "loss": 0.9293,
+      "loss": 1.0212,
       "step": 3053
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9657324839576643e-05,
-      "loss": 1.0021,
+      "loss": 1.0059,
       "step": 3054
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9657019070917478e-05,
-      "loss": 0.7386,
+      "loss": 0.8838,
       "step": 3055
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965671316828088e-05,
-      "loss": 0.763,
+      "loss": 0.9011,
       "step": 3056
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965640713167108e-05,
-      "loss": 0.8314,
+      "loss": 0.9929,
       "step": 3057
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965610096109234e-05,
-      "loss": 0.9409,
+      "loss": 1.0135,
       "step": 3058
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9655794656548894e-05,
-      "loss": 0.8958,
+      "loss": 1.0355,
       "step": 3059
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9655488218045e-05,
-      "loss": 0.8691,
+      "loss": 0.9703,
       "step": 3060
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9655181645584903e-05,
-      "loss": 0.8499,
+      "loss": 0.8938,
       "step": 3061
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9654874939172862e-05,
-      "loss": 0.8368,
+      "loss": 0.9604,
       "step": 3062
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9654568098813128e-05,
-      "loss": 0.8817,
+      "loss": 0.9155,
       "step": 3063
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9654261124509966e-05,
-      "loss": 0.7481,
+      "loss": 0.908,
       "step": 3064
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9653954016267623e-05,
-      "loss": 0.9944,
+      "loss": 1.0179,
       "step": 3065
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9653646774090366e-05,
-      "loss": 0.8711,
+      "loss": 0.9119,
       "step": 3066
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9653339397982452e-05,
-      "loss": 0.9185,
+      "loss": 0.8765,
       "step": 3067
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9653031887948157e-05,
-      "loss": 0.7661,
+      "loss": 0.9112,
       "step": 3068
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9652724243991738e-05,
-      "loss": 0.878,
+      "loss": 0.9828,
       "step": 3069
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9652416466117466e-05,
-      "loss": 0.9578,
+      "loss": 1.0347,
       "step": 3070
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965210855432961e-05,
-      "loss": 1.0037,
+      "loss": 1.0145,
       "step": 3071
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9651800508632443e-05,
-      "loss": 0.913,
+      "loss": 1.0114,
       "step": 3072
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9651492329030238e-05,
-      "loss": 0.9013,
+      "loss": 0.9999,
       "step": 3073
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965118401552727e-05,
-      "loss": 0.7672,
+      "loss": 0.904,
       "step": 3074
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9650875568127817e-05,
-      "loss": 0.9522,
+      "loss": 1.0563,
       "step": 3075
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.965056698683616e-05,
-      "loss": 0.9003,
+      "loss": 0.9894,
       "step": 3076
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9650258271656577e-05,
-      "loss": 0.8851,
+      "loss": 1.0274,
       "step": 3077
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964994942259335e-05,
-      "loss": 0.7953,
+      "loss": 0.9266,
       "step": 3078
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964964043965077e-05,
-      "loss": 0.7831,
+      "loss": 0.8458,
       "step": 3079
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964933132283312e-05,
-      "loss": 0.9057,
+      "loss": 0.885,
       "step": 3080
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9649022072144684e-05,
-      "loss": 0.767,
+      "loss": 0.9451,
       "step": 3081
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9648712687589762e-05,
-      "loss": 0.7523,
+      "loss": 0.9089,
       "step": 3082
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9648403169172637e-05,
-      "loss": 0.8327,
+      "loss": 0.9609,
       "step": 3083
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964809351689761e-05,
-      "loss": 0.8854,
+      "loss": 0.9238,
       "step": 3084
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9647783730768973e-05,
-      "loss": 0.8278,
+      "loss": 0.9288,
       "step": 3085
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9647473810791027e-05,
-      "loss": 0.9061,
+      "loss": 0.9662,
       "step": 3086
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964716375696807e-05,
-      "loss": 0.8812,
+      "loss": 0.9226,
       "step": 3087
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96468535693044e-05,
-      "loss": 0.8943,
+      "loss": 0.9802,
       "step": 3088
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9646543247804327e-05,
-      "loss": 0.9118,
+      "loss": 1.0078,
       "step": 3089
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9646232792472153e-05,
-      "loss": 0.849,
+      "loss": 0.9956,
       "step": 3090
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9645922203312185e-05,
-      "loss": 0.8671,
+      "loss": 1.005,
       "step": 3091
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964561148032873e-05,
-      "loss": 0.9152,
+      "loss": 0.9333,
       "step": 3092
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9645300623526102e-05,
-      "loss": 0.9452,
+      "loss": 1.0166,
       "step": 3093
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9644989632908614e-05,
-      "loss": 0.9519,
+      "loss": 0.9487,
       "step": 3094
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9644678508480578e-05,
-      "loss": 0.8732,
+      "loss": 0.9318,
       "step": 3095
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9644367250246314e-05,
-      "loss": 0.9294,
+      "loss": 1.0385,
       "step": 3096
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9644055858210137e-05,
-      "loss": 0.9118,
+      "loss": 1.0941,
       "step": 3097
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9643744332376365e-05,
-      "loss": 0.8393,
+      "loss": 0.9533,
       "step": 3098
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9643432672749325e-05,
-      "loss": 0.7751,
+      "loss": 0.8279,
       "step": 3099
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9643120879333342e-05,
-      "loss": 0.7936,
+      "loss": 0.9482,
       "step": 3100
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9642808952132734e-05,
-      "loss": 0.8266,
+      "loss": 0.9321,
       "step": 3101
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9642496891151833e-05,
-      "loss": 0.8832,
+      "loss": 0.9668,
       "step": 3102
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9642184696394974e-05,
-      "loss": 0.8213,
+      "loss": 0.9114,
       "step": 3103
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964187236786648e-05,
-      "loss": 0.9429,
+      "loss": 0.9461,
       "step": 3104
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9641559905570682e-05,
-      "loss": 0.9954,
+      "loss": 1.0358,
       "step": 3105
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9641247309511927e-05,
-      "loss": 0.9693,
+      "loss": 1.0252,
       "step": 3106
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9640934579694543e-05,
-      "loss": 0.9503,
+      "loss": 1.0217,
       "step": 3107
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964062171612287e-05,
-      "loss": 0.8152,
+      "loss": 0.9642,
       "step": 3108
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.964030871880125e-05,
-      "loss": 0.9536,
+      "loss": 0.9945,
       "step": 3109
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963999558773402e-05,
-      "loss": 1.0042,
+      "loss": 1.0055,
       "step": 3110
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963968232292553e-05,
-      "loss": 0.7976,
+      "loss": 0.9324,
       "step": 3111
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9639368924380126e-05,
-      "loss": 0.8172,
+      "loss": 0.8772,
       "step": 3112
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9639055392102156e-05,
-      "loss": 0.9863,
+      "loss": 1.005,
       "step": 3113
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9638741726095966e-05,
-      "loss": 0.8434,
+      "loss": 0.9876,
       "step": 3114
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9638427926365913e-05,
-      "loss": 0.8761,
+      "loss": 0.9953,
       "step": 3115
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9638113992916344e-05,
-      "loss": 0.8863,
+      "loss": 0.9816,
       "step": 3116
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963779992575162e-05,
-      "loss": 0.8658,
+      "loss": 1.0011,
       "step": 3117
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9637485724876097e-05,
-      "loss": 0.7783,
+      "loss": 0.9192,
       "step": 3118
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9637171390294133e-05,
-      "loss": 0.8695,
+      "loss": 1.0163,
       "step": 3119
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963685692201009e-05,
-      "loss": 0.7617,
+      "loss": 0.927,
       "step": 3120
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963654232002833e-05,
-      "loss": 0.9069,
+      "loss": 1.014,
       "step": 3121
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9636227584353215e-05,
-      "loss": 0.8605,
+      "loss": 1.0472,
       "step": 3122
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9635912714989116e-05,
-      "loss": 0.7751,
+      "loss": 0.9007,
       "step": 3123
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96355977119404e-05,
-      "loss": 0.8356,
+      "loss": 0.921,
       "step": 3124
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9635282575211434e-05,
-      "loss": 0.8971,
+      "loss": 1.0168,
       "step": 3125
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96349673048066e-05,
-      "loss": 0.8503,
+      "loss": 1.0876,
       "step": 3126
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9634651900730258e-05,
-      "loss": 0.871,
+      "loss": 0.9752,
       "step": 3127
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9634336362986792e-05,
-      "loss": 0.8294,
+      "loss": 0.9623,
       "step": 3128
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963402069158058e-05,
-      "loss": 0.8089,
+      "loss": 0.9769,
       "step": 3129
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9633704886516e-05,
-      "loss": 0.9155,
+      "loss": 0.9766,
       "step": 3130
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963338894779743e-05,
-      "loss": 0.8872,
+      "loss": 0.9346,
       "step": 3131
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963307287542926e-05,
-      "loss": 0.9259,
+      "loss": 1.0438,
       "step": 3132
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963275666941587e-05,
-      "loss": 0.9517,
+      "loss": 0.9648,
       "step": 3133
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9632440329761647e-05,
-      "loss": 0.8688,
+      "loss": 0.9067,
       "step": 3134
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9632123856470982e-05,
-      "loss": 0.9187,
+      "loss": 0.9783,
       "step": 3135
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9631807249548266e-05,
-      "loss": 0.9395,
+      "loss": 1.022,
       "step": 3136
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.963149050899789e-05,
-      "loss": 0.9615,
+      "loss": 1.0008,
       "step": 3137
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9631173634824247e-05,
-      "loss": 0.7526,
+      "loss": 0.9848,
       "step": 3138
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9630856627031736e-05,
-      "loss": 0.738,
+      "loss": 0.8623,
       "step": 3139
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9630539485624752e-05,
-      "loss": 0.8675,
+      "loss": 0.951,
       "step": 3140
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96302222106077e-05,
-      "loss": 0.9145,
+      "loss": 0.9328,
       "step": 3141
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9629904801984975e-05,
-      "loss": 0.8749,
+      "loss": 1.0408,
       "step": 3142
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9629587259760983e-05,
-      "loss": 0.8442,
+      "loss": 1.0006,
       "step": 3143
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9629269583940137e-05,
-      "loss": 0.8135,
+      "loss": 0.9591,
       "step": 3144
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962895177452683e-05,
-      "loss": 0.8285,
+      "loss": 0.9948,
       "step": 3145
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9628633831525482e-05,
-      "loss": 0.8824,
+      "loss": 1.0035,
       "step": 3146
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96283157549405e-05,
-      "loss": 0.8215,
+      "loss": 1.026,
       "step": 3147
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.96279975447763e-05,
-      "loss": 0.821,
+      "loss": 0.9318,
       "step": 3148
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9627679201037294e-05,
-      "loss": 0.8154,
+      "loss": 0.9801,
       "step": 3149
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9627360723727896e-05,
-      "loss": 0.7964,
+      "loss": 0.8978,
       "step": 3150
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962704211285253e-05,
-      "loss": 0.8893,
+      "loss": 1.014,
       "step": 3151
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962672336841561e-05,
-      "loss": 0.9462,
+      "loss": 0.9892,
       "step": 3152
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9626404490421563e-05,
-      "loss": 0.752,
+      "loss": 0.8655,
       "step": 3153
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9626085478874817e-05,
-      "loss": 0.7895,
+      "loss": 0.9424,
       "step": 3154
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962576633377979e-05,
-      "loss": 0.9078,
+      "loss": 0.9992,
       "step": 3155
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962544705514091e-05,
-      "loss": 0.779,
+      "loss": 0.8641,
       "step": 3156
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962512764296261e-05,
-      "loss": 0.9304,
+      "loss": 1.0097,
       "step": 3157
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.962480809724932e-05,
-      "loss": 0.9641,
+      "loss": 1.0352,
       "step": 3158
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9624488418005473e-05,
-      "loss": 0.9187,
+      "loss": 0.9819,
       "step": 3159
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9624168605235504e-05,
-      "loss": 0.8402,
+      "loss": 0.9685,
       "step": 3160
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9623848658943855e-05,
-      "loss": 0.8466,
+      "loss": 0.9722,
       "step": 3161
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9623528579134955e-05,
-      "loss": 0.8302,
+      "loss": 0.9444,
       "step": 3162
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9623208365813253e-05,
-      "loss": 0.8236,
+      "loss": 0.9191,
       "step": 3163
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9622888018983186e-05,
-      "loss": 0.9317,
+      "loss": 0.9871,
       "step": 3164
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9622567538649204e-05,
-      "loss": 0.9426,
+      "loss": 0.9415,
       "step": 3165
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9622246924815748e-05,
-      "loss": 0.8219,
+      "loss": 1.0616,
       "step": 3166
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.962192617748727e-05,
-      "loss": 0.7462,
+      "loss": 0.9473,
       "step": 3167
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9621605296668217e-05,
-      "loss": 1.0306,
+      "loss": 1.0341,
       "step": 3168
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9621284282363042e-05,
-      "loss": 0.969,
+      "loss": 1.0229,
       "step": 3169
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9620963134576198e-05,
-      "loss": 0.9336,
+      "loss": 1.0535,
       "step": 3170
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.962064185331214e-05,
-      "loss": 0.7699,
+      "loss": 0.8506,
       "step": 3171
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9620320438575327e-05,
-      "loss": 0.9368,
+      "loss": 0.9825,
       "step": 3172
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9619998890370215e-05,
-      "loss": 0.6603,
+      "loss": 0.9133,
       "step": 3173
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961967720870127e-05,
-      "loss": 0.84,
+      "loss": 0.92,
       "step": 3174
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9619355393572952e-05,
-      "loss": 1.0092,
+      "loss": 1.0609,
       "step": 3175
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9619033444989724e-05,
-      "loss": 0.8047,
+      "loss": 0.9331,
       "step": 3176
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9618711362956056e-05,
-      "loss": 0.9371,
+      "loss": 1.0374,
       "step": 3177
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9618389147476417e-05,
-      "loss": 0.8919,
+      "loss": 0.911,
       "step": 3178
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9618066798555274e-05,
-      "loss": 0.8815,
+      "loss": 0.9486,
       "step": 3179
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.96177443161971e-05,
-      "loss": 0.9791,
+      "loss": 0.989,
       "step": 3180
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961742170040637e-05,
-      "loss": 0.8705,
+      "loss": 0.9536,
       "step": 3181
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9617098951187558e-05,
-      "loss": 0.8499,
+      "loss": 0.9447,
       "step": 3182
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9616776068545144e-05,
-      "loss": 0.7899,
+      "loss": 0.9362,
       "step": 3183
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9616453052483607e-05,
-      "loss": 0.6791,
+      "loss": 0.951,
       "step": 3184
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9616129903007425e-05,
-      "loss": 0.9539,
+      "loss": 1.0419,
       "step": 3185
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961580662012109e-05,
-      "loss": 0.826,
+      "loss": 0.9492,
       "step": 3186
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9615483203829076e-05,
-      "loss": 0.7235,
+      "loss": 0.9125,
       "step": 3187
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9615159654135877e-05,
-      "loss": 0.9365,
+      "loss": 0.9712,
       "step": 3188
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9614835971045977e-05,
-      "loss": 0.7999,
+      "loss": 0.9796,
       "step": 3189
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9614512154563875e-05,
-      "loss": 1.0644,
+      "loss": 1.0742,
       "step": 3190
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9614188204694053e-05,
-      "loss": 0.8572,
+      "loss": 0.9658,
       "step": 3191
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9613864121441015e-05,
-      "loss": 0.8796,
+      "loss": 1.0089,
       "step": 3192
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961353990480925e-05,
-      "loss": 0.9106,
+      "loss": 0.999,
       "step": 3193
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9613215554803258e-05,
-      "loss": 0.8622,
+      "loss": 0.9692,
       "step": 3194
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961289107142754e-05,
-      "loss": 0.8236,
+      "loss": 0.8848,
       "step": 3195
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9612566454686598e-05,
-      "loss": 0.9861,
+      "loss": 1.0216,
       "step": 3196
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9612241704584934e-05,
-      "loss": 0.7417,
+      "loss": 0.9549,
       "step": 3197
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9611916821127053e-05,
-      "loss": 0.9411,
+      "loss": 1.0344,
       "step": 3198
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9611591804317464e-05,
-      "loss": 0.8419,
+      "loss": 0.9077,
       "step": 3199
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961126665416068e-05,
-      "loss": 0.817,
+      "loss": 0.9228,
       "step": 3200
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9610941370661205e-05,
-      "loss": 0.8719,
+      "loss": 0.9572,
       "step": 3201
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9610615953823554e-05,
-      "loss": 0.8058,
+      "loss": 0.8574,
       "step": 3202
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.961029040365224e-05,
-      "loss": 0.8159,
+      "loss": 0.9086,
       "step": 3203
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9609964720151788e-05,
-      "loss": 0.8954,
+      "loss": 0.9815,
       "step": 3204
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9609638903326703e-05,
-      "loss": 0.7605,
+      "loss": 0.9215,
       "step": 3205
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960931295318152e-05,
-      "loss": 0.8812,
+      "loss": 0.9385,
       "step": 3206
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9608986869720748e-05,
-      "loss": 0.7802,
+      "loss": 0.9261,
       "step": 3207
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960866065294892e-05,
-      "loss": 0.8612,
+      "loss": 0.9978,
       "step": 3208
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960833430287056e-05,
-      "loss": 0.9785,
+      "loss": 0.999,
       "step": 3209
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960800781949019e-05,
-      "loss": 0.753,
+      "loss": 0.9446,
       "step": 3210
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9607681202812342e-05,
-      "loss": 0.9116,
+      "loss": 0.9784,
       "step": 3211
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9607354452841553e-05,
-      "loss": 0.8975,
+      "loss": 0.9636,
       "step": 3212
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960702756958235e-05,
-      "loss": 0.7728,
+      "loss": 0.9424,
       "step": 3213
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960670055303927e-05,
-      "loss": 0.6329,
+      "loss": 0.9034,
       "step": 3214
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960637340321685e-05,
-      "loss": 0.8091,
+      "loss": 0.9598,
       "step": 3215
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960604612011963e-05,
-      "loss": 0.7988,
+      "loss": 0.8808,
       "step": 3216
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9605718703752147e-05,
-      "loss": 0.7516,
+      "loss": 0.9378,
       "step": 3217
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9605391154118948e-05,
-      "loss": 0.8864,
+      "loss": 0.9659,
       "step": 3218
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9605063471224573e-05,
-      "loss": 0.8252,
+      "loss": 0.8824,
       "step": 3219
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960473565507357e-05,
-      "loss": 0.8672,
+      "loss": 0.9585,
       "step": 3220
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960440770567049e-05,
-      "loss": 0.7479,
+      "loss": 0.8618,
       "step": 3221
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960407962301988e-05,
-      "loss": 0.8324,
+      "loss": 0.9419,
       "step": 3222
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9603751407126287e-05,
-      "loss": 0.9248,
+      "loss": 0.9561,
       "step": 3223
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9603423057994275e-05,
-      "loss": 0.8607,
+      "loss": 0.926,
       "step": 3224
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960309457562839e-05,
-      "loss": 0.9434,
+      "loss": 1.0259,
       "step": 3225
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9602765960033193e-05,
-      "loss": 0.8195,
+      "loss": 0.9413,
       "step": 3226
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9602437211213244e-05,
-      "loss": 0.7843,
+      "loss": 0.9269,
       "step": 3227
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.96021083291731e-05,
-      "loss": 0.9945,
+      "loss": 1.0557,
       "step": 3228
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960177931391733e-05,
-      "loss": 0.9955,
+      "loss": 1.0558,
       "step": 3229
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9601450165450493e-05,
-      "loss": 0.86,
+      "loss": 0.8784,
       "step": 3230
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9601120883777158e-05,
-      "loss": 0.8712,
+      "loss": 0.9326,
       "step": 3231
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9600791468901894e-05,
-      "loss": 0.9158,
+      "loss": 0.9738,
       "step": 3232
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.960046192082927e-05,
-      "loss": 0.8862,
+      "loss": 0.9501,
       "step": 3233
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9600132239563855e-05,
-      "loss": 0.8725,
+      "loss": 0.9327,
       "step": 3234
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959980242511023e-05,
-      "loss": 0.7902,
+      "loss": 1.0031,
       "step": 3235
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9599472477472965e-05,
-      "loss": 0.8555,
+      "loss": 0.9844,
       "step": 3236
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9599142396656637e-05,
-      "loss": 0.9177,
+      "loss": 1.0043,
       "step": 3237
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9598812182665835e-05,
-      "loss": 0.8779,
+      "loss": 0.9721,
       "step": 3238
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9598481835505127e-05,
-      "loss": 0.9145,
+      "loss": 1.011,
       "step": 3239
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9598151355179103e-05,
-      "loss": 0.8791,
+      "loss": 0.9316,
       "step": 3240
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9597820741692347e-05,
-      "loss": 0.8813,
+      "loss": 0.9281,
       "step": 3241
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9597489995049444e-05,
-      "loss": 0.8035,
+      "loss": 0.9091,
       "step": 3242
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9597159115254988e-05,
-      "loss": 0.8391,
+      "loss": 0.9725,
       "step": 3243
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9596828102313562e-05,
-      "loss": 0.8964,
+      "loss": 0.9795,
       "step": 3244
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9596496956229765e-05,
-      "loss": 0.8772,
+      "loss": 0.9557,
       "step": 3245
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9596165677008188e-05,
-      "loss": 0.8531,
+      "loss": 0.9672,
       "step": 3246
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9595834264653428e-05,
-      "loss": 0.8351,
+      "loss": 0.9203,
       "step": 3247
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959550271917008e-05,
-      "loss": 0.8471,
+      "loss": 0.9749,
       "step": 3248
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9595171040562747e-05,
-      "loss": 0.8154,
+      "loss": 0.9709,
       "step": 3249
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9594839228836028e-05,
-      "loss": 0.9338,
+      "loss": 1.0412,
       "step": 3250
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9594507283994534e-05,
-      "loss": 0.9395,
+      "loss": 0.9347,
       "step": 3251
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959417520604286e-05,
-      "loss": 0.9262,
+      "loss": 1.0313,
       "step": 3252
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9593842994985615e-05,
-      "loss": 0.8692,
+      "loss": 0.9744,
       "step": 3253
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9593510650827413e-05,
-      "loss": 0.7394,
+      "loss": 0.8025,
       "step": 3254
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959317817357286e-05,
-      "loss": 0.7926,
+      "loss": 0.9835,
       "step": 3255
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9592845563226572e-05,
-      "loss": 0.8368,
+      "loss": 0.8594,
       "step": 3256
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9592512819793166e-05,
-      "loss": 0.908,
+      "loss": 0.9599,
       "step": 3257
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959217994327725e-05,
-      "loss": 0.8843,
+      "loss": 0.92,
       "step": 3258
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9591846933683446e-05,
-      "loss": 0.82,
+      "loss": 0.93,
       "step": 3259
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959151379101638e-05,
-      "loss": 0.857,
+      "loss": 0.9292,
       "step": 3260
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9591180515280667e-05,
-      "loss": 0.9759,
+      "loss": 0.9877,
       "step": 3261
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.959084710648093e-05,
-      "loss": 0.7884,
+      "loss": 0.917,
       "step": 3262
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95905135646218e-05,
-      "loss": 0.8622,
+      "loss": 0.9438,
       "step": 3263
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9590179889707897e-05,
-      "loss": 0.883,
+      "loss": 0.9995,
       "step": 3264
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958984608174386e-05,
-      "loss": 0.9434,
+      "loss": 0.9821,
       "step": 3265
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958951214073431e-05,
-      "loss": 0.9486,
+      "loss": 0.9995,
       "step": 3266
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958917806668389e-05,
-      "loss": 0.8585,
+      "loss": 0.9508,
       "step": 3267
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9588843859597223e-05,
-      "loss": 0.9465,
+      "loss": 1.0119,
       "step": 3268
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958850951947896e-05,
-      "loss": 0.9558,
+      "loss": 1.0059,
       "step": 3269
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9588175046333727e-05,
-      "loss": 0.9076,
+      "loss": 0.9923,
       "step": 3270
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958784044016617e-05,
-      "loss": 0.9509,
+      "loss": 0.9861,
       "step": 3271
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958750570098093e-05,
-      "loss": 0.9844,
+      "loss": 1.0524,
       "step": 3272
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9587170828782652e-05,
-      "loss": 0.8199,
+      "loss": 0.9405,
       "step": 3273
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9586835823575982e-05,
-      "loss": 0.9073,
+      "loss": 0.9776,
       "step": 3274
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9586500685365562e-05,
-      "loss": 0.8619,
+      "loss": 1.0854,
       "step": 3275
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958616541415605e-05,
-      "loss": 0.8614,
+      "loss": 0.9651,
       "step": 3276
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9585830009952095e-05,
-      "loss": 0.7379,
+      "loss": 0.9687,
       "step": 3277
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958549447275835e-05,
-      "loss": 0.8557,
+      "loss": 0.9685,
       "step": 3278
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9585158802579464e-05,
-      "loss": 0.8546,
+      "loss": 0.9507,
       "step": 3279
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95848229994201e-05,
-      "loss": 0.7411,
+      "loss": 0.9347,
       "step": 3280
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958448706328492e-05,
-      "loss": 0.8056,
+      "loss": 0.9976,
       "step": 3281
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958415099417858e-05,
-      "loss": 0.8659,
+      "loss": 0.9601,
       "step": 3282
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9583814792105743e-05,
-      "loss": 0.8072,
+      "loss": 0.9289,
       "step": 3283
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958347845707107e-05,
-      "loss": 0.9122,
+      "loss": 1.0245,
       "step": 3284
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9583141989079232e-05,
-      "loss": 0.8133,
+      "loss": 0.9308,
       "step": 3285
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9582805388134896e-05,
-      "loss": 0.9467,
+      "loss": 1.0322,
       "step": 3286
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9582468654242736e-05,
-      "loss": 0.926,
+      "loss": 1.0266,
       "step": 3287
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9582131787407413e-05,
-      "loss": 0.8686,
+      "loss": 0.9783,
       "step": 3288
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.958179478763361e-05,
-      "loss": 0.7994,
+      "loss": 0.913,
       "step": 3289
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9581457654925998e-05,
-      "loss": 0.7718,
+      "loss": 0.9501,
       "step": 3290
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9581120389289254e-05,
-      "loss": 0.886,
+      "loss": 1.0,
       "step": 3291
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9580782990728063e-05,
-      "loss": 0.7545,
+      "loss": 0.9002,
       "step": 3292
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9580445459247098e-05,
-      "loss": 0.9123,
+      "loss": 0.9714,
       "step": 3293
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9580107794851044e-05,
-      "loss": 0.9583,
+      "loss": 1.0583,
       "step": 3294
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957976999754459e-05,
-      "loss": 0.649,
+      "loss": 0.825,
       "step": 3295
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9579432067332416e-05,
-      "loss": 0.8676,
+      "loss": 0.9428,
       "step": 3296
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9579094004219214e-05,
-      "loss": 0.8771,
+      "loss": 0.9488,
       "step": 3297
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9578755808209673e-05,
-      "loss": 0.8698,
+      "loss": 1.0179,
       "step": 3298
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9578417479308486e-05,
-      "loss": 0.9319,
+      "loss": 1.0416,
       "step": 3299
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957807901752035e-05,
-      "loss": 0.7849,
+      "loss": 0.8698,
       "step": 3300
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9577740422849954e-05,
-      "loss": 0.7871,
+      "loss": 1.0282,
       "step": 3301
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9577401695302e-05,
-      "loss": 0.7194,
+      "loss": 0.8419,
       "step": 3302
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9577062834881184e-05,
-      "loss": 0.9407,
+      "loss": 0.9594,
       "step": 3303
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9576723841592212e-05,
-      "loss": 0.9549,
+      "loss": 1.0231,
       "step": 3304
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957638471543978e-05,
-      "loss": 0.9342,
+      "loss": 1.0902,
       "step": 3305
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9576045456428603e-05,
-      "loss": 0.7986,
+      "loss": 0.9227,
       "step": 3306
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9575706064563378e-05,
-      "loss": 1.0282,
+      "loss": 1.1091,
       "step": 3307
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9575366539848818e-05,
-      "loss": 0.992,
+      "loss": 1.024,
       "step": 3308
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9575026882289632e-05,
-      "loss": 0.9058,
+      "loss": 1.02,
       "step": 3309
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9574687091890537e-05,
-      "loss": 0.8577,
+      "loss": 1.0159,
       "step": 3310
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957434716865624e-05,
-      "loss": 0.9695,
+      "loss": 1.0238,
       "step": 3311
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9574007112591463e-05,
-      "loss": 0.9166,
+      "loss": 1.0305,
       "step": 3312
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9573666923700917e-05,
-      "loss": 0.8757,
+      "loss": 1.0214,
       "step": 3313
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957332660198933e-05,
-      "loss": 0.8666,
+      "loss": 1.0264,
       "step": 3314
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9572986147461415e-05,
-      "loss": 1.0671,
+      "loss": 1.0826,
       "step": 3315
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95726455601219e-05,
-      "loss": 0.8371,
+      "loss": 1.0088,
       "step": 3316
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9572304839975513e-05,
-      "loss": 0.9232,
+      "loss": 0.9779,
       "step": 3317
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9571963987026975e-05,
-      "loss": 0.8109,
+      "loss": 0.9502,
       "step": 3318
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957162300128102e-05,
-      "loss": 0.7683,
+      "loss": 0.8574,
       "step": 3319
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9571281882742374e-05,
-      "loss": 0.7856,
+      "loss": 0.8951,
       "step": 3320
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9570940631415773e-05,
-      "loss": 0.8841,
+      "loss": 0.9818,
       "step": 3321
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957059924730595e-05,
-      "loss": 0.8587,
+      "loss": 0.9124,
       "step": 3322
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.957025773041764e-05,
-      "loss": 0.7951,
+      "loss": 0.9266,
       "step": 3323
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9569916080755583e-05,
-      "loss": 0.7645,
+      "loss": 0.9429,
       "step": 3324
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9569574298324518e-05,
-      "loss": 0.9079,
+      "loss": 1.0299,
       "step": 3325
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9569232383129187e-05,
-      "loss": 0.7415,
+      "loss": 0.9638,
       "step": 3326
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9568890335174336e-05,
-      "loss": 1.047,
+      "loss": 1.085,
       "step": 3327
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9568548154464707e-05,
-      "loss": 0.8721,
+      "loss": 0.9107,
       "step": 3328
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9568205841005045e-05,
-      "loss": 0.8039,
+      "loss": 0.9729,
       "step": 3329
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9567863394800107e-05,
-      "loss": 0.8982,
+      "loss": 1.0513,
       "step": 3330
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9567520815854637e-05,
-      "loss": 0.8655,
+      "loss": 0.9753,
       "step": 3331
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9567178104173388e-05,
-      "loss": 0.8389,
+      "loss": 0.9622,
       "step": 3332
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9566835259761125e-05,
-      "loss": 0.8292,
+      "loss": 0.9373,
       "step": 3333
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.956649228262259e-05,
-      "loss": 0.8503,
+      "loss": 0.9464,
       "step": 3334
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9566149172762547e-05,
-      "loss": 0.844,
+      "loss": 0.9771,
       "step": 3335
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.956580593018576e-05,
-      "loss": 0.9934,
+      "loss": 1.0661,
       "step": 3336
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9565462554896986e-05,
-      "loss": 0.7814,
+      "loss": 0.9173,
       "step": 3337
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.956511904690099e-05,
-      "loss": 0.931,
+      "loss": 1.029,
       "step": 3338
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9564775406202542e-05,
-      "loss": 0.9073,
+      "loss": 1.0421,
       "step": 3339
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95644316328064e-05,
-      "loss": 0.7715,
+      "loss": 0.9191,
       "step": 3340
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9564087726717344e-05,
-      "loss": 0.806,
+      "loss": 0.9437,
       "step": 3341
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9563743687940137e-05,
-      "loss": 0.8471,
+      "loss": 1.0065,
       "step": 3342
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.956339951647956e-05,
-      "loss": 1.05,
+      "loss": 1.0707,
       "step": 3343
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9563055212340378e-05,
-      "loss": 0.8197,
+      "loss": 0.8882,
       "step": 3344
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9562710775527377e-05,
-      "loss": 0.7233,
+      "loss": 0.8771,
       "step": 3345
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9562366206045333e-05,
-      "loss": 0.813,
+      "loss": 0.8093,
       "step": 3346
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9562021503899024e-05,
-      "loss": 0.9032,
+      "loss": 0.9956,
       "step": 3347
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9561676669093226e-05,
-      "loss": 0.8545,
+      "loss": 0.9052,
       "step": 3348
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.956133170163274e-05,
-      "loss": 0.8057,
+      "loss": 0.9475,
       "step": 3349
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9560986601522338e-05,
-      "loss": 0.8672,
+      "loss": 0.9025,
       "step": 3350
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9560641368766814e-05,
-      "loss": 0.8538,
+      "loss": 1.0242,
       "step": 3351
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9560296003370953e-05,
-      "loss": 0.8747,
+      "loss": 0.9364,
       "step": 3352
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9559950505339554e-05,
-      "loss": 0.9432,
+      "loss": 0.9251,
       "step": 3353
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95596048746774e-05,
-      "loss": 0.8666,
+      "loss": 0.9244,
       "step": 3354
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9559259111389293e-05,
-      "loss": 0.803,
+      "loss": 0.9287,
       "step": 3355
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9558913215480034e-05,
-      "loss": 0.9654,
+      "loss": 1.0781,
       "step": 3356
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.955856718695441e-05,
-      "loss": 0.9419,
+      "loss": 1.028,
       "step": 3357
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.955822102581723e-05,
-      "loss": 0.751,
+      "loss": 0.8879,
       "step": 3358
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95578747320733e-05,
-      "loss": 0.8672,
+      "loss": 1.0303,
       "step": 3359
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.955752830572741e-05,
-      "loss": 0.8409,
+      "loss": 0.9051,
       "step": 3360
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9557181746784383e-05,
-      "loss": 0.9062,
+      "loss": 0.9896,
       "step": 3361
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9556835055249013e-05,
-      "loss": 0.8414,
+      "loss": 0.929,
       "step": 3362
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9556488231126118e-05,
-      "loss": 1.0218,
+      "loss": 1.0576,
       "step": 3363
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9556141274420507e-05,
-      "loss": 0.696,
+      "loss": 0.8693,
       "step": 3364
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9555794185136994e-05,
-      "loss": 0.8629,
+      "loss": 0.9525,
       "step": 3365
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95554469632804e-05,
-      "loss": 0.8149,
+      "loss": 0.9717,
       "step": 3366
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.955509960885553e-05,
-      "loss": 0.8269,
+      "loss": 0.9968,
       "step": 3367
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9554752121867214e-05,
-      "loss": 0.9294,
+      "loss": 0.9992,
       "step": 3368
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9554404502320266e-05,
-      "loss": 1.0015,
+      "loss": 1.0515,
       "step": 3369
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9554056750219513e-05,
-      "loss": 0.8641,
+      "loss": 1.0176,
       "step": 3370
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.955370886556978e-05,
-      "loss": 0.932,
+      "loss": 1.0369,
       "step": 3371
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9553360848375886e-05,
-      "loss": 0.8492,
+      "loss": 1.0417,
       "step": 3372
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9553012698642672e-05,
-      "loss": 0.9014,
+      "loss": 1.0847,
       "step": 3373
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9552664416374954e-05,
-      "loss": 0.8923,
+      "loss": 1.0033,
       "step": 3374
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9552316001577575e-05,
-      "loss": 0.7914,
+      "loss": 0.9752,
       "step": 3375
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9551967454255362e-05,
-      "loss": 0.9267,
+      "loss": 1.0102,
       "step": 3376
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9551618774413154e-05,
-      "loss": 0.8214,
+      "loss": 0.9726,
       "step": 3377
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9551269962055785e-05,
-      "loss": 0.7774,
+      "loss": 0.9712,
       "step": 3378
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95509210171881e-05,
-      "loss": 0.8675,
+      "loss": 0.9128,
       "step": 3379
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9550571939814937e-05,
-      "loss": 0.8796,
+      "loss": 1.036,
       "step": 3380
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9550222729941137e-05,
-      "loss": 0.9098,
+      "loss": 1.0127,
       "step": 3381
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9549873387571546e-05,
-      "loss": 0.917,
+      "loss": 1.0223,
       "step": 3382
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9549523912711014e-05,
-      "loss": 0.8129,
+      "loss": 0.954,
       "step": 3383
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9549174305364384e-05,
-      "loss": 0.7485,
+      "loss": 0.8618,
       "step": 3384
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954882456553651e-05,
-      "loss": 0.8313,
+      "loss": 0.9366,
       "step": 3385
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9548474693232245e-05,
-      "loss": 0.8372,
+      "loss": 0.9357,
       "step": 3386
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954812468845644e-05,
-      "loss": 0.8705,
+      "loss": 0.9429,
       "step": 3387
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954777455121395e-05,
-      "loss": 0.8136,
+      "loss": 0.9234,
       "step": 3388
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9547424281509636e-05,
-      "loss": 0.8751,
+      "loss": 0.9317,
       "step": 3389
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9547073879348356e-05,
-      "loss": 0.7966,
+      "loss": 0.893,
       "step": 3390
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9546723344734973e-05,
-      "loss": 0.7259,
+      "loss": 0.8581,
       "step": 3391
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9546372677674347e-05,
-      "loss": 0.7732,
+      "loss": 0.9669,
       "step": 3392
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9546021878171345e-05,
-      "loss": 0.9607,
+      "loss": 1.0253,
       "step": 3393
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9545670946230832e-05,
-      "loss": 0.7888,
+      "loss": 0.86,
       "step": 3394
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954531988185768e-05,
-      "loss": 0.8911,
+      "loss": 0.9857,
       "step": 3395
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9544968685056757e-05,
-      "loss": 0.8884,
+      "loss": 1.0036,
       "step": 3396
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9544617355832936e-05,
-      "loss": 0.8189,
+      "loss": 0.9631,
       "step": 3397
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9544265894191093e-05,
-      "loss": 0.8062,
+      "loss": 0.9978,
       "step": 3398
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95439143001361e-05,
-      "loss": 1.0028,
+      "loss": 1.0508,
       "step": 3399
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954356257367284e-05,
-      "loss": 0.8784,
+      "loss": 0.929,
       "step": 3400
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9543210714806188e-05,
-      "loss": 0.9103,
+      "loss": 0.9783,
       "step": 3401
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9542858723541026e-05,
-      "loss": 0.9163,
+      "loss": 1.0134,
       "step": 3402
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954250659988224e-05,
-      "loss": 0.8638,
+      "loss": 0.9956,
       "step": 3403
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954215434383471e-05,
-      "loss": 0.7905,
+      "loss": 0.9901,
       "step": 3404
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9541801955403334e-05,
-      "loss": 0.7924,
+      "loss": 0.9006,
       "step": 3405
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954144943459299e-05,
-      "loss": 0.7604,
+      "loss": 0.9841,
       "step": 3406
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9541096781408572e-05,
-      "loss": 0.8141,
+      "loss": 0.9265,
       "step": 3407
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9540743995854978e-05,
-      "loss": 0.8796,
+      "loss": 0.9423,
       "step": 3408
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9540391077937093e-05,
-      "loss": 0.915,
+      "loss": 1.0504,
       "step": 3409
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.954003802765982e-05,
-      "loss": 0.8305,
+      "loss": 0.9876,
       "step": 3410
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9539684845028052e-05,
-      "loss": 0.7029,
+      "loss": 0.8584,
       "step": 3411
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9539331530046695e-05,
-      "loss": 0.8619,
+      "loss": 0.9241,
       "step": 3412
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9538978082720645e-05,
-      "loss": 0.8025,
+      "loss": 0.9828,
       "step": 3413
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953862450305481e-05,
-      "loss": 0.7865,
+      "loss": 0.8499,
       "step": 3414
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9538270791054095e-05,
-      "loss": 0.9713,
+      "loss": 0.9696,
       "step": 3415
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9537916946723404e-05,
-      "loss": 0.9225,
+      "loss": 1.017,
       "step": 3416
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9537562970067644e-05,
-      "loss": 0.8006,
+      "loss": 1.0342,
       "step": 3417
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9537208861091738e-05,
-      "loss": 0.7804,
+      "loss": 0.881,
       "step": 3418
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9536854619800584e-05,
-      "loss": 0.7932,
+      "loss": 0.892,
       "step": 3419
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9536500246199106e-05,
-      "loss": 0.8724,
+      "loss": 0.9385,
       "step": 3420
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9536145740292217e-05,
-      "loss": 0.8177,
+      "loss": 0.9342,
       "step": 3421
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9535791102084837e-05,
-      "loss": 0.7418,
+      "loss": 0.9221,
       "step": 3422
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953543633158188e-05,
-      "loss": 0.7647,
+      "loss": 0.9203,
       "step": 3423
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9535081428788278e-05,
-      "loss": 0.8405,
+      "loss": 0.9901,
       "step": 3424
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953472639370895e-05,
-      "loss": 0.7947,
+      "loss": 0.8549,
       "step": 3425
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953437122634882e-05,
-      "loss": 0.8975,
+      "loss": 1.0238,
       "step": 3426
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9534015926712816e-05,
-      "loss": 0.9102,
+      "loss": 0.9595,
       "step": 3427
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953366049480587e-05,
-      "loss": 0.7907,
+      "loss": 0.9143,
       "step": 3428
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953330493063291e-05,
-      "loss": 0.8404,
+      "loss": 0.8999,
       "step": 3429
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9532949234198865e-05,
-      "loss": 0.7766,
+      "loss": 0.9052,
       "step": 3430
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9532593405508683e-05,
-      "loss": 0.8351,
+      "loss": 0.996,
       "step": 3431
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953223744456729e-05,
-      "loss": 0.9283,
+      "loss": 1.0005,
       "step": 3432
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9531881351379624e-05,
-      "loss": 0.8302,
+      "loss": 0.939,
       "step": 3433
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953152512595063e-05,
-      "loss": 0.902,
+      "loss": 0.9411,
       "step": 3434
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.953116876828525e-05,
-      "loss": 0.8862,
+      "loss": 0.9308,
       "step": 3435
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9530812278388422e-05,
-      "loss": 0.788,
+      "loss": 0.8968,
       "step": 3436
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.95304556562651e-05,
-      "loss": 0.8038,
+      "loss": 0.9296,
       "step": 3437
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9530098901920227e-05,
-      "loss": 0.8763,
+      "loss": 0.9508,
       "step": 3438
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9529742015358752e-05,
-      "loss": 0.9223,
+      "loss": 1.0546,
       "step": 3439
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9529384996585626e-05,
-      "loss": 0.7506,
+      "loss": 0.8247,
       "step": 3440
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952902784560581e-05,
-      "loss": 0.7926,
+      "loss": 0.9404,
       "step": 3441
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952867056242425e-05,
-      "loss": 0.8245,
+      "loss": 0.8954,
       "step": 3442
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9528313147045902e-05,
-      "loss": 0.8363,
+      "loss": 0.9197,
       "step": 3443
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952795559947573e-05,
-      "loss": 0.9147,
+      "loss": 0.9746,
       "step": 3444
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9527597919718692e-05,
-      "loss": 0.7357,
+      "loss": 0.8939,
       "step": 3445
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952724010777975e-05,
-      "loss": 0.8673,
+      "loss": 0.981,
       "step": 3446
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952688216366387e-05,
-      "loss": 0.7667,
+      "loss": 0.9606,
       "step": 3447
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9526524087376018e-05,
-      "loss": 0.7722,
+      "loss": 0.9181,
       "step": 3448
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952616587892116e-05,
-      "loss": 0.9255,
+      "loss": 0.996,
       "step": 3449
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9525807538304266e-05,
-      "loss": 0.7494,
+      "loss": 0.8821,
       "step": 3450
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9525449065530306e-05,
-      "loss": 0.7544,
+      "loss": 0.8587,
       "step": 3451
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9525090460604255e-05,
-      "loss": 0.9093,
+      "loss": 0.9798,
       "step": 3452
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9524731723531093e-05,
-      "loss": 0.7575,
+      "loss": 0.8876,
       "step": 3453
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9524372854315786e-05,
-      "loss": 0.7451,
+      "loss": 0.8819,
       "step": 3454
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9524013852963323e-05,
-      "loss": 0.8452,
+      "loss": 0.946,
       "step": 3455
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952365471947868e-05,
-      "loss": 0.8962,
+      "loss": 1.0091,
       "step": 3456
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9523295453866836e-05,
-      "loss": 0.9866,
+      "loss": 1.0634,
       "step": 3457
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9522936056132783e-05,
-      "loss": 0.891,
+      "loss": 0.9817,
       "step": 3458
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952257652628151e-05,
-      "loss": 0.9042,
+      "loss": 0.9947,
       "step": 3459
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9522216864317987e-05,
-      "loss": 0.8246,
+      "loss": 0.952,
       "step": 3460
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952185707024722e-05,
-      "loss": 0.8458,
+      "loss": 0.9361,
       "step": 3461
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9521497144074197e-05,
-      "loss": 0.881,
+      "loss": 0.957,
       "step": 3462
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.952113708580391e-05,
-      "loss": 0.8896,
+      "loss": 1.0408,
       "step": 3463
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9520776895441353e-05,
-      "loss": 0.8121,
+      "loss": 0.9548,
       "step": 3464
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9520416572991525e-05,
-      "loss": 0.9085,
+      "loss": 1.0106,
       "step": 3465
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9520056118459427e-05,
-      "loss": 0.8106,
+      "loss": 0.9351,
       "step": 3466
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9519695531850055e-05,
-      "loss": 0.893,
+      "loss": 0.9704,
       "step": 3467
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9519334813168415e-05,
-      "loss": 0.896,
+      "loss": 1.0433,
       "step": 3468
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9518973962419513e-05,
-      "loss": 0.8017,
+      "loss": 0.9207,
       "step": 3469
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9518612979608348e-05,
-      "loss": 0.9422,
+      "loss": 0.983,
       "step": 3470
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9518251864739936e-05,
-      "loss": 0.8816,
+      "loss": 0.9756,
       "step": 3471
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9517890617819287e-05,
-      "loss": 0.8224,
+      "loss": 0.9133,
       "step": 3472
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9517529238851406e-05,
-      "loss": 0.7399,
+      "loss": 0.8601,
       "step": 3473
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951716772784131e-05,
-      "loss": 0.8505,
+      "loss": 0.9693,
       "step": 3474
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9516806084794018e-05,
-      "loss": 0.8091,
+      "loss": 0.9399,
       "step": 3475
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9516444309714543e-05,
-      "loss": 0.886,
+      "loss": 1.0042,
       "step": 3476
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9516082402607904e-05,
-      "loss": 0.9801,
+      "loss": 1.0822,
       "step": 3477
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9515720363479122e-05,
-      "loss": 0.8885,
+      "loss": 1.0836,
       "step": 3478
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951535819233322e-05,
-      "loss": 0.8803,
+      "loss": 0.9495,
       "step": 3479
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951499588917523e-05,
-      "loss": 0.9397,
+      "loss": 0.9723,
       "step": 3480
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951463345401017e-05,
-      "loss": 0.7477,
+      "loss": 0.9122,
       "step": 3481
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9514270886843064e-05,
-      "loss": 0.8532,
+      "loss": 0.9556,
       "step": 3482
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9513908187678955e-05,
-      "loss": 0.8677,
+      "loss": 0.8765,
       "step": 3483
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9513545356522864e-05,
-      "loss": 0.8907,
+      "loss": 0.9196,
       "step": 3484
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9513182393379836e-05,
-      "loss": 0.8945,
+      "loss": 1.0575,
       "step": 3485
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9512819298254896e-05,
-      "loss": 0.7693,
+      "loss": 0.9643,
       "step": 3486
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951245607115308e-05,
-      "loss": 0.7996,
+      "loss": 0.8713,
       "step": 3487
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9512092712079437e-05,
-      "loss": 0.9283,
+      "loss": 1.0732,
       "step": 3488
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9511729221039005e-05,
-      "loss": 0.8882,
+      "loss": 1.046,
       "step": 3489
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9511365598036823e-05,
-      "loss": 0.7743,
+      "loss": 0.8802,
       "step": 3490
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.951100184307794e-05,
-      "loss": 0.775,
+      "loss": 0.886,
       "step": 3491
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.95106379561674e-05,
-      "loss": 0.7743,
+      "loss": 0.9243,
       "step": 3492
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9510273937310252e-05,
-      "loss": 0.8464,
+      "loss": 0.9514,
       "step": 3493
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9509909786511544e-05,
-      "loss": 0.9064,
+      "loss": 0.9429,
       "step": 3494
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9509545503776335e-05,
-      "loss": 0.8302,
+      "loss": 0.9521,
       "step": 3495
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9509181089109673e-05,
-      "loss": 0.7681,
+      "loss": 0.9158,
       "step": 3496
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9508816542516613e-05,
-      "loss": 0.9442,
+      "loss": 0.9993,
       "step": 3497
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950845186400222e-05,
-      "loss": 0.9429,
+      "loss": 1.0047,
       "step": 3498
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950808705357154e-05,
-      "loss": 0.9311,
+      "loss": 1.001,
       "step": 3499
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950772211122965e-05,
-      "loss": 0.8838,
+      "loss": 1.0179,
       "step": 3500
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.95073570369816e-05,
-      "loss": 1.0859,
+      "loss": 1.1487,
       "step": 3501
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9506991830832463e-05,
-      "loss": 0.8623,
+      "loss": 0.9918,
       "step": 3502
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9506626492787304e-05,
-      "loss": 0.8477,
+      "loss": 0.9196,
       "step": 3503
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9506261022851187e-05,
-      "loss": 0.908,
+      "loss": 0.8986,
       "step": 3504
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9505895421029187e-05,
-      "loss": 0.8346,
+      "loss": 0.9081,
       "step": 3505
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950552968732638e-05,
-      "loss": 0.8359,
+      "loss": 0.9465,
       "step": 3506
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9505163821747828e-05,
-      "loss": 0.961,
+      "loss": 0.9975,
       "step": 3507
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950479782429862e-05,
-      "loss": 0.8319,
+      "loss": 1.0114,
       "step": 3508
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9504431694983822e-05,
-      "loss": 0.8385,
+      "loss": 0.9265,
       "step": 3509
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9504065433808523e-05,
-      "loss": 0.9662,
+      "loss": 0.967,
       "step": 3510
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.95036990407778e-05,
-      "loss": 0.7142,
+      "loss": 0.8396,
       "step": 3511
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9503332515896737e-05,
-      "loss": 0.9328,
+      "loss": 0.9872,
       "step": 3512
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9502965859170418e-05,
-      "loss": 0.8866,
+      "loss": 0.9645,
       "step": 3513
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9502599070603932e-05,
-      "loss": 0.7784,
+      "loss": 0.8793,
       "step": 3514
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9502232150202362e-05,
-      "loss": 0.8363,
+      "loss": 0.9137,
       "step": 3515
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9501865097970807e-05,
-      "loss": 0.8821,
+      "loss": 0.9601,
       "step": 3516
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9501497913914355e-05,
-      "loss": 0.858,
+      "loss": 1.0165,
       "step": 3517
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.95011305980381e-05,
-      "loss": 0.9142,
+      "loss": 1.0912,
       "step": 3518
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.950076315034714e-05,
-      "loss": 0.8765,
+      "loss": 0.9322,
       "step": 3519
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9500395570846566e-05,
-      "loss": 0.8882,
+      "loss": 0.992,
       "step": 3520
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9500027859541486e-05,
-      "loss": 0.9956,
+      "loss": 1.0179,
       "step": 3521
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9499660016437e-05,
-      "loss": 0.8396,
+      "loss": 0.9378,
       "step": 3522
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9499292041538207e-05,
-      "loss": 0.8938,
+      "loss": 1.0341,
       "step": 3523
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9498923934850213e-05,
-      "loss": 0.8674,
+      "loss": 0.9872,
       "step": 3524
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949855569637813e-05,
-      "loss": 0.8739,
+      "loss": 0.974,
       "step": 3525
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949818732612706e-05,
-      "loss": 0.8754,
+      "loss": 0.9666,
       "step": 3526
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9497818824102123e-05,
-      "loss": 0.8666,
+      "loss": 1.0573,
       "step": 3527
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949745019030842e-05,
-      "loss": 0.8461,
+      "loss": 0.9469,
       "step": 3528
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9497081424751073e-05,
-      "loss": 0.8677,
+      "loss": 0.9685,
       "step": 3529
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9496712527435194e-05,
-      "loss": 0.9142,
+      "loss": 0.9784,
       "step": 3530
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9496343498365905e-05,
-      "loss": 0.9527,
+      "loss": 0.9814,
       "step": 3531
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9495974337548325e-05,
-      "loss": 0.7808,
+      "loss": 0.9299,
       "step": 3532
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949560504498757e-05,
-      "loss": 0.8004,
+      "loss": 0.9904,
       "step": 3533
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949523562068877e-05,
-      "loss": 0.804,
+      "loss": 0.8615,
       "step": 3534
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9494866064657046e-05,
-      "loss": 0.9367,
+      "loss": 0.9898,
       "step": 3535
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949449637689753e-05,
-      "loss": 0.8531,
+      "loss": 0.9698,
       "step": 3536
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949412655741534e-05,
-      "loss": 0.9379,
+      "loss": 1.0135,
       "step": 3537
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9493756606215625e-05,
-      "loss": 0.8855,
+      "loss": 0.9817,
       "step": 3538
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9493386523303498e-05,
-      "loss": 0.8673,
+      "loss": 0.9702,
       "step": 3539
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9493016308684107e-05,
-      "loss": 0.8501,
+      "loss": 0.9265,
       "step": 3540
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9492645962362582e-05,
-      "loss": 0.7911,
+      "loss": 0.9445,
       "step": 3541
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9492275484344066e-05,
-      "loss": 0.82,
+      "loss": 0.9323,
       "step": 3542
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9491904874633692e-05,
-      "loss": 0.8712,
+      "loss": 1.0182,
       "step": 3543
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9491534133236606e-05,
-      "loss": 0.8842,
+      "loss": 0.9379,
       "step": 3544
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949116326015795e-05,
-      "loss": 0.8831,
+      "loss": 0.9552,
       "step": 3545
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9490792255402868e-05,
-      "loss": 0.8361,
+      "loss": 0.9102,
       "step": 3546
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9490421118976514e-05,
-      "loss": 0.9408,
+      "loss": 1.0143,
       "step": 3547
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.949004985088403e-05,
-      "loss": 0.6861,
+      "loss": 0.8286,
       "step": 3548
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9489678451130568e-05,
-      "loss": 0.7728,
+      "loss": 0.9449,
       "step": 3549
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948930691972128e-05,
-      "loss": 0.8059,
+      "loss": 0.9487,
       "step": 3550
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9488935256661325e-05,
-      "loss": 0.9488,
+      "loss": 1.0994,
       "step": 3551
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9488563461955856e-05,
-      "loss": 0.9081,
+      "loss": 1.0419,
       "step": 3552
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9488191535610032e-05,
-      "loss": 0.9852,
+      "loss": 1.0011,
       "step": 3553
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948781947762901e-05,
-      "loss": 0.8278,
+      "loss": 0.9772,
       "step": 3554
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9487447288017956e-05,
-      "loss": 0.7819,
+      "loss": 0.9106,
       "step": 3555
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948707496678203e-05,
-      "loss": 0.8044,
+      "loss": 0.9015,
       "step": 3556
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9486702513926402e-05,
-      "loss": 0.7931,
+      "loss": 0.9466,
       "step": 3557
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9486329929456235e-05,
-      "loss": 1.0338,
+      "loss": 1.1062,
       "step": 3558
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9485957213376702e-05,
-      "loss": 0.7683,
+      "loss": 0.8019,
       "step": 3559
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9485584365692968e-05,
-      "loss": 0.9845,
+      "loss": 1.0352,
       "step": 3560
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948521138641021e-05,
-      "loss": 0.8004,
+      "loss": 0.954,
       "step": 3561
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.94848382755336e-05,
-      "loss": 0.7697,
+      "loss": 0.8929,
       "step": 3562
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948446503306832e-05,
-      "loss": 0.8485,
+      "loss": 0.9582,
       "step": 3563
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9484091659019545e-05,
-      "loss": 0.9145,
+      "loss": 1.002,
       "step": 3564
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948371815339245e-05,
-      "loss": 0.8178,
+      "loss": 0.9986,
       "step": 3565
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9483344516192223e-05,
-      "loss": 0.9934,
+      "loss": 1.0117,
       "step": 3566
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9482970747424045e-05,
-      "loss": 0.7759,
+      "loss": 0.952,
       "step": 3567
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9482596847093105e-05,
-      "loss": 0.9356,
+      "loss": 1.0427,
       "step": 3568
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9482222815204585e-05,
-      "loss": 0.8611,
+      "loss": 1.0308,
       "step": 3569
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948184865176368e-05,
-      "loss": 0.8588,
+      "loss": 0.9561,
       "step": 3570
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948147435677558e-05,
-      "loss": 0.8419,
+      "loss": 0.9542,
       "step": 3571
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9481099930245467e-05,
-      "loss": 0.8816,
+      "loss": 1.009,
       "step": 3572
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9480725372178552e-05,
-      "loss": 0.8612,
+      "loss": 0.9357,
       "step": 3573
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.948035068258002e-05,
-      "loss": 0.7827,
+      "loss": 0.8841,
       "step": 3574
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9479975861455077e-05,
-      "loss": 0.8555,
+      "loss": 0.9741,
       "step": 3575
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9479600908808916e-05,
-      "loss": 1.0015,
+      "loss": 1.0716,
       "step": 3576
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947922582464674e-05,
-      "loss": 0.894,
+      "loss": 0.9748,
       "step": 3577
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9478850608973762e-05,
-      "loss": 0.7531,
+      "loss": 0.8981,
       "step": 3578
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9478475261795175e-05,
-      "loss": 0.7693,
+      "loss": 0.9709,
       "step": 3579
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9478099783116193e-05,
-      "loss": 0.9097,
+      "loss": 0.9685,
       "step": 3580
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9477724172942024e-05,
-      "loss": 0.9366,
+      "loss": 1.0215,
       "step": 3581
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9477348431277883e-05,
-      "loss": 0.7738,
+      "loss": 0.9913,
       "step": 3582
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9476972558128976e-05,
-      "loss": 0.9138,
+      "loss": 0.9959,
       "step": 3583
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9476596553500518e-05,
-      "loss": 0.7814,
+      "loss": 1.0045,
       "step": 3584
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947622041739773e-05,
-      "loss": 0.887,
+      "loss": 1.0085,
       "step": 3585
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947584414982583e-05,
-      "loss": 0.9259,
+      "loss": 0.9547,
       "step": 3586
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9475467750790038e-05,
-      "loss": 0.7576,
+      "loss": 0.8612,
       "step": 3587
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9475091220295575e-05,
-      "loss": 0.9602,
+      "loss": 1.017,
       "step": 3588
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9474714558347663e-05,
-      "loss": 0.7835,
+      "loss": 0.9563,
       "step": 3589
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947433776495153e-05,
-      "loss": 0.8735,
+      "loss": 0.9229,
       "step": 3590
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.94739608401124e-05,
-      "loss": 0.8037,
+      "loss": 0.9149,
       "step": 3591
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9473583783835507e-05,
-      "loss": 0.9181,
+      "loss": 1.0066,
       "step": 3592
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9473206596126083e-05,
-      "loss": 0.8409,
+      "loss": 0.9503,
       "step": 3593
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9472829276989356e-05,
-      "loss": 0.8089,
+      "loss": 0.9658,
       "step": 3594
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9472451826430564e-05,
-      "loss": 1.0293,
+      "loss": 1.092,
       "step": 3595
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947207424445494e-05,
-      "loss": 0.8888,
+      "loss": 0.9503,
       "step": 3596
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9471696531067728e-05,
-      "loss": 0.845,
+      "loss": 0.9863,
       "step": 3597
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9471318686274165e-05,
-      "loss": 0.8687,
+      "loss": 0.9221,
       "step": 3598
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.947094071007949e-05,
-      "loss": 0.7857,
+      "loss": 0.8873,
       "step": 3599
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9470562602488954e-05,
-      "loss": 0.8249,
+      "loss": 0.9703,
       "step": 3600
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9470184363507796e-05,
-      "loss": 0.9579,
+      "loss": 1.0349,
       "step": 3601
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946980599314127e-05,
-      "loss": 0.832,
+      "loss": 1.0155,
       "step": 3602
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946942749139462e-05,
-      "loss": 0.7517,
+      "loss": 0.8605,
       "step": 3603
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9469048858273096e-05,
-      "loss": 0.7458,
+      "loss": 0.9562,
       "step": 3604
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9468670093781953e-05,
-      "loss": 0.8984,
+      "loss": 1.0174,
       "step": 3605
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9468291197926453e-05,
-      "loss": 0.8587,
+      "loss": 0.9129,
       "step": 3606
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9467912170711843e-05,
-      "loss": 0.7354,
+      "loss": 0.9042,
       "step": 3607
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9467533012143386e-05,
-      "loss": 0.7661,
+      "loss": 0.9423,
       "step": 3608
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946715372222634e-05,
-      "loss": 0.9187,
+      "loss": 0.9768,
       "step": 3609
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9466774300965968e-05,
-      "loss": 0.8363,
+      "loss": 1.0018,
       "step": 3610
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9466394748367535e-05,
-      "loss": 0.8581,
+      "loss": 0.9222,
       "step": 3611
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9466015064436304e-05,
-      "loss": 0.9518,
+      "loss": 0.993,
       "step": 3612
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9465635249177545e-05,
-      "loss": 0.7235,
+      "loss": 0.8797,
       "step": 3613
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946525530259653e-05,
-      "loss": 0.893,
+      "loss": 1.0061,
       "step": 3614
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9464875224698524e-05,
-      "loss": 0.9053,
+      "loss": 0.961,
       "step": 3615
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9464495015488803e-05,
-      "loss": 0.9031,
+      "loss": 1.0189,
       "step": 3616
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9464114674972645e-05,
-      "loss": 1.0008,
+      "loss": 1.0496,
       "step": 3617
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946373420315532e-05,
-      "loss": 0.8072,
+      "loss": 0.9659,
       "step": 3618
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9463353600042113e-05,
-      "loss": 0.966,
+      "loss": 0.9512,
       "step": 3619
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.94629728656383e-05,
-      "loss": 0.6963,
+      "loss": 0.877,
       "step": 3620
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9462591999949168e-05,
-      "loss": 0.8591,
+      "loss": 1.0025,
       "step": 3621
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9462211002979992e-05,
-      "loss": 0.7449,
+      "loss": 0.8258,
       "step": 3622
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9461829874736068e-05,
-      "loss": 0.9117,
+      "loss": 0.9915,
       "step": 3623
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9461448615222677e-05,
-      "loss": 0.9477,
+      "loss": 1.0194,
       "step": 3624
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9461067224445112e-05,
-      "loss": 0.7818,
+      "loss": 1.0114,
       "step": 3625
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9460685702408662e-05,
-      "loss": 0.9544,
+      "loss": 0.9922,
       "step": 3626
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.946030404911862e-05,
-      "loss": 0.9391,
+      "loss": 0.9977,
       "step": 3627
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9459922264580282e-05,
-      "loss": 0.8062,
+      "loss": 0.9428,
       "step": 3628
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9459540348798947e-05,
-      "loss": 0.8736,
+      "loss": 0.9377,
       "step": 3629
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945915830177991e-05,
-      "loss": 1.0251,
+      "loss": 1.0134,
       "step": 3630
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9458776123528468e-05,
-      "loss": 0.7374,
+      "loss": 0.9063,
       "step": 3631
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945839381404993e-05,
-      "loss": 0.769,
+      "loss": 0.9438,
       "step": 3632
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9458011373349598e-05,
-      "loss": 0.9576,
+      "loss": 1.0468,
       "step": 3633
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9457628801432778e-05,
-      "loss": 0.8686,
+      "loss": 0.9658,
       "step": 3634
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9457246098304777e-05,
-      "loss": 0.94,
+      "loss": 1.0179,
       "step": 3635
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9456863263970905e-05,
-      "loss": 0.8143,
+      "loss": 0.9542,
       "step": 3636
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945648029843647e-05,
-      "loss": 0.9402,
+      "loss": 1.0537,
       "step": 3637
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9456097201706788e-05,
-      "loss": 0.8919,
+      "loss": 0.8927,
       "step": 3638
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945571397378718e-05,
-      "loss": 0.82,
+      "loss": 0.9064,
       "step": 3639
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945533061468295e-05,
-      "loss": 0.8614,
+      "loss": 0.9856,
       "step": 3640
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9454947124399424e-05,
-      "loss": 1.086,
+      "loss": 1.1167,
       "step": 3641
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9454563502941924e-05,
-      "loss": 0.886,
+      "loss": 0.9412,
       "step": 3642
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9454179750315768e-05,
-      "loss": 0.7887,
+      "loss": 0.9899,
       "step": 3643
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9453795866526282e-05,
-      "loss": 0.8469,
+      "loss": 0.9881,
       "step": 3644
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9453411851578793e-05,
-      "loss": 0.7674,
+      "loss": 0.8685,
       "step": 3645
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9453027705478623e-05,
-      "loss": 0.8905,
+      "loss": 0.9441,
       "step": 3646
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9452643428231108e-05,
-      "loss": 0.7747,
+      "loss": 0.8142,
       "step": 3647
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945225901984158e-05,
-      "loss": 0.7924,
+      "loss": 0.9174,
       "step": 3648
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9451874480315364e-05,
-      "loss": 0.9163,
+      "loss": 1.0012,
       "step": 3649
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9451489809657803e-05,
-      "loss": 0.9029,
+      "loss": 0.9128,
       "step": 3650
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.945110500787423e-05,
-      "loss": 0.8334,
+      "loss": 0.9635,
       "step": 3651
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9450720074969983e-05,
-      "loss": 0.7622,
+      "loss": 0.9673,
       "step": 3652
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9450335010950406e-05,
-      "loss": 0.6499,
+      "loss": 0.8838,
       "step": 3653
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9449949815820843e-05,
-      "loss": 0.6859,
+      "loss": 0.9207,
       "step": 3654
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9449564489586625e-05,
-      "loss": 0.7981,
+      "loss": 0.8846,
       "step": 3655
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9449179032253113e-05,
-      "loss": 1.0137,
+      "loss": 1.0686,
       "step": 3656
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9448793443825647e-05,
-      "loss": 0.8295,
+      "loss": 0.9453,
       "step": 3657
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9448407724309577e-05,
-      "loss": 0.832,
+      "loss": 0.9329,
       "step": 3658
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.944802187371026e-05,
-      "loss": 0.8847,
+      "loss": 0.9493,
       "step": 3659
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9447635892033035e-05,
-      "loss": 0.9072,
+      "loss": 1.0372,
       "step": 3660
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9447249779283274e-05,
-      "loss": 0.9006,
+      "loss": 1.008,
       "step": 3661
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9446863535466325e-05,
-      "loss": 0.745,
+      "loss": 0.8785,
       "step": 3662
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9446477160587548e-05,
-      "loss": 0.8474,
+      "loss": 0.9565,
       "step": 3663
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.94460906546523e-05,
-      "loss": 0.8575,
+      "loss": 1.0508,
       "step": 3664
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.944570401766595e-05,
-      "loss": 0.7878,
+      "loss": 0.8418,
       "step": 3665
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.944531724963386e-05,
-      "loss": 0.9917,
+      "loss": 1.0096,
       "step": 3666
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.944493035056139e-05,
-      "loss": 0.8937,
+      "loss": 0.9107,
       "step": 3667
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9444543320453913e-05,
-      "loss": 0.7449,
+      "loss": 0.9454,
       "step": 3668
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.94441561593168e-05,
-      "loss": 0.7251,
+      "loss": 0.8268,
       "step": 3669
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9443768867155415e-05,
-      "loss": 1.017,
+      "loss": 1.028,
       "step": 3670
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9443381443975137e-05,
-      "loss": 0.7729,
+      "loss": 0.9735,
       "step": 3671
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9442993889781342e-05,
-      "loss": 0.9643,
+      "loss": 1.0094,
       "step": 3672
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9442606204579408e-05,
-      "loss": 0.9992,
+      "loss": 1.0335,
       "step": 3673
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9442218388374704e-05,
-      "loss": 0.9302,
+      "loss": 1.0862,
       "step": 3674
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9441830441172617e-05,
-      "loss": 0.8566,
+      "loss": 0.9331,
       "step": 3675
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9441442362978532e-05,
-      "loss": 0.7525,
+      "loss": 0.8858,
       "step": 3676
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.944105415379783e-05,
-      "loss": 0.8479,
+      "loss": 0.993,
       "step": 3677
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9440665813635894e-05,
-      "loss": 0.847,
+      "loss": 0.8809,
       "step": 3678
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9440277342498116e-05,
-      "loss": 0.9457,
+      "loss": 1.0203,
       "step": 3679
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9439888740389884e-05,
-      "loss": 1.0082,
+      "loss": 1.0547,
       "step": 3680
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943950000731659e-05,
-      "loss": 1.0025,
+      "loss": 1.0253,
       "step": 3681
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9439111143283624e-05,
-      "loss": 0.8354,
+      "loss": 0.9804,
       "step": 3682
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9438722148296385e-05,
-      "loss": 0.9642,
+      "loss": 0.9717,
       "step": 3683
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9438333022360267e-05,
-      "loss": 0.7913,
+      "loss": 0.9284,
       "step": 3684
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9437943765480673e-05,
-      "loss": 0.7534,
+      "loss": 0.908,
       "step": 3685
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9437554377662995e-05,
-      "loss": 0.872,
+      "loss": 1.0112,
       "step": 3686
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943716485891264e-05,
-      "loss": 0.851,
+      "loss": 0.9987,
       "step": 3687
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9436775209235015e-05,
-      "loss": 0.8247,
+      "loss": 0.9594,
       "step": 3688
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9436385428635525e-05,
-      "loss": 0.9175,
+      "loss": 1.0062,
       "step": 3689
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9435995517119572e-05,
-      "loss": 0.7667,
+      "loss": 0.9081,
       "step": 3690
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943560547469257e-05,
-      "loss": 0.8634,
+      "loss": 0.9457,
       "step": 3691
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9435215301359927e-05,
-      "loss": 0.8183,
+      "loss": 0.9403,
       "step": 3692
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9434824997127065e-05,
-      "loss": 0.7295,
+      "loss": 0.8372,
       "step": 3693
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9434434561999386e-05,
-      "loss": 0.8345,
+      "loss": 0.9516,
       "step": 3694
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9434043995982317e-05,
-      "loss": 0.7933,
+      "loss": 0.8821,
       "step": 3695
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943365329908127e-05,
-      "loss": 0.8216,
+      "loss": 0.9705,
       "step": 3696
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943326247130167e-05,
-      "loss": 0.8046,
+      "loss": 0.9013,
       "step": 3697
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9432871512648936e-05,
-      "loss": 0.7949,
+      "loss": 0.884,
       "step": 3698
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9432480423128495e-05,
-      "loss": 0.9555,
+      "loss": 1.0712,
       "step": 3699
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943208920274577e-05,
-      "loss": 0.9843,
+      "loss": 1.046,
       "step": 3700
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9431697851506187e-05,
-      "loss": 0.8948,
+      "loss": 1.007,
       "step": 3701
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9431306369415183e-05,
-      "loss": 0.8888,
+      "loss": 0.9377,
       "step": 3702
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9430914756478182e-05,
-      "loss": 0.7972,
+      "loss": 0.9745,
       "step": 3703
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9430523012700618e-05,
-      "loss": 0.8719,
+      "loss": 0.9721,
       "step": 3704
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.943013113808793e-05,
-      "loss": 0.7919,
+      "loss": 0.9117,
       "step": 3705
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.942973913264555e-05,
-      "loss": 0.8066,
+      "loss": 0.9081,
       "step": 3706
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9429346996378923e-05,
-      "loss": 0.8027,
+      "loss": 0.9718,
       "step": 3707
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9428954729293478e-05,
-      "loss": 0.8989,
+      "loss": 0.997,
       "step": 3708
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.942856233139467e-05,
-      "loss": 1.0044,
+      "loss": 1.0572,
       "step": 3709
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9428169802687937e-05,
-      "loss": 0.9683,
+      "loss": 0.9955,
       "step": 3710
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.942777714317872e-05,
-      "loss": 0.9474,
+      "loss": 0.9707,
       "step": 3711
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9427384352872475e-05,
-      "loss": 0.7907,
+      "loss": 0.8921,
       "step": 3712
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9426991431774646e-05,
-      "loss": 0.8549,
+      "loss": 0.9586,
       "step": 3713
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9426598379890685e-05,
-      "loss": 0.8021,
+      "loss": 0.9258,
       "step": 3714
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.942620519722605e-05,
-      "loss": 0.9905,
+      "loss": 1.0537,
       "step": 3715
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9425811883786188e-05,
-      "loss": 0.8052,
+      "loss": 0.9792,
       "step": 3716
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9425418439576562e-05,
-      "loss": 0.7836,
+      "loss": 0.9392,
       "step": 3717
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9425024864602627e-05,
-      "loss": 0.7607,
+      "loss": 0.9625,
       "step": 3718
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9424631158869847e-05,
-      "loss": 0.9022,
+      "loss": 0.9404,
       "step": 3719
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9424237322383678e-05,
-      "loss": 0.7192,
+      "loss": 0.8593,
       "step": 3720
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9423843355149588e-05,
-      "loss": 0.8547,
+      "loss": 0.9685,
       "step": 3721
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9423449257173042e-05,
-      "loss": 0.9075,
+      "loss": 1.0144,
       "step": 3722
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.942305502845951e-05,
-      "loss": 0.8792,
+      "loss": 0.9455,
       "step": 3723
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9422660669014455e-05,
-      "loss": 0.8934,
+      "loss": 0.9621,
       "step": 3724
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9422266178843355e-05,
-      "loss": 0.833,
+      "loss": 0.9375,
       "step": 3725
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.942187155795168e-05,
-      "loss": 0.8927,
+      "loss": 1.0091,
       "step": 3726
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9421476806344904e-05,
-      "loss": 0.9444,
+      "loss": 0.9614,
       "step": 3727
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9421081924028507e-05,
-      "loss": 0.8317,
+      "loss": 0.9311,
       "step": 3728
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9420686911007964e-05,
-      "loss": 1.0234,
+      "loss": 1.045,
       "step": 3729
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9420291767288756e-05,
-      "loss": 0.9265,
+      "loss": 1.0526,
       "step": 3730
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9419896492876365e-05,
-      "loss": 0.9358,
+      "loss": 1.0094,
       "step": 3731
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9419501087776278e-05,
-      "loss": 0.8507,
+      "loss": 0.9876,
       "step": 3732
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9419105551993972e-05,
-      "loss": 0.8638,
+      "loss": 0.9354,
       "step": 3733
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9418709885534944e-05,
-      "loss": 0.9516,
+      "loss": 0.9877,
       "step": 3734
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941831408840468e-05,
-      "loss": 0.7994,
+      "loss": 0.8432,
       "step": 3735
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9417918160608673e-05,
-      "loss": 0.8185,
+      "loss": 0.9313,
       "step": 3736
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941752210215241e-05,
-      "loss": 0.7741,
+      "loss": 0.8231,
       "step": 3737
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941712591304139e-05,
-      "loss": 0.9386,
+      "loss": 0.9728,
       "step": 3738
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9416729593281113e-05,
-      "loss": 0.8963,
+      "loss": 0.8994,
       "step": 3739
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941633314287707e-05,
-      "loss": 0.8456,
+      "loss": 0.981,
       "step": 3740
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9415936561834765e-05,
-      "loss": 0.8393,
+      "loss": 0.9172,
       "step": 3741
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.94155398501597e-05,
-      "loss": 1.0344,
+      "loss": 1.0734,
       "step": 3742
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9415143007857377e-05,
-      "loss": 0.994,
+      "loss": 1.0021,
       "step": 3743
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9414746034933304e-05,
-      "loss": 0.9446,
+      "loss": 0.9951,
       "step": 3744
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941434893139299e-05,
-      "loss": 0.9744,
+      "loss": 1.01,
       "step": 3745
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941395169724194e-05,
-      "loss": 0.8216,
+      "loss": 0.9824,
       "step": 3746
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9413554332485667e-05,
-      "loss": 0.7905,
+      "loss": 0.9051,
       "step": 3747
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941315683712968e-05,
-      "loss": 0.945,
+      "loss": 0.9615,
       "step": 3748
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9412759211179503e-05,
-      "loss": 0.6736,
+      "loss": 0.8549,
       "step": 3749
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9412361454640646e-05,
-      "loss": 1.0822,
+      "loss": 1.0656,
       "step": 3750
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9411963567518626e-05,
-      "loss": 0.8572,
+      "loss": 0.9944,
       "step": 3751
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9411565549818966e-05,
-      "loss": 0.7968,
+      "loss": 0.9581,
       "step": 3752
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9411167401547188e-05,
-      "loss": 0.8742,
+      "loss": 0.8869,
       "step": 3753
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9410769122708815e-05,
-      "loss": 0.8754,
+      "loss": 0.9234,
       "step": 3754
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.941037071330937e-05,
-      "loss": 0.8682,
+      "loss": 0.9639,
       "step": 3755
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9409972173354385e-05,
-      "loss": 0.8487,
+      "loss": 0.9743,
       "step": 3756
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9409573502849387e-05,
-      "loss": 0.7576,
+      "loss": 0.9026,
       "step": 3757
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940917470179991e-05,
-      "loss": 0.701,
+      "loss": 0.8507,
       "step": 3758
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9408775770211478e-05,
-      "loss": 0.8902,
+      "loss": 0.9829,
       "step": 3759
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9408376708089637e-05,
-      "loss": 0.8899,
+      "loss": 1.0042,
       "step": 3760
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9407977515439916e-05,
-      "loss": 0.7962,
+      "loss": 0.9709,
       "step": 3761
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9407578192267853e-05,
-      "loss": 0.9433,
+      "loss": 1.0244,
       "step": 3762
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940717873857899e-05,
-      "loss": 0.8294,
+      "loss": 0.9161,
       "step": 3763
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940677915437887e-05,
-      "loss": 0.7801,
+      "loss": 0.8937,
       "step": 3764
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9406379439673038e-05,
-      "loss": 0.8142,
+      "loss": 0.925,
       "step": 3765
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9405979594467034e-05,
-      "loss": 0.7985,
+      "loss": 0.9518,
       "step": 3766
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9405579618766413e-05,
-      "loss": 0.8291,
+      "loss": 0.9666,
       "step": 3767
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9405179512576717e-05,
-      "loss": 0.8519,
+      "loss": 0.9487,
       "step": 3768
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.94047792759035e-05,
-      "loss": 0.902,
+      "loss": 1.0637,
       "step": 3769
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940437890875231e-05,
-      "loss": 0.8953,
+      "loss": 0.9745,
       "step": 3770
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940397841112871e-05,
-      "loss": 0.958,
+      "loss": 0.9961,
       "step": 3771
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940357778303825e-05,
-      "loss": 0.6975,
+      "loss": 0.9278,
       "step": 3772
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9403177024486494e-05,
-      "loss": 0.9317,
+      "loss": 1.027,
       "step": 3773
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9402776135478994e-05,
-      "loss": 0.9417,
+      "loss": 1.0723,
       "step": 3774
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940237511602132e-05,
-      "loss": 0.913,
+      "loss": 1.0088,
       "step": 3775
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9401973966119028e-05,
-      "loss": 0.7579,
+      "loss": 0.8896,
       "step": 3776
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940157268577769e-05,
-      "loss": 0.7813,
+      "loss": 0.9696,
       "step": 3777
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9401171275002864e-05,
-      "loss": 0.9433,
+      "loss": 1.0114,
       "step": 3778
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940076973380013e-05,
-      "loss": 0.841,
+      "loss": 0.957,
       "step": 3779
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.940036806217505e-05,
-      "loss": 0.8485,
+      "loss": 0.96,
       "step": 3780
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9399966260133208e-05,
-      "loss": 0.8046,
+      "loss": 0.8953,
       "step": 3781
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9399564327680165e-05,
-      "loss": 0.9277,
+      "loss": 0.9758,
       "step": 3782
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9399162264821502e-05,
-      "loss": 0.9874,
+      "loss": 1.0715,
       "step": 3783
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93987600715628e-05,
-      "loss": 0.7893,
+      "loss": 0.913,
       "step": 3784
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.939835774790964e-05,
-      "loss": 0.9633,
+      "loss": 1.0007,
       "step": 3785
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93979552938676e-05,
-      "loss": 0.8388,
+      "loss": 0.9386,
       "step": 3786
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.939755270944226e-05,
-      "loss": 0.8546,
+      "loss": 0.9267,
       "step": 3787
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9397149994639214e-05,
-      "loss": 0.7861,
+      "loss": 0.9328,
       "step": 3788
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9396747149464044e-05,
-      "loss": 0.8832,
+      "loss": 0.9868,
       "step": 3789
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9396344173922335e-05,
-      "loss": 0.7569,
+      "loss": 0.8833,
       "step": 3790
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9395941068019687e-05,
-      "loss": 0.8301,
+      "loss": 0.9045,
       "step": 3791
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9395537831761687e-05,
-      "loss": 0.7474,
+      "loss": 0.8836,
       "step": 3792
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.939513446515393e-05,
-      "loss": 0.8507,
+      "loss": 0.9274,
       "step": 3793
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9394730968202014e-05,
-      "loss": 0.9214,
+      "loss": 0.9494,
       "step": 3794
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9394327340911532e-05,
-      "loss": 0.8256,
+      "loss": 0.9471,
       "step": 3795
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9393923583288087e-05,
-      "loss": 0.7467,
+      "loss": 0.8822,
       "step": 3796
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9393519695337283e-05,
-      "loss": 0.8626,
+      "loss": 0.9859,
       "step": 3797
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9393115677064717e-05,
-      "loss": 0.9166,
+      "loss": 1.0601,
       "step": 3798
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9392711528476004e-05,
-      "loss": 0.862,
+      "loss": 1.0281,
       "step": 3799
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.939230724957674e-05,
-      "loss": 0.8829,
+      "loss": 1.0182,
       "step": 3800
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.939190284037254e-05,
-      "loss": 0.8688,
+      "loss": 0.98,
       "step": 3801
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9391498300869017e-05,
-      "loss": 0.9012,
+      "loss": 0.9566,
       "step": 3802
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9391093631071776e-05,
-      "loss": 0.8324,
+      "loss": 0.9151,
       "step": 3803
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9390688830986434e-05,
-      "loss": 0.7748,
+      "loss": 0.8849,
       "step": 3804
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9390283900618613e-05,
-      "loss": 0.8281,
+      "loss": 0.9185,
       "step": 3805
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938987883997392e-05,
-      "loss": 0.7042,
+      "loss": 0.8482,
       "step": 3806
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9389473649057984e-05,
-      "loss": 0.8974,
+      "loss": 0.9858,
       "step": 3807
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9389068327876425e-05,
-      "loss": 0.7871,
+      "loss": 0.8962,
       "step": 3808
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938866287643486e-05,
-      "loss": 0.8054,
+      "loss": 0.9826,
       "step": 3809
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9388257294738922e-05,
-      "loss": 0.8195,
+      "loss": 0.8841,
       "step": 3810
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938785158279423e-05,
-      "loss": 0.854,
+      "loss": 0.9406,
       "step": 3811
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9387445740606422e-05,
-      "loss": 0.9663,
+      "loss": 1.0134,
       "step": 3812
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9387039768181118e-05,
-      "loss": 0.81,
+      "loss": 1.0124,
       "step": 3813
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938663366552396e-05,
-      "loss": 0.8934,
+      "loss": 1.0135,
       "step": 3814
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9386227432640575e-05,
-      "loss": 1.0259,
+      "loss": 1.0303,
       "step": 3815
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9385821069536605e-05,
-      "loss": 0.8424,
+      "loss": 1.0245,
       "step": 3816
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938541457621768e-05,
-      "loss": 0.9139,
+      "loss": 0.9521,
       "step": 3817
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9385007952689446e-05,
-      "loss": 0.8533,
+      "loss": 0.9327,
       "step": 3818
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9384601198957543e-05,
-      "loss": 0.9176,
+      "loss": 0.9681,
       "step": 3819
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9384194315027613e-05,
-      "loss": 0.9536,
+      "loss": 1.0304,
       "step": 3820
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93837873009053e-05,
-      "loss": 0.752,
+      "loss": 0.9289,
       "step": 3821
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9383380156596253e-05,
-      "loss": 0.853,
+      "loss": 1.0085,
       "step": 3822
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9382972882106122e-05,
-      "loss": 0.9159,
+      "loss": 1.0043,
       "step": 3823
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938256547744055e-05,
-      "loss": 0.8835,
+      "loss": 0.9476,
       "step": 3824
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9382157942605197e-05,
-      "loss": 0.8611,
+      "loss": 1.0385,
       "step": 3825
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9381750277605715e-05,
-      "loss": 0.9362,
+      "loss": 0.9574,
       "step": 3826
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.938134248244776e-05,
-      "loss": 0.8371,
+      "loss": 0.9306,
       "step": 3827
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9380934557136984e-05,
-      "loss": 0.9005,
+      "loss": 0.9931,
       "step": 3828
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9380526501679054e-05,
-      "loss": 0.875,
+      "loss": 0.9822,
       "step": 3829
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9380118316079628e-05,
-      "loss": 0.8861,
+      "loss": 0.8481,
       "step": 3830
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9379710000344367e-05,
-      "loss": 0.9169,
+      "loss": 0.9871,
       "step": 3831
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937930155447894e-05,
-      "loss": 0.949,
+      "loss": 0.9333,
       "step": 3832
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937889297848901e-05,
-      "loss": 0.8534,
+      "loss": 0.9323,
       "step": 3833
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9378484272380246e-05,
-      "loss": 0.9074,
+      "loss": 1.0233,
       "step": 3834
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937807543615832e-05,
-      "loss": 0.79,
+      "loss": 1.0179,
       "step": 3835
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9377666469828904e-05,
-      "loss": 0.7583,
+      "loss": 0.9463,
       "step": 3836
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937725737339767e-05,
-      "loss": 0.8191,
+      "loss": 0.8933,
       "step": 3837
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9376848146870294e-05,
-      "loss": 0.8741,
+      "loss": 0.9067,
       "step": 3838
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9376438790252456e-05,
-      "loss": 0.9191,
+      "loss": 0.9475,
       "step": 3839
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9376029303549833e-05,
-      "loss": 0.8998,
+      "loss": 0.9142,
       "step": 3840
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9375619686768107e-05,
-      "loss": 0.9678,
+      "loss": 1.0157,
       "step": 3841
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9375209939912954e-05,
-      "loss": 0.7823,
+      "loss": 0.9004,
       "step": 3842
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9374800062990072e-05,
-      "loss": 0.9392,
+      "loss": 0.9309,
       "step": 3843
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9374390056005135e-05,
-      "loss": 0.7946,
+      "loss": 0.8865,
       "step": 3844
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9373979918963843e-05,
-      "loss": 0.8567,
+      "loss": 0.9391,
       "step": 3845
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9373569651871873e-05,
-      "loss": 0.8654,
+      "loss": 0.9688,
       "step": 3846
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9373159254734924e-05,
-      "loss": 0.8262,
+      "loss": 0.9814,
       "step": 3847
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9372748727558692e-05,
-      "loss": 0.9859,
+      "loss": 1.0226,
       "step": 3848
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937233807034887e-05,
-      "loss": 0.7297,
+      "loss": 0.9065,
       "step": 3849
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.937192728311115e-05,
-      "loss": 0.8815,
+      "loss": 1.0412,
       "step": 3850
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9371516365851243e-05,
-      "loss": 0.8662,
+      "loss": 0.9513,
       "step": 3851
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9371105318574836e-05,
-      "loss": 0.718,
+      "loss": 0.895,
       "step": 3852
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9370694141287644e-05,
-      "loss": 0.6597,
+      "loss": 0.8151,
       "step": 3853
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9370282833995363e-05,
-      "loss": 0.8203,
+      "loss": 0.89,
       "step": 3854
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9369871396703703e-05,
-      "loss": 0.8033,
+      "loss": 0.9433,
       "step": 3855
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936945982941837e-05,
-      "loss": 0.8674,
+      "loss": 0.9606,
       "step": 3856
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936904813214508e-05,
-      "loss": 0.7802,
+      "loss": 0.8642,
       "step": 3857
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9368636304889538e-05,
-      "loss": 0.9789,
+      "loss": 0.9805,
       "step": 3858
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9368224347657457e-05,
-      "loss": 0.7502,
+      "loss": 0.9068,
       "step": 3859
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9367812260454558e-05,
-      "loss": 0.8873,
+      "loss": 1.0336,
       "step": 3860
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9367400043286552e-05,
-      "loss": 0.8157,
+      "loss": 0.93,
       "step": 3861
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936698769615916e-05,
-      "loss": 0.8365,
+      "loss": 0.9564,
       "step": 3862
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9366575219078113e-05,
-      "loss": 0.8219,
+      "loss": 0.9129,
       "step": 3863
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9366162612049115e-05,
-      "loss": 0.9499,
+      "loss": 1.0213,
       "step": 3864
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93657498750779e-05,
-      "loss": 0.8911,
+      "loss": 1.0356,
       "step": 3865
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93653370081702e-05,
-      "loss": 0.9579,
+      "loss": 0.9815,
       "step": 3866
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9364924011331733e-05,
-      "loss": 0.729,
+      "loss": 0.8442,
       "step": 3867
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936451088456823e-05,
-      "loss": 0.882,
+      "loss": 0.9892,
       "step": 3868
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936409762788543e-05,
-      "loss": 0.7836,
+      "loss": 0.9334,
       "step": 3869
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9363684241289056e-05,
-      "loss": 0.809,
+      "loss": 0.8792,
       "step": 3870
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9363270724784852e-05,
-      "loss": 0.7617,
+      "loss": 0.8809,
       "step": 3871
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936285707837855e-05,
-      "loss": 0.8675,
+      "loss": 1.0081,
       "step": 3872
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9362443302075893e-05,
-      "loss": 1.0233,
+      "loss": 1.0664,
       "step": 3873
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9362029395882615e-05,
-      "loss": 0.7778,
+      "loss": 0.8607,
       "step": 3874
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9361615359804463e-05,
-      "loss": 0.8242,
+      "loss": 0.919,
       "step": 3875
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9361201193847183e-05,
-      "loss": 0.6818,
+      "loss": 0.9046,
       "step": 3876
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9360786898016515e-05,
-      "loss": 0.867,
+      "loss": 1.0342,
       "step": 3877
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.936037247231821e-05,
-      "loss": 0.7663,
+      "loss": 0.9016,
       "step": 3878
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935995791675802e-05,
-      "loss": 0.8274,
+      "loss": 0.9403,
       "step": 3879
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9359543231341694e-05,
-      "loss": 0.8155,
+      "loss": 0.8928,
       "step": 3880
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9359128416074983e-05,
-      "loss": 0.9899,
+      "loss": 1.0427,
       "step": 3881
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9358713470963645e-05,
-      "loss": 0.754,
+      "loss": 0.9311,
       "step": 3882
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9358298396013434e-05,
-      "loss": 0.763,
+      "loss": 0.9166,
       "step": 3883
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9357883191230112e-05,
-      "loss": 0.8472,
+      "loss": 0.9101,
       "step": 3884
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935746785661944e-05,
-      "loss": 0.8775,
+      "loss": 0.9375,
       "step": 3885
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9357052392187175e-05,
-      "loss": 0.9169,
+      "loss": 0.9931,
       "step": 3886
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9356636797939086e-05,
-      "loss": 0.799,
+      "loss": 0.9445,
       "step": 3887
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9356221073880938e-05,
-      "loss": 0.8709,
+      "loss": 0.9495,
       "step": 3888
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9355805220018494e-05,
-      "loss": 0.8711,
+      "loss": 1.0027,
       "step": 3889
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9355389236357528e-05,
-      "loss": 0.9373,
+      "loss": 0.9429,
       "step": 3890
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9354973122903806e-05,
-      "loss": 0.8135,
+      "loss": 0.9126,
       "step": 3891
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935455687966311e-05,
-      "loss": 0.7687,
+      "loss": 0.9474,
       "step": 3892
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9354140506641208e-05,
-      "loss": 0.7259,
+      "loss": 0.9868,
       "step": 3893
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935372400384388e-05,
-      "loss": 0.7798,
+      "loss": 0.9279,
       "step": 3894
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93533073712769e-05,
-      "loss": 0.8993,
+      "loss": 1.0016,
       "step": 3895
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935289060894605e-05,
-      "loss": 0.7879,
+      "loss": 0.8845,
       "step": 3896
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9352473716857117e-05,
-      "loss": 0.847,
+      "loss": 0.9785,
       "step": 3897
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935205669501588e-05,
-      "loss": 1.0239,
+      "loss": 1.1282,
       "step": 3898
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9351639543428122e-05,
-      "loss": 0.8357,
+      "loss": 0.9321,
       "step": 3899
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9351222262099636e-05,
-      "loss": 0.8553,
+      "loss": 0.9903,
       "step": 3900
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9350804851036207e-05,
-      "loss": 0.7819,
+      "loss": 0.9159,
       "step": 3901
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.935038731024363e-05,
-      "loss": 0.8501,
+      "loss": 0.9813,
       "step": 3902
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9349969639727693e-05,
-      "loss": 0.8323,
+      "loss": 0.9394,
       "step": 3903
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9349551839494193e-05,
-      "loss": 0.8437,
+      "loss": 0.966,
       "step": 3904
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934913390954893e-05,
-      "loss": 0.8548,
+      "loss": 0.9843,
       "step": 3905
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93487158498977e-05,
-      "loss": 0.7435,
+      "loss": 0.8985,
       "step": 3906
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9348297660546295e-05,
-      "loss": 0.8369,
+      "loss": 0.9557,
       "step": 3907
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9347879341500527e-05,
-      "loss": 0.9493,
+      "loss": 0.9919,
       "step": 3908
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9347460892766195e-05,
-      "loss": 0.8767,
+      "loss": 0.9706,
       "step": 3909
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9347042314349107e-05,
-      "loss": 0.8814,
+      "loss": 0.9086,
       "step": 3910
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934662360625507e-05,
-      "loss": 0.8622,
+      "loss": 0.9835,
       "step": 3911
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9346204768489888e-05,
-      "loss": 0.8216,
+      "loss": 0.925,
       "step": 3912
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9345785801059376e-05,
-      "loss": 0.8932,
+      "loss": 1.0101,
       "step": 3913
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9345366703969347e-05,
-      "loss": 0.9268,
+      "loss": 0.9095,
       "step": 3914
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9344947477225613e-05,
-      "loss": 0.9005,
+      "loss": 0.9964,
       "step": 3915
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934452812083399e-05,
-      "loss": 0.9306,
+      "loss": 1.0229,
       "step": 3916
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9344108634800303e-05,
-      "loss": 0.8313,
+      "loss": 0.9062,
       "step": 3917
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9343689019130364e-05,
-      "loss": 0.9551,
+      "loss": 1.0408,
       "step": 3918
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9343269273829994e-05,
-      "loss": 0.9094,
+      "loss": 0.955,
       "step": 3919
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9342849398905025e-05,
-      "loss": 0.8741,
+      "loss": 0.9783,
       "step": 3920
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934242939436127e-05,
-      "loss": 0.8824,
+      "loss": 0.9586,
       "step": 3921
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9342009260204568e-05,
-      "loss": 0.9208,
+      "loss": 0.9988,
       "step": 3922
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9341588996440738e-05,
-      "loss": 0.8564,
+      "loss": 0.976,
       "step": 3923
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9341168603075613e-05,
-      "loss": 0.8528,
+      "loss": 1.0147,
       "step": 3924
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934074808011503e-05,
-      "loss": 0.9253,
+      "loss": 0.9944,
       "step": 3925
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.934032742756482e-05,
-      "loss": 0.7941,
+      "loss": 0.9557,
       "step": 3926
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933990664543082e-05,
-      "loss": 0.9308,
+      "loss": 0.9798,
       "step": 3927
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9339485733718864e-05,
-      "loss": 0.9914,
+      "loss": 0.9645,
       "step": 3928
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93390646924348e-05,
-      "loss": 0.8197,
+      "loss": 0.9478,
       "step": 3929
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9338643521584457e-05,
-      "loss": 0.9275,
+      "loss": 0.9396,
       "step": 3930
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9338222221173686e-05,
-      "loss": 0.8637,
+      "loss": 1.0238,
       "step": 3931
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9337800791208336e-05,
-      "loss": 0.8204,
+      "loss": 0.9932,
       "step": 3932
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9337379231694243e-05,
-      "loss": 0.8393,
+      "loss": 0.9373,
       "step": 3933
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9336957542637265e-05,
-      "loss": 0.8363,
+      "loss": 0.9581,
       "step": 3934
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9336535724043248e-05,
-      "loss": 0.8047,
+      "loss": 0.9717,
       "step": 3935
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9336113775918045e-05,
-      "loss": 0.8335,
+      "loss": 0.9585,
       "step": 3936
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9335691698267508e-05,
-      "loss": 0.7629,
+      "loss": 0.8696,
       "step": 3937
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9335269491097495e-05,
-      "loss": 0.7336,
+      "loss": 0.889,
       "step": 3938
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9334847154413864e-05,
-      "loss": 0.841,
+      "loss": 0.9697,
       "step": 3939
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9334424688222467e-05,
-      "loss": 0.8723,
+      "loss": 0.9991,
       "step": 3940
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933400209252918e-05,
-      "loss": 0.909,
+      "loss": 1.0454,
       "step": 3941
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9333579367339853e-05,
-      "loss": 0.9381,
+      "loss": 0.9822,
       "step": 3942
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9333156512660355e-05,
-      "loss": 1.0192,
+      "loss": 1.0884,
       "step": 3943
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9332733528496554e-05,
-      "loss": 0.8732,
+      "loss": 0.9293,
       "step": 3944
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9332310414854313e-05,
-      "loss": 0.8492,
+      "loss": 0.9513,
       "step": 3945
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933188717173951e-05,
-      "loss": 0.8923,
+      "loss": 1.015,
       "step": 3946
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933146379915801e-05,
-      "loss": 0.8294,
+      "loss": 0.9927,
       "step": 3947
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933104029711569e-05,
-      "loss": 0.8289,
+      "loss": 0.8951,
       "step": 3948
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9330616665618424e-05,
-      "loss": 0.7826,
+      "loss": 0.9599,
       "step": 3949
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.933019290467209e-05,
-      "loss": 0.9356,
+      "loss": 0.9884,
       "step": 3950
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9329769014282572e-05,
-      "loss": 0.8788,
+      "loss": 0.9643,
       "step": 3951
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932934499445574e-05,
-      "loss": 0.8638,
+      "loss": 0.9359,
       "step": 3952
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9328920845197486e-05,
-      "loss": 0.8921,
+      "loss": 0.9519,
       "step": 3953
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932849656651369e-05,
-      "loss": 0.7669,
+      "loss": 0.8509,
       "step": 3954
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932807215841024e-05,
-      "loss": 0.8045,
+      "loss": 0.9607,
       "step": 3955
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9327647620893026e-05,
-      "loss": 0.7197,
+      "loss": 0.873,
       "step": 3956
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932722295396793e-05,
-      "loss": 0.9368,
+      "loss": 1.0732,
       "step": 3957
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9326798157640855e-05,
-      "loss": 0.8344,
+      "loss": 0.9685,
       "step": 3958
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9326373231917687e-05,
-      "loss": 0.8487,
+      "loss": 0.9394,
       "step": 3959
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932594817680432e-05,
-      "loss": 0.8102,
+      "loss": 0.9232,
       "step": 3960
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9325522992306657e-05,
-      "loss": 0.8855,
+      "loss": 0.926,
       "step": 3961
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9325097678430593e-05,
-      "loss": 0.8159,
+      "loss": 0.9164,
       "step": 3962
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9324672235182027e-05,
-      "loss": 0.8392,
+      "loss": 0.9379,
       "step": 3963
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9324246662566867e-05,
-      "loss": 0.8151,
+      "loss": 0.929,
       "step": 3964
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932382096059101e-05,
-      "loss": 0.8286,
+      "loss": 0.9398,
       "step": 3965
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9323395129260367e-05,
-      "loss": 0.8845,
+      "loss": 1.0049,
       "step": 3966
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9322969168580847e-05,
-      "loss": 0.8641,
+      "loss": 1.0229,
       "step": 3967
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9322543078558356e-05,
-      "loss": 0.9263,
+      "loss": 0.9789,
       "step": 3968
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9322116859198807e-05,
-      "loss": 0.9422,
+      "loss": 0.9338,
       "step": 3969
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9321690510508114e-05,
-      "loss": 0.8376,
+      "loss": 0.943,
       "step": 3970
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932126403249219e-05,
-      "loss": 0.9178,
+      "loss": 1.037,
       "step": 3971
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9320837425156954e-05,
-      "loss": 0.9052,
+      "loss": 1.0068,
       "step": 3972
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.932041068850832e-05,
-      "loss": 0.6829,
+      "loss": 0.835,
       "step": 3973
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9319983822552213e-05,
-      "loss": 0.7175,
+      "loss": 0.8962,
       "step": 3974
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9319556827294556e-05,
-      "loss": 0.672,
+      "loss": 0.9185,
       "step": 3975
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9319129702741266e-05,
-      "loss": 0.8614,
+      "loss": 0.997,
       "step": 3976
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9318702448898277e-05,
-      "loss": 0.8102,
+      "loss": 0.8875,
       "step": 3977
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.931827506577151e-05,
-      "loss": 0.7549,
+      "loss": 0.9083,
       "step": 3978
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.93178475533669e-05,
-      "loss": 0.9103,
+      "loss": 0.9877,
       "step": 3979
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9317419911690378e-05,
-      "loss": 0.7462,
+      "loss": 0.9592,
       "step": 3980
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9316992140747873e-05,
-      "loss": 0.8199,
+      "loss": 0.9262,
       "step": 3981
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9316564240545318e-05,
-      "loss": 0.9366,
+      "loss": 0.9377,
       "step": 3982
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9316136211088655e-05,
-      "loss": 0.8072,
+      "loss": 0.89,
       "step": 3983
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.931570805238382e-05,
-      "loss": 0.8085,
+      "loss": 0.9992,
       "step": 3984
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9315279764436752e-05,
-      "loss": 0.8384,
+      "loss": 0.9538,
       "step": 3985
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9314851347253394e-05,
-      "loss": 0.9324,
+      "loss": 1.0034,
       "step": 3986
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9314422800839693e-05,
-      "loss": 0.8992,
+      "loss": 1.027,
       "step": 3987
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9313994125201585e-05,
-      "loss": 0.9448,
+      "loss": 0.9821,
       "step": 3988
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931356532034503e-05,
-      "loss": 0.8393,
+      "loss": 0.935,
       "step": 3989
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9313136386275967e-05,
-      "loss": 0.9094,
+      "loss": 0.9986,
       "step": 3990
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931270732300035e-05,
-      "loss": 0.9081,
+      "loss": 1.0752,
       "step": 3991
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931227813052413e-05,
-      "loss": 0.9207,
+      "loss": 0.9382,
       "step": 3992
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9311848808853267e-05,
-      "loss": 0.8305,
+      "loss": 1.0023,
       "step": 3993
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931141935799371e-05,
-      "loss": 1.0095,
+      "loss": 1.0273,
       "step": 3994
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9310989777951422e-05,
-      "loss": 0.8943,
+      "loss": 0.9366,
       "step": 3995
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931056006873236e-05,
-      "loss": 0.7628,
+      "loss": 0.8593,
       "step": 3996
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.931013023034249e-05,
-      "loss": 1.0501,
+      "loss": 1.0381,
       "step": 3997
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9309700262787773e-05,
-      "loss": 0.9641,
+      "loss": 1.0226,
       "step": 3998
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9309270166074168e-05,
-      "loss": 0.7568,
+      "loss": 0.9028,
       "step": 3999
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9308839940207656e-05,
-      "loss": 0.8684,
+      "loss": 0.9419,
       "step": 4000
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9308409585194193e-05,
-      "loss": 0.8015,
+      "loss": 0.8839,
       "step": 4001
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930797910103975e-05,
-      "loss": 1.0649,
+      "loss": 1.0616,
       "step": 4002
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9307548487750308e-05,
-      "loss": 0.9362,
+      "loss": 1.045,
       "step": 4003
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9307117745331836e-05,
-      "loss": 0.8536,
+      "loss": 0.9993,
       "step": 4004
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930668687379031e-05,
-      "loss": 0.9114,
+      "loss": 1.049,
       "step": 4005
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9306255873131708e-05,
-      "loss": 0.7583,
+      "loss": 0.924,
       "step": 4006
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930582474336201e-05,
-      "loss": 0.8042,
+      "loss": 0.9373,
       "step": 4007
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9305393484487197e-05,
-      "loss": 0.8585,
+      "loss": 0.9979,
       "step": 4008
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930496209651325e-05,
-      "loss": 0.7517,
+      "loss": 0.9653,
       "step": 4009
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930453057944616e-05,
-      "loss": 0.7519,
+      "loss": 0.8959,
       "step": 4010
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9304098933291904e-05,
-      "loss": 0.9312,
+      "loss": 1.0042,
       "step": 4011
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930366715805648e-05,
-      "loss": 0.7255,
+      "loss": 0.7955,
       "step": 4012
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9303235253745874e-05,
-      "loss": 0.8441,
+      "loss": 0.9631,
       "step": 4013
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9302803220366076e-05,
-      "loss": 0.8875,
+      "loss": 0.969,
       "step": 4014
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9302371057923085e-05,
-      "loss": 0.9312,
+      "loss": 1.0009,
       "step": 4015
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.930193876642289e-05,
-      "loss": 0.8325,
+      "loss": 0.9651,
       "step": 4016
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.93015063458715e-05,
-      "loss": 0.9334,
+      "loss": 1.0231,
       "step": 4017
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.93010737962749e-05,
-      "loss": 0.9626,
+      "loss": 0.9913,
       "step": 4018
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9300641117639102e-05,
-      "loss": 0.9217,
+      "loss": 0.9621,
       "step": 4019
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9300208309970098e-05,
-      "loss": 0.7562,
+      "loss": 0.8783,
       "step": 4020
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9299775373273904e-05,
-      "loss": 0.8281,
+      "loss": 0.9118,
       "step": 4021
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929934230755652e-05,
-      "loss": 0.9567,
+      "loss": 1.033,
       "step": 4022
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929890911282396e-05,
-      "loss": 0.6655,
+      "loss": 0.8754,
       "step": 4023
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929847578908222e-05,
-      "loss": 0.7842,
+      "loss": 0.9728,
       "step": 4024
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9298042336337327e-05,
-      "loss": 0.8989,
+      "loss": 1.0283,
       "step": 4025
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9297608754595288e-05,
-      "loss": 0.7868,
+      "loss": 0.9318,
       "step": 4026
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929717504386212e-05,
-      "loss": 0.8773,
+      "loss": 0.9258,
       "step": 4027
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9296741204143837e-05,
-      "loss": 0.8134,
+      "loss": 0.868,
       "step": 4028
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9296307235446465e-05,
-      "loss": 0.9264,
+      "loss": 1.0199,
       "step": 4029
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9295873137776016e-05,
-      "loss": 0.7001,
+      "loss": 0.9626,
       "step": 4030
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9295438911138514e-05,
-      "loss": 0.9499,
+      "loss": 0.9674,
       "step": 4031
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929500455553999e-05,
-      "loss": 0.6846,
+      "loss": 0.9325,
       "step": 4032
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9294570070986465e-05,
-      "loss": 0.8288,
+      "loss": 0.9764,
       "step": 4033
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9294135457483966e-05,
-      "loss": 0.9244,
+      "loss": 1.1076,
       "step": 4034
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9293700715038527e-05,
-      "loss": 0.9879,
+      "loss": 1.0463,
       "step": 4035
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9293265843656174e-05,
-      "loss": 0.8729,
+      "loss": 0.9439,
       "step": 4036
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929283084334294e-05,
-      "loss": 0.813,
+      "loss": 1.0075,
       "step": 4037
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929239571410487e-05,
-      "loss": 0.9837,
+      "loss": 1.0655,
       "step": 4038
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9291960455947988e-05,
-      "loss": 0.7879,
+      "loss": 0.9126,
       "step": 4039
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.929152506887834e-05,
-      "loss": 0.8332,
+      "loss": 0.8882,
       "step": 4040
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9291089552901965e-05,
-      "loss": 0.7868,
+      "loss": 0.9139,
       "step": 4041
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9290653908024906e-05,
-      "loss": 0.9323,
+      "loss": 0.957,
       "step": 4042
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9290218134253204e-05,
-      "loss": 0.8577,
+      "loss": 1.0045,
       "step": 4043
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9289782231592906e-05,
-      "loss": 0.884,
+      "loss": 1.0523,
       "step": 4044
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928934620005006e-05,
-      "loss": 0.8804,
+      "loss": 0.969,
       "step": 4045
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9288910039630715e-05,
-      "loss": 0.7778,
+      "loss": 0.8969,
       "step": 4046
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9288473750340927e-05,
-      "loss": 0.8398,
+      "loss": 0.9288,
       "step": 4047
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928803733218674e-05,
-      "loss": 0.7745,
+      "loss": 0.8658,
       "step": 4048
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928760078517421e-05,
-      "loss": 0.819,
+      "loss": 0.927,
       "step": 4049
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92871641093094e-05,
-      "loss": 0.8646,
+      "loss": 1.0122,
       "step": 4050
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9286727304598367e-05,
-      "loss": 0.7751,
+      "loss": 0.9259,
       "step": 4051
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9286290371047168e-05,
-      "loss": 0.9592,
+      "loss": 0.992,
       "step": 4052
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9285853308661865e-05,
-      "loss": 0.8074,
+      "loss": 0.9063,
       "step": 4053
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928541611744852e-05,
-      "loss": 0.6889,
+      "loss": 0.8182,
       "step": 4054
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9284978797413204e-05,
-      "loss": 0.9202,
+      "loss": 0.9601,
       "step": 4055
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928454134856198e-05,
-      "loss": 0.9317,
+      "loss": 1.0543,
       "step": 4056
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9284103770900917e-05,
-      "loss": 0.9709,
+      "loss": 1.0652,
       "step": 4057
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9283666064436083e-05,
-      "loss": 0.8096,
+      "loss": 0.9342,
       "step": 4058
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9283228229173558e-05,
-      "loss": 0.7896,
+      "loss": 0.9042,
       "step": 4059
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9282790265119416e-05,
-      "loss": 0.8291,
+      "loss": 0.9774,
       "step": 4060
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9282352172279724e-05,
-      "loss": 0.8889,
+      "loss": 1.0009,
       "step": 4061
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.928191395066057e-05,
-      "loss": 0.9218,
+      "loss": 1.0139,
       "step": 4062
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9281475600268024e-05,
-      "loss": 0.9052,
+      "loss": 0.9889,
       "step": 4063
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9281037121108176e-05,
-      "loss": 0.8135,
+      "loss": 0.9113,
       "step": 4064
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9280598513187106e-05,
-      "loss": 0.7954,
+      "loss": 0.8993,
       "step": 4065
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92801597765109e-05,
-      "loss": 0.8233,
+      "loss": 0.9169,
       "step": 4066
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927972091108564e-05,
-      "loss": 0.9013,
+      "loss": 0.9902,
       "step": 4067
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9279281916917422e-05,
-      "loss": 0.7447,
+      "loss": 0.8834,
       "step": 4068
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927884279401233e-05,
-      "loss": 0.8643,
+      "loss": 0.9671,
       "step": 4069
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9278403542376463e-05,
-      "loss": 0.9704,
+      "loss": 1.0617,
       "step": 4070
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927796416201591e-05,
-      "loss": 1.0188,
+      "loss": 1.0959,
       "step": 4071
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9277524652936766e-05,
-      "loss": 0.8336,
+      "loss": 0.958,
       "step": 4072
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9277085015145133e-05,
-      "loss": 0.9225,
+      "loss": 0.9585,
       "step": 4073
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9276645248647105e-05,
-      "loss": 1.001,
+      "loss": 1.0771,
       "step": 4074
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927620535344879e-05,
-      "loss": 1.019,
+      "loss": 1.0599,
       "step": 4075
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9275765329556285e-05,
-      "loss": 0.7139,
+      "loss": 0.8792,
       "step": 4076
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9275325176975693e-05,
-      "loss": 1.0155,
+      "loss": 1.061,
       "step": 4077
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9274884895713127e-05,
-      "loss": 0.8069,
+      "loss": 0.9401,
       "step": 4078
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9274444485774696e-05,
-      "loss": 0.7909,
+      "loss": 0.9673,
       "step": 4079
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9274003947166503e-05,
-      "loss": 0.8236,
+      "loss": 0.9678,
       "step": 4080
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9273563279894665e-05,
-      "loss": 0.7869,
+      "loss": 0.9041,
       "step": 4081
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927312248396529e-05,
-      "loss": 1.008,
+      "loss": 1.0898,
       "step": 4082
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92726815593845e-05,
-      "loss": 0.8229,
+      "loss": 0.9609,
       "step": 4083
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927224050615841e-05,
-      "loss": 0.7769,
+      "loss": 0.9744,
       "step": 4084
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9271799324293138e-05,
-      "loss": 0.8834,
+      "loss": 0.9993,
       "step": 4085
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9271358013794804e-05,
-      "loss": 0.8679,
+      "loss": 0.9442,
       "step": 4086
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9270916574669536e-05,
-      "loss": 0.8393,
+      "loss": 0.9699,
       "step": 4087
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927047500692345e-05,
-      "loss": 0.812,
+      "loss": 0.8824,
       "step": 4088
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.927003331056268e-05,
-      "loss": 0.73,
+      "loss": 0.8496,
       "step": 4089
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9269591485593348e-05,
-      "loss": 0.9198,
+      "loss": 1.0251,
       "step": 4090
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9269149532021586e-05,
-      "loss": 0.9265,
+      "loss": 0.9985,
       "step": 4091
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9268707449853526e-05,
-      "loss": 0.8947,
+      "loss": 1.001,
       "step": 4092
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9268265239095298e-05,
-      "loss": 0.853,
+      "loss": 0.9834,
       "step": 4093
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9267822899753043e-05,
-      "loss": 0.9064,
+      "loss": 1.026,
       "step": 4094
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9267380431832894e-05,
-      "loss": 0.649,
+      "loss": 0.8044,
       "step": 4095
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926693783534099e-05,
-      "loss": 0.8459,
+      "loss": 0.9496,
       "step": 4096
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9266495110283472e-05,
-      "loss": 0.7546,
+      "loss": 0.8589,
       "step": 4097
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926605225666648e-05,
-      "loss": 0.9219,
+      "loss": 0.9835,
       "step": 4098
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9265609274496163e-05,
-      "loss": 0.8343,
+      "loss": 0.9379,
       "step": 4099
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9265166163778662e-05,
-      "loss": 0.8551,
+      "loss": 0.9947,
       "step": 4100
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9264722924520127e-05,
-      "loss": 0.8419,
+      "loss": 0.9424,
       "step": 4101
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9264279556726703e-05,
-      "loss": 0.8792,
+      "loss": 0.9124,
       "step": 4102
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9263836060404547e-05,
-      "loss": 0.8159,
+      "loss": 0.8921,
       "step": 4103
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926339243555981e-05,
-      "loss": 0.7741,
+      "loss": 0.9709,
       "step": 4104
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9262948682198645e-05,
-      "loss": 0.9366,
+      "loss": 1.0275,
       "step": 4105
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9262504800327207e-05,
-      "loss": 0.9406,
+      "loss": 1.0305,
       "step": 4106
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9262060789951658e-05,
-      "loss": 0.8227,
+      "loss": 0.9277,
       "step": 4107
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9261616651078156e-05,
-      "loss": 0.8143,
+      "loss": 0.9208,
       "step": 4108
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926117238371287e-05,
-      "loss": 0.9647,
+      "loss": 1.0259,
       "step": 4109
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926072798786195e-05,
-      "loss": 0.8672,
+      "loss": 0.9436,
       "step": 4110
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.926028346353157e-05,
-      "loss": 0.8334,
+      "loss": 0.9281,
       "step": 4111
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9259838810727894e-05,
-      "loss": 1.0186,
+      "loss": 1.0424,
       "step": 4112
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9259394029457096e-05,
-      "loss": 0.9484,
+      "loss": 0.9691,
       "step": 4113
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925894911972534e-05,
-      "loss": 0.8782,
+      "loss": 1.0729,
       "step": 4114
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9258504081538806e-05,
-      "loss": 0.8042,
+      "loss": 0.9348,
       "step": 4115
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925805891490366e-05,
-      "loss": 0.9029,
+      "loss": 0.9299,
       "step": 4116
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9257613619826085e-05,
-      "loss": 0.8416,
+      "loss": 0.955,
       "step": 4117
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9257168196312254e-05,
-      "loss": 0.8746,
+      "loss": 1.0177,
       "step": 4118
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925672264436835e-05,
-      "loss": 0.9299,
+      "loss": 1.0857,
       "step": 4119
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9256276964000552e-05,
-      "loss": 0.8189,
+      "loss": 0.8915,
       "step": 4120
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925583115521504e-05,
-      "loss": 0.789,
+      "loss": 0.8817,
       "step": 4121
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9255385218018008e-05,
-      "loss": 0.8174,
+      "loss": 0.8429,
       "step": 4122
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9254939152415638e-05,
-      "loss": 0.7731,
+      "loss": 0.9156,
       "step": 4123
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9254492958414115e-05,
-      "loss": 0.7942,
+      "loss": 0.9319,
       "step": 4124
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9254046636019637e-05,
-      "loss": 0.934,
+      "loss": 0.8956,
       "step": 4125
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9253600185238385e-05,
-      "loss": 0.816,
+      "loss": 0.9269,
       "step": 4126
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9253153606076565e-05,
-      "loss": 0.9011,
+      "loss": 0.9921,
       "step": 4127
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9252706898540366e-05,
-      "loss": 0.9033,
+      "loss": 0.9091,
       "step": 4128
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925226006263599e-05,
-      "loss": 0.8235,
+      "loss": 0.8812,
       "step": 4129
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9251813098369626e-05,
-      "loss": 0.8184,
+      "loss": 0.9457,
       "step": 4130
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9251366005747485e-05,
-      "loss": 0.9069,
+      "loss": 1.0262,
       "step": 4131
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9250918784775766e-05,
-      "loss": 0.9803,
+      "loss": 1.0576,
       "step": 4132
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9250471435460677e-05,
-      "loss": 0.8584,
+      "loss": 0.9809,
       "step": 4133
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.925002395780842e-05,
-      "loss": 0.8475,
+      "loss": 0.9468,
       "step": 4134
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9249576351825205e-05,
-      "loss": 0.7921,
+      "loss": 0.9594,
       "step": 4135
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.924912861751724e-05,
-      "loss": 0.9236,
+      "loss": 0.9808,
       "step": 4136
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9248680754890738e-05,
-      "loss": 0.8103,
+      "loss": 0.9531,
       "step": 4137
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9248232763951914e-05,
-      "loss": 0.766,
+      "loss": 0.9355,
       "step": 4138
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9247784644706983e-05,
-      "loss": 0.9064,
+      "loss": 1.025,
       "step": 4139
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.924733639716216e-05,
-      "loss": 0.9884,
+      "loss": 1.0886,
       "step": 4140
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9246888021323666e-05,
-      "loss": 0.9072,
+      "loss": 0.915,
       "step": 4141
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9246439517197717e-05,
-      "loss": 0.982,
+      "loss": 1.0901,
       "step": 4142
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9245990884790543e-05,
-      "loss": 0.897,
+      "loss": 1.0829,
       "step": 4143
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9245542124108357e-05,
-      "loss": 0.9028,
+      "loss": 0.9504,
       "step": 4144
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9245093235157397e-05,
-      "loss": 0.8382,
+      "loss": 0.9541,
       "step": 4145
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9244644217943884e-05,
-      "loss": 0.7138,
+      "loss": 0.9102,
       "step": 4146
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.924419507247405e-05,
-      "loss": 0.811,
+      "loss": 0.9676,
       "step": 4147
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9243745798754126e-05,
-      "loss": 0.7814,
+      "loss": 0.9956,
       "step": 4148
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.924329639679034e-05,
-      "loss": 0.8342,
+      "loss": 0.934,
       "step": 4149
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9242846866588936e-05,
-      "loss": 0.8598,
+      "loss": 0.9747,
       "step": 4150
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9242397208156144e-05,
-      "loss": 0.8633,
+      "loss": 0.9617,
       "step": 4151
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9241947421498202e-05,
-      "loss": 0.9113,
+      "loss": 1.0388,
       "step": 4152
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9241497506621353e-05,
-      "loss": 0.8777,
+      "loss": 1.0325,
       "step": 4153
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9241047463531838e-05,
-      "loss": 0.9613,
+      "loss": 1.0759,
       "step": 4154
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9240597292235903e-05,
-      "loss": 0.87,
+      "loss": 0.9428,
       "step": 4155
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.924014699273979e-05,
-      "loss": 0.8592,
+      "loss": 0.9824,
       "step": 4156
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9239696565049747e-05,
-      "loss": 0.8924,
+      "loss": 0.9441,
       "step": 4157
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9239246009172025e-05,
-      "loss": 0.9026,
+      "loss": 1.0129,
       "step": 4158
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9238795325112867e-05,
-      "loss": 0.7128,
+      "loss": 0.8164,
       "step": 4159
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.923834451287854e-05,
-      "loss": 0.8142,
+      "loss": 1.0048,
       "step": 4160
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9237893572475286e-05,
-      "loss": 0.8714,
+      "loss": 0.9215,
       "step": 4161
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9237442503909368e-05,
-      "loss": 0.8634,
+      "loss": 1.0252,
       "step": 4162
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.923699130718704e-05,
-      "loss": 0.8683,
+      "loss": 1.0043,
       "step": 4163
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9236539982314563e-05,
-      "loss": 0.9368,
+      "loss": 0.9833,
       "step": 4164
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92360885292982e-05,
-      "loss": 0.79,
+      "loss": 0.8778,
       "step": 4165
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9235636948144213e-05,
-      "loss": 1.0055,
+      "loss": 1.0666,
       "step": 4166
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9235185238858864e-05,
-      "loss": 0.9492,
+      "loss": 1.0658,
       "step": 4167
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9234733401448427e-05,
-      "loss": 0.8679,
+      "loss": 0.9421,
       "step": 4168
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9234281435919165e-05,
-      "loss": 0.8025,
+      "loss": 0.9102,
       "step": 4169
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.923382934227735e-05,
-      "loss": 0.9018,
+      "loss": 0.9332,
       "step": 4170
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9233377120529253e-05,
-      "loss": 0.8768,
+      "loss": 1.0106,
       "step": 4171
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.923292477068115e-05,
-      "loss": 0.7956,
+      "loss": 0.9511,
       "step": 4172
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9232472292739312e-05,
-      "loss": 0.7957,
+      "loss": 0.9419,
       "step": 4173
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9232019686710024e-05,
-      "loss": 0.913,
+      "loss": 0.9788,
       "step": 4174
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9231566952599562e-05,
-      "loss": 0.8269,
+      "loss": 0.97,
       "step": 4175
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9231114090414202e-05,
-      "loss": 0.8453,
+      "loss": 0.9085,
       "step": 4176
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9230661100160235e-05,
-      "loss": 0.8963,
+      "loss": 0.9878,
       "step": 4177
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.923020798184394e-05,
-      "loss": 0.8513,
+      "loss": 0.9092,
       "step": 4178
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9229754735471608e-05,
-      "loss": 0.8073,
+      "loss": 0.9563,
       "step": 4179
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922930136104952e-05,
-      "loss": 0.8489,
+      "loss": 0.951,
       "step": 4180
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9228847858583976e-05,
-      "loss": 0.7591,
+      "loss": 0.8872,
       "step": 4181
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9228394228081254e-05,
-      "loss": 0.9596,
+      "loss": 1.0189,
       "step": 4182
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922794046954766e-05,
-      "loss": 0.8795,
+      "loss": 0.9732,
       "step": 4183
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9227486582989488e-05,
-      "loss": 0.7648,
+      "loss": 0.9097,
       "step": 4184
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9227032568413028e-05,
-      "loss": 1.0368,
+      "loss": 1.0651,
       "step": 4185
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9226578425824584e-05,
-      "loss": 0.9927,
+      "loss": 1.073,
       "step": 4186
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922612415523045e-05,
-      "loss": 0.8941,
+      "loss": 0.8932,
       "step": 4187
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922566975663694e-05,
-      "loss": 0.7157,
+      "loss": 0.8528,
       "step": 4188
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9225215230050348e-05,
-      "loss": 0.7826,
+      "loss": 0.9063,
       "step": 4189
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9224760575476986e-05,
-      "loss": 0.9925,
+      "loss": 1.1071,
       "step": 4190
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922430579292316e-05,
-      "loss": 0.792,
+      "loss": 0.9457,
       "step": 4191
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9223850882395174e-05,
-      "loss": 0.8144,
+      "loss": 0.9477,
       "step": 4192
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9223395843899348e-05,
-      "loss": 0.9634,
+      "loss": 1.0595,
       "step": 4193
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922294067744199e-05,
-      "loss": 0.8269,
+      "loss": 0.8991,
       "step": 4194
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9222485383029417e-05,
-      "loss": 0.8625,
+      "loss": 0.9637,
       "step": 4195
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922202996066794e-05,
-      "loss": 0.7898,
+      "loss": 0.9615,
       "step": 4196
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9221574410363886e-05,
-      "loss": 0.8553,
+      "loss": 0.9469,
       "step": 4197
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922111873212357e-05,
-      "loss": 0.8853,
+      "loss": 0.9576,
       "step": 4198
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9220662925953314e-05,
-      "loss": 0.8566,
+      "loss": 0.9911,
       "step": 4199
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.922020699185944e-05,
-      "loss": 0.8271,
+      "loss": 0.9227,
       "step": 4200
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921975092984828e-05,
-      "loss": 0.8307,
+      "loss": 0.9556,
       "step": 4201
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9219294739926154e-05,
-      "loss": 0.8755,
+      "loss": 0.9652,
       "step": 4202
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9218838422099395e-05,
-      "loss": 0.8294,
+      "loss": 0.9453,
       "step": 4203
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9218381976374333e-05,
-      "loss": 0.8697,
+      "loss": 0.9776,
       "step": 4204
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92179254027573e-05,
-      "loss": 0.9296,
+      "loss": 0.9626,
       "step": 4205
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921746870125463e-05,
-      "loss": 0.7043,
+      "loss": 0.9187,
       "step": 4206
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921701187187266e-05,
-      "loss": 0.8527,
+      "loss": 0.9755,
       "step": 4207
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9216554914617726e-05,
-      "loss": 0.9391,
+      "loss": 1.0231,
       "step": 4208
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921609782949617e-05,
-      "loss": 0.7964,
+      "loss": 0.9241,
       "step": 4209
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9215640616514336e-05,
-      "loss": 0.8421,
+      "loss": 0.8994,
       "step": 4210
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921518327567856e-05,
-      "loss": 0.7986,
+      "loss": 0.9247,
       "step": 4211
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921472580699519e-05,
-      "loss": 0.9448,
+      "loss": 1.0437,
       "step": 4212
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9214268210470573e-05,
-      "loss": 0.74,
+      "loss": 0.9328,
       "step": 4213
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9213810486111063e-05,
-      "loss": 0.9701,
+      "loss": 0.9832,
       "step": 4214
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9213352633923002e-05,
-      "loss": 0.8537,
+      "loss": 0.887,
       "step": 4215
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9212894653912746e-05,
-      "loss": 0.8027,
+      "loss": 0.952,
       "step": 4216
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921243654608665e-05,
-      "loss": 0.8345,
+      "loss": 0.9706,
       "step": 4217
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921197831045106e-05,
-      "loss": 0.8625,
+      "loss": 1.038,
       "step": 4218
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9211519947012348e-05,
-      "loss": 0.87,
+      "loss": 0.9269,
       "step": 4219
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9211061455776865e-05,
-      "loss": 0.7567,
+      "loss": 0.9231,
       "step": 4220
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.921060283675097e-05,
-      "loss": 0.7968,
+      "loss": 0.9446,
       "step": 4221
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9210144089941033e-05,
-      "loss": 0.765,
+      "loss": 0.92,
       "step": 4222
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9209685215353413e-05,
-      "loss": 0.8394,
+      "loss": 0.9722,
       "step": 4223
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9209226212994475e-05,
-      "loss": 0.9075,
+      "loss": 1.0,
       "step": 4224
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.920876708287059e-05,
-      "loss": 0.8401,
+      "loss": 0.9377,
       "step": 4225
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.920830782498813e-05,
-      "loss": 0.6863,
+      "loss": 0.8454,
       "step": 4226
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9207848439353466e-05,
-      "loss": 0.7864,
+      "loss": 0.9478,
       "step": 4227
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9207388925972964e-05,
-      "loss": 0.7382,
+      "loss": 0.9021,
       "step": 4228
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9206929284853006e-05,
-      "loss": 0.7353,
+      "loss": 0.8985,
       "step": 4229
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9206469515999964e-05,
-      "loss": 0.9721,
+      "loss": 0.9846,
       "step": 4230
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9206009619420226e-05,
-      "loss": 0.8796,
+      "loss": 1.0009,
       "step": 4231
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9205549595120163e-05,
-      "loss": 0.7705,
+      "loss": 0.8934,
       "step": 4232
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.920508944310616e-05,
-      "loss": 0.9345,
+      "loss": 0.9346,
       "step": 4233
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.92046291633846e-05,
-      "loss": 0.8586,
+      "loss": 0.9523,
       "step": 4234
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9204168755961873e-05,
-      "loss": 0.9514,
+      "loss": 1.0008,
       "step": 4235
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9203708220844358e-05,
-      "loss": 0.7559,
+      "loss": 0.9021,
       "step": 4236
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9203247558038456e-05,
-      "loss": 0.84,
+      "loss": 0.9072,
       "step": 4237
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.920278676755055e-05,
-      "loss": 0.8675,
+      "loss": 0.9399,
       "step": 4238
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9202325849387034e-05,
-      "loss": 0.89,
+      "loss": 1.0323,
       "step": 4239
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9201864803554304e-05,
-      "loss": 0.916,
+      "loss": 1.0327,
       "step": 4240
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9201403630058753e-05,
-      "loss": 0.7917,
+      "loss": 0.9037,
       "step": 4241
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9200942328906786e-05,
-      "loss": 0.8156,
+      "loss": 0.9272,
       "step": 4242
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9200480900104795e-05,
-      "loss": 0.7646,
+      "loss": 0.9442,
       "step": 4243
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9200019343659188e-05,
-      "loss": 0.8512,
+      "loss": 1.0095,
       "step": 4244
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9199557659576364e-05,
-      "loss": 0.7627,
+      "loss": 0.7885,
       "step": 4245
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919909584786273e-05,
-      "loss": 0.8752,
+      "loss": 0.9213,
       "step": 4246
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9198633908524693e-05,
-      "loss": 0.9955,
+      "loss": 1.0734,
       "step": 4247
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919817184156866e-05,
-      "loss": 0.888,
+      "loss": 0.9963,
       "step": 4248
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9197709647001047e-05,
-      "loss": 0.8725,
+      "loss": 0.9198,
       "step": 4249
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919724732482826e-05,
-      "loss": 0.7902,
+      "loss": 0.9072,
       "step": 4250
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919678487505672e-05,
-      "loss": 0.8465,
+      "loss": 0.9232,
       "step": 4251
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919632229769283e-05,
-      "loss": 0.836,
+      "loss": 0.908,
       "step": 4252
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919585959274302e-05,
-      "loss": 0.887,
+      "loss": 1.0146,
       "step": 4253
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9195396760213708e-05,
-      "loss": 0.87,
+      "loss": 0.986,
       "step": 4254
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9194933800111308e-05,
-      "loss": 0.6766,
+      "loss": 0.8512,
       "step": 4255
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9194470712442253e-05,
-      "loss": 0.7262,
+      "loss": 0.8741,
       "step": 4256
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919400749721296e-05,
-      "loss": 0.9193,
+      "loss": 1.0313,
       "step": 4257
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9193544154429853e-05,
-      "loss": 1.0011,
+      "loss": 1.0235,
       "step": 4258
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.919308068409937e-05,
-      "loss": 0.9745,
+      "loss": 1.0506,
       "step": 4259
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9192617086227933e-05,
-      "loss": 0.8467,
+      "loss": 0.93,
       "step": 4260
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9192153360821975e-05,
-      "loss": 0.7567,
+      "loss": 0.8633,
       "step": 4261
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9191689507887936e-05,
-      "loss": 0.8577,
+      "loss": 1.0162,
       "step": 4262
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9191225527432246e-05,
-      "loss": 0.8144,
+      "loss": 0.8726,
       "step": 4263
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.919076141946134e-05,
-      "loss": 0.8084,
+      "loss": 0.9335,
       "step": 4264
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.919029718398166e-05,
-      "loss": 0.9074,
+      "loss": 1.0491,
       "step": 4265
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9189832820999645e-05,
-      "loss": 0.8676,
+      "loss": 0.9505,
       "step": 4266
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918936833052174e-05,
-      "loss": 0.9203,
+      "loss": 1.0355,
       "step": 4267
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9188903712554385e-05,
-      "loss": 0.9414,
+      "loss": 1.0101,
       "step": 4268
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918843896710403e-05,
-      "loss": 0.841,
+      "loss": 1.0297,
       "step": 4269
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918797409417712e-05,
-      "loss": 0.8813,
+      "loss": 0.928,
       "step": 4270
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9187509093780107e-05,
-      "loss": 0.882,
+      "loss": 0.9933,
       "step": 4271
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918704396591944e-05,
-      "loss": 0.8869,
+      "loss": 0.9103,
       "step": 4272
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9186578710601574e-05,
-      "loss": 0.685,
+      "loss": 0.8745,
       "step": 4273
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918611332783296e-05,
-      "loss": 0.7961,
+      "loss": 0.8812,
       "step": 4274
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918564781762006e-05,
-      "loss": 0.6819,
+      "loss": 0.8678,
       "step": 4275
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9185182179969325e-05,
-      "loss": 0.7344,
+      "loss": 0.885,
       "step": 4276
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918471641488722e-05,
-      "loss": 0.9051,
+      "loss": 1.0193,
       "step": 4277
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918425052238021e-05,
-      "loss": 0.9951,
+      "loss": 1.05,
       "step": 4278
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9183784502454754e-05,
-      "loss": 0.8524,
+      "loss": 0.8813,
       "step": 4279
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9183318355117315e-05,
-      "loss": 0.8968,
+      "loss": 1.0204,
       "step": 4280
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9182852080374364e-05,
-      "loss": 0.8337,
+      "loss": 0.9315,
       "step": 4281
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9182385678232373e-05,
-      "loss": 0.817,
+      "loss": 0.8799,
       "step": 4282
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.91819191486978e-05,
-      "loss": 0.8405,
+      "loss": 0.9823,
       "step": 4283
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9181452491777136e-05,
-      "loss": 0.78,
+      "loss": 0.9545,
       "step": 4284
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918098570747684e-05,
-      "loss": 0.8121,
+      "loss": 0.9216,
       "step": 4285
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9180518795803395e-05,
-      "loss": 0.7961,
+      "loss": 0.9427,
       "step": 4286
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.918005175676328e-05,
-      "loss": 0.7416,
+      "loss": 0.8669,
       "step": 4287
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917958459036297e-05,
-      "loss": 0.8769,
+      "loss": 0.9562,
       "step": 4288
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9179117296608943e-05,
-      "loss": 0.8836,
+      "loss": 0.9916,
       "step": 4289
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917864987550769e-05,
-      "loss": 0.9497,
+      "loss": 1.069,
       "step": 4290
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9178182327065695e-05,
-      "loss": 0.8818,
+      "loss": 0.9224,
       "step": 4291
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917771465128944e-05,
-      "loss": 0.9684,
+      "loss": 1.1542,
       "step": 4292
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917724684818542e-05,
-      "loss": 0.9309,
+      "loss": 0.9969,
       "step": 4293
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9176778917760116e-05,
-      "loss": 0.7712,
+      "loss": 0.9536,
       "step": 4294
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9176310860020028e-05,
-      "loss": 0.9531,
+      "loss": 1.0231,
       "step": 4295
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917584267497164e-05,
-      "loss": 0.8838,
+      "loss": 0.9,
       "step": 4296
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917537436262146e-05,
-      "loss": 0.7797,
+      "loss": 0.8692,
       "step": 4297
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917490592297598e-05,
-      "loss": 0.8613,
+      "loss": 1.0242,
       "step": 4298
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9174437356041692e-05,
-      "loss": 0.7695,
+      "loss": 0.9595,
       "step": 4299
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9173968661825103e-05,
-      "loss": 0.8582,
+      "loss": 0.8622,
       "step": 4300
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917349984033272e-05,
-      "loss": 0.7515,
+      "loss": 0.9476,
       "step": 4301
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9173030891571037e-05,
-      "loss": 0.8835,
+      "loss": 1.021,
       "step": 4302
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9172561815546568e-05,
-      "loss": 0.8316,
+      "loss": 0.96,
       "step": 4303
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9172092612265816e-05,
-      "loss": 0.8082,
+      "loss": 0.8991,
       "step": 4304
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9171623281735295e-05,
-      "loss": 0.9479,
+      "loss": 1.0462,
       "step": 4305
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9171153823961513e-05,
-      "loss": 0.7577,
+      "loss": 0.845,
       "step": 4306
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.917068423895098e-05,
-      "loss": 0.8476,
+      "loss": 0.9045,
       "step": 4307
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9170214526710223e-05,
-      "loss": 0.8699,
+      "loss": 0.9154,
       "step": 4308
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9169744687245746e-05,
-      "loss": 0.8678,
+      "loss": 0.9927,
       "step": 4309
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916927472056407e-05,
-      "loss": 0.7891,
+      "loss": 0.9201,
       "step": 4310
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9168804626671718e-05,
-      "loss": 0.7532,
+      "loss": 0.9144,
       "step": 4311
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9168334405575212e-05,
-      "loss": 0.9488,
+      "loss": 1.0306,
       "step": 4312
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9167864057281072e-05,
-      "loss": 0.706,
+      "loss": 0.9376,
       "step": 4313
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9167393581795828e-05,
-      "loss": 0.8941,
+      "loss": 0.992,
       "step": 4314
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9166922979126005e-05,
-      "loss": 0.7332,
+      "loss": 0.8755,
       "step": 4315
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916645224927813e-05,
-      "loss": 0.9208,
+      "loss": 1.0017,
       "step": 4316
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9165981392258736e-05,
-      "loss": 0.9367,
+      "loss": 1.0259,
       "step": 4317
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9165510408074354e-05,
-      "loss": 0.7638,
+      "loss": 0.906,
       "step": 4318
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916503929673152e-05,
-      "loss": 0.7858,
+      "loss": 0.9443,
       "step": 4319
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916456805823677e-05,
-      "loss": 0.7189,
+      "loss": 0.8488,
       "step": 4320
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916409669259664e-05,
-      "loss": 0.8449,
+      "loss": 1.0329,
       "step": 4321
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916362519981767e-05,
-      "loss": 0.8941,
+      "loss": 0.9432,
       "step": 4322
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9163153579906404e-05,
-      "loss": 0.8462,
+      "loss": 0.8884,
       "step": 4323
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916268183286938e-05,
-      "loss": 0.809,
+      "loss": 0.9375,
       "step": 4324
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916220995871315e-05,
-      "loss": 0.8524,
+      "loss": 0.9553,
       "step": 4325
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9161737957444252e-05,
-      "loss": 0.7889,
+      "loss": 0.9399,
       "step": 4326
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.916126582906924e-05,
-      "loss": 0.9287,
+      "loss": 1.0736,
       "step": 4327
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9160793573594664e-05,
-      "loss": 0.8356,
+      "loss": 0.9116,
       "step": 4328
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9160321191027073e-05,
-      "loss": 0.857,
+      "loss": 0.9989,
       "step": 4329
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9159848681373022e-05,
-      "loss": 0.7641,
+      "loss": 0.903,
       "step": 4330
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915937604463907e-05,
-      "loss": 0.8737,
+      "loss": 0.9546,
       "step": 4331
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915890328083177e-05,
-      "loss": 0.8642,
+      "loss": 1.0157,
       "step": 4332
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9158430389957677e-05,
-      "loss": 0.8239,
+      "loss": 0.946,
       "step": 4333
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915795737202336e-05,
-      "loss": 0.9114,
+      "loss": 1.0001,
       "step": 4334
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9157484227035378e-05,
-      "loss": 0.857,
+      "loss": 0.9685,
       "step": 4335
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9157010955000294e-05,
-      "loss": 0.8012,
+      "loss": 0.8752,
       "step": 4336
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9156537555924677e-05,
-      "loss": 0.8079,
+      "loss": 0.9042,
       "step": 4337
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915606402981509e-05,
-      "loss": 0.9219,
+      "loss": 1.0075,
       "step": 4338
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915559037667811e-05,
-      "loss": 0.7815,
+      "loss": 0.8759,
       "step": 4339
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.91551165965203e-05,
-      "loss": 0.8285,
+      "loss": 0.9553,
       "step": 4340
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9154642689348236e-05,
-      "loss": 0.9287,
+      "loss": 1.0323,
       "step": 4341
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9154168655168495e-05,
-      "loss": 0.894,
+      "loss": 1.0528,
       "step": 4342
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9153694493987653e-05,
-      "loss": 0.8806,
+      "loss": 0.9388,
       "step": 4343
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9153220205812285e-05,
-      "loss": 0.7087,
+      "loss": 0.9705,
       "step": 4344
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9152745790648973e-05,
-      "loss": 0.9071,
+      "loss": 0.945,
       "step": 4345
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9152271248504302e-05,
-      "loss": 0.8505,
+      "loss": 0.9583,
       "step": 4346
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.915179657938485e-05,
-      "loss": 0.9917,
+      "loss": 1.0212,
       "step": 4347
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9151321783297206e-05,
-      "loss": 0.7794,
+      "loss": 0.9805,
       "step": 4348
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9150846860247958e-05,
-      "loss": 0.9208,
+      "loss": 0.9884,
       "step": 4349
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9150371810243694e-05,
-      "loss": 0.8936,
+      "loss": 0.9996,
       "step": 4350
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9149896633291e-05,
-      "loss": 0.9306,
+      "loss": 1.0132,
       "step": 4351
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9149421329396474e-05,
-      "loss": 0.8806,
+      "loss": 0.9537,
       "step": 4352
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9148945898566708e-05,
-      "loss": 0.805,
+      "loss": 0.8964,
       "step": 4353
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.91484703408083e-05,
-      "loss": 0.8701,
+      "loss": 0.9845,
       "step": 4354
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9147994656127846e-05,
-      "loss": 0.7199,
+      "loss": 0.8213,
       "step": 4355
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9147518844531942e-05,
-      "loss": 0.8837,
+      "loss": 1.0087,
       "step": 4356
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9147042906027193e-05,
-      "loss": 0.8499,
+      "loss": 0.8662,
       "step": 4357
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9146566840620204e-05,
-      "loss": 0.817,
+      "loss": 0.9213,
       "step": 4358
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914609064831758e-05,
-      "loss": 0.7166,
+      "loss": 0.8984,
       "step": 4359
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914561432912592e-05,
-      "loss": 0.8511,
+      "loss": 0.963,
       "step": 4360
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914513788305184e-05,
-      "loss": 0.9267,
+      "loss": 0.9581,
       "step": 4361
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9144661310101943e-05,
-      "loss": 0.7984,
+      "loss": 0.9349,
       "step": 4362
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914418461028285e-05,
-      "loss": 0.9869,
+      "loss": 1.0054,
       "step": 4363
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9143707783601167e-05,
-      "loss": 0.8282,
+      "loss": 0.8641,
       "step": 4364
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914323083006351e-05,
-      "loss": 0.9152,
+      "loss": 1.0462,
       "step": 4365
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9142753749676502e-05,
-      "loss": 0.829,
+      "loss": 0.9091,
       "step": 4366
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9142276542446753e-05,
-      "loss": 0.8046,
+      "loss": 0.9339,
       "step": 4367
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9141799208380894e-05,
-      "loss": 0.8342,
+      "loss": 0.9512,
       "step": 4368
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9141321747485536e-05,
-      "loss": 0.812,
+      "loss": 0.9343,
       "step": 4369
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9140844159767312e-05,
-      "loss": 0.7902,
+      "loss": 0.9532,
       "step": 4370
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.914036644523284e-05,
-      "loss": 0.7743,
+      "loss": 0.9118,
       "step": 4371
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9139888603888755e-05,
-      "loss": 0.8628,
+      "loss": 0.9619,
       "step": 4372
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9139410635741686e-05,
-      "loss": 0.824,
+      "loss": 0.9943,
       "step": 4373
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913893254079826e-05,
-      "loss": 0.6902,
+      "loss": 0.927,
       "step": 4374
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913845431906511e-05,
-      "loss": 0.8304,
+      "loss": 0.9123,
       "step": 4375
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913797597054887e-05,
-      "loss": 0.8429,
+      "loss": 0.8741,
       "step": 4376
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9137497495256183e-05,
-      "loss": 0.8585,
+      "loss": 0.9684,
       "step": 4377
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913701889319368e-05,
-      "loss": 0.7386,
+      "loss": 0.9137,
       "step": 4378
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913654016436801e-05,
-      "loss": 0.8784,
+      "loss": 0.9736,
       "step": 4379
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9136061308785798e-05,
-      "loss": 0.9116,
+      "loss": 1.0171,
       "step": 4380
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9135582326453702e-05,
-      "loss": 0.8527,
+      "loss": 0.9293,
       "step": 4381
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9135103217378362e-05,
-      "loss": 0.8262,
+      "loss": 0.9716,
       "step": 4382
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913462398156643e-05,
-      "loss": 0.8514,
+      "loss": 0.8489,
       "step": 4383
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9134144619024547e-05,
-      "loss": 0.8671,
+      "loss": 0.9747,
       "step": 4384
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913366512975937e-05,
-      "loss": 0.9441,
+      "loss": 1.0319,
       "step": 4385
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9133185513777542e-05,
-      "loss": 0.8219,
+      "loss": 1.0171,
       "step": 4386
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9132705771085728e-05,
-      "loss": 0.808,
+      "loss": 0.8867,
       "step": 4387
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913222590169058e-05,
-      "loss": 0.8136,
+      "loss": 0.8807,
       "step": 4388
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913174590559875e-05,
-      "loss": 0.8831,
+      "loss": 1.0166,
       "step": 4389
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9131265782816908e-05,
-      "loss": 0.8215,
+      "loss": 0.9341,
       "step": 4390
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9130785533351706e-05,
-      "loss": 0.8813,
+      "loss": 1.002,
       "step": 4391
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.913030515720981e-05,
-      "loss": 0.9075,
+      "loss": 0.95,
       "step": 4392
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9129824654397882e-05,
-      "loss": 0.7686,
+      "loss": 0.9767,
       "step": 4393
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9129344024922593e-05,
-      "loss": 0.7282,
+      "loss": 0.9307,
       "step": 4394
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9128863268790606e-05,
-      "loss": 0.952,
+      "loss": 1.035,
       "step": 4395
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9128382386008596e-05,
-      "loss": 0.9817,
+      "loss": 1.0051,
       "step": 4396
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9127901376583228e-05,
-      "loss": 0.6801,
+      "loss": 0.9453,
       "step": 4397
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9127420240521183e-05,
-      "loss": 0.6677,
+      "loss": 0.9167,
       "step": 4398
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912693897782913e-05,
-      "loss": 0.9025,
+      "loss": 1.012,
       "step": 4399
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912645758851375e-05,
-      "loss": 0.8448,
+      "loss": 0.9033,
       "step": 4400
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912597607258172e-05,
-      "loss": 0.8853,
+      "loss": 0.9902,
       "step": 4401
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9125494430039717e-05,
-      "loss": 0.9534,
+      "loss": 0.9971,
       "step": 4402
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9125012660894428e-05,
-      "loss": 0.954,
+      "loss": 1.0556,
       "step": 4403
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9124530765152537e-05,
-      "loss": 0.948,
+      "loss": 1.0575,
       "step": 4404
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9124048742820725e-05,
-      "loss": 0.8027,
+      "loss": 0.905,
       "step": 4405
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9123566593905684e-05,
-      "loss": 0.7828,
+      "loss": 0.9043,
       "step": 4406
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9123084318414097e-05,
-      "loss": 0.7313,
+      "loss": 0.9062,
       "step": 4407
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912260191635266e-05,
-      "loss": 0.7004,
+      "loss": 0.8419,
       "step": 4408
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912211938772807e-05,
-      "loss": 0.8763,
+      "loss": 0.9775,
       "step": 4409
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9121636732547014e-05,
-      "loss": 0.9139,
+      "loss": 1.0287,
       "step": 4410
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9121153950816188e-05,
-      "loss": 0.7791,
+      "loss": 0.9683,
       "step": 4411
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9120671042542293e-05,
-      "loss": 0.8697,
+      "loss": 1.0038,
       "step": 4412
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.912018800773203e-05,
-      "loss": 0.7665,
+      "loss": 0.9907,
       "step": 4413
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9119704846392096e-05,
-      "loss": 0.8087,
+      "loss": 0.8563,
       "step": 4414
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9119221558529198e-05,
-      "loss": 0.8131,
+      "loss": 0.922,
       "step": 4415
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9118738144150037e-05,
-      "loss": 0.8246,
+      "loss": 0.8933,
       "step": 4416
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9118254603261324e-05,
-      "loss": 0.8123,
+      "loss": 0.9506,
       "step": 4417
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9117770935869763e-05,
-      "loss": 0.8395,
+      "loss": 0.9646,
       "step": 4418
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9117287141982067e-05,
-      "loss": 0.821,
+      "loss": 0.9163,
       "step": 4419
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911680322160495e-05,
-      "loss": 0.9215,
+      "loss": 0.9948,
       "step": 4420
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911631917474512e-05,
-      "loss": 0.7842,
+      "loss": 0.8413,
       "step": 4421
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.91158350014093e-05,
-      "loss": 0.714,
+      "loss": 0.8756,
       "step": 4422
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.91153507016042e-05,
-      "loss": 0.8308,
+      "loss": 0.8597,
       "step": 4423
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9114866275336546e-05,
-      "loss": 0.8016,
+      "loss": 0.9091,
       "step": 4424
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911438172261305e-05,
-      "loss": 0.7433,
+      "loss": 0.8959,
       "step": 4425
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9113897043440442e-05,
-      "loss": 0.7648,
+      "loss": 0.9363,
       "step": 4426
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911341223782544e-05,
-      "loss": 0.8355,
+      "loss": 0.9787,
       "step": 4427
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911292730577478e-05,
-      "loss": 0.7916,
+      "loss": 0.8846,
       "step": 4428
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9112442247295177e-05,
-      "loss": 0.9196,
+      "loss": 0.9658,
       "step": 4429
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.911195706239337e-05,
-      "loss": 0.8644,
+      "loss": 0.9719,
       "step": 4430
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9111471751076083e-05,
-      "loss": 0.8212,
+      "loss": 0.8492,
       "step": 4431
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9110986313350056e-05,
-      "loss": 0.8422,
+      "loss": 0.956,
       "step": 4432
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9110500749222024e-05,
-      "loss": 0.8796,
+      "loss": 1.0121,
       "step": 4433
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9110015058698715e-05,
-      "loss": 0.9633,
+      "loss": 1.0125,
       "step": 4434
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9109529241786876e-05,
-      "loss": 0.7617,
+      "loss": 0.9158,
       "step": 4435
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910904329849324e-05,
-      "loss": 0.9145,
+      "loss": 0.9516,
       "step": 4436
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9108557228824552e-05,
-      "loss": 0.8126,
+      "loss": 1.0084,
       "step": 4437
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910807103278756e-05,
-      "loss": 0.6796,
+      "loss": 0.8561,
       "step": 4438
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9107584710389e-05,
-      "loss": 0.7262,
+      "loss": 0.9189,
       "step": 4439
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910709826163563e-05,
-      "loss": 0.881,
+      "loss": 0.8928,
       "step": 4440
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9106611686534188e-05,
-      "loss": 0.8949,
+      "loss": 0.9375,
       "step": 4441
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9106124985091432e-05,
-      "loss": 0.8466,
+      "loss": 0.9702,
       "step": 4442
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910563815731411e-05,
-      "loss": 0.7547,
+      "loss": 0.8729,
       "step": 4443
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910515120320898e-05,
-      "loss": 0.8272,
+      "loss": 0.8389,
       "step": 4444
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910466412278279e-05,
-      "loss": 0.7762,
+      "loss": 0.9257,
       "step": 4445
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910417691604231e-05,
-      "loss": 0.8266,
+      "loss": 0.95,
       "step": 4446
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910368958299429e-05,
-      "loss": 0.707,
+      "loss": 0.956,
       "step": 4447
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.910320212364549e-05,
-      "loss": 0.8246,
+      "loss": 0.8942,
       "step": 4448
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9102714538002684e-05,
-      "loss": 0.7836,
+      "loss": 0.9393,
       "step": 4449
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9102226826072625e-05,
-      "loss": 0.8969,
+      "loss": 0.9261,
       "step": 4450
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9101738987862083e-05,
-      "loss": 0.6713,
+      "loss": 0.8331,
       "step": 4451
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9101251023377824e-05,
-      "loss": 0.8383,
+      "loss": 0.9077,
       "step": 4452
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9100762932626624e-05,
-      "loss": 0.7888,
+      "loss": 0.9355,
       "step": 4453
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9100274715615246e-05,
-      "loss": 0.7843,
+      "loss": 0.86,
       "step": 4454
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9099786372350474e-05,
-      "loss": 0.923,
+      "loss": 1.0165,
       "step": 4455
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909929790283907e-05,
-      "loss": 0.789,
+      "loss": 0.937,
       "step": 4456
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9098809307087823e-05,
-      "loss": 0.8068,
+      "loss": 0.9432,
       "step": 4457
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9098320585103503e-05,
-      "loss": 0.8416,
+      "loss": 0.9459,
       "step": 4458
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9097831736892894e-05,
-      "loss": 0.9239,
+      "loss": 1.0,
       "step": 4459
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909734276246278e-05,
-      "loss": 0.741,
+      "loss": 0.9648,
       "step": 4460
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9096853661819942e-05,
-      "loss": 0.8928,
+      "loss": 0.9319,
       "step": 4461
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909636443497117e-05,
-      "loss": 0.7728,
+      "loss": 0.753,
       "step": 4462
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909587508192324e-05,
-      "loss": 0.7482,
+      "loss": 0.8466,
       "step": 4463
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9095385602682952e-05,
-      "loss": 0.7823,
+      "loss": 0.9836,
       "step": 4464
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9094895997257095e-05,
-      "loss": 0.8718,
+      "loss": 0.9148,
       "step": 4465
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9094406265652457e-05,
-      "loss": 0.8874,
+      "loss": 0.996,
       "step": 4466
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9093916407875838e-05,
-      "loss": 0.9127,
+      "loss": 0.9763,
       "step": 4467
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909342642393403e-05,
-      "loss": 0.753,
+      "loss": 0.9309,
       "step": 4468
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909293631383383e-05,
-      "loss": 0.6867,
+      "loss": 0.9249,
       "step": 4469
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9092446077582044e-05,
-      "loss": 0.8009,
+      "loss": 0.9615,
       "step": 4470
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909195571518547e-05,
-      "loss": 0.7709,
+      "loss": 0.8845,
       "step": 4471
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.909146522665091e-05,
-      "loss": 0.8282,
+      "loss": 0.8994,
       "step": 4472
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9090974611985163e-05,
-      "loss": 0.9082,
+      "loss": 1.0072,
       "step": 4473
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9090483871195046e-05,
-      "loss": 0.8528,
+      "loss": 0.9675,
       "step": 4474
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908999300428736e-05,
-      "loss": 0.7796,
+      "loss": 0.9183,
       "step": 4475
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9089502011268923e-05,
-      "loss": 0.823,
+      "loss": 0.9012,
       "step": 4476
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908901089214654e-05,
-      "loss": 0.7833,
+      "loss": 0.9146,
       "step": 4477
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9088519646927025e-05,
-      "loss": 0.9086,
+      "loss": 0.9994,
       "step": 4478
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9088028275617193e-05,
-      "loss": 0.8583,
+      "loss": 0.9462,
       "step": 4479
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9087536778223868e-05,
-      "loss": 0.9685,
+      "loss": 1.0525,
       "step": 4480
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9087045154753863e-05,
-      "loss": 0.8234,
+      "loss": 0.9016,
       "step": 4481
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9086553405213995e-05,
-      "loss": 0.9075,
+      "loss": 0.9966,
       "step": 4482
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9086061529611094e-05,
-      "loss": 0.8355,
+      "loss": 1.0056,
       "step": 4483
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9085569527951978e-05,
-      "loss": 0.9297,
+      "loss": 0.9108,
       "step": 4484
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908507740024348e-05,
-      "loss": 0.7123,
+      "loss": 0.8493,
       "step": 4485
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908458514649242e-05,
-      "loss": 0.9279,
+      "loss": 0.9502,
       "step": 4486
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9084092766705632e-05,
-      "loss": 0.9235,
+      "loss": 1.0156,
       "step": 4487
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9083600260889944e-05,
-      "loss": 0.87,
+      "loss": 0.9173,
       "step": 4488
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9083107629052192e-05,
-      "loss": 0.7553,
+      "loss": 0.838,
       "step": 4489
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9082614871199208e-05,
-      "loss": 0.7774,
+      "loss": 0.9359,
       "step": 4490
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9082121987337828e-05,
-      "loss": 0.8494,
+      "loss": 0.9425,
       "step": 4491
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9081628977474893e-05,
-      "loss": 0.9305,
+      "loss": 0.9542,
       "step": 4492
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908113584161724e-05,
-      "loss": 0.7259,
+      "loss": 0.8583,
       "step": 4493
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9080642579771714e-05,
-      "loss": 0.9001,
+      "loss": 0.9229,
       "step": 4494
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.908014919194515e-05,
-      "loss": 0.7962,
+      "loss": 0.9053,
       "step": 4495
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9079655678144403e-05,
-      "loss": 0.8538,
+      "loss": 0.9694,
       "step": 4496
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9079162038376316e-05,
-      "loss": 0.8849,
+      "loss": 0.9613,
       "step": 4497
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907866827264774e-05,
-      "loss": 0.7961,
+      "loss": 0.9167,
       "step": 4498
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9078174380965518e-05,
-      "loss": 0.7965,
+      "loss": 0.8619,
       "step": 4499
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907768036333651e-05,
-      "loss": 0.9677,
+      "loss": 1.0461,
       "step": 4500
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9077186219767566e-05,
-      "loss": 0.8796,
+      "loss": 1.0087,
       "step": 4501
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907669195026554e-05,
-      "loss": 0.7886,
+      "loss": 1.007,
       "step": 4502
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907619755483729e-05,
-      "loss": 0.8867,
+      "loss": 1.0006,
       "step": 4503
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907570303348968e-05,
-      "loss": 0.5957,
+      "loss": 0.7973,
       "step": 4504
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9075208386229567e-05,
-      "loss": 0.7834,
+      "loss": 0.9446,
       "step": 4505
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9074713613063814e-05,
-      "loss": 0.795,
+      "loss": 0.944,
       "step": 4506
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907421871399928e-05,
-      "loss": 0.7715,
+      "loss": 0.9059,
       "step": 4507
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9073723689042844e-05,
-      "loss": 0.8131,
+      "loss": 0.9138,
       "step": 4508
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907322853820136e-05,
-      "loss": 0.9478,
+      "loss": 0.9855,
       "step": 4509
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9072733261481705e-05,
-      "loss": 0.9565,
+      "loss": 1.0387,
       "step": 4510
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907223785889075e-05,
-      "loss": 0.7518,
+      "loss": 0.8601,
       "step": 4511
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907174233043536e-05,
-      "loss": 0.8051,
+      "loss": 0.9158,
       "step": 4512
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907124667612242e-05,
-      "loss": 0.8296,
+      "loss": 0.9448,
       "step": 4513
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9070750895958804e-05,
-      "loss": 0.7862,
+      "loss": 0.8862,
       "step": 4514
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.907025498995139e-05,
-      "loss": 0.8238,
+      "loss": 0.9368,
       "step": 4515
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9069758958107058e-05,
-      "loss": 0.8309,
+      "loss": 0.9139,
       "step": 4516
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9069262800432688e-05,
-      "loss": 0.8124,
+      "loss": 0.9935,
       "step": 4517
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906876651693516e-05,
-      "loss": 0.7893,
+      "loss": 0.9151,
       "step": 4518
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9068270107621367e-05,
-      "loss": 0.9481,
+      "loss": 1.0869,
       "step": 4519
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9067773572498192e-05,
-      "loss": 0.8252,
+      "loss": 0.8992,
       "step": 4520
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906727691157252e-05,
-      "loss": 0.726,
+      "loss": 0.9183,
       "step": 4521
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9066780124851254e-05,
-      "loss": 0.824,
+      "loss": 0.9113,
       "step": 4522
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906628321234127e-05,
-      "loss": 0.8896,
+      "loss": 1.0099,
       "step": 4523
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9065786174049473e-05,
-      "loss": 0.8339,
+      "loss": 0.9673,
       "step": 4524
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906528900998276e-05,
-      "loss": 0.8388,
+      "loss": 0.9239,
       "step": 4525
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9064791720148018e-05,
-      "loss": 1.0197,
+      "loss": 1.0993,
       "step": 4526
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9064294304552153e-05,
-      "loss": 0.89,
+      "loss": 0.9839,
       "step": 4527
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9063796763202064e-05,
-      "loss": 0.8049,
+      "loss": 0.8946,
       "step": 4528
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9063299096104658e-05,
-      "loss": 0.7586,
+      "loss": 0.8981,
       "step": 4529
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9062801303266837e-05,
-      "loss": 0.775,
+      "loss": 0.9386,
       "step": 4530
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.90623033846955e-05,
-      "loss": 0.7605,
+      "loss": 0.886,
       "step": 4531
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906180534039757e-05,
-      "loss": 0.8669,
+      "loss": 0.987,
       "step": 4532
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906130717037994e-05,
-      "loss": 0.894,
+      "loss": 0.9666,
       "step": 4533
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9060808874649536e-05,
-      "loss": 0.8153,
+      "loss": 0.8827,
       "step": 4534
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.906031045321326e-05,
-      "loss": 0.7797,
+      "loss": 0.9048,
       "step": 4535
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9059811906078032e-05,
-      "loss": 0.9504,
+      "loss": 0.9542,
       "step": 4536
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9059313233250767e-05,
-      "loss": 0.7709,
+      "loss": 0.967,
       "step": 4537
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9058814434738386e-05,
-      "loss": 0.8416,
+      "loss": 0.9905,
       "step": 4538
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9058315510547806e-05,
-      "loss": 0.6615,
+      "loss": 0.8989,
       "step": 4539
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9057816460685953e-05,
-      "loss": 0.6971,
+      "loss": 0.7831,
       "step": 4540
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9057317285159745e-05,
-      "loss": 0.925,
+      "loss": 0.9544,
       "step": 4541
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905681798397611e-05,
-      "loss": 0.7818,
+      "loss": 0.8867,
       "step": 4542
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905631855714198e-05,
-      "loss": 0.7216,
+      "loss": 0.8765,
       "step": 4543
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905581900466428e-05,
-      "loss": 0.7825,
+      "loss": 0.8922,
       "step": 4544
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9055319326549933e-05,
-      "loss": 0.8715,
+      "loss": 0.9196,
       "step": 4545
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9054819522805884e-05,
-      "loss": 0.8782,
+      "loss": 0.9431,
       "step": 4546
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905431959343906e-05,
-      "loss": 0.8502,
+      "loss": 0.9915,
       "step": 4547
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9053819538456394e-05,
-      "loss": 1.0392,
+      "loss": 0.9857,
       "step": 4548
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905331935786483e-05,
-      "loss": 0.8321,
+      "loss": 0.9724,
       "step": 4549
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9052819051671307e-05,
-      "loss": 0.7773,
+      "loss": 0.8394,
       "step": 4550
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9052318619882763e-05,
-      "loss": 0.8484,
+      "loss": 0.9102,
       "step": 4551
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.905181806250614e-05,
-      "loss": 0.7533,
+      "loss": 0.8743,
       "step": 4552
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9051317379548387e-05,
-      "loss": 0.9256,
+      "loss": 0.9155,
       "step": 4553
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9050816571016444e-05,
-      "loss": 0.8785,
+      "loss": 0.9919,
       "step": 4554
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9050315636917264e-05,
-      "loss": 0.9921,
+      "loss": 1.0048,
       "step": 4555
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9049814577257795e-05,
-      "loss": 0.9094,
+      "loss": 0.9242,
       "step": 4556
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904931339204499e-05,
-      "loss": 0.7483,
+      "loss": 0.9094,
       "step": 4557
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90488120812858e-05,
-      "loss": 0.7364,
+      "loss": 0.8631,
       "step": 4558
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9048310644987182e-05,
-      "loss": 0.8334,
+      "loss": 0.9105,
       "step": 4559
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904780908315609e-05,
-      "loss": 0.8255,
+      "loss": 0.9214,
       "step": 4560
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9047307395799484e-05,
-      "loss": 0.7089,
+      "loss": 0.8107,
       "step": 4561
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904680558292433e-05,
-      "loss": 0.7937,
+      "loss": 0.8925,
       "step": 4562
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9046303644537577e-05,
-      "loss": 0.7899,
+      "loss": 0.8827,
       "step": 4563
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90458015806462e-05,
-      "loss": 0.8131,
+      "loss": 0.9304,
       "step": 4564
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9045299391257155e-05,
-      "loss": 0.7915,
+      "loss": 0.8952,
       "step": 4565
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904479707637742e-05,
-      "loss": 0.8292,
+      "loss": 0.9173,
       "step": 4566
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9044294636013957e-05,
-      "loss": 0.6812,
+      "loss": 0.9233,
       "step": 4567
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9043792070173736e-05,
-      "loss": 0.8885,
+      "loss": 0.9976,
       "step": 4568
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9043289378863737e-05,
-      "loss": 0.9028,
+      "loss": 1.0504,
       "step": 4569
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9042786562090923e-05,
-      "loss": 0.9014,
+      "loss": 1.0243,
       "step": 4570
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9042283619862278e-05,
-      "loss": 0.8184,
+      "loss": 0.9437,
       "step": 4571
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904178055218478e-05,
-      "loss": 0.7881,
+      "loss": 0.8927,
       "step": 4572
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90412773590654e-05,
-      "loss": 0.813,
+      "loss": 0.9243,
       "step": 4573
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9040774040511127e-05,
-      "loss": 0.9073,
+      "loss": 1.0239,
       "step": 4574
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.904027059652894e-05,
-      "loss": 0.7462,
+      "loss": 0.8549,
       "step": 4575
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9039767027125827e-05,
-      "loss": 0.9076,
+      "loss": 1.0208,
       "step": 4576
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903926333230877e-05,
-      "loss": 0.8417,
+      "loss": 1.0209,
       "step": 4577
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9038759512084762e-05,
-      "loss": 0.8351,
+      "loss": 0.9396,
       "step": 4578
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903825556646079e-05,
-      "loss": 0.8422,
+      "loss": 1.0289,
       "step": 4579
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903775149544384e-05,
-      "loss": 0.7414,
+      "loss": 0.9165,
       "step": 4580
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9037247299040916e-05,
-      "loss": 0.8244,
+      "loss": 0.8986,
       "step": 4581
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903674297725901e-05,
-      "loss": 0.8135,
+      "loss": 0.8994,
       "step": 4582
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9036238530105114e-05,
-      "loss": 0.8529,
+      "loss": 0.9507,
       "step": 4583
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9035733957586226e-05,
-      "loss": 0.8652,
+      "loss": 0.9208,
       "step": 4584
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9035229259709354e-05,
-      "loss": 0.8075,
+      "loss": 0.9219,
       "step": 4585
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9034724436481494e-05,
-      "loss": 0.8979,
+      "loss": 0.9832,
       "step": 4586
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903421948790965e-05,
-      "loss": 0.8042,
+      "loss": 0.9274,
       "step": 4587
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9033714414000826e-05,
-      "loss": 0.8488,
+      "loss": 0.982,
       "step": 4588
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903320921476204e-05,
-      "loss": 0.8367,
+      "loss": 0.9228,
       "step": 4589
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9032703890200284e-05,
-      "loss": 0.8522,
+      "loss": 0.9499,
       "step": 4590
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903219844032258e-05,
-      "loss": 0.8502,
+      "loss": 0.9877,
       "step": 4591
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.903169286513594e-05,
-      "loss": 0.9959,
+      "loss": 1.021,
       "step": 4592
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9031187164647375e-05,
-      "loss": 0.9313,
+      "loss": 1.0039,
       "step": 4593
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9030681338863898e-05,
-      "loss": 0.7942,
+      "loss": 1.0054,
       "step": 4594
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9030175387792535e-05,
-      "loss": 0.8175,
+      "loss": 0.9174,
       "step": 4595
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9029669311440296e-05,
-      "loss": 0.8659,
+      "loss": 0.9589,
       "step": 4596
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9029163109814213e-05,
-      "loss": 0.9509,
+      "loss": 1.0197,
       "step": 4597
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90286567829213e-05,
-      "loss": 0.699,
+      "loss": 0.8491,
       "step": 4598
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9028150330768584e-05,
-      "loss": 0.9108,
+      "loss": 0.9647,
       "step": 4599
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.902764375336309e-05,
-      "loss": 0.7709,
+      "loss": 0.9821,
       "step": 4600
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9027137050711846e-05,
-      "loss": 0.8068,
+      "loss": 0.8965,
       "step": 4601
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9026630222821883e-05,
-      "loss": 0.9457,
+      "loss": 1.0325,
       "step": 4602
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9026123269700237e-05,
-      "loss": 0.8804,
+      "loss": 1.0837,
       "step": 4603
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9025616191353935e-05,
-      "loss": 0.8731,
+      "loss": 0.9263,
       "step": 4604
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9025108987790016e-05,
-      "loss": 0.8074,
+      "loss": 0.8466,
       "step": 4605
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9024601659015513e-05,
-      "loss": 0.769,
+      "loss": 0.9463,
       "step": 4606
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9024094205037464e-05,
-      "loss": 0.7976,
+      "loss": 0.9135,
       "step": 4607
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9023586625862914e-05,
-      "loss": 0.9098,
+      "loss": 0.9744,
       "step": 4608
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90230789214989e-05,
-      "loss": 0.8036,
+      "loss": 0.9059,
       "step": 4609
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.902257109195247e-05,
-      "loss": 0.8211,
+      "loss": 0.9126,
       "step": 4610
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9022063137230668e-05,
-      "loss": 0.8513,
+      "loss": 0.9837,
       "step": 4611
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.902155505734054e-05,
-      "loss": 0.7791,
+      "loss": 0.9431,
       "step": 4612
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9021046852289134e-05,
-      "loss": 0.8407,
+      "loss": 0.9663,
       "step": 4613
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9020538522083505e-05,
-      "loss": 0.8903,
+      "loss": 0.9334,
       "step": 4614
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9020030066730702e-05,
-      "loss": 0.7178,
+      "loss": 0.8461,
       "step": 4615
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9019521486237777e-05,
-      "loss": 0.832,
+      "loss": 0.9239,
       "step": 4616
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9019012780611793e-05,
-      "loss": 0.7971,
+      "loss": 0.9875,
       "step": 4617
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9018503949859795e-05,
-      "loss": 1.0213,
+      "loss": 1.0455,
       "step": 4618
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9017994993988857e-05,
-      "loss": 0.8301,
+      "loss": 0.9267,
       "step": 4619
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9017485913006034e-05,
-      "loss": 0.8546,
+      "loss": 0.8978,
       "step": 4620
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9016976706918386e-05,
-      "loss": 0.8471,
+      "loss": 0.9183,
       "step": 4621
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9016467375732976e-05,
-      "loss": 0.713,
+      "loss": 0.8638,
       "step": 4622
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9015957919456878e-05,
-      "loss": 0.9539,
+      "loss": 1.0247,
       "step": 4623
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9015448338097157e-05,
-      "loss": 0.8625,
+      "loss": 0.8636,
       "step": 4624
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.901493863166088e-05,
-      "loss": 0.7183,
+      "loss": 0.9102,
       "step": 4625
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9014428800155117e-05,
-      "loss": 0.7502,
+      "loss": 0.9313,
       "step": 4626
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9013918843586946e-05,
-      "loss": 0.8542,
+      "loss": 0.9979,
       "step": 4627
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9013408761963445e-05,
-      "loss": 0.9534,
+      "loss": 1.0563,
       "step": 4628
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.901289855529168e-05,
-      "loss": 0.9269,
+      "loss": 0.9584,
       "step": 4629
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9012388223578736e-05,
-      "loss": 0.8338,
+      "loss": 1.0294,
       "step": 4630
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.901187776683169e-05,
-      "loss": 0.7944,
+      "loss": 0.8552,
       "step": 4631
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9011367185057633e-05,
-      "loss": 0.869,
+      "loss": 0.9964,
       "step": 4632
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9010856478263636e-05,
-      "loss": 0.9721,
+      "loss": 1.0367,
       "step": 4633
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9010345646456795e-05,
-      "loss": 0.8282,
+      "loss": 0.8336,
       "step": 4634
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9009834689644185e-05,
-      "loss": 0.7713,
+      "loss": 0.8433,
       "step": 4635
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.900932360783291e-05,
-      "loss": 0.8909,
+      "loss": 0.92,
       "step": 4636
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.900881240103005e-05,
-      "loss": 0.7933,
+      "loss": 0.8793,
       "step": 4637
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9008301069242695e-05,
-      "loss": 0.9407,
+      "loss": 0.9772,
       "step": 4638
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9007789612477948e-05,
-      "loss": 0.8522,
+      "loss": 0.9633,
       "step": 4639
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9007278030742898e-05,
-      "loss": 0.8054,
+      "loss": 1.0042,
       "step": 4640
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9006766324044647e-05,
-      "loss": 0.9155,
+      "loss": 1.0198,
       "step": 4641
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9006254492390293e-05,
-      "loss": 0.7541,
+      "loss": 0.9289,
       "step": 4642
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9005742535786935e-05,
-      "loss": 0.7955,
+      "loss": 0.9075,
       "step": 4643
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.900523045424168e-05,
-      "loss": 0.9219,
+      "loss": 0.9998,
       "step": 4644
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9004718247761623e-05,
-      "loss": 0.8593,
+      "loss": 0.946,
       "step": 4645
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.900420591635388e-05,
-      "loss": 0.8389,
+      "loss": 0.9142,
       "step": 4646
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9003693460025555e-05,
-      "loss": 0.8161,
+      "loss": 0.9803,
       "step": 4647
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9003180878783757e-05,
-      "loss": 0.8342,
+      "loss": 0.9345,
       "step": 4648
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.90026681726356e-05,
-      "loss": 0.8305,
+      "loss": 0.9019,
       "step": 4649
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9002155341588194e-05,
-      "loss": 0.8001,
+      "loss": 0.9358,
       "step": 4650
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9001642385648652e-05,
-      "loss": 0.7455,
+      "loss": 0.903,
       "step": 4651
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9001129304824097e-05,
-      "loss": 0.922,
+      "loss": 0.9636,
       "step": 4652
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9000616099121645e-05,
-      "loss": 0.8078,
+      "loss": 0.8876,
       "step": 4653
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9000102768548413e-05,
-      "loss": 0.8856,
+      "loss": 0.9745,
       "step": 4654
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8999589313111523e-05,
-      "loss": 0.9736,
+      "loss": 1.0755,
       "step": 4655
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8999075732818104e-05,
-      "loss": 0.8853,
+      "loss": 0.9766,
       "step": 4656
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8998562027675274e-05,
-      "loss": 0.6555,
+      "loss": 0.8502,
       "step": 4657
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8998048197690163e-05,
-      "loss": 0.989,
+      "loss": 0.9486,
       "step": 4658
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.89975342428699e-05,
-      "loss": 0.8008,
+      "loss": 0.9431,
       "step": 4659
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.899702016322162e-05,
-      "loss": 0.8168,
+      "loss": 0.9055,
       "step": 4660
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8996505958752443e-05,
-      "loss": 0.875,
+      "loss": 0.9539,
       "step": 4661
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8995991629469514e-05,
-      "loss": 0.6618,
+      "loss": 0.831,
       "step": 4662
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8995477175379962e-05,
-      "loss": 0.7792,
+      "loss": 0.9609,
       "step": 4663
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.899496259649093e-05,
-      "loss": 0.8151,
+      "loss": 0.9733,
       "step": 4664
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8994447892809556e-05,
-      "loss": 0.7965,
+      "loss": 0.8831,
       "step": 4665
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8993933064342976e-05,
-      "loss": 0.818,
+      "loss": 0.9498,
       "step": 4666
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8993418111098335e-05,
-      "loss": 0.8968,
+      "loss": 0.9729,
       "step": 4667
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.899290303308278e-05,
-      "loss": 0.9059,
+      "loss": 1.0085,
       "step": 4668
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8992387830303453e-05,
-      "loss": 0.9434,
+      "loss": 0.9753,
       "step": 4669
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8991872502767506e-05,
-      "loss": 0.7561,
+      "loss": 0.8806,
       "step": 4670
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.899135705048208e-05,
-      "loss": 0.7144,
+      "loss": 0.8582,
       "step": 4671
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8990841473454338e-05,
-      "loss": 0.8402,
+      "loss": 0.8813,
       "step": 4672
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8990325771691425e-05,
-      "loss": 0.8925,
+      "loss": 0.9595,
       "step": 4673
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8989809945200495e-05,
-      "loss": 0.9449,
+      "loss": 1.0007,
       "step": 4674
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.898929399398871e-05,
-      "loss": 0.8891,
+      "loss": 0.8932,
       "step": 4675
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8988777918063223e-05,
-      "loss": 0.9254,
+      "loss": 0.9624,
       "step": 4676
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8988261717431196e-05,
-      "loss": 1.0014,
+      "loss": 1.0194,
       "step": 4677
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8987745392099792e-05,
-      "loss": 0.7256,
+      "loss": 0.8636,
       "step": 4678
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8987228942076173e-05,
-      "loss": 0.8155,
+      "loss": 0.8804,
       "step": 4679
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.89867123673675e-05,
-      "loss": 0.6963,
+      "loss": 0.8634,
       "step": 4680
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8986195667980947e-05,
-      "loss": 0.781,
+      "loss": 0.8569,
       "step": 4681
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.898567884392368e-05,
-      "loss": 0.8118,
+      "loss": 0.9514,
       "step": 4682
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8985161895202863e-05,
-      "loss": 0.9169,
+      "loss": 1.046,
       "step": 4683
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8984644821825675e-05,
-      "loss": 0.7642,
+      "loss": 0.8945,
       "step": 4684
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8984127623799288e-05,
-      "loss": 0.9372,
+      "loss": 0.976,
       "step": 4685
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8983610301130878e-05,
-      "loss": 0.9017,
+      "loss": 0.9415,
       "step": 4686
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.898309285382762e-05,
-      "loss": 0.9533,
+      "loss": 1.0109,
       "step": 4687
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8982575281896697e-05,
-      "loss": 0.7683,
+      "loss": 0.9186,
       "step": 4688
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.898205758534528e-05,
-      "loss": 0.9174,
+      "loss": 1.0931,
       "step": 4689
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8981539764180563e-05,
-      "loss": 0.8161,
+      "loss": 0.8765,
       "step": 4690
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8981021818409723e-05,
-      "loss": 0.8934,
+      "loss": 0.9982,
       "step": 4691
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.898050374803995e-05,
-      "loss": 0.8753,
+      "loss": 0.9814,
       "step": 4692
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8979985553078427e-05,
-      "loss": 0.8018,
+      "loss": 0.8886,
       "step": 4693
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8979467233532343e-05,
-      "loss": 0.8264,
+      "loss": 0.9592,
       "step": 4694
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8978948789408896e-05,
-      "loss": 0.822,
+      "loss": 0.8416,
       "step": 4695
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8978430220715275e-05,
-      "loss": 0.9222,
+      "loss": 1.0096,
       "step": 4696
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.897791152745867e-05,
-      "loss": 0.7289,
+      "loss": 0.8288,
       "step": 4697
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.897739270964628e-05,
-      "loss": 0.9964,
+      "loss": 0.9969,
       "step": 4698
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8976873767285308e-05,
-      "loss": 0.7982,
+      "loss": 0.8978,
       "step": 4699
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8976354700382942e-05,
-      "loss": 0.7841,
+      "loss": 0.9312,
       "step": 4700
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8975835508946395e-05,
-      "loss": 0.8397,
+      "loss": 0.9567,
       "step": 4701
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8975316192982865e-05,
-      "loss": 0.8205,
+      "loss": 0.9948,
       "step": 4702
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8974796752499556e-05,
-      "loss": 0.7829,
+      "loss": 0.8695,
       "step": 4703
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8974277187503675e-05,
-      "loss": 0.8006,
+      "loss": 0.9492,
       "step": 4704
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.897375749800243e-05,
-      "loss": 0.9249,
+      "loss": 1.0391,
       "step": 4705
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8973237684003036e-05,
-      "loss": 0.7472,
+      "loss": 0.8559,
       "step": 4706
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.89727177455127e-05,
-      "loss": 0.6936,
+      "loss": 0.896,
       "step": 4707
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.897219768253863e-05,
-      "loss": 0.7946,
+      "loss": 0.8969,
       "step": 4708
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8971677495088054e-05,
-      "loss": 0.7615,
+      "loss": 0.869,
       "step": 4709
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.897115718316818e-05,
-      "loss": 0.7917,
+      "loss": 0.9622,
       "step": 4710
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8970636746786225e-05,
-      "loss": 0.7926,
+      "loss": 0.9565,
       "step": 4711
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8970116185949418e-05,
-      "loss": 0.8501,
+      "loss": 0.9985,
       "step": 4712
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.896959550066497e-05,
-      "loss": 0.89,
+      "loss": 0.9672,
       "step": 4713
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8969074690940113e-05,
-      "loss": 0.8714,
+      "loss": 1.0323,
       "step": 4714
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8968553756782073e-05,
-      "loss": 0.8356,
+      "loss": 0.9153,
       "step": 4715
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8968032698198074e-05,
-      "loss": 0.9382,
+      "loss": 0.9561,
       "step": 4716
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8967511515195343e-05,
-      "loss": 0.9046,
+      "loss": 1.0264,
       "step": 4717
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.896699020778111e-05,
-      "loss": 0.7621,
+      "loss": 0.8573,
       "step": 4718
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8966468775962615e-05,
-      "loss": 0.843,
+      "loss": 0.9788,
       "step": 4719
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.896594721974709e-05,
-      "loss": 0.8002,
+      "loss": 0.9528,
       "step": 4720
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.896542553914176e-05,
-      "loss": 0.906,
+      "loss": 1.0731,
       "step": 4721
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8964903734153878e-05,
-      "loss": 0.9412,
+      "loss": 1.03,
       "step": 4722
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8964381804790674e-05,
-      "loss": 0.8422,
+      "loss": 0.9278,
       "step": 4723
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8963859751059392e-05,
-      "loss": 0.8692,
+      "loss": 0.9108,
       "step": 4724
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.896333757296727e-05,
-      "loss": 0.8253,
+      "loss": 0.9686,
       "step": 4725
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8962815270521563e-05,
-      "loss": 0.9277,
+      "loss": 1.0449,
       "step": 4726
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8962292843729506e-05,
-      "loss": 0.8624,
+      "loss": 0.9507,
       "step": 4727
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8961770292598353e-05,
-      "loss": 0.9027,
+      "loss": 0.961,
       "step": 4728
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8961247617135353e-05,
-      "loss": 0.7468,
+      "loss": 0.9636,
       "step": 4729
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8960724817347755e-05,
-      "loss": 0.8613,
+      "loss": 0.8906,
       "step": 4730
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8960201893242814e-05,
-      "loss": 0.7823,
+      "loss": 0.9616,
       "step": 4731
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8959678844827786e-05,
-      "loss": 0.8138,
+      "loss": 0.9774,
       "step": 4732
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8959155672109927e-05,
-      "loss": 0.774,
+      "loss": 0.9111,
       "step": 4733
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8958632375096492e-05,
-      "loss": 0.8767,
+      "loss": 1.0056,
       "step": 4734
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8958108953794744e-05,
-      "loss": 0.8604,
+      "loss": 0.9238,
       "step": 4735
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8957585408211945e-05,
-      "loss": 0.8783,
+      "loss": 0.9305,
       "step": 4736
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8957061738355356e-05,
-      "loss": 0.7561,
+      "loss": 0.939,
       "step": 4737
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8956537944232246e-05,
-      "loss": 0.7198,
+      "loss": 0.7809,
       "step": 4738
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8956014025849878e-05,
-      "loss": 0.7643,
+      "loss": 0.8253,
       "step": 4739
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8955489983215525e-05,
-      "loss": 0.8321,
+      "loss": 0.8735,
       "step": 4740
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.895496581633645e-05,
-      "loss": 0.8595,
+      "loss": 1.004,
       "step": 4741
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8954441525219935e-05,
-      "loss": 0.7848,
+      "loss": 0.964,
       "step": 4742
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8953917109873243e-05,
-      "loss": 0.7873,
+      "loss": 0.872,
       "step": 4743
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.895339257030366e-05,
-      "loss": 0.7237,
+      "loss": 0.8618,
       "step": 4744
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8952867906518454e-05,
-      "loss": 0.8447,
+      "loss": 0.9529,
       "step": 4745
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.895234311852491e-05,
-      "loss": 0.9746,
+      "loss": 1.0185,
       "step": 4746
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8951818206330307e-05,
-      "loss": 0.8238,
+      "loss": 0.9262,
       "step": 4747
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8951293169941928e-05,
-      "loss": 0.8014,
+      "loss": 0.8449,
       "step": 4748
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8950768009367057e-05,
-      "loss": 0.8983,
+      "loss": 0.9248,
       "step": 4749
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8950242724612978e-05,
-      "loss": 1.0399,
+      "loss": 1.0159,
       "step": 4750
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8949717315686983e-05,
-      "loss": 0.8998,
+      "loss": 0.9865,
       "step": 4751
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8949191782596356e-05,
-      "loss": 0.9068,
+      "loss": 0.9695,
       "step": 4752
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8948666125348388e-05,
-      "loss": 1.0051,
+      "loss": 1.0157,
       "step": 4753
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8948140343950375e-05,
-      "loss": 0.9123,
+      "loss": 0.9776,
       "step": 4754
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8947614438409613e-05,
-      "loss": 0.7995,
+      "loss": 0.9215,
       "step": 4755
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.894708840873339e-05,
-      "loss": 0.8432,
+      "loss": 0.9091,
       "step": 4756
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8946562254929018e-05,
-      "loss": 0.8207,
+      "loss": 0.886,
       "step": 4757
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8946035977003784e-05,
-      "loss": 0.7746,
+      "loss": 0.9738,
       "step": 4758
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8945509574964994e-05,
-      "loss": 0.9077,
+      "loss": 1.0076,
       "step": 4759
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.894498304881995e-05,
-      "loss": 0.8535,
+      "loss": 0.9932,
       "step": 4760
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8944456398575955e-05,
-      "loss": 0.7707,
+      "loss": 0.8856,
       "step": 4761
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.894392962424032e-05,
-      "loss": 0.7972,
+      "loss": 0.933,
       "step": 4762
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.894340272582035e-05,
-      "loss": 0.775,
+      "loss": 0.8681,
       "step": 4763
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.894287570332336e-05,
-      "loss": 0.8655,
+      "loss": 0.9457,
       "step": 4764
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8942348556756655e-05,
-      "loss": 0.8609,
+      "loss": 0.9846,
       "step": 4765
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8941821286127556e-05,
-      "loss": 0.9215,
+      "loss": 1.0017,
       "step": 4766
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8941293891443368e-05,
-      "loss": 0.8548,
+      "loss": 0.9574,
       "step": 4767
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8940766372711414e-05,
-      "loss": 0.7764,
+      "loss": 0.8912,
       "step": 4768
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8940238729939014e-05,
-      "loss": 0.8929,
+      "loss": 0.9722,
       "step": 4769
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8939710963133485e-05,
-      "loss": 0.8103,
+      "loss": 0.9109,
       "step": 4770
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.893918307230215e-05,
-      "loss": 0.8002,
+      "loss": 0.96,
       "step": 4771
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8938655057452333e-05,
-      "loss": 0.9008,
+      "loss": 0.9849,
       "step": 4772
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.893812691859136e-05,
-      "loss": 0.8343,
+      "loss": 0.9375,
       "step": 4773
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8937598655726556e-05,
-      "loss": 0.7936,
+      "loss": 0.9144,
       "step": 4774
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8937070268865252e-05,
-      "loss": 0.8916,
+      "loss": 0.981,
       "step": 4775
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8936541758014778e-05,
-      "loss": 0.7961,
+      "loss": 0.9323,
       "step": 4776
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.893601312318247e-05,
-      "loss": 0.8417,
+      "loss": 0.9675,
       "step": 4777
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8935484364375652e-05,
-      "loss": 0.8785,
+      "loss": 0.9765,
       "step": 4778
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.893495548160167e-05,
-      "loss": 0.8284,
+      "loss": 0.9033,
       "step": 4779
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.893442647486786e-05,
-      "loss": 0.9175,
+      "loss": 1.0042,
       "step": 4780
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8933897344181552e-05,
-      "loss": 0.885,
+      "loss": 0.9247,
       "step": 4781
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.89333680895501e-05,
-      "loss": 0.8741,
+      "loss": 0.9561,
       "step": 4782
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8932838710980838e-05,
-      "loss": 0.7899,
+      "loss": 0.8951,
       "step": 4783
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8932309208481117e-05,
-      "loss": 0.9779,
+      "loss": 1.0357,
       "step": 4784
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8931779582058273e-05,
-      "loss": 0.8706,
+      "loss": 0.9901,
       "step": 4785
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8931249831719666e-05,
-      "loss": 0.7238,
+      "loss": 0.8508,
       "step": 4786
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8930719957472634e-05,
-      "loss": 0.7538,
+      "loss": 0.9206,
       "step": 4787
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8930189959324536e-05,
-      "loss": 0.8125,
+      "loss": 0.9138,
       "step": 4788
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892965983728272e-05,
-      "loss": 0.7863,
+      "loss": 0.8924,
       "step": 4789
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892912959135455e-05,
-      "loss": 0.881,
+      "loss": 0.9307,
       "step": 4790
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8928599221547367e-05,
-      "loss": 0.7912,
+      "loss": 0.9026,
       "step": 4791
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8928068727868542e-05,
-      "loss": 0.7365,
+      "loss": 0.8719,
       "step": 4792
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892753811032543e-05,
-      "loss": 0.9434,
+      "loss": 0.9735,
       "step": 4793
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8927007368925396e-05,
-      "loss": 0.8718,
+      "loss": 0.9402,
       "step": 4794
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8926476503675796e-05,
-      "loss": 0.8475,
+      "loss": 0.9604,
       "step": 4795
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8925945514584002e-05,
-      "loss": 0.8893,
+      "loss": 1.0012,
       "step": 4796
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8925414401657377e-05,
-      "loss": 0.8292,
+      "loss": 0.9327,
       "step": 4797
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8924883164903294e-05,
-      "loss": 0.8081,
+      "loss": 0.9611,
       "step": 4798
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8924351804329118e-05,
-      "loss": 0.7352,
+      "loss": 0.9542,
       "step": 4799
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892382031994222e-05,
-      "loss": 0.8378,
+      "loss": 0.9303,
       "step": 4800
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892328871174998e-05,
-      "loss": 0.7247,
+      "loss": 0.9012,
       "step": 4801
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.892275697975977e-05,
-      "loss": 0.9361,
+      "loss": 0.9975,
       "step": 4802
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8922225123978967e-05,
-      "loss": 0.7944,
+      "loss": 0.9163,
       "step": 4803
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8921693144414943e-05,
-      "loss": 0.968,
+      "loss": 1.0519,
       "step": 4804
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8921161041075092e-05,
-      "loss": 0.7728,
+      "loss": 0.8923,
       "step": 4805
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8920628813966786e-05,
-      "loss": 0.8207,
+      "loss": 0.9393,
       "step": 4806
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8920096463097413e-05,
-      "loss": 0.8514,
+      "loss": 0.9033,
       "step": 4807
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8919563988474354e-05,
-      "loss": 0.8061,
+      "loss": 0.9853,
       "step": 4808
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8919031390105003e-05,
-      "loss": 0.7391,
+      "loss": 0.9541,
       "step": 4809
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8918498667996745e-05,
-      "loss": 0.8225,
+      "loss": 0.934,
       "step": 4810
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8917965822156974e-05,
-      "loss": 0.8224,
+      "loss": 0.9111,
       "step": 4811
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.891743285259308e-05,
-      "loss": 0.8138,
+      "loss": 0.9076,
       "step": 4812
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8916899759312452e-05,
-      "loss": 0.9648,
+      "loss": 1.0476,
       "step": 4813
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8916366542322495e-05,
-      "loss": 0.9485,
+      "loss": 0.9726,
       "step": 4814
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8915833201630606e-05,
-      "loss": 0.8215,
+      "loss": 0.914,
       "step": 4815
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8915299737244174e-05,
-      "loss": 0.8868,
+      "loss": 0.967,
       "step": 4816
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8914766149170614e-05,
-      "loss": 0.8583,
+      "loss": 0.9072,
       "step": 4817
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8914232437417318e-05,
-      "loss": 0.9322,
+      "loss": 1.0237,
       "step": 4818
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8913698601991695e-05,
-      "loss": 0.8388,
+      "loss": 0.9353,
       "step": 4819
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8913164642901153e-05,
-      "loss": 0.8219,
+      "loss": 0.9916,
       "step": 4820
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8912630560153094e-05,
-      "loss": 0.8252,
+      "loss": 0.913,
       "step": 4821
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8912096353754935e-05,
-      "loss": 0.8751,
+      "loss": 0.9571,
       "step": 4822
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8911562023714083e-05,
-      "loss": 0.7757,
+      "loss": 0.9426,
       "step": 4823
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8911027570037954e-05,
-      "loss": 0.8657,
+      "loss": 0.9471,
       "step": 4824
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8910492992733957e-05,
-      "loss": 1.0287,
+      "loss": 1.0943,
       "step": 4825
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890995829180951e-05,
-      "loss": 0.9006,
+      "loss": 0.9527,
       "step": 4826
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890942346727204e-05,
-      "loss": 0.719,
+      "loss": 0.8809,
       "step": 4827
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8908888519128955e-05,
-      "loss": 0.9464,
+      "loss": 1.0547,
       "step": 4828
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8908353447387683e-05,
-      "loss": 0.9967,
+      "loss": 1.0001,
       "step": 4829
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8907818252055644e-05,
-      "loss": 0.8236,
+      "loss": 0.8317,
       "step": 4830
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890728293314027e-05,
-      "loss": 0.8261,
+      "loss": 0.9353,
       "step": 4831
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890674749064898e-05,
-      "loss": 0.7588,
+      "loss": 0.9153,
       "step": 4832
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8906211924589207e-05,
-      "loss": 0.8009,
+      "loss": 0.9122,
       "step": 4833
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8905676234968378e-05,
-      "loss": 0.8495,
+      "loss": 0.886,
       "step": 4834
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890514042179393e-05,
-      "loss": 0.8328,
+      "loss": 0.92,
       "step": 4835
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890460448507329e-05,
-      "loss": 0.9797,
+      "loss": 1.0392,
       "step": 4836
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.89040684248139e-05,
-      "loss": 0.9866,
+      "loss": 1.0321,
       "step": 4837
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890353224102319e-05,
-      "loss": 0.9235,
+      "loss": 0.911,
       "step": 4838
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8902995933708607e-05,
-      "loss": 0.8749,
+      "loss": 1.0064,
       "step": 4839
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.890245950287758e-05,
-      "loss": 0.9542,
+      "loss": 1.0504,
       "step": 4840
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8901922948537567e-05,
-      "loss": 0.8383,
+      "loss": 0.9287,
       "step": 4841
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8901386270696e-05,
-      "loss": 0.8666,
+      "loss": 1.0062,
       "step": 4842
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8900849469360324e-05,
-      "loss": 0.8227,
+      "loss": 0.9443,
       "step": 4843
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8900312544537995e-05,
-      "loss": 0.6634,
+      "loss": 0.8296,
       "step": 4844
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8899775496236458e-05,
-      "loss": 0.937,
+      "loss": 0.9873,
       "step": 4845
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.889923832446316e-05,
-      "loss": 0.9233,
+      "loss": 0.9912,
       "step": 4846
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8898701029225556e-05,
-      "loss": 0.8369,
+      "loss": 0.9919,
       "step": 4847
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8898163610531103e-05,
-      "loss": 0.8898,
+      "loss": 0.9385,
       "step": 4848
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8897626068387253e-05,
-      "loss": 0.9642,
+      "loss": 0.9991,
       "step": 4849
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8897088402801466e-05,
-      "loss": 0.9965,
+      "loss": 1.0266,
       "step": 4850
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8896550613781203e-05,
-      "loss": 0.9584,
+      "loss": 1.0264,
       "step": 4851
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8896012701333922e-05,
-      "loss": 0.754,
+      "loss": 0.9179,
       "step": 4852
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8895474665467082e-05,
-      "loss": 0.719,
+      "loss": 0.9429,
       "step": 4853
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8894936506188156e-05,
-      "loss": 0.8801,
+      "loss": 0.966,
       "step": 4854
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8894398223504606e-05,
-      "loss": 0.867,
+      "loss": 0.9904,
       "step": 4855
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88938598174239e-05,
-      "loss": 0.8161,
+      "loss": 0.9435,
       "step": 4856
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.889332128795351e-05,
-      "loss": 0.9521,
+      "loss": 1.0131,
       "step": 4857
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8892782635100903e-05,
-      "loss": 0.8792,
+      "loss": 1.0165,
       "step": 4858
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8892243858873552e-05,
-      "loss": 0.9646,
+      "loss": 1.072,
       "step": 4859
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8891704959278936e-05,
-      "loss": 0.9719,
+      "loss": 1.0417,
       "step": 4860
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8891165936324532e-05,
-      "loss": 0.8843,
+      "loss": 0.9646,
       "step": 4861
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.889062679001781e-05,
-      "loss": 0.7932,
+      "loss": 0.9797,
       "step": 4862
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.889008752036626e-05,
-      "loss": 0.7944,
+      "loss": 0.8872,
       "step": 4863
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8889548127377357e-05,
-      "loss": 0.8368,
+      "loss": 0.894,
       "step": 4864
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8889008611058587e-05,
-      "loss": 0.8787,
+      "loss": 0.96,
       "step": 4865
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8888468971417435e-05,
-      "loss": 0.9036,
+      "loss": 0.9843,
       "step": 4866
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8887929208461387e-05,
-      "loss": 0.8853,
+      "loss": 1.0045,
       "step": 4867
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8887389322197928e-05,
-      "loss": 0.7101,
+      "loss": 0.9135,
       "step": 4868
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8886849312634556e-05,
-      "loss": 0.883,
+      "loss": 0.9618,
       "step": 4869
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8886309179778756e-05,
-      "loss": 0.7402,
+      "loss": 0.8923,
       "step": 4870
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8885768923638025e-05,
-      "loss": 0.8189,
+      "loss": 0.9324,
       "step": 4871
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8885228544219858e-05,
-      "loss": 0.7467,
+      "loss": 0.8981,
       "step": 4872
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8884688041531753e-05,
-      "loss": 0.6907,
+      "loss": 0.822,
       "step": 4873
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8884147415581204e-05,
-      "loss": 0.7814,
+      "loss": 0.9178,
       "step": 4874
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8883606666375717e-05,
-      "loss": 0.7928,
+      "loss": 0.8875,
       "step": 4875
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.888306579392279e-05,
-      "loss": 0.8442,
+      "loss": 0.9807,
       "step": 4876
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8882524798229928e-05,
-      "loss": 0.7818,
+      "loss": 0.8873,
       "step": 4877
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8881983679304635e-05,
-      "loss": 0.7684,
+      "loss": 0.8906,
       "step": 4878
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8881442437154426e-05,
-      "loss": 0.8593,
+      "loss": 0.9803,
       "step": 4879
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88809010717868e-05,
-      "loss": 0.8128,
+      "loss": 0.922,
       "step": 4880
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8880359583209277e-05,
-      "loss": 0.7622,
+      "loss": 0.9271,
       "step": 4881
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.887981797142936e-05,
-      "loss": 0.765,
+      "loss": 0.8822,
       "step": 4882
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8879276236454568e-05,
-      "loss": 0.9677,
+      "loss": 1.017,
       "step": 4883
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8878734378292415e-05,
-      "loss": 0.972,
+      "loss": 1.0032,
       "step": 4884
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8878192396950423e-05,
-      "loss": 0.7381,
+      "loss": 0.9353,
       "step": 4885
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8877650292436106e-05,
-      "loss": 0.9436,
+      "loss": 1.0319,
       "step": 4886
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8877108064756988e-05,
-      "loss": 0.9458,
+      "loss": 0.9613,
       "step": 4887
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.887656571392059e-05,
-      "loss": 0.7857,
+      "loss": 0.8629,
       "step": 4888
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8876023239934435e-05,
-      "loss": 0.8939,
+      "loss": 0.9882,
       "step": 4889
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8875480642806054e-05,
-      "loss": 0.8706,
+      "loss": 0.9822,
       "step": 4890
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8874937922542972e-05,
-      "loss": 0.7302,
+      "loss": 0.8253,
       "step": 4891
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8874395079152718e-05,
-      "loss": 1.0444,
+      "loss": 1.0734,
       "step": 4892
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.887385211264282e-05,
-      "loss": 0.8773,
+      "loss": 0.9948,
       "step": 4893
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8873309023020816e-05,
-      "loss": 0.777,
+      "loss": 0.9436,
       "step": 4894
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8872765810294243e-05,
-      "loss": 0.8393,
+      "loss": 0.9709,
       "step": 4895
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8872222474470627e-05,
-      "loss": 0.8313,
+      "loss": 0.9156,
       "step": 4896
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8871679015557515e-05,
-      "loss": 0.7947,
+      "loss": 0.8917,
       "step": 4897
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8871135433562446e-05,
-      "loss": 0.7772,
+      "loss": 0.8891,
       "step": 4898
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8870591728492954e-05,
-      "loss": 0.7088,
+      "loss": 0.9441,
       "step": 4899
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.887004790035659e-05,
-      "loss": 0.8201,
+      "loss": 0.9546,
       "step": 4900
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88695039491609e-05,
-      "loss": 0.8335,
+      "loss": 0.9597,
       "step": 4901
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8868959874913423e-05,
-      "loss": 0.8538,
+      "loss": 0.991,
       "step": 4902
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.886841567762171e-05,
-      "loss": 0.8549,
+      "loss": 0.9523,
       "step": 4903
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8867871357293314e-05,
-      "loss": 0.9058,
+      "loss": 1.0178,
       "step": 4904
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8867326913935786e-05,
-      "loss": 0.7755,
+      "loss": 0.9362,
       "step": 4905
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8866782347556678e-05,
-      "loss": 0.8953,
+      "loss": 0.9823,
       "step": 4906
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8866237658163543e-05,
-      "loss": 0.8389,
+      "loss": 0.8847,
       "step": 4907
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8865692845763942e-05,
-      "loss": 0.923,
+      "loss": 0.9852,
       "step": 4908
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.886514791036543e-05,
-      "loss": 0.8933,
+      "loss": 1.0172,
       "step": 4909
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8864602851975572e-05,
-      "loss": 0.9029,
+      "loss": 0.9569,
       "step": 4910
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8864057670601924e-05,
-      "loss": 0.8922,
+      "loss": 0.9874,
       "step": 4911
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.886351236625205e-05,
-      "loss": 0.8421,
+      "loss": 0.8999,
       "step": 4912
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.886296693893352e-05,
-      "loss": 0.8947,
+      "loss": 0.978,
       "step": 4913
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8862421388653904e-05,
-      "loss": 0.816,
+      "loss": 0.9456,
       "step": 4914
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8861875715420756e-05,
-      "loss": 0.8911,
+      "loss": 1.0875,
       "step": 4915
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.886132991924166e-05,
-      "loss": 0.7758,
+      "loss": 0.8607,
       "step": 4916
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8860784000124182e-05,
-      "loss": 0.936,
+      "loss": 1.0556,
       "step": 4917
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88602379580759e-05,
-      "loss": 0.8901,
+      "loss": 0.9623,
       "step": 4918
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.885969179310439e-05,
-      "loss": 0.8575,
+      "loss": 0.9592,
       "step": 4919
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8859145505217223e-05,
-      "loss": 0.8024,
+      "loss": 0.9652,
       "step": 4920
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.885859909442198e-05,
-      "loss": 0.7858,
+      "loss": 0.939,
       "step": 4921
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8858052560726247e-05,
-      "loss": 0.8104,
+      "loss": 0.9854,
       "step": 4922
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8857505904137602e-05,
-      "loss": 0.8852,
+      "loss": 1.001,
       "step": 4923
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8856959124663626e-05,
-      "loss": 1.0507,
+      "loss": 1.075,
       "step": 4924
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8856412222311916e-05,
-      "loss": 0.85,
+      "loss": 0.9498,
       "step": 4925
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8855865197090046e-05,
-      "loss": 0.8053,
+      "loss": 0.9187,
       "step": 4926
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8855318049005614e-05,
-      "loss": 0.905,
+      "loss": 1.0663,
       "step": 4927
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8854770778066206e-05,
-      "loss": 0.8578,
+      "loss": 0.9356,
       "step": 4928
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8854223384279417e-05,
-      "loss": 0.7126,
+      "loss": 0.8738,
       "step": 4929
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.885367586765284e-05,
-      "loss": 0.7359,
+      "loss": 0.8626,
       "step": 4930
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8853128228194075e-05,
-      "loss": 0.8588,
+      "loss": 0.9772,
       "step": 4931
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8852580465910715e-05,
-      "loss": 0.6845,
+      "loss": 0.8666,
       "step": 4932
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8852032580810362e-05,
-      "loss": 0.858,
+      "loss": 0.9098,
       "step": 4933
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8851484572900617e-05,
-      "loss": 0.7495,
+      "loss": 0.8848,
       "step": 4934
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.885093644218908e-05,
-      "loss": 0.6681,
+      "loss": 0.8502,
       "step": 4935
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.885038818868336e-05,
-      "loss": 0.8894,
+      "loss": 0.921,
       "step": 4936
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884983981239106e-05,
-      "loss": 0.9475,
+      "loss": 0.9664,
       "step": 4937
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884929131331979e-05,
-      "loss": 1.01,
+      "loss": 1.0797,
       "step": 4938
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8848742691477152e-05,
-      "loss": 0.8092,
+      "loss": 0.9423,
       "step": 4939
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8848193946870774e-05,
-      "loss": 0.7863,
+      "loss": 0.8892,
       "step": 4940
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884764507950825e-05,
-      "loss": 0.9701,
+      "loss": 1.0474,
       "step": 4941
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884709608939721e-05,
-      "loss": 0.7138,
+      "loss": 0.7911,
       "step": 4942
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884654697654526e-05,
-      "loss": 0.8888,
+      "loss": 0.9801,
       "step": 4943
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884599774096003e-05,
-      "loss": 0.7277,
+      "loss": 0.8807,
       "step": 4944
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884544838264912e-05,
-      "loss": 0.8932,
+      "loss": 1.0602,
       "step": 4945
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8844898901620173e-05,
-      "loss": 0.8027,
+      "loss": 0.8565,
       "step": 4946
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88443492978808e-05,
-      "loss": 0.8352,
+      "loss": 0.9628,
       "step": 4947
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.884379957143863e-05,
-      "loss": 0.9816,
+      "loss": 1.0285,
       "step": 4948
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8843249722301285e-05,
-      "loss": 0.7558,
+      "loss": 0.8392,
       "step": 4949
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88426997504764e-05,
-      "loss": 0.8486,
+      "loss": 1.0173,
       "step": 4950
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8842149655971602e-05,
-      "loss": 0.7903,
+      "loss": 0.9005,
       "step": 4951
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8841599438794525e-05,
-      "loss": 0.7051,
+      "loss": 0.931,
       "step": 4952
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8841049098952797e-05,
-      "loss": 0.8689,
+      "loss": 0.9702,
       "step": 4953
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8840498636454056e-05,
-      "loss": 0.8463,
+      "loss": 1.0052,
       "step": 4954
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.883994805130594e-05,
-      "loss": 0.8342,
+      "loss": 0.9205,
       "step": 4955
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8839397343516092e-05,
-      "loss": 0.8936,
+      "loss": 0.905,
       "step": 4956
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.883884651309214e-05,
-      "loss": 0.7733,
+      "loss": 0.8954,
       "step": 4957
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.883829556004174e-05,
-      "loss": 0.8337,
+      "loss": 0.938,
       "step": 4958
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8837744484372526e-05,
-      "loss": 0.8977,
+      "loss": 1.0433,
       "step": 4959
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8837193286092145e-05,
-      "loss": 0.6831,
+      "loss": 0.8864,
       "step": 4960
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8836641965208247e-05,
-      "loss": 0.9126,
+      "loss": 1.0188,
       "step": 4961
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8836090521728477e-05,
-      "loss": 0.8286,
+      "loss": 0.9875,
       "step": 4962
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.883553895566049e-05,
-      "loss": 0.9478,
+      "loss": 1.0105,
       "step": 4963
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8834987267011937e-05,
-      "loss": 0.7397,
+      "loss": 0.8252,
       "step": 4964
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8834435455790468e-05,
-      "loss": 0.8225,
+      "loss": 0.9218,
       "step": 4965
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8833883522003745e-05,
-      "loss": 0.7149,
+      "loss": 0.8731,
       "step": 4966
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8833331465659422e-05,
-      "loss": 0.8692,
+      "loss": 0.9243,
       "step": 4967
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8832779286765155e-05,
-      "loss": 0.6888,
+      "loss": 0.8061,
       "step": 4968
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8832226985328606e-05,
-      "loss": 0.8565,
+      "loss": 0.9452,
       "step": 4969
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8831674561357442e-05,
-      "loss": 0.8137,
+      "loss": 0.9547,
       "step": 4970
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8831122014859323e-05,
-      "loss": 0.9201,
+      "loss": 0.9972,
       "step": 4971
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8830569345841917e-05,
-      "loss": 0.9543,
+      "loss": 0.945,
       "step": 4972
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8830016554312887e-05,
-      "loss": 0.8444,
+      "loss": 0.9785,
       "step": 4973
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8829463640279908e-05,
-      "loss": 0.7611,
+      "loss": 0.9423,
       "step": 4974
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8828910603750647e-05,
-      "loss": 0.9758,
+      "loss": 0.9948,
       "step": 4975
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882835744473278e-05,
-      "loss": 0.7731,
+      "loss": 0.9493,
       "step": 4976
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882780416323398e-05,
-      "loss": 0.7152,
+      "loss": 0.9041,
       "step": 4977
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8827250759261918e-05,
-      "loss": 0.8891,
+      "loss": 0.9481,
       "step": 4978
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8826697232824277e-05,
-      "loss": 0.8564,
+      "loss": 0.9434,
       "step": 4979
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882614358392874e-05,
-      "loss": 0.898,
+      "loss": 1.0083,
       "step": 4980
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8825589812582977e-05,
-      "loss": 0.8057,
+      "loss": 0.9623,
       "step": 4981
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8825035918794682e-05,
-      "loss": 0.6281,
+      "loss": 0.8197,
       "step": 4982
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8824481902571533e-05,
-      "loss": 0.8041,
+      "loss": 0.9312,
       "step": 4983
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8823927763921216e-05,
-      "loss": 0.7662,
+      "loss": 0.8887,
       "step": 4984
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8823373502851422e-05,
-      "loss": 0.7431,
+      "loss": 0.8979,
       "step": 4985
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8822819119369843e-05,
-      "loss": 0.8438,
+      "loss": 0.8843,
       "step": 4986
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8822264613484164e-05,
-      "loss": 0.8751,
+      "loss": 0.8867,
       "step": 4987
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882170998520208e-05,
-      "loss": 0.7831,
+      "loss": 0.9416,
       "step": 4988
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8821155234531286e-05,
-      "loss": 0.8828,
+      "loss": 0.9188,
       "step": 4989
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882060036147948e-05,
-      "loss": 0.903,
+      "loss": 1.0587,
       "step": 4990
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8820045366054357e-05,
-      "loss": 0.8393,
+      "loss": 0.9104,
       "step": 4991
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.881949024826362e-05,
-      "loss": 0.9298,
+      "loss": 0.9631,
       "step": 4992
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8818935008114968e-05,
-      "loss": 0.9003,
+      "loss": 1.0148,
       "step": 4993
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8818379645616107e-05,
-      "loss": 0.9016,
+      "loss": 0.9673,
       "step": 4994
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8817824160774738e-05,
-      "loss": 0.9149,
+      "loss": 1.0128,
       "step": 4995
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8817268553598572e-05,
-      "loss": 0.73,
+      "loss": 0.9302,
       "step": 4996
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8816712824095314e-05,
-      "loss": 0.782,
+      "loss": 0.9578,
       "step": 4997
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.881615697227267e-05,
-      "loss": 0.9003,
+      "loss": 1.0211,
       "step": 4998
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8815600998138363e-05,
-      "loss": 0.8524,
+      "loss": 0.9596,
       "step": 4999
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8815044901700097e-05,
-      "loss": 0.8929,
+      "loss": 1.0212,
       "step": 5000
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.881448868296559e-05,
-      "loss": 0.6615,
+      "loss": 0.8855,
       "step": 5001
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8813932341942563e-05,
-      "loss": 0.8591,
+      "loss": 0.9285,
       "step": 5002
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8813375878638725e-05,
-      "loss": 0.9016,
+      "loss": 0.9664,
       "step": 5003
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88128192930618e-05,
-      "loss": 0.7717,
+      "loss": 0.9069,
       "step": 5004
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8812262585219514e-05,
-      "loss": 0.8988,
+      "loss": 0.9826,
       "step": 5005
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8811705755119588e-05,
-      "loss": 0.8433,
+      "loss": 0.9582,
       "step": 5006
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8811148802769748e-05,
-      "loss": 0.8313,
+      "loss": 0.9258,
       "step": 5007
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.881059172817772e-05,
-      "loss": 0.7902,
+      "loss": 0.8694,
       "step": 5008
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8810034531351226e-05,
-      "loss": 0.9026,
+      "loss": 0.9649,
       "step": 5009
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8809477212298007e-05,
-      "loss": 0.8929,
+      "loss": 0.9429,
       "step": 5010
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880891977102579e-05,
-      "loss": 0.9004,
+      "loss": 0.9074,
       "step": 5011
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880836220754231e-05,
-      "loss": 0.9311,
+      "loss": 1.1125,
       "step": 5012
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8807804521855302e-05,
-      "loss": 0.7678,
+      "loss": 0.9791,
       "step": 5013
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8807246713972502e-05,
-      "loss": 0.8461,
+      "loss": 0.9669,
       "step": 5014
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880668878390165e-05,
-      "loss": 0.8362,
+      "loss": 0.9358,
       "step": 5015
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8806130731650487e-05,
-      "loss": 0.9766,
+      "loss": 1.055,
       "step": 5016
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8805572557226755e-05,
-      "loss": 0.8962,
+      "loss": 0.9541,
       "step": 5017
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8805014260638196e-05,
-      "loss": 0.8027,
+      "loss": 0.8597,
       "step": 5018
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8804455841892558e-05,
-      "loss": 0.8194,
+      "loss": 0.9538,
       "step": 5019
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8803897300997586e-05,
-      "loss": 0.9299,
+      "loss": 1.0954,
       "step": 5020
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8803338637961032e-05,
-      "loss": 0.7761,
+      "loss": 0.8906,
       "step": 5021
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880277985279064e-05,
-      "loss": 0.8411,
+      "loss": 1.0096,
       "step": 5022
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880222094549417e-05,
-      "loss": 0.8616,
+      "loss": 0.9756,
       "step": 5023
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8801661916079373e-05,
-      "loss": 0.7962,
+      "loss": 0.9748,
       "step": 5024
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8801102764554006e-05,
-      "loss": 0.8539,
+      "loss": 0.9995,
       "step": 5025
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8800543490925824e-05,
-      "loss": 0.7084,
+      "loss": 0.8726,
       "step": 5026
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8799984095202586e-05,
-      "loss": 0.8116,
+      "loss": 0.9116,
       "step": 5027
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8799424577392057e-05,
-      "loss": 1.0159,
+      "loss": 1.0067,
       "step": 5028
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8798864937501994e-05,
-      "loss": 0.8475,
+      "loss": 0.904,
       "step": 5029
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8798305175540165e-05,
-      "loss": 0.9368,
+      "loss": 0.9533,
       "step": 5030
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8797745291514335e-05,
-      "loss": 0.895,
+      "loss": 0.9987,
       "step": 5031
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.879718528543227e-05,
-      "loss": 0.92,
+      "loss": 0.9686,
       "step": 5032
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8796625157301744e-05,
-      "loss": 0.8105,
+      "loss": 0.8914,
       "step": 5033
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8796064907130524e-05,
-      "loss": 0.8494,
+      "loss": 0.9203,
       "step": 5034
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8795504534926383e-05,
-      "loss": 0.8431,
+      "loss": 0.9156,
       "step": 5035
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8794944040697096e-05,
-      "loss": 0.8219,
+      "loss": 0.901,
       "step": 5036
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8794383424450437e-05,
-      "loss": 0.8314,
+      "loss": 0.9544,
       "step": 5037
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8793822686194184e-05,
-      "loss": 0.7822,
+      "loss": 0.8895,
       "step": 5038
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8793261825936123e-05,
-      "loss": 0.8459,
+      "loss": 0.9539,
       "step": 5039
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8792700843684027e-05,
-      "loss": 0.6987,
+      "loss": 0.8758,
       "step": 5040
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8792139739445684e-05,
-      "loss": 1.0302,
+      "loss": 1.082,
       "step": 5041
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8791578513228875e-05,
-      "loss": 0.8645,
+      "loss": 0.9803,
       "step": 5042
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8791017165041388e-05,
-      "loss": 0.8076,
+      "loss": 0.9274,
       "step": 5043
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.879045569489101e-05,
-      "loss": 0.9125,
+      "loss": 0.9551,
       "step": 5044
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8789894102785532e-05,
-      "loss": 0.7603,
+      "loss": 0.938,
       "step": 5045
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8789332388732744e-05,
-      "loss": 0.8039,
+      "loss": 0.9352,
       "step": 5046
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8788770552740443e-05,
-      "loss": 0.7759,
+      "loss": 0.8944,
       "step": 5047
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8788208594816418e-05,
-      "loss": 0.8631,
+      "loss": 0.8884,
       "step": 5048
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8787646514968465e-05,
-      "loss": 0.7813,
+      "loss": 0.8194,
       "step": 5049
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8787084313204386e-05,
-      "loss": 0.8713,
+      "loss": 0.9688,
       "step": 5050
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878652198953198e-05,
-      "loss": 0.7274,
+      "loss": 0.9497,
       "step": 5051
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878595954395905e-05,
-      "loss": 0.8368,
+      "loss": 0.9582,
       "step": 5052
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8785396976493392e-05,
-      "loss": 0.7665,
+      "loss": 0.8871,
       "step": 5053
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878483428714282e-05,
-      "loss": 0.8493,
+      "loss": 0.9923,
       "step": 5054
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8784271475915138e-05,
-      "loss": 0.8131,
+      "loss": 0.961,
       "step": 5055
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8783708542818147e-05,
-      "loss": 0.9057,
+      "loss": 0.9986,
       "step": 5056
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8783145487859664e-05,
-      "loss": 0.7683,
+      "loss": 0.8974,
       "step": 5057
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.87825823110475e-05,
-      "loss": 0.7338,
+      "loss": 0.8627,
       "step": 5058
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878201901238947e-05,
-      "loss": 0.8579,
+      "loss": 0.9701,
       "step": 5059
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878145559189338e-05,
-      "loss": 0.8836,
+      "loss": 1.0358,
       "step": 5060
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8780892049567057e-05,
-      "loss": 0.931,
+      "loss": 0.9842,
       "step": 5061
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8780328385418314e-05,
-      "loss": 0.8047,
+      "loss": 0.9101,
       "step": 5062
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8779764599454972e-05,
-      "loss": 0.907,
+      "loss": 1.0445,
       "step": 5063
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8779200691684856e-05,
-      "loss": 0.7599,
+      "loss": 0.849,
       "step": 5064
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8778636662115787e-05,
-      "loss": 0.9894,
+      "loss": 1.0172,
       "step": 5065
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8778072510755584e-05,
-      "loss": 0.7846,
+      "loss": 0.9439,
       "step": 5066
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877750823761208e-05,
-      "loss": 0.8235,
+      "loss": 0.9214,
       "step": 5067
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8776943842693107e-05,
-      "loss": 0.7621,
+      "loss": 0.9091,
       "step": 5068
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877637932600649e-05,
-      "loss": 0.806,
+      "loss": 0.961,
       "step": 5069
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877581468756006e-05,
-      "loss": 0.8746,
+      "loss": 0.9455,
       "step": 5070
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8775249927361652e-05,
-      "loss": 0.8102,
+      "loss": 0.9161,
       "step": 5071
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8774685045419107e-05,
-      "loss": 0.8831,
+      "loss": 0.992,
       "step": 5072
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8774120041740254e-05,
-      "loss": 0.9523,
+      "loss": 1.0809,
       "step": 5073
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877355491633293e-05,
-      "loss": 0.8536,
+      "loss": 0.9827,
       "step": 5074
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8772989669204985e-05,
-      "loss": 0.8542,
+      "loss": 0.9577,
       "step": 5075
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8772424300364252e-05,
-      "loss": 0.8604,
+      "loss": 0.9366,
       "step": 5076
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8771858809818583e-05,
-      "loss": 0.8638,
+      "loss": 0.9154,
       "step": 5077
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8771293197575814e-05,
-      "loss": 0.8822,
+      "loss": 0.9922,
       "step": 5078
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8770727463643796e-05,
-      "loss": 0.7397,
+      "loss": 0.8775,
       "step": 5079
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8770161608030382e-05,
-      "loss": 0.9606,
+      "loss": 1.0172,
       "step": 5080
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8769595630743417e-05,
-      "loss": 0.7854,
+      "loss": 0.918,
       "step": 5081
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8769029531790756e-05,
-      "loss": 0.8062,
+      "loss": 1.025,
       "step": 5082
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.876846331118025e-05,
-      "loss": 0.7237,
+      "loss": 0.9148,
       "step": 5083
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8767896968919758e-05,
-      "loss": 0.736,
+      "loss": 0.8646,
       "step": 5084
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8767330505017135e-05,
-      "loss": 0.7027,
+      "loss": 0.8447,
       "step": 5085
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.876676391948024e-05,
-      "loss": 0.8687,
+      "loss": 0.9223,
       "step": 5086
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8766197212316938e-05,
-      "loss": 0.7449,
+      "loss": 0.9008,
       "step": 5087
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.876563038353508e-05,
-      "loss": 0.8158,
+      "loss": 0.974,
       "step": 5088
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8765063433142545e-05,
-      "loss": 0.7937,
+      "loss": 0.969,
       "step": 5089
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8764496361147183e-05,
-      "loss": 0.8066,
+      "loss": 0.8872,
       "step": 5090
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8763929167556877e-05,
-      "loss": 0.7861,
+      "loss": 0.956,
       "step": 5091
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8763361852379484e-05,
-      "loss": 1.0162,
+      "loss": 1.0468,
       "step": 5092
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.876279441562288e-05,
-      "loss": 0.9363,
+      "loss": 0.9845,
       "step": 5093
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8762226857294937e-05,
-      "loss": 0.8274,
+      "loss": 0.9396,
       "step": 5094
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8761659177403527e-05,
-      "loss": 0.8665,
+      "loss": 0.9683,
       "step": 5095
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8761091375956523e-05,
-      "loss": 0.7485,
+      "loss": 0.8455,
       "step": 5096
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8760523452961813e-05,
-      "loss": 0.9259,
+      "loss": 1.0165,
       "step": 5097
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875995540842727e-05,
-      "loss": 0.9706,
+      "loss": 1.0566,
       "step": 5098
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8759387242360768e-05,
-      "loss": 0.85,
+      "loss": 1.0193,
       "step": 5099
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87588189547702e-05,
-      "loss": 0.895,
+      "loss": 1.049,
       "step": 5100
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875825054566345e-05,
-      "loss": 0.9907,
+      "loss": 1.0692,
       "step": 5101
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875768201504839e-05,
-      "loss": 0.8168,
+      "loss": 0.843,
       "step": 5102
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8757113362932926e-05,
-      "loss": 0.9333,
+      "loss": 0.9694,
       "step": 5103
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8756544589324938e-05,
-      "loss": 0.7677,
+      "loss": 0.8719,
       "step": 5104
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8755975694232313e-05,
-      "loss": 0.928,
+      "loss": 0.9921,
       "step": 5105
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8755406677662953e-05,
-      "loss": 0.8405,
+      "loss": 0.9336,
       "step": 5106
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8754837539624745e-05,
-      "loss": 0.7574,
+      "loss": 0.9093,
       "step": 5107
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875426828012559e-05,
-      "loss": 0.9266,
+      "loss": 0.9959,
       "step": 5108
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8753698899173383e-05,
-      "loss": 0.8611,
+      "loss": 0.9316,
       "step": 5109
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875312939677602e-05,
-      "loss": 0.6931,
+      "loss": 0.8545,
       "step": 5110
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875255977294141e-05,
-      "loss": 0.8647,
+      "loss": 0.9263,
       "step": 5111
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8751990027677447e-05,
-      "loss": 0.9091,
+      "loss": 1.0037,
       "step": 5112
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8751420160992045e-05,
-      "loss": 0.7999,
+      "loss": 0.9265,
       "step": 5113
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87508501728931e-05,
-      "loss": 0.8278,
+      "loss": 0.8748,
       "step": 5114
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8750280063388528e-05,
-      "loss": 0.76,
+      "loss": 0.905,
       "step": 5115
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8749709832486233e-05,
-      "loss": 0.9963,
+      "loss": 0.9961,
       "step": 5116
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.874913948019413e-05,
-      "loss": 0.8828,
+      "loss": 0.9596,
       "step": 5117
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8748569006520128e-05,
-      "loss": 0.8322,
+      "loss": 1.0092,
       "step": 5118
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8747998411472148e-05,
-      "loss": 0.889,
+      "loss": 0.9673,
       "step": 5119
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87474276950581e-05,
-      "loss": 0.7654,
+      "loss": 0.9883,
       "step": 5120
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87468568572859e-05,
-      "loss": 0.6894,
+      "loss": 0.8612,
       "step": 5121
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8746285898163474e-05,
-      "loss": 0.7169,
+      "loss": 0.9141,
       "step": 5122
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8745714817698744e-05,
-      "loss": 0.7674,
+      "loss": 0.8888,
       "step": 5123
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8745143615899623e-05,
-      "loss": 0.8728,
+      "loss": 1.0453,
       "step": 5124
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8744572292774048e-05,
-      "loss": 0.7923,
+      "loss": 0.8677,
       "step": 5125
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8744000848329936e-05,
-      "loss": 0.8275,
+      "loss": 0.9774,
       "step": 5126
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8743429282575216e-05,
-      "loss": 0.8691,
+      "loss": 0.954,
       "step": 5127
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8742857595517825e-05,
-      "loss": 0.8705,
+      "loss": 0.96,
       "step": 5128
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8742285787165686e-05,
-      "loss": 0.9943,
+      "loss": 1.0801,
       "step": 5129
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8741713857526736e-05,
-      "loss": 0.8912,
+      "loss": 0.9168,
       "step": 5130
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8741141806608908e-05,
-      "loss": 0.8169,
+      "loss": 0.9318,
       "step": 5131
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8740569634420142e-05,
-      "loss": 0.6987,
+      "loss": 0.8749,
       "step": 5132
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8739997340968375e-05,
-      "loss": 0.8218,
+      "loss": 0.9504,
       "step": 5133
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8739424926261544e-05,
-      "loss": 0.7649,
+      "loss": 0.9114,
       "step": 5134
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8738852390307588e-05,
-      "loss": 0.8532,
+      "loss": 0.954,
       "step": 5135
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873827973311446e-05,
-      "loss": 0.7797,
+      "loss": 0.9274,
       "step": 5136
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8737706954690094e-05,
-      "loss": 0.8364,
+      "loss": 0.9742,
       "step": 5137
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873713405504244e-05,
-      "loss": 0.9514,
+      "loss": 1.0228,
       "step": 5138
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873656103417945e-05,
-      "loss": 0.9151,
+      "loss": 0.9328,
       "step": 5139
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8735987892109072e-05,
-      "loss": 0.6314,
+      "loss": 0.825,
       "step": 5140
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8735414628839258e-05,
-      "loss": 0.809,
+      "loss": 0.9742,
       "step": 5141
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8734841244377958e-05,
-      "loss": 0.7104,
+      "loss": 0.8605,
       "step": 5142
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8734267738733128e-05,
-      "loss": 0.7737,
+      "loss": 0.8709,
       "step": 5143
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8733694111912727e-05,
-      "loss": 0.7916,
+      "loss": 0.9822,
       "step": 5144
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8733120363924707e-05,
-      "loss": 0.9639,
+      "loss": 1.0391,
       "step": 5145
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873254649477704e-05,
-      "loss": 0.8042,
+      "loss": 0.9722,
       "step": 5146
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8731972504477675e-05,
-      "loss": 0.8659,
+      "loss": 0.9453,
       "step": 5147
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873139839303458e-05,
-      "loss": 0.7668,
+      "loss": 0.8443,
       "step": 5148
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8730824160455724e-05,
-      "loss": 0.9325,
+      "loss": 0.9466,
       "step": 5149
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873024980674907e-05,
-      "loss": 1.0605,
+      "loss": 1.0731,
       "step": 5150
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8729675331922586e-05,
-      "loss": 0.8107,
+      "loss": 0.8759,
       "step": 5151
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872910073598424e-05,
-      "loss": 0.8684,
+      "loss": 0.9536,
       "step": 5152
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872852601894201e-05,
-      "loss": 0.8698,
+      "loss": 0.9045,
       "step": 5153
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8727951180803865e-05,
-      "loss": 0.8679,
+      "loss": 0.9433,
       "step": 5154
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872737622157778e-05,
-      "loss": 0.8284,
+      "loss": 0.883,
       "step": 5155
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872680114127173e-05,
-      "loss": 0.8868,
+      "loss": 0.8969,
       "step": 5156
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87262259398937e-05,
-      "loss": 0.8642,
+      "loss": 0.8957,
       "step": 5157
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8725650617451663e-05,
-      "loss": 0.8552,
+      "loss": 0.9861,
       "step": 5158
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8725075173953604e-05,
-      "loss": 0.8503,
+      "loss": 0.9545,
       "step": 5159
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8724499609407507e-05,
-      "loss": 0.8075,
+      "loss": 0.9359,
       "step": 5160
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8723923923821358e-05,
-      "loss": 0.7361,
+      "loss": 0.852,
       "step": 5161
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8723348117203138e-05,
-      "loss": 0.7536,
+      "loss": 0.8504,
       "step": 5162
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8722772189560844e-05,
-      "loss": 0.697,
+      "loss": 0.8451,
       "step": 5163
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872219614090246e-05,
-      "loss": 0.7813,
+      "loss": 0.9078,
       "step": 5164
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872161997123598e-05,
-      "loss": 0.942,
+      "loss": 0.99,
       "step": 5165
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8721043680569394e-05,
-      "loss": 0.6698,
+      "loss": 0.7777,
       "step": 5166
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872046726891071e-05,
-      "loss": 0.88,
+      "loss": 0.9537,
       "step": 5167
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8719890736267905e-05,
-      "loss": 0.8945,
+      "loss": 0.9571,
       "step": 5168
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8719314082648993e-05,
-      "loss": 0.8747,
+      "loss": 0.9604,
       "step": 5169
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871873730806197e-05,
-      "loss": 0.8085,
+      "loss": 0.899,
       "step": 5170
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871816041251483e-05,
-      "loss": 0.7972,
+      "loss": 0.8881,
       "step": 5171
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871758339601559e-05,
-      "loss": 0.9139,
+      "loss": 1.0003,
       "step": 5172
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871700625857225e-05,
-      "loss": 0.8838,
+      "loss": 0.9699,
       "step": 5173
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8716429000192814e-05,
-      "loss": 0.7869,
+      "loss": 0.9227,
       "step": 5174
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8715851620885292e-05,
-      "loss": 0.8759,
+      "loss": 0.9648,
       "step": 5175
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8715274120657694e-05,
-      "loss": 0.772,
+      "loss": 0.9152,
       "step": 5176
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8714696499518034e-05,
-      "loss": 0.7237,
+      "loss": 0.8837,
       "step": 5177
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8714118757474328e-05,
-      "loss": 0.8318,
+      "loss": 0.9357,
       "step": 5178
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8713540894534584e-05,
-      "loss": 0.7793,
+      "loss": 0.9078,
       "step": 5179
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8712962910706824e-05,
-      "loss": 0.8365,
+      "loss": 1.0056,
       "step": 5180
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8712384805999063e-05,
-      "loss": 0.7492,
+      "loss": 0.8806,
       "step": 5181
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8711806580419328e-05,
-      "loss": 0.7961,
+      "loss": 0.9395,
       "step": 5182
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8711228233975634e-05,
-      "loss": 0.8407,
+      "loss": 0.983,
       "step": 5183
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8710649766676013e-05,
-      "loss": 0.7572,
+      "loss": 0.9205,
       "step": 5184
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871007117852848e-05,
-      "loss": 0.8538,
+      "loss": 0.9626,
       "step": 5185
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8709492469541072e-05,
-      "loss": 0.7629,
+      "loss": 0.8596,
       "step": 5186
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8708913639721808e-05,
-      "loss": 0.7739,
+      "loss": 0.9253,
       "step": 5187
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8708334689078726e-05,
-      "loss": 0.8852,
+      "loss": 0.8969,
       "step": 5188
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8707755617619855e-05,
-      "loss": 0.7476,
+      "loss": 0.8694,
       "step": 5189
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870717642535323e-05,
-      "loss": 0.8518,
+      "loss": 0.9584,
       "step": 5190
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870659711228689e-05,
-      "loss": 0.8781,
+      "loss": 1.0492,
       "step": 5191
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870601767842886e-05,
-      "loss": 0.8411,
+      "loss": 0.9614,
       "step": 5192
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8705438123787193e-05,
-      "loss": 0.9355,
+      "loss": 0.985,
       "step": 5193
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870485844836992e-05,
-      "loss": 0.6325,
+      "loss": 0.7848,
       "step": 5194
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870427865218509e-05,
-      "loss": 0.8302,
+      "loss": 0.9503,
       "step": 5195
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8703698735240742e-05,
-      "loss": 0.7532,
+      "loss": 0.8246,
       "step": 5196
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8703118697544923e-05,
-      "loss": 0.9563,
+      "loss": 1.039,
       "step": 5197
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870253853910568e-05,
-      "loss": 0.7612,
+      "loss": 0.8863,
       "step": 5198
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870195825993106e-05,
-      "loss": 0.888,
+      "loss": 0.9783,
       "step": 5199
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870137786002912e-05,
-      "loss": 0.7668,
+      "loss": 0.8529,
       "step": 5200
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8700797339407905e-05,
-      "loss": 0.833,
+      "loss": 0.9416,
       "step": 5201
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870021669807547e-05,
-      "loss": 0.7807,
+      "loss": 0.8732,
       "step": 5202
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8699635936039876e-05,
-      "loss": 0.8582,
+      "loss": 0.9083,
       "step": 5203
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8699055053309173e-05,
-      "loss": 0.8629,
+      "loss": 0.9667,
       "step": 5204
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.869847404989143e-05,
-      "loss": 0.8011,
+      "loss": 0.8674,
       "step": 5205
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8697892925794693e-05,
-      "loss": 0.8315,
+      "loss": 0.9043,
       "step": 5206
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8697311681027037e-05,
-      "loss": 0.8671,
+      "loss": 0.9581,
       "step": 5207
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.869673031559652e-05,
-      "loss": 0.7777,
+      "loss": 0.922,
       "step": 5208
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.869614882951121e-05,
-      "loss": 0.8969,
+      "loss": 0.9344,
       "step": 5209
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8695567222779173e-05,
-      "loss": 0.8639,
+      "loss": 0.9859,
       "step": 5210
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8694985495408475e-05,
-      "loss": 0.8172,
+      "loss": 0.9561,
       "step": 5211
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8694403647407194e-05,
-      "loss": 0.8248,
+      "loss": 0.9137,
       "step": 5212
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8693821678783396e-05,
-      "loss": 0.8445,
+      "loss": 0.8843,
       "step": 5213
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8693239589545156e-05,
-      "loss": 0.8046,
+      "loss": 0.9523,
       "step": 5214
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8692657379700554e-05,
-      "loss": 0.827,
+      "loss": 0.985,
       "step": 5215
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.869207504925766e-05,
-      "loss": 0.7835,
+      "loss": 0.9489,
       "step": 5216
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8691492598224565e-05,
-      "loss": 0.904,
+      "loss": 0.8794,
       "step": 5217
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8690910026609335e-05,
-      "loss": 0.8516,
+      "loss": 0.9754,
       "step": 5218
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8690327334420056e-05,
-      "loss": 0.8178,
+      "loss": 0.916,
       "step": 5219
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8689744521664823e-05,
-      "loss": 0.9487,
+      "loss": 0.9685,
       "step": 5220
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8689161588351708e-05,
-      "loss": 0.7569,
+      "loss": 0.9257,
       "step": 5221
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8688578534488807e-05,
-      "loss": 0.7442,
+      "loss": 0.8739,
       "step": 5222
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.868799536008421e-05,
-      "loss": 0.7402,
+      "loss": 0.9995,
       "step": 5223
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8687412065145998e-05,
-      "loss": 0.8155,
+      "loss": 0.9283,
       "step": 5224
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8686828649682272e-05,
-      "loss": 0.801,
+      "loss": 0.9294,
       "step": 5225
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8686245113701123e-05,
-      "loss": 0.7657,
+      "loss": 0.8548,
       "step": 5226
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8685661457210648e-05,
-      "loss": 0.9096,
+      "loss": 0.9058,
       "step": 5227
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8685077680218945e-05,
-      "loss": 0.8626,
+      "loss": 0.9418,
       "step": 5228
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.868449378273411e-05,
-      "loss": 0.916,
+      "loss": 1.0143,
       "step": 5229
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8683909764764246e-05,
-      "loss": 0.9028,
+      "loss": 1.007,
       "step": 5230
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8683325626317452e-05,
-      "loss": 0.8023,
+      "loss": 0.9477,
       "step": 5231
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8682741367401837e-05,
-      "loss": 0.8761,
+      "loss": 0.8857,
       "step": 5232
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8682156988025503e-05,
-      "loss": 0.7663,
+      "loss": 0.9282,
       "step": 5233
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.868157248819656e-05,
-      "loss": 0.7562,
+      "loss": 0.7931,
       "step": 5234
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.868098786792312e-05,
-      "loss": 0.7192,
+      "loss": 0.8663,
       "step": 5235
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8680403127213286e-05,
-      "loss": 0.8079,
+      "loss": 0.889,
       "step": 5236
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8679818266075176e-05,
-      "loss": 0.7491,
+      "loss": 0.8434,
       "step": 5237
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8679233284516903e-05,
-      "loss": 0.7871,
+      "loss": 0.8845,
       "step": 5238
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8678648182546583e-05,
-      "loss": 0.8538,
+      "loss": 0.9477,
       "step": 5239
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.867806296017233e-05,
-      "loss": 0.789,
+      "loss": 0.8783,
       "step": 5240
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.867747761740227e-05,
-      "loss": 0.8927,
+      "loss": 1.0274,
       "step": 5241
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8676892154244516e-05,
-      "loss": 0.6416,
+      "loss": 0.7636,
       "step": 5242
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8676306570707194e-05,
-      "loss": 0.7058,
+      "loss": 0.8944,
       "step": 5243
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.867572086679843e-05,
-      "loss": 0.9185,
+      "loss": 1.0196,
       "step": 5244
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8675135042526352e-05,
-      "loss": 0.938,
+      "loss": 0.9379,
       "step": 5245
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8674549097899078e-05,
-      "loss": 0.7869,
+      "loss": 0.935,
       "step": 5246
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8673963032924743e-05,
-      "loss": 0.8092,
+      "loss": 0.9694,
       "step": 5247
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8673376847611482e-05,
-      "loss": 0.7681,
+      "loss": 0.8789,
       "step": 5248
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.867279054196742e-05,
-      "loss": 0.8035,
+      "loss": 0.9316,
       "step": 5249
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8672204116000693e-05,
-      "loss": 0.9401,
+      "loss": 0.9625,
       "step": 5250
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.867161756971944e-05,
-      "loss": 0.8131,
+      "loss": 0.8603,
       "step": 5251
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.86710309031318e-05,
-      "loss": 0.7881,
+      "loss": 0.9097,
       "step": 5252
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8670444116245904e-05,
-      "loss": 0.7745,
+      "loss": 0.9157,
       "step": 5253
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8669857209069898e-05,
-      "loss": 0.7985,
+      "loss": 0.9176,
       "step": 5254
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8669270181611927e-05,
-      "loss": 0.8105,
+      "loss": 0.9176,
       "step": 5255
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8668683033880132e-05,
-      "loss": 0.7742,
+      "loss": 0.7702,
       "step": 5256
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8668095765882656e-05,
-      "loss": 0.8252,
+      "loss": 0.9267,
       "step": 5257
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8667508377627653e-05,
-      "loss": 0.8088,
+      "loss": 0.9298,
       "step": 5258
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8666920869123265e-05,
-      "loss": 0.7921,
+      "loss": 0.9569,
       "step": 5259
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.866633324037765e-05,
-      "loss": 0.9306,
+      "loss": 1.0581,
       "step": 5260
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8665745491398954e-05,
-      "loss": 0.9209,
+      "loss": 1.0412,
       "step": 5261
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8665157622195337e-05,
-      "loss": 0.7002,
+      "loss": 0.926,
       "step": 5262
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8664569632774953e-05,
-      "loss": 0.7573,
+      "loss": 0.9035,
       "step": 5263
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8663981523145957e-05,
-      "loss": 0.8385,
+      "loss": 0.9389,
       "step": 5264
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.866339329331651e-05,
-      "loss": 0.7726,
+      "loss": 0.85,
       "step": 5265
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8662804943294773e-05,
-      "loss": 0.7308,
+      "loss": 0.787,
       "step": 5266
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.866221647308891e-05,
-      "loss": 0.8485,
+      "loss": 0.9128,
       "step": 5267
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8661627882707083e-05,
-      "loss": 0.8647,
+      "loss": 0.8941,
       "step": 5268
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8661039172157457e-05,
-      "loss": 0.7674,
+      "loss": 0.8239,
       "step": 5269
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8660450341448202e-05,
-      "loss": 0.7063,
+      "loss": 0.8051,
       "step": 5270
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8659861390587486e-05,
-      "loss": 0.9537,
+      "loss": 1.0453,
       "step": 5271
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865927231958348e-05,
-      "loss": 0.9323,
+      "loss": 1.0082,
       "step": 5272
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8658683128444352e-05,
-      "loss": 0.808,
+      "loss": 0.9071,
       "step": 5273
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8658093817178286e-05,
-      "loss": 0.8519,
+      "loss": 0.9952,
       "step": 5274
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865750438579345e-05,
-      "loss": 0.5863,
+      "loss": 0.8906,
       "step": 5275
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8656914834298026e-05,
-      "loss": 0.7796,
+      "loss": 0.9711,
       "step": 5276
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865632516270019e-05,
-      "loss": 0.9034,
+      "loss": 0.9908,
       "step": 5277
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8655735371008126e-05,
-      "loss": 0.9075,
+      "loss": 0.9868,
       "step": 5278
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8655145459230012e-05,
-      "loss": 0.9016,
+      "loss": 1.0204,
       "step": 5279
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8654555427374037e-05,
-      "loss": 0.9201,
+      "loss": 0.9779,
       "step": 5280
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865396527544838e-05,
-      "loss": 0.8515,
+      "loss": 0.9821,
       "step": 5281
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865337500346124e-05,
-      "loss": 0.7123,
+      "loss": 0.857,
       "step": 5282
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8652784611420797e-05,
-      "loss": 0.782,
+      "loss": 0.9756,
       "step": 5283
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8652194099335244e-05,
-      "loss": 0.7287,
+      "loss": 0.9062,
       "step": 5284
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865160346721277e-05,
-      "loss": 0.8702,
+      "loss": 1.0145,
       "step": 5285
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.865101271506158e-05,
-      "loss": 0.7681,
+      "loss": 0.8795,
       "step": 5286
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8650421842889853e-05,
-      "loss": 0.8691,
+      "loss": 0.999,
       "step": 5287
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8649830850705805e-05,
-      "loss": 0.8458,
+      "loss": 0.98,
       "step": 5288
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8649239738517623e-05,
-      "loss": 0.7798,
+      "loss": 0.9417,
       "step": 5289
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864864850633351e-05,
-      "loss": 0.8306,
+      "loss": 0.9249,
       "step": 5290
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8648057154161674e-05,
-      "loss": 0.8143,
+      "loss": 0.9288,
       "step": 5291
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864746568201031e-05,
-      "loss": 0.8604,
+      "loss": 0.9403,
       "step": 5292
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8646874089887636e-05,
-      "loss": 0.8414,
+      "loss": 0.9248,
       "step": 5293
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8646282377801843e-05,
-      "loss": 0.7838,
+      "loss": 0.9539,
       "step": 5294
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8645690545761156e-05,
-      "loss": 0.8506,
+      "loss": 0.9135,
       "step": 5295
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864509859377378e-05,
-      "loss": 0.8368,
+      "loss": 0.9877,
       "step": 5296
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8644506521847923e-05,
-      "loss": 0.8258,
+      "loss": 0.9643,
       "step": 5297
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8643914329991803e-05,
-      "loss": 0.7835,
+      "loss": 0.9353,
       "step": 5298
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8643322018213636e-05,
-      "loss": 0.7584,
+      "loss": 0.9347,
       "step": 5299
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864272958652164e-05,
-      "loss": 0.8422,
+      "loss": 0.9724,
       "step": 5300
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8642137034924036e-05,
-      "loss": 0.8419,
+      "loss": 0.9386,
       "step": 5301
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864154436342904e-05,
-      "loss": 0.8539,
+      "loss": 0.9166,
       "step": 5302
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8640951572044876e-05,
-      "loss": 0.8539,
+      "loss": 0.893,
       "step": 5303
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.864035866077977e-05,
-      "loss": 0.9509,
+      "loss": 1.0056,
       "step": 5304
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8639765629641943e-05,
-      "loss": 0.7577,
+      "loss": 0.916,
       "step": 5305
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8639172478639632e-05,
-      "loss": 0.8716,
+      "loss": 0.9169,
       "step": 5306
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8638579207781055e-05,
-      "loss": 0.8778,
+      "loss": 0.9749,
       "step": 5307
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.863798581707445e-05,
-      "loss": 0.7767,
+      "loss": 0.8767,
       "step": 5308
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8637392306528047e-05,
-      "loss": 0.7872,
+      "loss": 0.9689,
       "step": 5309
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8636798676150083e-05,
-      "loss": 0.9622,
+      "loss": 1.0622,
       "step": 5310
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8636204925948788e-05,
-      "loss": 0.9856,
+      "loss": 0.9651,
       "step": 5311
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8635611055932406e-05,
-      "loss": 0.8541,
+      "loss": 0.9448,
       "step": 5312
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8635017066109173e-05,
-      "loss": 0.7976,
+      "loss": 0.9227,
       "step": 5313
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8634422956487327e-05,
-      "loss": 0.8867,
+      "loss": 0.9472,
       "step": 5314
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8633828727075112e-05,
-      "loss": 0.9609,
+      "loss": 1.056,
       "step": 5315
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8633234377880777e-05,
-      "loss": 0.7949,
+      "loss": 0.875,
       "step": 5316
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8632639908912565e-05,
-      "loss": 0.8714,
+      "loss": 0.9588,
       "step": 5317
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.863204532017872e-05,
-      "loss": 0.9813,
+      "loss": 1.0527,
       "step": 5318
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8631450611687495e-05,
-      "loss": 0.8756,
+      "loss": 0.9764,
       "step": 5319
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8630855783447134e-05,
-      "loss": 0.9439,
+      "loss": 1.0262,
       "step": 5320
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.86302608354659e-05,
-      "loss": 0.9636,
+      "loss": 0.984,
       "step": 5321
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862966576775204e-05,
-      "loss": 0.8431,
+      "loss": 0.9966,
       "step": 5322
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862907058031381e-05,
-      "loss": 0.814,
+      "loss": 0.9597,
       "step": 5323
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8628475273159473e-05,
-      "loss": 0.8256,
+      "loss": 0.9249,
       "step": 5324
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862787984629728e-05,
-      "loss": 0.6734,
+      "loss": 0.7922,
       "step": 5325
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8627284299735497e-05,
-      "loss": 0.7625,
+      "loss": 0.8786,
       "step": 5326
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862668863348238e-05,
-      "loss": 0.8134,
+      "loss": 0.9485,
       "step": 5327
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8626092847546204e-05,
-      "loss": 0.848,
+      "loss": 0.9391,
       "step": 5328
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862549694193523e-05,
-      "loss": 0.8939,
+      "loss": 0.9669,
       "step": 5329
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862490091665772e-05,
-      "loss": 0.7601,
+      "loss": 0.9082,
       "step": 5330
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8624304771721945e-05,
-      "loss": 0.7427,
+      "loss": 0.8646,
       "step": 5331
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862370850713618e-05,
-      "loss": 0.7987,
+      "loss": 0.9604,
       "step": 5332
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8623112122908696e-05,
-      "loss": 0.9259,
+      "loss": 0.9427,
       "step": 5333
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8622515619047763e-05,
-      "loss": 0.832,
+      "loss": 0.9626,
       "step": 5334
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8621918995561663e-05,
-      "loss": 0.7637,
+      "loss": 0.9543,
       "step": 5335
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862132225245867e-05,
-      "loss": 0.8464,
+      "loss": 0.94,
       "step": 5336
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.862072538974706e-05,
-      "loss": 0.8132,
+      "loss": 0.9141,
       "step": 5337
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8620128407435114e-05,
-      "loss": 0.852,
+      "loss": 1.0635,
       "step": 5338
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861953130553112e-05,
-      "loss": 0.8137,
+      "loss": 0.876,
       "step": 5339
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861893408404336e-05,
-      "loss": 0.8049,
+      "loss": 0.9135,
       "step": 5340
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8618336742980116e-05,
-      "loss": 0.7849,
+      "loss": 0.8707,
       "step": 5341
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861773928234968e-05,
-      "loss": 0.8606,
+      "loss": 1.0089,
       "step": 5342
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861714170216034e-05,
-      "loss": 0.8162,
+      "loss": 0.914,
       "step": 5343
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861654400242038e-05,
-      "loss": 0.7583,
+      "loss": 0.8676,
       "step": 5344
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.86159461831381e-05,
-      "loss": 0.8613,
+      "loss": 0.9641,
       "step": 5345
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861534824432179e-05,
-      "loss": 0.9615,
+      "loss": 1.0568,
       "step": 5346
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8614750185979744e-05,
-      "loss": 0.9162,
+      "loss": 0.9225,
       "step": 5347
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861415200812027e-05,
-      "loss": 0.8407,
+      "loss": 0.9727,
       "step": 5348
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861355371075165e-05,
-      "loss": 0.7353,
+      "loss": 0.869,
       "step": 5349
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8612955293882197e-05,
-      "loss": 0.8161,
+      "loss": 0.8725,
       "step": 5350
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8612356757520207e-05,
-      "loss": 0.6253,
+      "loss": 0.7485,
       "step": 5351
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.861175810167399e-05,
-      "loss": 0.8456,
+      "loss": 0.9956,
       "step": 5352
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8611159326351847e-05,
-      "loss": 0.8013,
+      "loss": 0.9176,
       "step": 5353
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8610560431562085e-05,
-      "loss": 0.8765,
+      "loss": 0.8819,
       "step": 5354
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8609961417313017e-05,
-      "loss": 0.7991,
+      "loss": 0.9817,
       "step": 5355
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8609362283612948e-05,
-      "loss": 0.7628,
+      "loss": 0.8591,
       "step": 5356
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.860876303047019e-05,
-      "loss": 0.8407,
+      "loss": 0.8933,
       "step": 5357
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8608163657893063e-05,
-      "loss": 0.8135,
+      "loss": 0.8935,
       "step": 5358
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8607564165889877e-05,
-      "loss": 0.8526,
+      "loss": 0.9613,
       "step": 5359
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8606964554468948e-05,
-      "loss": 0.8607,
+      "loss": 0.9434,
       "step": 5360
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8606364823638602e-05,
-      "loss": 0.9,
+      "loss": 0.8979,
       "step": 5361
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8605764973407155e-05,
-      "loss": 0.8083,
+      "loss": 0.9171,
       "step": 5362
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8605165003782927e-05,
-      "loss": 0.6573,
+      "loss": 0.8248,
       "step": 5363
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.860456491477424e-05,
-      "loss": 0.8074,
+      "loss": 0.9373,
       "step": 5364
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8603964706389427e-05,
-      "loss": 0.8117,
+      "loss": 0.926,
       "step": 5365
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8603364378636813e-05,
-      "loss": 0.9456,
+      "loss": 1.0175,
       "step": 5366
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8602763931524724e-05,
-      "loss": 0.8044,
+      "loss": 1.0084,
       "step": 5367
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8602163365061488e-05,
-      "loss": 0.8344,
+      "loss": 0.9086,
       "step": 5368
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.860156267925544e-05,
-      "loss": 0.8428,
+      "loss": 0.9788,
       "step": 5369
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.860096187411492e-05,
-      "loss": 0.815,
+      "loss": 0.9752,
       "step": 5370
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.860036094964825e-05,
-      "loss": 0.6829,
+      "loss": 0.8389,
       "step": 5371
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859975990586378e-05,
-      "loss": 0.7662,
+      "loss": 0.9075,
       "step": 5372
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8599158742769837e-05,
-      "loss": 0.7849,
+      "loss": 0.9346,
       "step": 5373
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8598557460374774e-05,
-      "loss": 0.8693,
+      "loss": 0.9608,
       "step": 5374
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859795605868692e-05,
-      "loss": 0.7682,
+      "loss": 0.8814,
       "step": 5375
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8597354537714628e-05,
-      "loss": 0.8527,
+      "loss": 1.0092,
       "step": 5376
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8596752897466238e-05,
-      "loss": 0.7219,
+      "loss": 0.8943,
       "step": 5377
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.85961511379501e-05,
-      "loss": 0.8177,
+      "loss": 0.9363,
       "step": 5378
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8595549259174558e-05,
-      "loss": 0.7485,
+      "loss": 0.9657,
       "step": 5379
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8594947261147968e-05,
-      "loss": 0.9096,
+      "loss": 0.9855,
       "step": 5380
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859434514387868e-05,
-      "loss": 0.6879,
+      "loss": 0.8746,
       "step": 5381
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859374290737505e-05,
-      "loss": 0.8231,
+      "loss": 0.9797,
       "step": 5382
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8593140551645422e-05,
-      "loss": 0.7338,
+      "loss": 0.8685,
       "step": 5383
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859253807669817e-05,
-      "loss": 0.8358,
+      "loss": 0.9238,
       "step": 5384
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8591935482541636e-05,
-      "loss": 0.676,
+      "loss": 0.8139,
       "step": 5385
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859133276918419e-05,
-      "loss": 0.805,
+      "loss": 0.9107,
       "step": 5386
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.859072993663419e-05,
-      "loss": 0.9286,
+      "loss": 0.9865,
       "step": 5387
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8590126984900005e-05,
-      "loss": 0.7309,
+      "loss": 0.879,
       "step": 5388
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8589523913989992e-05,
-      "loss": 0.768,
+      "loss": 0.8601,
       "step": 5389
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8588920723912523e-05,
-      "loss": 0.8035,
+      "loss": 0.9007,
       "step": 5390
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8588317414675966e-05,
-      "loss": 0.9083,
+      "loss": 0.9122,
       "step": 5391
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8587713986288688e-05,
-      "loss": 0.915,
+      "loss": 0.967,
       "step": 5392
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8587110438759063e-05,
-      "loss": 0.7792,
+      "loss": 1.002,
       "step": 5393
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8586506772095467e-05,
-      "loss": 0.7713,
+      "loss": 0.9378,
       "step": 5394
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.858590298630627e-05,
-      "loss": 0.7605,
+      "loss": 0.9494,
       "step": 5395
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.858529908139985e-05,
-      "loss": 0.7306,
+      "loss": 0.8559,
       "step": 5396
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8584695057384586e-05,
-      "loss": 0.8451,
+      "loss": 0.8912,
       "step": 5397
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.858409091426886e-05,
-      "loss": 0.865,
+      "loss": 0.9153,
       "step": 5398
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8583486652061054e-05,
-      "loss": 0.8247,
+      "loss": 0.9032,
       "step": 5399
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8582882270769547e-05,
-      "loss": 0.8616,
+      "loss": 1.009,
       "step": 5400
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8582277770402727e-05,
-      "loss": 0.7987,
+      "loss": 0.9455,
       "step": 5401
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8581673150968974e-05,
-      "loss": 0.8775,
+      "loss": 1.0384,
       "step": 5402
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.858106841247669e-05,
-      "loss": 0.7963,
+      "loss": 0.9416,
       "step": 5403
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.858046355493425e-05,
-      "loss": 0.8644,
+      "loss": 0.9881,
       "step": 5404
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8579858578350054e-05,
-      "loss": 0.7496,
+      "loss": 0.9047,
       "step": 5405
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8579253482732494e-05,
-      "loss": 0.8882,
+      "loss": 0.9973,
       "step": 5406
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8578648268089966e-05,
-      "loss": 0.9423,
+      "loss": 1.0662,
       "step": 5407
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8578042934430863e-05,
-      "loss": 0.7364,
+      "loss": 0.7615,
       "step": 5408
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8577437481763586e-05,
-      "loss": 0.6965,
+      "loss": 0.8143,
       "step": 5409
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8576831910096532e-05,
-      "loss": 0.808,
+      "loss": 0.8802,
       "step": 5410
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8576226219438106e-05,
-      "loss": 0.8939,
+      "loss": 0.9779,
       "step": 5411
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8575620409796705e-05,
-      "loss": 0.7762,
+      "loss": 0.9208,
       "step": 5412
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8575014481180744e-05,
-      "loss": 0.8876,
+      "loss": 0.9922,
       "step": 5413
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.857440843359862e-05,
-      "loss": 0.8342,
+      "loss": 0.9958,
       "step": 5414
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.857380226705875e-05,
-      "loss": 0.7844,
+      "loss": 0.9258,
       "step": 5415
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8573195981569528e-05,
-      "loss": 0.7682,
+      "loss": 0.9818,
       "step": 5416
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8572589577139382e-05,
-      "loss": 0.8201,
+      "loss": 0.9905,
       "step": 5417
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.857198305377672e-05,
-      "loss": 0.8751,
+      "loss": 0.9802,
       "step": 5418
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.857137641148995e-05,
-      "loss": 0.9471,
+      "loss": 0.9852,
       "step": 5419
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8570769650287496e-05,
-      "loss": 0.7803,
+      "loss": 0.8724,
       "step": 5420
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8570162770177775e-05,
-      "loss": 0.8856,
+      "loss": 1.0437,
       "step": 5421
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8569555771169203e-05,
-      "loss": 0.8838,
+      "loss": 0.9051,
       "step": 5422
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8568948653270206e-05,
-      "loss": 0.7766,
+      "loss": 0.9161,
       "step": 5423
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8568341416489202e-05,
-      "loss": 0.7496,
+      "loss": 0.8813,
       "step": 5424
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8567734060834616e-05,
-      "loss": 0.8324,
+      "loss": 0.922,
       "step": 5425
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.856712658631488e-05,
-      "loss": 0.7545,
+      "loss": 0.9907,
       "step": 5426
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8566518992938416e-05,
-      "loss": 0.6893,
+      "loss": 0.864,
       "step": 5427
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8565911280713656e-05,
-      "loss": 0.9214,
+      "loss": 0.9765,
       "step": 5428
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8565303449649027e-05,
-      "loss": 0.9448,
+      "loss": 1.0157,
       "step": 5429
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8564695499752968e-05,
-      "loss": 0.9204,
+      "loss": 0.9788,
       "step": 5430
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.856408743103391e-05,
-      "loss": 0.8386,
+      "loss": 0.9594,
       "step": 5431
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.856347924350029e-05,
-      "loss": 0.8126,
+      "loss": 0.895,
       "step": 5432
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8562870937160545e-05,
-      "loss": 0.7509,
+      "loss": 0.8898,
       "step": 5433
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8562262512023115e-05,
-      "loss": 0.8229,
+      "loss": 0.9548,
       "step": 5434
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8561653968096443e-05,
-      "loss": 0.7672,
+      "loss": 0.8844,
       "step": 5435
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8561045305388968e-05,
-      "loss": 0.9529,
+      "loss": 1.0092,
       "step": 5436
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8560436523909137e-05,
-      "loss": 0.8387,
+      "loss": 0.9997,
       "step": 5437
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.855982762366539e-05,
-      "loss": 0.8678,
+      "loss": 1.0031,
       "step": 5438
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8559218604666185e-05,
-      "loss": 0.8752,
+      "loss": 0.8998,
       "step": 5439
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8558609466919963e-05,
-      "loss": 0.8726,
+      "loss": 1.0468,
       "step": 5440
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8558000210435182e-05,
-      "loss": 0.9009,
+      "loss": 0.9636,
       "step": 5441
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8557390835220282e-05,
-      "loss": 0.9282,
+      "loss": 0.9754,
       "step": 5442
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8556781341283733e-05,
-      "loss": 0.8499,
+      "loss": 0.9549,
       "step": 5443
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.855617172863398e-05,
-      "loss": 0.8698,
+      "loss": 1.0324,
       "step": 5444
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8555561997279484e-05,
-      "loss": 0.8716,
+      "loss": 0.942,
       "step": 5445
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8554952147228705e-05,
-      "loss": 0.7337,
+      "loss": 0.8901,
       "step": 5446
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8554342178490104e-05,
-      "loss": 0.7104,
+      "loss": 0.8832,
       "step": 5447
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.855373209107214e-05,
-      "loss": 0.95,
+      "loss": 1.0037,
       "step": 5448
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8553121884983284e-05,
-      "loss": 0.9146,
+      "loss": 1.0265,
       "step": 5449
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.855251156023199e-05,
-      "loss": 0.7008,
+      "loss": 0.86,
       "step": 5450
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.855190111682674e-05,
-      "loss": 0.8419,
+      "loss": 1.0005,
       "step": 5451
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8551290554775988e-05,
-      "loss": 0.7532,
+      "loss": 0.9282,
       "step": 5452
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8550679874088217e-05,
-      "loss": 0.9654,
+      "loss": 1.0626,
       "step": 5453
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8550069074771893e-05,
-      "loss": 0.8188,
+      "loss": 0.9799,
       "step": 5454
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854945815683549e-05,
-      "loss": 0.8411,
+      "loss": 0.8706,
       "step": 5455
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854884712028749e-05,
-      "loss": 0.8416,
+      "loss": 0.9128,
       "step": 5456
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854823596513636e-05,
-      "loss": 0.8984,
+      "loss": 0.9958,
       "step": 5457
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854762469139059e-05,
-      "loss": 0.784,
+      "loss": 0.9366,
       "step": 5458
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8547013299058652e-05,
-      "loss": 0.899,
+      "loss": 0.9373,
       "step": 5459
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854640178814903e-05,
-      "loss": 0.8458,
+      "loss": 0.8599,
       "step": 5460
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854579015867021e-05,
-      "loss": 0.8271,
+      "loss": 0.9554,
       "step": 5461
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8545178410630676e-05,
-      "loss": 0.8739,
+      "loss": 0.9546,
       "step": 5462
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8544566544038917e-05,
-      "loss": 0.8802,
+      "loss": 0.9476,
       "step": 5463
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854395455890342e-05,
-      "loss": 0.8573,
+      "loss": 0.8973,
       "step": 5464
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8543342455232674e-05,
-      "loss": 0.8309,
+      "loss": 0.9491,
       "step": 5465
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8542730233035176e-05,
-      "loss": 0.66,
+      "loss": 0.8163,
       "step": 5466
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854211789231941e-05,
-      "loss": 0.7603,
+      "loss": 0.9518,
       "step": 5467
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8541505433093884e-05,
-      "loss": 0.7895,
+      "loss": 0.8613,
       "step": 5468
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8540892855367086e-05,
-      "loss": 0.8845,
+      "loss": 0.9659,
       "step": 5469
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.854028015914752e-05,
-      "loss": 0.8336,
+      "loss": 0.918,
       "step": 5470
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8539667344443682e-05,
-      "loss": 0.8491,
+      "loss": 0.9951,
       "step": 5471
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8539054411264075e-05,
-      "loss": 0.8546,
+      "loss": 1.0038,
       "step": 5472
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8538441359617207e-05,
-      "loss": 0.7859,
+      "loss": 0.9066,
       "step": 5473
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.853782818951158e-05,
-      "loss": 0.7548,
+      "loss": 0.8874,
       "step": 5474
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.85372149009557e-05,
-      "loss": 0.7974,
+      "loss": 0.9528,
       "step": 5475
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8536601493958072e-05,
-      "loss": 0.9264,
+      "loss": 0.9875,
       "step": 5476
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.853598796852721e-05,
-      "loss": 0.781,
+      "loss": 0.924,
       "step": 5477
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.853537432467163e-05,
-      "loss": 0.9462,
+      "loss": 0.9825,
       "step": 5478
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.853476056239984e-05,
-      "loss": 0.8103,
+      "loss": 0.8458,
       "step": 5479
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8534146681720354e-05,
-      "loss": 0.8169,
+      "loss": 0.967,
       "step": 5480
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8533532682641693e-05,
-      "loss": 0.9013,
+      "loss": 1.0178,
       "step": 5481
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8532918565172372e-05,
-      "loss": 0.9406,
+      "loss": 1.0058,
       "step": 5482
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8532304329320917e-05,
-      "loss": 0.9349,
+      "loss": 0.9691,
       "step": 5483
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8531689975095842e-05,
-      "loss": 0.8543,
+      "loss": 0.9218,
       "step": 5484
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.853107550250567e-05,
-      "loss": 0.7883,
+      "loss": 0.968,
       "step": 5485
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8530460911558937e-05,
-      "loss": 0.7709,
+      "loss": 0.901,
       "step": 5486
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8529846202264155e-05,
-      "loss": 0.9344,
+      "loss": 1.0041,
       "step": 5487
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8529231374629862e-05,
-      "loss": 0.9473,
+      "loss": 1.0085,
       "step": 5488
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852861642866459e-05,
-      "loss": 0.8366,
+      "loss": 0.9039,
       "step": 5489
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852800136437686e-05,
-      "loss": 0.7815,
+      "loss": 0.9143,
       "step": 5490
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852738618177521e-05,
-      "loss": 0.8551,
+      "loss": 0.9592,
       "step": 5491
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8526770880868178e-05,
-      "loss": 0.7932,
+      "loss": 0.8979,
       "step": 5492
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8526155461664296e-05,
-      "loss": 0.8751,
+      "loss": 1.0083,
       "step": 5493
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8525539924172106e-05,
-      "loss": 0.8559,
+      "loss": 0.9889,
       "step": 5494
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8524924268400145e-05,
-      "loss": 0.8885,
+      "loss": 0.9433,
       "step": 5495
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8524308494356953e-05,
-      "loss": 0.8022,
+      "loss": 0.9655,
       "step": 5496
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8523692602051078e-05,
-      "loss": 0.8749,
+      "loss": 0.9741,
       "step": 5497
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8523076591491062e-05,
-      "loss": 0.5826,
+      "loss": 0.7651,
       "step": 5498
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8522460462685447e-05,
-      "loss": 0.7147,
+      "loss": 0.8375,
       "step": 5499
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852184421564279e-05,
-      "loss": 0.8602,
+      "loss": 0.98,
       "step": 5500
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852122785037163e-05,
-      "loss": 0.7853,
+      "loss": 0.8871,
       "step": 5501
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.852061136688053e-05,
-      "loss": 0.8906,
+      "loss": 0.9285,
       "step": 5502
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851999476517803e-05,
-      "loss": 0.8746,
+      "loss": 0.9749,
       "step": 5503
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8519378045272692e-05,
-      "loss": 0.8497,
+      "loss": 0.9725,
       "step": 5504
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8518761207173074e-05,
-      "loss": 0.8288,
+      "loss": 0.8492,
       "step": 5505
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8518144250887732e-05,
-      "loss": 0.7537,
+      "loss": 0.8805,
       "step": 5506
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8517527176425224e-05,
-      "loss": 0.9522,
+      "loss": 1.0486,
       "step": 5507
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8516909983794108e-05,
-      "loss": 0.8906,
+      "loss": 0.9152,
       "step": 5508
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851629267300295e-05,
-      "loss": 0.7954,
+      "loss": 0.9142,
       "step": 5509
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8515675244060314e-05,
-      "loss": 0.7395,
+      "loss": 0.9347,
       "step": 5510
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851505769697477e-05,
-      "loss": 0.8142,
+      "loss": 0.8979,
       "step": 5511
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8514440031754875e-05,
-      "loss": 0.7936,
+      "loss": 0.9221,
       "step": 5512
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8513822248409213e-05,
-      "loss": 0.8688,
+      "loss": 0.9449,
       "step": 5513
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851320434694634e-05,
-      "loss": 0.9324,
+      "loss": 0.9736,
       "step": 5514
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851258632737484e-05,
-      "loss": 0.9054,
+      "loss": 0.9589,
       "step": 5515
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851196818970328e-05,
-      "loss": 0.8974,
+      "loss": 1.0238,
       "step": 5516
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.851134993394024e-05,
-      "loss": 0.7462,
+      "loss": 0.8623,
       "step": 5517
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8510731560094296e-05,
-      "loss": 0.8325,
+      "loss": 0.9396,
       "step": 5518
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8510113068174026e-05,
-      "loss": 0.8689,
+      "loss": 0.9579,
       "step": 5519
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8509494458188005e-05,
-      "loss": 0.8421,
+      "loss": 0.9295,
       "step": 5520
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.850887573014483e-05,
-      "loss": 0.8722,
+      "loss": 0.9925,
       "step": 5521
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8508256884053074e-05,
-      "loss": 0.8629,
+      "loss": 1.0135,
       "step": 5522
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8507637919921325e-05,
-      "loss": 0.7603,
+      "loss": 0.9722,
       "step": 5523
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8507018837758174e-05,
-      "loss": 0.7627,
+      "loss": 0.9155,
       "step": 5524
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8506399637572203e-05,
-      "loss": 0.789,
+      "loss": 0.9567,
       "step": 5525
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8505780319372008e-05,
-      "loss": 0.7797,
+      "loss": 0.9513,
       "step": 5526
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8505160883166178e-05,
-      "loss": 0.6206,
+      "loss": 0.757,
       "step": 5527
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8504541328963308e-05,
-      "loss": 0.8938,
+      "loss": 0.9722,
       "step": 5528
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8503921656771995e-05,
-      "loss": 0.8394,
+      "loss": 0.9068,
       "step": 5529
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.850330186660083e-05,
-      "loss": 0.8506,
+      "loss": 0.9424,
       "step": 5530
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.850268195845842e-05,
-      "loss": 0.8631,
+      "loss": 0.9602,
       "step": 5531
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.850206193235336e-05,
-      "loss": 0.8965,
+      "loss": 0.9827,
       "step": 5532
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8501441788294256e-05,
-      "loss": 0.8733,
+      "loss": 0.9374,
       "step": 5533
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.850082152628971e-05,
-      "loss": 0.7289,
+      "loss": 0.9674,
       "step": 5534
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8500201146348322e-05,
-      "loss": 0.8087,
+      "loss": 0.8913,
       "step": 5535
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8499580648478704e-05,
-      "loss": 0.7604,
+      "loss": 0.8715,
       "step": 5536
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8498960032689465e-05,
-      "loss": 0.9019,
+      "loss": 0.998,
       "step": 5537
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8498339298989216e-05,
-      "loss": 0.676,
+      "loss": 0.7635,
       "step": 5538
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8497718447386563e-05,
-      "loss": 0.9119,
+      "loss": 0.9313,
       "step": 5539
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8497097477890125e-05,
-      "loss": 0.7145,
+      "loss": 0.9041,
       "step": 5540
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8496476390508517e-05,
-      "loss": 0.8868,
+      "loss": 1.032,
       "step": 5541
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8495855185250355e-05,
-      "loss": 0.8306,
+      "loss": 0.8887,
       "step": 5542
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8495233862124248e-05,
-      "loss": 0.8418,
+      "loss": 1.0026,
       "step": 5543
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8494612421138832e-05,
-      "loss": 0.7853,
+      "loss": 0.8588,
       "step": 5544
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.849399086230272e-05,
-      "loss": 0.9394,
+      "loss": 1.0494,
       "step": 5545
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8493369185624536e-05,
-      "loss": 0.859,
+      "loss": 1.0029,
       "step": 5546
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8492747391112905e-05,
-      "loss": 0.9116,
+      "loss": 1.0075,
       "step": 5547
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.849212547877645e-05,
-      "loss": 0.7598,
+      "loss": 0.8649,
       "step": 5548
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8491503448623806e-05,
-      "loss": 0.7927,
+      "loss": 0.8927,
       "step": 5549
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8490881300663602e-05,
-      "loss": 0.8525,
+      "loss": 0.9523,
       "step": 5550
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8490259034904462e-05,
-      "loss": 0.829,
+      "loss": 0.9425,
       "step": 5551
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8489636651355026e-05,
-      "loss": 0.9101,
+      "loss": 0.9781,
       "step": 5552
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8489014150023928e-05,
-      "loss": 0.8744,
+      "loss": 0.9758,
       "step": 5553
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.84883915309198e-05,
-      "loss": 0.7366,
+      "loss": 0.8046,
       "step": 5554
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8487768794051286e-05,
-      "loss": 0.7498,
+      "loss": 0.8405,
       "step": 5555
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.848714593942702e-05,
-      "loss": 0.7731,
+      "loss": 0.8656,
       "step": 5556
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8486522967055646e-05,
-      "loss": 0.9021,
+      "loss": 0.9813,
       "step": 5557
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.848589987694581e-05,
-      "loss": 0.8552,
+      "loss": 0.9602,
       "step": 5558
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.848527666910615e-05,
-      "loss": 0.8447,
+      "loss": 0.933,
       "step": 5559
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8484653343545312e-05,
-      "loss": 0.8593,
+      "loss": 0.9509,
       "step": 5560
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8484029900271953e-05,
-      "loss": 0.7986,
+      "loss": 0.9283,
       "step": 5561
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8483406339294717e-05,
-      "loss": 0.831,
+      "loss": 0.962,
       "step": 5562
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.848278266062225e-05,
-      "loss": 0.7983,
+      "loss": 0.935,
       "step": 5563
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8482158864263213e-05,
-      "loss": 0.944,
+      "loss": 0.9621,
       "step": 5564
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8481534950226252e-05,
-      "loss": 0.8667,
+      "loss": 1.0006,
       "step": 5565
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8480910918520033e-05,
-      "loss": 0.8478,
+      "loss": 1.0419,
       "step": 5566
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8480286769153203e-05,
-      "loss": 0.8676,
+      "loss": 0.9973,
       "step": 5567
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8479662502134424e-05,
-      "loss": 0.8352,
+      "loss": 0.9312,
       "step": 5568
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8479038117472365e-05,
-      "loss": 0.8345,
+      "loss": 0.9476,
       "step": 5569
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8478413615175677e-05,
-      "loss": 0.9726,
+      "loss": 1.0925,
       "step": 5570
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8477788995253035e-05,
-      "loss": 0.7831,
+      "loss": 0.9289,
       "step": 5571
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8477164257713096e-05,
-      "loss": 0.7797,
+      "loss": 0.8823,
       "step": 5572
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.847653940256453e-05,
-      "loss": 0.7949,
+      "loss": 0.9526,
       "step": 5573
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8475914429816005e-05,
-      "loss": 0.9686,
+      "loss": 0.9948,
       "step": 5574
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8475289339476198e-05,
-      "loss": 0.8679,
+      "loss": 0.9557,
       "step": 5575
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.847466413155377e-05,
-      "loss": 0.7078,
+      "loss": 0.9036,
       "step": 5576
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8474038806057407e-05,
-      "loss": 0.8091,
+      "loss": 0.9661,
       "step": 5577
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8473413362995777e-05,
-      "loss": 0.9272,
+      "loss": 1.0269,
       "step": 5578
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8472787802377555e-05,
-      "loss": 0.8439,
+      "loss": 0.9767,
       "step": 5579
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8472162124211427e-05,
-      "loss": 0.8911,
+      "loss": 0.9399,
       "step": 5580
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.847153632850607e-05,
-      "loss": 0.8348,
+      "loss": 0.9602,
       "step": 5581
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8470910415270164e-05,
-      "loss": 0.8733,
+      "loss": 0.9352,
       "step": 5582
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8470284384512397e-05,
-      "loss": 0.7849,
+      "loss": 0.93,
       "step": 5583
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.846965823624145e-05,
-      "loss": 0.7857,
+      "loss": 0.9234,
       "step": 5584
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8469031970466014e-05,
-      "loss": 0.7259,
+      "loss": 0.7419,
       "step": 5585
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8468405587194774e-05,
-      "loss": 0.878,
+      "loss": 0.9059,
       "step": 5586
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.846777908643642e-05,
-      "loss": 0.8559,
+      "loss": 0.9344,
       "step": 5587
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8467152468199647e-05,
-      "loss": 0.8309,
+      "loss": 0.94,
       "step": 5588
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.846652573249315e-05,
-      "loss": 0.7889,
+      "loss": 0.8586,
       "step": 5589
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8465898879325618e-05,
-      "loss": 0.8411,
+      "loss": 0.991,
       "step": 5590
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.846527190870575e-05,
-      "loss": 0.8026,
+      "loss": 0.9929,
       "step": 5591
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8464644820642243e-05,
-      "loss": 0.8676,
+      "loss": 0.9917,
       "step": 5592
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8464017615143805e-05,
-      "loss": 0.7531,
+      "loss": 0.9617,
       "step": 5593
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8463390292219126e-05,
-      "loss": 0.7306,
+      "loss": 0.8373,
       "step": 5594
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8462762851876913e-05,
-      "loss": 0.8615,
+      "loss": 0.9996,
       "step": 5595
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.846213529412588e-05,
-      "loss": 0.7908,
+      "loss": 0.9072,
       "step": 5596
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8461507618974722e-05,
-      "loss": 0.7572,
+      "loss": 0.8927,
       "step": 5597
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8460879826432148e-05,
-      "loss": 0.9168,
+      "loss": 0.9724,
       "step": 5598
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8460251916506874e-05,
-      "loss": 0.743,
+      "loss": 0.8342,
       "step": 5599
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8459623889207607e-05,
-      "loss": 0.9422,
+      "loss": 0.9302,
       "step": 5600
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8458995744543065e-05,
-      "loss": 0.7901,
+      "loss": 0.9722,
       "step": 5601
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8458367482521953e-05,
-      "loss": 0.9366,
+      "loss": 0.9673,
       "step": 5602
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8457739103152998e-05,
-      "loss": 0.7988,
+      "loss": 0.9383,
       "step": 5603
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.845711060644491e-05,
-      "loss": 0.862,
+      "loss": 0.9374,
       "step": 5604
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8456481992406413e-05,
-      "loss": 0.7373,
+      "loss": 0.8993,
       "step": 5605
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8455853261046225e-05,
-      "loss": 0.9087,
+      "loss": 0.9453,
       "step": 5606
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8455224412373072e-05,
-      "loss": 0.8247,
+      "loss": 0.9215,
       "step": 5607
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8454595446395675e-05,
-      "loss": 0.8566,
+      "loss": 0.9268,
       "step": 5608
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8453966363122764e-05,
-      "loss": 0.9828,
+      "loss": 1.1026,
       "step": 5609
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.845333716256306e-05,
-      "loss": 0.8455,
+      "loss": 0.9282,
       "step": 5610
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.84527078447253e-05,
-      "loss": 0.7476,
+      "loss": 0.9187,
       "step": 5611
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8452078409618214e-05,
-      "loss": 0.7958,
+      "loss": 0.8828,
       "step": 5612
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.845144885725053e-05,
-      "loss": 0.8768,
+      "loss": 1.0297,
       "step": 5613
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8450819187630983e-05,
-      "loss": 0.8742,
+      "loss": 0.9402,
       "step": 5614
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8450189400768308e-05,
-      "loss": 0.7844,
+      "loss": 0.8696,
       "step": 5615
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8449559496671245e-05,
-      "loss": 0.785,
+      "loss": 0.8823,
       "step": 5616
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8448929475348535e-05,
-      "loss": 0.8352,
+      "loss": 0.909,
       "step": 5617
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8448299336808914e-05,
-      "loss": 0.8229,
+      "loss": 0.8885,
       "step": 5618
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8447669081061126e-05,
-      "loss": 0.8954,
+      "loss": 0.9764,
       "step": 5619
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8447038708113915e-05,
-      "loss": 0.8821,
+      "loss": 0.9599,
       "step": 5620
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844640821797603e-05,
-      "loss": 0.7403,
+      "loss": 0.8642,
       "step": 5621
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844577761065621e-05,
-      "loss": 0.8151,
+      "loss": 0.9403,
       "step": 5622
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8445146886163213e-05,
-      "loss": 0.7518,
+      "loss": 0.8532,
       "step": 5623
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8444516044505784e-05,
-      "loss": 0.83,
+      "loss": 0.9088,
       "step": 5624
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8443885085692674e-05,
-      "loss": 0.7429,
+      "loss": 0.9177,
       "step": 5625
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844325400973264e-05,
-      "loss": 0.6854,
+      "loss": 0.8925,
       "step": 5626
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8442622816634434e-05,
-      "loss": 0.6862,
+      "loss": 0.8023,
       "step": 5627
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844199150640682e-05,
-      "loss": 0.7554,
+      "loss": 0.9079,
       "step": 5628
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844136007905855e-05,
-      "loss": 0.8844,
+      "loss": 0.9869,
       "step": 5629
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.844072853459838e-05,
-      "loss": 0.7819,
+      "loss": 0.9001,
       "step": 5630
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8440096873035082e-05,
-      "loss": 0.7471,
+      "loss": 0.8269,
       "step": 5631
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8439465094377417e-05,
-      "loss": 0.9685,
+      "loss": 0.9918,
       "step": 5632
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8438833198634146e-05,
-      "loss": 0.8226,
+      "loss": 0.9568,
       "step": 5633
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8438201185814036e-05,
-      "loss": 0.7721,
+      "loss": 0.8567,
       "step": 5634
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8437569055925855e-05,
-      "loss": 0.8156,
+      "loss": 0.9598,
       "step": 5635
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8436936808978377e-05,
-      "loss": 0.8376,
+      "loss": 0.9176,
       "step": 5636
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8436304444980372e-05,
-      "loss": 0.723,
+      "loss": 0.8456,
       "step": 5637
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8435671963940612e-05,
-      "loss": 0.8387,
+      "loss": 0.9197,
       "step": 5638
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.843503936586787e-05,
-      "loss": 0.829,
+      "loss": 0.9985,
       "step": 5639
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8434406650770928e-05,
-      "loss": 0.7402,
+      "loss": 0.8405,
       "step": 5640
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.843377381865856e-05,
-      "loss": 0.8263,
+      "loss": 0.9249,
       "step": 5641
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8433140869539545e-05,
-      "loss": 0.8275,
+      "loss": 0.9404,
       "step": 5642
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8432507803422666e-05,
-      "loss": 0.7508,
+      "loss": 0.8419,
       "step": 5643
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8431874620316706e-05,
-      "loss": 0.8596,
+      "loss": 0.9438,
       "step": 5644
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8431241320230448e-05,
-      "loss": 0.8739,
+      "loss": 0.9618,
       "step": 5645
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8430607903172677e-05,
-      "loss": 0.7908,
+      "loss": 0.9863,
       "step": 5646
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8429974369152185e-05,
-      "loss": 0.8163,
+      "loss": 0.8706,
       "step": 5647
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8429340718177758e-05,
-      "loss": 0.8302,
+      "loss": 0.9097,
       "step": 5648
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.842870695025819e-05,
-      "loss": 0.9502,
+      "loss": 1.0061,
       "step": 5649
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8428073065402274e-05,
-      "loss": 0.8018,
+      "loss": 0.8487,
       "step": 5650
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.84274390636188e-05,
-      "loss": 0.8698,
+      "loss": 0.9756,
       "step": 5651
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8426804944916565e-05,
-      "loss": 0.8173,
+      "loss": 0.9422,
       "step": 5652
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8426170709304365e-05,
-      "loss": 0.9005,
+      "loss": 1.0461,
       "step": 5653
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8425536356791006e-05,
-      "loss": 0.8913,
+      "loss": 0.9885,
       "step": 5654
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.842490188738528e-05,
-      "loss": 1.0006,
+      "loss": 1.0042,
       "step": 5655
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8424267301095997e-05,
-      "loss": 0.8286,
+      "loss": 0.97,
       "step": 5656
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8423632597931957e-05,
-      "loss": 0.718,
+      "loss": 0.8841,
       "step": 5657
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.842299777790197e-05,
-      "loss": 0.7961,
+      "loss": 0.8653,
       "step": 5658
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8422362841014833e-05,
-      "loss": 0.7454,
+      "loss": 0.8989,
       "step": 5659
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8421727787279364e-05,
-      "loss": 0.8638,
+      "loss": 1.0063,
       "step": 5660
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8421092616704374e-05,
-      "loss": 0.8591,
+      "loss": 0.9558,
       "step": 5661
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8420457329298668e-05,
-      "loss": 0.9145,
+      "loss": 0.9608,
       "step": 5662
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8419821925071062e-05,
-      "loss": 0.9236,
+      "loss": 0.9883,
       "step": 5663
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8419186404030375e-05,
-      "loss": 0.884,
+      "loss": 0.9483,
       "step": 5664
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8418550766185424e-05,
-      "loss": 0.9861,
+      "loss": 1.028,
       "step": 5665
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.841791501154502e-05,
-      "loss": 0.9113,
+      "loss": 0.9647,
       "step": 5666
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8417279140117993e-05,
-      "loss": 0.8775,
+      "loss": 0.957,
       "step": 5667
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8416643151913164e-05,
-      "loss": 0.8416,
+      "loss": 0.9003,
       "step": 5668
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8416007046939348e-05,
-      "loss": 0.8603,
+      "loss": 0.9391,
       "step": 5669
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8415370825205374e-05,
-      "loss": 0.8373,
+      "loss": 0.9568,
       "step": 5670
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8414734486720075e-05,
-      "loss": 0.7497,
+      "loss": 0.8884,
       "step": 5671
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.841409803149227e-05,
-      "loss": 0.878,
+      "loss": 0.9325,
       "step": 5672
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8413461459530792e-05,
-      "loss": 0.9087,
+      "loss": 1.072,
       "step": 5673
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.841282477084447e-05,
-      "loss": 0.7417,
+      "loss": 0.8964,
       "step": 5674
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.841218796544215e-05,
-      "loss": 0.8396,
+      "loss": 0.9505,
       "step": 5675
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.841155104333265e-05,
-      "loss": 0.8462,
+      "loss": 0.9593,
       "step": 5676
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8410914004524818e-05,
-      "loss": 0.9076,
+      "loss": 1.0347,
       "step": 5677
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8410276849027486e-05,
-      "loss": 0.802,
+      "loss": 0.8551,
       "step": 5678
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8409639576849496e-05,
-      "loss": 0.8937,
+      "loss": 0.9801,
       "step": 5679
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8409002187999686e-05,
-      "loss": 0.7236,
+      "loss": 0.8321,
       "step": 5680
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8408364682486907e-05,
-      "loss": 0.7804,
+      "loss": 0.8734,
       "step": 5681
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.840772706031999e-05,
-      "loss": 0.717,
+      "loss": 0.8733,
       "step": 5682
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8407089321507796e-05,
-      "loss": 0.7636,
+      "loss": 0.942,
       "step": 5683
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8406451466059164e-05,
-      "loss": 0.8597,
+      "loss": 0.9858,
       "step": 5684
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.840581349398294e-05,
-      "loss": 0.837,
+      "loss": 0.8407,
       "step": 5685
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.840517540528799e-05,
-      "loss": 0.8203,
+      "loss": 0.9007,
       "step": 5686
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8404537199983148e-05,
-      "loss": 0.6948,
+      "loss": 0.8684,
       "step": 5687
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.840389887807728e-05,
-      "loss": 0.8515,
+      "loss": 0.9731,
       "step": 5688
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8403260439579238e-05,
-      "loss": 0.7755,
+      "loss": 0.8891,
       "step": 5689
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.840262188449788e-05,
-      "loss": 0.7806,
+      "loss": 0.9051,
       "step": 5690
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8401983212842064e-05,
-      "loss": 0.7627,
+      "loss": 0.9786,
       "step": 5691
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8401344424620655e-05,
-      "loss": 0.7868,
+      "loss": 0.8138,
       "step": 5692
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8400705519842512e-05,
-      "loss": 0.7622,
+      "loss": 0.9287,
       "step": 5693
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8400066498516498e-05,
-      "loss": 0.8241,
+      "loss": 0.9159,
       "step": 5694
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839942736065148e-05,
-      "loss": 0.8879,
+      "loss": 0.9712,
       "step": 5695
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839878810625632e-05,
-      "loss": 0.865,
+      "loss": 0.9804,
       "step": 5696
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8398148735339894e-05,
-      "loss": 0.7676,
+      "loss": 0.8812,
       "step": 5697
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8397509247911073e-05,
-      "loss": 0.7007,
+      "loss": 0.8856,
       "step": 5698
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8396869643978725e-05,
-      "loss": 0.6964,
+      "loss": 0.8609,
       "step": 5699
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839622992355172e-05,
-      "loss": 0.7228,
+      "loss": 0.8368,
       "step": 5700
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8395590086638938e-05,
-      "loss": 0.919,
+      "loss": 0.9616,
       "step": 5701
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839495013324926e-05,
-      "loss": 0.9311,
+      "loss": 0.9727,
       "step": 5702
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8394310063391555e-05,
-      "loss": 0.9151,
+      "loss": 1.0103,
       "step": 5703
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839366987707471e-05,
-      "loss": 0.8741,
+      "loss": 0.9331,
       "step": 5704
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8393029574307607e-05,
-      "loss": 0.8447,
+      "loss": 0.92,
       "step": 5705
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8392389155099122e-05,
-      "loss": 0.8815,
+      "loss": 0.924,
       "step": 5706
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839174861945815e-05,
-      "loss": 0.7581,
+      "loss": 0.9575,
       "step": 5707
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8391107967393568e-05,
-      "loss": 0.7546,
+      "loss": 0.895,
       "step": 5708
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.839046719891427e-05,
-      "loss": 0.9933,
+      "loss": 1.0637,
       "step": 5709
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8389826314029143e-05,
-      "loss": 0.8324,
+      "loss": 0.8988,
       "step": 5710
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838918531274708e-05,
-      "loss": 0.5941,
+      "loss": 0.7329,
       "step": 5711
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8388544195076972e-05,
-      "loss": 0.8455,
+      "loss": 0.9348,
       "step": 5712
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8387902961027718e-05,
-      "loss": 0.6503,
+      "loss": 0.8275,
       "step": 5713
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838726161060821e-05,
-      "loss": 0.8155,
+      "loss": 0.9185,
       "step": 5714
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8386620143827346e-05,
-      "loss": 0.8782,
+      "loss": 0.9157,
       "step": 5715
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8385978560694028e-05,
-      "loss": 0.7903,
+      "loss": 0.9259,
       "step": 5716
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8385336861217154e-05,
-      "loss": 0.846,
+      "loss": 0.9017,
       "step": 5717
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838469504540563e-05,
-      "loss": 0.8831,
+      "loss": 0.9194,
       "step": 5718
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838405311326836e-05,
-      "loss": 0.8065,
+      "loss": 0.9044,
       "step": 5719
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8383411064814244e-05,
-      "loss": 0.8355,
+      "loss": 0.9347,
       "step": 5720
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.83827689000522e-05,
-      "loss": 0.7665,
+      "loss": 0.9093,
       "step": 5721
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8382126618991125e-05,
-      "loss": 0.8489,
+      "loss": 0.8883,
       "step": 5722
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838148422163994e-05,
-      "loss": 0.7031,
+      "loss": 0.8946,
       "step": 5723
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.838084170800755e-05,
-      "loss": 0.7325,
+      "loss": 0.8382,
       "step": 5724
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8380199078102876e-05,
-      "loss": 0.8302,
+      "loss": 0.9291,
       "step": 5725
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8379556331934826e-05,
-      "loss": 0.8383,
+      "loss": 0.9307,
       "step": 5726
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8378913469512324e-05,
-      "loss": 0.8546,
+      "loss": 0.9648,
       "step": 5727
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8378270490844285e-05,
-      "loss": 1.0276,
+      "loss": 1.098,
       "step": 5728
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8377627395939628e-05,
-      "loss": 0.9063,
+      "loss": 0.925,
       "step": 5729
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.837698418480728e-05,
-      "loss": 0.9251,
+      "loss": 1.007,
       "step": 5730
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.837634085745616e-05,
-      "loss": 0.8752,
+      "loss": 0.9588,
       "step": 5731
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8375697413895195e-05,
-      "loss": 0.8112,
+      "loss": 0.9605,
       "step": 5732
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8375053854133314e-05,
-      "loss": 0.8956,
+      "loss": 0.9267,
       "step": 5733
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.837441017817944e-05,
-      "loss": 1.0042,
+      "loss": 0.9507,
       "step": 5734
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.837376638604251e-05,
-      "loss": 0.9416,
+      "loss": 0.9782,
       "step": 5735
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.837312247773145e-05,
-      "loss": 0.817,
+      "loss": 0.9525,
       "step": 5736
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8372478453255194e-05,
-      "loss": 0.7608,
+      "loss": 0.8226,
       "step": 5737
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8371834312622682e-05,
-      "loss": 0.7979,
+      "loss": 0.9997,
       "step": 5738
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8371190055842846e-05,
-      "loss": 0.8453,
+      "loss": 0.9865,
       "step": 5739
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8370545682924626e-05,
-      "loss": 0.8978,
+      "loss": 0.9595,
       "step": 5740
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836990119387696e-05,
-      "loss": 0.8908,
+      "loss": 0.9514,
       "step": 5741
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836925658870879e-05,
-      "loss": 0.7492,
+      "loss": 0.9183,
       "step": 5742
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836861186742906e-05,
-      "loss": 0.8581,
+      "loss": 0.9543,
       "step": 5743
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8367967030046712e-05,
-      "loss": 0.8648,
+      "loss": 0.9783,
       "step": 5744
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8367322076570694e-05,
-      "loss": 1.026,
+      "loss": 0.9716,
       "step": 5745
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8366677007009955e-05,
-      "loss": 0.9533,
+      "loss": 0.9375,
       "step": 5746
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836603182137344e-05,
-      "loss": 0.7769,
+      "loss": 0.8957,
       "step": 5747
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836538651967011e-05,
-      "loss": 0.7752,
+      "loss": 0.923,
       "step": 5748
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8364741101908908e-05,
-      "loss": 0.8091,
+      "loss": 0.8812,
       "step": 5749
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.836409556809879e-05,
-      "loss": 0.8433,
+      "loss": 0.9573,
       "step": 5750
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8363449918248715e-05,
-      "loss": 0.7751,
+      "loss": 0.8561,
       "step": 5751
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8362804152367638e-05,
-      "loss": 0.8021,
+      "loss": 0.8364,
       "step": 5752
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8362158270464515e-05,
-      "loss": 0.8486,
+      "loss": 0.9528,
       "step": 5753
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8361512272548317e-05,
-      "loss": 0.7963,
+      "loss": 0.9669,
       "step": 5754
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8360866158627996e-05,
-      "loss": 0.8553,
+      "loss": 0.9404,
       "step": 5755
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8360219928712517e-05,
-      "loss": 0.8332,
+      "loss": 0.9572,
       "step": 5756
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835957358281085e-05,
-      "loss": 0.9872,
+      "loss": 1.0323,
       "step": 5757
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8358927120931963e-05,
-      "loss": 0.903,
+      "loss": 0.9599,
       "step": 5758
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8358280543084822e-05,
-      "loss": 0.6591,
+      "loss": 0.8462,
       "step": 5759
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8357633849278393e-05,
-      "loss": 0.8742,
+      "loss": 0.9305,
       "step": 5760
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8356987039521654e-05,
-      "loss": 0.9252,
+      "loss": 0.9827,
       "step": 5761
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8356340113823578e-05,
-      "loss": 0.8207,
+      "loss": 0.8985,
       "step": 5762
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835569307219314e-05,
-      "loss": 0.8072,
+      "loss": 0.9078,
       "step": 5763
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8355045914639315e-05,
-      "loss": 0.9131,
+      "loss": 1.0457,
       "step": 5764
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835439864117108e-05,
-      "loss": 0.9188,
+      "loss": 0.9186,
       "step": 5765
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835375125179742e-05,
-      "loss": 0.8424,
+      "loss": 0.9395,
       "step": 5766
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835310374652731e-05,
-      "loss": 0.8908,
+      "loss": 1.0302,
       "step": 5767
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835245612536974e-05,
-      "loss": 0.9138,
+      "loss": 1.0033,
       "step": 5768
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835180838833369e-05,
-      "loss": 0.8585,
+      "loss": 0.933,
       "step": 5769
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.835116053542815e-05,
-      "loss": 0.9633,
+      "loss": 1.0026,
       "step": 5770
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8350512566662107e-05,
-      "loss": 0.8169,
+      "loss": 0.9784,
       "step": 5771
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834986448204455e-05,
-      "loss": 0.7648,
+      "loss": 0.8853,
       "step": 5772
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834921628158447e-05,
-      "loss": 0.8807,
+      "loss": 0.9189,
       "step": 5773
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8348567965290858e-05,
-      "loss": 0.6805,
+      "loss": 0.8684,
       "step": 5774
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834791953317271e-05,
-      "loss": 0.9032,
+      "loss": 0.9432,
       "step": 5775
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8347270985239025e-05,
-      "loss": 0.8583,
+      "loss": 0.9764,
       "step": 5776
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8346622321498797e-05,
-      "loss": 0.8306,
+      "loss": 0.8932,
       "step": 5777
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8345973541961027e-05,
-      "loss": 0.7686,
+      "loss": 0.82,
       "step": 5778
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8345324646634718e-05,
-      "loss": 0.8129,
+      "loss": 0.8442,
       "step": 5779
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8344675635528863e-05,
-      "loss": 0.8479,
+      "loss": 0.9982,
       "step": 5780
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834402650865248e-05,
-      "loss": 0.7606,
+      "loss": 0.8766,
       "step": 5781
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8343377266014565e-05,
-      "loss": 0.8126,
+      "loss": 0.9366,
       "step": 5782
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8342727907624128e-05,
-      "loss": 0.889,
+      "loss": 0.8232,
       "step": 5783
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834207843349018e-05,
-      "loss": 0.9251,
+      "loss": 0.9495,
       "step": 5784
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834142884362173e-05,
-      "loss": 0.8501,
+      "loss": 0.9103,
       "step": 5785
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8340779138027788e-05,
-      "loss": 0.8234,
+      "loss": 0.9861,
       "step": 5786
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.834012931671737e-05,
-      "loss": 0.799,
+      "loss": 0.9046,
       "step": 5787
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.833947937969949e-05,
-      "loss": 0.8691,
+      "loss": 0.865,
       "step": 5788
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8338829326983165e-05,
-      "loss": 0.9733,
+      "loss": 1.0427,
       "step": 5789
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8338179158577415e-05,
-      "loss": 0.9191,
+      "loss": 1.0526,
       "step": 5790
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.833752887449126e-05,
-      "loss": 0.776,
+      "loss": 0.8967,
       "step": 5791
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8336878474733724e-05,
-      "loss": 0.9691,
+      "loss": 1.0151,
       "step": 5792
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8336227959313825e-05,
-      "loss": 0.8138,
+      "loss": 0.8818,
       "step": 5793
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8335577328240586e-05,
-      "loss": 0.8409,
+      "loss": 0.9815,
       "step": 5794
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8334926581523044e-05,
-      "loss": 0.7966,
+      "loss": 0.9309,
       "step": 5795
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8334275719170224e-05,
-      "loss": 0.9553,
+      "loss": 0.991,
       "step": 5796
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8333624741191148e-05,
-      "loss": 0.8815,
+      "loss": 0.9444,
       "step": 5797
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8332973647594854e-05,
-      "loss": 0.7761,
+      "loss": 0.9922,
       "step": 5798
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8332322438390373e-05,
-      "loss": 0.7173,
+      "loss": 0.9162,
       "step": 5799
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.833167111358674e-05,
-      "loss": 0.8732,
+      "loss": 0.9311,
       "step": 5800
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.833101967319299e-05,
-      "loss": 0.8452,
+      "loss": 0.9874,
       "step": 5801
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8330368117218168e-05,
-      "loss": 0.8092,
+      "loss": 0.93,
       "step": 5802
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.83297164456713e-05,
-      "loss": 0.811,
+      "loss": 0.9109,
       "step": 5803
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.832906465856144e-05,
-      "loss": 0.9011,
+      "loss": 1.0366,
       "step": 5804
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8328412755897625e-05,
-      "loss": 0.7517,
+      "loss": 0.8793,
       "step": 5805
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8327760737688896e-05,
-      "loss": 0.8853,
+      "loss": 1.048,
       "step": 5806
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8327108603944305e-05,
-      "loss": 0.7206,
+      "loss": 0.8722,
       "step": 5807
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.83264563546729e-05,
-      "loss": 0.8054,
+      "loss": 0.9107,
       "step": 5808
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8325803989883724e-05,
-      "loss": 0.7523,
+      "loss": 0.9322,
       "step": 5809
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8325151509585827e-05,
-      "loss": 0.8956,
+      "loss": 0.9515,
       "step": 5810
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.832449891378827e-05,
-      "loss": 0.7979,
+      "loss": 0.9235,
       "step": 5811
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8323846202500094e-05,
-      "loss": 0.9998,
+      "loss": 1.0406,
       "step": 5812
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.832319337573037e-05,
-      "loss": 0.8873,
+      "loss": 0.9376,
       "step": 5813
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8322540433488144e-05,
-      "loss": 0.8678,
+      "loss": 0.9605,
       "step": 5814
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8321887375782477e-05,
-      "loss": 0.9494,
+      "loss": 1.0596,
       "step": 5815
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8321234202622427e-05,
-      "loss": 0.8023,
+      "loss": 0.9639,
       "step": 5816
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8320580914017062e-05,
-      "loss": 0.8604,
+      "loss": 0.8609,
       "step": 5817
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8319927509975445e-05,
-      "loss": 0.8276,
+      "loss": 0.9633,
       "step": 5818
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8319273990506636e-05,
-      "loss": 0.8301,
+      "loss": 0.9171,
       "step": 5819
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.83186203556197e-05,
-      "loss": 0.8843,
+      "loss": 0.9681,
       "step": 5820
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8317966605323714e-05,
-      "loss": 0.771,
+      "loss": 0.9329,
       "step": 5821
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.831731273962774e-05,
-      "loss": 0.8217,
+      "loss": 0.9375,
       "step": 5822
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8316658758540856e-05,
-      "loss": 0.8788,
+      "loss": 0.9702,
       "step": 5823
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.831600466207213e-05,
-      "loss": 0.8911,
+      "loss": 0.9667,
       "step": 5824
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8315350450230637e-05,
-      "loss": 0.9819,
+      "loss": 0.969,
       "step": 5825
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8314696123025456e-05,
-      "loss": 0.939,
+      "loss": 0.9397,
       "step": 5826
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.831404168046566e-05,
-      "loss": 0.8027,
+      "loss": 0.9321,
       "step": 5827
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8313387122560336e-05,
-      "loss": 0.8116,
+      "loss": 0.8536,
       "step": 5828
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8312732449318557e-05,
-      "loss": 0.7225,
+      "loss": 0.9066,
       "step": 5829
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.831207766074941e-05,
-      "loss": 0.8497,
+      "loss": 0.9326,
       "step": 5830
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8311422756861983e-05,
-      "loss": 0.7627,
+      "loss": 0.8923,
       "step": 5831
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8310767737665355e-05,
-      "loss": 0.8852,
+      "loss": 1.0402,
       "step": 5832
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8310112603168613e-05,
-      "loss": 0.8317,
+      "loss": 0.9066,
       "step": 5833
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8309457353380852e-05,
-      "loss": 0.8269,
+      "loss": 0.9693,
       "step": 5834
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.830880198831116e-05,
-      "loss": 0.8535,
+      "loss": 0.9592,
       "step": 5835
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.830814650796863e-05,
-      "loss": 0.7054,
+      "loss": 0.7519,
       "step": 5836
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8307490912362352e-05,
-      "loss": 0.7147,
+      "loss": 0.855,
       "step": 5837
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8306835201501428e-05,
-      "loss": 0.8333,
+      "loss": 0.9705,
       "step": 5838
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8306179375394946e-05,
-      "loss": 0.8686,
+      "loss": 0.9181,
       "step": 5839
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8305523434052018e-05,
-      "loss": 0.8143,
+      "loss": 0.9692,
       "step": 5840
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.830486737748173e-05,
-      "loss": 0.9998,
+      "loss": 0.995,
       "step": 5841
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8304211205693193e-05,
-      "loss": 0.8827,
+      "loss": 0.9418,
       "step": 5842
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8303554918695505e-05,
-      "loss": 0.7193,
+      "loss": 0.8651,
       "step": 5843
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8302898516497776e-05,
-      "loss": 0.8455,
+      "loss": 0.9354,
       "step": 5844
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8302241999109108e-05,
-      "loss": 0.9043,
+      "loss": 0.9467,
       "step": 5845
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8301585366538614e-05,
-      "loss": 0.7449,
+      "loss": 0.8539,
       "step": 5846
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8300928618795403e-05,
-      "loss": 0.7498,
+      "loss": 0.8282,
       "step": 5847
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8300271755888583e-05,
-      "loss": 0.7979,
+      "loss": 0.9565,
       "step": 5848
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8299614777827267e-05,
-      "loss": 0.8687,
+      "loss": 0.9961,
       "step": 5849
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8298957684620574e-05,
-      "loss": 0.8889,
+      "loss": 0.9794,
       "step": 5850
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8298300476277618e-05,
-      "loss": 0.8366,
+      "loss": 0.9425,
       "step": 5851
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8297643152807515e-05,
-      "loss": 0.6904,
+      "loss": 0.8276,
       "step": 5852
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8296985714219387e-05,
-      "loss": 0.9242,
+      "loss": 1.0099,
       "step": 5853
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8296328160522355e-05,
-      "loss": 0.7738,
+      "loss": 0.9135,
       "step": 5854
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.829567049172554e-05,
-      "loss": 0.763,
+      "loss": 0.8743,
       "step": 5855
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8295012707838068e-05,
-      "loss": 0.82,
+      "loss": 0.8962,
       "step": 5856
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8294354808869058e-05,
-      "loss": 0.7182,
+      "loss": 0.8734,
       "step": 5857
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.829369679482765e-05,
-      "loss": 0.8361,
+      "loss": 0.8513,
       "step": 5858
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8293038665722963e-05,
-      "loss": 0.7598,
+      "loss": 0.8939,
       "step": 5859
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.829238042156413e-05,
-      "loss": 0.8239,
+      "loss": 0.9159,
       "step": 5860
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8291722062360287e-05,
-      "loss": 0.8774,
+      "loss": 0.9311,
       "step": 5861
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8291063588120563e-05,
-      "loss": 0.801,
+      "loss": 0.943,
       "step": 5862
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8290404998854097e-05,
-      "loss": 0.8261,
+      "loss": 0.8906,
       "step": 5863
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8289746294570024e-05,
-      "loss": 0.7209,
+      "loss": 0.8899,
       "step": 5864
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828908747527748e-05,
-      "loss": 0.8044,
+      "loss": 0.8962,
       "step": 5865
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8288428540985607e-05,
-      "loss": 0.9392,
+      "loss": 1.0479,
       "step": 5866
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828776949170355e-05,
-      "loss": 0.7814,
+      "loss": 0.929,
       "step": 5867
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828711032744045e-05,
-      "loss": 0.7626,
+      "loss": 0.8882,
       "step": 5868
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8286451048205453e-05,
-      "loss": 0.7312,
+      "loss": 0.8607,
       "step": 5869
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8285791654007704e-05,
-      "loss": 0.8006,
+      "loss": 0.9084,
       "step": 5870
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8285132144856352e-05,
-      "loss": 0.7701,
+      "loss": 0.9009,
       "step": 5871
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828447252076055e-05,
-      "loss": 0.896,
+      "loss": 0.8995,
       "step": 5872
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828381278172944e-05,
-      "loss": 0.8416,
+      "loss": 0.9792,
       "step": 5873
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8283152927772182e-05,
-      "loss": 0.8165,
+      "loss": 0.8927,
       "step": 5874
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828249295889793e-05,
-      "loss": 0.8013,
+      "loss": 0.9147,
       "step": 5875
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8281832875115843e-05,
-      "loss": 0.7431,
+      "loss": 0.867,
       "step": 5876
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.828117267643507e-05,
-      "loss": 0.9431,
+      "loss": 0.9915,
       "step": 5877
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8280512362864776e-05,
-      "loss": 0.8931,
+      "loss": 0.996,
       "step": 5878
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8279851934414127e-05,
-      "loss": 0.8322,
+      "loss": 0.8728,
       "step": 5879
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8279191391092274e-05,
-      "loss": 0.8715,
+      "loss": 0.9933,
       "step": 5880
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.827853073290839e-05,
-      "loss": 0.8723,
+      "loss": 0.9168,
       "step": 5881
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8277869959871636e-05,
-      "loss": 0.8497,
+      "loss": 0.985,
       "step": 5882
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8277209071991183e-05,
-      "loss": 0.7725,
+      "loss": 0.8472,
       "step": 5883
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8276548069276197e-05,
-      "loss": 0.7566,
+      "loss": 0.8517,
       "step": 5884
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.827588695173585e-05,
-      "loss": 0.7028,
+      "loss": 0.798,
       "step": 5885
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.827522571937931e-05,
-      "loss": 0.8278,
+      "loss": 0.9186,
       "step": 5886
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8274564372215758e-05,
-      "loss": 0.8264,
+      "loss": 0.9294,
       "step": 5887
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8273902910254365e-05,
-      "loss": 0.9346,
+      "loss": 0.9549,
       "step": 5888
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.827324133350431e-05,
-      "loss": 0.9387,
+      "loss": 1.0474,
       "step": 5889
     },
     {
@@ -35350,83233 +35350,83233 @@
     {
       "epoch": 0.21,
       "learning_rate": 1.827191783567492e-05,
-      "loss": 0.8369,
+      "loss": 0.9388,
       "step": 5891
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8271255914613948e-05,
-      "loss": 0.8031,
+      "loss": 0.896,
       "step": 5892
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8270593878801035e-05,
-      "loss": 0.8039,
+      "loss": 0.8637,
       "step": 5893
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.826993172824537e-05,
-      "loss": 0.9357,
+      "loss": 1.0223,
       "step": 5894
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8269269462956133e-05,
-      "loss": 0.8529,
+      "loss": 0.9631,
       "step": 5895
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8268607082942512e-05,
-      "loss": 0.9265,
+      "loss": 0.9993,
       "step": 5896
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8267944588213704e-05,
-      "loss": 0.6821,
+      "loss": 0.8501,
       "step": 5897
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8267281978778895e-05,
-      "loss": 0.7714,
+      "loss": 0.8102,
       "step": 5898
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8266619254647277e-05,
-      "loss": 0.8164,
+      "loss": 0.8861,
       "step": 5899
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8265956415828044e-05,
-      "loss": 0.9062,
+      "loss": 0.948,
       "step": 5900
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8265293462330397e-05,
-      "loss": 0.812,
+      "loss": 0.9987,
       "step": 5901
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.826463039416353e-05,
-      "loss": 0.8592,
+      "loss": 0.9359,
       "step": 5902
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.826396721133664e-05,
-      "loss": 0.8409,
+      "loss": 0.9501,
       "step": 5903
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.826330391385893e-05,
-      "loss": 0.7174,
+      "loss": 0.8579,
       "step": 5904
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8262640501739603e-05,
-      "loss": 0.8039,
+      "loss": 0.9597,
       "step": 5905
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8261976974987862e-05,
-      "loss": 0.987,
+      "loss": 1.061,
       "step": 5906
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8261313333612915e-05,
-      "loss": 0.8199,
+      "loss": 0.973,
       "step": 5907
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8260649577623964e-05,
-      "loss": 0.9804,
+      "loss": 0.9957,
       "step": 5908
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8259985707030223e-05,
-      "loss": 0.8268,
+      "loss": 0.9323,
       "step": 5909
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8259321721840896e-05,
-      "loss": 0.8402,
+      "loss": 1.0203,
       "step": 5910
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8258657622065202e-05,
-      "loss": 0.8676,
+      "loss": 0.9302,
       "step": 5911
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8257993407712348e-05,
-      "loss": 0.8501,
+      "loss": 0.947,
       "step": 5912
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8257329078791555e-05,
-      "loss": 0.8932,
+      "loss": 0.9651,
       "step": 5913
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8256664635312033e-05,
-      "loss": 0.8569,
+      "loss": 0.9554,
       "step": 5914
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8256000077283004e-05,
-      "loss": 0.7629,
+      "loss": 0.8901,
       "step": 5915
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.825533540471369e-05,
-      "loss": 0.7417,
+      "loss": 0.8496,
       "step": 5916
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.825467061761331e-05,
-      "loss": 0.8341,
+      "loss": 0.8953,
       "step": 5917
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8254005715991088e-05,
-      "loss": 0.9651,
+      "loss": 1.0211,
       "step": 5918
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.825334069985625e-05,
-      "loss": 0.8171,
+      "loss": 0.8784,
       "step": 5919
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8252675569218013e-05,
-      "loss": 0.7675,
+      "loss": 0.8733,
       "step": 5920
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8252010324085615e-05,
-      "loss": 0.6616,
+      "loss": 0.7944,
       "step": 5921
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.825134496446828e-05,
-      "loss": 0.9472,
+      "loss": 1.0338,
       "step": 5922
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.825067949037524e-05,
-      "loss": 0.9088,
+      "loss": 0.9164,
       "step": 5923
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8250013901815733e-05,
-      "loss": 0.7828,
+      "loss": 0.8845,
       "step": 5924
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8249348198798985e-05,
-      "loss": 0.8898,
+      "loss": 0.9318,
       "step": 5925
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8248682381334237e-05,
-      "loss": 0.7778,
+      "loss": 0.9044,
       "step": 5926
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8248016449430724e-05,
-      "loss": 0.8573,
+      "loss": 0.9877,
       "step": 5927
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8247350403097685e-05,
-      "loss": 0.8457,
+      "loss": 0.9692,
       "step": 5928
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8246684242344358e-05,
-      "loss": 0.7222,
+      "loss": 0.8503,
       "step": 5929
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824601796717999e-05,
-      "loss": 0.8566,
+      "loss": 0.9847,
       "step": 5930
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8245351577613823e-05,
-      "loss": 0.929,
+      "loss": 0.9284,
       "step": 5931
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.82446850736551e-05,
-      "loss": 0.8682,
+      "loss": 0.9571,
       "step": 5932
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824401845531307e-05,
-      "loss": 0.8508,
+      "loss": 0.9652,
       "step": 5933
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824335172259698e-05,
-      "loss": 0.7778,
+      "loss": 0.9074,
       "step": 5934
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824268487551608e-05,
-      "loss": 0.8617,
+      "loss": 0.9319,
       "step": 5935
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8242017914079622e-05,
-      "loss": 0.7417,
+      "loss": 0.9007,
       "step": 5936
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824135083829686e-05,
-      "loss": 0.7094,
+      "loss": 0.895,
       "step": 5937
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.824068364817705e-05,
-      "loss": 0.8424,
+      "loss": 0.9222,
       "step": 5938
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8240016343729445e-05,
-      "loss": 0.8601,
+      "loss": 0.9625,
       "step": 5939
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8239348924963303e-05,
-      "loss": 0.7673,
+      "loss": 0.9405,
       "step": 5940
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8238681391887887e-05,
-      "loss": 0.9182,
+      "loss": 0.9686,
       "step": 5941
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8238013744512453e-05,
-      "loss": 0.9591,
+      "loss": 1.0249,
       "step": 5942
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.823734598284627e-05,
-      "loss": 0.7506,
+      "loss": 0.8855,
       "step": 5943
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8236678106898598e-05,
-      "loss": 0.8837,
+      "loss": 0.9767,
       "step": 5944
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.82360101166787e-05,
-      "loss": 0.8064,
+      "loss": 0.9042,
       "step": 5945
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.823534201219585e-05,
-      "loss": 0.9355,
+      "loss": 0.9917,
       "step": 5946
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8234673793459316e-05,
-      "loss": 0.873,
+      "loss": 0.9841,
       "step": 5947
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8234005460478365e-05,
-      "loss": 0.8678,
+      "loss": 0.9922,
       "step": 5948
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8233337013262268e-05,
-      "loss": 0.8384,
+      "loss": 0.9813,
       "step": 5949
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.82326684518203e-05,
-      "loss": 0.8042,
+      "loss": 0.8598,
       "step": 5950
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8231999776161743e-05,
-      "loss": 0.9022,
+      "loss": 0.8978,
       "step": 5951
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.823133098629587e-05,
-      "loss": 0.7128,
+      "loss": 0.8722,
       "step": 5952
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8230662082231955e-05,
-      "loss": 1.0698,
+      "loss": 1.1277,
       "step": 5953
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822999306397928e-05,
-      "loss": 0.8971,
+      "loss": 0.9586,
       "step": 5954
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822932393154713e-05,
-      "loss": 0.8311,
+      "loss": 1.0126,
       "step": 5955
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8228654684944788e-05,
-      "loss": 0.7305,
+      "loss": 0.8901,
       "step": 5956
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8227985324181533e-05,
-      "loss": 0.7705,
+      "loss": 0.9264,
       "step": 5957
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822731584926666e-05,
-      "loss": 0.8102,
+      "loss": 0.9145,
       "step": 5958
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822664626020945e-05,
-      "loss": 0.8154,
+      "loss": 0.9358,
       "step": 5959
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8225976557019196e-05,
-      "loss": 0.9065,
+      "loss": 1.0038,
       "step": 5960
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8225306739705185e-05,
-      "loss": 0.8063,
+      "loss": 0.9196,
       "step": 5961
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8224636808276718e-05,
-      "loss": 0.8301,
+      "loss": 0.8816,
       "step": 5962
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8223966762743083e-05,
-      "loss": 0.777,
+      "loss": 0.8753,
       "step": 5963
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822329660311358e-05,
-      "loss": 0.8139,
+      "loss": 0.8866,
       "step": 5964
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8222626329397498e-05,
-      "loss": 0.8117,
+      "loss": 0.9317,
       "step": 5965
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8221955941604145e-05,
-      "loss": 0.8039,
+      "loss": 0.9128,
       "step": 5966
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.822128543974282e-05,
-      "loss": 0.8559,
+      "loss": 0.9218,
       "step": 5967
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8220614823822823e-05,
-      "loss": 0.8456,
+      "loss": 0.9714,
       "step": 5968
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.821994409385346e-05,
-      "loss": 0.8787,
+      "loss": 0.9298,
       "step": 5969
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8219273249844036e-05,
-      "loss": 0.8139,
+      "loss": 0.8945,
       "step": 5970
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8218602291803854e-05,
-      "loss": 0.8773,
+      "loss": 0.9977,
       "step": 5971
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.821793121974223e-05,
-      "loss": 0.8152,
+      "loss": 0.8816,
       "step": 5972
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8217260033668465e-05,
-      "loss": 0.9669,
+      "loss": 0.9888,
       "step": 5973
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.821658873359188e-05,
-      "loss": 0.8591,
+      "loss": 0.9753,
       "step": 5974
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.821591731952178e-05,
-      "loss": 0.9075,
+      "loss": 1.0029,
       "step": 5975
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8215245791467485e-05,
-      "loss": 0.715,
+      "loss": 0.9045,
       "step": 5976
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8214574149438315e-05,
-      "loss": 0.6367,
+      "loss": 0.865,
       "step": 5977
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.821390239344358e-05,
-      "loss": 0.9354,
+      "loss": 0.9907,
       "step": 5978
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.82132305234926e-05,
-      "loss": 0.7819,
+      "loss": 0.8641,
       "step": 5979
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8212558539594706e-05,
-      "loss": 0.8386,
+      "loss": 0.9095,
       "step": 5980
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8211886441759215e-05,
-      "loss": 0.7258,
+      "loss": 0.8324,
       "step": 5981
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8211214229995447e-05,
-      "loss": 0.7614,
+      "loss": 0.8652,
       "step": 5982
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8210541904312734e-05,
-      "loss": 0.8771,
+      "loss": 1.0173,
       "step": 5983
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8209869464720397e-05,
-      "loss": 0.6754,
+      "loss": 0.8566,
       "step": 5984
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8209196911227773e-05,
-      "loss": 0.8992,
+      "loss": 0.9597,
       "step": 5985
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8208524243844186e-05,
-      "loss": 0.8965,
+      "loss": 0.9622,
       "step": 5986
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8207851462578975e-05,
-      "loss": 0.9428,
+      "loss": 1.038,
       "step": 5987
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.820717856744147e-05,
-      "loss": 0.8452,
+      "loss": 0.9695,
       "step": 5988
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8206505558441005e-05,
-      "loss": 0.7812,
+      "loss": 0.9396,
       "step": 5989
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8205832435586922e-05,
-      "loss": 0.8655,
+      "loss": 0.9067,
       "step": 5990
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8205159198888555e-05,
-      "loss": 0.866,
+      "loss": 0.9054,
       "step": 5991
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.820448584835524e-05,
-      "loss": 0.8778,
+      "loss": 0.9935,
       "step": 5992
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.820381238399633e-05,
-      "loss": 0.8427,
+      "loss": 0.9778,
       "step": 5993
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8203138805821165e-05,
-      "loss": 0.6915,
+      "loss": 0.9015,
       "step": 5994
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8202465113839085e-05,
-      "loss": 0.9738,
+      "loss": 0.9945,
       "step": 5995
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.820179130805944e-05,
-      "loss": 0.7716,
+      "loss": 0.9841,
       "step": 5996
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8201117388491577e-05,
-      "loss": 0.8381,
+      "loss": 0.888,
       "step": 5997
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8200443355144845e-05,
-      "loss": 0.8742,
+      "loss": 0.9106,
       "step": 5998
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8199769208028598e-05,
-      "loss": 0.9484,
+      "loss": 0.9583,
       "step": 5999
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8199094947152188e-05,
-      "loss": 0.8357,
+      "loss": 0.9545,
       "step": 6000
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8198420572524968e-05,
-      "loss": 0.8739,
+      "loss": 0.9107,
       "step": 6001
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.819774608415629e-05,
-      "loss": 0.7068,
+      "loss": 0.8986,
       "step": 6002
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8197071482055523e-05,
-      "loss": 0.7751,
+      "loss": 0.8756,
       "step": 6003
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8196396766232018e-05,
-      "loss": 0.9301,
+      "loss": 0.9861,
       "step": 6004
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8195721936695135e-05,
-      "loss": 0.8373,
+      "loss": 0.9387,
       "step": 6005
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8195046993454237e-05,
-      "loss": 0.7481,
+      "loss": 0.8717,
       "step": 6006
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8194371936518692e-05,
-      "loss": 0.8431,
+      "loss": 0.8916,
       "step": 6007
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.819369676589786e-05,
-      "loss": 0.8096,
+      "loss": 0.9013,
       "step": 6008
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8193021481601112e-05,
-      "loss": 0.9497,
+      "loss": 1.0537,
       "step": 6009
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8192346083637817e-05,
-      "loss": 0.7049,
+      "loss": 0.8695,
       "step": 6010
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.819167057201734e-05,
-      "loss": 0.7832,
+      "loss": 0.8928,
       "step": 6011
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.819099494674906e-05,
-      "loss": 0.7502,
+      "loss": 0.909,
       "step": 6012
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8190319207842346e-05,
-      "loss": 0.8088,
+      "loss": 0.9537,
       "step": 6013
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818964335530657e-05,
-      "loss": 0.7362,
+      "loss": 0.9698,
       "step": 6014
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8188967389151115e-05,
-      "loss": 0.8692,
+      "loss": 0.932,
       "step": 6015
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818829130938536e-05,
-      "loss": 0.8162,
+      "loss": 0.8878,
       "step": 6016
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818761511601867e-05,
-      "loss": 0.9096,
+      "loss": 0.949,
       "step": 6017
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8186938809060448e-05,
-      "loss": 0.8808,
+      "loss": 0.9705,
       "step": 6018
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818626238852006e-05,
-      "loss": 0.7995,
+      "loss": 0.9296,
       "step": 6019
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8185585854406897e-05,
-      "loss": 0.8378,
+      "loss": 0.908,
       "step": 6020
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8184909206730345e-05,
-      "loss": 0.723,
+      "loss": 0.9345,
       "step": 6021
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8184232445499786e-05,
-      "loss": 0.8375,
+      "loss": 0.9742,
       "step": 6022
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818355557072462e-05,
-      "loss": 0.8613,
+      "loss": 0.9297,
       "step": 6023
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8182878582414232e-05,
-      "loss": 0.7085,
+      "loss": 0.8115,
       "step": 6024
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818220148057801e-05,
-      "loss": 0.8223,
+      "loss": 0.9032,
       "step": 6025
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8181524265225353e-05,
-      "loss": 0.8691,
+      "loss": 0.9667,
       "step": 6026
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8180846936365653e-05,
-      "loss": 0.8352,
+      "loss": 1.0799,
       "step": 6027
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.818016949400831e-05,
-      "loss": 0.7475,
+      "loss": 0.8405,
       "step": 6028
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.817949193816272e-05,
-      "loss": 0.8157,
+      "loss": 0.8844,
       "step": 6029
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8178814268838287e-05,
-      "loss": 0.9258,
+      "loss": 1.0352,
       "step": 6030
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.817813648604441e-05,
-      "loss": 0.797,
+      "loss": 0.9209,
       "step": 6031
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.817745858979049e-05,
-      "loss": 0.8719,
+      "loss": 0.9725,
       "step": 6032
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8176780580085937e-05,
-      "loss": 0.8369,
+      "loss": 0.9944,
       "step": 6033
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8176102456940156e-05,
-      "loss": 0.9316,
+      "loss": 1.0235,
       "step": 6034
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.817542422036255e-05,
-      "loss": 0.8211,
+      "loss": 0.9048,
       "step": 6035
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8174745870362532e-05,
-      "loss": 0.919,
+      "loss": 1.0622,
       "step": 6036
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8174067406949518e-05,
-      "loss": 0.7618,
+      "loss": 0.8679,
       "step": 6037
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8173388830132914e-05,
-      "loss": 0.886,
+      "loss": 0.9253,
       "step": 6038
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8172710139922135e-05,
-      "loss": 0.7608,
+      "loss": 0.9229,
       "step": 6039
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8172031336326598e-05,
-      "loss": 0.918,
+      "loss": 1.0376,
       "step": 6040
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8171352419355725e-05,
-      "loss": 0.7706,
+      "loss": 0.9479,
       "step": 6041
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8170673389018927e-05,
-      "loss": 0.7451,
+      "loss": 0.9159,
       "step": 6042
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.816999424532563e-05,
-      "loss": 0.7796,
+      "loss": 0.9577,
       "step": 6043
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8169314988285257e-05,
-      "loss": 0.8112,
+      "loss": 0.9052,
       "step": 6044
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8168635617907225e-05,
-      "loss": 0.7836,
+      "loss": 0.8705,
       "step": 6045
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8167956134200965e-05,
-      "loss": 0.9733,
+      "loss": 0.9975,
       "step": 6046
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.81672765371759e-05,
-      "loss": 0.8272,
+      "loss": 0.9662,
       "step": 6047
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8166596826841466e-05,
-      "loss": 0.7151,
+      "loss": 0.8864,
       "step": 6048
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8165917003207085e-05,
-      "loss": 0.8416,
+      "loss": 0.9404,
       "step": 6049
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.816523706628219e-05,
-      "loss": 0.9602,
+      "loss": 1.0303,
       "step": 6050
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.816455701607622e-05,
-      "loss": 0.7502,
+      "loss": 0.9111,
       "step": 6051
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8163876852598603e-05,
-      "loss": 0.8302,
+      "loss": 0.8839,
       "step": 6052
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8163196575858776e-05,
-      "loss": 0.8597,
+      "loss": 0.9709,
       "step": 6053
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.816251618586618e-05,
-      "loss": 0.9086,
+      "loss": 0.9615,
       "step": 6054
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8161835682630254e-05,
-      "loss": 0.6893,
+      "loss": 0.8348,
       "step": 6055
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8161155066160437e-05,
-      "loss": 0.6706,
+      "loss": 0.8121,
       "step": 6056
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.816047433646617e-05,
-      "loss": 0.8797,
+      "loss": 0.9628,
       "step": 6057
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8159793493556902e-05,
-      "loss": 0.7572,
+      "loss": 0.9898,
       "step": 6058
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.815911253744208e-05,
-      "loss": 0.8907,
+      "loss": 0.9834,
       "step": 6059
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8158431468131143e-05,
-      "loss": 0.9014,
+      "loss": 0.9643,
       "step": 6060
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8157750285633548e-05,
-      "loss": 0.828,
+      "loss": 0.9395,
       "step": 6061
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.815706898995874e-05,
-      "loss": 0.7517,
+      "loss": 0.8572,
       "step": 6062
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8156387581116175e-05,
-      "loss": 0.8809,
+      "loss": 0.9745,
       "step": 6063
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8155706059115303e-05,
-      "loss": 0.8222,
+      "loss": 0.8795,
       "step": 6064
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.815502442396558e-05,
-      "loss": 0.621,
+      "loss": 0.8319,
       "step": 6065
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8154342675676466e-05,
-      "loss": 0.9598,
+      "loss": 0.9393,
       "step": 6066
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8153660814257417e-05,
-      "loss": 0.823,
+      "loss": 0.9234,
       "step": 6067
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8152978839717893e-05,
-      "loss": 0.814,
+      "loss": 0.943,
       "step": 6068
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8152296752067352e-05,
-      "loss": 0.8428,
+      "loss": 0.8975,
       "step": 6069
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8151614551315262e-05,
-      "loss": 0.809,
+      "loss": 0.8561,
       "step": 6070
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8150932237471088e-05,
-      "loss": 0.8279,
+      "loss": 0.9154,
       "step": 6071
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8150249810544288e-05,
-      "loss": 0.8445,
+      "loss": 0.9491,
       "step": 6072
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8149567270544342e-05,
-      "loss": 0.7988,
+      "loss": 0.9152,
       "step": 6073
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8148884617480708e-05,
-      "loss": 0.8186,
+      "loss": 0.9755,
       "step": 6074
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8148201851362864e-05,
-      "loss": 0.7533,
+      "loss": 0.8638,
       "step": 6075
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.814751897220028e-05,
-      "loss": 0.8765,
+      "loss": 0.9437,
       "step": 6076
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.814683598000243e-05,
-      "loss": 0.8488,
+      "loss": 0.9093,
       "step": 6077
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.814615287477879e-05,
-      "loss": 0.9931,
+      "loss": 1.0828,
       "step": 6078
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8145469656538835e-05,
-      "loss": 0.7988,
+      "loss": 0.8593,
       "step": 6079
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8144786325292047e-05,
-      "loss": 0.7291,
+      "loss": 0.8377,
       "step": 6080
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8144102881047903e-05,
-      "loss": 0.7702,
+      "loss": 0.8966,
       "step": 6081
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8143419323815886e-05,
-      "loss": 0.7229,
+      "loss": 0.8192,
       "step": 6082
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8142735653605483e-05,
-      "loss": 0.8301,
+      "loss": 0.9187,
       "step": 6083
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8142051870426177e-05,
-      "loss": 0.7228,
+      "loss": 0.8562,
       "step": 6084
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.814136797428745e-05,
-      "loss": 0.8307,
+      "loss": 0.8583,
       "step": 6085
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8140683965198794e-05,
-      "loss": 0.8118,
+      "loss": 0.8953,
       "step": 6086
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8139999843169697e-05,
-      "loss": 0.9406,
+      "loss": 0.9801,
       "step": 6087
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8139315608209655e-05,
-      "loss": 0.801,
+      "loss": 0.9464,
       "step": 6088
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8138631260328154e-05,
-      "loss": 0.8596,
+      "loss": 0.9556,
       "step": 6089
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8137946799534694e-05,
-      "loss": 0.7937,
+      "loss": 0.9551,
       "step": 6090
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8137262225838766e-05,
-      "loss": 0.8513,
+      "loss": 0.984,
       "step": 6091
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.813657753924987e-05,
-      "loss": 0.8766,
+      "loss": 0.9612,
       "step": 6092
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8135892739777504e-05,
-      "loss": 0.7422,
+      "loss": 0.952,
       "step": 6093
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8135207827431173e-05,
-      "loss": 0.8034,
+      "loss": 0.9999,
       "step": 6094
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8134522802220372e-05,
-      "loss": 0.8528,
+      "loss": 0.923,
       "step": 6095
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.813383766415461e-05,
-      "loss": 0.9539,
+      "loss": 0.9866,
       "step": 6096
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.813315241324339e-05,
-      "loss": 0.6067,
+      "loss": 0.8171,
       "step": 6097
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8132467049496224e-05,
-      "loss": 0.6787,
+      "loss": 0.8143,
       "step": 6098
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8131781572922612e-05,
-      "loss": 0.8813,
+      "loss": 0.8788,
       "step": 6099
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.813109598353207e-05,
-      "loss": 0.9086,
+      "loss": 0.9931,
       "step": 6100
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8130410281334105e-05,
-      "loss": 0.883,
+      "loss": 0.9975,
       "step": 6101
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8129724466338237e-05,
-      "loss": 0.8076,
+      "loss": 0.9229,
       "step": 6102
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8129038538553972e-05,
-      "loss": 0.7727,
+      "loss": 0.8841,
       "step": 6103
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8128352497990835e-05,
-      "loss": 0.831,
+      "loss": 0.9454,
       "step": 6104
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8127666344658336e-05,
-      "loss": 0.8478,
+      "loss": 0.9363,
       "step": 6105
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8126980078566002e-05,
-      "loss": 0.8022,
+      "loss": 0.9106,
       "step": 6106
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.812629369972335e-05,
-      "loss": 0.9008,
+      "loss": 0.9186,
       "step": 6107
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.81256072081399e-05,
-      "loss": 0.9123,
+      "loss": 0.9605,
       "step": 6108
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.812492060382518e-05,
-      "loss": 0.8959,
+      "loss": 0.996,
       "step": 6109
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8124233886788715e-05,
-      "loss": 0.7505,
+      "loss": 0.8467,
       "step": 6110
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.812354705704003e-05,
-      "loss": 0.7604,
+      "loss": 0.9413,
       "step": 6111
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8122860114588653e-05,
-      "loss": 0.801,
+      "loss": 0.8821,
       "step": 6112
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8122173059444122e-05,
-      "loss": 0.8762,
+      "loss": 0.9203,
       "step": 6113
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.812148589161596e-05,
-      "loss": 0.8181,
+      "loss": 0.8601,
       "step": 6114
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8120798611113704e-05,
-      "loss": 0.6072,
+      "loss": 0.8002,
       "step": 6115
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8120111217946893e-05,
-      "loss": 0.867,
+      "loss": 0.9893,
       "step": 6116
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8119423712125055e-05,
-      "loss": 0.743,
+      "loss": 0.8512,
       "step": 6117
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8118736093657736e-05,
-      "loss": 0.758,
+      "loss": 0.9175,
       "step": 6118
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8118048362554472e-05,
-      "loss": 0.7553,
+      "loss": 0.9022,
       "step": 6119
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8117360518824805e-05,
-      "loss": 0.748,
+      "loss": 0.8749,
       "step": 6120
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.811667256247828e-05,
-      "loss": 0.859,
+      "loss": 0.9403,
       "step": 6121
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8115984493524435e-05,
-      "loss": 0.8052,
+      "loss": 0.8769,
       "step": 6122
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8115296311972824e-05,
-      "loss": 0.8858,
+      "loss": 1.0461,
       "step": 6123
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8114608017832987e-05,
-      "loss": 0.9463,
+      "loss": 0.9548,
       "step": 6124
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.811391961111448e-05,
-      "loss": 0.8814,
+      "loss": 1.0075,
       "step": 6125
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8113231091826848e-05,
-      "loss": 0.7035,
+      "loss": 0.8412,
       "step": 6126
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8112542459979648e-05,
-      "loss": 0.8393,
+      "loss": 0.9326,
       "step": 6127
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.811185371558243e-05,
-      "loss": 0.8411,
+      "loss": 0.9734,
       "step": 6128
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.811116485864475e-05,
-      "loss": 0.9539,
+      "loss": 1.053,
       "step": 6129
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8110475889176167e-05,
-      "loss": 0.7343,
+      "loss": 0.8431,
       "step": 6130
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8109786807186243e-05,
-      "loss": 0.8764,
+      "loss": 1.0203,
       "step": 6131
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.810909761268453e-05,
-      "loss": 0.8824,
+      "loss": 0.9823,
       "step": 6132
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.810840830568059e-05,
-      "loss": 0.8349,
+      "loss": 1.0057,
       "step": 6133
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.810771888618399e-05,
-      "loss": 0.9204,
+      "loss": 1.0068,
       "step": 6134
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8107029354204297e-05,
-      "loss": 0.8711,
+      "loss": 0.9324,
       "step": 6135
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.810633970975107e-05,
-      "loss": 0.8083,
+      "loss": 0.9232,
       "step": 6136
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8105649952833883e-05,
-      "loss": 0.8412,
+      "loss": 0.8799,
       "step": 6137
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8104960083462308e-05,
-      "loss": 0.8394,
+      "loss": 0.9933,
       "step": 6138
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8104270101645905e-05,
-      "loss": 0.9514,
+      "loss": 0.9883,
       "step": 6139
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.810358000739425e-05,
-      "loss": 0.8538,
+      "loss": 0.9193,
       "step": 6140
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8102889800716927e-05,
-      "loss": 0.7935,
+      "loss": 0.9487,
       "step": 6141
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.81021994816235e-05,
-      "loss": 0.766,
+      "loss": 0.8644,
       "step": 6142
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8101509050123555e-05,
-      "loss": 0.7587,
+      "loss": 0.9481,
       "step": 6143
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8100818506226663e-05,
-      "loss": 0.7217,
+      "loss": 0.8444,
       "step": 6144
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8100127849942407e-05,
-      "loss": 0.7892,
+      "loss": 0.9188,
       "step": 6145
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8099437081280373e-05,
-      "loss": 0.8143,
+      "loss": 0.8995,
       "step": 6146
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8098746200250138e-05,
-      "loss": 0.8507,
+      "loss": 0.914,
       "step": 6147
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8098055206861294e-05,
-      "loss": 0.8467,
+      "loss": 0.9232,
       "step": 6148
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.809736410112342e-05,
-      "loss": 0.8294,
+      "loss": 0.9251,
       "step": 6149
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.809667288304611e-05,
-      "loss": 0.8955,
+      "loss": 0.9527,
       "step": 6150
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.809598155263895e-05,
-      "loss": 0.8207,
+      "loss": 0.9942,
       "step": 6151
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8095290109911533e-05,
-      "loss": 0.7375,
+      "loss": 0.8612,
       "step": 6152
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.809459855487345e-05,
-      "loss": 0.6811,
+      "loss": 0.888,
       "step": 6153
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.80939068875343e-05,
-      "loss": 0.7597,
+      "loss": 0.9397,
       "step": 6154
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8093215107903672e-05,
-      "loss": 0.7474,
+      "loss": 0.8106,
       "step": 6155
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8092523215991172e-05,
-      "loss": 0.7286,
+      "loss": 0.9457,
       "step": 6156
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.809183121180639e-05,
-      "loss": 0.9243,
+      "loss": 1.0068,
       "step": 6157
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8091139095358932e-05,
-      "loss": 0.8046,
+      "loss": 0.8959,
       "step": 6158
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.80904468666584e-05,
-      "loss": 0.9586,
+      "loss": 1.0299,
       "step": 6159
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8089754525714392e-05,
-      "loss": 0.7982,
+      "loss": 0.8293,
       "step": 6160
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8089062072536526e-05,
-      "loss": 0.916,
+      "loss": 0.9349,
       "step": 6161
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8088369507134394e-05,
-      "loss": 0.9992,
+      "loss": 1.0861,
       "step": 6162
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8087676829517614e-05,
-      "loss": 0.8959,
+      "loss": 1.0354,
       "step": 6163
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.808698403969579e-05,
-      "loss": 0.8105,
+      "loss": 0.9399,
       "step": 6164
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8086291137678536e-05,
-      "loss": 0.9173,
+      "loss": 1.0045,
       "step": 6165
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.808559812347547e-05,
-      "loss": 0.7797,
+      "loss": 0.8483,
       "step": 6166
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.80849049970962e-05,
-      "loss": 0.6609,
+      "loss": 0.9157,
       "step": 6167
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8084211758550343e-05,
-      "loss": 0.7871,
+      "loss": 0.915,
       "step": 6168
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.808351840784752e-05,
-      "loss": 0.7722,
+      "loss": 0.9597,
       "step": 6169
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8082824944997344e-05,
-      "loss": 0.8262,
+      "loss": 0.9354,
       "step": 6170
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8082131370009442e-05,
-      "loss": 0.7475,
+      "loss": 0.9262,
       "step": 6171
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8081437682893436e-05,
-      "loss": 0.7343,
+      "loss": 0.8978,
       "step": 6172
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8080743883658945e-05,
-      "loss": 0.9024,
+      "loss": 0.9344,
       "step": 6173
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8080049972315602e-05,
-      "loss": 0.9485,
+      "loss": 1.0688,
       "step": 6174
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.807935594887303e-05,
-      "loss": 0.7901,
+      "loss": 0.9303,
       "step": 6175
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8078661813340853e-05,
-      "loss": 0.784,
+      "loss": 0.8669,
       "step": 6176
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.807796756572871e-05,
-      "loss": 0.8156,
+      "loss": 0.9521,
       "step": 6177
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8077273206046223e-05,
-      "loss": 0.8028,
+      "loss": 0.8748,
       "step": 6178
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8076578734303032e-05,
-      "loss": 0.8289,
+      "loss": 0.8816,
       "step": 6179
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8075884150508773e-05,
-      "loss": 0.8483,
+      "loss": 0.9741,
       "step": 6180
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8075189454673075e-05,
-      "loss": 0.8009,
+      "loss": 0.8968,
       "step": 6181
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8074494646805585e-05,
-      "loss": 0.8501,
+      "loss": 0.9861,
       "step": 6182
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8073799726915937e-05,
-      "loss": 0.8463,
+      "loss": 0.9675,
       "step": 6183
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.807310469501377e-05,
-      "loss": 0.8441,
+      "loss": 0.9727,
       "step": 6184
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8072409551108734e-05,
-      "loss": 0.9533,
+      "loss": 0.9982,
       "step": 6185
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8071714295210464e-05,
-      "loss": 0.8641,
+      "loss": 0.9824,
       "step": 6186
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8071018927328614e-05,
-      "loss": 0.9232,
+      "loss": 0.964,
       "step": 6187
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8070323447472827e-05,
-      "loss": 0.9168,
+      "loss": 0.9583,
       "step": 6188
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8069627855652752e-05,
-      "loss": 0.7698,
+      "loss": 0.8713,
       "step": 6189
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806893215187804e-05,
-      "loss": 0.8597,
+      "loss": 0.9612,
       "step": 6190
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8068236336158343e-05,
-      "loss": 0.6769,
+      "loss": 0.897,
       "step": 6191
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806754040850331e-05,
-      "loss": 0.7399,
+      "loss": 0.8859,
       "step": 6192
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8066844368922606e-05,
-      "loss": 0.9766,
+      "loss": 1.0581,
       "step": 6193
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806614821742588e-05,
-      "loss": 0.8927,
+      "loss": 0.9616,
       "step": 6194
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806545195402279e-05,
-      "loss": 0.8262,
+      "loss": 0.9885,
       "step": 6195
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8064755578723e-05,
-      "loss": 0.6992,
+      "loss": 0.8985,
       "step": 6196
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8064059091536167e-05,
-      "loss": 0.7061,
+      "loss": 0.8925,
       "step": 6197
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806336249247196e-05,
-      "loss": 0.8352,
+      "loss": 0.912,
       "step": 6198
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8062665781540034e-05,
-      "loss": 0.7897,
+      "loss": 0.9338,
       "step": 6199
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806196895875006e-05,
-      "loss": 0.6945,
+      "loss": 0.8489,
       "step": 6200
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8061272024111707e-05,
-      "loss": 0.8689,
+      "loss": 0.9701,
       "step": 6201
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.806057497763464e-05,
-      "loss": 0.9565,
+      "loss": 0.9911,
       "step": 6202
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8059877819328533e-05,
-      "loss": 0.8017,
+      "loss": 0.9416,
       "step": 6203
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8059180549203054e-05,
-      "loss": 0.8007,
+      "loss": 0.9449,
       "step": 6204
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8058483167267886e-05,
-      "loss": 1.0932,
+      "loss": 1.1021,
       "step": 6205
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.805778567353269e-05,
-      "loss": 0.8997,
+      "loss": 0.9611,
       "step": 6206
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8057088068007155e-05,
-      "loss": 0.7507,
+      "loss": 0.9285,
       "step": 6207
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8056390350700952e-05,
-      "loss": 0.8206,
+      "loss": 0.9203,
       "step": 6208
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8055692521623765e-05,
-      "loss": 0.8463,
+      "loss": 1.0095,
       "step": 6209
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8054994580785273e-05,
-      "loss": 0.8141,
+      "loss": 0.9225,
       "step": 6210
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.805429652819516e-05,
-      "loss": 0.8083,
+      "loss": 0.8808,
       "step": 6211
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.805359836386311e-05,
-      "loss": 0.663,
+      "loss": 0.8636,
       "step": 6212
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8052900087798812e-05,
-      "loss": 0.8233,
+      "loss": 0.9195,
       "step": 6213
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.805220170001195e-05,
-      "loss": 0.8872,
+      "loss": 0.9803,
       "step": 6214
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.805150320051221e-05,
-      "loss": 0.8973,
+      "loss": 0.9009,
       "step": 6215
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8050804589309286e-05,
-      "loss": 0.6906,
+      "loss": 0.8608,
       "step": 6216
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8050105866412877e-05,
-      "loss": 0.8191,
+      "loss": 0.9213,
       "step": 6217
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8049407031832667e-05,
-      "loss": 0.8007,
+      "loss": 0.9692,
       "step": 6218
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804870808557836e-05,
-      "loss": 0.6937,
+      "loss": 0.8688,
       "step": 6219
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804800902765964e-05,
-      "loss": 0.8578,
+      "loss": 0.8869,
       "step": 6220
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8047309858086216e-05,
-      "loss": 0.7158,
+      "loss": 0.8555,
       "step": 6221
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804661057686779e-05,
-      "loss": 0.7416,
+      "loss": 0.9669,
       "step": 6222
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804591118401405e-05,
-      "loss": 0.8746,
+      "loss": 0.9903,
       "step": 6223
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8045211679534714e-05,
-      "loss": 0.827,
+      "loss": 0.9876,
       "step": 6224
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804451206343948e-05,
-      "loss": 0.8236,
+      "loss": 0.8835,
       "step": 6225
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8043812335738054e-05,
-      "loss": 0.7436,
+      "loss": 0.9047,
       "step": 6226
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804311249644015e-05,
-      "loss": 0.8104,
+      "loss": 0.8932,
       "step": 6227
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8042412545555464e-05,
-      "loss": 0.6106,
+      "loss": 0.8204,
       "step": 6228
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804171248309372e-05,
-      "loss": 0.8578,
+      "loss": 0.9715,
       "step": 6229
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.804101230906462e-05,
-      "loss": 0.8602,
+      "loss": 0.9458,
       "step": 6230
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8040312023477887e-05,
-      "loss": 0.8103,
+      "loss": 0.9458,
       "step": 6231
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8039611626343232e-05,
-      "loss": 0.7435,
+      "loss": 0.9193,
       "step": 6232
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.803891111767037e-05,
-      "loss": 0.8261,
+      "loss": 1.0204,
       "step": 6233
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8038210497469023e-05,
-      "loss": 0.7701,
+      "loss": 0.8755,
       "step": 6234
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8037509765748914e-05,
-      "loss": 0.7515,
+      "loss": 0.8526,
       "step": 6235
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8036808922519756e-05,
-      "loss": 0.9587,
+      "loss": 0.971,
       "step": 6236
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8036107967791278e-05,
-      "loss": 0.8146,
+      "loss": 0.9125,
       "step": 6237
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8035406901573202e-05,
-      "loss": 0.7709,
+      "loss": 0.9334,
       "step": 6238
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.803470572387526e-05,
-      "loss": 0.7618,
+      "loss": 0.9225,
       "step": 6239
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.803400443470717e-05,
-      "loss": 0.8419,
+      "loss": 0.9945,
       "step": 6240
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8033303034078674e-05,
-      "loss": 0.9162,
+      "loss": 0.9099,
       "step": 6241
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.803260152199949e-05,
-      "loss": 0.7386,
+      "loss": 0.8206,
       "step": 6242
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.803189989847936e-05,
-      "loss": 0.7737,
+      "loss": 0.8945,
       "step": 6243
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8031198163528017e-05,
-      "loss": 0.831,
+      "loss": 0.8907,
       "step": 6244
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8030496317155192e-05,
-      "loss": 0.8967,
+      "loss": 0.9438,
       "step": 6245
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8029794359370623e-05,
-      "loss": 0.8066,
+      "loss": 0.9223,
       "step": 6246
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802909229018405e-05,
-      "loss": 0.7826,
+      "loss": 0.922,
       "step": 6247
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8028390109605217e-05,
-      "loss": 0.7588,
+      "loss": 0.8807,
       "step": 6248
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8027687817643856e-05,
-      "loss": 0.8042,
+      "loss": 0.9127,
       "step": 6249
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802698541430972e-05,
-      "loss": 0.7356,
+      "loss": 0.8987,
       "step": 6250
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802628289961255e-05,
-      "loss": 0.7283,
+      "loss": 0.9292,
       "step": 6251
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802558027356209e-05,
-      "loss": 0.8259,
+      "loss": 0.8923,
       "step": 6252
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8024877536168096e-05,
-      "loss": 0.7807,
+      "loss": 0.8998,
       "step": 6253
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802417468744031e-05,
-      "loss": 0.9233,
+      "loss": 0.9865,
       "step": 6254
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8023471727388485e-05,
-      "loss": 0.7446,
+      "loss": 0.7615,
       "step": 6255
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.802276865602237e-05,
-      "loss": 0.8488,
+      "loss": 1.0043,
       "step": 6256
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8022065473351724e-05,
-      "loss": 0.9488,
+      "loss": 1.0108,
       "step": 6257
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8021362179386304e-05,
-      "loss": 0.8228,
+      "loss": 0.8994,
       "step": 6258
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8020658774135866e-05,
-      "loss": 0.7157,
+      "loss": 0.8804,
       "step": 6259
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8019955257610164e-05,
-      "loss": 0.9668,
+      "loss": 0.9822,
       "step": 6260
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.801925162981896e-05,
-      "loss": 0.8011,
+      "loss": 0.9215,
       "step": 6261
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.801854789077202e-05,
-      "loss": 0.7883,
+      "loss": 0.8464,
       "step": 6262
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8017844040479104e-05,
-      "loss": 0.8424,
+      "loss": 0.9415,
       "step": 6263
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8017140078949976e-05,
-      "loss": 0.9208,
+      "loss": 0.9407,
       "step": 6264
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8016436006194407e-05,
-      "loss": 0.8197,
+      "loss": 0.8951,
       "step": 6265
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8015731822222165e-05,
-      "loss": 0.7958,
+      "loss": 0.8388,
       "step": 6266
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8015027527043016e-05,
-      "loss": 0.8162,
+      "loss": 0.9384,
       "step": 6267
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8014323120666726e-05,
-      "loss": 0.8315,
+      "loss": 0.9826,
       "step": 6268
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8013618603103077e-05,
-      "loss": 0.7,
+      "loss": 0.937,
       "step": 6269
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.801291397436184e-05,
-      "loss": 0.8826,
+      "loss": 1.0179,
       "step": 6270
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.801220923445279e-05,
-      "loss": 0.8443,
+      "loss": 0.9403,
       "step": 6271
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8011504383385706e-05,
-      "loss": 0.7604,
+      "loss": 0.7935,
       "step": 6272
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8010799421170363e-05,
-      "loss": 0.8257,
+      "loss": 0.9075,
       "step": 6273
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8010094347816545e-05,
-      "loss": 0.8268,
+      "loss": 0.9557,
       "step": 6274
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8009389163334035e-05,
-      "loss": 0.7534,
+      "loss": 0.9248,
       "step": 6275
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.800868386773261e-05,
-      "loss": 0.7604,
+      "loss": 0.9017,
       "step": 6276
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8007978461022063e-05,
-      "loss": 0.691,
+      "loss": 0.7872,
       "step": 6277
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8007272943212176e-05,
-      "loss": 0.8295,
+      "loss": 0.8765,
       "step": 6278
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8006567314312735e-05,
-      "loss": 0.9789,
+      "loss": 0.9951,
       "step": 6279
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8005861574333536e-05,
-      "loss": 0.8326,
+      "loss": 0.9317,
       "step": 6280
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8005155723284363e-05,
-      "loss": 0.8908,
+      "loss": 0.9685,
       "step": 6281
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8004449761175015e-05,
-      "loss": 0.8075,
+      "loss": 0.9764,
       "step": 6282
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8003743688015284e-05,
-      "loss": 0.7768,
+      "loss": 0.8844,
       "step": 6283
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.800303750381496e-05,
-      "loss": 0.9738,
+      "loss": 1.0614,
       "step": 6284
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.800233120858385e-05,
-      "loss": 0.7168,
+      "loss": 0.8326,
       "step": 6285
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8001624802331745e-05,
-      "loss": 0.8057,
+      "loss": 0.9416,
       "step": 6286
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8000918285068452e-05,
-      "loss": 0.9277,
+      "loss": 1.0386,
       "step": 6287
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8000211656803768e-05,
-      "loss": 0.8633,
+      "loss": 0.9004,
       "step": 6288
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79995049175475e-05,
-      "loss": 0.9333,
+      "loss": 0.9655,
       "step": 6289
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.799879806730945e-05,
-      "loss": 0.9594,
+      "loss": 0.998,
       "step": 6290
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7998091106099423e-05,
-      "loss": 0.6733,
+      "loss": 0.7988,
       "step": 6291
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7997384033927235e-05,
-      "loss": 0.8359,
+      "loss": 0.8711,
       "step": 6292
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7996676850802685e-05,
-      "loss": 0.97,
+      "loss": 0.9995,
       "step": 6293
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7995969556735593e-05,
-      "loss": 0.8582,
+      "loss": 0.9686,
       "step": 6294
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7995262151735767e-05,
-      "loss": 0.9024,
+      "loss": 0.9786,
       "step": 6295
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7994554635813023e-05,
-      "loss": 0.6308,
+      "loss": 0.8459,
       "step": 6296
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7993847008977176e-05,
-      "loss": 0.7401,
+      "loss": 0.9518,
       "step": 6297
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.799313927123804e-05,
-      "loss": 0.901,
+      "loss": 1.0123,
       "step": 6298
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.799243142260544e-05,
-      "loss": 0.8663,
+      "loss": 0.9622,
       "step": 6299
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7991723463089194e-05,
-      "loss": 0.7368,
+      "loss": 0.9122,
       "step": 6300
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7991015392699127e-05,
-      "loss": 0.6328,
+      "loss": 0.8106,
       "step": 6301
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7990307211445057e-05,
-      "loss": 0.7339,
+      "loss": 0.8493,
       "step": 6302
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7989598919336807e-05,
-      "loss": 0.8701,
+      "loss": 0.9294,
       "step": 6303
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7988890516384213e-05,
-      "loss": 1.0082,
+      "loss": 1.0609,
       "step": 6304
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7988182002597097e-05,
-      "loss": 0.8351,
+      "loss": 0.9565,
       "step": 6305
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7987473377985288e-05,
-      "loss": 0.7499,
+      "loss": 0.8619,
       "step": 6306
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7986764642558618e-05,
-      "loss": 0.7714,
+      "loss": 0.9293,
       "step": 6307
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7986055796326923e-05,
-      "loss": 0.8219,
+      "loss": 0.9289,
       "step": 6308
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7985346839300032e-05,
-      "loss": 0.7868,
+      "loss": 0.8816,
       "step": 6309
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7984637771487785e-05,
-      "loss": 0.7744,
+      "loss": 0.8629,
       "step": 6310
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7983928592900016e-05,
-      "loss": 0.8754,
+      "loss": 0.9771,
       "step": 6311
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7983219303546565e-05,
-      "loss": 0.9137,
+      "loss": 0.9328,
       "step": 6312
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7982509903437275e-05,
-      "loss": 0.7735,
+      "loss": 0.9433,
       "step": 6313
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7981800392581988e-05,
-      "loss": 0.8816,
+      "loss": 0.9276,
       "step": 6314
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.798109077099054e-05,
-      "loss": 0.7866,
+      "loss": 0.9131,
       "step": 6315
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7980381038672786e-05,
-      "loss": 0.7766,
+      "loss": 0.8765,
       "step": 6316
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7979671195638564e-05,
-      "loss": 0.9153,
+      "loss": 0.9964,
       "step": 6317
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7978961241897727e-05,
-      "loss": 0.7546,
+      "loss": 0.8458,
       "step": 6318
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7978251177460124e-05,
-      "loss": 0.8308,
+      "loss": 0.9393,
       "step": 6319
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7977541002335606e-05,
-      "loss": 0.8472,
+      "loss": 0.9273,
       "step": 6320
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7976830716534023e-05,
-      "loss": 0.8945,
+      "loss": 0.9589,
       "step": 6321
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7976120320065233e-05,
-      "loss": 0.8788,
+      "loss": 0.9991,
       "step": 6322
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.797540981293909e-05,
-      "loss": 0.8104,
+      "loss": 0.9267,
       "step": 6323
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7974699195165454e-05,
-      "loss": 0.9061,
+      "loss": 1.0008,
       "step": 6324
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7973988466754177e-05,
-      "loss": 0.73,
+      "loss": 0.9197,
       "step": 6325
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7973277627715128e-05,
-      "loss": 0.7266,
+      "loss": 0.8495,
       "step": 6326
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.797256667805816e-05,
-      "loss": 0.8774,
+      "loss": 0.9672,
       "step": 6327
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.797185561779314e-05,
-      "loss": 0.7457,
+      "loss": 0.8561,
       "step": 6328
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.797114444692994e-05,
-      "loss": 0.8513,
+      "loss": 0.9477,
       "step": 6329
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7970433165478417e-05,
-      "loss": 0.9624,
+      "loss": 0.9911,
       "step": 6330
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7969721773448443e-05,
-      "loss": 0.9085,
+      "loss": 0.9267,
       "step": 6331
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7969010270849885e-05,
-      "loss": 0.7649,
+      "loss": 0.911,
       "step": 6332
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.796829865769262e-05,
-      "loss": 0.858,
+      "loss": 0.9772,
       "step": 6333
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7967586933986513e-05,
-      "loss": 0.735,
+      "loss": 0.8134,
       "step": 6334
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7966875099741444e-05,
-      "loss": 0.8365,
+      "loss": 0.9364,
       "step": 6335
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7966163154967283e-05,
-      "loss": 0.9087,
+      "loss": 0.9854,
       "step": 6336
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7965451099673913e-05,
-      "loss": 0.7955,
+      "loss": 0.9343,
       "step": 6337
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7964738933871214e-05,
-      "loss": 0.9208,
+      "loss": 0.9999,
       "step": 6338
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.796402665756906e-05,
-      "loss": 0.9306,
+      "loss": 1.0309,
       "step": 6339
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7963314270777334e-05,
-      "loss": 0.7594,
+      "loss": 0.8998,
       "step": 6340
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7962601773505923e-05,
-      "loss": 0.7261,
+      "loss": 0.8403,
       "step": 6341
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.796188916576471e-05,
-      "loss": 0.9822,
+      "loss": 1.0116,
       "step": 6342
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.796117644756358e-05,
-      "loss": 0.6385,
+      "loss": 0.8247,
       "step": 6343
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.796046361891242e-05,
-      "loss": 0.8256,
+      "loss": 0.9451,
       "step": 6344
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7959750679821125e-05,
-      "loss": 0.6792,
+      "loss": 0.7816,
       "step": 6345
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.795903763029958e-05,
-      "loss": 0.78,
+      "loss": 0.9326,
       "step": 6346
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7958324470357683e-05,
-      "loss": 0.8171,
+      "loss": 0.9007,
       "step": 6347
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7957611200005325e-05,
-      "loss": 0.873,
+      "loss": 0.9321,
       "step": 6348
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79568978192524e-05,
-      "loss": 0.7489,
+      "loss": 0.8577,
       "step": 6349
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7956184328108812e-05,
-      "loss": 0.6737,
+      "loss": 0.8234,
       "step": 6350
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.795547072658445e-05,
-      "loss": 0.8444,
+      "loss": 0.8557,
       "step": 6351
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.795475701468922e-05,
-      "loss": 0.8933,
+      "loss": 0.997,
       "step": 6352
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7954043192433023e-05,
-      "loss": 0.9198,
+      "loss": 0.931,
       "step": 6353
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7953329259825763e-05,
-      "loss": 0.6953,
+      "loss": 0.8855,
       "step": 6354
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7952615216877344e-05,
-      "loss": 0.8302,
+      "loss": 0.9507,
       "step": 6355
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.795190106359767e-05,
-      "loss": 0.7615,
+      "loss": 0.9323,
       "step": 6356
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7951186799996652e-05,
-      "loss": 0.862,
+      "loss": 0.9509,
       "step": 6357
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79504724260842e-05,
-      "loss": 0.8328,
+      "loss": 1.0408,
       "step": 6358
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794975794187022e-05,
-      "loss": 0.8133,
+      "loss": 0.9549,
       "step": 6359
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794904334736463e-05,
-      "loss": 0.7718,
+      "loss": 0.9191,
       "step": 6360
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7948328642577343e-05,
-      "loss": 0.8776,
+      "loss": 0.9973,
       "step": 6361
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794761382751827e-05,
-      "loss": 0.8667,
+      "loss": 0.9118,
       "step": 6362
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7946898902197337e-05,
-      "loss": 0.5626,
+      "loss": 0.7244,
       "step": 6363
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794618386662445e-05,
-      "loss": 0.8266,
+      "loss": 0.9864,
       "step": 6364
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794546872080954e-05,
-      "loss": 0.7519,
+      "loss": 0.8767,
       "step": 6365
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7944753464762522e-05,
-      "loss": 0.8261,
+      "loss": 0.8635,
       "step": 6366
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7944038098493327e-05,
-      "loss": 0.7621,
+      "loss": 0.924,
       "step": 6367
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794332262201187e-05,
-      "loss": 0.9834,
+      "loss": 1.0026,
       "step": 6368
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.794260703532808e-05,
-      "loss": 0.7961,
+      "loss": 0.9349,
       "step": 6369
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7941891338451893e-05,
-      "loss": 0.9545,
+      "loss": 1.0469,
       "step": 6370
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7941175531393223e-05,
-      "loss": 0.9092,
+      "loss": 1.0098,
       "step": 6371
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7940459614162014e-05,
-      "loss": 0.9119,
+      "loss": 1.0506,
       "step": 6372
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7939743586768198e-05,
-      "loss": 0.6872,
+      "loss": 0.9488,
       "step": 6373
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79390274492217e-05,
-      "loss": 0.8027,
+      "loss": 0.8593,
       "step": 6374
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7938311201532463e-05,
-      "loss": 0.7247,
+      "loss": 0.8508,
       "step": 6375
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7937594843710416e-05,
-      "loss": 0.8492,
+      "loss": 0.8658,
       "step": 6376
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7936878375765507e-05,
-      "loss": 0.8499,
+      "loss": 0.9203,
       "step": 6377
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.793616179770767e-05,
-      "loss": 0.7494,
+      "loss": 0.9288,
       "step": 6378
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7935445109546853e-05,
-      "loss": 0.8825,
+      "loss": 0.9192,
       "step": 6379
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7934728311292986e-05,
-      "loss": 0.7646,
+      "loss": 0.888,
       "step": 6380
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7934011402956028e-05,
-      "loss": 0.882,
+      "loss": 0.9098,
       "step": 6381
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7933294384545914e-05,
-      "loss": 0.7419,
+      "loss": 0.8946,
       "step": 6382
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7932577256072597e-05,
-      "loss": 0.8626,
+      "loss": 1.0013,
       "step": 6383
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7931860017546024e-05,
-      "loss": 0.9096,
+      "loss": 0.9405,
       "step": 6384
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7931142668976152e-05,
-      "loss": 0.8948,
+      "loss": 0.9958,
       "step": 6385
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7930425210372927e-05,
-      "loss": 0.8855,
+      "loss": 0.9279,
       "step": 6386
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79297076417463e-05,
-      "loss": 0.7561,
+      "loss": 0.8643,
       "step": 6387
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7928989963106236e-05,
-      "loss": 0.8129,
+      "loss": 0.8892,
       "step": 6388
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.792827217446268e-05,
-      "loss": 0.8279,
+      "loss": 0.8528,
       "step": 6389
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79275542758256e-05,
-      "loss": 0.8993,
+      "loss": 0.9903,
       "step": 6390
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7926836267204954e-05,
-      "loss": 0.811,
+      "loss": 0.9196,
       "step": 6391
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7926118148610698e-05,
-      "loss": 0.8449,
+      "loss": 0.9686,
       "step": 6392
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7925399920052798e-05,
-      "loss": 0.7476,
+      "loss": 0.9434,
       "step": 6393
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.792468158154122e-05,
-      "loss": 0.7061,
+      "loss": 0.8521,
       "step": 6394
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.792396313308593e-05,
-      "loss": 0.8649,
+      "loss": 0.9748,
       "step": 6395
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.792324457469689e-05,
-      "loss": 0.7412,
+      "loss": 0.8495,
       "step": 6396
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.792252590638408e-05,
-      "loss": 0.759,
+      "loss": 0.8735,
       "step": 6397
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7921807128157458e-05,
-      "loss": 0.9088,
+      "loss": 0.9426,
       "step": 6398
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7921088240027003e-05,
-      "loss": 0.8556,
+      "loss": 0.9308,
       "step": 6399
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7920369242002687e-05,
-      "loss": 0.8239,
+      "loss": 0.8812,
       "step": 6400
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7919650134094488e-05,
-      "loss": 0.8802,
+      "loss": 0.942,
       "step": 6401
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.791893091631238e-05,
-      "loss": 0.7236,
+      "loss": 0.8874,
       "step": 6402
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7918211588666337e-05,
-      "loss": 0.7334,
+      "loss": 0.8486,
       "step": 6403
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7917492151166345e-05,
-      "loss": 0.8019,
+      "loss": 0.909,
       "step": 6404
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7916772603822383e-05,
-      "loss": 0.7616,
+      "loss": 0.9318,
       "step": 6405
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7916052946644433e-05,
-      "loss": 0.8525,
+      "loss": 0.9019,
       "step": 6406
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.791533317964248e-05,
-      "loss": 0.8585,
+      "loss": 0.9285,
       "step": 6407
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7914613302826508e-05,
-      "loss": 0.8041,
+      "loss": 0.9922,
       "step": 6408
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.791389331620651e-05,
-      "loss": 0.6557,
+      "loss": 0.7695,
       "step": 6409
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7913173219792466e-05,
-      "loss": 0.8004,
+      "loss": 0.8956,
       "step": 6410
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.791245301359437e-05,
-      "loss": 0.8947,
+      "loss": 0.9682,
       "step": 6411
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7911732697622218e-05,
-      "loss": 0.819,
+      "loss": 0.9357,
       "step": 6412
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7911012271886e-05,
-      "loss": 0.9925,
+      "loss": 1.0531,
       "step": 6413
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7910291736395712e-05,
-      "loss": 0.7238,
+      "loss": 0.7983,
       "step": 6414
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7909571091161348e-05,
-      "loss": 0.9117,
+      "loss": 0.9873,
       "step": 6415
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7908850336192907e-05,
-      "loss": 0.9232,
+      "loss": 0.9695,
       "step": 6416
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7908129471500387e-05,
-      "loss": 0.8794,
+      "loss": 0.9556,
       "step": 6417
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7907408497093793e-05,
-      "loss": 0.8739,
+      "loss": 0.9478,
       "step": 6418
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7906687412983125e-05,
-      "loss": 0.9171,
+      "loss": 0.993,
       "step": 6419
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.790596621917839e-05,
-      "loss": 0.8232,
+      "loss": 0.9773,
       "step": 6420
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7905244915689585e-05,
-      "loss": 0.8343,
+      "loss": 0.8828,
       "step": 6421
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.790452350252673e-05,
-      "loss": 0.8045,
+      "loss": 0.9013,
       "step": 6422
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7903801979699817e-05,
-      "loss": 0.9277,
+      "loss": 0.9493,
       "step": 6423
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7903080347218873e-05,
-      "loss": 0.7897,
+      "loss": 0.927,
       "step": 6424
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.79023586050939e-05,
-      "loss": 0.7431,
+      "loss": 0.8855,
       "step": 6425
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7901636753334915e-05,
-      "loss": 0.7374,
+      "loss": 0.9036,
       "step": 6426
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7900914791951928e-05,
-      "loss": 0.8509,
+      "loss": 0.9191,
       "step": 6427
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.790019272095496e-05,
-      "loss": 0.8277,
+      "loss": 0.8949,
       "step": 6428
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7899470540354028e-05,
-      "loss": 0.7533,
+      "loss": 0.7615,
       "step": 6429
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7898748250159146e-05,
-      "loss": 0.8798,
+      "loss": 0.9915,
       "step": 6430
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7898025850380342e-05,
-      "loss": 0.821,
+      "loss": 0.9185,
       "step": 6431
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7897303341027635e-05,
-      "loss": 0.8627,
+      "loss": 0.983,
       "step": 6432
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.789658072211105e-05,
-      "loss": 0.8792,
+      "loss": 0.9052,
       "step": 6433
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.789585799364061e-05,
-      "loss": 0.836,
+      "loss": 0.9021,
       "step": 6434
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7895135155626342e-05,
-      "loss": 0.8124,
+      "loss": 0.918,
       "step": 6435
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7894412208078277e-05,
-      "loss": 0.7749,
+      "loss": 0.9521,
       "step": 6436
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7893689151006443e-05,
-      "loss": 0.6955,
+      "loss": 0.9133,
       "step": 6437
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7892965984420872e-05,
-      "loss": 0.7898,
+      "loss": 0.8809,
       "step": 6438
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7892242708331596e-05,
-      "loss": 0.9733,
+      "loss": 1.0331,
       "step": 6439
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.789151932274865e-05,
-      "loss": 0.6972,
+      "loss": 0.7933,
       "step": 6440
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7890795827682067e-05,
-      "loss": 0.8243,
+      "loss": 0.8597,
       "step": 6441
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7890072223141892e-05,
-      "loss": 0.8292,
+      "loss": 0.8911,
       "step": 6442
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.788934850913816e-05,
-      "loss": 0.7392,
+      "loss": 0.9472,
       "step": 6443
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7888624685680905e-05,
-      "loss": 0.6933,
+      "loss": 0.839,
       "step": 6444
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.788790075278018e-05,
-      "loss": 0.7871,
+      "loss": 0.8945,
       "step": 6445
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7887176710446023e-05,
-      "loss": 0.7738,
+      "loss": 0.8911,
       "step": 6446
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.788645255868848e-05,
-      "loss": 0.9666,
+      "loss": 1.0007,
       "step": 6447
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7885728297517596e-05,
-      "loss": 0.8569,
+      "loss": 0.9636,
       "step": 6448
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7885003926943422e-05,
-      "loss": 0.6846,
+      "loss": 0.8973,
       "step": 6449
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7884279446976003e-05,
-      "loss": 0.8919,
+      "loss": 0.9363,
       "step": 6450
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7883554857625393e-05,
-      "loss": 0.7116,
+      "loss": 0.9115,
       "step": 6451
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7882830158901646e-05,
-      "loss": 0.8461,
+      "loss": 0.9413,
       "step": 6452
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.788210535081482e-05,
-      "loss": 0.7778,
+      "loss": 0.9141,
       "step": 6453
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.788138043337496e-05,
-      "loss": 0.7551,
+      "loss": 0.9207,
       "step": 6454
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7880655406592125e-05,
-      "loss": 0.6156,
+      "loss": 0.7605,
       "step": 6455
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7879930270476384e-05,
-      "loss": 0.8764,
+      "loss": 0.9633,
       "step": 6456
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.787920502503779e-05,
-      "loss": 0.7524,
+      "loss": 0.8583,
       "step": 6457
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.78784796702864e-05,
-      "loss": 0.9155,
+      "loss": 0.9657,
       "step": 6458
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.7877754206232286e-05,
-      "loss": 0.7525,
+      "loss": 0.8558,
       "step": 6459
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.787702863288551e-05,
-      "loss": 0.6669,
+      "loss": 0.8542,
       "step": 6460
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.787630295025614e-05,
-      "loss": 0.8885,
+      "loss": 0.9394,
       "step": 6461
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7875577158354235e-05,
-      "loss": 0.7369,
+      "loss": 0.9155,
       "step": 6462
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7874851257189875e-05,
-      "loss": 0.743,
+      "loss": 0.9287,
       "step": 6463
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.787412524677313e-05,
-      "loss": 0.9448,
+      "loss": 0.9792,
       "step": 6464
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7873399127114063e-05,
-      "loss": 0.8576,
+      "loss": 0.907,
       "step": 6465
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.787267289822276e-05,
-      "loss": 0.7367,
+      "loss": 0.7757,
       "step": 6466
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7871946560109284e-05,
-      "loss": 0.8652,
+      "loss": 0.9849,
       "step": 6467
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.787122011278372e-05,
-      "loss": 0.8292,
+      "loss": 0.8875,
       "step": 6468
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.787049355625615e-05,
-      "loss": 0.6832,
+      "loss": 0.8828,
       "step": 6469
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7869766890536647e-05,
-      "loss": 0.7894,
+      "loss": 0.9312,
       "step": 6470
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.786904011563529e-05,
-      "loss": 0.8882,
+      "loss": 0.8945,
       "step": 6471
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7868313231562167e-05,
-      "loss": 0.817,
+      "loss": 0.9591,
       "step": 6472
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7867586238327366e-05,
-      "loss": 0.8268,
+      "loss": 0.9512,
       "step": 6473
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7866859135940963e-05,
-      "loss": 0.7328,
+      "loss": 0.8425,
       "step": 6474
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7866131924413052e-05,
-      "loss": 0.9112,
+      "loss": 0.9567,
       "step": 6475
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7865404603753725e-05,
-      "loss": 0.8039,
+      "loss": 0.9152,
       "step": 6476
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7864677173973065e-05,
-      "loss": 0.7656,
+      "loss": 0.9705,
       "step": 6477
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.786394963508117e-05,
-      "loss": 0.8554,
+      "loss": 0.9333,
       "step": 6478
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.786322198708813e-05,
-      "loss": 0.8728,
+      "loss": 0.8931,
       "step": 6479
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7862494230004042e-05,
-      "loss": 0.8433,
+      "loss": 0.9268,
       "step": 6480
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7861766363839e-05,
-      "loss": 0.7793,
+      "loss": 0.8548,
       "step": 6481
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7861038388603108e-05,
-      "loss": 0.8379,
+      "loss": 0.9992,
       "step": 6482
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.786031030430646e-05,
-      "loss": 0.8236,
+      "loss": 0.9205,
       "step": 6483
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7859582110959156e-05,
-      "loss": 0.7705,
+      "loss": 0.9264,
       "step": 6484
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7858853808571303e-05,
-      "loss": 0.8036,
+      "loss": 0.9164,
       "step": 6485
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7858125397153006e-05,
-      "loss": 0.7655,
+      "loss": 0.9159,
       "step": 6486
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.785739687671437e-05,
-      "loss": 0.7714,
+      "loss": 0.854,
       "step": 6487
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7856668247265492e-05,
-      "loss": 0.8358,
+      "loss": 0.9437,
       "step": 6488
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.78559395088165e-05,
-      "loss": 0.9141,
+      "loss": 0.9645,
       "step": 6489
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7855210661377483e-05,
-      "loss": 0.7253,
+      "loss": 0.8788,
       "step": 6490
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7854481704958567e-05,
-      "loss": 0.8774,
+      "loss": 0.8867,
       "step": 6491
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7853752639569866e-05,
-      "loss": 0.7594,
+      "loss": 0.8379,
       "step": 6492
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7853023465221485e-05,
-      "loss": 0.8297,
+      "loss": 0.984,
       "step": 6493
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.785229418192355e-05,
-      "loss": 0.8661,
+      "loss": 0.924,
       "step": 6494
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7851564789686172e-05,
-      "loss": 0.9012,
+      "loss": 1.0322,
       "step": 6495
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.785083528851947e-05,
-      "loss": 0.7046,
+      "loss": 0.8015,
       "step": 6496
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.785010567843357e-05,
-      "loss": 0.7907,
+      "loss": 0.9651,
       "step": 6497
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.784937595943859e-05,
-      "loss": 0.9149,
+      "loss": 1.0026,
       "step": 6498
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7848646131544653e-05,
-      "loss": 0.8396,
+      "loss": 0.9197,
       "step": 6499
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7847916194761893e-05,
-      "loss": 0.7195,
+      "loss": 0.8159,
       "step": 6500
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7847186149100427e-05,
-      "loss": 0.799,
+      "loss": 0.8829,
       "step": 6501
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7846455994570386e-05,
-      "loss": 0.7321,
+      "loss": 0.8646,
       "step": 6502
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7845725731181902e-05,
-      "loss": 0.8807,
+      "loss": 1.0325,
       "step": 6503
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7844995358945105e-05,
-      "loss": 0.7264,
+      "loss": 0.919,
       "step": 6504
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7844264877870126e-05,
-      "loss": 0.8103,
+      "loss": 0.9759,
       "step": 6505
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7843534287967105e-05,
-      "loss": 0.7768,
+      "loss": 0.8972,
       "step": 6506
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.784280358924617e-05,
-      "loss": 0.7698,
+      "loss": 0.9149,
       "step": 6507
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7842072781717464e-05,
-      "loss": 0.7628,
+      "loss": 0.8647,
       "step": 6508
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.784134186539113e-05,
-      "loss": 0.7265,
+      "loss": 0.8925,
       "step": 6509
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7840610840277298e-05,
-      "loss": 0.6721,
+      "loss": 0.822,
       "step": 6510
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7839879706386116e-05,
-      "loss": 0.8727,
+      "loss": 0.8768,
       "step": 6511
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7839148463727728e-05,
-      "loss": 0.8682,
+      "loss": 0.9734,
       "step": 6512
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7838417112312274e-05,
-      "loss": 0.8379,
+      "loss": 1.0035,
       "step": 6513
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7837685652149907e-05,
-      "loss": 0.7963,
+      "loss": 0.8864,
       "step": 6514
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.783695408325077e-05,
-      "loss": 0.8678,
+      "loss": 0.9544,
       "step": 6515
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.783622240562501e-05,
-      "loss": 0.7434,
+      "loss": 0.9597,
       "step": 6516
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7835490619282788e-05,
-      "loss": 0.807,
+      "loss": 0.9229,
       "step": 6517
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.783475872423425e-05,
-      "loss": 0.7216,
+      "loss": 0.8303,
       "step": 6518
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.783402672048955e-05,
-      "loss": 0.8745,
+      "loss": 0.9656,
       "step": 6519
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7833294608058845e-05,
-      "loss": 0.7943,
+      "loss": 0.8642,
       "step": 6520
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7832562386952293e-05,
-      "loss": 0.7367,
+      "loss": 0.9127,
       "step": 6521
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7831830057180046e-05,
-      "loss": 0.7167,
+      "loss": 0.8657,
       "step": 6522
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7831097618752274e-05,
-      "loss": 0.8935,
+      "loss": 0.9611,
       "step": 6523
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.783036507167913e-05,
-      "loss": 0.7,
+      "loss": 0.8599,
       "step": 6524
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7829632415970783e-05,
-      "loss": 0.8451,
+      "loss": 0.9196,
       "step": 6525
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7828899651637393e-05,
-      "loss": 0.8524,
+      "loss": 0.9764,
       "step": 6526
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7828166778689125e-05,
-      "loss": 0.8925,
+      "loss": 0.9274,
       "step": 6527
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7827433797136156e-05,
-      "loss": 0.8194,
+      "loss": 0.9072,
       "step": 6528
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7826700706988646e-05,
-      "loss": 0.966,
+      "loss": 1.015,
       "step": 6529
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7825967508256768e-05,
-      "loss": 0.6363,
+      "loss": 0.8165,
       "step": 6530
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7825234200950692e-05,
-      "loss": 0.8336,
+      "loss": 0.9481,
       "step": 6531
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7824500785080597e-05,
-      "loss": 0.8242,
+      "loss": 0.9391,
       "step": 6532
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7823767260656656e-05,
-      "loss": 0.7521,
+      "loss": 0.8895,
       "step": 6533
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7823033627689045e-05,
-      "loss": 0.8379,
+      "loss": 0.8868,
       "step": 6534
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.782229988618794e-05,
-      "loss": 0.7748,
+      "loss": 0.9551,
       "step": 6535
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7821566036163522e-05,
-      "loss": 0.7198,
+      "loss": 0.8931,
       "step": 6536
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7820832077625977e-05,
-      "loss": 0.6872,
+      "loss": 0.8985,
       "step": 6537
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7820098010585476e-05,
-      "loss": 0.7901,
+      "loss": 0.8795,
       "step": 6538
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7819363835052215e-05,
-      "loss": 0.8021,
+      "loss": 0.946,
       "step": 6539
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7818629551036374e-05,
-      "loss": 0.8958,
+      "loss": 0.9383,
       "step": 6540
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7817895158548143e-05,
-      "loss": 0.7395,
+      "loss": 0.815,
       "step": 6541
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7817160657597705e-05,
-      "loss": 0.861,
+      "loss": 0.953,
       "step": 6542
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7816426048195255e-05,
-      "loss": 0.7167,
+      "loss": 0.7533,
       "step": 6543
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7815691330350986e-05,
-      "loss": 0.8057,
+      "loss": 1.0156,
       "step": 6544
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7814956504075088e-05,
-      "loss": 0.8185,
+      "loss": 0.8817,
       "step": 6545
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7814221569377753e-05,
-      "loss": 0.7364,
+      "loss": 0.8732,
       "step": 6546
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7813486526269186e-05,
-      "loss": 0.9235,
+      "loss": 1.0635,
       "step": 6547
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.781275137475958e-05,
-      "loss": 0.6778,
+      "loss": 0.8978,
       "step": 6548
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7812016114859127e-05,
-      "loss": 0.7327,
+      "loss": 0.8657,
       "step": 6549
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.781128074657804e-05,
-      "loss": 0.8298,
+      "loss": 0.9149,
       "step": 6550
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.781054526992651e-05,
-      "loss": 0.84,
+      "loss": 0.9638,
       "step": 6551
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7809809684914753e-05,
-      "loss": 0.9122,
+      "loss": 0.9263,
       "step": 6552
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.780907399155296e-05,
-      "loss": 0.7877,
+      "loss": 0.9196,
       "step": 6553
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.780833818985135e-05,
-      "loss": 0.9465,
+      "loss": 1.0419,
       "step": 6554
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7807602279820123e-05,
-      "loss": 0.7972,
+      "loss": 0.956,
       "step": 6555
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7806866261469497e-05,
-      "loss": 0.7765,
+      "loss": 0.8675,
       "step": 6556
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7806130134809675e-05,
-      "loss": 0.7417,
+      "loss": 0.9422,
       "step": 6557
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.780539389985087e-05,
-      "loss": 0.848,
+      "loss": 0.9346,
       "step": 6558
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.78046575566033e-05,
-      "loss": 0.7872,
+      "loss": 0.9107,
       "step": 6559
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7803921105077183e-05,
-      "loss": 0.6803,
+      "loss": 0.8908,
       "step": 6560
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7803184545282734e-05,
-      "loss": 0.8222,
+      "loss": 0.878,
       "step": 6561
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7802447877230166e-05,
-      "loss": 0.8347,
+      "loss": 0.9062,
       "step": 6562
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7801711100929708e-05,
-      "loss": 0.8491,
+      "loss": 0.8733,
       "step": 6563
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7800974216391575e-05,
-      "loss": 0.8835,
+      "loss": 0.9766,
       "step": 6564
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7800237223625992e-05,
-      "loss": 0.9122,
+      "loss": 0.9803,
       "step": 6565
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7799500122643186e-05,
-      "loss": 0.895,
+      "loss": 0.9343,
       "step": 6566
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779876291345338e-05,
-      "loss": 0.7819,
+      "loss": 1.0081,
       "step": 6567
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779802559606681e-05,
-      "loss": 0.7485,
+      "loss": 0.9011,
       "step": 6568
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7797288170493692e-05,
-      "loss": 0.7797,
+      "loss": 0.8691,
       "step": 6569
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7796550636744267e-05,
-      "loss": 0.7814,
+      "loss": 0.8552,
       "step": 6570
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779581299482876e-05,
-      "loss": 0.7952,
+      "loss": 0.9396,
       "step": 6571
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779507524475741e-05,
-      "loss": 0.83,
+      "loss": 0.9767,
       "step": 6572
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7794337386540454e-05,
-      "loss": 0.9522,
+      "loss": 0.9929,
       "step": 6573
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7793599420188125e-05,
-      "loss": 0.9074,
+      "loss": 0.9707,
       "step": 6574
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779286134571066e-05,
-      "loss": 0.723,
+      "loss": 0.9058,
       "step": 6575
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.77921231631183e-05,
-      "loss": 0.7318,
+      "loss": 0.8464,
       "step": 6576
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7791384872421286e-05,
-      "loss": 0.7426,
+      "loss": 0.8802,
       "step": 6577
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.779064647362986e-05,
-      "loss": 0.7996,
+      "loss": 0.9294,
       "step": 6578
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7789907966754274e-05,
-      "loss": 0.6748,
+      "loss": 0.9542,
       "step": 6579
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.778916935180476e-05,
-      "loss": 0.6334,
+      "loss": 0.8706,
       "step": 6580
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.778843062879158e-05,
-      "loss": 0.8453,
+      "loss": 1.002,
       "step": 6581
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7787691797724972e-05,
-      "loss": 0.7188,
+      "loss": 0.9376,
       "step": 6582
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.778695285861519e-05,
-      "loss": 0.9793,
+      "loss": 0.9592,
       "step": 6583
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7786213811472483e-05,
-      "loss": 0.7741,
+      "loss": 0.8886,
       "step": 6584
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7785474656307108e-05,
-      "loss": 0.8046,
+      "loss": 0.8951,
       "step": 6585
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7784735393129317e-05,
-      "loss": 0.8648,
+      "loss": 0.9691,
       "step": 6586
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.778399602194937e-05,
-      "loss": 0.8688,
+      "loss": 0.9225,
       "step": 6587
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7783256542777523e-05,
-      "loss": 0.7624,
+      "loss": 0.9283,
       "step": 6588
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7782516955624033e-05,
-      "loss": 0.8868,
+      "loss": 1.0042,
       "step": 6589
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7781777260499162e-05,
-      "loss": 0.8074,
+      "loss": 0.8918,
       "step": 6590
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.778103745741317e-05,
-      "loss": 0.6971,
+      "loss": 0.7967,
       "step": 6591
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7780297546376327e-05,
-      "loss": 0.8929,
+      "loss": 0.9778,
       "step": 6592
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7779557527398892e-05,
-      "loss": 0.7493,
+      "loss": 0.848,
       "step": 6593
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7778817400491136e-05,
-      "loss": 0.7927,
+      "loss": 0.9084,
       "step": 6594
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7778077165663323e-05,
-      "loss": 0.8516,
+      "loss": 0.9779,
       "step": 6595
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7777336822925727e-05,
-      "loss": 0.8604,
+      "loss": 0.9447,
       "step": 6596
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7776596372288616e-05,
-      "loss": 0.7801,
+      "loss": 0.9421,
       "step": 6597
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7775855813762263e-05,
-      "loss": 0.936,
+      "loss": 0.9989,
       "step": 6598
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7775115147356946e-05,
-      "loss": 0.7137,
+      "loss": 0.7489,
       "step": 6599
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7774374373082935e-05,
-      "loss": 0.8459,
+      "loss": 1.018,
       "step": 6600
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.777363349095051e-05,
-      "loss": 0.8116,
+      "loss": 0.8858,
       "step": 6601
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.777289250096995e-05,
-      "loss": 0.8452,
+      "loss": 0.9501,
       "step": 6602
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7772151403151537e-05,
-      "loss": 0.7678,
+      "loss": 0.8816,
       "step": 6603
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.777141019750555e-05,
-      "loss": 0.9594,
+      "loss": 1.0129,
       "step": 6604
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.777066888404227e-05,
-      "loss": 0.8149,
+      "loss": 0.8926,
       "step": 6605
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7769927462771986e-05,
-      "loss": 0.9404,
+      "loss": 1.0071,
       "step": 6606
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7769185933704985e-05,
-      "loss": 0.8842,
+      "loss": 0.924,
       "step": 6607
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7768444296851547e-05,
-      "loss": 0.7206,
+      "loss": 0.8972,
       "step": 6608
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.776770255222197e-05,
-      "loss": 0.6871,
+      "loss": 0.8133,
       "step": 6609
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7766960699826537e-05,
-      "loss": 0.8974,
+      "loss": 0.9725,
       "step": 6610
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.776621873967555e-05,
-      "loss": 0.9397,
+      "loss": 0.9831,
       "step": 6611
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7765476671779295e-05,
-      "loss": 0.8674,
+      "loss": 0.9286,
       "step": 6612
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7764734496148067e-05,
-      "loss": 0.7689,
+      "loss": 0.8994,
       "step": 6613
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7763992212792165e-05,
-      "loss": 0.8303,
+      "loss": 0.9681,
       "step": 6614
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.776324982172189e-05,
-      "loss": 0.8477,
+      "loss": 0.9041,
       "step": 6615
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7762507322947534e-05,
-      "loss": 0.7256,
+      "loss": 0.9639,
       "step": 6616
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7761764716479405e-05,
-      "loss": 0.7239,
+      "loss": 0.8721,
       "step": 6617
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.77610220023278e-05,
-      "loss": 0.7787,
+      "loss": 0.891,
       "step": 6618
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7760279180503032e-05,
-      "loss": 0.8637,
+      "loss": 0.9794,
       "step": 6619
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7759536251015395e-05,
-      "loss": 0.728,
+      "loss": 0.9106,
       "step": 6620
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7758793213875204e-05,
-      "loss": 0.8878,
+      "loss": 0.9787,
       "step": 6621
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7758050069092764e-05,
-      "loss": 0.741,
+      "loss": 0.8981,
       "step": 6622
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.775730681667839e-05,
-      "loss": 0.7783,
+      "loss": 0.8296,
       "step": 6623
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7756563456642388e-05,
-      "loss": 1.016,
+      "loss": 1.0189,
       "step": 6624
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7755819988995072e-05,
-      "loss": 0.8054,
+      "loss": 0.9752,
       "step": 6625
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7755076413746758e-05,
-      "loss": 0.762,
+      "loss": 0.9656,
       "step": 6626
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7754332730907764e-05,
-      "loss": 0.6522,
+      "loss": 0.8054,
       "step": 6627
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7753588940488406e-05,
-      "loss": 0.9741,
+      "loss": 1.0399,
       "step": 6628
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7752845042498997e-05,
-      "loss": 0.7372,
+      "loss": 0.8867,
       "step": 6629
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.775210103694987e-05,
-      "loss": 0.8715,
+      "loss": 0.9319,
       "step": 6630
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7751356923851335e-05,
-      "loss": 0.9331,
+      "loss": 1.0173,
       "step": 6631
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.775061270321372e-05,
-      "loss": 0.8523,
+      "loss": 0.9703,
       "step": 6632
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774986837504735e-05,
-      "loss": 0.7714,
+      "loss": 0.8891,
       "step": 6633
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774912393936256e-05,
-      "loss": 0.7605,
+      "loss": 0.9637,
       "step": 6634
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774837939616966e-05,
-      "loss": 0.7022,
+      "loss": 0.7995,
       "step": 6635
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7747634745478995e-05,
-      "loss": 0.8154,
+      "loss": 0.9394,
       "step": 6636
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774688998730089e-05,
-      "loss": 0.6825,
+      "loss": 0.8317,
       "step": 6637
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7746145121645674e-05,
-      "loss": 0.6832,
+      "loss": 0.8003,
       "step": 6638
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7745400148523688e-05,
-      "loss": 0.8118,
+      "loss": 0.8867,
       "step": 6639
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7744655067945263e-05,
-      "loss": 0.8582,
+      "loss": 0.8842,
       "step": 6640
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774390987992074e-05,
-      "loss": 0.9051,
+      "loss": 0.9963,
       "step": 6641
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774316458446045e-05,
-      "loss": 0.8151,
+      "loss": 0.9232,
       "step": 6642
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.774241918157474e-05,
-      "loss": 0.8362,
+      "loss": 0.9545,
       "step": 6643
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7741673671273945e-05,
-      "loss": 0.7671,
+      "loss": 0.9126,
       "step": 6644
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7740928053568418e-05,
-      "loss": 0.8911,
+      "loss": 0.9081,
       "step": 6645
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7740182328468495e-05,
-      "loss": 0.7862,
+      "loss": 0.9082,
       "step": 6646
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7739436495984527e-05,
-      "loss": 0.8204,
+      "loss": 0.9183,
       "step": 6647
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7738690556126853e-05,
-      "loss": 0.8715,
+      "loss": 1.0022,
       "step": 6648
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773794450890583e-05,
-      "loss": 0.8147,
+      "loss": 0.8983,
       "step": 6649
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7737198354331803e-05,
-      "loss": 0.8891,
+      "loss": 0.9874,
       "step": 6650
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7736452092415127e-05,
-      "loss": 0.7643,
+      "loss": 0.875,
       "step": 6651
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773570572316616e-05,
-      "loss": 0.7919,
+      "loss": 0.9769,
       "step": 6652
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773495924659525e-05,
-      "loss": 0.9458,
+      "loss": 0.9646,
       "step": 6653
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773421266271275e-05,
-      "loss": 0.7727,
+      "loss": 0.8838,
       "step": 6654
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7733465971529026e-05,
-      "loss": 0.6718,
+      "loss": 0.9287,
       "step": 6655
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7732719173054434e-05,
-      "loss": 0.796,
+      "loss": 0.9004,
       "step": 6656
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7731972267299335e-05,
-      "loss": 0.7541,
+      "loss": 0.8236,
       "step": 6657
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773122525427409e-05,
-      "loss": 0.7903,
+      "loss": 0.856,
       "step": 6658
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.773047813398906e-05,
-      "loss": 0.6596,
+      "loss": 0.9214,
       "step": 6659
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7729730906454622e-05,
-      "loss": 0.85,
+      "loss": 0.913,
       "step": 6660
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772898357168113e-05,
-      "loss": 0.9424,
+      "loss": 0.9865,
       "step": 6661
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7728236129678958e-05,
-      "loss": 0.8394,
+      "loss": 0.8769,
       "step": 6662
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7727488580458475e-05,
-      "loss": 0.7423,
+      "loss": 0.8815,
       "step": 6663
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7726740924030052e-05,
-      "loss": 0.8635,
+      "loss": 0.9726,
       "step": 6664
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772599316040406e-05,
-      "loss": 0.721,
+      "loss": 0.8479,
       "step": 6665
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7725245289590875e-05,
-      "loss": 0.9408,
+      "loss": 0.9933,
       "step": 6666
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772449731160087e-05,
-      "loss": 0.7653,
+      "loss": 0.8999,
       "step": 6667
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7723749226444428e-05,
-      "loss": 0.7532,
+      "loss": 0.8112,
       "step": 6668
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7723001034131927e-05,
-      "loss": 0.8231,
+      "loss": 0.9121,
       "step": 6669
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772225273467374e-05,
-      "loss": 0.9479,
+      "loss": 0.9807,
       "step": 6670
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772150432808025e-05,
-      "loss": 0.9943,
+      "loss": 1.0447,
       "step": 6671
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7720755814361848e-05,
-      "loss": 0.7979,
+      "loss": 0.9801,
       "step": 6672
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.772000719352891e-05,
-      "loss": 0.7771,
+      "loss": 0.8698,
       "step": 6673
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771925846559183e-05,
-      "loss": 0.7376,
+      "loss": 0.7995,
       "step": 6674
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7718509630560987e-05,
-      "loss": 0.9461,
+      "loss": 0.9677,
       "step": 6675
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7717760688446776e-05,
-      "loss": 0.8148,
+      "loss": 0.9124,
       "step": 6676
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7717011639259584e-05,
-      "loss": 0.8916,
+      "loss": 0.9987,
       "step": 6677
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7716262483009802e-05,
-      "loss": 0.779,
+      "loss": 0.845,
       "step": 6678
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771551321970783e-05,
-      "loss": 0.826,
+      "loss": 0.8813,
       "step": 6679
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7714763849364057e-05,
-      "loss": 0.7169,
+      "loss": 0.8909,
       "step": 6680
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7714014371988883e-05,
-      "loss": 0.8089,
+      "loss": 0.8982,
       "step": 6681
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7713264787592702e-05,
-      "loss": 0.8163,
+      "loss": 0.9305,
       "step": 6682
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771251509618592e-05,
-      "loss": 0.8649,
+      "loss": 0.9549,
       "step": 6683
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771176529777893e-05,
-      "loss": 0.7265,
+      "loss": 0.8023,
       "step": 6684
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771101539238214e-05,
-      "loss": 0.7027,
+      "loss": 0.8454,
       "step": 6685
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.771026538000595e-05,
-      "loss": 0.7627,
+      "loss": 0.8146,
       "step": 6686
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7709515260660767e-05,
-      "loss": 0.7854,
+      "loss": 0.8421,
       "step": 6687
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7708765034357e-05,
-      "loss": 0.8654,
+      "loss": 0.9543,
       "step": 6688
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7708014701105053e-05,
-      "loss": 0.8504,
+      "loss": 0.891,
       "step": 6689
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7707264260915344e-05,
-      "loss": 0.6976,
+      "loss": 0.7902,
       "step": 6690
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7706513713798273e-05,
-      "loss": 0.8677,
+      "loss": 0.9869,
       "step": 6691
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.770576305976426e-05,
-      "loss": 0.716,
+      "loss": 0.846,
       "step": 6692
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7705012298823717e-05,
-      "loss": 0.7998,
+      "loss": 0.8772,
       "step": 6693
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7704261430987063e-05,
-      "loss": 0.8446,
+      "loss": 0.8992,
       "step": 6694
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.770351045626471e-05,
-      "loss": 0.8502,
+      "loss": 0.9227,
       "step": 6695
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7702759374667078e-05,
-      "loss": 0.6493,
+      "loss": 0.8256,
       "step": 6696
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.770200818620459e-05,
-      "loss": 0.8321,
+      "loss": 0.9506,
       "step": 6697
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7701256890887667e-05,
-      "loss": 0.8224,
+      "loss": 0.893,
       "step": 6698
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7700505488726733e-05,
-      "loss": 0.8286,
+      "loss": 0.8848,
       "step": 6699
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.769975397973221e-05,
-      "loss": 0.768,
+      "loss": 0.8315,
       "step": 6700
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7699002363914522e-05,
-      "loss": 0.9392,
+      "loss": 0.9962,
       "step": 6701
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7698250641284097e-05,
-      "loss": 0.8513,
+      "loss": 0.9111,
       "step": 6702
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7697498811851375e-05,
-      "loss": 0.9616,
+      "loss": 1.011,
       "step": 6703
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7696746875626772e-05,
-      "loss": 0.7847,
+      "loss": 0.9613,
       "step": 6704
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.769599483262073e-05,
-      "loss": 0.831,
+      "loss": 0.9591,
       "step": 6705
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.769524268284368e-05,
-      "loss": 0.8231,
+      "loss": 0.976,
       "step": 6706
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.769449042630605e-05,
-      "loss": 0.799,
+      "loss": 0.8816,
       "step": 6707
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7693738063018286e-05,
-      "loss": 0.8062,
+      "loss": 0.9509,
       "step": 6708
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7692985592990823e-05,
-      "loss": 0.6311,
+      "loss": 0.8829,
       "step": 6709
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7692233016234098e-05,
-      "loss": 0.7741,
+      "loss": 0.87,
       "step": 6710
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7691480332758554e-05,
-      "loss": 0.8095,
+      "loss": 0.986,
       "step": 6711
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7690727542574636e-05,
-      "loss": 0.8322,
+      "loss": 0.9245,
       "step": 6712
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7689974645692786e-05,
-      "loss": 0.9614,
+      "loss": 1.0143,
       "step": 6713
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7689221642123444e-05,
-      "loss": 0.6853,
+      "loss": 0.8551,
       "step": 6714
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.768846853187706e-05,
-      "loss": 0.7543,
+      "loss": 0.91,
       "step": 6715
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.768771531496409e-05,
-      "loss": 0.8002,
+      "loss": 0.8717,
       "step": 6716
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7686961991394973e-05,
-      "loss": 0.7367,
+      "loss": 0.873,
       "step": 6717
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7686208561180163e-05,
-      "loss": 0.829,
+      "loss": 0.9086,
       "step": 6718
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.768545502433012e-05,
-      "loss": 0.7479,
+      "loss": 0.8679,
       "step": 6719
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7684701380855286e-05,
-      "loss": 0.7645,
+      "loss": 0.8712,
       "step": 6720
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.768394763076613e-05,
-      "loss": 0.8866,
+      "loss": 0.9428,
       "step": 6721
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7683193774073098e-05,
-      "loss": 0.9035,
+      "loss": 1.0138,
       "step": 6722
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7682439810786654e-05,
-      "loss": 0.8889,
+      "loss": 1.0403,
       "step": 6723
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7681685740917262e-05,
-      "loss": 0.8478,
+      "loss": 1.0164,
       "step": 6724
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7680931564475377e-05,
-      "loss": 0.9907,
+      "loss": 1.0006,
       "step": 6725
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7680177281471462e-05,
-      "loss": 0.9052,
+      "loss": 1.0209,
       "step": 6726
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7679422891915985e-05,
-      "loss": 0.9283,
+      "loss": 0.9837,
       "step": 6727
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7678668395819414e-05,
-      "loss": 0.7968,
+      "loss": 0.9636,
       "step": 6728
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.767791379319221e-05,
-      "loss": 0.8612,
+      "loss": 0.9685,
       "step": 6729
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7677159084044847e-05,
-      "loss": 0.9084,
+      "loss": 0.9689,
       "step": 6730
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7676404268387793e-05,
-      "loss": 0.8696,
+      "loss": 0.9296,
       "step": 6731
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7675649346231525e-05,
-      "loss": 0.8606,
+      "loss": 0.9344,
       "step": 6732
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7674894317586512e-05,
-      "loss": 0.8352,
+      "loss": 0.9353,
       "step": 6733
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7674139182463227e-05,
-      "loss": 0.7364,
+      "loss": 0.8824,
       "step": 6734
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7673383940872146e-05,
-      "loss": 0.7646,
+      "loss": 0.9134,
       "step": 6735
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.7672628592823754e-05,
-      "loss": 0.7414,
+      "loss": 0.8161,
       "step": 6736
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7671873138328526e-05,
-      "loss": 0.8027,
+      "loss": 0.9016,
       "step": 6737
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7671117577396943e-05,
-      "loss": 0.728,
+      "loss": 0.9614,
       "step": 6738
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.767036191003949e-05,
-      "loss": 0.7969,
+      "loss": 0.8768,
       "step": 6739
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7669606136266643e-05,
-      "loss": 0.7674,
+      "loss": 0.8541,
       "step": 6740
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7668850256088895e-05,
-      "loss": 0.8737,
+      "loss": 0.9114,
       "step": 6741
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.766809426951673e-05,
-      "loss": 0.7301,
+      "loss": 0.8864,
       "step": 6742
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7667338176560637e-05,
-      "loss": 0.7568,
+      "loss": 0.9156,
       "step": 6743
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.766658197723111e-05,
-      "loss": 0.6936,
+      "loss": 0.8687,
       "step": 6744
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7665825671538628e-05,
-      "loss": 0.8651,
+      "loss": 0.9297,
       "step": 6745
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7665069259493693e-05,
-      "loss": 0.8636,
+      "loss": 0.9461,
       "step": 6746
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.76643127411068e-05,
-      "loss": 0.7925,
+      "loss": 0.9002,
       "step": 6747
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7663556116388444e-05,
-      "loss": 0.8118,
+      "loss": 0.8798,
       "step": 6748
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7662799385349115e-05,
-      "loss": 0.828,
+      "loss": 0.8795,
       "step": 6749
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.766204254799932e-05,
-      "loss": 0.8825,
+      "loss": 0.9164,
       "step": 6750
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7661285604349553e-05,
-      "loss": 0.8135,
+      "loss": 0.8874,
       "step": 6751
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.766052855441032e-05,
-      "loss": 0.7854,
+      "loss": 0.8409,
       "step": 6752
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.765977139819212e-05,
-      "loss": 0.7379,
+      "loss": 0.9174,
       "step": 6753
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7659014135705462e-05,
-      "loss": 0.8317,
+      "loss": 0.9534,
       "step": 6754
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7658256766960845e-05,
-      "loss": 0.8391,
+      "loss": 0.8976,
       "step": 6755
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7657499291968786e-05,
-      "loss": 0.8841,
+      "loss": 0.9218,
       "step": 6756
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7656741710739788e-05,
-      "loss": 0.8918,
+      "loss": 0.9899,
       "step": 6757
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.765598402328436e-05,
-      "loss": 0.8045,
+      "loss": 0.8894,
       "step": 6758
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.765522622961302e-05,
-      "loss": 0.9066,
+      "loss": 0.9733,
       "step": 6759
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7654468329736277e-05,
-      "loss": 0.7171,
+      "loss": 0.8673,
       "step": 6760
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7653710323664643e-05,
-      "loss": 1.0573,
+      "loss": 1.0831,
       "step": 6761
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.765295221140864e-05,
-      "loss": 0.7859,
+      "loss": 0.9245,
       "step": 6762
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.765219399297878e-05,
-      "loss": 0.7692,
+      "loss": 0.8798,
       "step": 6763
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7651435668385592e-05,
-      "loss": 0.7481,
+      "loss": 0.9454,
       "step": 6764
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7650677237639583e-05,
-      "loss": 0.8562,
+      "loss": 0.9749,
       "step": 6765
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7649918700751287e-05,
-      "loss": 0.8833,
+      "loss": 1.0004,
       "step": 6766
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.764916005773122e-05,
-      "loss": 0.839,
+      "loss": 0.9425,
       "step": 6767
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.764840130858991e-05,
-      "loss": 0.7997,
+      "loss": 0.9457,
       "step": 6768
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7647642453337883e-05,
-      "loss": 0.8315,
+      "loss": 0.8982,
       "step": 6769
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.764688349198567e-05,
-      "loss": 0.91,
+      "loss": 0.8935,
       "step": 6770
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7646124424543795e-05,
-      "loss": 0.7371,
+      "loss": 0.8437,
       "step": 6771
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7645365251022794e-05,
-      "loss": 0.7974,
+      "loss": 0.908,
       "step": 6772
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7644605971433197e-05,
-      "loss": 0.785,
+      "loss": 0.8669,
       "step": 6773
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7643846585785538e-05,
-      "loss": 0.8666,
+      "loss": 0.9365,
       "step": 6774
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7643087094090353e-05,
-      "loss": 0.7936,
+      "loss": 0.8761,
       "step": 6775
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7642327496358183e-05,
-      "loss": 0.7388,
+      "loss": 0.9104,
       "step": 6776
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7641567792599558e-05,
-      "loss": 0.7391,
+      "loss": 0.8579,
       "step": 6777
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.764080798282502e-05,
-      "loss": 0.7909,
+      "loss": 0.9206,
       "step": 6778
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7640048067045116e-05,
-      "loss": 0.9122,
+      "loss": 0.94,
       "step": 6779
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7639288045270383e-05,
-      "loss": 0.7878,
+      "loss": 0.8383,
       "step": 6780
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7638527917511365e-05,
-      "loss": 0.7916,
+      "loss": 0.9671,
       "step": 6781
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.763776768377861e-05,
-      "loss": 0.9687,
+      "loss": 1.019,
       "step": 6782
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.763700734408267e-05,
-      "loss": 0.8362,
+      "loss": 0.986,
       "step": 6783
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7636246898434083e-05,
-      "loss": 0.8453,
+      "loss": 0.9128,
       "step": 6784
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.763548634684341e-05,
-      "loss": 0.8539,
+      "loss": 0.9096,
       "step": 6785
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7634725689321193e-05,
-      "loss": 0.8575,
+      "loss": 0.949,
       "step": 6786
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7633964925877992e-05,
-      "loss": 0.7062,
+      "loss": 0.816,
       "step": 6787
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.763320405652436e-05,
-      "loss": 0.6788,
+      "loss": 0.8914,
       "step": 6788
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7632443081270848e-05,
-      "loss": 0.7864,
+      "loss": 0.9115,
       "step": 6789
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7631682000128017e-05,
-      "loss": 0.8271,
+      "loss": 0.8901,
       "step": 6790
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.763092081310643e-05,
-      "loss": 0.86,
+      "loss": 0.9092,
       "step": 6791
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7630159520216642e-05,
-      "loss": 0.7282,
+      "loss": 0.7616,
       "step": 6792
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7629398121469216e-05,
-      "loss": 0.7774,
+      "loss": 0.9413,
       "step": 6793
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7628636616874713e-05,
-      "loss": 0.7155,
+      "loss": 0.8601,
       "step": 6794
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7627875006443704e-05,
-      "loss": 0.8511,
+      "loss": 0.9926,
       "step": 6795
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.762711329018675e-05,
-      "loss": 0.7979,
+      "loss": 0.8825,
       "step": 6796
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7626351468114422e-05,
-      "loss": 0.7434,
+      "loss": 0.9138,
       "step": 6797
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7625589540237288e-05,
-      "loss": 0.8682,
+      "loss": 0.8939,
       "step": 6798
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7624827506565918e-05,
-      "loss": 0.9424,
+      "loss": 0.9733,
       "step": 6799
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7624065367110887e-05,
-      "loss": 0.8606,
+      "loss": 1.0006,
       "step": 6800
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7623303121882762e-05,
-      "loss": 0.8705,
+      "loss": 0.9819,
       "step": 6801
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.762254077089212e-05,
-      "loss": 0.8465,
+      "loss": 0.9217,
       "step": 6802
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7621778314149546e-05,
-      "loss": 0.8128,
+      "loss": 0.8557,
       "step": 6803
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.762101575166561e-05,
-      "loss": 0.9779,
+      "loss": 0.9651,
       "step": 6804
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.762025308345089e-05,
-      "loss": 0.7658,
+      "loss": 0.8917,
       "step": 6805
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7619490309515975e-05,
-      "loss": 0.876,
+      "loss": 0.9045,
       "step": 6806
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7618727429871442e-05,
-      "loss": 0.8085,
+      "loss": 0.9163,
       "step": 6807
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7617964444527873e-05,
-      "loss": 0.8153,
+      "loss": 0.9857,
       "step": 6808
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7617201353495855e-05,
-      "loss": 0.8016,
+      "loss": 0.9507,
       "step": 6809
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761643815678598e-05,
-      "loss": 0.6939,
+      "loss": 0.8597,
       "step": 6810
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761567485440883e-05,
-      "loss": 0.7971,
+      "loss": 0.8738,
       "step": 6811
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7614911446374992e-05,
-      "loss": 0.8383,
+      "loss": 0.8921,
       "step": 6812
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7614147932695066e-05,
-      "loss": 0.8094,
+      "loss": 0.9617,
       "step": 6813
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761338431337964e-05,
-      "loss": 0.7533,
+      "loss": 0.9162,
       "step": 6814
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761262058843931e-05,
-      "loss": 0.8111,
+      "loss": 0.9321,
       "step": 6815
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761185675788467e-05,
-      "loss": 0.9319,
+      "loss": 0.9864,
       "step": 6816
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7611092821726313e-05,
-      "loss": 0.7423,
+      "loss": 0.8774,
       "step": 6817
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.761032877997485e-05,
-      "loss": 0.7868,
+      "loss": 0.9304,
       "step": 6818
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7609564632640865e-05,
-      "loss": 0.8995,
+      "loss": 1.0347,
       "step": 6819
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.760880037973497e-05,
-      "loss": 0.7785,
+      "loss": 0.9163,
       "step": 6820
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7608036021267767e-05,
-      "loss": 0.7512,
+      "loss": 0.8565,
       "step": 6821
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7607271557249858e-05,
-      "loss": 0.7302,
+      "loss": 0.8618,
       "step": 6822
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7606506987691846e-05,
-      "loss": 0.7368,
+      "loss": 0.886,
       "step": 6823
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7605742312604344e-05,
-      "loss": 0.7353,
+      "loss": 0.8629,
       "step": 6824
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.760497753199796e-05,
-      "loss": 0.8725,
+      "loss": 0.9635,
       "step": 6825
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.76042126458833e-05,
-      "loss": 0.7517,
+      "loss": 0.9696,
       "step": 6826
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7603447654270983e-05,
-      "loss": 0.867,
+      "loss": 0.9775,
       "step": 6827
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7602682557171612e-05,
-      "loss": 0.7614,
+      "loss": 0.9039,
       "step": 6828
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7601917354595813e-05,
-      "loss": 0.76,
+      "loss": 0.9315,
       "step": 6829
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7601152046554193e-05,
-      "loss": 0.8401,
+      "loss": 0.8873,
       "step": 6830
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7600386633057375e-05,
-      "loss": 0.7964,
+      "loss": 0.8159,
       "step": 6831
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7599621114115973e-05,
-      "loss": 0.9095,
+      "loss": 0.9538,
       "step": 6832
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7598855489740612e-05,
-      "loss": 0.9014,
+      "loss": 0.9511,
       "step": 6833
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.759808975994191e-05,
-      "loss": 0.9184,
+      "loss": 0.9302,
       "step": 6834
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7597323924730498e-05,
-      "loss": 0.6886,
+      "loss": 0.8394,
       "step": 6835
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7596557984116992e-05,
-      "loss": 0.7412,
+      "loss": 0.9191,
       "step": 6836
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7595791938112022e-05,
-      "loss": 0.8678,
+      "loss": 0.9218,
       "step": 6837
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.759502578672622e-05,
-      "loss": 0.7989,
+      "loss": 0.9274,
       "step": 6838
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.759425952997021e-05,
-      "loss": 0.8652,
+      "loss": 0.9063,
       "step": 6839
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.759349316785462e-05,
-      "loss": 0.8039,
+      "loss": 0.9326,
       "step": 6840
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7592726700390087e-05,
-      "loss": 0.771,
+      "loss": 0.8751,
       "step": 6841
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7591960127587242e-05,
-      "loss": 0.7777,
+      "loss": 0.9228,
       "step": 6842
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7591193449456728e-05,
-      "loss": 0.6811,
+      "loss": 0.7996,
       "step": 6843
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.759042666600917e-05,
-      "loss": 0.7621,
+      "loss": 0.8962,
       "step": 6844
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7589659777255213e-05,
-      "loss": 0.8065,
+      "loss": 0.8942,
       "step": 6845
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7588892783205496e-05,
-      "loss": 0.7891,
+      "loss": 0.9144,
       "step": 6846
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7588125683870658e-05,
-      "loss": 0.8649,
+      "loss": 1.0204,
       "step": 6847
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7587358479261344e-05,
-      "loss": 0.7762,
+      "loss": 0.9215,
       "step": 6848
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7586591169388193e-05,
-      "loss": 0.6766,
+      "loss": 0.8111,
       "step": 6849
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7585823754261853e-05,
-      "loss": 0.8562,
+      "loss": 0.9438,
       "step": 6850
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7585056233892978e-05,
-      "loss": 0.767,
+      "loss": 0.9133,
       "step": 6851
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7584288608292202e-05,
-      "loss": 0.9419,
+      "loss": 1.0234,
       "step": 6852
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7583520877470185e-05,
-      "loss": 0.7559,
+      "loss": 0.8954,
       "step": 6853
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7582753041437575e-05,
-      "loss": 0.7735,
+      "loss": 0.8513,
       "step": 6854
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7581985100205025e-05,
-      "loss": 0.6633,
+      "loss": 0.7554,
       "step": 6855
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.758121705378319e-05,
-      "loss": 0.8847,
+      "loss": 1.0013,
       "step": 6856
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7580448902182724e-05,
-      "loss": 0.8925,
+      "loss": 0.9203,
       "step": 6857
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7579680645414287e-05,
-      "loss": 0.7716,
+      "loss": 0.8717,
       "step": 6858
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7578912283488532e-05,
-      "loss": 0.8165,
+      "loss": 0.8974,
       "step": 6859
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7578143816416126e-05,
-      "loss": 0.6752,
+      "loss": 0.8352,
       "step": 6860
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7577375244207724e-05,
-      "loss": 0.6487,
+      "loss": 0.7677,
       "step": 6861
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.757660656687399e-05,
-      "loss": 0.8339,
+      "loss": 0.931,
       "step": 6862
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7575837784425594e-05,
-      "loss": 0.7357,
+      "loss": 0.9618,
       "step": 6863
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7575068896873192e-05,
-      "loss": 0.8135,
+      "loss": 0.8869,
       "step": 6864
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7574299904227457e-05,
-      "loss": 0.799,
+      "loss": 0.9472,
       "step": 6865
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7573530806499062e-05,
-      "loss": 0.8172,
+      "loss": 0.8741,
       "step": 6866
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7572761603698673e-05,
-      "loss": 0.8911,
+      "loss": 0.9299,
       "step": 6867
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.757199229583696e-05,
-      "loss": 0.8694,
+      "loss": 0.9583,
       "step": 6868
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7571222882924592e-05,
-      "loss": 0.7459,
+      "loss": 0.9663,
       "step": 6869
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7570453364972252e-05,
-      "loss": 0.8093,
+      "loss": 0.9369,
       "step": 6870
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7569683741990612e-05,
-      "loss": 0.8186,
+      "loss": 0.9553,
       "step": 6871
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.756891401399035e-05,
-      "loss": 0.8373,
+      "loss": 0.9086,
       "step": 6872
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7568144180982148e-05,
-      "loss": 0.8544,
+      "loss": 0.9528,
       "step": 6873
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.756737424297668e-05,
-      "loss": 0.9043,
+      "loss": 0.9659,
       "step": 6874
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.756660419998463e-05,
-      "loss": 0.8627,
+      "loss": 0.9421,
       "step": 6875
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7565834052016686e-05,
-      "loss": 0.7401,
+      "loss": 0.9011,
       "step": 6876
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7565063799083528e-05,
-      "loss": 0.8659,
+      "loss": 0.9247,
       "step": 6877
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.756429344119584e-05,
-      "loss": 0.894,
+      "loss": 0.9345,
       "step": 6878
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7563522978364317e-05,
-      "loss": 0.7624,
+      "loss": 0.8283,
       "step": 6879
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7562752410599644e-05,
-      "loss": 0.8237,
+      "loss": 0.8692,
       "step": 6880
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.756198173791251e-05,
-      "loss": 0.6624,
+      "loss": 0.8729,
       "step": 6881
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7561210960313607e-05,
-      "loss": 0.7886,
+      "loss": 0.9168,
       "step": 6882
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7560440077813632e-05,
-      "loss": 0.7042,
+      "loss": 0.8578,
       "step": 6883
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7559669090423277e-05,
-      "loss": 0.7252,
+      "loss": 0.8885,
       "step": 6884
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7558897998153238e-05,
-      "loss": 0.7347,
+      "loss": 0.8747,
       "step": 6885
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7558126801014218e-05,
-      "loss": 0.9087,
+      "loss": 0.9316,
       "step": 6886
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7557355499016907e-05,
-      "loss": 0.8991,
+      "loss": 0.9673,
       "step": 6887
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7556584092172015e-05,
-      "loss": 0.7879,
+      "loss": 0.9557,
       "step": 6888
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7555812580490234e-05,
-      "loss": 0.7749,
+      "loss": 0.8964,
       "step": 6889
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7555040963982278e-05,
-      "loss": 0.8865,
+      "loss": 1.0079,
       "step": 6890
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.755426924265885e-05,
-      "loss": 0.8069,
+      "loss": 0.8777,
       "step": 6891
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7553497416530653e-05,
-      "loss": 0.8678,
+      "loss": 0.9454,
       "step": 6892
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7552725485608394e-05,
-      "loss": 1.0218,
+      "loss": 1.0128,
       "step": 6893
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.755195344990279e-05,
-      "loss": 0.7237,
+      "loss": 0.7963,
       "step": 6894
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.755118130942454e-05,
-      "loss": 0.8347,
+      "loss": 1.0142,
       "step": 6895
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7550409064184367e-05,
-      "loss": 0.857,
+      "loss": 1.01,
       "step": 6896
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7549636714192977e-05,
-      "loss": 1.028,
+      "loss": 1.0773,
       "step": 6897
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7548864259461095e-05,
-      "loss": 0.8687,
+      "loss": 0.9022,
       "step": 6898
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.754809169999943e-05,
-      "loss": 1.1027,
+      "loss": 1.1418,
       "step": 6899
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7547319035818698e-05,
-      "loss": 0.8546,
+      "loss": 0.9226,
       "step": 6900
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7546546266929628e-05,
-      "loss": 0.6943,
+      "loss": 0.83,
       "step": 6901
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.754577339334293e-05,
-      "loss": 0.818,
+      "loss": 0.8902,
       "step": 6902
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7545000415069338e-05,
-      "loss": 0.9053,
+      "loss": 1.0123,
       "step": 6903
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7544227332119566e-05,
-      "loss": 0.8086,
+      "loss": 0.8915,
       "step": 6904
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.754345414450435e-05,
-      "loss": 0.7106,
+      "loss": 0.8928,
       "step": 6905
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7542680852234403e-05,
-      "loss": 0.7109,
+      "loss": 0.8313,
       "step": 6906
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7541907455320465e-05,
-      "loss": 0.8499,
+      "loss": 0.8976,
       "step": 6907
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.754113395377326e-05,
-      "loss": 0.6943,
+      "loss": 0.7756,
       "step": 6908
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7540360347603517e-05,
-      "loss": 0.753,
+      "loss": 0.8508,
       "step": 6909
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7539586636821978e-05,
-      "loss": 0.7734,
+      "loss": 0.9145,
       "step": 6910
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.753881282143937e-05,
-      "loss": 0.7799,
+      "loss": 0.8628,
       "step": 6911
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7538038901466433e-05,
-      "loss": 0.8342,
+      "loss": 0.9396,
       "step": 6912
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.75372648769139e-05,
-      "loss": 0.8554,
+      "loss": 1.0206,
       "step": 6913
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7536490747792507e-05,
-      "loss": 0.8163,
+      "loss": 0.8603,
       "step": 6914
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7535716514113003e-05,
-      "loss": 0.881,
+      "loss": 0.9838,
       "step": 6915
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.753494217588612e-05,
-      "loss": 0.8159,
+      "loss": 0.9358,
       "step": 6916
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7534167733122607e-05,
-      "loss": 0.8405,
+      "loss": 0.9523,
       "step": 6917
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7533393185833206e-05,
-      "loss": 0.7124,
+      "loss": 0.8706,
       "step": 6918
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7532618534028664e-05,
-      "loss": 0.8812,
+      "loss": 0.9478,
       "step": 6919
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7531843777719728e-05,
-      "loss": 0.7098,
+      "loss": 0.8412,
       "step": 6920
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.753106891691714e-05,
-      "loss": 0.7879,
+      "loss": 0.8157,
       "step": 6921
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.753029395163166e-05,
-      "loss": 0.7391,
+      "loss": 0.9404,
       "step": 6922
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7529518881874036e-05,
-      "loss": 0.7788,
+      "loss": 0.9274,
       "step": 6923
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7528743707655022e-05,
-      "loss": 0.8586,
+      "loss": 0.9445,
       "step": 6924
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.752796842898537e-05,
-      "loss": 0.9019,
+      "loss": 0.9059,
       "step": 6925
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7527193045875834e-05,
-      "loss": 0.7224,
+      "loss": 0.8624,
       "step": 6926
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7526417558337175e-05,
-      "loss": 0.7506,
+      "loss": 0.8768,
       "step": 6927
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7525641966380153e-05,
-      "loss": 0.7205,
+      "loss": 0.8924,
       "step": 6928
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7524866270015522e-05,
-      "loss": 0.8038,
+      "loss": 0.9499,
       "step": 6929
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7524090469254057e-05,
-      "loss": 0.8717,
+      "loss": 0.9561,
       "step": 6930
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7523314564106505e-05,
-      "loss": 0.7925,
+      "loss": 0.8996,
       "step": 6931
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7522538554583636e-05,
-      "loss": 0.8271,
+      "loss": 0.9884,
       "step": 6932
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.752176244069622e-05,
-      "loss": 0.9161,
+      "loss": 0.9987,
       "step": 6933
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7520986222455025e-05,
-      "loss": 0.8448,
+      "loss": 0.9447,
       "step": 6934
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7520209899870815e-05,
-      "loss": 0.9258,
+      "loss": 0.9579,
       "step": 6935
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7519433472954362e-05,
-      "loss": 0.8042,
+      "loss": 0.8893,
       "step": 6936
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.751865694171644e-05,
-      "loss": 0.8487,
+      "loss": 0.9661,
       "step": 6937
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7517880306167817e-05,
-      "loss": 0.8605,
+      "loss": 0.9402,
       "step": 6938
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7517103566319273e-05,
-      "loss": 0.6939,
+      "loss": 0.8644,
       "step": 6939
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.751632672218158e-05,
-      "loss": 0.7735,
+      "loss": 0.8788,
       "step": 6940
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.751554977376552e-05,
-      "loss": 0.8133,
+      "loss": 0.9319,
       "step": 6941
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7514772721081873e-05,
-      "loss": 0.8248,
+      "loss": 0.9564,
       "step": 6942
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7513995564141414e-05,
-      "loss": 0.861,
+      "loss": 0.9756,
       "step": 6943
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.751321830295493e-05,
-      "loss": 0.78,
+      "loss": 0.9031,
       "step": 6944
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7512440937533198e-05,
-      "loss": 0.7385,
+      "loss": 0.8484,
       "step": 6945
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7511663467887006e-05,
-      "loss": 0.8318,
+      "loss": 1.0005,
       "step": 6946
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7510885894027143e-05,
-      "loss": 0.6152,
+      "loss": 0.7957,
       "step": 6947
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7510108215964395e-05,
-      "loss": 0.7907,
+      "loss": 0.9146,
       "step": 6948
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7509330433709554e-05,
-      "loss": 0.7641,
+      "loss": 0.9074,
       "step": 6949
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7508552547273404e-05,
-      "loss": 0.8774,
+      "loss": 0.9391,
       "step": 6950
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.750777455666674e-05,
-      "loss": 0.9729,
+      "loss": 1.0134,
       "step": 6951
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7506996461900357e-05,
-      "loss": 0.7531,
+      "loss": 0.8972,
       "step": 6952
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.750621826298505e-05,
-      "loss": 0.8044,
+      "loss": 0.8936,
       "step": 6953
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7505439959931614e-05,
-      "loss": 0.8068,
+      "loss": 0.8504,
       "step": 6954
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7504661552750847e-05,
-      "loss": 0.8329,
+      "loss": 0.9747,
       "step": 6955
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.750388304145355e-05,
-      "loss": 0.8123,
+      "loss": 0.9877,
       "step": 6956
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7503104426050518e-05,
-      "loss": 0.7337,
+      "loss": 0.9066,
       "step": 6957
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7502325706552562e-05,
-      "loss": 0.8674,
+      "loss": 0.8555,
       "step": 6958
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7501546882970483e-05,
-      "loss": 0.7866,
+      "loss": 0.9184,
       "step": 6959
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7500767955315078e-05,
-      "loss": 0.9297,
+      "loss": 1.0255,
       "step": 6960
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7499988923597166e-05,
-      "loss": 0.9609,
+      "loss": 1.0211,
       "step": 6961
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7499209787827545e-05,
-      "loss": 0.6627,
+      "loss": 0.8618,
       "step": 6962
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7498430548017032e-05,
-      "loss": 0.8078,
+      "loss": 1.0193,
       "step": 6963
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7497651204176427e-05,
-      "loss": 0.9557,
+      "loss": 0.9988,
       "step": 6964
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7496871756316555e-05,
-      "loss": 0.7235,
+      "loss": 0.8864,
       "step": 6965
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7496092204448225e-05,
-      "loss": 0.737,
+      "loss": 0.8469,
       "step": 6966
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7495312548582245e-05,
-      "loss": 0.7642,
+      "loss": 0.8342,
       "step": 6967
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7494532788729444e-05,
-      "loss": 0.7577,
+      "loss": 0.9025,
       "step": 6968
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7493752924900632e-05,
-      "loss": 0.7133,
+      "loss": 0.8741,
       "step": 6969
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.749297295710663e-05,
-      "loss": 0.8876,
+      "loss": 1.0162,
       "step": 6970
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.749219288535826e-05,
-      "loss": 0.7052,
+      "loss": 0.8513,
       "step": 6971
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7491412709666344e-05,
-      "loss": 0.7933,
+      "loss": 0.8895,
       "step": 6972
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7490632430041703e-05,
-      "loss": 0.8806,
+      "loss": 0.9269,
       "step": 6973
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748985204649517e-05,
-      "loss": 0.788,
+      "loss": 0.8792,
       "step": 6974
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7489071559037563e-05,
-      "loss": 0.795,
+      "loss": 0.8993,
       "step": 6975
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748829096767971e-05,
-      "loss": 1.0934,
+      "loss": 1.1229,
       "step": 6976
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748751027243245e-05,
-      "loss": 0.8371,
+      "loss": 0.8875,
       "step": 6977
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7486729473306608e-05,
-      "loss": 0.6825,
+      "loss": 0.7894,
       "step": 6978
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7485948570313015e-05,
-      "loss": 0.8197,
+      "loss": 0.9042,
       "step": 6979
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7485167563462505e-05,
-      "loss": 0.9635,
+      "loss": 1.0416,
       "step": 6980
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748438645276592e-05,
-      "loss": 0.8226,
+      "loss": 0.8744,
       "step": 6981
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7483605238234088e-05,
-      "loss": 0.8011,
+      "loss": 0.8739,
       "step": 6982
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7482823919877853e-05,
-      "loss": 0.6633,
+      "loss": 0.8454,
       "step": 6983
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7482042497708053e-05,
-      "loss": 0.7957,
+      "loss": 0.9202,
       "step": 6984
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748126097173553e-05,
-      "loss": 0.7715,
+      "loss": 0.9573,
       "step": 6985
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.748047934197112e-05,
-      "loss": 0.6803,
+      "loss": 0.7442,
       "step": 6986
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7479697608425677e-05,
-      "loss": 0.5837,
+      "loss": 0.8066,
       "step": 6987
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.747891577111004e-05,
-      "loss": 0.7507,
+      "loss": 0.8416,
       "step": 6988
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.747813383003506e-05,
-      "loss": 0.6944,
+      "loss": 0.8741,
       "step": 6989
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7477351785211586e-05,
-      "loss": 0.9416,
+      "loss": 1.0029,
       "step": 6990
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7476569636650458e-05,
-      "loss": 0.8146,
+      "loss": 0.9112,
       "step": 6991
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.747578738436254e-05,
-      "loss": 0.7006,
+      "loss": 0.7736,
       "step": 6992
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7475005028358675e-05,
-      "loss": 0.8028,
+      "loss": 0.9405,
       "step": 6993
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7474222568649724e-05,
-      "loss": 0.7373,
+      "loss": 0.8837,
       "step": 6994
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7473440005246536e-05,
-      "loss": 0.7322,
+      "loss": 0.8531,
       "step": 6995
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7472657338159973e-05,
-      "loss": 0.8066,
+      "loss": 0.9303,
       "step": 6996
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.747187456740089e-05,
-      "loss": 0.7105,
+      "loss": 0.8519,
       "step": 6997
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7471091692980153e-05,
-      "loss": 0.9026,
+      "loss": 0.9259,
       "step": 6998
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7470308714908614e-05,
-      "loss": 0.8419,
+      "loss": 0.918,
       "step": 6999
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7469525633197143e-05,
-      "loss": 0.6685,
+      "loss": 0.8931,
       "step": 7000
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.74687424478566e-05,
-      "loss": 0.828,
+      "loss": 0.9844,
       "step": 7001
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7467959158897856e-05,
-      "loss": 0.8562,
+      "loss": 0.9783,
       "step": 7002
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746717576633177e-05,
-      "loss": 0.7518,
+      "loss": 0.8782,
       "step": 7003
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746639227016922e-05,
-      "loss": 0.8237,
+      "loss": 1.0042,
       "step": 7004
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7465608670421065e-05,
-      "loss": 0.8571,
+      "loss": 0.9841,
       "step": 7005
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7464824967098185e-05,
-      "loss": 0.8413,
+      "loss": 0.9474,
       "step": 7006
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746404116021145e-05,
-      "loss": 0.7943,
+      "loss": 0.9238,
       "step": 7007
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.7463257249771732e-05,
-      "loss": 0.8289,
+      "loss": 0.9125,
       "step": 7008
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746247323578991e-05,
-      "loss": 1.0504,
+      "loss": 1.1134,
       "step": 7009
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746168911827686e-05,
-      "loss": 0.7888,
+      "loss": 0.9662,
       "step": 7010
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.746090489724346e-05,
-      "loss": 0.8506,
+      "loss": 0.9314,
       "step": 7011
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.746012057270059e-05,
-      "loss": 0.8091,
+      "loss": 0.9832,
       "step": 7012
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7459336144659132e-05,
-      "loss": 0.7651,
+      "loss": 0.9569,
       "step": 7013
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7458551613129968e-05,
-      "loss": 0.7332,
+      "loss": 0.9295,
       "step": 7014
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7457766978123982e-05,
-      "loss": 0.8806,
+      "loss": 0.9244,
       "step": 7015
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.745698223965206e-05,
-      "loss": 0.8341,
+      "loss": 0.9704,
       "step": 7016
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7456197397725092e-05,
-      "loss": 0.7621,
+      "loss": 0.843,
       "step": 7017
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7455412452353963e-05,
-      "loss": 0.7379,
+      "loss": 0.9244,
       "step": 7018
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7454627403549564e-05,
-      "loss": 0.875,
+      "loss": 0.9227,
       "step": 7019
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7453842251322787e-05,
-      "loss": 0.8479,
+      "loss": 0.8921,
       "step": 7020
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7453056995684523e-05,
-      "loss": 0.7712,
+      "loss": 0.848,
       "step": 7021
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7452271636645667e-05,
-      "loss": 0.7935,
+      "loss": 0.9181,
       "step": 7022
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7451486174217115e-05,
-      "loss": 0.8133,
+      "loss": 0.9424,
       "step": 7023
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.745070060840977e-05,
-      "loss": 0.9692,
+      "loss": 1.0366,
       "step": 7024
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.744991493923452e-05,
-      "loss": 0.8401,
+      "loss": 0.8746,
       "step": 7025
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7449129166702272e-05,
-      "loss": 0.9003,
+      "loss": 0.9651,
       "step": 7026
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7448343290823925e-05,
-      "loss": 0.8699,
+      "loss": 0.9776,
       "step": 7027
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7447557311610385e-05,
-      "loss": 0.7158,
+      "loss": 0.8479,
       "step": 7028
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.744677122907255e-05,
-      "loss": 0.7481,
+      "loss": 0.8767,
       "step": 7029
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7445985043221334e-05,
-      "loss": 0.8701,
+      "loss": 0.9572,
       "step": 7030
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7445198754067637e-05,
-      "loss": 0.716,
+      "loss": 0.8193,
       "step": 7031
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.744441236162237e-05,
-      "loss": 0.8082,
+      "loss": 0.9496,
       "step": 7032
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7443625865896447e-05,
-      "loss": 0.8286,
+      "loss": 0.9278,
       "step": 7033
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7442839266900775e-05,
-      "loss": 0.9286,
+      "loss": 0.9513,
       "step": 7034
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7442052564646266e-05,
-      "loss": 0.9365,
+      "loss": 0.9863,
       "step": 7035
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.744126575914384e-05,
-      "loss": 0.7451,
+      "loss": 0.987,
       "step": 7036
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7440478850404406e-05,
-      "loss": 0.9032,
+      "loss": 0.9897,
       "step": 7037
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7439691838438887e-05,
-      "loss": 0.8589,
+      "loss": 0.9137,
       "step": 7038
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.74389047232582e-05,
-      "loss": 0.7641,
+      "loss": 0.8246,
       "step": 7039
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.743811750487326e-05,
-      "loss": 0.6729,
+      "loss": 0.8852,
       "step": 7040
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7437330183294998e-05,
-      "loss": 0.9104,
+      "loss": 0.9217,
       "step": 7041
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7436542758534328e-05,
-      "loss": 0.7666,
+      "loss": 0.9271,
       "step": 7042
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7435755230602182e-05,
-      "loss": 0.757,
+      "loss": 0.9225,
       "step": 7043
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.743496759950948e-05,
-      "loss": 0.6986,
+      "loss": 0.9149,
       "step": 7044
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.743417986526715e-05,
-      "loss": 0.8226,
+      "loss": 0.9433,
       "step": 7045
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7433392027886123e-05,
-      "loss": 0.7655,
+      "loss": 0.8902,
       "step": 7046
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.743260408737733e-05,
-      "loss": 0.6907,
+      "loss": 0.7934,
       "step": 7047
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.74318160437517e-05,
-      "loss": 0.7354,
+      "loss": 0.8647,
       "step": 7048
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7431027897020167e-05,
-      "loss": 0.9302,
+      "loss": 0.9796,
       "step": 7049
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7430239647193668e-05,
-      "loss": 0.8998,
+      "loss": 0.9767,
       "step": 7050
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742945129428313e-05,
-      "loss": 0.7768,
+      "loss": 0.8707,
       "step": 7051
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.74286628382995e-05,
-      "loss": 0.9274,
+      "loss": 0.9738,
       "step": 7052
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742787427925371e-05,
-      "loss": 0.9181,
+      "loss": 1.0013,
       "step": 7053
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7427085617156706e-05,
-      "loss": 0.8062,
+      "loss": 0.8773,
       "step": 7054
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7426296852019425e-05,
-      "loss": 0.9456,
+      "loss": 0.9325,
       "step": 7055
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742550798385281e-05,
-      "loss": 0.9643,
+      "loss": 1.0136,
       "step": 7056
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7424719012667816e-05,
-      "loss": 0.8069,
+      "loss": 0.9039,
       "step": 7057
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742392993847537e-05,
-      "loss": 0.7924,
+      "loss": 0.9215,
       "step": 7058
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7423140761286435e-05,
-      "loss": 1.0001,
+      "loss": 1.0948,
       "step": 7059
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742235148111195e-05,
-      "loss": 0.878,
+      "loss": 0.9241,
       "step": 7060
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7421562097962873e-05,
-      "loss": 0.8374,
+      "loss": 0.9447,
       "step": 7061
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.742077261185015e-05,
-      "loss": 0.9151,
+      "loss": 0.9718,
       "step": 7062
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7419983022784735e-05,
-      "loss": 0.8496,
+      "loss": 0.9209,
       "step": 7063
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7419193330777582e-05,
-      "loss": 0.7844,
+      "loss": 0.8154,
       "step": 7064
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7418403535839653e-05,
-      "loss": 0.8655,
+      "loss": 0.9576,
       "step": 7065
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7417613637981895e-05,
-      "loss": 0.7513,
+      "loss": 0.9071,
       "step": 7066
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7416823637215275e-05,
-      "loss": 0.7594,
+      "loss": 0.9348,
       "step": 7067
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7416033533550747e-05,
-      "loss": 0.8932,
+      "loss": 0.9349,
       "step": 7068
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7415243326999278e-05,
-      "loss": 0.8105,
+      "loss": 0.8913,
       "step": 7069
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.741445301757183e-05,
-      "loss": 0.7823,
+      "loss": 0.8068,
       "step": 7070
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7413662605279365e-05,
-      "loss": 0.9528,
+      "loss": 0.9894,
       "step": 7071
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7412872090132848e-05,
-      "loss": 0.8299,
+      "loss": 0.8562,
       "step": 7072
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.741208147214325e-05,
-      "loss": 0.7726,
+      "loss": 0.8892,
       "step": 7073
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7411290751321538e-05,
-      "loss": 0.7226,
+      "loss": 0.9007,
       "step": 7074
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7410499927678682e-05,
-      "loss": 0.8745,
+      "loss": 0.8184,
       "step": 7075
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7409709001225655e-05,
-      "loss": 0.872,
+      "loss": 0.9025,
       "step": 7076
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7408917971973426e-05,
-      "loss": 0.7348,
+      "loss": 0.8302,
       "step": 7077
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7408126839932973e-05,
-      "loss": 0.837,
+      "loss": 0.9521,
       "step": 7078
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7407335605115273e-05,
-      "loss": 0.9345,
+      "loss": 0.9518,
       "step": 7079
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7406544267531295e-05,
-      "loss": 0.6241,
+      "loss": 0.749,
       "step": 7080
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7405752827192028e-05,
-      "loss": 0.8548,
+      "loss": 0.8958,
       "step": 7081
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.740496128410845e-05,
-      "loss": 0.8557,
+      "loss": 0.9389,
       "step": 7082
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7404169638291537e-05,
-      "loss": 0.6436,
+      "loss": 0.8343,
       "step": 7083
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.740337788975228e-05,
-      "loss": 0.8024,
+      "loss": 0.847,
       "step": 7084
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7402586038501652e-05,
-      "loss": 0.8472,
+      "loss": 0.9531,
       "step": 7085
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.740179408455065e-05,
-      "loss": 0.8236,
+      "loss": 0.9298,
       "step": 7086
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7401002027910254e-05,
-      "loss": 0.9982,
+      "loss": 0.9877,
       "step": 7087
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.740020986859146e-05,
-      "loss": 0.9006,
+      "loss": 0.9861,
       "step": 7088
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739941760660525e-05,
-      "loss": 0.7275,
+      "loss": 0.8789,
       "step": 7089
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739862524196262e-05,
-      "loss": 0.8767,
+      "loss": 0.9818,
       "step": 7090
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739783277467456e-05,
-      "loss": 0.8149,
+      "loss": 0.8719,
       "step": 7091
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739704020475207e-05,
-      "loss": 0.8636,
+      "loss": 0.9277,
       "step": 7092
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7396247532206142e-05,
-      "loss": 0.8303,
+      "loss": 0.9687,
       "step": 7093
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7395454757047772e-05,
-      "loss": 0.7954,
+      "loss": 0.9909,
       "step": 7094
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739466187928796e-05,
-      "loss": 0.7463,
+      "loss": 0.8946,
       "step": 7095
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7393868898937705e-05,
-      "loss": 0.8464,
+      "loss": 0.9681,
       "step": 7096
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739307581600801e-05,
-      "loss": 0.7551,
+      "loss": 0.9172,
       "step": 7097
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739228263050988e-05,
-      "loss": 0.7352,
+      "loss": 0.8988,
       "step": 7098
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7391489342454314e-05,
-      "loss": 0.8176,
+      "loss": 0.9149,
       "step": 7099
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.739069595185232e-05,
-      "loss": 0.6908,
+      "loss": 0.7797,
       "step": 7100
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.738990245871491e-05,
-      "loss": 0.9731,
+      "loss": 1.0319,
       "step": 7101
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7389108863053086e-05,
-      "loss": 0.8211,
+      "loss": 0.8618,
       "step": 7102
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.738831516487786e-05,
-      "loss": 0.7339,
+      "loss": 0.805,
       "step": 7103
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.738752136420024e-05,
-      "loss": 0.8904,
+      "loss": 1.0074,
       "step": 7104
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.738672746103125e-05,
-      "loss": 1.0106,
+      "loss": 1.0146,
       "step": 7105
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7385933455381893e-05,
-      "loss": 0.6738,
+      "loss": 0.8456,
       "step": 7106
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7385139347263187e-05,
-      "loss": 0.8363,
+      "loss": 0.906,
       "step": 7107
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7384345136686153e-05,
-      "loss": 0.7191,
+      "loss": 0.8605,
       "step": 7108
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7383550823661805e-05,
-      "loss": 0.9301,
+      "loss": 0.9544,
       "step": 7109
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7382756408201168e-05,
-      "loss": 0.7294,
+      "loss": 0.9092,
       "step": 7110
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.738196189031526e-05,
-      "loss": 0.7719,
+      "loss": 0.8662,
       "step": 7111
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.73811672700151e-05,
-      "loss": 0.9217,
+      "loss": 0.9999,
       "step": 7112
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7380372547311722e-05,
-      "loss": 0.8444,
+      "loss": 0.9424,
       "step": 7113
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7379577722216147e-05,
-      "loss": 0.5874,
+      "loss": 0.6926,
       "step": 7114
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7378782794739398e-05,
-      "loss": 0.8277,
+      "loss": 0.9111,
       "step": 7115
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7377987764892506e-05,
-      "loss": 0.8173,
+      "loss": 0.9006,
       "step": 7116
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7377192632686505e-05,
-      "loss": 0.9082,
+      "loss": 0.94,
       "step": 7117
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.737639739813242e-05,
-      "loss": 0.681,
+      "loss": 0.8428,
       "step": 7118
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.737560206124129e-05,
-      "loss": 0.8185,
+      "loss": 0.9309,
       "step": 7119
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7374806622024145e-05,
-      "loss": 0.8641,
+      "loss": 0.9993,
       "step": 7120
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.737401108049202e-05,
-      "loss": 0.7161,
+      "loss": 0.8686,
       "step": 7121
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7373215436655956e-05,
-      "loss": 0.8375,
+      "loss": 0.9035,
       "step": 7122
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7372419690526987e-05,
-      "loss": 0.7278,
+      "loss": 0.865,
       "step": 7123
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7371623842116156e-05,
-      "loss": 0.873,
+      "loss": 0.9616,
       "step": 7124
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7370827891434503e-05,
-      "loss": 1.0042,
+      "loss": 1.0034,
       "step": 7125
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7370031838493072e-05,
-      "loss": 0.7031,
+      "loss": 0.854,
       "step": 7126
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7369235683302904e-05,
-      "loss": 0.729,
+      "loss": 0.8582,
       "step": 7127
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7368439425875045e-05,
-      "loss": 0.7933,
+      "loss": 0.9095,
       "step": 7128
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.736764306622055e-05,
-      "loss": 0.7744,
+      "loss": 0.8909,
       "step": 7129
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7366846604350453e-05,
-      "loss": 0.8449,
+      "loss": 0.9227,
       "step": 7130
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7366050040275813e-05,
-      "loss": 0.8022,
+      "loss": 0.9309,
       "step": 7131
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7365253374007682e-05,
-      "loss": 0.7462,
+      "loss": 0.8656,
       "step": 7132
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.736445660555711e-05,
-      "loss": 0.866,
+      "loss": 0.9609,
       "step": 7133
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7363659734935156e-05,
-      "loss": 0.8607,
+      "loss": 1.0003,
       "step": 7134
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7362862762152862e-05,
-      "loss": 0.821,
+      "loss": 0.891,
       "step": 7135
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7362065687221298e-05,
-      "loss": 0.7564,
+      "loss": 0.9144,
       "step": 7136
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.736126851015152e-05,
-      "loss": 0.8309,
+      "loss": 0.9298,
       "step": 7137
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7360471230954583e-05,
-      "loss": 0.7335,
+      "loss": 0.892,
       "step": 7138
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.735967384964155e-05,
-      "loss": 0.8032,
+      "loss": 0.9562,
       "step": 7139
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.735887636622349e-05,
-      "loss": 0.8293,
+      "loss": 0.9389,
       "step": 7140
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7358078780711452e-05,
-      "loss": 0.7616,
+      "loss": 0.8384,
       "step": 7141
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.735728109311652e-05,
-      "loss": 0.7643,
+      "loss": 0.8693,
       "step": 7142
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7356483303449746e-05,
-      "loss": 0.8081,
+      "loss": 0.9683,
       "step": 7143
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7355685411722203e-05,
-      "loss": 0.8608,
+      "loss": 0.9946,
       "step": 7144
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7354887417944962e-05,
-      "loss": 0.7561,
+      "loss": 0.8817,
       "step": 7145
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7354089322129097e-05,
-      "loss": 0.832,
+      "loss": 0.8937,
       "step": 7146
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7353291124285674e-05,
-      "loss": 0.732,
+      "loss": 0.8206,
       "step": 7147
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.735249282442577e-05,
-      "loss": 0.7999,
+      "loss": 0.89,
       "step": 7148
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.735169442256046e-05,
-      "loss": 0.6995,
+      "loss": 0.9591,
       "step": 7149
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7350895918700818e-05,
-      "loss": 0.7784,
+      "loss": 0.8609,
       "step": 7150
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7350097312857928e-05,
-      "loss": 0.7779,
+      "loss": 0.9395,
       "step": 7151
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7349298605042862e-05,
-      "loss": 0.8629,
+      "loss": 0.934,
       "step": 7152
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734849979526671e-05,
-      "loss": 0.8875,
+      "loss": 0.9423,
       "step": 7153
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7347700883540552e-05,
-      "loss": 0.7888,
+      "loss": 0.9087,
       "step": 7154
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734690186987546e-05,
-      "loss": 0.887,
+      "loss": 0.9955,
       "step": 7155
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7346102754282537e-05,
-      "loss": 0.8617,
+      "loss": 0.9915,
       "step": 7156
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7345303536772856e-05,
-      "loss": 0.7896,
+      "loss": 0.9041,
       "step": 7157
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734450421735751e-05,
-      "loss": 0.8248,
+      "loss": 0.8712,
       "step": 7158
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734370479604759e-05,
-      "loss": 0.8278,
+      "loss": 0.983,
       "step": 7159
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7342905272854188e-05,
-      "loss": 0.7173,
+      "loss": 0.8418,
       "step": 7160
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734210564778839e-05,
-      "loss": 0.7668,
+      "loss": 0.9061,
       "step": 7161
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7341305920861294e-05,
-      "loss": 0.714,
+      "loss": 0.8012,
       "step": 7162
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.734050609208399e-05,
-      "loss": 0.848,
+      "loss": 0.9055,
       "step": 7163
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.733970616146759e-05,
-      "loss": 0.8575,
+      "loss": 0.9197,
       "step": 7164
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7338906129023173e-05,
-      "loss": 0.907,
+      "loss": 1.0163,
       "step": 7165
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7338105994761846e-05,
-      "loss": 0.712,
+      "loss": 0.9276,
       "step": 7166
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.733730575869471e-05,
-      "loss": 0.7702,
+      "loss": 0.85,
       "step": 7167
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7336505420832866e-05,
-      "loss": 0.7993,
+      "loss": 0.9284,
       "step": 7168
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7335704981187423e-05,
-      "loss": 0.8224,
+      "loss": 0.9938,
       "step": 7169
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7334904439769474e-05,
-      "loss": 0.8677,
+      "loss": 0.9413,
       "step": 7170
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.733410379659014e-05,
-      "loss": 0.8238,
+      "loss": 0.9467,
       "step": 7171
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7333303051660516e-05,
-      "loss": 0.7159,
+      "loss": 0.8632,
       "step": 7172
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.733250220499172e-05,
-      "loss": 0.8085,
+      "loss": 0.8887,
       "step": 7173
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.733170125659486e-05,
-      "loss": 0.7469,
+      "loss": 0.9286,
       "step": 7174
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7330900206481046e-05,
-      "loss": 0.8663,
+      "loss": 0.9367,
       "step": 7175
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7330099054661393e-05,
-      "loss": 0.7952,
+      "loss": 0.9438,
       "step": 7176
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732929780114702e-05,
-      "loss": 0.8026,
+      "loss": 0.8945,
       "step": 7177
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732849644594903e-05,
-      "loss": 0.7338,
+      "loss": 0.8976,
       "step": 7178
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7327694989078553e-05,
-      "loss": 0.6808,
+      "loss": 0.7447,
       "step": 7179
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7326893430546707e-05,
-      "loss": 0.8867,
+      "loss": 0.9581,
       "step": 7180
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732609177036461e-05,
-      "loss": 0.7867,
+      "loss": 0.8666,
       "step": 7181
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7325290008543382e-05,
-      "loss": 0.8364,
+      "loss": 0.9979,
       "step": 7182
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732448814509415e-05,
-      "loss": 0.6872,
+      "loss": 0.8668,
       "step": 7183
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7323686180028037e-05,
-      "loss": 0.9616,
+      "loss": 0.9466,
       "step": 7184
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732288411335617e-05,
-      "loss": 0.8431,
+      "loss": 0.923,
       "step": 7185
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7322081945089672e-05,
-      "loss": 0.7081,
+      "loss": 0.7909,
       "step": 7186
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.732127967523968e-05,
-      "loss": 0.7494,
+      "loss": 0.9037,
       "step": 7187
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7320477303817316e-05,
-      "loss": 0.9134,
+      "loss": 0.934,
       "step": 7188
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.731967483083372e-05,
-      "loss": 0.7548,
+      "loss": 0.8645,
       "step": 7189
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7318872256300018e-05,
-      "loss": 0.7128,
+      "loss": 0.8816,
       "step": 7190
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.731806958022735e-05,
-      "loss": 0.7504,
+      "loss": 0.9763,
       "step": 7191
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7317266802626847e-05,
-      "loss": 0.8311,
+      "loss": 0.9856,
       "step": 7192
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.731646392350965e-05,
-      "loss": 0.8654,
+      "loss": 0.9301,
       "step": 7193
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7315660942886897e-05,
-      "loss": 0.704,
+      "loss": 0.8961,
       "step": 7194
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7314857860769726e-05,
-      "loss": 0.7298,
+      "loss": 0.8383,
       "step": 7195
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7314054677169284e-05,
-      "loss": 0.6728,
+      "loss": 0.8093,
       "step": 7196
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.731325139209671e-05,
-      "loss": 0.9613,
+      "loss": 1.0186,
       "step": 7197
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7312448005563145e-05,
-      "loss": 0.8074,
+      "loss": 0.9275,
       "step": 7198
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7311644517579746e-05,
-      "loss": 0.7947,
+      "loss": 0.859,
       "step": 7199
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7310840928157646e-05,
-      "loss": 0.8911,
+      "loss": 0.9985,
       "step": 7200
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7310037237308006e-05,
-      "loss": 0.7747,
+      "loss": 0.8646,
       "step": 7201
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7309233445041967e-05,
-      "loss": 0.6894,
+      "loss": 0.8425,
       "step": 7202
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7308429551370687e-05,
-      "loss": 0.7014,
+      "loss": 0.8166,
       "step": 7203
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7307625556305316e-05,
-      "loss": 0.8415,
+      "loss": 0.9248,
       "step": 7204
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.730682145985701e-05,
-      "loss": 0.8397,
+      "loss": 0.9378,
       "step": 7205
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.730601726203692e-05,
-      "loss": 0.782,
+      "loss": 0.9455,
       "step": 7206
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7305212962856206e-05,
-      "loss": 0.8153,
+      "loss": 0.8783,
       "step": 7207
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.730440856232603e-05,
-      "loss": 0.8899,
+      "loss": 1.0069,
       "step": 7208
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7303604060457545e-05,
-      "loss": 0.799,
+      "loss": 1.0217,
       "step": 7209
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.730279945726192e-05,
-      "loss": 0.8364,
+      "loss": 0.9496,
       "step": 7210
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7301994752750315e-05,
-      "loss": 0.7085,
+      "loss": 0.927,
       "step": 7211
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.730118994693389e-05,
-      "loss": 0.7672,
+      "loss": 0.8253,
       "step": 7212
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7300385039823813e-05,
-      "loss": 0.6441,
+      "loss": 0.8213,
       "step": 7213
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7299580031431253e-05,
-      "loss": 0.6662,
+      "loss": 0.8997,
       "step": 7214
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7298774921767376e-05,
-      "loss": 0.7616,
+      "loss": 0.9368,
       "step": 7215
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7297969710843352e-05,
-      "loss": 0.804,
+      "loss": 0.8584,
       "step": 7216
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7297164398670354e-05,
-      "loss": 0.7018,
+      "loss": 0.8893,
       "step": 7217
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.729635898525955e-05,
-      "loss": 0.7894,
+      "loss": 0.8691,
       "step": 7218
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7295553470622123e-05,
-      "loss": 0.7121,
+      "loss": 0.83,
       "step": 7219
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.729474785476924e-05,
-      "loss": 0.8844,
+      "loss": 0.9432,
       "step": 7220
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7293942137712082e-05,
-      "loss": 0.857,
+      "loss": 0.9702,
       "step": 7221
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.729313631946182e-05,
-      "loss": 0.8118,
+      "loss": 0.9832,
       "step": 7222
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7292330400029645e-05,
-      "loss": 0.7803,
+      "loss": 0.8888,
       "step": 7223
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.729152437942673e-05,
-      "loss": 0.9287,
+      "loss": 0.9942,
       "step": 7224
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7290718257664258e-05,
-      "loss": 0.86,
+      "loss": 0.9866,
       "step": 7225
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7289912034753418e-05,
-      "loss": 0.8601,
+      "loss": 0.9384,
       "step": 7226
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728910571070539e-05,
-      "loss": 0.8032,
+      "loss": 0.8411,
       "step": 7227
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728829928553136e-05,
-      "loss": 0.9371,
+      "loss": 0.9178,
       "step": 7228
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728749275924252e-05,
-      "loss": 0.7603,
+      "loss": 0.8719,
       "step": 7229
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728668613185006e-05,
-      "loss": 0.7171,
+      "loss": 0.8229,
       "step": 7230
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7285879403365166e-05,
-      "loss": 0.8223,
+      "loss": 0.8655,
       "step": 7231
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728507257379903e-05,
-      "loss": 0.719,
+      "loss": 0.8655,
       "step": 7232
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728426564316285e-05,
-      "loss": 0.7837,
+      "loss": 0.8477,
       "step": 7233
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.728345861146782e-05,
-      "loss": 0.9272,
+      "loss": 1.0,
       "step": 7234
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7282651478725135e-05,
-      "loss": 0.8025,
+      "loss": 0.935,
       "step": 7235
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7281844244945992e-05,
-      "loss": 0.8528,
+      "loss": 0.9332,
       "step": 7236
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.72810369101416e-05,
-      "loss": 0.7257,
+      "loss": 0.929,
       "step": 7237
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7280229474323143e-05,
-      "loss": 0.6726,
+      "loss": 0.8458,
       "step": 7238
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7279421937501833e-05,
-      "loss": 0.9001,
+      "loss": 1.0593,
       "step": 7239
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7278614299688872e-05,
-      "loss": 0.7567,
+      "loss": 0.8414,
       "step": 7240
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7277806560895466e-05,
-      "loss": 0.8296,
+      "loss": 0.9998,
       "step": 7241
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7276998721132815e-05,
-      "loss": 0.8512,
+      "loss": 0.994,
       "step": 7242
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7276190780412134e-05,
-      "loss": 0.6471,
+      "loss": 0.8641,
       "step": 7243
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.727538273874463e-05,
-      "loss": 0.9717,
+      "loss": 1.0748,
       "step": 7244
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7274574596141515e-05,
-      "loss": 0.7533,
+      "loss": 0.9269,
       "step": 7245
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7273766352613996e-05,
-      "loss": 0.7967,
+      "loss": 0.8416,
       "step": 7246
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7272958008173293e-05,
-      "loss": 0.8906,
+      "loss": 0.9366,
       "step": 7247
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7272149562830613e-05,
-      "loss": 0.6955,
+      "loss": 0.8965,
       "step": 7248
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7271341016597178e-05,
-      "loss": 0.8229,
+      "loss": 0.9322,
       "step": 7249
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.72705323694842e-05,
-      "loss": 0.9249,
+      "loss": 1.0234,
       "step": 7250
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.72697236215029e-05,
-      "loss": 0.7302,
+      "loss": 0.889,
       "step": 7251
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7268914772664503e-05,
-      "loss": 0.7212,
+      "loss": 0.8411,
       "step": 7252
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7268105822980222e-05,
-      "loss": 0.889,
+      "loss": 0.9388,
       "step": 7253
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.726729677246129e-05,
-      "loss": 0.8375,
+      "loss": 0.8907,
       "step": 7254
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7266487621118922e-05,
-      "loss": 0.7046,
+      "loss": 0.7906,
       "step": 7255
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7265678368964353e-05,
-      "loss": 0.9124,
+      "loss": 0.9374,
       "step": 7256
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7264869016008804e-05,
-      "loss": 0.8927,
+      "loss": 0.9382,
       "step": 7257
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7264059562263504e-05,
-      "loss": 0.7615,
+      "loss": 0.9004,
       "step": 7258
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7263250007739683e-05,
-      "loss": 0.8109,
+      "loss": 0.9584,
       "step": 7259
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.726244035244857e-05,
-      "loss": 0.8066,
+      "loss": 0.9559,
       "step": 7260
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7261630596401407e-05,
-      "loss": 0.9622,
+      "loss": 0.9795,
       "step": 7261
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.726082073960942e-05,
-      "loss": 0.7702,
+      "loss": 0.9219,
       "step": 7262
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7260010782083848e-05,
-      "loss": 0.7326,
+      "loss": 0.8714,
       "step": 7263
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7259200723835924e-05,
-      "loss": 0.7431,
+      "loss": 0.8852,
       "step": 7264
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7258390564876894e-05,
-      "loss": 0.8081,
+      "loss": 0.9498,
       "step": 7265
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.725758030521799e-05,
-      "loss": 0.8326,
+      "loss": 0.8936,
       "step": 7266
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7256769944870456e-05,
-      "loss": 0.9339,
+      "loss": 1.0067,
       "step": 7267
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7255959483845538e-05,
-      "loss": 0.8587,
+      "loss": 0.9863,
       "step": 7268
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7255148922154478e-05,
-      "loss": 0.8419,
+      "loss": 0.9025,
       "step": 7269
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7254338259808518e-05,
-      "loss": 0.7876,
+      "loss": 0.9129,
       "step": 7270
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7253527496818905e-05,
-      "loss": 0.8853,
+      "loss": 0.9255,
       "step": 7271
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.725271663319689e-05,
-      "loss": 0.8925,
+      "loss": 0.9062,
       "step": 7272
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7251905668953727e-05,
-      "loss": 0.7301,
+      "loss": 0.8866,
       "step": 7273
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7251094604100654e-05,
-      "loss": 0.7491,
+      "loss": 0.8777,
       "step": 7274
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7250283438648937e-05,
-      "loss": 0.8336,
+      "loss": 0.9951,
       "step": 7275
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7249472172609826e-05,
-      "loss": 0.8839,
+      "loss": 0.9754,
       "step": 7276
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7248660805994567e-05,
-      "loss": 0.7905,
+      "loss": 0.8255,
       "step": 7277
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.724784933881443e-05,
-      "loss": 0.7104,
+      "loss": 0.8389,
       "step": 7278
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7247037771080662e-05,
-      "loss": 0.7822,
+      "loss": 0.9302,
       "step": 7279
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.724622610280453e-05,
-      "loss": 0.8757,
+      "loss": 0.946,
       "step": 7280
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7245414333997296e-05,
-      "loss": 0.7464,
+      "loss": 0.8571,
       "step": 7281
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.724460246467021e-05,
-      "loss": 0.6997,
+      "loss": 0.8882,
       "step": 7282
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7243790494834546e-05,
-      "loss": 0.854,
+      "loss": 0.9097,
       "step": 7283
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.724297842450157e-05,
-      "loss": 0.7395,
+      "loss": 0.9121,
       "step": 7284
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.7242166253682542e-05,
-      "loss": 0.8081,
+      "loss": 0.8631,
       "step": 7285
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.724135398238873e-05,
-      "loss": 0.688,
+      "loss": 0.872,
       "step": 7286
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.724054161063141e-05,
-      "loss": 0.8068,
+      "loss": 0.9156,
       "step": 7287
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.723972913842185e-05,
-      "loss": 0.9091,
+      "loss": 1.0256,
       "step": 7288
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7238916565771316e-05,
-      "loss": 0.7608,
+      "loss": 0.8282,
       "step": 7289
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7238103892691084e-05,
-      "loss": 0.9324,
+      "loss": 0.9971,
       "step": 7290
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7237291119192434e-05,
-      "loss": 0.8556,
+      "loss": 0.9014,
       "step": 7291
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7236478245286633e-05,
-      "loss": 0.6117,
+      "loss": 0.8464,
       "step": 7292
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.723566527098497e-05,
-      "loss": 0.7052,
+      "loss": 0.81,
       "step": 7293
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7234852196298715e-05,
-      "loss": 0.7876,
+      "loss": 0.9556,
       "step": 7294
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7234039021239146e-05,
-      "loss": 0.7618,
+      "loss": 0.8677,
       "step": 7295
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7233225745817556e-05,
-      "loss": 0.9582,
+      "loss": 0.962,
       "step": 7296
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7232412370045216e-05,
-      "loss": 0.8758,
+      "loss": 0.8515,
       "step": 7297
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.723159889393342e-05,
-      "loss": 0.706,
+      "loss": 0.8668,
       "step": 7298
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7230785317493448e-05,
-      "loss": 0.7441,
+      "loss": 0.9267,
       "step": 7299
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.722997164073659e-05,
-      "loss": 0.7817,
+      "loss": 0.9014,
       "step": 7300
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.722915786367413e-05,
-      "loss": 0.6625,
+      "loss": 0.7733,
       "step": 7301
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7228343986317365e-05,
-      "loss": 0.6735,
+      "loss": 0.8755,
       "step": 7302
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.722753000867758e-05,
-      "loss": 0.8829,
+      "loss": 0.9313,
       "step": 7303
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7226715930766077e-05,
-      "loss": 0.8389,
+      "loss": 0.9258,
       "step": 7304
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7225901752594137e-05,
-      "loss": 0.9205,
+      "loss": 1.0025,
       "step": 7305
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7225087474173062e-05,
-      "loss": 0.886,
+      "loss": 1.009,
       "step": 7306
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7224273095514156e-05,
-      "loss": 0.9814,
+      "loss": 1.0421,
       "step": 7307
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7223458616628707e-05,
-      "loss": 0.8642,
+      "loss": 0.9369,
       "step": 7308
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7222644037528015e-05,
-      "loss": 0.746,
+      "loss": 0.8861,
       "step": 7309
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.722182935822339e-05,
-      "loss": 0.8116,
+      "loss": 0.9285,
       "step": 7310
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.722101457872613e-05,
-      "loss": 0.9173,
+      "loss": 0.9644,
       "step": 7311
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7220199699047532e-05,
-      "loss": 0.8259,
+      "loss": 0.9313,
       "step": 7312
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.721938471919891e-05,
-      "loss": 0.7334,
+      "loss": 0.8918,
       "step": 7313
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7218569639191567e-05,
-      "loss": 0.8138,
+      "loss": 0.8305,
       "step": 7314
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7217754459036814e-05,
-      "loss": 0.7573,
+      "loss": 0.8449,
       "step": 7315
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7216939178745956e-05,
-      "loss": 0.8972,
+      "loss": 0.9302,
       "step": 7316
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.721612379833031e-05,
-      "loss": 0.8029,
+      "loss": 0.8925,
       "step": 7317
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.721530831780118e-05,
-      "loss": 0.7226,
+      "loss": 0.7818,
       "step": 7318
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.721449273716989e-05,
-      "loss": 0.7573,
+      "loss": 0.9499,
       "step": 7319
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7213677056447747e-05,
-      "loss": 0.8764,
+      "loss": 1.0012,
       "step": 7320
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7212861275646065e-05,
-      "loss": 0.8099,
+      "loss": 0.9352,
       "step": 7321
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7212045394776172e-05,
-      "loss": 0.7879,
+      "loss": 0.9106,
       "step": 7322
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.721122941384938e-05,
-      "loss": 0.9061,
+      "loss": 0.9318,
       "step": 7323
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7210413332877012e-05,
-      "loss": 0.8176,
+      "loss": 0.8789,
       "step": 7324
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720959715187039e-05,
-      "loss": 0.7873,
+      "loss": 0.9237,
       "step": 7325
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7208780870840837e-05,
-      "loss": 0.8253,
+      "loss": 0.9147,
       "step": 7326
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7207964489799675e-05,
-      "loss": 0.9801,
+      "loss": 1.0398,
       "step": 7327
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720714800875823e-05,
-      "loss": 0.9093,
+      "loss": 1.0273,
       "step": 7328
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720633142772784e-05,
-      "loss": 0.6233,
+      "loss": 0.7671,
       "step": 7329
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720551474671982e-05,
-      "loss": 0.8362,
+      "loss": 0.9539,
       "step": 7330
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7204697965745506e-05,
-      "loss": 0.8405,
+      "loss": 0.9489,
       "step": 7331
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720388108481623e-05,
-      "loss": 0.7855,
+      "loss": 0.9343,
       "step": 7332
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7203064103943326e-05,
-      "loss": 0.8403,
+      "loss": 0.9141,
       "step": 7333
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7202247023138126e-05,
-      "loss": 0.8117,
+      "loss": 0.8878,
       "step": 7334
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720142984241197e-05,
-      "loss": 0.8026,
+      "loss": 0.8561,
       "step": 7335
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.720061256177619e-05,
-      "loss": 0.8686,
+      "loss": 1.0054,
       "step": 7336
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.719979518124213e-05,
-      "loss": 0.779,
+      "loss": 0.8791,
       "step": 7337
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7198977700821125e-05,
-      "loss": 0.7462,
+      "loss": 0.8623,
       "step": 7338
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.719816012052452e-05,
-      "loss": 0.6991,
+      "loss": 0.8567,
       "step": 7339
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7197342440363654e-05,
-      "loss": 0.7592,
+      "loss": 0.8696,
       "step": 7340
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.719652466034987e-05,
-      "loss": 0.7509,
+      "loss": 0.9007,
       "step": 7341
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7195706780494523e-05,
-      "loss": 0.7402,
+      "loss": 0.8822,
       "step": 7342
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.719488880080895e-05,
-      "loss": 0.9457,
+      "loss": 0.9806,
       "step": 7343
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7194070721304506e-05,
-      "loss": 0.6793,
+      "loss": 0.8176,
       "step": 7344
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7193252541992536e-05,
-      "loss": 0.8252,
+      "loss": 0.9176,
       "step": 7345
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7192434262884395e-05,
-      "loss": 0.7493,
+      "loss": 0.8768,
       "step": 7346
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7191615883991432e-05,
-      "loss": 0.8243,
+      "loss": 0.8779,
       "step": 7347
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7190797405325e-05,
-      "loss": 0.7057,
+      "loss": 0.907,
       "step": 7348
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718997882689646e-05,
-      "loss": 0.9167,
+      "loss": 0.9947,
       "step": 7349
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718916014871716e-05,
-      "loss": 0.7197,
+      "loss": 0.8544,
       "step": 7350
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718834137079847e-05,
-      "loss": 0.8483,
+      "loss": 0.9274,
       "step": 7351
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7187522493151738e-05,
-      "loss": 0.7126,
+      "loss": 0.8919,
       "step": 7352
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7186703515788327e-05,
-      "loss": 0.7077,
+      "loss": 0.9343,
       "step": 7353
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7185884438719608e-05,
-      "loss": 0.8399,
+      "loss": 0.9309,
       "step": 7354
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7185065261956933e-05,
-      "loss": 0.8634,
+      "loss": 0.8862,
       "step": 7355
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718424598551167e-05,
-      "loss": 0.8606,
+      "loss": 0.9289,
       "step": 7356
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7183426609395194e-05,
-      "loss": 0.7783,
+      "loss": 0.9408,
       "step": 7357
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718260713361886e-05,
-      "loss": 0.683,
+      "loss": 0.8345,
       "step": 7358
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7181787558194045e-05,
-      "loss": 0.635,
+      "loss": 0.8522,
       "step": 7359
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718096788313212e-05,
-      "loss": 0.7774,
+      "loss": 0.8705,
       "step": 7360
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.718014810844445e-05,
-      "loss": 0.8925,
+      "loss": 0.9293,
       "step": 7361
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.717932823414241e-05,
-      "loss": 0.781,
+      "loss": 0.9342,
       "step": 7362
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7178508260237378e-05,
-      "loss": 0.8932,
+      "loss": 0.9664,
       "step": 7363
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7177688186740733e-05,
-      "loss": 0.7937,
+      "loss": 0.9302,
       "step": 7364
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7176868013663845e-05,
-      "loss": 0.8516,
+      "loss": 0.9421,
       "step": 7365
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.71760477410181e-05,
-      "loss": 0.6343,
+      "loss": 0.7898,
       "step": 7366
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7175227368814867e-05,
-      "loss": 0.8302,
+      "loss": 0.9716,
       "step": 7367
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.717440689706554e-05,
-      "loss": 0.7645,
+      "loss": 0.9017,
       "step": 7368
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7173586325781495e-05,
-      "loss": 0.8638,
+      "loss": 0.9763,
       "step": 7369
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.717276565497412e-05,
-      "loss": 0.8278,
+      "loss": 0.9426,
       "step": 7370
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7171944884654794e-05,
-      "loss": 0.8051,
+      "loss": 0.9136,
       "step": 7371
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7171124014834913e-05,
-      "loss": 0.8185,
+      "loss": 0.9355,
       "step": 7372
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7170303045525858e-05,
-      "loss": 0.8138,
+      "loss": 0.8849,
       "step": 7373
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716948197673902e-05,
-      "loss": 0.8656,
+      "loss": 0.955,
       "step": 7374
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7168660808485797e-05,
-      "loss": 0.7806,
+      "loss": 0.8915,
       "step": 7375
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7167839540777572e-05,
-      "loss": 0.8059,
+      "loss": 0.9722,
       "step": 7376
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716701817362574e-05,
-      "loss": 0.7694,
+      "loss": 0.8518,
       "step": 7377
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7166196707041705e-05,
-      "loss": 0.7909,
+      "loss": 0.9033,
       "step": 7378
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716537514103686e-05,
-      "loss": 0.814,
+      "loss": 0.8967,
       "step": 7379
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.71645534756226e-05,
-      "loss": 0.891,
+      "loss": 0.9549,
       "step": 7380
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716373171081032e-05,
-      "loss": 0.906,
+      "loss": 0.9747,
       "step": 7381
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7162909846611432e-05,
-      "loss": 0.7546,
+      "loss": 0.867,
       "step": 7382
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716208788303733e-05,
-      "loss": 0.8899,
+      "loss": 0.9456,
       "step": 7383
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7161265820099423e-05,
-      "loss": 0.8008,
+      "loss": 0.8995,
       "step": 7384
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.716044365780911e-05,
-      "loss": 0.7112,
+      "loss": 0.921,
       "step": 7385
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7159621396177802e-05,
-      "loss": 0.792,
+      "loss": 0.8949,
       "step": 7386
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7158799035216906e-05,
-      "loss": 0.8427,
+      "loss": 0.9911,
       "step": 7387
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7157976574937828e-05,
-      "loss": 0.6363,
+      "loss": 0.8349,
       "step": 7388
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7157154015351984e-05,
-      "loss": 0.73,
+      "loss": 0.8841,
       "step": 7389
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7156331356470782e-05,
-      "loss": 0.8432,
+      "loss": 0.9703,
       "step": 7390
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7155508598305636e-05,
-      "loss": 0.8567,
+      "loss": 0.9785,
       "step": 7391
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.715468574086796e-05,
-      "loss": 0.6998,
+      "loss": 0.8953,
       "step": 7392
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7153862784169172e-05,
-      "loss": 0.6995,
+      "loss": 0.8689,
       "step": 7393
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7153039728220683e-05,
-      "loss": 0.7945,
+      "loss": 0.81,
       "step": 7394
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.715221657303392e-05,
-      "loss": 0.805,
+      "loss": 0.8944,
       "step": 7395
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.71513933186203e-05,
-      "loss": 0.7923,
+      "loss": 0.9151,
       "step": 7396
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7150569964991243e-05,
-      "loss": 0.7299,
+      "loss": 0.9009,
       "step": 7397
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7149746512158175e-05,
-      "loss": 0.7586,
+      "loss": 0.8539,
       "step": 7398
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7148922960132518e-05,
-      "loss": 0.8365,
+      "loss": 0.8967,
       "step": 7399
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7148099308925696e-05,
-      "loss": 0.7985,
+      "loss": 0.9016,
       "step": 7400
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.714727555854914e-05,
-      "loss": 0.9517,
+      "loss": 0.9438,
       "step": 7401
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7146451709014277e-05,
-      "loss": 0.852,
+      "loss": 0.9896,
       "step": 7402
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7145627760332533e-05,
-      "loss": 0.8542,
+      "loss": 0.9757,
       "step": 7403
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7144803712515344e-05,
-      "loss": 0.8384,
+      "loss": 0.904,
       "step": 7404
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.714397956557414e-05,
-      "loss": 0.6888,
+      "loss": 0.8548,
       "step": 7405
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7143155319520356e-05,
-      "loss": 0.6818,
+      "loss": 0.7706,
       "step": 7406
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7142330974365427e-05,
-      "loss": 0.7764,
+      "loss": 0.9317,
       "step": 7407
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.714150653012079e-05,
-      "loss": 0.7448,
+      "loss": 0.8433,
       "step": 7408
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.714068198679788e-05,
-      "loss": 0.8423,
+      "loss": 0.9586,
       "step": 7409
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7139857344408137e-05,
-      "loss": 0.8101,
+      "loss": 0.9288,
       "step": 7410
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713903260296301e-05,
-      "loss": 0.6658,
+      "loss": 0.8421,
       "step": 7411
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713820776247393e-05,
-      "loss": 0.6937,
+      "loss": 0.8076,
       "step": 7412
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7137382822952346e-05,
-      "loss": 0.8124,
+      "loss": 0.9323,
       "step": 7413
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.71365577844097e-05,
-      "loss": 0.8092,
+      "loss": 0.8827,
       "step": 7414
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7135732646857442e-05,
-      "loss": 0.7362,
+      "loss": 0.9179,
       "step": 7415
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713490741030702e-05,
-      "loss": 0.817,
+      "loss": 0.9815,
       "step": 7416
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713408207476988e-05,
-      "loss": 0.8055,
+      "loss": 0.9285,
       "step": 7417
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713325664025747e-05,
-      "loss": 0.6976,
+      "loss": 0.8495,
       "step": 7418
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7132431106781246e-05,
-      "loss": 0.8326,
+      "loss": 0.9454,
       "step": 7419
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713160547435266e-05,
-      "loss": 0.7489,
+      "loss": 0.8595,
       "step": 7420
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.713077974298317e-05,
-      "loss": 0.9203,
+      "loss": 0.9822,
       "step": 7421
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7129953912684226e-05,
-      "loss": 0.7476,
+      "loss": 0.9324,
       "step": 7422
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7129127983467287e-05,
-      "loss": 0.6721,
+      "loss": 0.8267,
       "step": 7423
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7128301955343812e-05,
-      "loss": 0.8176,
+      "loss": 0.8713,
       "step": 7424
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7127475828325262e-05,
-      "loss": 0.9226,
+      "loss": 0.9025,
       "step": 7425
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7126649602423097e-05,
-      "loss": 0.8207,
+      "loss": 0.9109,
       "step": 7426
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7125823277648778e-05,
-      "loss": 0.9809,
+      "loss": 1.0069,
       "step": 7427
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7124996854013775e-05,
-      "loss": 0.7562,
+      "loss": 0.8596,
       "step": 7428
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7124170331529545e-05,
-      "loss": 0.8548,
+      "loss": 1.0068,
       "step": 7429
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.712334371020756e-05,
-      "loss": 0.8346,
+      "loss": 0.913,
       "step": 7430
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7122516990059293e-05,
-      "loss": 0.8683,
+      "loss": 0.9286,
       "step": 7431
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.712169017109621e-05,
-      "loss": 0.791,
+      "loss": 0.951,
       "step": 7432
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7120863253329773e-05,
-      "loss": 0.7522,
+      "loss": 0.8784,
       "step": 7433
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7120036236771466e-05,
-      "loss": 0.8096,
+      "loss": 0.9137,
       "step": 7434
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7119209121432752e-05,
-      "loss": 0.6819,
+      "loss": 0.8056,
       "step": 7435
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.711838190732512e-05,
-      "loss": 0.7806,
+      "loss": 0.9247,
       "step": 7436
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7117554594460036e-05,
-      "loss": 0.7215,
+      "loss": 0.8044,
       "step": 7437
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.711672718284898e-05,
-      "loss": 0.875,
+      "loss": 1.0024,
       "step": 7438
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7115899672503432e-05,
-      "loss": 0.7706,
+      "loss": 0.942,
       "step": 7439
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.711507206343487e-05,
-      "loss": 0.7466,
+      "loss": 0.9107,
       "step": 7440
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7114244355654782e-05,
-      "loss": 0.6466,
+      "loss": 0.9206,
       "step": 7441
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7113416549174643e-05,
-      "loss": 0.718,
+      "loss": 0.8837,
       "step": 7442
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7112588644005946e-05,
-      "loss": 0.7795,
+      "loss": 0.8515,
       "step": 7443
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.711176064016017e-05,
-      "loss": 0.8948,
+      "loss": 0.9344,
       "step": 7444
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7110932537648807e-05,
-      "loss": 0.73,
+      "loss": 0.8243,
       "step": 7445
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7110104336483342e-05,
-      "loss": 0.8211,
+      "loss": 0.9572,
       "step": 7446
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.710927603667527e-05,
-      "loss": 0.6059,
+      "loss": 0.8114,
       "step": 7447
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.710844763823608e-05,
-      "loss": 0.837,
+      "loss": 0.9375,
       "step": 7448
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7107619141177262e-05,
-      "loss": 0.6973,
+      "loss": 0.8586,
       "step": 7449
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7106790545510315e-05,
-      "loss": 0.8085,
+      "loss": 0.8753,
       "step": 7450
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.710596185124673e-05,
-      "loss": 0.8102,
+      "loss": 0.9012,
       "step": 7451
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.710513305839801e-05,
-      "loss": 0.8609,
+      "loss": 0.8914,
       "step": 7452
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7104304166975648e-05,
-      "loss": 0.8029,
+      "loss": 0.8637,
       "step": 7453
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7103475176991144e-05,
-      "loss": 0.7528,
+      "loss": 0.8952,
       "step": 7454
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7102646088456e-05,
-      "loss": 0.8316,
+      "loss": 0.9268,
       "step": 7455
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7101816901381722e-05,
-      "loss": 0.7812,
+      "loss": 0.8871,
       "step": 7456
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7100987615779806e-05,
-      "loss": 1.0248,
+      "loss": 1.0935,
       "step": 7457
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7100158231661767e-05,
-      "loss": 0.8823,
+      "loss": 0.9587,
       "step": 7458
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7099328749039107e-05,
-      "loss": 0.7036,
+      "loss": 0.7777,
       "step": 7459
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709849916792333e-05,
-      "loss": 0.8902,
+      "loss": 0.9766,
       "step": 7460
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709766948832595e-05,
-      "loss": 0.7905,
+      "loss": 0.9257,
       "step": 7461
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709683971025847e-05,
-      "loss": 0.6823,
+      "loss": 0.8227,
       "step": 7462
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7096009833732414e-05,
-      "loss": 0.7306,
+      "loss": 0.9647,
       "step": 7463
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709517985875929e-05,
-      "loss": 0.7944,
+      "loss": 0.9211,
       "step": 7464
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709434978535061e-05,
-      "loss": 0.7652,
+      "loss": 0.962,
       "step": 7465
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7093519613517894e-05,
-      "loss": 0.7547,
+      "loss": 0.8875,
       "step": 7466
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7092689343272658e-05,
-      "loss": 0.8138,
+      "loss": 0.878,
       "step": 7467
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.709185897462642e-05,
-      "loss": 0.824,
+      "loss": 0.8901,
       "step": 7468
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7091028507590697e-05,
-      "loss": 0.6994,
+      "loss": 0.824,
       "step": 7469
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7090197942177017e-05,
-      "loss": 0.7233,
+      "loss": 0.8119,
       "step": 7470
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70893672783969e-05,
-      "loss": 0.6784,
+      "loss": 0.8346,
       "step": 7471
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7088536516261874e-05,
-      "loss": 0.936,
+      "loss": 1.0003,
       "step": 7472
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7087705655783458e-05,
-      "loss": 0.9791,
+      "loss": 1.1072,
       "step": 7473
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7086874696973183e-05,
-      "loss": 0.6628,
+      "loss": 0.8064,
       "step": 7474
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7086043639842572e-05,
-      "loss": 0.8425,
+      "loss": 0.9847,
       "step": 7475
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7085212484403167e-05,
-      "loss": 0.7054,
+      "loss": 0.816,
       "step": 7476
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7084381230666485e-05,
-      "loss": 0.747,
+      "loss": 0.8191,
       "step": 7477
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.708354987864407e-05,
-      "loss": 0.7799,
+      "loss": 0.8776,
       "step": 7478
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7082718428347446e-05,
-      "loss": 0.9309,
+      "loss": 1.0033,
       "step": 7479
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7081886879788157e-05,
-      "loss": 0.7035,
+      "loss": 0.8457,
       "step": 7480
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7081055232977735e-05,
-      "loss": 0.7272,
+      "loss": 0.8542,
       "step": 7481
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7080223487927714e-05,
-      "loss": 0.8554,
+      "loss": 0.9644,
       "step": 7482
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.707939164464964e-05,
-      "loss": 0.7886,
+      "loss": 0.9073,
       "step": 7483
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7078559703155054e-05,
-      "loss": 0.7009,
+      "loss": 0.8637,
       "step": 7484
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.707772766345549e-05,
-      "loss": 0.7463,
+      "loss": 0.8974,
       "step": 7485
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70768955255625e-05,
-      "loss": 0.9378,
+      "loss": 1.0051,
       "step": 7486
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7076063289487628e-05,
-      "loss": 0.824,
+      "loss": 0.9291,
       "step": 7487
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7075230955242413e-05,
-      "loss": 0.8139,
+      "loss": 0.8189,
       "step": 7488
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.707439852283841e-05,
-      "loss": 0.939,
+      "loss": 1.0141,
       "step": 7489
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.707356599228716e-05,
-      "loss": 0.7683,
+      "loss": 0.9281,
       "step": 7490
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7072733363600226e-05,
-      "loss": 0.8522,
+      "loss": 0.9568,
       "step": 7491
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7071900636789146e-05,
-      "loss": 0.8675,
+      "loss": 0.9417,
       "step": 7492
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7071067811865477e-05,
-      "loss": 0.8658,
+      "loss": 0.9562,
       "step": 7493
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7070234888840777e-05,
-      "loss": 0.7917,
+      "loss": 0.8871,
       "step": 7494
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7069401867726598e-05,
-      "loss": 0.8857,
+      "loss": 0.9567,
       "step": 7495
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7068568748534498e-05,
-      "loss": 0.8027,
+      "loss": 0.8435,
       "step": 7496
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.706773553127604e-05,
-      "loss": 0.8572,
+      "loss": 0.9624,
       "step": 7497
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7066902215962776e-05,
-      "loss": 0.8347,
+      "loss": 0.9382,
       "step": 7498
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.706606880260627e-05,
-      "loss": 0.7138,
+      "loss": 0.7984,
       "step": 7499
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7065235291218085e-05,
-      "loss": 0.7288,
+      "loss": 0.8358,
       "step": 7500
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7064401681809784e-05,
-      "loss": 0.8225,
+      "loss": 0.9308,
       "step": 7501
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7063567974392933e-05,
-      "loss": 0.9133,
+      "loss": 1.0112,
       "step": 7502
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70627341689791e-05,
-      "loss": 0.7644,
+      "loss": 0.9084,
       "step": 7503
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7061900265579852e-05,
-      "loss": 0.7732,
+      "loss": 0.851,
       "step": 7504
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7061066264206753e-05,
-      "loss": 0.9772,
+      "loss": 0.977,
       "step": 7505
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7060232164871383e-05,
-      "loss": 0.8707,
+      "loss": 0.9136,
       "step": 7506
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7059397967585302e-05,
-      "loss": 0.8693,
+      "loss": 0.9581,
       "step": 7507
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7058563672360095e-05,
-      "loss": 0.7635,
+      "loss": 0.813,
       "step": 7508
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7057729279207332e-05,
-      "loss": 0.7417,
+      "loss": 0.8753,
       "step": 7509
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705689478813859e-05,
-      "loss": 0.6925,
+      "loss": 0.7879,
       "step": 7510
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705606019916544e-05,
-      "loss": 0.7616,
+      "loss": 0.9109,
       "step": 7511
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705522551229947e-05,
-      "loss": 0.8386,
+      "loss": 0.9104,
       "step": 7512
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705439072755226e-05,
-      "loss": 0.74,
+      "loss": 0.8651,
       "step": 7513
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7053555844935378e-05,
-      "loss": 0.8434,
+      "loss": 0.918,
       "step": 7514
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705272086446042e-05,
-      "loss": 0.7956,
+      "loss": 0.9244,
       "step": 7515
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705188578613897e-05,
-      "loss": 0.9283,
+      "loss": 1.0233,
       "step": 7516
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7051050609982608e-05,
-      "loss": 0.8344,
+      "loss": 1.0101,
       "step": 7517
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.705021533600292e-05,
-      "loss": 0.8741,
+      "loss": 1.0007,
       "step": 7518
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70493799642115e-05,
-      "loss": 0.6693,
+      "loss": 0.8844,
       "step": 7519
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7048544494619938e-05,
-      "loss": 0.8166,
+      "loss": 0.9482,
       "step": 7520
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7047708927239813e-05,
-      "loss": 0.6991,
+      "loss": 0.8457,
       "step": 7521
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7046873262082735e-05,
-      "loss": 0.7582,
+      "loss": 0.8878,
       "step": 7522
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7046037499160284e-05,
-      "loss": 0.6814,
+      "loss": 0.8457,
       "step": 7523
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.704520163848406e-05,
-      "loss": 0.9355,
+      "loss": 1.0149,
       "step": 7524
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7044365680065664e-05,
-      "loss": 0.8325,
+      "loss": 0.9348,
       "step": 7525
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7043529623916683e-05,
-      "loss": 0.8504,
+      "loss": 0.9283,
       "step": 7526
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7042693470048723e-05,
-      "loss": 0.8913,
+      "loss": 0.9422,
       "step": 7527
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7041857218473385e-05,
-      "loss": 0.7874,
+      "loss": 0.8909,
       "step": 7528
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7041020869202266e-05,
-      "loss": 0.734,
+      "loss": 0.7739,
       "step": 7529
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7040184422246974e-05,
-      "loss": 0.7792,
+      "loss": 0.9102,
       "step": 7530
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7039347877619114e-05,
-      "loss": 0.7794,
+      "loss": 0.8788,
       "step": 7531
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7038511235330286e-05,
-      "loss": 0.8793,
+      "loss": 0.9516,
       "step": 7532
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7037674495392105e-05,
-      "loss": 0.7796,
+      "loss": 0.8756,
       "step": 7533
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7036837657816172e-05,
-      "loss": 0.8775,
+      "loss": 0.9858,
       "step": 7534
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70360007226141e-05,
-      "loss": 0.6764,
+      "loss": 0.8423,
       "step": 7535
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7035163689797504e-05,
-      "loss": 0.7476,
+      "loss": 0.8511,
       "step": 7536
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7034326559377992e-05,
-      "loss": 0.8956,
+      "loss": 0.9215,
       "step": 7537
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.703348933136718e-05,
-      "loss": 0.8476,
+      "loss": 0.8406,
       "step": 7538
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7032652005776682e-05,
-      "loss": 0.936,
+      "loss": 1.0547,
       "step": 7539
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7031814582618115e-05,
-      "loss": 0.8893,
+      "loss": 0.931,
       "step": 7540
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70309770619031e-05,
-      "loss": 0.8246,
+      "loss": 1.0255,
       "step": 7541
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7030139443643252e-05,
-      "loss": 0.7835,
+      "loss": 0.9683,
       "step": 7542
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7029301727850193e-05,
-      "loss": 0.6134,
+      "loss": 0.8674,
       "step": 7543
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7028463914535547e-05,
-      "loss": 0.6854,
+      "loss": 0.8943,
       "step": 7544
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7027626003710937e-05,
-      "loss": 0.9843,
+      "loss": 1.0046,
       "step": 7545
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7026787995387985e-05,
-      "loss": 0.8537,
+      "loss": 0.9786,
       "step": 7546
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.702594988957832e-05,
-      "loss": 0.8332,
+      "loss": 0.9118,
       "step": 7547
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.702511168629357e-05,
-      "loss": 0.9456,
+      "loss": 0.9449,
       "step": 7548
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7024273385545364e-05,
-      "loss": 0.8505,
+      "loss": 0.9415,
       "step": 7549
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7023434987345324e-05,
-      "loss": 0.833,
+      "loss": 0.9261,
       "step": 7550
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7022596491705097e-05,
-      "loss": 0.8217,
+      "loss": 0.8939,
       "step": 7551
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.70217578986363e-05,
-      "loss": 0.7542,
+      "loss": 0.9157,
       "step": 7552
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7020919208150578e-05,
-      "loss": 0.8146,
+      "loss": 0.9318,
       "step": 7553
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7020080420259563e-05,
-      "loss": 0.8455,
+      "loss": 0.9241,
       "step": 7554
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7019241534974897e-05,
-      "loss": 0.7416,
+      "loss": 0.9038,
       "step": 7555
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7018402552308208e-05,
-      "loss": 0.7284,
+      "loss": 0.8916,
       "step": 7556
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7017563472271145e-05,
-      "loss": 0.8346,
+      "loss": 0.8685,
       "step": 7557
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7016724294875342e-05,
-      "loss": 0.9006,
+      "loss": 0.9499,
       "step": 7558
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7015885020132445e-05,
-      "loss": 0.7139,
+      "loss": 0.8522,
       "step": 7559
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7015045648054102e-05,
-      "loss": 0.9134,
+      "loss": 0.9423,
       "step": 7560
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.7014206178651948e-05,
-      "loss": 0.7558,
+      "loss": 0.928,
       "step": 7561
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7013366611937637e-05,
-      "loss": 0.859,
+      "loss": 0.9514,
       "step": 7562
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.701252694792282e-05,
-      "loss": 0.8151,
+      "loss": 0.9134,
       "step": 7563
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7011687186619135e-05,
-      "loss": 0.8988,
+      "loss": 1.002,
       "step": 7564
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.701084732803824e-05,
-      "loss": 0.8505,
+      "loss": 0.8845,
       "step": 7565
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7010007372191787e-05,
-      "loss": 0.7642,
+      "loss": 0.8661,
       "step": 7566
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7009167319091426e-05,
-      "loss": 0.8415,
+      "loss": 0.8679,
       "step": 7567
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7008327168748812e-05,
-      "loss": 0.7278,
+      "loss": 0.9214,
       "step": 7568
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7007486921175606e-05,
-      "loss": 0.9467,
+      "loss": 1.0147,
       "step": 7569
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7006646576383457e-05,
-      "loss": 0.8519,
+      "loss": 0.9297,
       "step": 7570
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7005806134384032e-05,
-      "loss": 0.5974,
+      "loss": 0.7884,
       "step": 7571
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7004965595188983e-05,
-      "loss": 0.7423,
+      "loss": 0.9236,
       "step": 7572
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7004124958809977e-05,
-      "loss": 0.7372,
+      "loss": 0.9077,
       "step": 7573
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7003284225258675e-05,
-      "loss": 0.7613,
+      "loss": 0.9144,
       "step": 7574
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.700244339454674e-05,
-      "loss": 0.7817,
+      "loss": 0.9442,
       "step": 7575
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.700160246668584e-05,
-      "loss": 0.7763,
+      "loss": 0.9698,
       "step": 7576
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.7000761441687636e-05,
-      "loss": 0.8256,
+      "loss": 0.9191,
       "step": 7577
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6999920319563804e-05,
-      "loss": 0.8202,
+      "loss": 0.9445,
       "step": 7578
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6999079100326005e-05,
-      "loss": 0.7983,
+      "loss": 0.8835,
       "step": 7579
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6998237783985917e-05,
-      "loss": 0.8247,
+      "loss": 0.8664,
       "step": 7580
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6997396370555207e-05,
-      "loss": 0.875,
+      "loss": 0.9692,
       "step": 7581
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6996554860045552e-05,
-      "loss": 0.7006,
+      "loss": 0.8162,
       "step": 7582
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6995713252468622e-05,
-      "loss": 0.8053,
+      "loss": 0.8521,
       "step": 7583
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6994871547836102e-05,
-      "loss": 0.7165,
+      "loss": 0.8049,
       "step": 7584
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6994029746159658e-05,
-      "loss": 0.9105,
+      "loss": 0.9755,
       "step": 7585
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6993187847450975e-05,
-      "loss": 0.7478,
+      "loss": 0.9374,
       "step": 7586
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6992345851721735e-05,
-      "loss": 0.6916,
+      "loss": 0.8383,
       "step": 7587
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6991503758983618e-05,
-      "loss": 0.7195,
+      "loss": 0.8151,
       "step": 7588
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6990661569248302e-05,
-      "loss": 0.7762,
+      "loss": 0.9439,
       "step": 7589
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.698981928252748e-05,
-      "loss": 0.7587,
+      "loss": 0.9468,
       "step": 7590
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.698897689883283e-05,
-      "loss": 0.7406,
+      "loss": 0.9039,
       "step": 7591
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6988134418176043e-05,
-      "loss": 0.8292,
+      "loss": 0.9443,
       "step": 7592
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6987291840568803e-05,
-      "loss": 0.9217,
+      "loss": 0.9615,
       "step": 7593
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6986449166022806e-05,
-      "loss": 0.8102,
+      "loss": 0.8857,
       "step": 7594
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6985606394549738e-05,
-      "loss": 0.7002,
+      "loss": 0.9151,
       "step": 7595
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.698476352616129e-05,
-      "loss": 0.7051,
+      "loss": 0.9213,
       "step": 7596
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6983920560869158e-05,
-      "loss": 0.879,
+      "loss": 1.0562,
       "step": 7597
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.698307749868504e-05,
-      "loss": 0.9135,
+      "loss": 0.9443,
       "step": 7598
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6982234339620633e-05,
-      "loss": 0.7526,
+      "loss": 0.8644,
       "step": 7599
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6981391083687623e-05,
-      "loss": 0.8297,
+      "loss": 0.9662,
       "step": 7600
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6980547730897723e-05,
-      "loss": 0.8708,
+      "loss": 0.983,
       "step": 7601
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6979704281262623e-05,
-      "loss": 0.6444,
+      "loss": 0.8078,
       "step": 7602
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6978860734794033e-05,
-      "loss": 0.8704,
+      "loss": 0.9715,
       "step": 7603
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.697801709150365e-05,
-      "loss": 0.6779,
+      "loss": 0.8193,
       "step": 7604
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.697717335140318e-05,
-      "loss": 0.7457,
+      "loss": 0.9133,
       "step": 7605
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6976329514504327e-05,
-      "loss": 0.8737,
+      "loss": 0.877,
       "step": 7606
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6975485580818806e-05,
-      "loss": 0.8315,
+      "loss": 0.9039,
       "step": 7607
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6974641550358316e-05,
-      "loss": 0.7088,
+      "loss": 0.9126,
       "step": 7608
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.697379742313457e-05,
-      "loss": 0.7657,
+      "loss": 0.925,
       "step": 7609
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.697295319915928e-05,
-      "loss": 0.7548,
+      "loss": 0.93,
       "step": 7610
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6972108878444156e-05,
-      "loss": 1.0174,
+      "loss": 0.9929,
       "step": 7611
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6971264461000917e-05,
-      "loss": 0.7537,
+      "loss": 0.9127,
       "step": 7612
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.697041994684127e-05,
-      "loss": 0.7152,
+      "loss": 0.9109,
       "step": 7613
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.696957533597694e-05,
-      "loss": 0.8781,
+      "loss": 0.9583,
       "step": 7614
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.696873062841964e-05,
-      "loss": 0.8296,
+      "loss": 0.9084,
       "step": 7615
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.696788582418109e-05,
-      "loss": 0.8875,
+      "loss": 0.9385,
       "step": 7616
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6967040923273005e-05,
-      "loss": 0.9943,
+      "loss": 1.0555,
       "step": 7617
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.696619592570712e-05,
-      "loss": 0.7268,
+      "loss": 0.8863,
       "step": 7618
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6965350831495143e-05,
-      "loss": 0.8845,
+      "loss": 1.0198,
       "step": 7619
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6964505640648813e-05,
-      "loss": 0.8434,
+      "loss": 0.8941,
       "step": 7620
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6963660353179842e-05,
-      "loss": 0.8153,
+      "loss": 0.9321,
       "step": 7621
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6962814969099968e-05,
-      "loss": 0.7605,
+      "loss": 0.9033,
       "step": 7622
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.696196948842091e-05,
-      "loss": 0.7675,
+      "loss": 0.9515,
       "step": 7623
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6961123911154407e-05,
-      "loss": 0.8057,
+      "loss": 0.9035,
       "step": 7624
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6960278237312188e-05,
-      "loss": 0.6426,
+      "loss": 0.746,
       "step": 7625
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6959432466905984e-05,
-      "loss": 0.772,
+      "loss": 0.8692,
       "step": 7626
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6958586599947527e-05,
-      "loss": 0.8273,
+      "loss": 0.9072,
       "step": 7627
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.695774063644855e-05,
-      "loss": 0.9078,
+      "loss": 0.9455,
       "step": 7628
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.69568945764208e-05,
-      "loss": 0.7414,
+      "loss": 0.9487,
       "step": 7629
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6956048419876007e-05,
-      "loss": 0.7412,
+      "loss": 0.8799,
       "step": 7630
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.695520216682591e-05,
-      "loss": 0.9457,
+      "loss": 0.9971,
       "step": 7631
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6954355817282245e-05,
-      "loss": 0.881,
+      "loss": 0.8964,
       "step": 7632
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6953509371256768e-05,
-      "loss": 0.654,
+      "loss": 0.8575,
       "step": 7633
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6952662828761214e-05,
-      "loss": 0.8263,
+      "loss": 0.9659,
       "step": 7634
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6951816189807323e-05,
-      "loss": 0.8293,
+      "loss": 0.8672,
       "step": 7635
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.695096945440685e-05,
-      "loss": 0.7483,
+      "loss": 0.7869,
       "step": 7636
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6950122622571536e-05,
-      "loss": 0.7311,
+      "loss": 0.8845,
       "step": 7637
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6949275694313133e-05,
-      "loss": 0.8285,
+      "loss": 0.8353,
       "step": 7638
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.694842866964339e-05,
-      "loss": 0.7399,
+      "loss": 0.8776,
       "step": 7639
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6947581548574056e-05,
-      "loss": 0.7016,
+      "loss": 0.8515,
       "step": 7640
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6946734331116886e-05,
-      "loss": 0.8528,
+      "loss": 0.9029,
       "step": 7641
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.694588701728363e-05,
-      "loss": 0.7508,
+      "loss": 0.8888,
       "step": 7642
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6945039607086055e-05,
-      "loss": 0.8124,
+      "loss": 0.949,
       "step": 7643
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.69441921005359e-05,
-      "loss": 0.788,
+      "loss": 0.8815,
       "step": 7644
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6943344497644936e-05,
-      "loss": 0.8724,
+      "loss": 0.9613,
       "step": 7645
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.694249679842492e-05,
-      "loss": 0.8085,
+      "loss": 0.8979,
       "step": 7646
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6941649002887608e-05,
-      "loss": 0.7668,
+      "loss": 0.8529,
       "step": 7647
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6940801111044764e-05,
-      "loss": 0.7205,
+      "loss": 0.845,
       "step": 7648
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.693995312290816e-05,
-      "loss": 1.0026,
+      "loss": 1.0163,
       "step": 7649
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6939105038489545e-05,
-      "loss": 0.7821,
+      "loss": 0.9053,
       "step": 7650
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.693825685780069e-05,
-      "loss": 0.822,
+      "loss": 0.925,
       "step": 7651
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6937408580853376e-05,
-      "loss": 0.8636,
+      "loss": 0.9443,
       "step": 7652
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6936560207659353e-05,
-      "loss": 0.6503,
+      "loss": 0.8698,
       "step": 7653
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.69357117382304e-05,
-      "loss": 0.8822,
+      "loss": 0.8971,
       "step": 7654
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6934863172578292e-05,
-      "loss": 0.8444,
+      "loss": 0.9719,
       "step": 7655
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.693401451071479e-05,
-      "loss": 0.7749,
+      "loss": 0.8572,
       "step": 7656
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6933165752651678e-05,
-      "loss": 0.9249,
+      "loss": 0.9245,
       "step": 7657
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.693231689840073e-05,
-      "loss": 0.7862,
+      "loss": 0.9641,
       "step": 7658
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.693146794797372e-05,
-      "loss": 1.0064,
+      "loss": 1.0415,
       "step": 7659
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6930618901382423e-05,
-      "loss": 0.8803,
+      "loss": 0.9028,
       "step": 7660
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6929769758638625e-05,
-      "loss": 0.8086,
+      "loss": 0.8736,
       "step": 7661
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6928920519754102e-05,
-      "loss": 0.787,
+      "loss": 0.9098,
       "step": 7662
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.692807118474064e-05,
-      "loss": 0.8651,
+      "loss": 1.0193,
       "step": 7663
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6927221753610018e-05,
-      "loss": 0.7903,
+      "loss": 0.847,
       "step": 7664
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6926372226374025e-05,
-      "loss": 0.7522,
+      "loss": 0.9223,
       "step": 7665
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.692552260304444e-05,
-      "loss": 0.6886,
+      "loss": 0.8375,
       "step": 7666
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.692467288363306e-05,
-      "loss": 0.8658,
+      "loss": 0.9089,
       "step": 7667
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6923823068151663e-05,
-      "loss": 0.7465,
+      "loss": 0.8972,
       "step": 7668
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6922973156612048e-05,
-      "loss": 0.8024,
+      "loss": 0.9669,
       "step": 7669
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6922123149026002e-05,
-      "loss": 0.8242,
+      "loss": 0.9659,
       "step": 7670
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6921273045405323e-05,
-      "loss": 0.788,
+      "loss": 0.9577,
       "step": 7671
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6920422845761793e-05,
-      "loss": 0.8262,
+      "loss": 0.8794,
       "step": 7672
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6919572550107218e-05,
-      "loss": 0.8809,
+      "loss": 0.9682,
       "step": 7673
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6918722158453394e-05,
-      "loss": 0.8725,
+      "loss": 0.9532,
       "step": 7674
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6917871670812113e-05,
-      "loss": 0.8622,
+      "loss": 0.9636,
       "step": 7675
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6917021087195183e-05,
-      "loss": 0.7829,
+      "loss": 0.992,
       "step": 7676
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6916170407614393e-05,
-      "loss": 0.7717,
+      "loss": 0.8568,
       "step": 7677
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6915319632081555e-05,
-      "loss": 0.7361,
+      "loss": 0.8572,
       "step": 7678
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6914468760608466e-05,
-      "loss": 0.867,
+      "loss": 0.97,
       "step": 7679
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6913617793206934e-05,
-      "loss": 0.799,
+      "loss": 0.8669,
       "step": 7680
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6912766729888767e-05,
-      "loss": 0.7952,
+      "loss": 0.9869,
       "step": 7681
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6911915570665766e-05,
-      "loss": 0.8492,
+      "loss": 1.0049,
       "step": 7682
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6911064315549747e-05,
-      "loss": 0.9047,
+      "loss": 0.9426,
       "step": 7683
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6910212964552513e-05,
-      "loss": 0.8977,
+      "loss": 0.9152,
       "step": 7684
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6909361517685878e-05,
-      "loss": 0.8707,
+      "loss": 0.9699,
       "step": 7685
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6908509974961655e-05,
-      "loss": 0.9527,
+      "loss": 0.9739,
       "step": 7686
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.690765833639166e-05,
-      "loss": 0.7262,
+      "loss": 0.8026,
       "step": 7687
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6906806601987707e-05,
-      "loss": 0.7981,
+      "loss": 0.8945,
       "step": 7688
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.690595477176161e-05,
-      "loss": 0.6866,
+      "loss": 0.8601,
       "step": 7689
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.690510284572519e-05,
-      "loss": 0.8305,
+      "loss": 0.8743,
       "step": 7690
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6904250823890262e-05,
-      "loss": 0.74,
+      "loss": 0.8243,
       "step": 7691
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.690339870626865e-05,
-      "loss": 0.8146,
+      "loss": 0.9347,
       "step": 7692
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.690254649287218e-05,
-      "loss": 0.7231,
+      "loss": 0.9738,
       "step": 7693
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6901694183712672e-05,
-      "loss": 0.952,
+      "loss": 0.9804,
       "step": 7694
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6900841778801944e-05,
-      "loss": 0.8135,
+      "loss": 0.9655,
       "step": 7695
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.689998927815183e-05,
-      "loss": 0.9012,
+      "loss": 0.8966,
       "step": 7696
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6899136681774157e-05,
-      "loss": 0.7792,
+      "loss": 0.9486,
       "step": 7697
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6898283989680747e-05,
-      "loss": 0.7745,
+      "loss": 0.8528,
       "step": 7698
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6897431201883435e-05,
-      "loss": 0.7223,
+      "loss": 0.8084,
       "step": 7699
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.689657831839405e-05,
-      "loss": 0.899,
+      "loss": 1.0093,
       "step": 7700
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6895725339224423e-05,
-      "loss": 0.7386,
+      "loss": 0.892,
       "step": 7701
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6894872264386396e-05,
-      "loss": 0.7784,
+      "loss": 0.8479,
       "step": 7702
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.68940190938918e-05,
-      "loss": 0.8194,
+      "loss": 0.9294,
       "step": 7703
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6893165827752465e-05,
-      "loss": 0.9023,
+      "loss": 0.9681,
       "step": 7704
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6892312465980238e-05,
-      "loss": 0.7519,
+      "loss": 0.9246,
       "step": 7705
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6891459008586956e-05,
-      "loss": 0.8382,
+      "loss": 0.9435,
       "step": 7706
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6890605455584454e-05,
-      "loss": 0.8747,
+      "loss": 0.9646,
       "step": 7707
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688975180698458e-05,
-      "loss": 0.7301,
+      "loss": 0.8921,
       "step": 7708
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6888898062799173e-05,
-      "loss": 0.7668,
+      "loss": 0.9095,
       "step": 7709
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688804422304008e-05,
-      "loss": 0.8607,
+      "loss": 0.989,
       "step": 7710
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688719028771915e-05,
-      "loss": 0.7977,
+      "loss": 0.93,
       "step": 7711
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6886336256848223e-05,
-      "loss": 0.8466,
+      "loss": 0.9521,
       "step": 7712
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688548213043915e-05,
-      "loss": 0.8582,
+      "loss": 0.894,
       "step": 7713
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6884627908503787e-05,
-      "loss": 0.9957,
+      "loss": 1.0413,
       "step": 7714
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6883773591053975e-05,
-      "loss": 0.7934,
+      "loss": 0.8855,
       "step": 7715
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688291917810157e-05,
-      "loss": 0.8298,
+      "loss": 0.9601,
       "step": 7716
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.688206466965843e-05,
-      "loss": 0.7808,
+      "loss": 0.9229,
       "step": 7717
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6881210065736405e-05,
-      "loss": 0.8224,
+      "loss": 0.9138,
       "step": 7718
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6880355366347356e-05,
-      "loss": 0.7797,
+      "loss": 0.8324,
       "step": 7719
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6879500571503138e-05,
-      "loss": 0.762,
+      "loss": 0.8711,
       "step": 7720
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.687864568121561e-05,
-      "loss": 0.8197,
+      "loss": 0.9194,
       "step": 7721
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.687779069549663e-05,
-      "loss": 0.8151,
+      "loss": 0.9137,
       "step": 7722
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6876935614358062e-05,
-      "loss": 0.8647,
+      "loss": 0.9521,
       "step": 7723
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6876080437811775e-05,
-      "loss": 0.7473,
+      "loss": 0.7663,
       "step": 7724
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6875225165869623e-05,
-      "loss": 0.7606,
+      "loss": 0.9225,
       "step": 7725
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.687436979854348e-05,
-      "loss": 0.7639,
+      "loss": 0.9189,
       "step": 7726
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6873514335845205e-05,
-      "loss": 0.7145,
+      "loss": 0.844,
       "step": 7727
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6872658777786673e-05,
-      "loss": 0.8492,
+      "loss": 0.9556,
       "step": 7728
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6871803124379753e-05,
-      "loss": 0.8839,
+      "loss": 0.9336,
       "step": 7729
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6870947375636313e-05,
-      "loss": 0.9162,
+      "loss": 0.9855,
       "step": 7730
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6870091531568226e-05,
-      "loss": 0.748,
+      "loss": 0.8844,
       "step": 7731
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.686923559218737e-05,
-      "loss": 0.7982,
+      "loss": 0.941,
       "step": 7732
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6868379557505612e-05,
-      "loss": 0.7613,
+      "loss": 0.8971,
       "step": 7733
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6867523427534835e-05,
-      "loss": 0.7879,
+      "loss": 0.9523,
       "step": 7734
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6866667202286916e-05,
-      "loss": 0.7873,
+      "loss": 0.8818,
       "step": 7735
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6865810881773727e-05,
-      "loss": 0.7289,
+      "loss": 0.8691,
       "step": 7736
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6864954466007156e-05,
-      "loss": 0.7815,
+      "loss": 0.9393,
       "step": 7737
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6864097954999082e-05,
-      "loss": 0.6884,
+      "loss": 0.8557,
       "step": 7738
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6863241348761388e-05,
-      "loss": 0.7773,
+      "loss": 0.9066,
       "step": 7739
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.686238464730596e-05,
-      "loss": 0.8063,
+      "loss": 0.876,
       "step": 7740
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6861527850644677e-05,
-      "loss": 0.8695,
+      "loss": 0.9803,
       "step": 7741
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.686067095878943e-05,
-      "loss": 0.8278,
+      "loss": 0.9167,
       "step": 7742
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.685981397175211e-05,
-      "loss": 0.7256,
+      "loss": 0.8477,
       "step": 7743
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6858956889544608e-05,
-      "loss": 0.6927,
+      "loss": 0.8005,
       "step": 7744
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6858099712178805e-05,
-      "loss": 0.7365,
+      "loss": 0.9504,
       "step": 7745
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.68572424396666e-05,
-      "loss": 0.8494,
+      "loss": 0.9409,
       "step": 7746
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6856385072019887e-05,
-      "loss": 0.7722,
+      "loss": 0.8877,
       "step": 7747
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6855527609250554e-05,
-      "loss": 0.8693,
+      "loss": 0.9167,
       "step": 7748
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6854670051370506e-05,
-      "loss": 0.9127,
+      "loss": 0.951,
       "step": 7749
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6853812398391637e-05,
-      "loss": 0.8021,
+      "loss": 0.8009,
       "step": 7750
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6852954650325846e-05,
-      "loss": 0.8217,
+      "loss": 0.8795,
       "step": 7751
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.685209680718503e-05,
-      "loss": 0.82,
+      "loss": 0.8867,
       "step": 7752
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6851238868981095e-05,
-      "loss": 0.7556,
+      "loss": 0.9127,
       "step": 7753
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6850380835725937e-05,
-      "loss": 0.8141,
+      "loss": 0.9755,
       "step": 7754
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.684952270743147e-05,
-      "loss": 0.9018,
+      "loss": 1.0143,
       "step": 7755
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.684866448410959e-05,
-      "loss": 0.7639,
+      "loss": 0.9321,
       "step": 7756
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.684780616577221e-05,
-      "loss": 0.8006,
+      "loss": 0.8665,
       "step": 7757
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6846947752431234e-05,
-      "loss": 0.8047,
+      "loss": 0.8904,
       "step": 7758
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6846089244098573e-05,
-      "loss": 0.7381,
+      "loss": 0.8576,
       "step": 7759
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.684523064078613e-05,
-      "loss": 0.8681,
+      "loss": 0.9091,
       "step": 7760
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6844371942505834e-05,
-      "loss": 0.7743,
+      "loss": 0.8954,
       "step": 7761
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6843513149269585e-05,
-      "loss": 0.7126,
+      "loss": 0.8336,
       "step": 7762
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6842654261089298e-05,
-      "loss": 0.8517,
+      "loss": 0.9406,
       "step": 7763
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6841795277976892e-05,
-      "loss": 0.8358,
+      "loss": 0.844,
       "step": 7764
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6840936199944287e-05,
-      "loss": 0.9033,
+      "loss": 0.9848,
       "step": 7765
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6840077027003395e-05,
-      "loss": 0.8485,
+      "loss": 0.9066,
       "step": 7766
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.683921775916614e-05,
-      "loss": 0.7386,
+      "loss": 0.9814,
       "step": 7767
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6838358396444443e-05,
-      "loss": 0.7463,
+      "loss": 0.8641,
       "step": 7768
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6837498938850228e-05,
-      "loss": 0.9174,
+      "loss": 0.9214,
       "step": 7769
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.683663938639541e-05,
-      "loss": 0.6883,
+      "loss": 0.86,
       "step": 7770
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6835779739091925e-05,
-      "loss": 0.7582,
+      "loss": 0.8688,
       "step": 7771
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6834919996951696e-05,
-      "loss": 0.7031,
+      "loss": 0.8316,
       "step": 7772
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6834060159986647e-05,
-      "loss": 0.9198,
+      "loss": 0.96,
       "step": 7773
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6833200228208708e-05,
-      "loss": 0.7978,
+      "loss": 0.9071,
       "step": 7774
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6832340201629814e-05,
-      "loss": 0.6659,
+      "loss": 0.8434,
       "step": 7775
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6831480080261893e-05,
-      "loss": 0.7778,
+      "loss": 0.8988,
       "step": 7776
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6830619864116876e-05,
-      "loss": 0.8506,
+      "loss": 0.9579,
       "step": 7777
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6829759553206705e-05,
-      "loss": 0.6873,
+      "loss": 0.8626,
       "step": 7778
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6828899147543305e-05,
-      "loss": 0.818,
+      "loss": 0.9205,
       "step": 7779
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.682803864713862e-05,
-      "loss": 0.84,
+      "loss": 0.9569,
       "step": 7780
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6827178052004588e-05,
-      "loss": 0.7315,
+      "loss": 0.9554,
       "step": 7781
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6826317362153143e-05,
-      "loss": 0.8482,
+      "loss": 0.9464,
       "step": 7782
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6825456577596235e-05,
-      "loss": 0.8067,
+      "loss": 0.8211,
       "step": 7783
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6824595698345798e-05,
-      "loss": 0.7954,
+      "loss": 0.899,
       "step": 7784
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6823734724413777e-05,
-      "loss": 0.6703,
+      "loss": 0.8046,
       "step": 7785
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6822873655812123e-05,
-      "loss": 0.8459,
+      "loss": 0.9339,
       "step": 7786
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6822012492552775e-05,
-      "loss": 0.7611,
+      "loss": 0.9069,
       "step": 7787
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6821151234647683e-05,
-      "loss": 0.8209,
+      "loss": 0.9133,
       "step": 7788
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6820289882108794e-05,
-      "loss": 0.7124,
+      "loss": 0.8398,
       "step": 7789
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6819428434948064e-05,
-      "loss": 0.6902,
+      "loss": 0.8639,
       "step": 7790
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6818566893177438e-05,
-      "loss": 0.7916,
+      "loss": 0.8721,
       "step": 7791
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6817705256808867e-05,
-      "loss": 0.742,
+      "loss": 0.8845,
       "step": 7792
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.681684352585431e-05,
-      "loss": 0.8224,
+      "loss": 0.9028,
       "step": 7793
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6815981700325724e-05,
-      "loss": 0.8676,
+      "loss": 0.9219,
       "step": 7794
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6815119780235058e-05,
-      "loss": 0.718,
+      "loss": 0.8389,
       "step": 7795
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.681425776559428e-05,
-      "loss": 0.832,
+      "loss": 0.8332,
       "step": 7796
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.681339565641534e-05,
-      "loss": 0.7488,
+      "loss": 0.8899,
       "step": 7797
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6812533452710202e-05,
-      "loss": 0.8415,
+      "loss": 0.986,
       "step": 7798
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6811671154490828e-05,
-      "loss": 0.8321,
+      "loss": 0.8746,
       "step": 7799
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6810808761769185e-05,
-      "loss": 0.8745,
+      "loss": 0.9487,
       "step": 7800
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6809946274557228e-05,
-      "loss": 0.8143,
+      "loss": 0.8725,
       "step": 7801
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6809083692866935e-05,
-      "loss": 0.8757,
+      "loss": 0.9403,
       "step": 7802
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.680822101671026e-05,
-      "loss": 0.8487,
+      "loss": 0.8966,
       "step": 7803
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.680735824609918e-05,
-      "loss": 0.8247,
+      "loss": 0.8636,
       "step": 7804
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6806495381045668e-05,
-      "loss": 0.8557,
+      "loss": 0.9168,
       "step": 7805
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6805632421561686e-05,
-      "loss": 0.6943,
+      "loss": 0.8749,
       "step": 7806
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6804769367659205e-05,
-      "loss": 0.8276,
+      "loss": 0.8822,
       "step": 7807
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.680390621935021e-05,
-      "loss": 0.8542,
+      "loss": 0.9398,
       "step": 7808
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6803042976646667e-05,
-      "loss": 0.8706,
+      "loss": 0.9266,
       "step": 7809
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6802179639560557e-05,
-      "loss": 0.7218,
+      "loss": 0.8885,
       "step": 7810
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6801316208103854e-05,
-      "loss": 0.9349,
+      "loss": 1.0282,
       "step": 7811
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.680045268228854e-05,
-      "loss": 0.8,
+      "loss": 0.9062,
       "step": 7812
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6799589062126595e-05,
-      "loss": 0.7566,
+      "loss": 0.8809,
       "step": 7813
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6798725347629993e-05,
-      "loss": 0.7069,
+      "loss": 0.78,
       "step": 7814
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6797861538810726e-05,
-      "loss": 0.7427,
+      "loss": 0.9036,
       "step": 7815
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6796997635680775e-05,
-      "loss": 0.6902,
+      "loss": 0.8322,
       "step": 7816
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6796133638252126e-05,
-      "loss": 0.7676,
+      "loss": 0.9309,
       "step": 7817
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6795269546536768e-05,
-      "loss": 0.8092,
+      "loss": 0.9572,
       "step": 7818
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6794405360546682e-05,
-      "loss": 0.8201,
+      "loss": 0.861,
       "step": 7819
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6793541080293865e-05,
-      "loss": 0.7986,
+      "loss": 0.9786,
       "step": 7820
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6792676705790302e-05,
-      "loss": 0.9039,
+      "loss": 0.933,
       "step": 7821
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6791812237047988e-05,
-      "loss": 0.6985,
+      "loss": 0.9149,
       "step": 7822
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6790947674078913e-05,
-      "loss": 0.7779,
+      "loss": 0.9168,
       "step": 7823
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.679008301689508e-05,
-      "loss": 0.7522,
+      "loss": 0.8691,
       "step": 7824
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6789218265508472e-05,
-      "loss": 0.8043,
+      "loss": 0.9306,
       "step": 7825
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.67883534199311e-05,
-      "loss": 0.8403,
+      "loss": 0.9784,
       "step": 7826
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.678748848017495e-05,
-      "loss": 0.7299,
+      "loss": 0.9244,
       "step": 7827
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.678662344625203e-05,
-      "loss": 0.6996,
+      "loss": 0.9156,
       "step": 7828
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.678575831817434e-05,
-      "loss": 0.8852,
+      "loss": 0.9451,
       "step": 7829
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6784893095953878e-05,
-      "loss": 0.801,
+      "loss": 0.9167,
       "step": 7830
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6784027779602654e-05,
-      "loss": 0.8922,
+      "loss": 0.9566,
       "step": 7831
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6783162369132667e-05,
-      "loss": 0.878,
+      "loss": 0.9305,
       "step": 7832
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6782296864555924e-05,
-      "loss": 0.7477,
+      "loss": 0.8156,
       "step": 7833
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.678143126588444e-05,
-      "loss": 0.7179,
+      "loss": 0.8409,
       "step": 7834
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6780565573130218e-05,
-      "loss": 0.8386,
+      "loss": 0.9118,
       "step": 7835
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.6779699786305268e-05,
-      "loss": 0.8999,
+      "loss": 0.9418,
       "step": 7836
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.67788339054216e-05,
-      "loss": 0.6483,
+      "loss": 0.8157,
       "step": 7837
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6777967930491233e-05,
-      "loss": 0.6859,
+      "loss": 0.8037,
       "step": 7838
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6777101861526177e-05,
-      "loss": 0.8882,
+      "loss": 0.9485,
       "step": 7839
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.677623569853845e-05,
-      "loss": 0.7032,
+      "loss": 0.8758,
       "step": 7840
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6775369441540065e-05,
-      "loss": 0.8257,
+      "loss": 0.9583,
       "step": 7841
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.677450309054304e-05,
-      "loss": 0.7892,
+      "loss": 0.9548,
       "step": 7842
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6773636645559398e-05,
-      "loss": 0.7477,
+      "loss": 0.8525,
       "step": 7843
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6772770106601157e-05,
-      "loss": 0.7413,
+      "loss": 0.8153,
       "step": 7844
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6771903473680342e-05,
-      "loss": 0.7199,
+      "loss": 0.8652,
       "step": 7845
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6771036746808974e-05,
-      "loss": 0.7909,
+      "loss": 0.918,
       "step": 7846
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6770169925999077e-05,
-      "loss": 0.8683,
+      "loss": 0.9117,
       "step": 7847
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6769303011262682e-05,
-      "loss": 0.7957,
+      "loss": 0.8671,
       "step": 7848
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6768436002611808e-05,
-      "loss": 0.8878,
+      "loss": 0.9698,
       "step": 7849
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6767568900058488e-05,
-      "loss": 0.7859,
+      "loss": 0.901,
       "step": 7850
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.676670170361475e-05,
-      "loss": 0.707,
+      "loss": 0.8482,
       "step": 7851
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.676583441329263e-05,
-      "loss": 0.8795,
+      "loss": 0.8883,
       "step": 7852
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6764967029104153e-05,
-      "loss": 0.8544,
+      "loss": 0.954,
       "step": 7853
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6764099551061358e-05,
-      "loss": 0.7883,
+      "loss": 0.9586,
       "step": 7854
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.676323197917628e-05,
-      "loss": 0.8587,
+      "loss": 0.9251,
       "step": 7855
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6762364313460956e-05,
-      "loss": 0.8115,
+      "loss": 0.9132,
       "step": 7856
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6761496553927416e-05,
-      "loss": 0.7559,
+      "loss": 0.8293,
       "step": 7857
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.676062870058771e-05,
-      "loss": 0.7917,
+      "loss": 0.8961,
       "step": 7858
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.675976075345387e-05,
-      "loss": 0.6814,
+      "loss": 0.858,
       "step": 7859
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6758892712537943e-05,
-      "loss": 0.7495,
+      "loss": 0.8428,
       "step": 7860
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6758024577851965e-05,
-      "loss": 0.8435,
+      "loss": 1.0247,
       "step": 7861
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6757156349407988e-05,
-      "loss": 0.886,
+      "loss": 0.9771,
       "step": 7862
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6756288027218053e-05,
-      "loss": 0.9252,
+      "loss": 0.9366,
       "step": 7863
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6755419611294206e-05,
-      "loss": 0.653,
+      "loss": 0.87,
       "step": 7864
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6754551101648497e-05,
-      "loss": 0.7448,
+      "loss": 0.8607,
       "step": 7865
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6753682498292975e-05,
-      "loss": 0.7962,
+      "loss": 0.7802,
       "step": 7866
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6752813801239692e-05,
-      "loss": 0.6986,
+      "loss": 0.8385,
       "step": 7867
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.67519450105007e-05,
-      "loss": 0.8234,
+      "loss": 0.9094,
       "step": 7868
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6751076126088048e-05,
-      "loss": 0.9022,
+      "loss": 1.007,
       "step": 7869
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6750207148013797e-05,
-      "loss": 0.8678,
+      "loss": 0.989,
       "step": 7870
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.674933807628999e-05,
-      "loss": 0.835,
+      "loss": 0.9334,
       "step": 7871
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.67484689109287e-05,
-      "loss": 0.87,
+      "loss": 0.9162,
       "step": 7872
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.674759965194198e-05,
-      "loss": 0.7918,
+      "loss": 0.9127,
       "step": 7873
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6746730299341886e-05,
-      "loss": 0.8376,
+      "loss": 0.9559,
       "step": 7874
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.674586085314048e-05,
-      "loss": 0.8755,
+      "loss": 1.0223,
       "step": 7875
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.674499131334983e-05,
-      "loss": 0.8363,
+      "loss": 0.9238,
       "step": 7876
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6744121679981992e-05,
-      "loss": 0.821,
+      "loss": 0.9797,
       "step": 7877
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6743251953049036e-05,
-      "loss": 0.8906,
+      "loss": 0.9796,
       "step": 7878
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6742382132563023e-05,
-      "loss": 0.7338,
+      "loss": 0.9485,
       "step": 7879
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.674151221853603e-05,
-      "loss": 0.7662,
+      "loss": 0.8609,
       "step": 7880
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6740642210980117e-05,
-      "loss": 0.8309,
+      "loss": 0.9594,
       "step": 7881
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6739772109907355e-05,
-      "loss": 0.7788,
+      "loss": 0.8961,
       "step": 7882
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.673890191532982e-05,
-      "loss": 0.6955,
+      "loss": 0.8219,
       "step": 7883
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6738031627259583e-05,
-      "loss": 0.6916,
+      "loss": 0.8804,
       "step": 7884
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6737161245708713e-05,
-      "loss": 0.74,
+      "loss": 0.7788,
       "step": 7885
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6736290770689295e-05,
-      "loss": 0.6856,
+      "loss": 0.8708,
       "step": 7886
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6735420202213396e-05,
-      "loss": 0.8638,
+      "loss": 0.9542,
       "step": 7887
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.67345495402931e-05,
-      "loss": 0.821,
+      "loss": 0.9622,
       "step": 7888
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.673367878494048e-05,
-      "loss": 0.7586,
+      "loss": 0.9497,
       "step": 7889
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6732807936167626e-05,
-      "loss": 0.6653,
+      "loss": 0.8887,
       "step": 7890
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6731936993986614e-05,
-      "loss": 0.7681,
+      "loss": 0.8953,
       "step": 7891
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6731065958409524e-05,
-      "loss": 0.6971,
+      "loss": 0.8425,
       "step": 7892
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6730194829448446e-05,
-      "loss": 0.9346,
+      "loss": 0.9986,
       "step": 7893
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6729323607115465e-05,
-      "loss": 0.7866,
+      "loss": 0.955,
       "step": 7894
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6728452291422664e-05,
-      "loss": 0.737,
+      "loss": 0.8035,
       "step": 7895
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6727580882382137e-05,
-      "loss": 0.7205,
+      "loss": 0.8858,
       "step": 7896
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6726709380005968e-05,
-      "loss": 0.7152,
+      "loss": 0.8733,
       "step": 7897
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6725837784306254e-05,
-      "loss": 0.8551,
+      "loss": 0.9889,
       "step": 7898
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.672496609529508e-05,
-      "loss": 0.6896,
+      "loss": 0.8264,
       "step": 7899
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.672409431298454e-05,
-      "loss": 0.8853,
+      "loss": 0.9454,
       "step": 7900
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6723222437386736e-05,
-      "loss": 0.7386,
+      "loss": 0.9297,
       "step": 7901
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6722350468513758e-05,
-      "loss": 0.8823,
+      "loss": 0.9443,
       "step": 7902
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6721478406377706e-05,
-      "loss": 0.6892,
+      "loss": 0.853,
       "step": 7903
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.672060625099068e-05,
-      "loss": 0.8123,
+      "loss": 0.9055,
       "step": 7904
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6719734002364775e-05,
-      "loss": 0.8065,
+      "loss": 0.9096,
       "step": 7905
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6718861660512094e-05,
-      "loss": 0.7631,
+      "loss": 0.867,
       "step": 7906
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6717989225444742e-05,
-      "loss": 0.7994,
+      "loss": 0.9167,
       "step": 7907
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.671711669717482e-05,
-      "loss": 0.9692,
+      "loss": 0.9762,
       "step": 7908
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6716244075714433e-05,
-      "loss": 0.7133,
+      "loss": 0.9192,
       "step": 7909
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.671537136107569e-05,
-      "loss": 0.8806,
+      "loss": 0.9016,
       "step": 7910
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6714498553270698e-05,
-      "loss": 0.9972,
+      "loss": 0.9581,
       "step": 7911
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6713625652311566e-05,
-      "loss": 0.9336,
+      "loss": 0.9837,
       "step": 7912
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.67127526582104e-05,
-      "loss": 0.7685,
+      "loss": 0.8305,
       "step": 7913
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6711879570979317e-05,
-      "loss": 0.8399,
+      "loss": 0.8865,
       "step": 7914
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6711006390630426e-05,
-      "loss": 0.8122,
+      "loss": 0.8576,
       "step": 7915
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6710133117175846e-05,
-      "loss": 0.9617,
+      "loss": 1.0194,
       "step": 7916
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.670925975062769e-05,
-      "loss": 0.8211,
+      "loss": 0.8725,
       "step": 7917
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.670838629099807e-05,
-      "loss": 0.8575,
+      "loss": 0.9557,
       "step": 7918
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.670751273829911e-05,
-      "loss": 0.8592,
+      "loss": 0.9065,
       "step": 7919
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6706639092542927e-05,
-      "loss": 0.7113,
+      "loss": 0.8404,
       "step": 7920
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6705765353741643e-05,
-      "loss": 0.8254,
+      "loss": 0.958,
       "step": 7921
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.670489152190738e-05,
-      "loss": 0.7101,
+      "loss": 0.8293,
       "step": 7922
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.670401759705226e-05,
-      "loss": 0.8252,
+      "loss": 0.958,
       "step": 7923
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6703143579188407e-05,
-      "loss": 0.8993,
+      "loss": 0.942,
       "step": 7924
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6702269468327948e-05,
-      "loss": 0.751,
+      "loss": 0.9788,
       "step": 7925
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6701395264483007e-05,
-      "loss": 0.8265,
+      "loss": 0.8631,
       "step": 7926
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6700520967665718e-05,
-      "loss": 0.7663,
+      "loss": 0.8857,
       "step": 7927
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6699646577888202e-05,
-      "loss": 0.7897,
+      "loss": 0.8979,
       "step": 7928
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66987720951626e-05,
-      "loss": 0.8419,
+      "loss": 0.9505,
       "step": 7929
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6697897519501038e-05,
-      "loss": 0.7731,
+      "loss": 0.9008,
       "step": 7930
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.669702285091565e-05,
-      "loss": 0.7143,
+      "loss": 0.889,
       "step": 7931
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6696148089418574e-05,
-      "loss": 0.6547,
+      "loss": 0.814,
       "step": 7932
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6695273235021945e-05,
-      "loss": 0.7552,
+      "loss": 0.9041,
       "step": 7933
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6694398287737897e-05,
-      "loss": 0.7259,
+      "loss": 0.895,
       "step": 7934
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.669352324757857e-05,
-      "loss": 0.7059,
+      "loss": 0.9339,
       "step": 7935
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6692648114556105e-05,
-      "loss": 0.7691,
+      "loss": 0.8457,
       "step": 7936
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6691772888682643e-05,
-      "loss": 0.8815,
+      "loss": 0.8935,
       "step": 7937
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.669089756997033e-05,
-      "loss": 0.9394,
+      "loss": 0.9823,
       "step": 7938
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6690022158431304e-05,
-      "loss": 0.8158,
+      "loss": 0.9924,
       "step": 7939
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6689146654077712e-05,
-      "loss": 0.8703,
+      "loss": 0.8672,
       "step": 7940
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66882710569217e-05,
-      "loss": 0.8167,
+      "loss": 0.922,
       "step": 7941
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6687395366975417e-05,
-      "loss": 0.7677,
+      "loss": 0.8403,
       "step": 7942
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6686519584251013e-05,
-      "loss": 0.6841,
+      "loss": 0.8264,
       "step": 7943
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.668564370876064e-05,
-      "loss": 0.7474,
+      "loss": 0.8676,
       "step": 7944
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.668476774051644e-05,
-      "loss": 0.7438,
+      "loss": 0.8865,
       "step": 7945
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6683891679530573e-05,
-      "loss": 0.8185,
+      "loss": 1.0022,
       "step": 7946
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.668301552581519e-05,
-      "loss": 0.8701,
+      "loss": 0.8839,
       "step": 7947
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.668213927938245e-05,
-      "loss": 0.6836,
+      "loss": 0.8303,
       "step": 7948
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6681262940244514e-05,
-      "loss": 0.7573,
+      "loss": 0.865,
       "step": 7949
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.668038650841353e-05,
-      "loss": 0.8627,
+      "loss": 0.9366,
       "step": 7950
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.667950998390166e-05,
-      "loss": 0.7852,
+      "loss": 0.9353,
       "step": 7951
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6678633366721068e-05,
-      "loss": 0.9348,
+      "loss": 0.9771,
       "step": 7952
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6677756656883915e-05,
-      "loss": 0.7885,
+      "loss": 0.9036,
       "step": 7953
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.667687985440236e-05,
-      "loss": 0.8658,
+      "loss": 0.9058,
       "step": 7954
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.667600295928857e-05,
-      "loss": 0.9199,
+      "loss": 0.9542,
       "step": 7955
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6675125971554713e-05,
-      "loss": 0.8336,
+      "loss": 0.9368,
       "step": 7956
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6674248891212954e-05,
-      "loss": 0.7446,
+      "loss": 0.8364,
       "step": 7957
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6673371718275458e-05,
-      "loss": 0.7719,
+      "loss": 0.9301,
       "step": 7958
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66724944527544e-05,
-      "loss": 0.8517,
+      "loss": 0.9409,
       "step": 7959
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.667161709466195e-05,
-      "loss": 0.6942,
+      "loss": 0.7897,
       "step": 7960
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6670739644010276e-05,
-      "loss": 0.7525,
+      "loss": 0.8998,
       "step": 7961
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6669862100811555e-05,
-      "loss": 0.7493,
+      "loss": 0.8768,
       "step": 7962
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6668984465077963e-05,
-      "loss": 0.8484,
+      "loss": 0.9426,
       "step": 7963
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666810673682167e-05,
-      "loss": 0.6705,
+      "loss": 0.8423,
       "step": 7964
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666722891605486e-05,
-      "loss": 0.7989,
+      "loss": 0.9901,
       "step": 7965
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6666351002789706e-05,
-      "loss": 0.7746,
+      "loss": 0.8916,
       "step": 7966
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666547299703839e-05,
-      "loss": 0.6852,
+      "loss": 0.7765,
       "step": 7967
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6664594898813094e-05,
-      "loss": 0.7084,
+      "loss": 0.8687,
       "step": 7968
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6663716708126e-05,
-      "loss": 0.82,
+      "loss": 0.9968,
       "step": 7969
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666283842498929e-05,
-      "loss": 0.7764,
+      "loss": 0.8457,
       "step": 7970
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666196004941515e-05,
-      "loss": 0.842,
+      "loss": 0.9231,
       "step": 7971
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6661081581415768e-05,
-      "loss": 0.8304,
+      "loss": 0.967,
       "step": 7972
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.666020302100333e-05,
-      "loss": 0.8858,
+      "loss": 0.9749,
       "step": 7973
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6659324368190024e-05,
-      "loss": 0.9171,
+      "loss": 0.9726,
       "step": 7974
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.665844562298804e-05,
-      "loss": 0.9284,
+      "loss": 0.9387,
       "step": 7975
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.665756678540957e-05,
-      "loss": 0.8028,
+      "loss": 0.9324,
       "step": 7976
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6656687855466807e-05,
-      "loss": 0.7486,
+      "loss": 0.8648,
       "step": 7977
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6655808833171947e-05,
-      "loss": 0.8007,
+      "loss": 0.9736,
       "step": 7978
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.665492971853718e-05,
-      "loss": 0.7665,
+      "loss": 0.8271,
       "step": 7979
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6654050511574707e-05,
-      "loss": 0.773,
+      "loss": 0.8716,
       "step": 7980
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6653171212296722e-05,
-      "loss": 0.7646,
+      "loss": 0.8188,
       "step": 7981
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.665229182071543e-05,
-      "loss": 0.9148,
+      "loss": 0.8976,
       "step": 7982
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6651412336843025e-05,
-      "loss": 0.8704,
+      "loss": 0.8838,
       "step": 7983
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6650532760691708e-05,
-      "loss": 0.7387,
+      "loss": 0.8713,
       "step": 7984
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6649653092273688e-05,
-      "loss": 0.8294,
+      "loss": 0.9215,
       "step": 7985
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6648773331601166e-05,
-      "loss": 0.8673,
+      "loss": 0.9332,
       "step": 7986
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6647893478686346e-05,
-      "loss": 0.7507,
+      "loss": 0.9088,
       "step": 7987
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6647013533541436e-05,
-      "loss": 0.7724,
+      "loss": 0.9122,
       "step": 7988
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6646133496178646e-05,
-      "loss": 0.8104,
+      "loss": 0.8918,
       "step": 7989
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.664525336661018e-05,
-      "loss": 0.8149,
+      "loss": 0.9193,
       "step": 7990
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6644373144848257e-05,
-      "loss": 0.8287,
+      "loss": 0.9436,
       "step": 7991
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.664349283090508e-05,
-      "loss": 0.7665,
+      "loss": 0.8813,
       "step": 7992
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.664261242479287e-05,
-      "loss": 0.7036,
+      "loss": 0.8154,
       "step": 7993
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.664173192652383e-05,
-      "loss": 0.8236,
+      "loss": 0.9586,
       "step": 7994
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.664085133611019e-05,
-      "loss": 0.7372,
+      "loss": 0.7699,
       "step": 7995
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6639970653564155e-05,
-      "loss": 0.8209,
+      "loss": 0.9119,
       "step": 7996
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.663908987889795e-05,
-      "loss": 0.9246,
+      "loss": 1.0046,
       "step": 7997
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.663820901212379e-05,
-      "loss": 0.9861,
+      "loss": 1.0188,
       "step": 7998
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66373280532539e-05,
-      "loss": 0.8057,
+      "loss": 0.9119,
       "step": 7999
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6636447002300505e-05,
-      "loss": 0.7098,
+      "loss": 0.906,
       "step": 8000
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.663556585927582e-05,
-      "loss": 0.8722,
+      "loss": 0.935,
       "step": 8001
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6634684624192075e-05,
-      "loss": 0.8651,
+      "loss": 0.8915,
       "step": 8002
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.663380329706149e-05,
-      "loss": 0.9437,
+      "loss": 0.9368,
       "step": 8003
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66329218778963e-05,
-      "loss": 0.8214,
+      "loss": 0.9179,
       "step": 8004
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6632040366708727e-05,
-      "loss": 0.7727,
+      "loss": 0.8642,
       "step": 8005
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6631158763511005e-05,
-      "loss": 0.716,
+      "loss": 0.9313,
       "step": 8006
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6630277068315365e-05,
-      "loss": 0.7013,
+      "loss": 0.8258,
       "step": 8007
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6629395281134037e-05,
-      "loss": 0.7319,
+      "loss": 0.9598,
       "step": 8008
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6628513401979252e-05,
-      "loss": 0.9102,
+      "loss": 0.9633,
       "step": 8009
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6627631430863252e-05,
-      "loss": 0.7829,
+      "loss": 0.8936,
       "step": 8010
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.662674936779827e-05,
-      "loss": 0.8607,
+      "loss": 0.9666,
       "step": 8011
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.662586721279654e-05,
-      "loss": 0.8375,
+      "loss": 0.9424,
       "step": 8012
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6624984965870305e-05,
-      "loss": 0.8535,
+      "loss": 0.9461,
       "step": 8013
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66241026270318e-05,
-      "loss": 0.7526,
+      "loss": 0.908,
       "step": 8014
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6623220196293274e-05,
-      "loss": 0.7708,
+      "loss": 0.9075,
       "step": 8015
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6622337673666963e-05,
-      "loss": 0.7595,
+      "loss": 0.8717,
       "step": 8016
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6621455059165113e-05,
-      "loss": 0.8713,
+      "loss": 0.9349,
       "step": 8017
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6620572352799968e-05,
-      "loss": 0.7906,
+      "loss": 0.9287,
       "step": 8018
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6619689554583773e-05,
-      "loss": 0.857,
+      "loss": 1.0157,
       "step": 8019
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6618806664528783e-05,
-      "loss": 0.8097,
+      "loss": 0.9075,
       "step": 8020
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.661792368264724e-05,
-      "loss": 0.62,
+      "loss": 0.8295,
       "step": 8021
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6617040608951393e-05,
-      "loss": 0.7768,
+      "loss": 0.9384,
       "step": 8022
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6616157443453492e-05,
-      "loss": 0.7204,
+      "loss": 0.8724,
       "step": 8023
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6615274186165803e-05,
-      "loss": 0.8396,
+      "loss": 0.9853,
       "step": 8024
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6614390837100565e-05,
-      "loss": 0.8912,
+      "loss": 0.9744,
       "step": 8025
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6613507396270038e-05,
-      "loss": 0.8119,
+      "loss": 0.8782,
       "step": 8026
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6612623863686483e-05,
-      "loss": 0.8679,
+      "loss": 0.9543,
       "step": 8027
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6611740239362148e-05,
-      "loss": 0.7076,
+      "loss": 0.8652,
       "step": 8028
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.66108565233093e-05,
-      "loss": 0.8845,
+      "loss": 1.0119,
       "step": 8029
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6609972715540198e-05,
-      "loss": 0.8339,
+      "loss": 0.9114,
       "step": 8030
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6609088816067104e-05,
-      "loss": 0.8624,
+      "loss": 0.9676,
       "step": 8031
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6608204824902283e-05,
-      "loss": 0.7664,
+      "loss": 0.9017,
       "step": 8032
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.660732074205799e-05,
-      "loss": 0.7868,
+      "loss": 0.9253,
       "step": 8033
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6606436567546497e-05,
-      "loss": 0.8084,
+      "loss": 0.9091,
       "step": 8034
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6605552301380073e-05,
-      "loss": 0.7229,
+      "loss": 0.8644,
       "step": 8035
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6604667943570983e-05,
-      "loss": 0.752,
+      "loss": 1.0053,
       "step": 8036
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6603783494131495e-05,
-      "loss": 0.7448,
+      "loss": 0.8601,
       "step": 8037
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6602898953073877e-05,
-      "loss": 0.8133,
+      "loss": 0.8684,
       "step": 8038
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.660201432041041e-05,
-      "loss": 0.9643,
+      "loss": 1.0018,
       "step": 8039
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.660112959615336e-05,
-      "loss": 0.79,
+      "loss": 0.8817,
       "step": 8040
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6600244780315e-05,
-      "loss": 0.7453,
+      "loss": 0.8837,
       "step": 8041
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.659935987290761e-05,
-      "loss": 0.8853,
+      "loss": 0.9433,
       "step": 8042
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6598474873943464e-05,
-      "loss": 0.7861,
+      "loss": 0.9593,
       "step": 8043
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6597589783434842e-05,
-      "loss": 0.728,
+      "loss": 0.9123,
       "step": 8044
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6596704601394024e-05,
-      "loss": 0.7139,
+      "loss": 0.8545,
       "step": 8045
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6595819327833287e-05,
-      "loss": 0.8213,
+      "loss": 0.8818,
       "step": 8046
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6594933962764915e-05,
-      "loss": 0.8511,
+      "loss": 0.9333,
       "step": 8047
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.659404850620119e-05,
-      "loss": 0.9496,
+      "loss": 1.0012,
       "step": 8048
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.65931629581544e-05,
-      "loss": 0.7208,
+      "loss": 0.8257,
       "step": 8049
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6592277318636828e-05,
-      "loss": 0.8129,
+      "loss": 0.9377,
       "step": 8050
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.659139158766076e-05,
-      "loss": 0.9394,
+      "loss": 0.9784,
       "step": 8051
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.659050576523849e-05,
-      "loss": 0.818,
+      "loss": 0.929,
       "step": 8052
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.65896198513823e-05,
-      "loss": 0.7747,
+      "loss": 0.8708,
       "step": 8053
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6588733846104483e-05,
-      "loss": 0.7123,
+      "loss": 0.8554,
       "step": 8054
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6587847749417332e-05,
-      "loss": 0.8794,
+      "loss": 0.9702,
       "step": 8055
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6586961561333144e-05,
-      "loss": 0.8838,
+      "loss": 0.899,
       "step": 8056
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6586075281864205e-05,
-      "loss": 0.8199,
+      "loss": 0.8771,
       "step": 8057
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.658518891102282e-05,
-      "loss": 0.8597,
+      "loss": 0.9065,
       "step": 8058
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.658430244882128e-05,
-      "loss": 0.706,
+      "loss": 0.7947,
       "step": 8059
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6583415895271886e-05,
-      "loss": 0.8099,
+      "loss": 0.9129,
       "step": 8060
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.658252925038694e-05,
-      "loss": 0.6626,
+      "loss": 0.8249,
       "step": 8061
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6581642514178735e-05,
-      "loss": 0.8517,
+      "loss": 0.8418,
       "step": 8062
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.658075568665958e-05,
-      "loss": 0.8135,
+      "loss": 0.8929,
       "step": 8063
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6579868767841776e-05,
-      "loss": 0.7477,
+      "loss": 0.8007,
       "step": 8064
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6578981757737632e-05,
-      "loss": 0.7864,
+      "loss": 0.935,
       "step": 8065
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6578094656359448e-05,
-      "loss": 0.9019,
+      "loss": 0.9556,
       "step": 8066
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.657720746371953e-05,
-      "loss": 0.8624,
+      "loss": 0.965,
       "step": 8067
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6576320179830198e-05,
-      "loss": 0.8047,
+      "loss": 0.9861,
       "step": 8068
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6575432804703748e-05,
-      "loss": 0.8154,
+      "loss": 0.9354,
       "step": 8069
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.65745453383525e-05,
-      "loss": 0.7209,
+      "loss": 0.9131,
       "step": 8070
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.657365778078876e-05,
-      "loss": 0.7431,
+      "loss": 0.8366,
       "step": 8071
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.657277013202485e-05,
-      "loss": 0.7524,
+      "loss": 0.7674,
       "step": 8072
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6571882392073076e-05,
-      "loss": 0.7153,
+      "loss": 0.7824,
       "step": 8073
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.657099456094576e-05,
-      "loss": 0.7655,
+      "loss": 0.9337,
       "step": 8074
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6570106638655218e-05,
-      "loss": 0.7649,
+      "loss": 0.8759,
       "step": 8075
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6569218625213768e-05,
-      "loss": 0.6043,
+      "loss": 0.7888,
       "step": 8076
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.656833052063373e-05,
-      "loss": 0.7862,
+      "loss": 0.9057,
       "step": 8077
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6567442324927427e-05,
-      "loss": 0.9048,
+      "loss": 1.0014,
       "step": 8078
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6566554038107178e-05,
-      "loss": 0.8252,
+      "loss": 0.8906,
       "step": 8079
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6565665660185306e-05,
-      "loss": 0.8796,
+      "loss": 0.9136,
       "step": 8080
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.656477719117414e-05,
-      "loss": 0.8227,
+      "loss": 0.9296,
       "step": 8081
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6563888631086003e-05,
-      "loss": 0.7769,
+      "loss": 0.8517,
       "step": 8082
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.656299997993323e-05,
-      "loss": 0.9571,
+      "loss": 1.0395,
       "step": 8083
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6562111237728137e-05,
-      "loss": 0.7415,
+      "loss": 0.8737,
       "step": 8084
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6561222404483067e-05,
-      "loss": 0.7487,
+      "loss": 0.9259,
       "step": 8085
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6560333480210342e-05,
-      "loss": 0.765,
+      "loss": 0.9373,
       "step": 8086
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.65594444649223e-05,
-      "loss": 0.9295,
+      "loss": 0.9072,
       "step": 8087
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.655855535863127e-05,
-      "loss": 0.7989,
+      "loss": 0.9125,
       "step": 8088
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.655766616134959e-05,
-      "loss": 0.8328,
+      "loss": 0.9086,
       "step": 8089
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6556776873089596e-05,
-      "loss": 0.7715,
+      "loss": 0.8562,
       "step": 8090
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6555887493863627e-05,
-      "loss": 0.6901,
+      "loss": 0.9212,
       "step": 8091
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6554998023684025e-05,
-      "loss": 0.7796,
+      "loss": 0.8981,
       "step": 8092
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6554108462563117e-05,
-      "loss": 0.7415,
+      "loss": 0.9112,
       "step": 8093
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6553218810513256e-05,
-      "loss": 0.8309,
+      "loss": 0.8548,
       "step": 8094
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6552329067546785e-05,
-      "loss": 0.8355,
+      "loss": 0.9619,
       "step": 8095
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6551439233676043e-05,
-      "loss": 0.8422,
+      "loss": 0.8838,
       "step": 8096
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6550549308913378e-05,
-      "loss": 0.7697,
+      "loss": 0.9267,
       "step": 8097
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6549659293271134e-05,
-      "loss": 0.8785,
+      "loss": 0.9434,
       "step": 8098
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.654876918676166e-05,
-      "loss": 0.67,
+      "loss": 0.8187,
       "step": 8099
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6547878989397306e-05,
-      "loss": 0.7493,
+      "loss": 0.8959,
       "step": 8100
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.654698870119042e-05,
-      "loss": 0.8375,
+      "loss": 0.9576,
       "step": 8101
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6546098322153354e-05,
-      "loss": 0.8149,
+      "loss": 0.9115,
       "step": 8102
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6545207852298464e-05,
-      "loss": 0.8968,
+      "loss": 0.9719,
       "step": 8103
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.65443172916381e-05,
-      "loss": 0.75,
+      "loss": 0.8877,
       "step": 8104
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6543426640184618e-05,
-      "loss": 0.7945,
+      "loss": 0.9062,
       "step": 8105
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6542535897950377e-05,
-      "loss": 0.7995,
+      "loss": 0.8638,
       "step": 8106
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6541645064947733e-05,
-      "loss": 0.8938,
+      "loss": 1.0262,
       "step": 8107
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6540754141189042e-05,
-      "loss": 0.8081,
+      "loss": 0.9317,
       "step": 8108
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.653986312668667e-05,
-      "loss": 0.8177,
+      "loss": 0.9282,
       "step": 8109
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.6538972021452975e-05,
-      "loss": 0.8899,
+      "loss": 0.9683,
       "step": 8110
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.653808082550032e-05,
-      "loss": 0.7646,
+      "loss": 0.9929,
       "step": 8111
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6537189538841075e-05,
-      "loss": 0.7379,
+      "loss": 0.8538,
       "step": 8112
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6536298161487594e-05,
-      "loss": 0.7575,
+      "loss": 0.8902,
       "step": 8113
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.653540669345225e-05,
-      "loss": 0.6495,
+      "loss": 0.8017,
       "step": 8114
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6534515134747414e-05,
-      "loss": 0.9422,
+      "loss": 0.9696,
       "step": 8115
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6533623485385454e-05,
-      "loss": 0.8152,
+      "loss": 0.9074,
       "step": 8116
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6532731745378738e-05,
-      "loss": 0.9089,
+      "loss": 0.9675,
       "step": 8117
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.653183991473963e-05,
-      "loss": 0.7479,
+      "loss": 0.9455,
       "step": 8118
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.653094799348052e-05,
-      "loss": 0.9401,
+      "loss": 0.9317,
       "step": 8119
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6530055981613763e-05,
-      "loss": 0.7301,
+      "loss": 0.796,
       "step": 8120
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6529163879151752e-05,
-      "loss": 0.8379,
+      "loss": 0.9042,
       "step": 8121
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652827168610685e-05,
-      "loss": 0.684,
+      "loss": 0.7846,
       "step": 8122
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6527379402491445e-05,
-      "loss": 0.8643,
+      "loss": 0.8715,
       "step": 8123
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6526487028317914e-05,
-      "loss": 0.8121,
+      "loss": 0.8916,
       "step": 8124
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652559456359863e-05,
-      "loss": 0.5734,
+      "loss": 0.7946,
       "step": 8125
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652470200834598e-05,
-      "loss": 0.8425,
+      "loss": 0.8569,
       "step": 8126
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652380936257235e-05,
-      "loss": 0.8667,
+      "loss": 0.9521,
       "step": 8127
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652291662629012e-05,
-      "loss": 0.8548,
+      "loss": 0.9963,
       "step": 8128
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.652202379951168e-05,
-      "loss": 0.7683,
+      "loss": 0.9142,
       "step": 8129
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6521130882249408e-05,
-      "loss": 0.6482,
+      "loss": 0.7715,
       "step": 8130
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.65202378745157e-05,
-      "loss": 0.8467,
+      "loss": 0.9353,
       "step": 8131
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.651934477632294e-05,
-      "loss": 0.8338,
+      "loss": 0.9176,
       "step": 8132
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6518451587683524e-05,
-      "loss": 0.8738,
+      "loss": 0.9076,
       "step": 8133
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6517558308609842e-05,
-      "loss": 0.6843,
+      "loss": 0.8403,
       "step": 8134
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6516664939114284e-05,
-      "loss": 0.8147,
+      "loss": 0.9238,
       "step": 8135
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6515771479209245e-05,
-      "loss": 0.7472,
+      "loss": 0.8891,
       "step": 8136
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6514877928907122e-05,
-      "loss": 0.7983,
+      "loss": 0.9114,
       "step": 8137
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.651398428822031e-05,
-      "loss": 0.7708,
+      "loss": 0.8515,
       "step": 8138
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.651309055716121e-05,
-      "loss": 0.6959,
+      "loss": 0.8049,
       "step": 8139
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6512196735742217e-05,
-      "loss": 0.8976,
+      "loss": 1.017,
       "step": 8140
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6511302823975736e-05,
-      "loss": 0.8593,
+      "loss": 0.9985,
       "step": 8141
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6510408821874167e-05,
-      "loss": 0.7814,
+      "loss": 0.8931,
       "step": 8142
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650951472944991e-05,
-      "loss": 0.8853,
+      "loss": 0.9598,
       "step": 8143
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6508620546715374e-05,
-      "loss": 0.6812,
+      "loss": 0.792,
       "step": 8144
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650772627368296e-05,
-      "loss": 0.8626,
+      "loss": 0.9321,
       "step": 8145
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650683191036508e-05,
-      "loss": 0.7508,
+      "loss": 0.9014,
       "step": 8146
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650593745677414e-05,
-      "loss": 0.8746,
+      "loss": 0.9216,
       "step": 8147
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6505042912922546e-05,
-      "loss": 0.627,
+      "loss": 0.7796,
       "step": 8148
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6504148278822712e-05,
-      "loss": 0.7807,
+      "loss": 0.8201,
       "step": 8149
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650325355448705e-05,
-      "loss": 0.7685,
+      "loss": 0.7969,
       "step": 8150
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6502358739927967e-05,
-      "loss": 0.6483,
+      "loss": 0.7762,
       "step": 8151
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6501463835157887e-05,
-      "loss": 0.8038,
+      "loss": 0.866,
       "step": 8152
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.650056884018922e-05,
-      "loss": 0.654,
+      "loss": 0.7546,
       "step": 8153
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649967375503438e-05,
-      "loss": 0.9352,
+      "loss": 0.9638,
       "step": 8154
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6498778579705793e-05,
-      "loss": 0.6547,
+      "loss": 0.7956,
       "step": 8155
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649788331421587e-05,
-      "loss": 0.8059,
+      "loss": 0.8767,
       "step": 8156
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649698795857704e-05,
-      "loss": 0.7201,
+      "loss": 0.8213,
       "step": 8157
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649609251280172e-05,
-      "loss": 0.8543,
+      "loss": 0.9406,
       "step": 8158
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649519697690233e-05,
-      "loss": 0.6821,
+      "loss": 0.8797,
       "step": 8159
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6494301350891298e-05,
-      "loss": 0.7553,
+      "loss": 0.9437,
       "step": 8160
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6493405634781047e-05,
-      "loss": 0.8967,
+      "loss": 0.9362,
       "step": 8161
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.649250982858401e-05,
-      "loss": 0.7302,
+      "loss": 0.8576,
       "step": 8162
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6491613932312606e-05,
-      "loss": 0.7182,
+      "loss": 0.835,
       "step": 8163
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6490717945979275e-05,
-      "loss": 0.7632,
+      "loss": 0.9013,
       "step": 8164
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6489821869596438e-05,
-      "loss": 0.8687,
+      "loss": 0.9093,
       "step": 8165
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.648892570317653e-05,
-      "loss": 0.8007,
+      "loss": 0.9465,
       "step": 8166
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6488029446731985e-05,
-      "loss": 0.7993,
+      "loss": 0.8373,
       "step": 8167
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6487133100275237e-05,
-      "loss": 0.6632,
+      "loss": 0.9392,
       "step": 8168
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6486236663818718e-05,
-      "loss": 0.7978,
+      "loss": 0.9078,
       "step": 8169
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6485340137374872e-05,
-      "loss": 0.722,
+      "loss": 0.7937,
       "step": 8170
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6484443520956134e-05,
-      "loss": 0.8288,
+      "loss": 0.8956,
       "step": 8171
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.648354681457494e-05,
-      "loss": 0.6063,
+      "loss": 0.9398,
       "step": 8172
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.648265001824373e-05,
-      "loss": 0.7598,
+      "loss": 0.8259,
       "step": 8173
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6481753131974957e-05,
-      "loss": 0.6684,
+      "loss": 0.8258,
       "step": 8174
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.648085615578105e-05,
-      "loss": 0.7901,
+      "loss": 0.8971,
       "step": 8175
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647995908967446e-05,
-      "loss": 0.8091,
+      "loss": 0.9287,
       "step": 8176
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647906193366763e-05,
-      "loss": 0.7935,
+      "loss": 0.949,
       "step": 8177
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6478164687773012e-05,
-      "loss": 0.776,
+      "loss": 0.85,
       "step": 8178
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6477267352003046e-05,
-      "loss": 0.7398,
+      "loss": 0.8376,
       "step": 8179
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647636992637019e-05,
-      "loss": 0.7868,
+      "loss": 0.9063,
       "step": 8180
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6475472410886886e-05,
-      "loss": 0.9094,
+      "loss": 0.9441,
       "step": 8181
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647457480556559e-05,
-      "loss": 0.7767,
+      "loss": 0.9425,
       "step": 8182
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647367711041876e-05,
-      "loss": 0.7621,
+      "loss": 0.8504,
       "step": 8183
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6472779325458836e-05,
-      "loss": 0.8085,
+      "loss": 0.8926,
       "step": 8184
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.647188145069829e-05,
-      "loss": 0.7952,
+      "loss": 0.9685,
       "step": 8185
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6470983486149568e-05,
-      "loss": 0.7924,
+      "loss": 0.881,
       "step": 8186
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6470085431825134e-05,
-      "loss": 0.8583,
+      "loss": 0.9032,
       "step": 8187
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.646918728773744e-05,
-      "loss": 0.6394,
+      "loss": 0.7487,
       "step": 8188
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6468289053898952e-05,
-      "loss": 0.8181,
+      "loss": 0.8626,
       "step": 8189
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6467390730322134e-05,
-      "loss": 0.8972,
+      "loss": 1.0225,
       "step": 8190
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6466492317019445e-05,
-      "loss": 0.7154,
+      "loss": 0.8586,
       "step": 8191
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.646559381400335e-05,
-      "loss": 0.6339,
+      "loss": 0.8241,
       "step": 8192
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6464695221286312e-05,
-      "loss": 0.8751,
+      "loss": 0.932,
       "step": 8193
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.64637965388808e-05,
-      "loss": 0.7421,
+      "loss": 0.8647,
       "step": 8194
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6462897766799282e-05,
-      "loss": 0.8008,
+      "loss": 0.9127,
       "step": 8195
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.646199890505423e-05,
-      "loss": 0.8296,
+      "loss": 1.0237,
       "step": 8196
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.646109995365811e-05,
-      "loss": 0.8714,
+      "loss": 0.8766,
       "step": 8197
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6460200912623393e-05,
-      "loss": 0.8297,
+      "loss": 0.9158,
       "step": 8198
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6459301781962558e-05,
-      "loss": 0.8778,
+      "loss": 0.9845,
       "step": 8199
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6458402561688073e-05,
-      "loss": 0.8081,
+      "loss": 0.9587,
       "step": 8200
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6457503251812417e-05,
-      "loss": 0.8049,
+      "loss": 0.9335,
       "step": 8201
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6456603852348065e-05,
-      "loss": 0.9507,
+      "loss": 0.9543,
       "step": 8202
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.645570436330749e-05,
-      "loss": 0.763,
+      "loss": 0.7738,
       "step": 8203
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6454804784703186e-05,
-      "loss": 0.8114,
+      "loss": 0.9384,
       "step": 8204
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6453905116547616e-05,
-      "loss": 0.7618,
+      "loss": 0.9128,
       "step": 8205
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6453005358853274e-05,
-      "loss": 0.8103,
+      "loss": 0.856,
       "step": 8206
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6452105511632635e-05,
-      "loss": 0.9477,
+      "loss": 0.984,
       "step": 8207
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.645120557489819e-05,
-      "loss": 0.923,
+      "loss": 0.9706,
       "step": 8208
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6450305548662418e-05,
-      "loss": 0.8791,
+      "loss": 0.9218,
       "step": 8209
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6449405432937807e-05,
-      "loss": 0.8448,
+      "loss": 0.9323,
       "step": 8210
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.644850522773685e-05,
-      "loss": 0.8811,
+      "loss": 0.9547,
       "step": 8211
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6447604933072032e-05,
-      "loss": 0.8577,
+      "loss": 0.9702,
       "step": 8212
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.644670454895584e-05,
-      "loss": 0.8756,
+      "loss": 0.9073,
       "step": 8213
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6445804075400774e-05,
-      "loss": 0.8057,
+      "loss": 0.9395,
       "step": 8214
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6444903512419318e-05,
-      "loss": 0.9009,
+      "loss": 0.9382,
       "step": 8215
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6444002860023972e-05,
-      "loss": 0.7468,
+      "loss": 0.8268,
       "step": 8216
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6443102118227226e-05,
-      "loss": 0.8743,
+      "loss": 0.9731,
       "step": 8217
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6442201287041583e-05,
-      "loss": 0.8699,
+      "loss": 0.9859,
       "step": 8218
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6441300366479533e-05,
-      "loss": 0.7985,
+      "loss": 0.9586,
       "step": 8219
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6440399356553586e-05,
-      "loss": 0.7602,
+      "loss": 0.8965,
       "step": 8220
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.643949825727623e-05,
-      "loss": 0.8156,
+      "loss": 0.9427,
       "step": 8221
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6438597068659973e-05,
-      "loss": 0.6913,
+      "loss": 0.8283,
       "step": 8222
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6437695790717318e-05,
-      "loss": 0.8228,
+      "loss": 0.9323,
       "step": 8223
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6436794423460768e-05,
-      "loss": 0.7945,
+      "loss": 0.9777,
       "step": 8224
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.643589296690283e-05,
-      "loss": 0.7107,
+      "loss": 0.8225,
       "step": 8225
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6434991421056005e-05,
-      "loss": 0.7045,
+      "loss": 0.8478,
       "step": 8226
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6434089785932804e-05,
-      "loss": 0.8353,
+      "loss": 0.9047,
       "step": 8227
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6433188061545737e-05,
-      "loss": 0.7199,
+      "loss": 0.8512,
       "step": 8228
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6432286247907313e-05,
-      "loss": 0.8411,
+      "loss": 0.8921,
       "step": 8229
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6431384345030044e-05,
-      "loss": 0.8077,
+      "loss": 0.9256,
       "step": 8230
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6430482352926444e-05,
-      "loss": 0.8239,
+      "loss": 0.9355,
       "step": 8231
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6429580271609024e-05,
-      "loss": 0.7694,
+      "loss": 0.918,
       "step": 8232
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.64286781010903e-05,
-      "loss": 0.8026,
+      "loss": 0.964,
       "step": 8233
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6427775841382787e-05,
-      "loss": 0.8502,
+      "loss": 0.9213,
       "step": 8234
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6426873492499006e-05,
-      "loss": 0.7878,
+      "loss": 0.8526,
       "step": 8235
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.642597105445147e-05,
-      "loss": 0.7356,
+      "loss": 0.9144,
       "step": 8236
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.642506852725271e-05,
-      "loss": 0.7943,
+      "loss": 0.8945,
       "step": 8237
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6424165910915232e-05,
-      "loss": 0.8096,
+      "loss": 0.8895,
       "step": 8238
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.642326320545157e-05,
-      "loss": 0.8747,
+      "loss": 0.9783,
       "step": 8239
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6422360410874247e-05,
-      "loss": 0.7018,
+      "loss": 0.8156,
       "step": 8240
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.642145752719578e-05,
-      "loss": 0.7852,
+      "loss": 0.8609,
       "step": 8241
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6420554554428706e-05,
-      "loss": 0.8468,
+      "loss": 0.9157,
       "step": 8242
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6419651492585543e-05,
-      "loss": 0.7939,
+      "loss": 0.8583,
       "step": 8243
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6418748341678827e-05,
-      "loss": 0.6999,
+      "loss": 0.8933,
       "step": 8244
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6417845101721085e-05,
-      "loss": 0.7678,
+      "loss": 0.9148,
       "step": 8245
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6416941772724845e-05,
-      "loss": 0.7186,
+      "loss": 0.8576,
       "step": 8246
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6416038354702647e-05,
-      "loss": 0.6705,
+      "loss": 0.8256,
       "step": 8247
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6415134847667014e-05,
-      "loss": 0.7366,
+      "loss": 0.9684,
       "step": 8248
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.641423125163049e-05,
-      "loss": 0.67,
+      "loss": 0.8104,
       "step": 8249
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.641332756660561e-05,
-      "loss": 0.7083,
+      "loss": 0.8007,
       "step": 8250
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6412423792604906e-05,
-      "loss": 0.8202,
+      "loss": 0.9097,
       "step": 8251
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6411519929640922e-05,
-      "loss": 0.7301,
+      "loss": 0.8195,
       "step": 8252
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6410615977726193e-05,
-      "loss": 0.8634,
+      "loss": 0.9578,
       "step": 8253
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6409711936873266e-05,
-      "loss": 0.6884,
+      "loss": 0.8928,
       "step": 8254
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6408807807094677e-05,
-      "loss": 0.8825,
+      "loss": 0.919,
       "step": 8255
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6407903588402974e-05,
-      "loss": 0.8137,
+      "loss": 0.8963,
       "step": 8256
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6406999280810702e-05,
-      "loss": 0.8925,
+      "loss": 0.9943,
       "step": 8257
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6406094884330402e-05,
-      "loss": 0.8269,
+      "loss": 0.9431,
       "step": 8258
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6405190398974625e-05,
-      "loss": 0.9932,
+      "loss": 0.9995,
       "step": 8259
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6404285824755924e-05,
-      "loss": 0.9283,
+      "loss": 0.9815,
       "step": 8260
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.640338116168684e-05,
-      "loss": 0.8451,
+      "loss": 0.9183,
       "step": 8261
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6402476409779925e-05,
-      "loss": 0.8076,
+      "loss": 0.9135,
       "step": 8262
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6401571569047738e-05,
-      "loss": 0.898,
+      "loss": 0.9837,
       "step": 8263
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.640066663950283e-05,
-      "loss": 0.8325,
+      "loss": 0.923,
       "step": 8264
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6399761621157746e-05,
-      "loss": 0.9021,
+      "loss": 0.9322,
       "step": 8265
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6398856514025054e-05,
-      "loss": 0.7974,
+      "loss": 0.9355,
       "step": 8266
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.639795131811731e-05,
-      "loss": 0.7092,
+      "loss": 0.849,
       "step": 8267
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6397046033447063e-05,
-      "loss": 0.7192,
+      "loss": 0.8747,
       "step": 8268
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.639614066002688e-05,
-      "loss": 0.7817,
+      "loss": 0.8895,
       "step": 8269
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6395235197869323e-05,
-      "loss": 0.7981,
+      "loss": 0.9158,
       "step": 8270
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.639432964698695e-05,
-      "loss": 0.7412,
+      "loss": 0.8992,
       "step": 8271
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6393424007392324e-05,
-      "loss": 0.9017,
+      "loss": 0.9595,
       "step": 8272
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6392518279098012e-05,
-      "loss": 0.6289,
+      "loss": 0.8279,
       "step": 8273
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.639161246211658e-05,
-      "loss": 0.8509,
+      "loss": 0.9616,
       "step": 8274
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.639070655646059e-05,
-      "loss": 0.7502,
+      "loss": 0.845,
       "step": 8275
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6389800562142618e-05,
-      "loss": 0.7929,
+      "loss": 0.8974,
       "step": 8276
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6388894479175227e-05,
-      "loss": 0.7887,
+      "loss": 0.965,
       "step": 8277
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.638798830757099e-05,
-      "loss": 0.84,
+      "loss": 0.9365,
       "step": 8278
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.638708204734248e-05,
-      "loss": 0.8375,
+      "loss": 0.9462,
       "step": 8279
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6386175698502265e-05,
-      "loss": 0.8491,
+      "loss": 0.9803,
       "step": 8280
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6385269261062923e-05,
-      "loss": 0.6801,
+      "loss": 0.8837,
       "step": 8281
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6384362735037033e-05,
-      "loss": 0.6906,
+      "loss": 0.8782,
       "step": 8282
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6383456120437168e-05,
-      "loss": 0.5909,
+      "loss": 0.8274,
       "step": 8283
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6382549417275906e-05,
-      "loss": 0.784,
+      "loss": 0.9549,
       "step": 8284
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6381642625565826e-05,
-      "loss": 0.8137,
+      "loss": 0.8936,
       "step": 8285
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6380735745319506e-05,
-      "loss": 0.9518,
+      "loss": 1.1117,
       "step": 8286
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6379828776549532e-05,
-      "loss": 0.8682,
+      "loss": 0.9716,
       "step": 8287
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6378921719268487e-05,
-      "loss": 0.7188,
+      "loss": 0.9151,
       "step": 8288
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.637801457348896e-05,
-      "loss": 0.8767,
+      "loss": 0.9462,
       "step": 8289
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.637710733922352e-05,
-      "loss": 0.9028,
+      "loss": 0.9018,
       "step": 8290
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6376200016484766e-05,
-      "loss": 0.7501,
+      "loss": 0.8687,
       "step": 8291
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6375292605285286e-05,
-      "loss": 0.7662,
+      "loss": 0.8763,
       "step": 8292
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6374385105637666e-05,
-      "loss": 0.8528,
+      "loss": 0.9224,
       "step": 8293
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6373477517554494e-05,
-      "loss": 0.8029,
+      "loss": 0.896,
       "step": 8294
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6372569841048366e-05,
-      "loss": 0.8236,
+      "loss": 0.8624,
       "step": 8295
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6371662076131874e-05,
-      "loss": 0.8202,
+      "loss": 0.9367,
       "step": 8296
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6370754222817607e-05,
-      "loss": 0.8489,
+      "loss": 0.9264,
       "step": 8297
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.636984628111817e-05,
-      "loss": 0.8744,
+      "loss": 0.9326,
       "step": 8298
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6368938251046148e-05,
-      "loss": 0.8054,
+      "loss": 0.9388,
       "step": 8299
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6368030132614147e-05,
-      "loss": 0.8599,
+      "loss": 0.9974,
       "step": 8300
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6367121925834762e-05,
-      "loss": 0.7842,
+      "loss": 0.9441,
       "step": 8301
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6366213630720595e-05,
-      "loss": 0.8054,
+      "loss": 0.9089,
       "step": 8302
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6365305247284243e-05,
-      "loss": 0.601,
+      "loss": 0.7135,
       "step": 8303
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6364396775538316e-05,
-      "loss": 0.8562,
+      "loss": 0.9509,
       "step": 8304
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6363488215495415e-05,
-      "loss": 0.6952,
+      "loss": 0.8659,
       "step": 8305
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.636257956716814e-05,
-      "loss": 0.7099,
+      "loss": 0.7772,
       "step": 8306
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.63616708305691e-05,
-      "loss": 0.6984,
+      "loss": 0.8259,
       "step": 8307
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6360762005710907e-05,
-      "loss": 0.8162,
+      "loss": 0.9341,
       "step": 8308
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6359853092606163e-05,
-      "loss": 0.5907,
+      "loss": 0.7657,
       "step": 8309
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6358944091267482e-05,
-      "loss": 0.8108,
+      "loss": 0.8739,
       "step": 8310
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6358035001707473e-05,
-      "loss": 0.8303,
+      "loss": 0.9216,
       "step": 8311
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.635712582393875e-05,
-      "loss": 0.6968,
+      "loss": 0.9076,
       "step": 8312
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6356216557973926e-05,
-      "loss": 0.8392,
+      "loss": 0.9406,
       "step": 8313
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6355307203825616e-05,
-      "loss": 0.8035,
+      "loss": 0.9661,
       "step": 8314
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6354397761506434e-05,
-      "loss": 0.7845,
+      "loss": 0.9093,
       "step": 8315
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6353488231029e-05,
-      "loss": 0.8239,
+      "loss": 0.916,
       "step": 8316
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.635257861240593e-05,
-      "loss": 0.9473,
+      "loss": 0.9943,
       "step": 8317
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6351668905649844e-05,
-      "loss": 0.8406,
+      "loss": 0.9124,
       "step": 8318
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.635075911077337e-05,
-      "loss": 0.904,
+      "loss": 0.9904,
       "step": 8319
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6349849227789116e-05,
-      "loss": 0.8119,
+      "loss": 0.8442,
       "step": 8320
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6348939256709715e-05,
-      "loss": 0.666,
+      "loss": 0.8118,
       "step": 8321
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.634802919754779e-05,
-      "loss": 0.7826,
+      "loss": 0.8981,
       "step": 8322
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6347119050315973e-05,
-      "loss": 0.7334,
+      "loss": 0.8384,
       "step": 8323
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6346208815026877e-05,
-      "loss": 0.8414,
+      "loss": 0.9667,
       "step": 8324
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.634529849169314e-05,
-      "loss": 0.8364,
+      "loss": 0.8709,
       "step": 8325
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.634438808032739e-05,
-      "loss": 0.8484,
+      "loss": 0.9355,
       "step": 8326
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6343477580942257e-05,
-      "loss": 0.8062,
+      "loss": 0.9548,
       "step": 8327
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6342566993550374e-05,
-      "loss": 0.8456,
+      "loss": 0.9168,
       "step": 8328
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.634165631816437e-05,
-      "loss": 0.8108,
+      "loss": 0.8993,
       "step": 8329
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.634074555479688e-05,
-      "loss": 0.7702,
+      "loss": 0.9023,
       "step": 8330
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.633983470346055e-05,
-      "loss": 0.788,
+      "loss": 0.9398,
       "step": 8331
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6338923764168e-05,
-      "loss": 0.968,
+      "loss": 1.0114,
       "step": 8332
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.633801273693188e-05,
-      "loss": 0.7985,
+      "loss": 0.9144,
       "step": 8333
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6337101621764826e-05,
-      "loss": 0.7386,
+      "loss": 0.8334,
       "step": 8334
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.633619041867948e-05,
-      "loss": 0.8321,
+      "loss": 0.891,
       "step": 8335
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6335279127688477e-05,
-      "loss": 0.6006,
+      "loss": 0.7937,
       "step": 8336
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6334367748804468e-05,
-      "loss": 0.9524,
+      "loss": 1.027,
       "step": 8337
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6333456282040093e-05,
-      "loss": 0.8082,
+      "loss": 0.901,
       "step": 8338
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6332544727408e-05,
-      "loss": 0.7464,
+      "loss": 0.9103,
       "step": 8339
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6331633084920828e-05,
-      "loss": 0.7044,
+      "loss": 0.9109,
       "step": 8340
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6330721354591234e-05,
-      "loss": 0.6648,
+      "loss": 0.9091,
       "step": 8341
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6329809536431866e-05,
-      "loss": 0.7047,
+      "loss": 0.8123,
       "step": 8342
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6328897630455368e-05,
-      "loss": 0.9705,
+      "loss": 0.9559,
       "step": 8343
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6327985636674398e-05,
-      "loss": 0.9538,
+      "loss": 1.0387,
       "step": 8344
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.63270735551016e-05,
-      "loss": 0.8934,
+      "loss": 0.9591,
       "step": 8345
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6326161385749642e-05,
-      "loss": 0.7665,
+      "loss": 0.8878,
       "step": 8346
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6325249128631164e-05,
-      "loss": 0.737,
+      "loss": 0.9039,
       "step": 8347
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.632433678375883e-05,
-      "loss": 0.7772,
+      "loss": 0.9002,
       "step": 8348
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6323424351145293e-05,
-      "loss": 0.8282,
+      "loss": 0.8922,
       "step": 8349
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6322511830803222e-05,
-      "loss": 0.9045,
+      "loss": 0.9438,
       "step": 8350
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6321599222745264e-05,
-      "loss": 0.772,
+      "loss": 0.8632,
       "step": 8351
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.632068652698409e-05,
-      "loss": 0.8888,
+      "loss": 1.0158,
       "step": 8352
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6319773743532354e-05,
-      "loss": 0.7675,
+      "loss": 0.8948,
       "step": 8353
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6318860872402727e-05,
-      "loss": 0.7961,
+      "loss": 0.944,
       "step": 8354
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.631794791360787e-05,
-      "loss": 0.8469,
+      "loss": 0.9861,
       "step": 8355
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.631703486716045e-05,
-      "loss": 0.7514,
+      "loss": 0.9017,
       "step": 8356
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.631612173307313e-05,
-      "loss": 0.9351,
+      "loss": 0.9109,
       "step": 8357
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6315208511358586e-05,
-      "loss": 0.8732,
+      "loss": 1.0117,
       "step": 8358
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6314295202029488e-05,
-      "loss": 0.854,
+      "loss": 0.9424,
       "step": 8359
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6313381805098494e-05,
-      "loss": 0.7554,
+      "loss": 0.8888,
       "step": 8360
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.631246832057829e-05,
-      "loss": 0.6908,
+      "loss": 0.8687,
       "step": 8361
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6311554748481544e-05,
-      "loss": 0.8068,
+      "loss": 0.9182,
       "step": 8362
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.631064108882093e-05,
-      "loss": 0.7079,
+      "loss": 0.8334,
       "step": 8363
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6309727341609124e-05,
-      "loss": 0.7965,
+      "loss": 0.8386,
       "step": 8364
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.630881350685881e-05,
-      "loss": 0.8764,
+      "loss": 0.9179,
       "step": 8365
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6307899584582652e-05,
-      "loss": 0.7747,
+      "loss": 0.8939,
       "step": 8366
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.630698557479334e-05,
-      "loss": 0.8332,
+      "loss": 0.9257,
       "step": 8367
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6306071477503552e-05,
-      "loss": 0.7691,
+      "loss": 0.9016,
       "step": 8368
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6305157292725967e-05,
-      "loss": 0.7953,
+      "loss": 0.8977,
       "step": 8369
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.630424302047327e-05,
-      "loss": 0.8318,
+      "loss": 0.9153,
       "step": 8370
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6303328660758147e-05,
-      "loss": 0.7885,
+      "loss": 0.8513,
       "step": 8371
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6302414213593283e-05,
-      "loss": 0.9502,
+      "loss": 0.9568,
       "step": 8372
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6301499678991365e-05,
-      "loss": 0.8551,
+      "loss": 0.8911,
       "step": 8373
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6300585056965077e-05,
-      "loss": 0.7866,
+      "loss": 0.935,
       "step": 8374
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6299670347527112e-05,
-      "loss": 0.7103,
+      "loss": 0.8965,
       "step": 8375
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6298755550690157e-05,
-      "loss": 0.7764,
+      "loss": 0.9009,
       "step": 8376
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.629784066646691e-05,
-      "loss": 0.7371,
+      "loss": 0.9439,
       "step": 8377
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6296925694870053e-05,
-      "loss": 0.8482,
+      "loss": 0.9271,
       "step": 8378
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.629601063591229e-05,
-      "loss": 0.7965,
+      "loss": 0.8643,
       "step": 8379
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6295095489606313e-05,
-      "loss": 0.7209,
+      "loss": 0.8672,
       "step": 8380
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6294180255964814e-05,
-      "loss": 0.862,
+      "loss": 0.8844,
       "step": 8381
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6293264935000497e-05,
-      "loss": 0.8211,
+      "loss": 0.8702,
       "step": 8382
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6292349526726056e-05,
-      "loss": 0.9234,
+      "loss": 1.0125,
       "step": 8383
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6291434031154198e-05,
-      "loss": 0.8102,
+      "loss": 0.9078,
       "step": 8384
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6290518448297614e-05,
-      "loss": 0.7651,
+      "loss": 0.9295,
       "step": 8385
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.6289602778169012e-05,
-      "loss": 0.7167,
+      "loss": 0.8841,
       "step": 8386
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6288687020781096e-05,
-      "loss": 0.7966,
+      "loss": 0.7774,
       "step": 8387
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.628777117614657e-05,
-      "loss": 0.9103,
+      "loss": 1.0225,
       "step": 8388
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6286855244278144e-05,
-      "loss": 0.8083,
+      "loss": 0.8571,
       "step": 8389
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6285939225188516e-05,
-      "loss": 0.8957,
+      "loss": 0.9378,
       "step": 8390
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6285023118890405e-05,
-      "loss": 0.781,
+      "loss": 0.8475,
       "step": 8391
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.628410692539651e-05,
-      "loss": 0.8169,
+      "loss": 1.0078,
       "step": 8392
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.628319064471955e-05,
-      "loss": 0.7552,
+      "loss": 0.8429,
       "step": 8393
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6282274276872235e-05,
-      "loss": 0.8048,
+      "loss": 0.9121,
       "step": 8394
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.628135782186728e-05,
-      "loss": 0.7318,
+      "loss": 0.9352,
       "step": 8395
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.628044127971739e-05,
-      "loss": 0.8227,
+      "loss": 0.9409,
       "step": 8396
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6279524650435294e-05,
-      "loss": 0.9356,
+      "loss": 1.0323,
       "step": 8397
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.62786079340337e-05,
-      "loss": 0.6956,
+      "loss": 0.8671,
       "step": 8398
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.627769113052533e-05,
-      "loss": 0.7948,
+      "loss": 0.8758,
       "step": 8399
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6276774239922904e-05,
-      "loss": 0.9279,
+      "loss": 1.007,
       "step": 8400
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6275857262239137e-05,
-      "loss": 0.8959,
+      "loss": 0.9124,
       "step": 8401
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.627494019748676e-05,
-      "loss": 0.7367,
+      "loss": 0.8684,
       "step": 8402
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6274023045678488e-05,
-      "loss": 0.8028,
+      "loss": 0.9301,
       "step": 8403
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6273105806827053e-05,
-      "loss": 0.7664,
+      "loss": 0.9326,
       "step": 8404
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.627218848094517e-05,
-      "loss": 0.6797,
+      "loss": 0.8578,
       "step": 8405
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.627127106804557e-05,
-      "loss": 0.7873,
+      "loss": 0.867,
       "step": 8406
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.627035356814098e-05,
-      "loss": 0.8686,
+      "loss": 0.988,
       "step": 8407
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6269435981244138e-05,
-      "loss": 0.9274,
+      "loss": 0.9926,
       "step": 8408
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6268518307367763e-05,
-      "loss": 0.8793,
+      "loss": 0.9794,
       "step": 8409
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6267600546524587e-05,
-      "loss": 0.7594,
+      "loss": 0.8943,
       "step": 8410
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.626668269872735e-05,
-      "loss": 0.8121,
+      "loss": 0.8957,
       "step": 8411
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6265764763988784e-05,
-      "loss": 0.8046,
+      "loss": 0.8723,
       "step": 8412
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6264846742321616e-05,
-      "loss": 0.9855,
+      "loss": 0.9971,
       "step": 8413
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.626392863373859e-05,
-      "loss": 0.7631,
+      "loss": 0.8762,
       "step": 8414
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6263010438252442e-05,
-      "loss": 0.7837,
+      "loss": 0.8845,
       "step": 8415
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6262092155875912e-05,
-      "loss": 0.7339,
+      "loss": 0.8515,
       "step": 8416
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6261173786621736e-05,
-      "loss": 0.8723,
+      "loss": 0.9914,
       "step": 8417
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.626025533050266e-05,
-      "loss": 0.8231,
+      "loss": 0.9485,
       "step": 8418
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.625933678753142e-05,
-      "loss": 0.7893,
+      "loss": 0.8356,
       "step": 8419
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6258418157720766e-05,
-      "loss": 0.8635,
+      "loss": 0.9684,
       "step": 8420
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6257499441083434e-05,
-      "loss": 0.914,
+      "loss": 0.9797,
       "step": 8421
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.625658063763218e-05,
-      "loss": 0.8094,
+      "loss": 0.9261,
       "step": 8422
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.625566174737975e-05,
-      "loss": 0.7174,
+      "loss": 0.8955,
       "step": 8423
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6254742770338885e-05,
-      "loss": 0.7326,
+      "loss": 0.923,
       "step": 8424
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6253823706522334e-05,
-      "loss": 0.7437,
+      "loss": 0.9413,
       "step": 8425
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6252904555942857e-05,
-      "loss": 0.5946,
+      "loss": 0.7975,
       "step": 8426
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.62519853186132e-05,
-      "loss": 0.853,
+      "loss": 0.9644,
       "step": 8427
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.625106599454612e-05,
-      "loss": 0.8324,
+      "loss": 0.9116,
       "step": 8428
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6250146583754366e-05,
-      "loss": 0.7688,
+      "loss": 0.9114,
       "step": 8429
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6249227086250696e-05,
-      "loss": 0.7576,
+      "loss": 0.853,
       "step": 8430
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6248307502047867e-05,
-      "loss": 0.7314,
+      "loss": 0.8456,
       "step": 8431
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6247387831158636e-05,
-      "loss": 0.8215,
+      "loss": 0.9581,
       "step": 8432
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.624646807359576e-05,
-      "loss": 0.8519,
+      "loss": 0.9038,
       "step": 8433
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6245548229372006e-05,
-      "loss": 0.8075,
+      "loss": 0.9624,
       "step": 8434
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.624462829850013e-05,
-      "loss": 0.6402,
+      "loss": 0.783,
       "step": 8435
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6243708280992898e-05,
-      "loss": 0.8388,
+      "loss": 0.947,
       "step": 8436
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.624278817686307e-05,
-      "loss": 0.7493,
+      "loss": 0.8,
       "step": 8437
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6241867986123415e-05,
-      "loss": 0.8561,
+      "loss": 0.9459,
       "step": 8438
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.62409477087867e-05,
-      "loss": 0.8704,
+      "loss": 0.9203,
       "step": 8439
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6240027344865684e-05,
-      "loss": 0.8576,
+      "loss": 0.9602,
       "step": 8440
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6239106894373146e-05,
-      "loss": 0.8826,
+      "loss": 0.9641,
       "step": 8441
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6238186357321853e-05,
-      "loss": 0.9174,
+      "loss": 0.9865,
       "step": 8442
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6237265733724573e-05,
-      "loss": 0.7821,
+      "loss": 0.8587,
       "step": 8443
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.623634502359408e-05,
-      "loss": 0.8026,
+      "loss": 0.9016,
       "step": 8444
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6235424226943152e-05,
-      "loss": 0.8202,
+      "loss": 0.9447,
       "step": 8445
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6234503343784554e-05,
-      "loss": 0.6383,
+      "loss": 0.7503,
       "step": 8446
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6233582374131072e-05,
-      "loss": 0.8842,
+      "loss": 0.8979,
       "step": 8447
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.623266131799548e-05,
-      "loss": 0.8704,
+      "loss": 0.9095,
       "step": 8448
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.623174017539055e-05,
-      "loss": 0.7641,
+      "loss": 0.8492,
       "step": 8449
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.623081894632907e-05,
-      "loss": 0.6969,
+      "loss": 0.7751,
       "step": 8450
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6229897630823814e-05,
-      "loss": 0.6808,
+      "loss": 0.858,
       "step": 8451
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.622897622888757e-05,
-      "loss": 0.803,
+      "loss": 0.8537,
       "step": 8452
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.622805474053312e-05,
-      "loss": 0.648,
+      "loss": 0.7807,
       "step": 8453
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6227133165773244e-05,
-      "loss": 0.7385,
+      "loss": 0.8722,
       "step": 8454
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6226211504620736e-05,
-      "loss": 0.8923,
+      "loss": 0.9974,
       "step": 8455
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.622528975708837e-05,
-      "loss": 0.7034,
+      "loss": 0.8638,
       "step": 8456
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6224367923188947e-05,
-      "loss": 0.7245,
+      "loss": 0.9344,
       "step": 8457
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6223446002935247e-05,
-      "loss": 0.8369,
+      "loss": 0.9278,
       "step": 8458
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6222523996340064e-05,
-      "loss": 0.7726,
+      "loss": 0.9551,
       "step": 8459
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.622160190341619e-05,
-      "loss": 0.7209,
+      "loss": 0.9354,
       "step": 8460
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6220679724176417e-05,
-      "loss": 0.8764,
+      "loss": 0.8831,
       "step": 8461
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6219757458633537e-05,
-      "loss": 0.8743,
+      "loss": 0.9579,
       "step": 8462
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6218835106800352e-05,
-      "loss": 0.8716,
+      "loss": 0.9655,
       "step": 8463
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6217912668689647e-05,
-      "loss": 0.8798,
+      "loss": 0.92,
       "step": 8464
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6216990144314228e-05,
-      "loss": 0.7339,
+      "loss": 0.7948,
       "step": 8465
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6216067533686893e-05,
-      "loss": 0.9574,
+      "loss": 0.9855,
       "step": 8466
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.621514483682044e-05,
-      "loss": 0.7372,
+      "loss": 0.848,
       "step": 8467
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6214222053727667e-05,
-      "loss": 0.8348,
+      "loss": 0.9435,
       "step": 8468
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6213299184421383e-05,
-      "loss": 0.6498,
+      "loss": 0.8544,
       "step": 8469
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6212376228914393e-05,
-      "loss": 0.7775,
+      "loss": 0.9697,
       "step": 8470
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.621145318721949e-05,
-      "loss": 0.8193,
+      "loss": 0.9429,
       "step": 8471
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.621053005934949e-05,
-      "loss": 0.8141,
+      "loss": 0.9237,
       "step": 8472
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6209606845317195e-05,
-      "loss": 0.7824,
+      "loss": 0.8674,
       "step": 8473
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6208683545135415e-05,
-      "loss": 0.791,
+      "loss": 0.9272,
       "step": 8474
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6207760158816965e-05,
-      "loss": 0.737,
+      "loss": 0.9474,
       "step": 8475
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6206836686374643e-05,
-      "loss": 0.7495,
+      "loss": 0.8969,
       "step": 8476
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6205913127821272e-05,
-      "loss": 0.747,
+      "loss": 0.9055,
       "step": 8477
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.620498948316966e-05,
-      "loss": 0.8036,
+      "loss": 0.9397,
       "step": 8478
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6204065752432625e-05,
-      "loss": 0.8137,
+      "loss": 0.8885,
       "step": 8479
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6203141935622978e-05,
-      "loss": 0.8538,
+      "loss": 0.966,
       "step": 8480
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.620221803275354e-05,
-      "loss": 0.7532,
+      "loss": 0.9147,
       "step": 8481
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6201294043837127e-05,
-      "loss": 0.7289,
+      "loss": 0.8203,
       "step": 8482
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6200369968886555e-05,
-      "loss": 0.5879,
+      "loss": 0.7997,
       "step": 8483
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6199445807914647e-05,
-      "loss": 0.9213,
+      "loss": 1.0043,
       "step": 8484
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6198521560934225e-05,
-      "loss": 0.779,
+      "loss": 0.9078,
       "step": 8485
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6197597227958112e-05,
-      "loss": 0.8363,
+      "loss": 0.9779,
       "step": 8486
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.619667280899913e-05,
-      "loss": 0.7658,
+      "loss": 0.9054,
       "step": 8487
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6195748304070106e-05,
-      "loss": 0.6728,
+      "loss": 0.8437,
       "step": 8488
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.619482371318386e-05,
-      "loss": 0.7489,
+      "loss": 0.9129,
       "step": 8489
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.619389903635323e-05,
-      "loss": 0.6176,
+      "loss": 0.7404,
       "step": 8490
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6192974273591037e-05,
-      "loss": 0.7934,
+      "loss": 0.8582,
       "step": 8491
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.619204942491011e-05,
-      "loss": 0.6114,
+      "loss": 0.8129,
       "step": 8492
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6191124490323284e-05,
-      "loss": 0.784,
+      "loss": 0.9268,
       "step": 8493
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6190199469843392e-05,
-      "loss": 0.753,
+      "loss": 0.8912,
       "step": 8494
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6189274363483264e-05,
-      "loss": 0.8453,
+      "loss": 0.9261,
       "step": 8495
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6188349171255736e-05,
-      "loss": 0.8338,
+      "loss": 0.9338,
       "step": 8496
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618742389317364e-05,
-      "loss": 0.8113,
+      "loss": 0.8708,
       "step": 8497
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618649852924982e-05,
-      "loss": 0.7915,
+      "loss": 0.8694,
       "step": 8498
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618557307949711e-05,
-      "loss": 0.7564,
+      "loss": 0.8447,
       "step": 8499
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618464754392835e-05,
-      "loss": 0.8796,
+      "loss": 0.9682,
       "step": 8500
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6183721922556376e-05,
-      "loss": 0.7624,
+      "loss": 0.859,
       "step": 8501
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618279621539404e-05,
-      "loss": 0.8339,
+      "loss": 0.9042,
       "step": 8502
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6181870422454172e-05,
-      "loss": 0.7452,
+      "loss": 0.8242,
       "step": 8503
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.618094454374963e-05,
-      "loss": 0.7486,
+      "loss": 0.8497,
       "step": 8504
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6180018579293246e-05,
-      "loss": 0.7084,
+      "loss": 0.8378,
       "step": 8505
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6179092529097877e-05,
-      "loss": 0.7715,
+      "loss": 0.8537,
       "step": 8506
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6178166393176364e-05,
-      "loss": 0.7508,
+      "loss": 0.8791,
       "step": 8507
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.617724017154156e-05,
-      "loss": 0.7611,
+      "loss": 0.876,
       "step": 8508
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.617631386420631e-05,
-      "loss": 0.871,
+      "loss": 1.0742,
       "step": 8509
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6175387471183472e-05,
-      "loss": 0.7841,
+      "loss": 0.8432,
       "step": 8510
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.617446099248589e-05,
-      "loss": 0.8952,
+      "loss": 0.9577,
       "step": 8511
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6173534428126423e-05,
-      "loss": 0.8885,
+      "loss": 0.9894,
       "step": 8512
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6172607778117927e-05,
-      "loss": 0.7409,
+      "loss": 0.8279,
       "step": 8513
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6171681042473255e-05,
-      "loss": 0.762,
+      "loss": 0.8495,
       "step": 8514
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6170754221205265e-05,
-      "loss": 0.7248,
+      "loss": 0.8889,
       "step": 8515
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6169827314326814e-05,
-      "loss": 0.9473,
+      "loss": 0.9739,
       "step": 8516
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6168900321850765e-05,
-      "loss": 0.8318,
+      "loss": 0.953,
       "step": 8517
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6167973243789975e-05,
-      "loss": 0.7697,
+      "loss": 0.8987,
       "step": 8518
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6167046080157308e-05,
-      "loss": 0.7081,
+      "loss": 0.8268,
       "step": 8519
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.616611883096563e-05,
-      "loss": 0.7596,
+      "loss": 0.9097,
       "step": 8520
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6165191496227795e-05,
-      "loss": 0.7429,
+      "loss": 0.8507,
       "step": 8521
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6164264075956682e-05,
-      "loss": 0.8709,
+      "loss": 0.9562,
       "step": 8522
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6163336570165148e-05,
-      "loss": 0.8705,
+      "loss": 0.9256,
       "step": 8523
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.616240897886606e-05,
-      "loss": 0.7312,
+      "loss": 0.9278,
       "step": 8524
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6161481302072294e-05,
-      "loss": 0.9048,
+      "loss": 0.9612,
       "step": 8525
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6160553539796717e-05,
-      "loss": 0.7953,
+      "loss": 0.8609,
       "step": 8526
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.61596256920522e-05,
-      "loss": 0.8583,
+      "loss": 0.9061,
       "step": 8527
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6158697758851617e-05,
-      "loss": 0.7829,
+      "loss": 0.8307,
       "step": 8528
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.615776974020784e-05,
-      "loss": 0.685,
+      "loss": 0.8347,
       "step": 8529
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6156841636133744e-05,
-      "loss": 0.7234,
+      "loss": 0.8767,
       "step": 8530
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6155913446642206e-05,
-      "loss": 0.7952,
+      "loss": 0.9134,
       "step": 8531
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6154985171746104e-05,
-      "loss": 0.9485,
+      "loss": 1.0084,
       "step": 8532
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6154056811458314e-05,
-      "loss": 0.836,
+      "loss": 0.9641,
       "step": 8533
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6153128365791716e-05,
-      "loss": 0.7229,
+      "loss": 0.8411,
       "step": 8534
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6152199834759197e-05,
-      "loss": 0.7746,
+      "loss": 0.8457,
       "step": 8535
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6151271218373628e-05,
-      "loss": 0.9023,
+      "loss": 1.0076,
       "step": 8536
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6150342516647904e-05,
-      "loss": 0.8885,
+      "loss": 0.9315,
       "step": 8537
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.61494137295949e-05,
-      "loss": 0.8001,
+      "loss": 0.8684,
       "step": 8538
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614848485722751e-05,
-      "loss": 0.8486,
+      "loss": 0.8722,
       "step": 8539
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614755589955861e-05,
-      "loss": 0.8673,
+      "loss": 0.9218,
       "step": 8540
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.61466268566011e-05,
-      "loss": 0.6367,
+      "loss": 0.7966,
       "step": 8541
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614569772836786e-05,
-      "loss": 0.815,
+      "loss": 0.9398,
       "step": 8542
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614476851487178e-05,
-      "loss": 0.7183,
+      "loss": 0.8115,
       "step": 8543
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6143839216125762e-05,
-      "loss": 0.7656,
+      "loss": 0.8233,
       "step": 8544
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614290983214269e-05,
-      "loss": 0.6952,
+      "loss": 0.8331,
       "step": 8545
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6141980362935455e-05,
-      "loss": 0.6634,
+      "loss": 0.8262,
       "step": 8546
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.614105080851696e-05,
-      "loss": 0.9107,
+      "loss": 0.9682,
       "step": 8547
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.61401211689001e-05,
-      "loss": 0.8188,
+      "loss": 0.9032,
       "step": 8548
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6139191444097767e-05,
-      "loss": 0.7349,
+      "loss": 0.8459,
       "step": 8549
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6138261634122865e-05,
-      "loss": 0.7442,
+      "loss": 0.8526,
       "step": 8550
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.613733173898829e-05,
-      "loss": 0.7652,
+      "loss": 0.9084,
       "step": 8551
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6136401758706946e-05,
-      "loss": 0.8753,
+      "loss": 0.9024,
       "step": 8552
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6135471693291736e-05,
-      "loss": 0.8302,
+      "loss": 0.8914,
       "step": 8553
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6134541542755563e-05,
-      "loss": 0.7148,
+      "loss": 0.8798,
       "step": 8554
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6133611307111325e-05,
-      "loss": 0.8186,
+      "loss": 0.893,
       "step": 8555
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6132680986371935e-05,
-      "loss": 0.8434,
+      "loss": 0.9037,
       "step": 8556
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6131750580550297e-05,
-      "loss": 0.9676,
+      "loss": 0.9855,
       "step": 8557
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.613082008965932e-05,
-      "loss": 0.7362,
+      "loss": 0.8945,
       "step": 8558
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6129889513711912e-05,
-      "loss": 0.8343,
+      "loss": 0.9799,
       "step": 8559
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6128958852720983e-05,
-      "loss": 0.771,
+      "loss": 0.9244,
       "step": 8560
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6128028106699448e-05,
-      "loss": 0.9868,
+      "loss": 0.989,
       "step": 8561
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.612709727566022e-05,
-      "loss": 0.7905,
+      "loss": 1.0012,
       "step": 8562
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6126166359616205e-05,
-      "loss": 0.8342,
+      "loss": 0.9694,
       "step": 8563
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6125235358580328e-05,
-      "loss": 0.8223,
+      "loss": 0.9228,
       "step": 8564
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6124304272565497e-05,
-      "loss": 0.6931,
+      "loss": 0.8419,
       "step": 8565
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6123373101584636e-05,
-      "loss": 0.7357,
+      "loss": 0.9294,
       "step": 8566
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.612244184565066e-05,
-      "loss": 0.8173,
+      "loss": 0.9016,
       "step": 8567
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.612151050477649e-05,
-      "loss": 0.7756,
+      "loss": 0.9954,
       "step": 8568
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6120579078975047e-05,
-      "loss": 0.7684,
+      "loss": 0.9184,
       "step": 8569
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6119647568259255e-05,
-      "loss": 0.7723,
+      "loss": 0.9072,
       "step": 8570
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6118715972642034e-05,
-      "loss": 0.763,
+      "loss": 0.8623,
       "step": 8571
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.611778429213631e-05,
-      "loss": 0.6973,
+      "loss": 0.9269,
       "step": 8572
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6116852526755008e-05,
-      "loss": 0.7724,
+      "loss": 0.9346,
       "step": 8573
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6115920676511057e-05,
-      "loss": 0.7757,
+      "loss": 0.8938,
       "step": 8574
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6114988741417386e-05,
-      "loss": 0.7113,
+      "loss": 0.8085,
       "step": 8575
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.611405672148692e-05,
-      "loss": 0.7044,
+      "loss": 0.8769,
       "step": 8576
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6113124616732592e-05,
-      "loss": 0.8372,
+      "loss": 0.9512,
       "step": 8577
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6112192427167338e-05,
-      "loss": 0.7725,
+      "loss": 0.8905,
       "step": 8578
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.611126015280408e-05,
-      "loss": 0.8916,
+      "loss": 0.9537,
       "step": 8579
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.611032779365576e-05,
-      "loss": 0.7879,
+      "loss": 0.911,
       "step": 8580
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610939534973531e-05,
-      "loss": 0.7101,
+      "loss": 0.8802,
       "step": 8581
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610846282105567e-05,
-      "loss": 0.8161,
+      "loss": 0.9689,
       "step": 8582
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610753020762977e-05,
-      "loss": 0.9039,
+      "loss": 0.9488,
       "step": 8583
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610659750947056e-05,
-      "loss": 0.9967,
+      "loss": 1.0618,
       "step": 8584
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610566472659097e-05,
-      "loss": 0.8789,
+      "loss": 0.9896,
       "step": 8585
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610473185900395e-05,
-      "loss": 0.8431,
+      "loss": 0.9754,
       "step": 8586
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6103798906722432e-05,
-      "loss": 0.7654,
+      "loss": 0.9077,
       "step": 8587
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6102865869759362e-05,
-      "loss": 0.7143,
+      "loss": 0.8085,
       "step": 8588
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610193274812769e-05,
-      "loss": 0.8058,
+      "loss": 0.9279,
       "step": 8589
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.610099954184036e-05,
-      "loss": 0.7865,
+      "loss": 0.9581,
       "step": 8590
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6100066250910318e-05,
-      "loss": 0.7897,
+      "loss": 0.8763,
       "step": 8591
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.609913287535051e-05,
-      "loss": 0.8236,
+      "loss": 0.8791,
       "step": 8592
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6098199415173885e-05,
-      "loss": 0.7012,
+      "loss": 0.8801,
       "step": 8593
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6097265870393398e-05,
-      "loss": 0.8746,
+      "loss": 0.9283,
       "step": 8594
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6096332241022e-05,
-      "loss": 0.7329,
+      "loss": 0.8338,
       "step": 8595
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6095398527072638e-05,
-      "loss": 0.8118,
+      "loss": 0.9265,
       "step": 8596
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.609446472855827e-05,
-      "loss": 0.8628,
+      "loss": 0.9894,
       "step": 8597
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6093530845491852e-05,
-      "loss": 0.6972,
+      "loss": 0.845,
       "step": 8598
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.609259687788634e-05,
-      "loss": 0.7906,
+      "loss": 0.9029,
       "step": 8599
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6091662825754687e-05,
-      "loss": 0.7015,
+      "loss": 0.9002,
       "step": 8600
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.609072868910986e-05,
-      "loss": 0.8346,
+      "loss": 1.0128,
       "step": 8601
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.608979446796481e-05,
-      "loss": 0.7509,
+      "loss": 0.8377,
       "step": 8602
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6088860162332504e-05,
-      "loss": 0.7275,
+      "loss": 0.8943,
       "step": 8603
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.60879257722259e-05,
-      "loss": 0.7188,
+      "loss": 0.7795,
       "step": 8604
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6086991297657965e-05,
-      "loss": 0.7811,
+      "loss": 0.9208,
       "step": 8605
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6086056738641663e-05,
-      "loss": 0.9063,
+      "loss": 0.9922,
       "step": 8606
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.608512209518996e-05,
-      "loss": 1.0734,
+      "loss": 1.0386,
       "step": 8607
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.608418736731582e-05,
-      "loss": 0.7011,
+      "loss": 0.8603,
       "step": 8608
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.608325255503221e-05,
-      "loss": 0.7559,
+      "loss": 0.9013,
       "step": 8609
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6082317658352106e-05,
-      "loss": 0.6079,
+      "loss": 0.8157,
       "step": 8610
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6081382677288472e-05,
-      "loss": 0.7665,
+      "loss": 0.8973,
       "step": 8611
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.608044761185428e-05,
-      "loss": 0.7644,
+      "loss": 0.8836,
       "step": 8612
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6079512462062504e-05,
-      "loss": 0.7661,
+      "loss": 0.9009,
       "step": 8613
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6078577227926123e-05,
-      "loss": 0.7349,
+      "loss": 0.8958,
       "step": 8614
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6077641909458105e-05,
-      "loss": 0.6835,
+      "loss": 0.891,
       "step": 8615
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.607670650667143e-05,
-      "loss": 0.6922,
+      "loss": 0.8219,
       "step": 8616
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6075771019579072e-05,
-      "loss": 0.9025,
+      "loss": 0.9589,
       "step": 8617
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.607483544819401e-05,
-      "loss": 0.7816,
+      "loss": 0.9097,
       "step": 8618
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.607389979252923e-05,
-      "loss": 0.8458,
+      "loss": 0.9351,
       "step": 8619
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6072964052597705e-05,
-      "loss": 0.9099,
+      "loss": 0.9812,
       "step": 8620
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.607202822841242e-05,
-      "loss": 0.8636,
+      "loss": 0.9671,
       "step": 8621
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6071092319986358e-05,
-      "loss": 0.8092,
+      "loss": 0.9647,
       "step": 8622
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6070156327332505e-05,
-      "loss": 0.9063,
+      "loss": 0.9365,
       "step": 8623
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6069220250463848e-05,
-      "loss": 0.8226,
+      "loss": 0.9099,
       "step": 8624
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6068284089393363e-05,
-      "loss": 0.6779,
+      "loss": 0.8563,
       "step": 8625
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6067347844134055e-05,
-      "loss": 0.7551,
+      "loss": 0.8977,
       "step": 8626
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.60664115146989e-05,
-      "loss": 0.7093,
+      "loss": 0.8467,
       "step": 8627
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.606547510110089e-05,
-      "loss": 0.8342,
+      "loss": 0.9186,
       "step": 8628
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6064538603353023e-05,
-      "loss": 0.8657,
+      "loss": 0.9155,
       "step": 8629
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6063602021468286e-05,
-      "loss": 0.7896,
+      "loss": 0.9053,
       "step": 8630
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6062665355459672e-05,
-      "loss": 0.8535,
+      "loss": 0.8742,
       "step": 8631
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.606172860534018e-05,
-      "loss": 0.7358,
+      "loss": 0.8507,
       "step": 8632
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6060791771122802e-05,
-      "loss": 0.9122,
+      "loss": 0.978,
       "step": 8633
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6059854852820543e-05,
-      "loss": 0.6931,
+      "loss": 0.795,
       "step": 8634
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.605891785044639e-05,
-      "loss": 0.58,
+      "loss": 0.7692,
       "step": 8635
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6057980764013354e-05,
-      "loss": 0.7828,
+      "loss": 0.8712,
       "step": 8636
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6057043593534424e-05,
-      "loss": 0.81,
+      "loss": 0.9172,
       "step": 8637
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.605610633902261e-05,
-      "loss": 0.7358,
+      "loss": 0.8689,
       "step": 8638
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6055169000490917e-05,
-      "loss": 0.836,
+      "loss": 0.8836,
       "step": 8639
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.605423157795234e-05,
-      "loss": 0.7579,
+      "loss": 0.8946,
       "step": 8640
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6053294071419888e-05,
-      "loss": 0.8373,
+      "loss": 0.8476,
       "step": 8641
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6052356480906573e-05,
-      "loss": 0.9084,
+      "loss": 0.9531,
       "step": 8642
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.60514188064254e-05,
-      "loss": 0.7559,
+      "loss": 0.8958,
       "step": 8643
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6050481047989376e-05,
-      "loss": 0.763,
+      "loss": 0.8441,
       "step": 8644
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6049543205611512e-05,
-      "loss": 0.7192,
+      "loss": 0.8026,
       "step": 8645
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6048605279304816e-05,
-      "loss": 0.7346,
+      "loss": 0.8638,
       "step": 8646
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6047667269082308e-05,
-      "loss": 0.7378,
+      "loss": 0.8703,
       "step": 8647
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6046729174956996e-05,
-      "loss": 0.8587,
+      "loss": 0.9333,
       "step": 8648
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6045790996941893e-05,
-      "loss": 0.9462,
+      "loss": 1.0568,
       "step": 8649
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.604485273505002e-05,
-      "loss": 0.7814,
+      "loss": 0.8528,
       "step": 8650
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6043914389294393e-05,
-      "loss": 0.6755,
+      "loss": 0.7761,
       "step": 8651
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6042975959688028e-05,
-      "loss": 0.6633,
+      "loss": 0.8362,
       "step": 8652
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6042037446243944e-05,
-      "loss": 0.8175,
+      "loss": 0.9581,
       "step": 8653
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6041098848975166e-05,
-      "loss": 0.78,
+      "loss": 0.863,
       "step": 8654
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.604016016789471e-05,
-      "loss": 0.7861,
+      "loss": 0.821,
       "step": 8655
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.60392214030156e-05,
-      "loss": 0.7809,
+      "loss": 0.9271,
       "step": 8656
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6038282554350866e-05,
-      "loss": 0.7087,
+      "loss": 0.8552,
       "step": 8657
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.603734362191353e-05,
-      "loss": 0.843,
+      "loss": 0.9259,
       "step": 8658
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.603640460571661e-05,
-      "loss": 0.7482,
+      "loss": 0.8719,
       "step": 8659
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.6035465505773148e-05,
-      "loss": 0.8878,
+      "loss": 0.9457,
       "step": 8660
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.603452632209616e-05,
-      "loss": 0.8123,
+      "loss": 0.9302,
       "step": 8661
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6033587054698684e-05,
-      "loss": 0.7499,
+      "loss": 0.8202,
       "step": 8662
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6032647703593743e-05,
-      "loss": 0.721,
+      "loss": 0.8547,
       "step": 8663
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.603170826879438e-05,
-      "loss": 0.8553,
+      "loss": 0.9342,
       "step": 8664
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6030768750313624e-05,
-      "loss": 0.8105,
+      "loss": 0.8763,
       "step": 8665
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6029829148164504e-05,
-      "loss": 0.8331,
+      "loss": 0.9836,
       "step": 8666
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602888946236006e-05,
-      "loss": 0.8064,
+      "loss": 0.9053,
       "step": 8667
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602794969291333e-05,
-      "loss": 0.8228,
+      "loss": 0.8454,
       "step": 8668
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602700983983735e-05,
-      "loss": 0.6574,
+      "loss": 0.8254,
       "step": 8669
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602606990314516e-05,
-      "loss": 0.8835,
+      "loss": 0.9419,
       "step": 8670
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.60251298828498e-05,
-      "loss": 0.8513,
+      "loss": 0.901,
       "step": 8671
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602418977896431e-05,
-      "loss": 0.7727,
+      "loss": 0.8228,
       "step": 8672
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602324959150174e-05,
-      "loss": 0.9355,
+      "loss": 0.9908,
       "step": 8673
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602230932047512e-05,
-      "loss": 0.8081,
+      "loss": 0.8568,
       "step": 8674
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6021368965897506e-05,
-      "loss": 0.6511,
+      "loss": 0.8615,
       "step": 8675
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.602042852778194e-05,
-      "loss": 0.7394,
+      "loss": 0.8693,
       "step": 8676
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6019488006141472e-05,
-      "loss": 0.737,
+      "loss": 0.8524,
       "step": 8677
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6018547400989146e-05,
-      "loss": 0.7901,
+      "loss": 0.9044,
       "step": 8678
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6017606712338016e-05,
-      "loss": 0.7356,
+      "loss": 0.911,
       "step": 8679
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.601666594020113e-05,
-      "loss": 0.8868,
+      "loss": 0.9546,
       "step": 8680
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.601572508459154e-05,
-      "loss": 0.9187,
+      "loss": 0.9569,
       "step": 8681
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.60147841455223e-05,
-      "loss": 0.7953,
+      "loss": 0.8728,
       "step": 8682
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6013843123006463e-05,
-      "loss": 0.7157,
+      "loss": 0.8515,
       "step": 8683
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6012902017057085e-05,
-      "loss": 0.7051,
+      "loss": 0.8002,
       "step": 8684
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6011960827687225e-05,
-      "loss": 0.8675,
+      "loss": 0.801,
       "step": 8685
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.601101955490994e-05,
-      "loss": 0.6137,
+      "loss": 0.7375,
       "step": 8686
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.601007819873828e-05,
-      "loss": 0.7765,
+      "loss": 0.8562,
       "step": 8687
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6009136759185318e-05,
-      "loss": 0.7245,
+      "loss": 0.8427,
       "step": 8688
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.600819523626411e-05,
-      "loss": 0.7388,
+      "loss": 0.843,
       "step": 8689
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6007253629987714e-05,
-      "loss": 0.7525,
+      "loss": 0.9534,
       "step": 8690
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6006311940369198e-05,
-      "loss": 0.7052,
+      "loss": 0.8619,
       "step": 8691
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6005370167421628e-05,
-      "loss": 0.7592,
+      "loss": 0.8721,
       "step": 8692
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6004428311158063e-05,
-      "loss": 0.838,
+      "loss": 0.9308,
       "step": 8693
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.600348637159158e-05,
-      "loss": 0.9479,
+      "loss": 0.9405,
       "step": 8694
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6002544348735236e-05,
-      "loss": 0.761,
+      "loss": 0.8894,
       "step": 8695
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6001602242602106e-05,
-      "loss": 0.733,
+      "loss": 0.8008,
       "step": 8696
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.6000660053205265e-05,
-      "loss": 0.6866,
+      "loss": 0.8573,
       "step": 8697
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5999717780557776e-05,
-      "loss": 0.7502,
+      "loss": 0.8972,
       "step": 8698
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5998775424672716e-05,
-      "loss": 0.6859,
+      "loss": 0.7836,
       "step": 8699
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.599783298556316e-05,
-      "loss": 0.6381,
+      "loss": 0.7469,
       "step": 8700
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.599689046324218e-05,
-      "loss": 0.7801,
+      "loss": 0.8787,
       "step": 8701
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.599594785772285e-05,
-      "loss": 0.6113,
+      "loss": 0.7592,
       "step": 8702
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5995005169018254e-05,
-      "loss": 0.7086,
+      "loss": 0.849,
       "step": 8703
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5994062397141466e-05,
-      "loss": 0.8723,
+      "loss": 0.9224,
       "step": 8704
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5993119542105565e-05,
-      "loss": 0.618,
+      "loss": 0.781,
       "step": 8705
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.599217660392364e-05,
-      "loss": 0.9219,
+      "loss": 1.0031,
       "step": 8706
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5991233582608757e-05,
-      "loss": 0.8648,
+      "loss": 0.8993,
       "step": 8707
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5990290478174017e-05,
-      "loss": 0.8248,
+      "loss": 0.9446,
       "step": 8708
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.598934729063249e-05,
-      "loss": 0.6793,
+      "loss": 0.8566,
       "step": 8709
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.598840401999727e-05,
-      "loss": 0.6368,
+      "loss": 0.8192,
       "step": 8710
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5987460666281442e-05,
-      "loss": 0.7142,
+      "loss": 0.8603,
       "step": 8711
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5986517229498088e-05,
-      "loss": 0.9701,
+      "loss": 1.0583,
       "step": 8712
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5985573709660305e-05,
-      "loss": 0.7746,
+      "loss": 0.9245,
       "step": 8713
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.598463010678118e-05,
-      "loss": 0.6973,
+      "loss": 0.8391,
       "step": 8714
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5983686420873798e-05,
-      "loss": 0.8045,
+      "loss": 0.9415,
       "step": 8715
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5982742651951258e-05,
-      "loss": 0.8133,
+      "loss": 0.9182,
       "step": 8716
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5981798800026655e-05,
-      "loss": 0.7341,
+      "loss": 0.8789,
       "step": 8717
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.598085486511308e-05,
-      "loss": 0.7643,
+      "loss": 0.8793,
       "step": 8718
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.597991084722363e-05,
-      "loss": 0.8336,
+      "loss": 0.9155,
       "step": 8719
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5978966746371397e-05,
-      "loss": 0.7035,
+      "loss": 0.8542,
       "step": 8720
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5978022562569487e-05,
-      "loss": 0.7733,
+      "loss": 0.8328,
       "step": 8721
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5977078295830997e-05,
-      "loss": 0.69,
+      "loss": 0.826,
       "step": 8722
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5976133946169024e-05,
-      "loss": 0.8334,
+      "loss": 0.9741,
       "step": 8723
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.597518951359667e-05,
-      "loss": 0.7317,
+      "loss": 0.8687,
       "step": 8724
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5974244998127044e-05,
-      "loss": 0.7796,
+      "loss": 0.8465,
       "step": 8725
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.597330039977324e-05,
-      "loss": 0.7202,
+      "loss": 0.7686,
       "step": 8726
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.597235571854837e-05,
-      "loss": 0.9736,
+      "loss": 1.0341,
       "step": 8727
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.597141095446554e-05,
-      "loss": 0.9187,
+      "loss": 0.9068,
       "step": 8728
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5970466107537852e-05,
-      "loss": 0.7662,
+      "loss": 0.9149,
       "step": 8729
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5969521177778423e-05,
-      "loss": 1.0125,
+      "loss": 0.9602,
       "step": 8730
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5968576165200352e-05,
-      "loss": 0.749,
+      "loss": 0.9278,
       "step": 8731
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5967631069816753e-05,
-      "loss": 0.8405,
+      "loss": 0.9324,
       "step": 8732
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5966685891640748e-05,
-      "loss": 0.8553,
+      "loss": 0.908,
       "step": 8733
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5965740630685433e-05,
-      "loss": 0.7667,
+      "loss": 0.8854,
       "step": 8734
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5964795286963937e-05,
-      "loss": 0.8766,
+      "loss": 0.9217,
       "step": 8735
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.596384986048937e-05,
-      "loss": 0.8928,
+      "loss": 0.9587,
       "step": 8736
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5962904351274843e-05,
-      "loss": 0.8838,
+      "loss": 0.9767,
       "step": 8737
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.596195875933348e-05,
-      "loss": 0.8596,
+      "loss": 0.9449,
       "step": 8738
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.59610130846784e-05,
-      "loss": 0.7805,
+      "loss": 0.9002,
       "step": 8739
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.596006732732272e-05,
-      "loss": 0.849,
+      "loss": 0.8899,
       "step": 8740
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.595912148727956e-05,
-      "loss": 0.8324,
+      "loss": 0.9215,
       "step": 8741
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5958175564562046e-05,
-      "loss": 0.8521,
+      "loss": 0.8651,
       "step": 8742
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.59572295591833e-05,
-      "loss": 0.8638,
+      "loss": 0.9076,
       "step": 8743
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5956283471156448e-05,
-      "loss": 0.7997,
+      "loss": 0.9771,
       "step": 8744
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5955337300494612e-05,
-      "loss": 0.8255,
+      "loss": 0.9708,
       "step": 8745
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.595439104721092e-05,
-      "loss": 0.8744,
+      "loss": 0.9104,
       "step": 8746
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.59534447113185e-05,
-      "loss": 0.6539,
+      "loss": 0.8644,
       "step": 8747
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5952498292830482e-05,
-      "loss": 0.7085,
+      "loss": 0.7432,
       "step": 8748
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5951551791759996e-05,
-      "loss": 0.8741,
+      "loss": 0.9305,
       "step": 8749
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5950605208120172e-05,
-      "loss": 0.6938,
+      "loss": 0.8421,
       "step": 8750
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5949658541924144e-05,
-      "loss": 0.7542,
+      "loss": 0.9134,
       "step": 8751
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5948711793185046e-05,
-      "loss": 0.7856,
+      "loss": 0.8465,
       "step": 8752
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.594776496191601e-05,
-      "loss": 0.793,
+      "loss": 0.8854,
       "step": 8753
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5946818048130178e-05,
-      "loss": 0.8401,
+      "loss": 0.9625,
       "step": 8754
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5945871051840676e-05,
-      "loss": 0.8412,
+      "loss": 0.9459,
       "step": 8755
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5944923973060655e-05,
-      "loss": 0.8128,
+      "loss": 0.8974,
       "step": 8756
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5943976811803247e-05,
-      "loss": 0.7736,
+      "loss": 0.9026,
       "step": 8757
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5943029568081595e-05,
-      "loss": 0.8034,
+      "loss": 0.8653,
       "step": 8758
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5942082241908837e-05,
-      "loss": 0.739,
+      "loss": 0.8024,
       "step": 8759
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5941134833298124e-05,
-      "loss": 0.7699,
+      "loss": 0.8336,
       "step": 8760
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.594018734226259e-05,
-      "loss": 0.8506,
+      "loss": 0.9968,
       "step": 8761
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5939239768815385e-05,
-      "loss": 0.7767,
+      "loss": 0.9424,
       "step": 8762
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5938292112969656e-05,
-      "loss": 0.919,
+      "loss": 0.9469,
       "step": 8763
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.593734437473855e-05,
-      "loss": 0.8762,
+      "loss": 0.9987,
       "step": 8764
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5936396554135213e-05,
-      "loss": 0.6957,
+      "loss": 0.8421,
       "step": 8765
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.59354486511728e-05,
-      "loss": 0.784,
+      "loss": 0.9137,
       "step": 8766
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5934500665864453e-05,
-      "loss": 0.878,
+      "loss": 1.0455,
       "step": 8767
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5933552598223333e-05,
-      "loss": 0.754,
+      "loss": 0.8816,
       "step": 8768
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5932604448262587e-05,
-      "loss": 0.9952,
+      "loss": 1.006,
       "step": 8769
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.593165621599537e-05,
-      "loss": 0.8662,
+      "loss": 0.9347,
       "step": 8770
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5930707901434847e-05,
-      "loss": 0.8749,
+      "loss": 0.8603,
       "step": 8771
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5929759504594158e-05,
-      "loss": 0.734,
+      "loss": 0.8991,
       "step": 8772
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5928811025486472e-05,
-      "loss": 0.7977,
+      "loss": 0.8558,
       "step": 8773
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5927862464124944e-05,
-      "loss": 0.7864,
+      "loss": 0.9033,
       "step": 8774
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5926913820522738e-05,
-      "loss": 0.8763,
+      "loss": 1.004,
       "step": 8775
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.592596509469301e-05,
-      "loss": 0.7232,
+      "loss": 0.8458,
       "step": 8776
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5925016286648924e-05,
-      "loss": 0.8219,
+      "loss": 0.9085,
       "step": 8777
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5924067396403642e-05,
-      "loss": 0.7755,
+      "loss": 0.9346,
       "step": 8778
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5923118423970333e-05,
-      "loss": 0.6792,
+      "loss": 0.837,
       "step": 8779
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5922169369362157e-05,
-      "loss": 0.6672,
+      "loss": 0.821,
       "step": 8780
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5921220232592283e-05,
-      "loss": 0.6882,
+      "loss": 0.8105,
       "step": 8781
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.592027101367388e-05,
-      "loss": 0.8864,
+      "loss": 0.9404,
       "step": 8782
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5919321712620117e-05,
-      "loss": 0.7064,
+      "loss": 0.7361,
       "step": 8783
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5918372329444168e-05,
-      "loss": 0.7844,
+      "loss": 0.8272,
       "step": 8784
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5917422864159196e-05,
-      "loss": 0.7691,
+      "loss": 0.8261,
       "step": 8785
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5916473316778374e-05,
-      "loss": 0.8481,
+      "loss": 0.8776,
       "step": 8786
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5915523687314887e-05,
-      "loss": 0.7968,
+      "loss": 0.954,
       "step": 8787
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.59145739757819e-05,
-      "loss": 0.8133,
+      "loss": 0.8538,
       "step": 8788
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5913624182192588e-05,
-      "loss": 0.7881,
+      "loss": 0.83,
       "step": 8789
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5912674306560126e-05,
-      "loss": 0.759,
+      "loss": 0.8985,
       "step": 8790
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5911724348897704e-05,
-      "loss": 0.7845,
+      "loss": 0.8636,
       "step": 8791
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5910774309218494e-05,
-      "loss": 0.8421,
+      "loss": 0.9072,
       "step": 8792
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5909824187535673e-05,
-      "loss": 0.871,
+      "loss": 0.9703,
       "step": 8793
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5908873983862428e-05,
-      "loss": 0.8845,
+      "loss": 0.9367,
       "step": 8794
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5907923698211942e-05,
-      "loss": 0.8232,
+      "loss": 0.9972,
       "step": 8795
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5906973330597394e-05,
-      "loss": 0.6904,
+      "loss": 0.9249,
       "step": 8796
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5906022881031976e-05,
-      "loss": 0.7758,
+      "loss": 0.8777,
       "step": 8797
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5905072349528865e-05,
-      "loss": 0.7936,
+      "loss": 0.8646,
       "step": 8798
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5904121736101257e-05,
-      "loss": 0.7125,
+      "loss": 0.8794,
       "step": 8799
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5903171040762337e-05,
-      "loss": 0.824,
+      "loss": 0.9796,
       "step": 8800
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.590222026352529e-05,
-      "loss": 0.8387,
+      "loss": 0.8909,
       "step": 8801
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5901269404403316e-05,
-      "loss": 0.8957,
+      "loss": 0.9479,
       "step": 8802
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5900318463409598e-05,
-      "loss": 0.7875,
+      "loss": 0.9263,
       "step": 8803
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5899367440557335e-05,
-      "loss": 0.9534,
+      "loss": 0.942,
       "step": 8804
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5898416335859718e-05,
-      "loss": 0.905,
+      "loss": 0.9472,
       "step": 8805
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5897465149329946e-05,
-      "loss": 0.8912,
+      "loss": 0.9121,
       "step": 8806
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.589651388098121e-05,
-      "loss": 0.673,
+      "loss": 0.8867,
       "step": 8807
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.589556253082671e-05,
-      "loss": 0.7138,
+      "loss": 0.7533,
       "step": 8808
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5894611098879643e-05,
-      "loss": 0.7045,
+      "loss": 0.9146,
       "step": 8809
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.589365958515321e-05,
-      "loss": 0.8066,
+      "loss": 0.9635,
       "step": 8810
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5892707989660617e-05,
-      "loss": 0.7926,
+      "loss": 0.9504,
       "step": 8811
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5891756312415057e-05,
-      "loss": 0.7462,
+      "loss": 0.8291,
       "step": 8812
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.589080455342974e-05,
-      "loss": 0.8804,
+      "loss": 0.9545,
       "step": 8813
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5889852712717866e-05,
-      "loss": 0.6752,
+      "loss": 0.8274,
       "step": 8814
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5888900790292642e-05,
-      "loss": 0.8945,
+      "loss": 0.9803,
       "step": 8815
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5887948786167276e-05,
-      "loss": 0.9075,
+      "loss": 0.9338,
       "step": 8816
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5886996700354974e-05,
-      "loss": 0.8049,
+      "loss": 0.9066,
       "step": 8817
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5886044532868946e-05,
-      "loss": 0.7211,
+      "loss": 0.8335,
       "step": 8818
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.58850922837224e-05,
-      "loss": 0.7566,
+      "loss": 0.8422,
       "step": 8819
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.588413995292855e-05,
-      "loss": 0.7644,
+      "loss": 0.9033,
       "step": 8820
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5883187540500606e-05,
-      "loss": 0.7386,
+      "loss": 0.9096,
       "step": 8821
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5882235046451782e-05,
-      "loss": 0.7182,
+      "loss": 0.8714,
       "step": 8822
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5881282470795292e-05,
-      "loss": 0.8237,
+      "loss": 0.9473,
       "step": 8823
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.588032981354435e-05,
-      "loss": 0.8274,
+      "loss": 0.9059,
       "step": 8824
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5879377074712185e-05,
-      "loss": 0.7849,
+      "loss": 0.8652,
       "step": 8825
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5878424254311996e-05,
-      "loss": 0.9309,
+      "loss": 0.9509,
       "step": 8826
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.587747135235701e-05,
-      "loss": 0.7255,
+      "loss": 0.8424,
       "step": 8827
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5876518368860452e-05,
-      "loss": 0.9613,
+      "loss": 0.9667,
       "step": 8828
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5875565303835535e-05,
-      "loss": 0.7694,
+      "loss": 0.8968,
       "step": 8829
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5874612157295487e-05,
-      "loss": 0.8,
+      "loss": 0.886,
       "step": 8830
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5873658929253536e-05,
-      "loss": 0.9371,
+      "loss": 0.9648,
       "step": 8831
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5872705619722896e-05,
-      "loss": 0.7855,
+      "loss": 0.9298,
       "step": 8832
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5871752228716797e-05,
-      "loss": 0.9519,
+      "loss": 1.0061,
       "step": 8833
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.587079875624847e-05,
-      "loss": 0.7971,
+      "loss": 0.8484,
       "step": 8834
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5869845202331138e-05,
-      "loss": 0.7523,
+      "loss": 0.9037,
       "step": 8835
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5868891566978032e-05,
-      "loss": 0.8989,
+      "loss": 0.9825,
       "step": 8836
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586793785020238e-05,
-      "loss": 0.8634,
+      "loss": 0.945,
       "step": 8837
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5866984052017417e-05,
-      "loss": 0.7595,
+      "loss": 0.881,
       "step": 8838
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5866030172436378e-05,
-      "loss": 0.743,
+      "loss": 0.8923,
       "step": 8839
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586507621147249e-05,
-      "loss": 1.0172,
+      "loss": 1.0135,
       "step": 8840
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586412216913899e-05,
-      "loss": 0.8185,
+      "loss": 0.9358,
       "step": 8841
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586316804544912e-05,
-      "loss": 0.8085,
+      "loss": 0.8222,
       "step": 8842
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586221384041611e-05,
-      "loss": 0.6876,
+      "loss": 0.9322,
       "step": 8843
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.58612595540532e-05,
-      "loss": 0.814,
+      "loss": 0.96,
       "step": 8844
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.586030518637363e-05,
-      "loss": 0.7695,
+      "loss": 0.9046,
       "step": 8845
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585935073739064e-05,
-      "loss": 0.7244,
+      "loss": 0.845,
       "step": 8846
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585839620711747e-05,
-      "loss": 0.6648,
+      "loss": 0.864,
       "step": 8847
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5857441595567365e-05,
-      "loss": 0.8484,
+      "loss": 0.9555,
       "step": 8848
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585648690275357e-05,
-      "loss": 0.6884,
+      "loss": 0.8547,
       "step": 8849
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5855532128689328e-05,
-      "loss": 0.8148,
+      "loss": 0.895,
       "step": 8850
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5854577273387882e-05,
-      "loss": 0.7423,
+      "loss": 0.9288,
       "step": 8851
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585362233686249e-05,
-      "loss": 0.8428,
+      "loss": 0.9438,
       "step": 8852
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585266731912639e-05,
-      "loss": 0.8577,
+      "loss": 0.9135,
       "step": 8853
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.585171222019283e-05,
-      "loss": 0.8193,
+      "loss": 0.9091,
       "step": 8854
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5850757040075073e-05,
-      "loss": 0.8804,
+      "loss": 0.9376,
       "step": 8855
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584980177878636e-05,
-      "loss": 0.646,
+      "loss": 0.7912,
       "step": 8856
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5848846436339945e-05,
-      "loss": 0.8536,
+      "loss": 0.9532,
       "step": 8857
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584789101274909e-05,
-      "loss": 0.7682,
+      "loss": 0.8746,
       "step": 8858
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5846935508027036e-05,
-      "loss": 0.7907,
+      "loss": 0.9622,
       "step": 8859
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5845979922187056e-05,
-      "loss": 0.7777,
+      "loss": 0.8771,
       "step": 8860
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584502425524239e-05,
-      "loss": 0.7234,
+      "loss": 0.8548,
       "step": 8861
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5844068507206313e-05,
-      "loss": 0.6502,
+      "loss": 0.8726,
       "step": 8862
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5843112678092074e-05,
-      "loss": 0.6613,
+      "loss": 0.8281,
       "step": 8863
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584215676791294e-05,
-      "loss": 0.8296,
+      "loss": 0.8109,
       "step": 8864
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584120077668217e-05,
-      "loss": 0.8751,
+      "loss": 0.9418,
       "step": 8865
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.584024470441302e-05,
-      "loss": 0.9108,
+      "loss": 0.9563,
       "step": 8866
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5839288551118768e-05,
-      "loss": 0.8607,
+      "loss": 1.0,
       "step": 8867
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.583833231681267e-05,
-      "loss": 0.8568,
+      "loss": 0.9194,
       "step": 8868
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5837376001507993e-05,
-      "loss": 0.7179,
+      "loss": 0.8515,
       "step": 8869
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5836419605218006e-05,
-      "loss": 0.8551,
+      "loss": 0.9331,
       "step": 8870
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.583546312795598e-05,
-      "loss": 0.7923,
+      "loss": 0.893,
       "step": 8871
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.583450656973518e-05,
-      "loss": 0.848,
+      "loss": 0.9327,
       "step": 8872
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.583354993056888e-05,
-      "loss": 0.8149,
+      "loss": 0.8291,
       "step": 8873
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5832593210470353e-05,
-      "loss": 0.8207,
+      "loss": 0.9661,
       "step": 8874
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5831636409452868e-05,
-      "loss": 0.7155,
+      "loss": 0.9096,
       "step": 8875
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5830679527529704e-05,
-      "loss": 0.7824,
+      "loss": 0.9347,
       "step": 8876
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5829722564714134e-05,
-      "loss": 0.8115,
+      "loss": 0.963,
       "step": 8877
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5828765521019437e-05,
-      "loss": 0.8836,
+      "loss": 0.9254,
       "step": 8878
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5827808396458883e-05,
-      "loss": 0.8196,
+      "loss": 0.9248,
       "step": 8879
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.582685119104576e-05,
-      "loss": 0.8549,
+      "loss": 0.9845,
       "step": 8880
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.582589390479334e-05,
-      "loss": 0.7784,
+      "loss": 0.9172,
       "step": 8881
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5824936537714913e-05,
-      "loss": 0.8102,
+      "loss": 0.9199,
       "step": 8882
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5823979089823757e-05,
-      "loss": 0.8027,
+      "loss": 0.8885,
       "step": 8883
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5823021561133146e-05,
-      "loss": 0.7902,
+      "loss": 0.8915,
       "step": 8884
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5822063951656378e-05,
-      "loss": 0.8027,
+      "loss": 0.8781,
       "step": 8885
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5821106261406737e-05,
-      "loss": 0.7744,
+      "loss": 0.8421,
       "step": 8886
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5820148490397496e-05,
-      "loss": 0.8566,
+      "loss": 0.9937,
       "step": 8887
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581919063864196e-05,
-      "loss": 0.7558,
+      "loss": 0.8779,
       "step": 8888
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581823270615341e-05,
-      "loss": 0.5747,
+      "loss": 0.8244,
       "step": 8889
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5817274692945136e-05,
-      "loss": 0.7913,
+      "loss": 0.9394,
       "step": 8890
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5816316599030425e-05,
-      "loss": 0.8007,
+      "loss": 0.9033,
       "step": 8891
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5815358424422575e-05,
-      "loss": 0.7403,
+      "loss": 0.8692,
       "step": 8892
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581440016913488e-05,
-      "loss": 0.6258,
+      "loss": 0.9079,
       "step": 8893
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581344183318063e-05,
-      "loss": 0.8099,
+      "loss": 0.9489,
       "step": 8894
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5812483416573123e-05,
-      "loss": 0.6795,
+      "loss": 0.7256,
       "step": 8895
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581152491932566e-05,
-      "loss": 0.8184,
+      "loss": 0.9113,
       "step": 8896
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.581056634145153e-05,
-      "loss": 0.7026,
+      "loss": 0.8442,
       "step": 8897
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5809607682964035e-05,
-      "loss": 0.821,
+      "loss": 0.8999,
       "step": 8898
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5808648943876478e-05,
-      "loss": 0.7945,
+      "loss": 0.9343,
       "step": 8899
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5807690124202156e-05,
-      "loss": 0.7636,
+      "loss": 0.8295,
       "step": 8900
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.580673122395438e-05,
-      "loss": 0.8873,
+      "loss": 0.9237,
       "step": 8901
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5805772243146442e-05,
-      "loss": 0.8534,
+      "loss": 1.0182,
       "step": 8902
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.580481318179165e-05,
-      "loss": 0.83,
+      "loss": 0.9696,
       "step": 8903
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5803854039903313e-05,
-      "loss": 0.6948,
+      "loss": 0.7959,
       "step": 8904
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5802894817494738e-05,
-      "loss": 0.7912,
+      "loss": 0.8874,
       "step": 8905
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.580193551457923e-05,
-      "loss": 0.8804,
+      "loss": 0.9474,
       "step": 8906
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5800976131170097e-05,
-      "loss": 0.7942,
+      "loss": 0.886,
       "step": 8907
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5800016667280648e-05,
-      "loss": 0.7609,
+      "loss": 0.9254,
       "step": 8908
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.57990571229242e-05,
-      "loss": 0.6662,
+      "loss": 0.8005,
       "step": 8909
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.579809749811406e-05,
-      "loss": 0.8342,
+      "loss": 0.9312,
       "step": 8910
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5797137792863546e-05,
-      "loss": 0.7438,
+      "loss": 0.8557,
       "step": 8911
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5796178007185973e-05,
-      "loss": 0.8787,
+      "loss": 0.9575,
       "step": 8912
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.579521814109465e-05,
-      "loss": 0.6405,
+      "loss": 0.8064,
       "step": 8913
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5794258194602896e-05,
-      "loss": 0.8635,
+      "loss": 0.9499,
       "step": 8914
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5793298167724034e-05,
-      "loss": 0.6974,
+      "loss": 0.8163,
       "step": 8915
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5792338060471378e-05,
-      "loss": 0.8182,
+      "loss": 0.9544,
       "step": 8916
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.579137787285825e-05,
-      "loss": 0.7517,
+      "loss": 0.8806,
       "step": 8917
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5790417604897968e-05,
-      "loss": 0.7804,
+      "loss": 0.8926,
       "step": 8918
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5789457256603862e-05,
-      "loss": 0.894,
+      "loss": 0.9432,
       "step": 8919
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.578849682798925e-05,
-      "loss": 0.7693,
+      "loss": 0.9313,
       "step": 8920
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5787536319067454e-05,
-      "loss": 0.7524,
+      "loss": 0.9425,
       "step": 8921
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5786575729851808e-05,
-      "loss": 0.8786,
+      "loss": 0.9028,
       "step": 8922
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.578561506035563e-05,
-      "loss": 0.6553,
+      "loss": 0.7551,
       "step": 8923
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5784654310592253e-05,
-      "loss": 0.8171,
+      "loss": 0.865,
       "step": 8924
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5783693480575005e-05,
-      "loss": 0.8739,
+      "loss": 0.8988,
       "step": 8925
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5782732570317215e-05,
-      "loss": 0.9373,
+      "loss": 1.0475,
       "step": 8926
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5781771579832216e-05,
-      "loss": 0.8557,
+      "loss": 0.9043,
       "step": 8927
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5780810509133338e-05,
-      "loss": 0.7554,
+      "loss": 0.9014,
       "step": 8928
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5779849358233918e-05,
-      "loss": 0.6729,
+      "loss": 0.8929,
       "step": 8929
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.577888812714729e-05,
-      "loss": 0.8465,
+      "loss": 0.9376,
       "step": 8930
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5777926815886785e-05,
-      "loss": 0.6078,
+      "loss": 0.7261,
       "step": 8931
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5776965424465746e-05,
-      "loss": 0.8033,
+      "loss": 0.8922,
       "step": 8932
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5776003952897504e-05,
-      "loss": 0.6267,
+      "loss": 0.7848,
       "step": 8933
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5775042401195406e-05,
-      "loss": 0.8361,
+      "loss": 0.8834,
       "step": 8934
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.5774080769372786e-05,
-      "loss": 0.6739,
+      "loss": 0.7824,
       "step": 8935
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.577311905744299e-05,
-      "loss": 0.7988,
+      "loss": 0.8642,
       "step": 8936
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5772157265419357e-05,
-      "loss": 0.6563,
+      "loss": 0.8525,
       "step": 8937
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.577119539331523e-05,
-      "loss": 0.7514,
+      "loss": 0.8498,
       "step": 8938
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5770233441143958e-05,
-      "loss": 0.7837,
+      "loss": 0.8421,
       "step": 8939
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5769271408918886e-05,
-      "loss": 0.7867,
+      "loss": 0.87,
       "step": 8940
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5768309296653355e-05,
-      "loss": 0.8975,
+      "loss": 0.9348,
       "step": 8941
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5767347104360716e-05,
-      "loss": 0.8237,
+      "loss": 0.8836,
       "step": 8942
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.576638483205432e-05,
-      "loss": 0.8518,
+      "loss": 0.9318,
       "step": 8943
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5765422479747515e-05,
-      "loss": 0.9792,
+      "loss": 1.0892,
       "step": 8944
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5764460047453657e-05,
-      "loss": 0.7935,
+      "loss": 0.8889,
       "step": 8945
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.576349753518609e-05,
-      "loss": 0.8594,
+      "loss": 0.9943,
       "step": 8946
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.576253494295817e-05,
-      "loss": 0.7661,
+      "loss": 0.8906,
       "step": 8947
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5761572270783262e-05,
-      "loss": 0.7404,
+      "loss": 0.8537,
       "step": 8948
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5760609518674707e-05,
-      "loss": 0.8223,
+      "loss": 0.9416,
       "step": 8949
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.575964668664587e-05,
-      "loss": 0.7133,
+      "loss": 0.8703,
       "step": 8950
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.575868377471011e-05,
-      "loss": 0.7349,
+      "loss": 0.9731,
       "step": 8951
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5757720782880777e-05,
-      "loss": 0.6587,
+      "loss": 0.9084,
       "step": 8952
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.575675771117124e-05,
-      "loss": 0.7462,
+      "loss": 0.8468,
       "step": 8953
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5755794559594857e-05,
-      "loss": 0.8227,
+      "loss": 0.9323,
       "step": 8954
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.575483132816499e-05,
-      "loss": 0.7213,
+      "loss": 0.9206,
       "step": 8955
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5753868016895008e-05,
-      "loss": 0.7428,
+      "loss": 0.8921,
       "step": 8956
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5752904625798266e-05,
-      "loss": 0.7988,
+      "loss": 0.9096,
       "step": 8957
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5751941154888136e-05,
-      "loss": 0.6545,
+      "loss": 0.7531,
       "step": 8958
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5750977604177985e-05,
-      "loss": 0.7393,
+      "loss": 0.8752,
       "step": 8959
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.575001397368118e-05,
-      "loss": 0.8152,
+      "loss": 0.9121,
       "step": 8960
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5749050263411086e-05,
-      "loss": 0.7965,
+      "loss": 0.9431,
       "step": 8961
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.574808647338108e-05,
-      "loss": 0.921,
+      "loss": 1.0157,
       "step": 8962
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.574712260360453e-05,
-      "loss": 0.8142,
+      "loss": 0.887,
       "step": 8963
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5746158654094804e-05,
-      "loss": 0.7229,
+      "loss": 0.9441,
       "step": 8964
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.574519462486528e-05,
-      "loss": 0.7712,
+      "loss": 0.8582,
       "step": 8965
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.574423051592933e-05,
-      "loss": 0.8153,
+      "loss": 0.9345,
       "step": 8966
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5743266327300333e-05,
-      "loss": 0.8441,
+      "loss": 1.0043,
       "step": 8967
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.574230205899167e-05,
-      "loss": 0.7591,
+      "loss": 0.9355,
       "step": 8968
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5741337711016707e-05,
-      "loss": 0.7696,
+      "loss": 0.8176,
       "step": 8969
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5740373283388834e-05,
-      "loss": 0.8301,
+      "loss": 0.8784,
       "step": 8970
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5739408776121425e-05,
-      "loss": 0.8251,
+      "loss": 0.9603,
       "step": 8971
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5738444189227857e-05,
-      "loss": 0.7344,
+      "loss": 0.8478,
       "step": 8972
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5737479522721524e-05,
-      "loss": 0.741,
+      "loss": 0.874,
       "step": 8973
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5736514776615797e-05,
-      "loss": 0.6914,
+      "loss": 0.8817,
       "step": 8974
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5735549950924075e-05,
-      "loss": 0.6395,
+      "loss": 0.8386,
       "step": 8975
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5734585045659727e-05,
-      "loss": 0.6968,
+      "loss": 0.8386,
       "step": 8976
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.573362006083615e-05,
-      "loss": 0.7106,
+      "loss": 0.915,
       "step": 8977
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.573265499646673e-05,
-      "loss": 0.7366,
+      "loss": 0.8944,
       "step": 8978
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.573168985256486e-05,
-      "loss": 0.7846,
+      "loss": 0.858,
       "step": 8979
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.573072462914392e-05,
-      "loss": 0.8476,
+      "loss": 0.9681,
       "step": 8980
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.572975932621731e-05,
-      "loss": 0.7689,
+      "loss": 0.8868,
       "step": 8981
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5728793943798416e-05,
-      "loss": 0.8229,
+      "loss": 0.8841,
       "step": 8982
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5727828481900634e-05,
-      "loss": 0.8041,
+      "loss": 0.933,
       "step": 8983
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.572686294053736e-05,
-      "loss": 0.8332,
+      "loss": 0.8803,
       "step": 8984
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5725897319721988e-05,
-      "loss": 0.6747,
+      "loss": 0.8221,
       "step": 8985
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5724931619467914e-05,
-      "loss": 0.7232,
+      "loss": 0.9014,
       "step": 8986
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5723965839788537e-05,
-      "loss": 0.8933,
+      "loss": 1.0263,
       "step": 8987
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5722999980697256e-05,
-      "loss": 0.8532,
+      "loss": 0.962,
       "step": 8988
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5722034042207468e-05,
-      "loss": 0.7129,
+      "loss": 0.938,
       "step": 8989
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5721068024332577e-05,
-      "loss": 0.7747,
+      "loss": 0.9024,
       "step": 8990
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5720101927085985e-05,
-      "loss": 0.8024,
+      "loss": 0.8849,
       "step": 8991
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5719135750481097e-05,
-      "loss": 0.8597,
+      "loss": 0.9559,
       "step": 8992
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.571816949453131e-05,
-      "loss": 1.0005,
+      "loss": 1.0525,
       "step": 8993
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5717203159250035e-05,
-      "loss": 0.9331,
+      "loss": 0.9886,
       "step": 8994
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.571623674465068e-05,
-      "loss": 0.8182,
+      "loss": 0.9258,
       "step": 8995
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5715270250746648e-05,
-      "loss": 0.8016,
+      "loss": 0.894,
       "step": 8996
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5714303677551353e-05,
-      "loss": 0.7966,
+      "loss": 0.8699,
       "step": 8997
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.57133370250782e-05,
-      "loss": 0.8468,
+      "loss": 0.9292,
       "step": 8998
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5712370293340605e-05,
-      "loss": 0.8516,
+      "loss": 0.9162,
       "step": 8999
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5711403482351973e-05,
-      "loss": 0.7849,
+      "loss": 0.8679,
       "step": 9000
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5710436592125724e-05,
-      "loss": 0.7783,
+      "loss": 0.856,
       "step": 9001
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.570946962267527e-05,
-      "loss": 0.934,
+      "loss": 0.9016,
       "step": 9002
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.570850257401402e-05,
-      "loss": 0.9242,
+      "loss": 0.9806,
       "step": 9003
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5707535446155402e-05,
-      "loss": 0.6786,
+      "loss": 0.8063,
       "step": 9004
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5706568239112826e-05,
-      "loss": 0.9263,
+      "loss": 0.9618,
       "step": 9005
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.570560095289971e-05,
-      "loss": 0.782,
+      "loss": 0.9184,
       "step": 9006
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5704633587529478e-05,
-      "loss": 0.8046,
+      "loss": 0.9454,
       "step": 9007
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.570366614301555e-05,
-      "loss": 0.8457,
+      "loss": 0.9181,
       "step": 9008
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5702698619371344e-05,
-      "loss": 0.8793,
+      "loss": 0.976,
       "step": 9009
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5701731016610286e-05,
-      "loss": 0.7395,
+      "loss": 0.8628,
       "step": 9010
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5700763334745804e-05,
-      "loss": 0.9274,
+      "loss": 0.9861,
       "step": 9011
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5699795573791314e-05,
-      "loss": 0.7432,
+      "loss": 0.9137,
       "step": 9012
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5698827733760248e-05,
-      "loss": 0.9621,
+      "loss": 1.0799,
       "step": 9013
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5697859814666035e-05,
-      "loss": 0.8311,
+      "loss": 0.9572,
       "step": 9014
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.56968918165221e-05,
-      "loss": 0.8994,
+      "loss": 0.8447,
       "step": 9015
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5695923739341872e-05,
-      "loss": 0.7601,
+      "loss": 0.8388,
       "step": 9016
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.569495558313879e-05,
-      "loss": 0.7292,
+      "loss": 0.8375,
       "step": 9017
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5693987347926275e-05,
-      "loss": 0.8647,
+      "loss": 0.886,
       "step": 9018
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5693019033717765e-05,
-      "loss": 0.8507,
+      "loss": 0.9864,
       "step": 9019
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5692050640526692e-05,
-      "loss": 0.8138,
+      "loss": 0.9404,
       "step": 9020
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.569108216836649e-05,
-      "loss": 0.6281,
+      "loss": 0.77,
       "step": 9021
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.56901136172506e-05,
-      "loss": 0.8545,
+      "loss": 1.0099,
       "step": 9022
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.568914498719246e-05,
-      "loss": 0.7275,
+      "loss": 0.8647,
       "step": 9023
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.56881762782055e-05,
-      "loss": 0.6182,
+      "loss": 0.7828,
       "step": 9024
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.568720749030317e-05,
-      "loss": 0.7537,
+      "loss": 0.8675,
       "step": 9025
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.56862386234989e-05,
-      "loss": 0.8559,
+      "loss": 0.9128,
       "step": 9026
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5685269677806138e-05,
-      "loss": 0.8128,
+      "loss": 0.8802,
       "step": 9027
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.568430065323833e-05,
-      "loss": 0.6793,
+      "loss": 0.8162,
       "step": 9028
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5683331549808907e-05,
-      "loss": 0.7626,
+      "loss": 0.8823,
       "step": 9029
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.568236236753133e-05,
-      "loss": 0.9191,
+      "loss": 0.9691,
       "step": 9030
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5681393106419034e-05,
-      "loss": 0.7715,
+      "loss": 0.9148,
       "step": 9031
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.568042376648547e-05,
-      "loss": 0.7396,
+      "loss": 0.8579,
       "step": 9032
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5679454347744086e-05,
-      "loss": 0.826,
+      "loss": 0.9194,
       "step": 9033
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.567848485020833e-05,
-      "loss": 0.736,
+      "loss": 0.8322,
       "step": 9034
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5677515273891656e-05,
-      "loss": 0.7278,
+      "loss": 0.8568,
       "step": 9035
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.567654561880751e-05,
-      "loss": 0.7445,
+      "loss": 0.8227,
       "step": 9036
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.567557588496935e-05,
-      "loss": 0.7103,
+      "loss": 0.8243,
       "step": 9037
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5674606072390624e-05,
-      "loss": 0.7796,
+      "loss": 0.8653,
       "step": 9038
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5673636181084792e-05,
-      "loss": 0.7454,
+      "loss": 0.8391,
       "step": 9039
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.567266621106531e-05,
-      "loss": 0.9124,
+      "loss": 0.9814,
       "step": 9040
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5671696162345628e-05,
-      "loss": 0.7436,
+      "loss": 0.8631,
       "step": 9041
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.567072603493921e-05,
-      "loss": 0.9425,
+      "loss": 0.9837,
       "step": 9042
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5669755828859517e-05,
-      "loss": 0.8023,
+      "loss": 1.0388,
       "step": 9043
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5668785544120003e-05,
-      "loss": 0.774,
+      "loss": 0.876,
       "step": 9044
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5667815180734138e-05,
-      "loss": 0.8772,
+      "loss": 0.9284,
       "step": 9045
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.566684473871537e-05,
-      "loss": 0.7457,
+      "loss": 0.851,
       "step": 9046
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5665874218077178e-05,
-      "loss": 0.8729,
+      "loss": 0.8458,
       "step": 9047
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.566490361883302e-05,
-      "loss": 0.7704,
+      "loss": 0.8707,
       "step": 9048
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.566393294099636e-05,
-      "loss": 0.7218,
+      "loss": 0.8362,
       "step": 9049
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5662962184580665e-05,
-      "loss": 0.7493,
+      "loss": 0.8705,
       "step": 9050
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5661991349599407e-05,
-      "loss": 0.7572,
+      "loss": 0.8921,
       "step": 9051
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.566102043606605e-05,
-      "loss": 0.7333,
+      "loss": 0.8551,
       "step": 9052
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5660049443994068e-05,
-      "loss": 0.7866,
+      "loss": 0.9435,
       "step": 9053
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.565907837339693e-05,
-      "loss": 0.8538,
+      "loss": 0.9124,
       "step": 9054
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.565810722428811e-05,
-      "loss": 0.7346,
+      "loss": 0.8877,
       "step": 9055
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5657135996681082e-05,
-      "loss": 0.766,
+      "loss": 0.9625,
       "step": 9056
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5656164690589312e-05,
-      "loss": 0.8054,
+      "loss": 0.8712,
       "step": 9057
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5655193306026285e-05,
-      "loss": 0.7316,
+      "loss": 0.8324,
       "step": 9058
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5654221843005476e-05,
-      "loss": 0.8099,
+      "loss": 0.9027,
       "step": 9059
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5653250301540363e-05,
-      "loss": 0.9912,
+      "loss": 1.0253,
       "step": 9060
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.565227868164442e-05,
-      "loss": 0.8682,
+      "loss": 0.938,
       "step": 9061
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5651306983331126e-05,
-      "loss": 0.7017,
+      "loss": 0.8229,
       "step": 9062
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5650335206613973e-05,
-      "loss": 0.7521,
+      "loss": 0.8618,
       "step": 9063
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5649363351506434e-05,
-      "loss": 0.7973,
+      "loss": 0.9164,
       "step": 9064
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.564839141802199e-05,
-      "loss": 0.9303,
+      "loss": 1.0471,
       "step": 9065
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5647419406174132e-05,
-      "loss": 0.7837,
+      "loss": 0.863,
       "step": 9066
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5646447315976342e-05,
-      "loss": 0.6916,
+      "loss": 0.8066,
       "step": 9067
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5645475147442106e-05,
-      "loss": 0.7318,
+      "loss": 0.9636,
       "step": 9068
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5644502900584916e-05,
-      "loss": 0.7475,
+      "loss": 0.8922,
       "step": 9069
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5643530575418252e-05,
-      "loss": 0.7539,
+      "loss": 0.8538,
       "step": 9070
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.564255817195561e-05,
-      "loss": 0.6168,
+      "loss": 0.8374,
       "step": 9071
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5641585690210477e-05,
-      "loss": 0.7706,
+      "loss": 0.8967,
       "step": 9072
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.564061313019635e-05,
-      "loss": 0.7853,
+      "loss": 0.931,
       "step": 9073
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5639640491926718e-05,
-      "loss": 0.9459,
+      "loss": 0.8999,
       "step": 9074
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5638667775415078e-05,
-      "loss": 0.7878,
+      "loss": 0.9111,
       "step": 9075
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.563769498067492e-05,
-      "loss": 0.8121,
+      "loss": 0.8477,
       "step": 9076
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.563672210771974e-05,
-      "loss": 0.7882,
+      "loss": 0.8954,
       "step": 9077
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5635749156563043e-05,
-      "loss": 0.7226,
+      "loss": 0.9079,
       "step": 9078
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5634776127218324e-05,
-      "loss": 0.819,
+      "loss": 0.8925,
       "step": 9079
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.563380301969908e-05,
-      "loss": 0.797,
+      "loss": 0.9774,
       "step": 9080
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5632829834018808e-05,
-      "loss": 0.7649,
+      "loss": 0.9057,
       "step": 9081
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5631856570191018e-05,
-      "loss": 0.826,
+      "loss": 0.903,
       "step": 9082
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5630883228229206e-05,
-      "loss": 0.8871,
+      "loss": 0.9697,
       "step": 9083
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.562990980814688e-05,
-      "loss": 0.71,
+      "loss": 0.8341,
       "step": 9084
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5628936309957542e-05,
-      "loss": 0.8192,
+      "loss": 0.9413,
       "step": 9085
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.56279627336747e-05,
-      "loss": 0.8068,
+      "loss": 0.9082,
       "step": 9086
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.562698907931186e-05,
-      "loss": 0.7838,
+      "loss": 0.9153,
       "step": 9087
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.562601534688253e-05,
-      "loss": 0.7564,
+      "loss": 0.8221,
       "step": 9088
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5625041536400218e-05,
-      "loss": 0.7827,
+      "loss": 0.8813,
       "step": 9089
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.562406764787844e-05,
-      "loss": 0.7213,
+      "loss": 0.7929,
       "step": 9090
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5623093681330698e-05,
-      "loss": 0.7795,
+      "loss": 0.9164,
       "step": 9091
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5622119636770512e-05,
-      "loss": 0.7016,
+      "loss": 0.9023,
       "step": 9092
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5621145514211393e-05,
-      "loss": 0.8197,
+      "loss": 0.8922,
       "step": 9093
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5620171313666855e-05,
-      "loss": 0.6579,
+      "loss": 0.8684,
       "step": 9094
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5619197035150417e-05,
-      "loss": 0.8732,
+      "loss": 0.9425,
       "step": 9095
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5618222678675587e-05,
-      "loss": 0.826,
+      "loss": 0.9127,
       "step": 9096
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.561724824425589e-05,
-      "loss": 0.817,
+      "loss": 0.9178,
       "step": 9097
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5616273731904847e-05,
-      "loss": 0.7839,
+      "loss": 0.8254,
       "step": 9098
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.561529914163597e-05,
-      "loss": 0.7213,
+      "loss": 0.8456,
       "step": 9099
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5614324473462784e-05,
-      "loss": 0.8505,
+      "loss": 0.8652,
       "step": 9100
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5613349727398816e-05,
-      "loss": 0.6982,
+      "loss": 0.8964,
       "step": 9101
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.561237490345758e-05,
-      "loss": 0.8522,
+      "loss": 0.927,
       "step": 9102
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5611400001652606e-05,
-      "loss": 0.8011,
+      "loss": 0.8722,
       "step": 9103
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5610425021997422e-05,
-      "loss": 0.6784,
+      "loss": 0.8169,
       "step": 9104
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.560944996450555e-05,
-      "loss": 0.7325,
+      "loss": 0.8651,
       "step": 9105
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5608474829190517e-05,
-      "loss": 0.8071,
+      "loss": 0.9119,
       "step": 9106
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5607499616065856e-05,
-      "loss": 0.8494,
+      "loss": 0.9542,
       "step": 9107
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.560652432514509e-05,
-      "loss": 0.8047,
+      "loss": 0.8862,
       "step": 9108
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5605548956441757e-05,
-      "loss": 0.7166,
+      "loss": 0.8171,
       "step": 9109
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5604573509969382e-05,
-      "loss": 0.8463,
+      "loss": 0.9555,
       "step": 9110
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5603597985741503e-05,
-      "loss": 0.894,
+      "loss": 0.8968,
       "step": 9111
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5602622383771654e-05,
-      "loss": 0.6623,
+      "loss": 0.8597,
       "step": 9112
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5601646704073368e-05,
-      "loss": 0.8086,
+      "loss": 0.9452,
       "step": 9113
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.560067094666018e-05,
-      "loss": 0.6556,
+      "loss": 0.7998,
       "step": 9114
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.559969511154563e-05,
-      "loss": 0.6636,
+      "loss": 0.8209,
       "step": 9115
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5598719198743256e-05,
-      "loss": 0.823,
+      "loss": 0.9389,
       "step": 9116
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5597743208266593e-05,
-      "loss": 0.899,
+      "loss": 0.9605,
       "step": 9117
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5596767140129186e-05,
-      "loss": 0.7474,
+      "loss": 0.888,
       "step": 9118
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.559579099434458e-05,
-      "loss": 0.7522,
+      "loss": 0.8168,
       "step": 9119
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5594814770926312e-05,
-      "loss": 0.6286,
+      "loss": 0.7753,
       "step": 9120
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5593838469887926e-05,
-      "loss": 0.8439,
+      "loss": 0.8852,
       "step": 9121
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5592862091242966e-05,
-      "loss": 0.6704,
+      "loss": 0.8257,
       "step": 9122
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5591885635004983e-05,
-      "loss": 0.7468,
+      "loss": 0.8325,
       "step": 9123
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5590909101187516e-05,
-      "loss": 0.7893,
+      "loss": 0.9329,
       "step": 9124
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.558993248980412e-05,
-      "loss": 0.7946,
+      "loss": 0.9176,
       "step": 9125
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5588955800868343e-05,
-      "loss": 0.8151,
+      "loss": 0.881,
       "step": 9126
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5587979034393735e-05,
-      "loss": 0.8241,
+      "loss": 0.8585,
       "step": 9127
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5587002190393844e-05,
-      "loss": 0.7261,
+      "loss": 0.886,
       "step": 9128
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5586025268882225e-05,
-      "loss": 0.9258,
+      "loss": 1.0096,
       "step": 9129
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5585048269872434e-05,
-      "loss": 0.8404,
+      "loss": 0.9367,
       "step": 9130
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5584071193378018e-05,
-      "loss": 0.7708,
+      "loss": 0.8568,
       "step": 9131
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5583094039412538e-05,
-      "loss": 0.745,
+      "loss": 0.8697,
       "step": 9132
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.558211680798955e-05,
-      "loss": 0.775,
+      "loss": 0.9414,
       "step": 9133
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5581139499122613e-05,
-      "loss": 0.8727,
+      "loss": 0.9567,
       "step": 9134
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5580162112825283e-05,
-      "loss": 0.8764,
+      "loss": 0.9532,
       "step": 9135
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.557918464911112e-05,
-      "loss": 0.7696,
+      "loss": 0.8822,
       "step": 9136
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5578207107993684e-05,
-      "loss": 0.6877,
+      "loss": 0.8196,
       "step": 9137
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.557722948948654e-05,
-      "loss": 0.825,
+      "loss": 0.9241,
       "step": 9138
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5576251793603252e-05,
-      "loss": 0.897,
+      "loss": 0.9442,
       "step": 9139
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5575274020357382e-05,
-      "loss": 0.7422,
+      "loss": 0.9121,
       "step": 9140
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5574296169762492e-05,
-      "loss": 0.6746,
+      "loss": 0.83,
       "step": 9141
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5573318241832157e-05,
-      "loss": 0.7561,
+      "loss": 0.8335,
       "step": 9142
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5572340236579934e-05,
-      "loss": 0.8299,
+      "loss": 0.9088,
       "step": 9143
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.55713621540194e-05,
-      "loss": 0.8474,
+      "loss": 0.907,
       "step": 9144
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5570383994164115e-05,
-      "loss": 0.7206,
+      "loss": 0.8486,
       "step": 9145
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.556940575702766e-05,
-      "loss": 0.7062,
+      "loss": 0.8683,
       "step": 9146
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.55684274426236e-05,
-      "loss": 0.8546,
+      "loss": 0.9635,
       "step": 9147
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5567449050965512e-05,
-      "loss": 0.6885,
+      "loss": 0.9055,
       "step": 9148
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.556647058206697e-05,
-      "loss": 0.7509,
+      "loss": 0.8561,
       "step": 9149
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.556549203594154e-05,
-      "loss": 0.7121,
+      "loss": 0.7783,
       "step": 9150
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5564513412602812e-05,
-      "loss": 0.7249,
+      "loss": 0.9135,
       "step": 9151
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5563534712064346e-05,
-      "loss": 0.7949,
+      "loss": 0.8803,
       "step": 9152
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5562555934339736e-05,
-      "loss": 0.6769,
+      "loss": 0.8812,
       "step": 9153
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.556157707944255e-05,
-      "loss": 0.6765,
+      "loss": 0.8607,
       "step": 9154
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5560598147386378e-05,
-      "loss": 0.6307,
+      "loss": 0.8544,
       "step": 9155
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5559619138184797e-05,
-      "loss": 0.835,
+      "loss": 0.964,
       "step": 9156
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5558640051851383e-05,
-      "loss": 0.7189,
+      "loss": 0.8033,
       "step": 9157
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5557660888399725e-05,
-      "loss": 0.8189,
+      "loss": 0.9863,
       "step": 9158
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5556681647843413e-05,
-      "loss": 0.8459,
+      "loss": 0.9402,
       "step": 9159
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5555702330196024e-05,
-      "loss": 0.6634,
+      "loss": 0.793,
       "step": 9160
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5554722935471148e-05,
-      "loss": 0.9239,
+      "loss": 0.9144,
       "step": 9161
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.555374346368237e-05,
-      "loss": 0.7961,
+      "loss": 0.8315,
       "step": 9162
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5552763914843285e-05,
-      "loss": 0.7942,
+      "loss": 0.8211,
       "step": 9163
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5551784288967477e-05,
-      "loss": 0.7742,
+      "loss": 0.917,
       "step": 9164
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5550804586068538e-05,
-      "loss": 0.7294,
+      "loss": 0.8642,
       "step": 9165
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5549824806160064e-05,
-      "loss": 0.7892,
+      "loss": 0.8625,
       "step": 9166
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.554884494925564e-05,
-      "loss": 0.7984,
+      "loss": 0.9164,
       "step": 9167
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.554786501536887e-05,
-      "loss": 0.8847,
+      "loss": 0.9572,
       "step": 9168
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5546885004513345e-05,
-      "loss": 0.7642,
+      "loss": 0.8922,
       "step": 9169
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5545904916702658e-05,
-      "loss": 0.854,
+      "loss": 0.8396,
       "step": 9170
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.554492475195041e-05,
-      "loss": 0.7854,
+      "loss": 0.8747,
       "step": 9171
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5543944510270197e-05,
-      "loss": 0.924,
+      "loss": 0.9247,
       "step": 9172
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.554296419167562e-05,
-      "loss": 0.7313,
+      "loss": 0.8071,
       "step": 9173
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.554198379618028e-05,
-      "loss": 0.7705,
+      "loss": 0.8277,
       "step": 9174
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5541003323797776e-05,
-      "loss": 0.6624,
+      "loss": 0.7991,
       "step": 9175
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5540022774541713e-05,
-      "loss": 0.7547,
+      "loss": 0.9187,
       "step": 9176
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5539042148425695e-05,
-      "loss": 0.6451,
+      "loss": 0.7994,
       "step": 9177
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5538061445463327e-05,
-      "loss": 0.8118,
+      "loss": 0.949,
       "step": 9178
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5537080665668214e-05,
-      "loss": 0.8156,
+      "loss": 0.884,
       "step": 9179
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.553609980905396e-05,
-      "loss": 0.6187,
+      "loss": 0.8777,
       "step": 9180
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.553511887563418e-05,
-      "loss": 0.7029,
+      "loss": 0.7916,
       "step": 9181
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5534137865422475e-05,
-      "loss": 0.6182,
+      "loss": 0.8139,
       "step": 9182
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5533156778432462e-05,
-      "loss": 0.7813,
+      "loss": 0.8211,
       "step": 9183
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5532175614677747e-05,
-      "loss": 0.8894,
+      "loss": 1.0203,
       "step": 9184
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5531194374171944e-05,
-      "loss": 0.7055,
+      "loss": 0.8515,
       "step": 9185
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.553021305692867e-05,
-      "loss": 0.7492,
+      "loss": 0.848,
       "step": 9186
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5529231662961533e-05,
-      "loss": 0.7502,
+      "loss": 0.9176,
       "step": 9187
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5528250192284155e-05,
-      "loss": 0.6892,
+      "loss": 0.8156,
       "step": 9188
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5527268644910146e-05,
-      "loss": 0.7822,
+      "loss": 0.965,
       "step": 9189
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5526287020853126e-05,
-      "loss": 0.742,
+      "loss": 0.8051,
       "step": 9190
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5525305320126716e-05,
-      "loss": 0.83,
+      "loss": 0.9534,
       "step": 9191
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5524323542744536e-05,
-      "loss": 0.9306,
+      "loss": 0.9692,
       "step": 9192
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5523341688720202e-05,
-      "loss": 0.8576,
+      "loss": 0.9355,
       "step": 9193
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.552235975806734e-05,
-      "loss": 0.7976,
+      "loss": 0.8403,
       "step": 9194
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5521377750799573e-05,
-      "loss": 0.9507,
+      "loss": 0.9897,
       "step": 9195
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5520395666930518e-05,
-      "loss": 0.8937,
+      "loss": 0.9801,
       "step": 9196
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.551941350647381e-05,
-      "loss": 0.8153,
+      "loss": 0.902,
       "step": 9197
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.551843126944307e-05,
-      "loss": 0.71,
+      "loss": 0.8656,
       "step": 9198
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5517448955851927e-05,
-      "loss": 0.7426,
+      "loss": 0.8575,
       "step": 9199
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5516466565714007e-05,
-      "loss": 0.809,
+      "loss": 0.9425,
       "step": 9200
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5515484099042943e-05,
-      "loss": 0.7303,
+      "loss": 0.8959,
       "step": 9201
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5514501555852356e-05,
-      "loss": 0.8202,
+      "loss": 0.9422,
       "step": 9202
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5513518936155886e-05,
-      "loss": 0.8289,
+      "loss": 0.8774,
       "step": 9203
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5512536239967168e-05,
-      "loss": 0.7172,
+      "loss": 0.8921,
       "step": 9204
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5511553467299827e-05,
-      "loss": 0.7957,
+      "loss": 0.9218,
       "step": 9205
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5510570618167503e-05,
-      "loss": 0.7044,
+      "loss": 0.8941,
       "step": 9206
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.550958769258383e-05,
-      "loss": 0.8593,
+      "loss": 0.8944,
       "step": 9207
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.550860469056244e-05,
-      "loss": 0.6793,
+      "loss": 0.869,
       "step": 9208
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5507621612116987e-05,
-      "loss": 0.742,
+      "loss": 0.8526,
       "step": 9209
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.550663845726109e-05,
-      "loss": 0.7757,
+      "loss": 0.8588,
       "step": 9210
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.5505655226008397e-05,
-      "loss": 0.7555,
+      "loss": 0.9337,
       "step": 9211
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.550467191837255e-05,
-      "loss": 0.7409,
+      "loss": 0.8797,
       "step": 9212
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.550368853436719e-05,
-      "loss": 0.8336,
+      "loss": 0.8958,
       "step": 9213
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5502705074005963e-05,
-      "loss": 0.8904,
+      "loss": 0.9388,
       "step": 9214
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.550172153730251e-05,
-      "loss": 0.7269,
+      "loss": 0.8797,
       "step": 9215
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5500737924270473e-05,
-      "loss": 0.7933,
+      "loss": 0.9383,
       "step": 9216
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5499754234923505e-05,
-      "loss": 0.6531,
+      "loss": 0.8434,
       "step": 9217
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.549877046927525e-05,
-      "loss": 0.8254,
+      "loss": 0.912,
       "step": 9218
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5497786627339358e-05,
-      "loss": 0.7863,
+      "loss": 0.8861,
       "step": 9219
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5496802709129474e-05,
-      "loss": 0.9573,
+      "loss": 1.01,
       "step": 9220
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5495818714659252e-05,
-      "loss": 0.7706,
+      "loss": 0.9216,
       "step": 9221
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5494834643942343e-05,
-      "loss": 0.836,
+      "loss": 0.9706,
       "step": 9222
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5493850496992398e-05,
-      "loss": 0.8574,
+      "loss": 0.913,
       "step": 9223
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5492866273823074e-05,
-      "loss": 0.6463,
+      "loss": 0.862,
       "step": 9224
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5491881974448025e-05,
-      "loss": 0.7743,
+      "loss": 0.9484,
       "step": 9225
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5490897598880903e-05,
-      "loss": 0.8365,
+      "loss": 0.9678,
       "step": 9226
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5489913147135367e-05,
-      "loss": 0.6861,
+      "loss": 0.7889,
       "step": 9227
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.548892861922508e-05,
-      "loss": 0.85,
+      "loss": 0.9225,
       "step": 9228
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.548794401516369e-05,
-      "loss": 0.8915,
+      "loss": 0.9395,
       "step": 9229
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5486959334964868e-05,
-      "loss": 0.7918,
+      "loss": 0.9255,
       "step": 9230
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5485974578642266e-05,
-      "loss": 0.8396,
+      "loss": 0.9151,
       "step": 9231
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5484989746209557e-05,
-      "loss": 0.8064,
+      "loss": 0.9099,
       "step": 9232
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.548400483768039e-05,
-      "loss": 0.9468,
+      "loss": 0.9824,
       "step": 9233
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.548301985306844e-05,
-      "loss": 0.7883,
+      "loss": 0.882,
       "step": 9234
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5482034792387367e-05,
-      "loss": 0.7499,
+      "loss": 0.8672,
       "step": 9235
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.548104965565084e-05,
-      "loss": 0.7773,
+      "loss": 0.9534,
       "step": 9236
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5480064442872528e-05,
-      "loss": 0.8185,
+      "loss": 0.8376,
       "step": 9237
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5479079154066096e-05,
-      "loss": 0.6112,
+      "loss": 0.7555,
       "step": 9238
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5478093789245213e-05,
-      "loss": 0.6405,
+      "loss": 0.8644,
       "step": 9239
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5477108348423554e-05,
-      "loss": 0.7964,
+      "loss": 0.9322,
       "step": 9240
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5476122831614785e-05,
-      "loss": 0.8982,
+      "loss": 1.0671,
       "step": 9241
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5475137238832583e-05,
-      "loss": 0.7888,
+      "loss": 0.8943,
       "step": 9242
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5474151570090617e-05,
-      "loss": 0.6953,
+      "loss": 0.8558,
       "step": 9243
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5473165825402565e-05,
-      "loss": 0.7001,
+      "loss": 0.8466,
       "step": 9244
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5472180004782107e-05,
-      "loss": 0.6765,
+      "loss": 0.835,
       "step": 9245
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.547119410824291e-05,
-      "loss": 0.7476,
+      "loss": 0.8898,
       "step": 9246
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5470208135798665e-05,
-      "loss": 0.9728,
+      "loss": 0.9843,
       "step": 9247
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5469222087463038e-05,
-      "loss": 0.831,
+      "loss": 0.9864,
       "step": 9248
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5468235963249714e-05,
-      "loss": 0.7201,
+      "loss": 0.7849,
       "step": 9249
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5467249763172373e-05,
-      "loss": 0.8199,
+      "loss": 0.9279,
       "step": 9250
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5466263487244702e-05,
-      "loss": 0.7983,
+      "loss": 0.976,
       "step": 9251
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.546527713548038e-05,
-      "loss": 0.7599,
+      "loss": 0.9138,
       "step": 9252
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.546429070789309e-05,
-      "loss": 0.6944,
+      "loss": 0.832,
       "step": 9253
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5463304204496524e-05,
-      "loss": 0.8648,
+      "loss": 0.9527,
       "step": 9254
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.546231762530436e-05,
-      "loss": 0.8547,
+      "loss": 0.9522,
       "step": 9255
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5461330970330284e-05,
-      "loss": 0.809,
+      "loss": 0.8714,
       "step": 9256
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5460344239587998e-05,
-      "loss": 0.7466,
+      "loss": 0.9147,
       "step": 9257
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.545935743309118e-05,
-      "loss": 0.8833,
+      "loss": 0.9982,
       "step": 9258
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5458370550853524e-05,
-      "loss": 0.7186,
+      "loss": 0.7999,
       "step": 9259
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5457383592888716e-05,
-      "loss": 0.7334,
+      "loss": 0.9087,
       "step": 9260
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.545639655921046e-05,
-      "loss": 0.8116,
+      "loss": 0.8511,
       "step": 9261
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.545540944983244e-05,
-      "loss": 0.8754,
+      "loss": 0.9434,
       "step": 9262
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5454422264768356e-05,
-      "loss": 0.8278,
+      "loss": 0.8963,
       "step": 9263
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.54534350040319e-05,
-      "loss": 0.8631,
+      "loss": 0.9795,
       "step": 9264
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5452447667636773e-05,
-      "loss": 0.7042,
+      "loss": 0.8724,
       "step": 9265
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.545146025559667e-05,
-      "loss": 0.7288,
+      "loss": 0.9192,
       "step": 9266
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5450472767925292e-05,
-      "loss": 0.7457,
+      "loss": 0.9029,
       "step": 9267
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5449485204636334e-05,
-      "loss": 0.7317,
+      "loss": 0.8513,
       "step": 9268
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5448497565743502e-05,
-      "loss": 0.8563,
+      "loss": 0.9543,
       "step": 9269
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5447509851260498e-05,
-      "loss": 0.8386,
+      "loss": 0.8657,
       "step": 9270
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5446522061201023e-05,
-      "loss": 0.8373,
+      "loss": 0.9704,
       "step": 9271
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.544553419557878e-05,
-      "loss": 0.6508,
+      "loss": 0.8691,
       "step": 9272
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5444546254407477e-05,
-      "loss": 0.812,
+      "loss": 0.9027,
       "step": 9273
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.544355823770082e-05,
-      "loss": 0.934,
+      "loss": 0.9792,
       "step": 9274
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5442570145472516e-05,
-      "loss": 0.7715,
+      "loss": 0.8468,
       "step": 9275
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5441581977736276e-05,
-      "loss": 0.7714,
+      "loss": 0.8694,
       "step": 9276
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5440593734505803e-05,
-      "loss": 0.8575,
+      "loss": 0.9626,
       "step": 9277
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.543960541579481e-05,
-      "loss": 0.7166,
+      "loss": 0.8947,
       "step": 9278
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5438617021617014e-05,
-      "loss": 0.8551,
+      "loss": 0.9348,
       "step": 9279
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5437628551986123e-05,
-      "loss": 0.804,
+      "loss": 0.9141,
       "step": 9280
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5436640006915848e-05,
-      "loss": 0.9128,
+      "loss": 0.9377,
       "step": 9281
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5435651386419905e-05,
-      "loss": 0.9487,
+      "loss": 0.9725,
       "step": 9282
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5434662690512015e-05,
-      "loss": 0.8536,
+      "loss": 1.0001,
       "step": 9283
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.543367391920589e-05,
-      "loss": 0.7804,
+      "loss": 0.8319,
       "step": 9284
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5432685072515247e-05,
-      "loss": 0.8134,
+      "loss": 0.9045,
       "step": 9285
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.543169615045381e-05,
-      "loss": 0.7802,
+      "loss": 0.8293,
       "step": 9286
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5430707153035293e-05,
-      "loss": 0.7576,
+      "loss": 0.9157,
       "step": 9287
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5429718080273422e-05,
-      "loss": 0.7551,
+      "loss": 0.9258,
       "step": 9288
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.542872893218191e-05,
-      "loss": 0.8674,
+      "loss": 0.9055,
       "step": 9289
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.542773970877449e-05,
-      "loss": 0.7343,
+      "loss": 0.9065,
       "step": 9290
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5426750410064885e-05,
-      "loss": 0.7869,
+      "loss": 0.9321,
       "step": 9291
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5425761036066816e-05,
-      "loss": 0.7121,
+      "loss": 0.9219,
       "step": 9292
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5424771586794013e-05,
-      "loss": 0.787,
+      "loss": 0.8982,
       "step": 9293
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.54237820622602e-05,
-      "loss": 0.8213,
+      "loss": 0.9518,
       "step": 9294
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.54227924624791e-05,
-      "loss": 0.8056,
+      "loss": 0.9288,
       "step": 9295
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.542180278746445e-05,
-      "loss": 0.8185,
+      "loss": 0.9358,
       "step": 9296
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5420813037229983e-05,
-      "loss": 0.7436,
+      "loss": 0.9487,
       "step": 9297
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.541982321178943e-05,
-      "loss": 0.8172,
+      "loss": 0.9218,
       "step": 9298
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5418833311156512e-05,
-      "loss": 0.8054,
+      "loss": 0.9088,
       "step": 9299
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5417843335344973e-05,
-      "loss": 0.7616,
+      "loss": 0.8481,
       "step": 9300
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5416853284368547e-05,
-      "loss": 0.772,
+      "loss": 0.9336,
       "step": 9301
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5415863158240965e-05,
-      "loss": 0.828,
+      "loss": 0.8909,
       "step": 9302
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5414872956975964e-05,
-      "loss": 0.7583,
+      "loss": 0.8708,
       "step": 9303
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5413882680587286e-05,
-      "loss": 0.7794,
+      "loss": 0.8781,
       "step": 9304
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.541289232908867e-05,
-      "loss": 0.807,
+      "loss": 0.8782,
       "step": 9305
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5411901902493846e-05,
-      "loss": 0.8424,
+      "loss": 0.8612,
       "step": 9306
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5410911400816565e-05,
-      "loss": 0.8967,
+      "loss": 0.8659,
       "step": 9307
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5409920824070566e-05,
-      "loss": 0.7248,
+      "loss": 0.8475,
       "step": 9308
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.540893017226959e-05,
-      "loss": 0.7731,
+      "loss": 0.9345,
       "step": 9309
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.540793944542738e-05,
-      "loss": 0.7868,
+      "loss": 0.9075,
       "step": 9310
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5406948643557688e-05,
-      "loss": 0.8577,
+      "loss": 0.9424,
       "step": 9311
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.540595776667425e-05,
-      "loss": 0.946,
+      "loss": 1.0115,
       "step": 9312
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5404966814790824e-05,
-      "loss": 0.7918,
+      "loss": 0.8583,
       "step": 9313
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.540397578792115e-05,
-      "loss": 0.7588,
+      "loss": 0.7938,
       "step": 9314
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5402984686078975e-05,
-      "loss": 0.6955,
+      "loss": 0.7591,
       "step": 9315
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5401993509278055e-05,
-      "loss": 0.5795,
+      "loss": 0.8206,
       "step": 9316
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.540100225753214e-05,
-      "loss": 0.9996,
+      "loss": 1.0134,
       "step": 9317
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5400010930854984e-05,
-      "loss": 0.7927,
+      "loss": 0.9055,
       "step": 9318
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5399019529260336e-05,
-      "loss": 0.7348,
+      "loss": 0.9045,
       "step": 9319
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.539802805276195e-05,
-      "loss": 0.7135,
+      "loss": 0.8797,
       "step": 9320
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5397036501373588e-05,
-      "loss": 0.783,
+      "loss": 0.8628,
       "step": 9321
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5396044875109e-05,
-      "loss": 0.8018,
+      "loss": 0.8767,
       "step": 9322
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5395053173981945e-05,
-      "loss": 0.871,
+      "loss": 0.9751,
       "step": 9323
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5394061398006185e-05,
-      "loss": 0.7781,
+      "loss": 0.9149,
       "step": 9324
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5393069547195473e-05,
-      "loss": 0.7261,
+      "loss": 0.8868,
       "step": 9325
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5392077621563573e-05,
-      "loss": 0.7212,
+      "loss": 0.8343,
       "step": 9326
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5391085621124244e-05,
-      "loss": 0.7848,
+      "loss": 0.8303,
       "step": 9327
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5390093545891255e-05,
-      "loss": 0.8793,
+      "loss": 0.9439,
       "step": 9328
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5389101395878365e-05,
-      "loss": 0.9734,
+      "loss": 0.929,
       "step": 9329
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5388109171099338e-05,
-      "loss": 0.6428,
+      "loss": 0.823,
       "step": 9330
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.538711687156794e-05,
-      "loss": 0.7864,
+      "loss": 0.8785,
       "step": 9331
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.538612449729794e-05,
-      "loss": 0.9481,
+      "loss": 0.9911,
       "step": 9332
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5385132048303104e-05,
-      "loss": 0.7616,
+      "loss": 0.9484,
       "step": 9333
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5384139524597205e-05,
-      "loss": 0.8681,
+      "loss": 0.9014,
       "step": 9334
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5383146926194005e-05,
-      "loss": 0.6694,
+      "loss": 0.8707,
       "step": 9335
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.538215425310728e-05,
-      "loss": 0.7483,
+      "loss": 0.9684,
       "step": 9336
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.53811615053508e-05,
-      "loss": 0.8388,
+      "loss": 0.9152,
       "step": 9337
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.538016868293834e-05,
-      "loss": 0.8426,
+      "loss": 0.943,
       "step": 9338
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5379175785883672e-05,
-      "loss": 0.8283,
+      "loss": 0.945,
       "step": 9339
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5378182814200577e-05,
-      "loss": 0.6637,
+      "loss": 0.8394,
       "step": 9340
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.537718976790282e-05,
-      "loss": 0.9642,
+      "loss": 1.0505,
       "step": 9341
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5376196647004188e-05,
-      "loss": 0.6748,
+      "loss": 0.831,
       "step": 9342
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.537520345151845e-05,
-      "loss": 0.9553,
+      "loss": 0.9475,
       "step": 9343
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5374210181459397e-05,
-      "loss": 0.8808,
+      "loss": 1.0202,
       "step": 9344
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.53732168368408e-05,
-      "loss": 0.7496,
+      "loss": 0.9571,
       "step": 9345
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5372223417676444e-05,
-      "loss": 0.8003,
+      "loss": 0.9329,
       "step": 9346
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5371229923980113e-05,
-      "loss": 0.7158,
+      "loss": 0.8574,
       "step": 9347
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5370236355765584e-05,
-      "loss": 0.7471,
+      "loss": 0.909,
       "step": 9348
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5369242713046643e-05,
-      "loss": 0.7635,
+      "loss": 0.8657,
       "step": 9349
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5368248995837077e-05,
-      "loss": 0.699,
+      "loss": 0.8707,
       "step": 9350
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5367255204150673e-05,
-      "loss": 0.7473,
+      "loss": 0.8458,
       "step": 9351
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5366261338001222e-05,
-      "loss": 0.8773,
+      "loss": 0.9171,
       "step": 9352
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.536526739740251e-05,
-      "loss": 0.9291,
+      "loss": 1.0119,
       "step": 9353
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.536427338236832e-05,
-      "loss": 0.8358,
+      "loss": 0.8981,
       "step": 9354
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5363279292912448e-05,
-      "loss": 0.8398,
+      "loss": 0.9103,
       "step": 9355
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5362285129048687e-05,
-      "loss": 0.7153,
+      "loss": 0.8363,
       "step": 9356
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5361290890790826e-05,
-      "loss": 0.7377,
+      "loss": 0.8104,
       "step": 9357
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5360296578152664e-05,
-      "loss": 0.7372,
+      "loss": 0.84,
       "step": 9358
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5359302191147993e-05,
-      "loss": 0.6877,
+      "loss": 0.8386,
       "step": 9359
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.53583077297906e-05,
-      "loss": 0.8193,
+      "loss": 0.9328,
       "step": 9360
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5357313194094298e-05,
-      "loss": 0.6837,
+      "loss": 0.8248,
       "step": 9361
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5356318584072877e-05,
-      "loss": 0.6432,
+      "loss": 0.8468,
       "step": 9362
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.535532389974013e-05,
-      "loss": 0.9417,
+      "loss": 0.9164,
       "step": 9363
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5354329141109867e-05,
-      "loss": 0.7066,
+      "loss": 0.8961,
       "step": 9364
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.535333430819588e-05,
-      "loss": 0.6999,
+      "loss": 0.8008,
       "step": 9365
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5352339401011976e-05,
-      "loss": 0.6704,
+      "loss": 0.7977,
       "step": 9366
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5351344419571957e-05,
-      "loss": 0.8157,
+      "loss": 0.881,
       "step": 9367
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5350349363889625e-05,
-      "loss": 0.7514,
+      "loss": 0.8604,
       "step": 9368
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.534935423397879e-05,
-      "loss": 0.9579,
+      "loss": 0.9427,
       "step": 9369
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.534835902985325e-05,
-      "loss": 0.7554,
+      "loss": 0.9097,
       "step": 9370
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.534736375152682e-05,
-      "loss": 0.7533,
+      "loss": 0.8571,
       "step": 9371
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5346368399013306e-05,
-      "loss": 0.7858,
+      "loss": 0.815,
       "step": 9372
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5345372972326512e-05,
-      "loss": 0.7165,
+      "loss": 0.9493,
       "step": 9373
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.534437747148026e-05,
-      "loss": 0.7781,
+      "loss": 0.9448,
       "step": 9374
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5343381896488346e-05,
-      "loss": 0.8227,
+      "loss": 0.853,
       "step": 9375
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.534238624736459e-05,
-      "loss": 0.8153,
+      "loss": 0.9146,
       "step": 9376
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5341390524122806e-05,
-      "loss": 0.8212,
+      "loss": 0.8459,
       "step": 9377
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5340394726776807e-05,
-      "loss": 0.8248,
+      "loss": 0.9364,
       "step": 9378
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.533939885534041e-05,
-      "loss": 0.8322,
+      "loss": 0.8319,
       "step": 9379
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5338402909827427e-05,
-      "loss": 0.7189,
+      "loss": 0.8901,
       "step": 9380
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.533740689025168e-05,
-      "loss": 0.7653,
+      "loss": 0.8576,
       "step": 9381
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5336410796626987e-05,
-      "loss": 0.7515,
+      "loss": 0.8826,
       "step": 9382
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5335414628967158e-05,
-      "loss": 0.9781,
+      "loss": 1.0405,
       "step": 9383
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5334418387286024e-05,
-      "loss": 0.8657,
+      "loss": 0.947,
       "step": 9384
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.533342207159741e-05,
-      "loss": 0.8845,
+      "loss": 0.9437,
       "step": 9385
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.533242568191512e-05,
-      "loss": 0.7936,
+      "loss": 0.8781,
       "step": 9386
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5331429218253e-05,
-      "loss": 0.6341,
+      "loss": 0.8487,
       "step": 9387
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.533043268062486e-05,
-      "loss": 0.7593,
+      "loss": 0.9472,
       "step": 9388
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.532943606904453e-05,
-      "loss": 0.8064,
+      "loss": 0.8843,
       "step": 9389
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5328439383525836e-05,
-      "loss": 0.8738,
+      "loss": 0.934,
       "step": 9390
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5327442624082603e-05,
-      "loss": 0.796,
+      "loss": 0.9149,
       "step": 9391
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5326445790728664e-05,
-      "loss": 0.6138,
+      "loss": 0.7967,
       "step": 9392
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.532544888347785e-05,
-      "loss": 0.8164,
+      "loss": 0.9204,
       "step": 9393
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5324451902343983e-05,
-      "loss": 0.7185,
+      "loss": 0.8683,
       "step": 9394
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5323454847340905e-05,
-      "loss": 0.8297,
+      "loss": 0.9868,
       "step": 9395
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5322457718482443e-05,
-      "loss": 0.8734,
+      "loss": 0.8817,
       "step": 9396
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5321460515782427e-05,
-      "loss": 0.7947,
+      "loss": 0.9192,
       "step": 9397
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.53204632392547e-05,
-      "loss": 0.8264,
+      "loss": 0.9308,
       "step": 9398
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5319465888913098e-05,
-      "loss": 0.7421,
+      "loss": 0.8587,
       "step": 9399
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.531846846477145e-05,
-      "loss": 0.7974,
+      "loss": 0.9672,
       "step": 9400
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.53174709668436e-05,
-      "loss": 0.793,
+      "loss": 0.8817,
       "step": 9401
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.531647339514338e-05,
-      "loss": 0.8959,
+      "loss": 1.0277,
       "step": 9402
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.531547574968464e-05,
-      "loss": 0.7805,
+      "loss": 0.8853,
       "step": 9403
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5314478030481216e-05,
-      "loss": 0.8602,
+      "loss": 0.9614,
       "step": 9404
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5313480237546945e-05,
-      "loss": 0.7966,
+      "loss": 0.8851,
       "step": 9405
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5312482370895677e-05,
-      "loss": 0.8781,
+      "loss": 0.9047,
       "step": 9406
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5311484430541254e-05,
-      "loss": 0.769,
+      "loss": 0.8781,
       "step": 9407
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.531048641649752e-05,
-      "loss": 0.709,
+      "loss": 0.8205,
       "step": 9408
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5309488328778322e-05,
-      "loss": 0.7426,
+      "loss": 0.8371,
       "step": 9409
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5308490167397505e-05,
-      "loss": 0.7968,
+      "loss": 0.9156,
       "step": 9410
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5307491932368922e-05,
-      "loss": 0.6775,
+      "loss": 0.8471,
       "step": 9411
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5306493623706416e-05,
-      "loss": 0.9519,
+      "loss": 1.0321,
       "step": 9412
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.530549524142384e-05,
-      "loss": 0.8473,
+      "loss": 0.9597,
       "step": 9413
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5304496785535046e-05,
-      "loss": 0.8076,
+      "loss": 0.9183,
       "step": 9414
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5303498256053884e-05,
-      "loss": 0.7675,
+      "loss": 0.8848,
       "step": 9415
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5302499652994208e-05,
-      "loss": 0.7447,
+      "loss": 0.8903,
       "step": 9416
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5301500976369873e-05,
-      "loss": 0.8652,
+      "loss": 0.9332,
       "step": 9417
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5300502226194737e-05,
-      "loss": 0.7441,
+      "loss": 0.8567,
       "step": 9418
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5299503402482653e-05,
-      "loss": 0.7254,
+      "loss": 0.8454,
       "step": 9419
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5298504505247476e-05,
-      "loss": 0.7033,
+      "loss": 0.9034,
       "step": 9420
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5297505534503066e-05,
-      "loss": 0.8357,
+      "loss": 0.8801,
       "step": 9421
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5296506490263285e-05,
-      "loss": 0.8105,
+      "loss": 0.8176,
       "step": 9422
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5295507372541985e-05,
-      "loss": 0.7789,
+      "loss": 0.8613,
       "step": 9423
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.529450818135304e-05,
-      "loss": 0.7698,
+      "loss": 0.9001,
       "step": 9424
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5293508916710305e-05,
-      "loss": 0.8738,
+      "loss": 0.9237,
       "step": 9425
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.529250957862764e-05,
-      "loss": 0.852,
+      "loss": 0.896,
       "step": 9426
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5291510167118918e-05,
-      "loss": 0.8084,
+      "loss": 0.9296,
       "step": 9427
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5290510682198e-05,
-      "loss": 0.8731,
+      "loss": 0.9236,
       "step": 9428
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528951112387875e-05,
-      "loss": 0.7338,
+      "loss": 0.8496,
       "step": 9429
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528851149217504e-05,
-      "loss": 0.7677,
+      "loss": 0.8312,
       "step": 9430
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528751178710074e-05,
-      "loss": 0.8363,
+      "loss": 0.9049,
       "step": 9431
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528651200866971e-05,
-      "loss": 0.776,
+      "loss": 0.8433,
       "step": 9432
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528551215689583e-05,
-      "loss": 0.588,
+      "loss": 0.8066,
       "step": 9433
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5284512231792963e-05,
-      "loss": 0.7584,
+      "loss": 0.8436,
       "step": 9434
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5283512233374992e-05,
-      "loss": 0.737,
+      "loss": 0.8901,
       "step": 9435
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5282512161655785e-05,
-      "loss": 0.7764,
+      "loss": 0.9231,
       "step": 9436
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.528151201664921e-05,
-      "loss": 0.7583,
+      "loss": 0.8701,
       "step": 9437
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5280511798369157e-05,
-      "loss": 0.7652,
+      "loss": 0.8363,
       "step": 9438
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.527951150682949e-05,
-      "loss": 0.8304,
+      "loss": 0.8196,
       "step": 9439
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5278511142044098e-05,
-      "loss": 0.7801,
+      "loss": 0.8709,
       "step": 9440
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5277510704026848e-05,
-      "loss": 1.0199,
+      "loss": 1.0518,
       "step": 9441
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5276510192791623e-05,
-      "loss": 0.7489,
+      "loss": 0.8485,
       "step": 9442
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5275509608352308e-05,
-      "loss": 0.7891,
+      "loss": 0.8552,
       "step": 9443
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5274508950722783e-05,
-      "loss": 0.6546,
+      "loss": 0.8508,
       "step": 9444
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5273508219916927e-05,
-      "loss": 0.563,
+      "loss": 0.8242,
       "step": 9445
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.527250741594863e-05,
-      "loss": 0.8449,
+      "loss": 0.9601,
       "step": 9446
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5271506538831774e-05,
-      "loss": 0.8617,
+      "loss": 0.9554,
       "step": 9447
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.527050558858024e-05,
-      "loss": 0.8301,
+      "loss": 0.9339,
       "step": 9448
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5269504565207922e-05,
-      "loss": 0.6773,
+      "loss": 0.8727,
       "step": 9449
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.52685034687287e-05,
-      "loss": 0.7607,
+      "loss": 0.8934,
       "step": 9450
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5267502299156473e-05,
-      "loss": 0.6535,
+      "loss": 0.7955,
       "step": 9451
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.526650105650512e-05,
-      "loss": 0.7658,
+      "loss": 0.9733,
       "step": 9452
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.526549974078854e-05,
-      "loss": 0.6061,
+      "loss": 0.7515,
       "step": 9453
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5264498352020622e-05,
-      "loss": 0.8514,
+      "loss": 1.0035,
       "step": 9454
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.526349689021526e-05,
-      "loss": 0.8548,
+      "loss": 0.9256,
       "step": 9455
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5262495355386345e-05,
-      "loss": 0.6945,
+      "loss": 0.8654,
       "step": 9456
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5261493747547776e-05,
-      "loss": 0.7675,
+      "loss": 0.8706,
       "step": 9457
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5260492066713442e-05,
-      "loss": 0.9043,
+      "loss": 0.9514,
       "step": 9458
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5259490312897245e-05,
-      "loss": 0.7657,
+      "loss": 0.9337,
       "step": 9459
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5258488486113086e-05,
-      "loss": 0.7229,
+      "loss": 0.9608,
       "step": 9460
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5257486586374857e-05,
-      "loss": 0.8046,
+      "loss": 0.9567,
       "step": 9461
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5256484613696461e-05,
-      "loss": 0.6703,
+      "loss": 0.8688,
       "step": 9462
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.52554825680918e-05,
-      "loss": 0.7606,
+      "loss": 0.8544,
       "step": 9463
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5254480449574773e-05,
-      "loss": 0.7967,
+      "loss": 0.9105,
       "step": 9464
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5253478258159287e-05,
-      "loss": 0.8715,
+      "loss": 0.9636,
       "step": 9465
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.525247599385924e-05,
-      "loss": 0.9092,
+      "loss": 0.9633,
       "step": 9466
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5251473656688544e-05,
-      "loss": 0.7269,
+      "loss": 0.8435,
       "step": 9467
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5250471246661103e-05,
-      "loss": 0.8064,
+      "loss": 0.9236,
       "step": 9468
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.524946876379082e-05,
-      "loss": 0.735,
+      "loss": 0.7796,
       "step": 9469
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5248466208091607e-05,
-      "loss": 0.8866,
+      "loss": 0.9483,
       "step": 9470
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5247463579577373e-05,
-      "loss": 0.7134,
+      "loss": 0.8413,
       "step": 9471
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5246460878262026e-05,
-      "loss": 0.67,
+      "loss": 0.8412,
       "step": 9472
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.524545810415948e-05,
-      "loss": 0.9198,
+      "loss": 0.9708,
       "step": 9473
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5244455257283644e-05,
-      "loss": 0.8094,
+      "loss": 0.8958,
       "step": 9474
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.524345233764843e-05,
-      "loss": 0.6194,
+      "loss": 0.7717,
       "step": 9475
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5242449345267759e-05,
-      "loss": 0.7802,
+      "loss": 0.8555,
       "step": 9476
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5241446280155536e-05,
-      "loss": 0.9003,
+      "loss": 0.8878,
       "step": 9477
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5240443142325687e-05,
-      "loss": 0.7214,
+      "loss": 0.8704,
       "step": 9478
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5239439931792124e-05,
-      "loss": 0.725,
+      "loss": 0.8163,
       "step": 9479
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5238436648568762e-05,
-      "loss": 0.7635,
+      "loss": 0.9042,
       "step": 9480
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5237433292669527e-05,
-      "loss": 0.7189,
+      "loss": 0.8155,
       "step": 9481
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5236429864108335e-05,
-      "loss": 0.8248,
+      "loss": 0.9042,
       "step": 9482
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.523542636289911e-05,
-      "loss": 0.6779,
+      "loss": 0.7664,
       "step": 9483
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5234422789055771e-05,
-      "loss": 0.7711,
+      "loss": 0.8911,
       "step": 9484
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5233419142592241e-05,
-      "loss": 0.8062,
+      "loss": 0.956,
       "step": 9485
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.5232415423522446e-05,
-      "loss": 0.9689,
+      "loss": 1.0219,
       "step": 9486
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5231411631860313e-05,
-      "loss": 0.809,
+      "loss": 0.9028,
       "step": 9487
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5230407767619764e-05,
-      "loss": 0.7685,
+      "loss": 0.9073,
       "step": 9488
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5229403830814731e-05,
-      "loss": 0.6933,
+      "loss": 0.8308,
       "step": 9489
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5228399821459137e-05,
-      "loss": 0.6599,
+      "loss": 0.851,
       "step": 9490
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5227395739566913e-05,
-      "loss": 0.9153,
+      "loss": 0.9695,
       "step": 9491
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5226391585151992e-05,
-      "loss": 0.8152,
+      "loss": 0.8475,
       "step": 9492
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5225387358228298e-05,
-      "loss": 0.8197,
+      "loss": 0.917,
       "step": 9493
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5224383058809773e-05,
-      "loss": 0.8833,
+      "loss": 0.9555,
       "step": 9494
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5223378686910346e-05,
-      "loss": 0.8431,
+      "loss": 0.9221,
       "step": 9495
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.522237424254395e-05,
-      "loss": 0.7042,
+      "loss": 0.8598,
       "step": 9496
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5221369725724522e-05,
-      "loss": 0.8094,
+      "loss": 0.8932,
       "step": 9497
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5220365136465993e-05,
-      "loss": 0.7826,
+      "loss": 0.9076,
       "step": 9498
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5219360474782307e-05,
-      "loss": 0.8076,
+      "loss": 0.9302,
       "step": 9499
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.52183557406874e-05,
-      "loss": 0.7742,
+      "loss": 0.9151,
       "step": 9500
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5217350934195212e-05,
-      "loss": 0.6903,
+      "loss": 0.812,
       "step": 9501
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5216346055319679e-05,
-      "loss": 0.6679,
+      "loss": 0.8872,
       "step": 9502
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5215341104074747e-05,
-      "loss": 0.6354,
+      "loss": 0.7972,
       "step": 9503
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5214336080474357e-05,
-      "loss": 0.7431,
+      "loss": 0.8379,
       "step": 9504
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5213330984532455e-05,
-      "loss": 0.8978,
+      "loss": 0.972,
       "step": 9505
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5212325816262978e-05,
-      "loss": 0.8274,
+      "loss": 0.9068,
       "step": 9506
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5211320575679875e-05,
-      "loss": 0.7605,
+      "loss": 0.9453,
       "step": 9507
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5210315262797097e-05,
-      "loss": 0.7846,
+      "loss": 0.8212,
       "step": 9508
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5209309877628584e-05,
-      "loss": 0.8073,
+      "loss": 0.9027,
       "step": 9509
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5208304420188289e-05,
-      "loss": 0.684,
+      "loss": 0.8366,
       "step": 9510
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5207298890490161e-05,
-      "loss": 0.8557,
+      "loss": 0.9459,
       "step": 9511
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5206293288548146e-05,
-      "loss": 0.7704,
+      "loss": 0.9005,
       "step": 9512
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5205287614376201e-05,
-      "loss": 0.913,
+      "loss": 1.0645,
       "step": 9513
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5204281867988276e-05,
-      "loss": 0.7227,
+      "loss": 0.9159,
       "step": 9514
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5203276049398321e-05,
-      "loss": 0.7031,
+      "loss": 0.9212,
       "step": 9515
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5202270158620299e-05,
-      "loss": 0.7232,
+      "loss": 0.9457,
       "step": 9516
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5201264195668153e-05,
-      "loss": 0.698,
+      "loss": 0.8403,
       "step": 9517
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.520025816055585e-05,
-      "loss": 0.737,
+      "loss": 0.8216,
       "step": 9518
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.519925205329734e-05,
-      "loss": 0.6862,
+      "loss": 0.8676,
       "step": 9519
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5198245873906588e-05,
-      "loss": 0.7693,
+      "loss": 0.8716,
       "step": 9520
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5197239622397545e-05,
-      "loss": 0.8329,
+      "loss": 0.9415,
       "step": 9521
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.519623329878418e-05,
-      "loss": 0.714,
+      "loss": 0.781,
       "step": 9522
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.519522690308045e-05,
-      "loss": 0.7793,
+      "loss": 0.8996,
       "step": 9523
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5194220435300317e-05,
-      "loss": 0.587,
+      "loss": 0.794,
       "step": 9524
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5193213895457746e-05,
-      "loss": 0.6842,
+      "loss": 0.8586,
       "step": 9525
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5192207283566701e-05,
-      "loss": 0.8277,
+      "loss": 0.8993,
       "step": 9526
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5191200599641147e-05,
-      "loss": 0.9238,
+      "loss": 1.0288,
       "step": 9527
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5190193843695046e-05,
-      "loss": 0.837,
+      "loss": 0.8843,
       "step": 9528
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5189187015742373e-05,
-      "loss": 0.8511,
+      "loss": 0.9003,
       "step": 9529
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5188180115797093e-05,
-      "loss": 0.8325,
+      "loss": 0.9133,
       "step": 9530
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5187173143873173e-05,
-      "loss": 0.8538,
+      "loss": 0.955,
       "step": 9531
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5186166099984588e-05,
-      "loss": 0.7147,
+      "loss": 0.823,
       "step": 9532
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5185158984145301e-05,
-      "loss": 0.825,
+      "loss": 0.827,
       "step": 9533
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5184151796369295e-05,
-      "loss": 0.8009,
+      "loss": 0.9156,
       "step": 9534
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.518314453667054e-05,
-      "loss": 0.7229,
+      "loss": 0.8498,
       "step": 9535
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5182137205063003e-05,
-      "loss": 0.7076,
+      "loss": 0.9102,
       "step": 9536
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5181129801560665e-05,
-      "loss": 0.7602,
+      "loss": 0.924,
       "step": 9537
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5180122326177505e-05,
-      "loss": 0.6469,
+      "loss": 0.849,
       "step": 9538
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5179114778927497e-05,
-      "loss": 0.8766,
+      "loss": 0.9006,
       "step": 9539
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5178107159824618e-05,
-      "loss": 0.6853,
+      "loss": 0.8485,
       "step": 9540
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.517709946888285e-05,
-      "loss": 0.7693,
+      "loss": 0.917,
       "step": 9541
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5176091706116171e-05,
-      "loss": 0.8648,
+      "loss": 0.8773,
       "step": 9542
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5175083871538566e-05,
-      "loss": 0.8625,
+      "loss": 0.9693,
       "step": 9543
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5174075965164012e-05,
-      "loss": 0.8405,
+      "loss": 0.9228,
       "step": 9544
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5173067987006497e-05,
-      "loss": 0.7079,
+      "loss": 0.8855,
       "step": 9545
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5172059937080003e-05,
-      "loss": 0.8527,
+      "loss": 0.941,
       "step": 9546
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5171051815398513e-05,
-      "loss": 0.8912,
+      "loss": 0.9767,
       "step": 9547
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.517004362197602e-05,
-      "loss": 0.7809,
+      "loss": 0.8639,
       "step": 9548
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5169035356826504e-05,
-      "loss": 0.7389,
+      "loss": 0.9022,
       "step": 9549
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5168027019963958e-05,
-      "loss": 0.8372,
+      "loss": 0.9315,
       "step": 9550
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5167018611402369e-05,
-      "loss": 0.7621,
+      "loss": 0.8732,
       "step": 9551
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5166010131155728e-05,
-      "loss": 0.641,
+      "loss": 0.7722,
       "step": 9552
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5165001579238029e-05,
-      "loss": 0.7939,
+      "loss": 0.9499,
       "step": 9553
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5163992955663256e-05,
-      "loss": 0.8644,
+      "loss": 0.9951,
       "step": 9554
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5162984260445411e-05,
-      "loss": 0.6957,
+      "loss": 0.8601,
       "step": 9555
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5161975493598484e-05,
-      "loss": 0.8335,
+      "loss": 0.8935,
       "step": 9556
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5160966655136469e-05,
-      "loss": 0.6764,
+      "loss": 0.9044,
       "step": 9557
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5159957745073363e-05,
-      "loss": 0.7798,
+      "loss": 0.8525,
       "step": 9558
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.515894876342317e-05,
-      "loss": 0.871,
+      "loss": 0.9459,
       "step": 9559
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5157939710199876e-05,
-      "loss": 0.8717,
+      "loss": 0.9666,
       "step": 9560
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5156930585417492e-05,
-      "loss": 0.8658,
+      "loss": 0.8393,
       "step": 9561
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5155921389090011e-05,
-      "loss": 0.7115,
+      "loss": 0.7858,
       "step": 9562
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5154912121231435e-05,
-      "loss": 0.8341,
+      "loss": 0.9456,
       "step": 9563
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5153902781855769e-05,
-      "loss": 0.8177,
+      "loss": 0.9338,
       "step": 9564
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5152893370977012e-05,
-      "loss": 0.6592,
+      "loss": 0.8186,
       "step": 9565
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5151883888609171e-05,
-      "loss": 0.7173,
+      "loss": 0.8359,
       "step": 9566
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5150874334766252e-05,
-      "loss": 0.6635,
+      "loss": 0.8057,
       "step": 9567
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5149864709462259e-05,
-      "loss": 0.7827,
+      "loss": 0.8546,
       "step": 9568
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5148855012711198e-05,
-      "loss": 0.7972,
+      "loss": 0.9512,
       "step": 9569
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5147845244527084e-05,
-      "loss": 0.7856,
+      "loss": 0.8212,
       "step": 9570
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5146835404923917e-05,
-      "loss": 0.7865,
+      "loss": 0.8358,
       "step": 9571
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5145825493915713e-05,
-      "loss": 0.7706,
+      "loss": 0.876,
       "step": 9572
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.514481551151648e-05,
-      "loss": 0.7817,
+      "loss": 0.8746,
       "step": 9573
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5143805457740233e-05,
-      "loss": 0.7523,
+      "loss": 0.8802,
       "step": 9574
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5142795332600981e-05,
-      "loss": 0.8324,
+      "loss": 0.9003,
       "step": 9575
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5141785136112742e-05,
-      "loss": 0.7815,
+      "loss": 0.8852,
       "step": 9576
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.514077486828953e-05,
-      "loss": 0.7278,
+      "loss": 0.8322,
       "step": 9577
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5139764529145358e-05,
-      "loss": 0.8311,
+      "loss": 0.9758,
       "step": 9578
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5138754118694245e-05,
-      "loss": 0.8,
+      "loss": 0.9443,
       "step": 9579
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5137743636950215e-05,
-      "loss": 0.7847,
+      "loss": 0.9008,
       "step": 9580
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5136733083927278e-05,
-      "loss": 0.8563,
+      "loss": 0.9367,
       "step": 9581
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5135722459639457e-05,
-      "loss": 0.7586,
+      "loss": 0.7884,
       "step": 9582
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5134711764100774e-05,
-      "loss": 0.7242,
+      "loss": 0.8298,
       "step": 9583
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.513370099732525e-05,
-      "loss": 0.7297,
+      "loss": 0.8583,
       "step": 9584
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.513269015932691e-05,
-      "loss": 0.8026,
+      "loss": 0.8844,
       "step": 9585
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5131679250119775e-05,
-      "loss": 0.8464,
+      "loss": 0.94,
       "step": 9586
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.513066826971787e-05,
-      "loss": 0.6789,
+      "loss": 0.9151,
       "step": 9587
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5129657218135225e-05,
-      "loss": 0.7666,
+      "loss": 0.8266,
       "step": 9588
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5128646095385861e-05,
-      "loss": 0.8365,
+      "loss": 0.9282,
       "step": 9589
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5127634901483812e-05,
-      "loss": 0.7146,
+      "loss": 0.8593,
       "step": 9590
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5126623636443106e-05,
-      "loss": 0.7537,
+      "loss": 0.8538,
       "step": 9591
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5125612300277768e-05,
-      "loss": 0.7372,
+      "loss": 0.8468,
       "step": 9592
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5124600893001831e-05,
-      "loss": 0.7435,
+      "loss": 0.9014,
       "step": 9593
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5123589414629325e-05,
-      "loss": 0.7911,
+      "loss": 0.9171,
       "step": 9594
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5122577865174291e-05,
-      "loss": 0.7811,
+      "loss": 0.8847,
       "step": 9595
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5121566244650754e-05,
-      "loss": 0.8229,
+      "loss": 0.9156,
       "step": 9596
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5120554553072753e-05,
-      "loss": 0.8308,
+      "loss": 0.9139,
       "step": 9597
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.511954279045432e-05,
-      "loss": 0.7095,
+      "loss": 0.8652,
       "step": 9598
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5118530956809498e-05,
-      "loss": 0.8383,
+      "loss": 0.8919,
       "step": 9599
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5117519052152316e-05,
-      "loss": 0.7419,
+      "loss": 0.8932,
       "step": 9600
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5116507076496825e-05,
-      "loss": 0.7446,
+      "loss": 0.8755,
       "step": 9601
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5115495029857052e-05,
-      "loss": 0.8939,
+      "loss": 0.9207,
       "step": 9602
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5114482912247045e-05,
-      "loss": 0.6044,
+      "loss": 0.8008,
       "step": 9603
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5113470723680847e-05,
-      "loss": 0.685,
+      "loss": 0.7719,
       "step": 9604
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5112458464172495e-05,
-      "loss": 0.7867,
+      "loss": 0.8676,
       "step": 9605
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5111446133736035e-05,
-      "loss": 0.8375,
+      "loss": 0.9113,
       "step": 9606
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5110433732385514e-05,
-      "loss": 0.6888,
+      "loss": 0.8413,
       "step": 9607
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5109421260134974e-05,
-      "loss": 0.7851,
+      "loss": 0.9109,
       "step": 9608
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5108408716998465e-05,
-      "loss": 0.8204,
+      "loss": 0.9366,
       "step": 9609
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.510739610299003e-05,
-      "loss": 0.8085,
+      "loss": 0.9549,
       "step": 9610
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5106383418123725e-05,
-      "loss": 0.7511,
+      "loss": 0.8727,
       "step": 9611
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5105370662413594e-05,
-      "loss": 0.8649,
+      "loss": 0.8888,
       "step": 9612
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5104357835873686e-05,
-      "loss": 0.8579,
+      "loss": 0.9575,
       "step": 9613
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5103344938518056e-05,
-      "loss": 0.9007,
+      "loss": 0.9512,
       "step": 9614
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5102331970360757e-05,
-      "loss": 0.7606,
+      "loss": 0.9208,
       "step": 9615
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5101318931415838e-05,
-      "loss": 0.8667,
+      "loss": 0.9154,
       "step": 9616
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.510030582169736e-05,
-      "loss": 0.8426,
+      "loss": 0.9183,
       "step": 9617
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5099292641219373e-05,
-      "loss": 0.7526,
+      "loss": 0.8951,
       "step": 9618
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5098279389995938e-05,
-      "loss": 0.7824,
+      "loss": 0.8758,
       "step": 9619
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5097266068041107e-05,
-      "loss": 0.8367,
+      "loss": 0.9443,
       "step": 9620
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5096252675368942e-05,
-      "loss": 0.9222,
+      "loss": 0.9463,
       "step": 9621
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5095239211993504e-05,
-      "loss": 0.6717,
+      "loss": 0.7556,
       "step": 9622
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.509422567792885e-05,
-      "loss": 0.9093,
+      "loss": 0.9591,
       "step": 9623
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5093212073189043e-05,
-      "loss": 0.8245,
+      "loss": 0.9277,
       "step": 9624
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5092198397788145e-05,
-      "loss": 0.7455,
+      "loss": 0.8525,
       "step": 9625
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5091184651740216e-05,
-      "loss": 0.824,
+      "loss": 0.9082,
       "step": 9626
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5090170835059326e-05,
-      "loss": 0.7887,
+      "loss": 0.9113,
       "step": 9627
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5089156947759538e-05,
-      "loss": 0.6873,
+      "loss": 0.8615,
       "step": 9628
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5088142989854917e-05,
-      "loss": 0.6839,
+      "loss": 0.827,
       "step": 9629
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5087128961359532e-05,
-      "loss": 0.8629,
+      "loss": 0.9368,
       "step": 9630
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5086114862287451e-05,
-      "loss": 0.6744,
+      "loss": 0.8646,
       "step": 9631
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5085100692652738e-05,
-      "loss": 0.7009,
+      "loss": 0.8158,
       "step": 9632
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5084086452469473e-05,
-      "loss": 0.7761,
+      "loss": 0.8784,
       "step": 9633
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5083072141751722e-05,
-      "loss": 0.8283,
+      "loss": 1.0085,
       "step": 9634
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5082057760513553e-05,
-      "loss": 0.7992,
+      "loss": 0.906,
       "step": 9635
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5081043308769045e-05,
-      "loss": 0.7104,
+      "loss": 0.8482,
       "step": 9636
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.508002878653227e-05,
-      "loss": 0.959,
+      "loss": 1.0462,
       "step": 9637
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5079014193817302e-05,
-      "loss": 0.6518,
+      "loss": 0.762,
       "step": 9638
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.507799953063822e-05,
-      "loss": 0.8057,
+      "loss": 0.8596,
       "step": 9639
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5076984797009098e-05,
-      "loss": 0.7429,
+      "loss": 0.8593,
       "step": 9640
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5075969992944018e-05,
-      "loss": 0.6444,
+      "loss": 0.8214,
       "step": 9641
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5074955118457055e-05,
-      "loss": 0.7863,
+      "loss": 0.9052,
       "step": 9642
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5073940173562286e-05,
-      "loss": 0.8509,
+      "loss": 0.9192,
       "step": 9643
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5072925158273801e-05,
-      "loss": 0.8191,
+      "loss": 0.9223,
       "step": 9644
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5071910072605675e-05,
-      "loss": 0.7471,
+      "loss": 0.8387,
       "step": 9645
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5070894916571993e-05,
-      "loss": 0.7877,
+      "loss": 0.9007,
       "step": 9646
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5069879690186839e-05,
-      "loss": 0.8977,
+      "loss": 1.0331,
       "step": 9647
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5068864393464297e-05,
-      "loss": 0.739,
+      "loss": 0.8766,
       "step": 9648
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5067849026418455e-05,
-      "loss": 0.8069,
+      "loss": 0.8938,
       "step": 9649
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5066833589063396e-05,
-      "loss": 0.8151,
+      "loss": 0.9061,
       "step": 9650
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5065818081413212e-05,
-      "loss": 0.8206,
+      "loss": 0.8824,
       "step": 9651
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5064802503481988e-05,
-      "loss": 0.8405,
+      "loss": 0.9013,
       "step": 9652
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5063786855283815e-05,
-      "loss": 0.8304,
+      "loss": 0.9044,
       "step": 9653
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5062771136832784e-05,
-      "loss": 0.7256,
+      "loss": 0.8413,
       "step": 9654
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5061755348142988e-05,
-      "loss": 0.8259,
+      "loss": 0.8556,
       "step": 9655
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5060739489228517e-05,
-      "loss": 0.8662,
+      "loss": 1.0027,
       "step": 9656
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5059723560103466e-05,
-      "loss": 0.8519,
+      "loss": 0.9786,
       "step": 9657
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5058707560781928e-05,
-      "loss": 0.8131,
+      "loss": 0.9238,
       "step": 9658
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5057691491278003e-05,
-      "loss": 0.6283,
+      "loss": 0.7788,
       "step": 9659
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5056675351605783e-05,
-      "loss": 0.7836,
+      "loss": 0.8824,
       "step": 9660
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5055659141779366e-05,
-      "loss": 0.9217,
+      "loss": 0.986,
       "step": 9661
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5054642861812854e-05,
-      "loss": 0.7719,
+      "loss": 0.9408,
       "step": 9662
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.505362651172034e-05,
-      "loss": 0.8253,
+      "loss": 0.9215,
       "step": 9663
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5052610091515931e-05,
-      "loss": 0.781,
+      "loss": 0.911,
       "step": 9664
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5051593601213728e-05,
-      "loss": 0.8247,
+      "loss": 0.9533,
       "step": 9665
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5050577040827827e-05,
-      "loss": 0.7528,
+      "loss": 0.8569,
       "step": 9666
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5049560410372336e-05,
-      "loss": 0.7589,
+      "loss": 0.8796,
       "step": 9667
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5048543709861361e-05,
-      "loss": 0.843,
+      "loss": 0.9046,
       "step": 9668
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5047526939309003e-05,
-      "loss": 0.8031,
+      "loss": 0.862,
       "step": 9669
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5046510098729372e-05,
-      "loss": 0.6695,
+      "loss": 0.8656,
       "step": 9670
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5045493188136573e-05,
-      "loss": 0.8136,
+      "loss": 0.8504,
       "step": 9671
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5044476207544711e-05,
-      "loss": 0.7887,
+      "loss": 0.8416,
       "step": 9672
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5043459156967903e-05,
-      "loss": 0.7584,
+      "loss": 0.8526,
       "step": 9673
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5042442036420255e-05,
-      "loss": 0.8359,
+      "loss": 0.8936,
       "step": 9674
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5041424845915874e-05,
-      "loss": 0.7378,
+      "loss": 0.9226,
       "step": 9675
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5040407585468878e-05,
-      "loss": 0.782,
+      "loss": 0.901,
       "step": 9676
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5039390255093377e-05,
-      "loss": 0.7569,
+      "loss": 0.9044,
       "step": 9677
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5038372854803492e-05,
-      "loss": 0.8427,
+      "loss": 0.8775,
       "step": 9678
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5037355384613327e-05,
-      "loss": 0.8681,
+      "loss": 0.8669,
       "step": 9679
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5036337844537002e-05,
-      "loss": 0.7162,
+      "loss": 0.8196,
       "step": 9680
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5035320234588637e-05,
-      "loss": 0.8371,
+      "loss": 0.9739,
       "step": 9681
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.503430255478235e-05,
-      "loss": 0.6389,
+      "loss": 0.7976,
       "step": 9682
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5033284805132255e-05,
-      "loss": 0.7544,
+      "loss": 0.8947,
       "step": 9683
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5032266985652478e-05,
-      "loss": 0.8597,
+      "loss": 0.9473,
       "step": 9684
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5031249096357132e-05,
-      "loss": 0.7713,
+      "loss": 0.8472,
       "step": 9685
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5030231137260344e-05,
-      "loss": 0.856,
+      "loss": 0.9297,
       "step": 9686
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.502921310837624e-05,
-      "loss": 0.8234,
+      "loss": 0.9511,
       "step": 9687
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5028195009718935e-05,
-      "loss": 0.6203,
+      "loss": 0.8382,
       "step": 9688
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5027176841302563e-05,
-      "loss": 0.6724,
+      "loss": 0.8389,
       "step": 9689
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.502615860314124e-05,
-      "loss": 0.8026,
+      "loss": 0.8657,
       "step": 9690
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.50251402952491e-05,
-      "loss": 0.8336,
+      "loss": 0.8511,
       "step": 9691
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5024121917640268e-05,
-      "loss": 0.7649,
+      "loss": 0.7945,
       "step": 9692
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5023103470328871e-05,
-      "loss": 0.8188,
+      "loss": 0.8535,
       "step": 9693
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.502208495332904e-05,
-      "loss": 0.744,
+      "loss": 0.835,
       "step": 9694
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5021066366654911e-05,
-      "loss": 0.8589,
+      "loss": 0.9556,
       "step": 9695
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5020047710320605e-05,
-      "loss": 0.7842,
+      "loss": 0.9517,
       "step": 9696
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5019028984340262e-05,
-      "loss": 0.7979,
+      "loss": 0.9706,
       "step": 9697
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5018010188728012e-05,
-      "loss": 0.7986,
+      "loss": 0.9413,
       "step": 9698
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5016991323497989e-05,
-      "loss": 0.8271,
+      "loss": 0.8992,
       "step": 9699
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5015972388664332e-05,
-      "loss": 0.7978,
+      "loss": 0.9092,
       "step": 9700
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5014953384241173e-05,
-      "loss": 0.6985,
+      "loss": 0.845,
       "step": 9701
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5013934310242651e-05,
-      "loss": 0.7542,
+      "loss": 0.8402,
       "step": 9702
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5012915166682907e-05,
-      "loss": 0.7839,
+      "loss": 0.8901,
       "step": 9703
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5011895953576074e-05,
-      "loss": 0.7512,
+      "loss": 0.9135,
       "step": 9704
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5010876670936297e-05,
-      "loss": 0.7872,
+      "loss": 0.8892,
       "step": 9705
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5009857318777716e-05,
-      "loss": 0.734,
+      "loss": 0.8259,
       "step": 9706
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5008837897114474e-05,
-      "loss": 0.683,
+      "loss": 0.8396,
       "step": 9707
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5007818405960711e-05,
-      "loss": 0.7878,
+      "loss": 0.9421,
       "step": 9708
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.500679884533057e-05,
-      "loss": 0.7895,
+      "loss": 0.8742,
       "step": 9709
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5005779215238203e-05,
-      "loss": 0.6871,
+      "loss": 0.7888,
       "step": 9710
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5004759515697753e-05,
-      "loss": 0.7696,
+      "loss": 0.9249,
       "step": 9711
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.500373974672336e-05,
-      "loss": 0.7715,
+      "loss": 0.8639,
       "step": 9712
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5002719908329183e-05,
-      "loss": 0.7235,
+      "loss": 0.8864,
       "step": 9713
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5001700000529362e-05,
-      "loss": 0.7469,
+      "loss": 0.9041,
       "step": 9714
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.500068002333805e-05,
-      "loss": 0.818,
+      "loss": 0.9633,
       "step": 9715
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4999659976769398e-05,
-      "loss": 0.7538,
+      "loss": 0.8692,
       "step": 9716
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4998639860837559e-05,
-      "loss": 0.7887,
+      "loss": 0.8806,
       "step": 9717
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.499761967555668e-05,
-      "loss": 0.6783,
+      "loss": 0.7485,
       "step": 9718
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4996599420940923e-05,
-      "loss": 0.7168,
+      "loss": 0.8653,
       "step": 9719
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4995579097004436e-05,
-      "loss": 0.8105,
+      "loss": 0.9063,
       "step": 9720
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4994558703761377e-05,
-      "loss": 0.758,
+      "loss": 0.878,
       "step": 9721
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4993538241225905e-05,
-      "loss": 0.8384,
+      "loss": 0.877,
       "step": 9722
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4992517709412171e-05,
-      "loss": 0.7318,
+      "loss": 0.9181,
       "step": 9723
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4991497108334339e-05,
-      "loss": 0.6795,
+      "loss": 0.8138,
       "step": 9724
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4990476438006567e-05,
-      "loss": 0.7639,
+      "loss": 0.794,
       "step": 9725
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4989455698443015e-05,
-      "loss": 0.7487,
+      "loss": 0.8436,
       "step": 9726
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4988434889657844e-05,
-      "loss": 0.8851,
+      "loss": 0.9504,
       "step": 9727
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4987414011665216e-05,
-      "loss": 0.879,
+      "loss": 0.993,
       "step": 9728
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4986393064479296e-05,
-      "loss": 0.7777,
+      "loss": 0.8271,
       "step": 9729
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4985372048114247e-05,
-      "loss": 0.705,
+      "loss": 0.8275,
       "step": 9730
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4984350962584233e-05,
-      "loss": 0.8155,
+      "loss": 0.921,
       "step": 9731
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4983329807903422e-05,
-      "loss": 0.7957,
+      "loss": 0.8268,
       "step": 9732
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4982308584085978e-05,
-      "loss": 0.7371,
+      "loss": 0.867,
       "step": 9733
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4981287291146074e-05,
-      "loss": 0.7978,
+      "loss": 0.9372,
       "step": 9734
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4980265929097875e-05,
-      "loss": 0.6359,
+      "loss": 0.8137,
       "step": 9735
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4979244497955554e-05,
-      "loss": 0.8904,
+      "loss": 1.001,
       "step": 9736
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4978222997733278e-05,
-      "loss": 0.8117,
+      "loss": 0.933,
       "step": 9737
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4977201428445223e-05,
-      "loss": 0.7685,
+      "loss": 0.9279,
       "step": 9738
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4976179790105559e-05,
-      "loss": 0.845,
+      "loss": 0.996,
       "step": 9739
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.497515808272846e-05,
-      "loss": 0.7206,
+      "loss": 0.8587,
       "step": 9740
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4974136306328105e-05,
-      "loss": 0.7916,
+      "loss": 0.863,
       "step": 9741
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.497311446091866e-05,
-      "loss": 0.7977,
+      "loss": 0.9208,
       "step": 9742
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4972092546514314e-05,
-      "loss": 0.7905,
+      "loss": 0.868,
       "step": 9743
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4971070563129232e-05,
-      "loss": 0.8056,
+      "loss": 0.8407,
       "step": 9744
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4970048510777607e-05,
-      "loss": 0.641,
+      "loss": 0.8661,
       "step": 9745
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4969026389473606e-05,
-      "loss": 0.7951,
+      "loss": 0.8459,
       "step": 9746
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4968004199231411e-05,
-      "loss": 0.7034,
+      "loss": 0.8281,
       "step": 9747
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4966981940065212e-05,
-      "loss": 0.8074,
+      "loss": 0.9273,
       "step": 9748
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4965959611989183e-05,
-      "loss": 0.7838,
+      "loss": 0.8595,
       "step": 9749
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.496493721501751e-05,
-      "loss": 0.8773,
+      "loss": 0.9692,
       "step": 9750
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.496391474916438e-05,
-      "loss": 0.6804,
+      "loss": 0.7999,
       "step": 9751
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4962892214443972e-05,
-      "loss": 0.7245,
+      "loss": 0.9321,
       "step": 9752
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4961869610870477e-05,
-      "loss": 0.7167,
+      "loss": 0.8349,
       "step": 9753
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4960846938458082e-05,
-      "loss": 0.8054,
+      "loss": 0.9017,
       "step": 9754
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4959824197220974e-05,
-      "loss": 0.8097,
+      "loss": 0.9243,
       "step": 9755
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4958801387173343e-05,
-      "loss": 0.7345,
+      "loss": 0.8997,
       "step": 9756
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4957778508329377e-05,
-      "loss": 0.7448,
+      "loss": 0.8561,
       "step": 9757
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4956755560703267e-05,
-      "loss": 0.8047,
+      "loss": 0.8725,
       "step": 9758
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4955732544309209e-05,
-      "loss": 0.649,
+      "loss": 0.8114,
       "step": 9759
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.4954709459161391e-05,
-      "loss": 0.9129,
+      "loss": 0.9828,
       "step": 9760
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.495368630527401e-05,
-      "loss": 0.727,
+      "loss": 0.8481,
       "step": 9761
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4952663082661263e-05,
-      "loss": 0.7857,
+      "loss": 0.8738,
       "step": 9762
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.495163979133734e-05,
-      "loss": 0.7063,
+      "loss": 0.7603,
       "step": 9763
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4950616431316442e-05,
-      "loss": 0.7383,
+      "loss": 0.8283,
       "step": 9764
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.494959300261276e-05,
-      "loss": 0.9018,
+      "loss": 1.0175,
       "step": 9765
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4948569505240504e-05,
-      "loss": 0.7099,
+      "loss": 0.8004,
       "step": 9766
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4947545939213866e-05,
-      "loss": 0.8512,
+      "loss": 0.9238,
       "step": 9767
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4946522304547044e-05,
-      "loss": 0.6686,
+      "loss": 0.796,
       "step": 9768
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4945498601254244e-05,
-      "loss": 0.758,
+      "loss": 0.8852,
       "step": 9769
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.494447482934967e-05,
-      "loss": 0.8401,
+      "loss": 0.9313,
       "step": 9770
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4943450988847522e-05,
-      "loss": 0.8077,
+      "loss": 0.8394,
       "step": 9771
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4942427079762008e-05,
-      "loss": 0.6739,
+      "loss": 0.845,
       "step": 9772
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4941403102107325e-05,
-      "loss": 0.8163,
+      "loss": 0.9041,
       "step": 9773
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4940379055897689e-05,
-      "loss": 0.7764,
+      "loss": 0.837,
       "step": 9774
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4939354941147302e-05,
-      "loss": 0.8955,
+      "loss": 0.9853,
       "step": 9775
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4938330757870372e-05,
-      "loss": 0.7295,
+      "loss": 0.8324,
       "step": 9776
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.493730650608111e-05,
-      "loss": 0.8023,
+      "loss": 0.916,
       "step": 9777
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4936282185793726e-05,
-      "loss": 0.6922,
+      "loss": 0.8326,
       "step": 9778
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4935257797022431e-05,
-      "loss": 0.7472,
+      "loss": 0.8313,
       "step": 9779
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4934233339781436e-05,
-      "loss": 0.6803,
+      "loss": 0.8832,
       "step": 9780
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4933208814084954e-05,
-      "loss": 0.875,
+      "loss": 0.9393,
       "step": 9781
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4932184219947196e-05,
-      "loss": 0.7424,
+      "loss": 0.854,
       "step": 9782
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4931159557382387e-05,
-      "loss": 0.7434,
+      "loss": 0.8658,
       "step": 9783
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.493013482640473e-05,
-      "loss": 0.6821,
+      "loss": 0.8495,
       "step": 9784
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4929110027028448e-05,
-      "loss": 0.779,
+      "loss": 0.8614,
       "step": 9785
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4928085159267758e-05,
-      "loss": 0.9166,
+      "loss": 0.9855,
       "step": 9786
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4927060223136878e-05,
-      "loss": 0.885,
+      "loss": 0.9641,
       "step": 9787
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.492603521865003e-05,
-      "loss": 0.866,
+      "loss": 0.9469,
       "step": 9788
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.492501014582143e-05,
-      "loss": 0.9257,
+      "loss": 0.9989,
       "step": 9789
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4923985004665302e-05,
-      "loss": 0.5931,
+      "loss": 0.7815,
       "step": 9790
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.492295979519587e-05,
-      "loss": 0.8354,
+      "loss": 0.9027,
       "step": 9791
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4921934517427355e-05,
-      "loss": 0.843,
+      "loss": 0.9224,
       "step": 9792
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.492090917137398e-05,
-      "loss": 0.7743,
+      "loss": 0.9083,
       "step": 9793
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4919883757049975e-05,
-      "loss": 0.7691,
+      "loss": 0.8269,
       "step": 9794
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4918858274469562e-05,
-      "loss": 0.869,
+      "loss": 0.9633,
       "step": 9795
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.491783272364697e-05,
-      "loss": 0.792,
+      "loss": 0.9178,
       "step": 9796
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4916807104596428e-05,
-      "loss": 0.7114,
+      "loss": 0.8824,
       "step": 9797
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4915781417332157e-05,
-      "loss": 0.83,
+      "loss": 0.8971,
       "step": 9798
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4914755661868401e-05,
-      "loss": 0.7195,
+      "loss": 0.8447,
       "step": 9799
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.491372983821938e-05,
-      "loss": 0.7577,
+      "loss": 0.8302,
       "step": 9800
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4912703946399328e-05,
-      "loss": 0.7979,
+      "loss": 0.8991,
       "step": 9801
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4911677986422483e-05,
-      "loss": 0.8072,
+      "loss": 0.9699,
       "step": 9802
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.491065195830307e-05,
-      "loss": 0.7449,
+      "loss": 0.8803,
       "step": 9803
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.490962586205533e-05,
-      "loss": 0.7901,
+      "loss": 0.9301,
       "step": 9804
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4908599697693497e-05,
-      "loss": 0.7048,
+      "loss": 0.8345,
       "step": 9805
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4907573465231808e-05,
-      "loss": 0.812,
+      "loss": 0.9319,
       "step": 9806
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4906547164684502e-05,
-      "loss": 0.6121,
+      "loss": 0.7852,
       "step": 9807
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.490552079606581e-05,
-      "loss": 0.7275,
+      "loss": 0.9066,
       "step": 9808
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4904494359389982e-05,
-      "loss": 0.8344,
+      "loss": 0.8918,
       "step": 9809
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4903467854671253e-05,
-      "loss": 0.7206,
+      "loss": 0.8231,
       "step": 9810
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4902441281923863e-05,
-      "loss": 0.707,
+      "loss": 0.8445,
       "step": 9811
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4901414641162055e-05,
-      "loss": 0.7347,
+      "loss": 0.8083,
       "step": 9812
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4900387932400079e-05,
-      "loss": 0.7266,
+      "loss": 0.8295,
       "step": 9813
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4899361155652165e-05,
-      "loss": 0.7562,
+      "loss": 0.8969,
       "step": 9814
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4898334310932572e-05,
-      "loss": 0.9094,
+      "loss": 0.8891,
       "step": 9815
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4897307398255537e-05,
-      "loss": 0.8325,
+      "loss": 0.9533,
       "step": 9816
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.489628041763531e-05,
-      "loss": 0.6286,
+      "loss": 0.7922,
       "step": 9817
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4895253369086142e-05,
-      "loss": 0.7755,
+      "loss": 0.9136,
       "step": 9818
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4894226252622277e-05,
-      "loss": 0.8509,
+      "loss": 0.8806,
       "step": 9819
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4893199068257966e-05,
-      "loss": 0.7848,
+      "loss": 0.929,
       "step": 9820
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4892171816007464e-05,
-      "loss": 0.7938,
+      "loss": 0.8647,
       "step": 9821
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4891144495885015e-05,
-      "loss": 0.7836,
+      "loss": 0.9081,
       "step": 9822
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.489011710790488e-05,
-      "loss": 0.7843,
+      "loss": 0.9361,
       "step": 9823
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4889089652081304e-05,
-      "loss": 0.7845,
+      "loss": 0.8999,
       "step": 9824
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4888062128428548e-05,
-      "loss": 0.7662,
+      "loss": 0.8914,
       "step": 9825
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4887034536960865e-05,
-      "loss": 0.797,
+      "loss": 0.8275,
       "step": 9826
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4886006877692512e-05,
-      "loss": 0.9418,
+      "loss": 0.9734,
       "step": 9827
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4884979150637744e-05,
-      "loss": 0.6702,
+      "loss": 0.8091,
       "step": 9828
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4883951355810824e-05,
-      "loss": 0.7371,
+      "loss": 0.9283,
       "step": 9829
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4882923493226005e-05,
-      "loss": 0.7959,
+      "loss": 0.9312,
       "step": 9830
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4881895562897555e-05,
-      "loss": 0.6934,
+      "loss": 0.9077,
       "step": 9831
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4880867564839725e-05,
-      "loss": 0.7045,
+      "loss": 0.899,
       "step": 9832
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4879839499066786e-05,
-      "loss": 0.8124,
+      "loss": 0.9241,
       "step": 9833
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4878811365592997e-05,
-      "loss": 0.8505,
+      "loss": 0.8786,
       "step": 9834
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4877783164432623e-05,
-      "loss": 0.848,
+      "loss": 0.9355,
       "step": 9835
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4876754895599928e-05,
-      "loss": 0.8304,
+      "loss": 0.8882,
       "step": 9836
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4875726559109177e-05,
-      "loss": 0.8343,
+      "loss": 0.975,
       "step": 9837
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4874698154974638e-05,
-      "loss": 0.91,
+      "loss": 0.9588,
       "step": 9838
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4873669683210578e-05,
-      "loss": 0.7754,
+      "loss": 0.8492,
       "step": 9839
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4872641143831271e-05,
-      "loss": 0.8111,
+      "loss": 0.924,
       "step": 9840
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4871612536850977e-05,
-      "loss": 0.6673,
+      "loss": 0.8558,
       "step": 9841
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4870583862283972e-05,
-      "loss": 0.6944,
+      "loss": 0.8554,
       "step": 9842
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4869555120144523e-05,
-      "loss": 0.7571,
+      "loss": 0.9338,
       "step": 9843
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4868526310446909e-05,
-      "loss": 0.7352,
+      "loss": 0.8445,
       "step": 9844
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.48674974332054e-05,
-      "loss": 0.8469,
+      "loss": 0.9444,
       "step": 9845
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.486646848843427e-05,
-      "loss": 0.8103,
+      "loss": 0.942,
       "step": 9846
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4865439476147794e-05,
-      "loss": 0.8816,
+      "loss": 0.9808,
       "step": 9847
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4864410396360247e-05,
-      "loss": 0.8054,
+      "loss": 0.9232,
       "step": 9848
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4863381249085909e-05,
-      "loss": 0.8218,
+      "loss": 0.9039,
       "step": 9849
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4862352034339057e-05,
-      "loss": 0.7567,
+      "loss": 0.9153,
       "step": 9850
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4861322752133966e-05,
-      "loss": 0.8236,
+      "loss": 0.9185,
       "step": 9851
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4860293402484925e-05,
-      "loss": 0.7642,
+      "loss": 0.8683,
       "step": 9852
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4859263985406205e-05,
-      "loss": 0.6776,
+      "loss": 0.8193,
       "step": 9853
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.485823450091209e-05,
-      "loss": 0.787,
+      "loss": 0.9091,
       "step": 9854
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4857204949016868e-05,
-      "loss": 0.7506,
+      "loss": 0.8817,
       "step": 9855
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4856175329734819e-05,
-      "loss": 0.7127,
+      "loss": 0.8076,
       "step": 9856
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4855145643080222e-05,
-      "loss": 0.7877,
+      "loss": 0.9619,
       "step": 9857
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4854115889067372e-05,
-      "loss": 0.7647,
+      "loss": 0.9196,
       "step": 9858
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4853086067710549e-05,
-      "loss": 0.7525,
+      "loss": 0.7867,
       "step": 9859
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4852056179024046e-05,
-      "loss": 0.7025,
+      "loss": 0.9072,
       "step": 9860
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4851026223022147e-05,
-      "loss": 0.8549,
+      "loss": 0.9035,
       "step": 9861
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4849996199719138e-05,
-      "loss": 0.7927,
+      "loss": 0.8507,
       "step": 9862
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4848966109129318e-05,
-      "loss": 0.8601,
+      "loss": 0.937,
       "step": 9863
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4847935951266968e-05,
-      "loss": 0.5757,
+      "loss": 0.8475,
       "step": 9864
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4846905726146388e-05,
-      "loss": 0.7914,
+      "loss": 0.8654,
       "step": 9865
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4845875433781869e-05,
-      "loss": 0.8936,
+      "loss": 0.8957,
       "step": 9866
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.48448450741877e-05,
-      "loss": 0.7955,
+      "loss": 0.9386,
       "step": 9867
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4843814647378181e-05,
-      "loss": 0.7989,
+      "loss": 0.8417,
       "step": 9868
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4842784153367608e-05,
-      "loss": 0.6953,
+      "loss": 0.8742,
       "step": 9869
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4841753592170273e-05,
-      "loss": 0.8849,
+      "loss": 0.9371,
       "step": 9870
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.484072296380048e-05,
-      "loss": 0.8077,
+      "loss": 0.9151,
       "step": 9871
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4839692268272523e-05,
-      "loss": 0.6828,
+      "loss": 0.7653,
       "step": 9872
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.48386615056007e-05,
-      "loss": 0.7312,
+      "loss": 0.9022,
       "step": 9873
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4837630675799315e-05,
-      "loss": 0.6646,
+      "loss": 0.8503,
       "step": 9874
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4836599778882667e-05,
-      "loss": 0.7357,
+      "loss": 0.9128,
       "step": 9875
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4835568814865062e-05,
-      "loss": 0.6901,
+      "loss": 0.8523,
       "step": 9876
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.48345377837608e-05,
-      "loss": 0.8386,
+      "loss": 0.8564,
       "step": 9877
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4833506685584182e-05,
-      "loss": 0.7975,
+      "loss": 0.9283,
       "step": 9878
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4832475520349523e-05,
-      "loss": 0.6602,
+      "loss": 0.8692,
       "step": 9879
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4831444288071117e-05,
-      "loss": 0.7344,
+      "loss": 0.8342,
       "step": 9880
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4830412988763278e-05,
-      "loss": 0.5338,
+      "loss": 0.732,
       "step": 9881
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4829381622440315e-05,
-      "loss": 0.8237,
+      "loss": 0.9074,
       "step": 9882
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.482835018911653e-05,
-      "loss": 0.7978,
+      "loss": 0.8539,
       "step": 9883
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4827318688806238e-05,
-      "loss": 0.6415,
+      "loss": 0.757,
       "step": 9884
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4826287121523748e-05,
-      "loss": 0.7508,
+      "loss": 0.8555,
       "step": 9885
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4825255487283372e-05,
-      "loss": 0.6695,
+      "loss": 0.8195,
       "step": 9886
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4824223786099424e-05,
-      "loss": 0.7351,
+      "loss": 0.8796,
       "step": 9887
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4823192017986211e-05,
-      "loss": 0.7911,
+      "loss": 0.9325,
       "step": 9888
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4822160182958057e-05,
-      "loss": 0.8757,
+      "loss": 0.9771,
       "step": 9889
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4821128281029272e-05,
-      "loss": 0.8246,
+      "loss": 0.9458,
       "step": 9890
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4820096312214167e-05,
-      "loss": 0.7736,
+      "loss": 0.8428,
       "step": 9891
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4819064276527069e-05,
-      "loss": 0.7793,
+      "loss": 0.8547,
       "step": 9892
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.481803217398229e-05,
-      "loss": 0.7902,
+      "loss": 0.9201,
       "step": 9893
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4817000004594151e-05,
-      "loss": 0.7752,
+      "loss": 0.8981,
       "step": 9894
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4815967768376973e-05,
-      "loss": 0.6994,
+      "loss": 0.818,
       "step": 9895
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4814935465345075e-05,
-      "loss": 0.7296,
+      "loss": 0.9279,
       "step": 9896
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4813903095512774e-05,
-      "loss": 0.6957,
+      "loss": 0.8092,
       "step": 9897
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4812870658894405e-05,
-      "loss": 0.6569,
+      "loss": 0.8051,
       "step": 9898
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4811838155504278e-05,
-      "loss": 0.9276,
+      "loss": 0.9577,
       "step": 9899
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4810805585356727e-05,
-      "loss": 0.7346,
+      "loss": 0.8709,
       "step": 9900
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4809772948466075e-05,
-      "loss": 0.8702,
+      "loss": 0.8934,
       "step": 9901
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4808740244846644e-05,
-      "loss": 0.7388,
+      "loss": 0.7908,
       "step": 9902
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4807707474512766e-05,
-      "loss": 0.7944,
+      "loss": 0.9269,
       "step": 9903
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4806674637478769e-05,
-      "loss": 0.7188,
+      "loss": 0.9004,
       "step": 9904
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.480564173375898e-05,
-      "loss": 0.7273,
+      "loss": 0.8287,
       "step": 9905
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4804608763367731e-05,
-      "loss": 0.8051,
+      "loss": 0.9448,
       "step": 9906
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4803575726319353e-05,
-      "loss": 0.8507,
+      "loss": 0.9807,
       "step": 9907
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4802542622628176e-05,
-      "loss": 0.7075,
+      "loss": 0.8195,
       "step": 9908
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4801509452308536e-05,
-      "loss": 0.7986,
+      "loss": 0.8807,
       "step": 9909
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4800476215374762e-05,
-      "loss": 0.8535,
+      "loss": 0.9927,
       "step": 9910
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4799442911841192e-05,
-      "loss": 0.7416,
+      "loss": 0.8541,
       "step": 9911
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4798409541722163e-05,
-      "loss": 0.6774,
+      "loss": 0.8022,
       "step": 9912
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4797376105032007e-05,
-      "loss": 0.8587,
+      "loss": 0.9365,
       "step": 9913
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4796342601785068e-05,
-      "loss": 0.7746,
+      "loss": 0.887,
       "step": 9914
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4795309031995675e-05,
-      "loss": 0.7546,
+      "loss": 0.8552,
       "step": 9915
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4794275395678177e-05,
-      "loss": 0.9721,
+      "loss": 0.9493,
       "step": 9916
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4793241692846911e-05,
-      "loss": 0.5584,
+      "loss": 0.7224,
       "step": 9917
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4792207923516215e-05,
-      "loss": 0.8576,
+      "loss": 0.9744,
       "step": 9918
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4791174087700433e-05,
-      "loss": 0.7632,
+      "loss": 0.8739,
       "step": 9919
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4790140185413908e-05,
-      "loss": 0.8279,
+      "loss": 1.0005,
       "step": 9920
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4789106216670987e-05,
-      "loss": 0.747,
+      "loss": 0.9105,
       "step": 9921
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4788072181486011e-05,
-      "loss": 0.8765,
+      "loss": 0.9853,
       "step": 9922
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4787038079873323e-05,
-      "loss": 0.7895,
+      "loss": 0.8847,
       "step": 9923
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.478600391184728e-05,
-      "loss": 0.7777,
+      "loss": 0.864,
       "step": 9924
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4784969677422218e-05,
-      "loss": 0.9601,
+      "loss": 1.0181,
       "step": 9925
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4783935376612493e-05,
-      "loss": 0.7684,
+      "loss": 0.8902,
       "step": 9926
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4782901009432454e-05,
-      "loss": 0.7888,
+      "loss": 0.8622,
       "step": 9927
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4781866575896446e-05,
-      "loss": 0.7506,
+      "loss": 0.7872,
       "step": 9928
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4780832076018824e-05,
-      "loss": 0.8576,
+      "loss": 0.9407,
       "step": 9929
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4779797509813942e-05,
-      "loss": 0.7076,
+      "loss": 0.8626,
       "step": 9930
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.477876287729615e-05,
-      "loss": 0.9445,
+      "loss": 0.9997,
       "step": 9931
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4777728178479804e-05,
-      "loss": 0.8161,
+      "loss": 0.8993,
       "step": 9932
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4776693413379257e-05,
-      "loss": 0.709,
+      "loss": 0.879,
       "step": 9933
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4775658582008867e-05,
-      "loss": 0.7788,
+      "loss": 0.93,
       "step": 9934
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.477462368438299e-05,
-      "loss": 0.7198,
+      "loss": 0.7555,
       "step": 9935
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4773588720515984e-05,
-      "loss": 0.7601,
+      "loss": 0.8865,
       "step": 9936
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.477255369042221e-05,
-      "loss": 0.7966,
+      "loss": 0.9073,
       "step": 9937
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4771518594116021e-05,
-      "loss": 0.7653,
+      "loss": 0.858,
       "step": 9938
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4770483431611781e-05,
-      "loss": 0.8976,
+      "loss": 0.8822,
       "step": 9939
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4769448202923853e-05,
-      "loss": 0.9059,
+      "loss": 0.9153,
       "step": 9940
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4768412908066601e-05,
-      "loss": 0.8092,
+      "loss": 0.8811,
       "step": 9941
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.476737754705438e-05,
-      "loss": 0.7973,
+      "loss": 0.8759,
       "step": 9942
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4766342119901562e-05,
-      "loss": 0.9189,
+      "loss": 1.0129,
       "step": 9943
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4765306626622512e-05,
-      "loss": 0.8867,
+      "loss": 0.8987,
       "step": 9944
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4764271067231591e-05,
-      "loss": 0.7831,
+      "loss": 0.843,
       "step": 9945
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.476323544174317e-05,
-      "loss": 0.6589,
+      "loss": 0.8515,
       "step": 9946
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.476219975017161e-05,
-      "loss": 0.8142,
+      "loss": 0.9155,
       "step": 9947
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.476116399253129e-05,
-      "loss": 0.9648,
+      "loss": 1.0421,
       "step": 9948
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4760128168836576e-05,
-      "loss": 0.6156,
+      "loss": 0.8057,
       "step": 9949
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4759092279101834e-05,
-      "loss": 0.6539,
+      "loss": 0.7256,
       "step": 9950
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4758056323341443e-05,
-      "loss": 0.8307,
+      "loss": 0.9164,
       "step": 9951
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.475702030156977e-05,
-      "loss": 0.7648,
+      "loss": 0.9487,
       "step": 9952
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4755984213801187e-05,
-      "loss": 0.6374,
+      "loss": 0.8494,
       "step": 9953
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4754948060050072e-05,
-      "loss": 0.8732,
+      "loss": 0.9012,
       "step": 9954
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4753911840330802e-05,
-      "loss": 0.9084,
+      "loss": 0.9802,
       "step": 9955
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4752875554657746e-05,
-      "loss": 0.6586,
+      "loss": 0.7702,
       "step": 9956
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4751839203045288e-05,
-      "loss": 0.6685,
+      "loss": 0.8896,
       "step": 9957
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4750802785507801e-05,
-      "loss": 0.8572,
+      "loss": 0.864,
       "step": 9958
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.474976630205967e-05,
-      "loss": 0.7027,
+      "loss": 0.8642,
       "step": 9959
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4748729752715264e-05,
-      "loss": 0.7702,
+      "loss": 0.8972,
       "step": 9960
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4747693137488974e-05,
-      "loss": 0.9469,
+      "loss": 1.0491,
       "step": 9961
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.474665645639518e-05,
-      "loss": 0.6648,
+      "loss": 0.8345,
       "step": 9962
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4745619709448256e-05,
-      "loss": 0.7768,
+      "loss": 0.8529,
       "step": 9963
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4744582896662598e-05,
-      "loss": 0.8084,
+      "loss": 0.856,
       "step": 9964
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.474354601805258e-05,
-      "loss": 0.6556,
+      "loss": 0.884,
       "step": 9965
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4742509073632594e-05,
-      "loss": 0.9082,
+      "loss": 0.9951,
       "step": 9966
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.474147206341702e-05,
-      "loss": 0.8094,
+      "loss": 0.9403,
       "step": 9967
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.474043498742025e-05,
-      "loss": 0.7767,
+      "loss": 0.853,
       "step": 9968
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4739397845656671e-05,
-      "loss": 0.7002,
+      "loss": 0.8628,
       "step": 9969
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4738360638140672e-05,
-      "loss": 0.9411,
+      "loss": 0.9761,
       "step": 9970
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.473732336488664e-05,
-      "loss": 0.7636,
+      "loss": 0.7924,
       "step": 9971
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.473628602590897e-05,
-      "loss": 0.8203,
+      "loss": 0.9573,
       "step": 9972
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.473524862122205e-05,
-      "loss": 0.7487,
+      "loss": 0.8487,
       "step": 9973
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4734211150840271e-05,
-      "loss": 0.6952,
+      "loss": 0.8106,
       "step": 9974
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4733173614778035e-05,
-      "loss": 0.6745,
+      "loss": 0.9213,
       "step": 9975
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4732136013049728e-05,
-      "loss": 0.7132,
+      "loss": 0.7757,
       "step": 9976
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4731098345669748e-05,
-      "loss": 0.7589,
+      "loss": 0.8766,
       "step": 9977
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.473006061265249e-05,
-      "loss": 0.6673,
+      "loss": 0.8354,
       "step": 9978
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.472902281401235e-05,
-      "loss": 0.8249,
+      "loss": 0.8843,
       "step": 9979
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4727984949763732e-05,
-      "loss": 0.8744,
+      "loss": 0.8914,
       "step": 9980
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4726947019921029e-05,
-      "loss": 0.8644,
+      "loss": 0.9253,
       "step": 9981
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4725909024498643e-05,
-      "loss": 0.6402,
+      "loss": 0.812,
       "step": 9982
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4724870963510975e-05,
-      "loss": 0.8331,
+      "loss": 0.8803,
       "step": 9983
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4723832836972427e-05,
-      "loss": 0.8655,
+      "loss": 0.9327,
       "step": 9984
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4722794644897399e-05,
-      "loss": 0.8405,
+      "loss": 0.9243,
       "step": 9985
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4721756387300297e-05,
-      "loss": 0.8404,
+      "loss": 0.8993,
       "step": 9986
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4720718064195523e-05,
-      "loss": 0.7034,
+      "loss": 0.8482,
       "step": 9987
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4719679675597485e-05,
-      "loss": 0.913,
+      "loss": 0.9734,
       "step": 9988
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4718641221520588e-05,
-      "loss": 0.8337,
+      "loss": 0.8542,
       "step": 9989
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4717602701979236e-05,
-      "loss": 0.8135,
+      "loss": 0.8795,
       "step": 9990
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4716564116987844e-05,
-      "loss": 0.6987,
+      "loss": 0.8982,
       "step": 9991
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4715525466560816e-05,
-      "loss": 0.827,
+      "loss": 0.9301,
       "step": 9992
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.471448675071256e-05,
-      "loss": 1.03,
+      "loss": 1.008,
       "step": 9993
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4713447969457493e-05,
-      "loss": 0.8371,
+      "loss": 0.9305,
       "step": 9994
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.471240912281002e-05,
-      "loss": 0.7525,
+      "loss": 0.9091,
       "step": 9995
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.471137021078456e-05,
-      "loss": 0.7866,
+      "loss": 0.9006,
       "step": 9996
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.471033123339552e-05,
-      "loss": 0.8632,
+      "loss": 0.9846,
       "step": 9997
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4709292190657316e-05,
-      "loss": 0.7758,
+      "loss": 0.8448,
       "step": 9998
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4708253082584366e-05,
-      "loss": 0.7062,
+      "loss": 0.854,
       "step": 9999
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4707213909191085e-05,
-      "loss": 0.7846,
+      "loss": 0.867,
       "step": 10000
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4706174670491888e-05,
-      "loss": 0.6795,
+      "loss": 0.7925,
       "step": 10001
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4705135366501197e-05,
-      "loss": 0.8099,
+      "loss": 0.9395,
       "step": 10002
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4704095997233427e-05,
-      "loss": 0.7348,
+      "loss": 0.9085,
       "step": 10003
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4703056562703e-05,
-      "loss": 0.7081,
+      "loss": 0.9971,
       "step": 10004
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4702017062924337e-05,
-      "loss": 0.7684,
+      "loss": 0.8981,
       "step": 10005
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4700977497911854e-05,
-      "loss": 0.7881,
+      "loss": 0.827,
       "step": 10006
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4699937867679983e-05,
-      "loss": 0.7973,
+      "loss": 0.8307,
       "step": 10007
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.469889817224314e-05,
-      "loss": 0.814,
+      "loss": 0.9243,
       "step": 10008
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4697858411615753e-05,
-      "loss": 0.7758,
+      "loss": 0.8235,
       "step": 10009
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4696818585812244e-05,
-      "loss": 0.7406,
+      "loss": 0.8823,
       "step": 10010
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4695778694847045e-05,
-      "loss": 0.7845,
+      "loss": 0.8778,
       "step": 10011
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4694738738734575e-05,
-      "loss": 0.7387,
+      "loss": 0.8288,
       "step": 10012
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4693698717489269e-05,
-      "loss": 0.6468,
+      "loss": 0.8126,
       "step": 10013
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.469265863112555e-05,
-      "loss": 0.7513,
+      "loss": 0.9294,
       "step": 10014
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4691618479657854e-05,
-      "loss": 0.897,
+      "loss": 0.9794,
       "step": 10015
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4690578263100606e-05,
-      "loss": 0.777,
+      "loss": 0.8718,
       "step": 10016
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4689537981468238e-05,
-      "loss": 0.6823,
+      "loss": 0.8334,
       "step": 10017
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4688497634775185e-05,
-      "loss": 0.7104,
+      "loss": 0.8994,
       "step": 10018
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.468745722303588e-05,
-      "loss": 0.8597,
+      "loss": 0.8998,
       "step": 10019
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4686416746264755e-05,
-      "loss": 0.7423,
+      "loss": 0.8876,
       "step": 10020
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4685376204476252e-05,
-      "loss": 0.6699,
+      "loss": 0.8448,
       "step": 10021
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4684335597684796e-05,
-      "loss": 0.7625,
+      "loss": 0.8894,
       "step": 10022
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4683294925904833e-05,
-      "loss": 0.7438,
+      "loss": 0.7812,
       "step": 10023
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4682254189150798e-05,
-      "loss": 0.683,
+      "loss": 0.855,
       "step": 10024
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4681213387437125e-05,
-      "loss": 0.7349,
+      "loss": 0.8761,
       "step": 10025
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4680172520778262e-05,
-      "loss": 0.6808,
+      "loss": 0.8225,
       "step": 10026
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4679131589188647e-05,
-      "loss": 0.7946,
+      "loss": 0.9652,
       "step": 10027
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4678090592682714e-05,
-      "loss": 0.7654,
+      "loss": 0.8589,
       "step": 10028
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4677049531274917e-05,
-      "loss": 0.6349,
+      "loss": 0.7832,
       "step": 10029
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4676008404979691e-05,
-      "loss": 0.8072,
+      "loss": 0.8456,
       "step": 10030
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4674967213811484e-05,
-      "loss": 0.8488,
+      "loss": 0.8628,
       "step": 10031
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.467392595778474e-05,
-      "loss": 0.6633,
+      "loss": 0.8161,
       "step": 10032
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4672884636913905e-05,
-      "loss": 0.7934,
+      "loss": 0.8611,
       "step": 10033
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4671843251213425e-05,
-      "loss": 0.9234,
+      "loss": 1.0029,
       "step": 10034
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4670801800697744e-05,
-      "loss": 0.8507,
+      "loss": 0.952,
       "step": 10035
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.4669760285381323e-05,
-      "loss": 0.8393,
+      "loss": 0.9749,
       "step": 10036
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4668718705278599e-05,
-      "loss": 0.7815,
+      "loss": 0.9242,
       "step": 10037
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4667677060404026e-05,
-      "loss": 0.7559,
+      "loss": 0.8728,
       "step": 10038
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4666635350772058e-05,
-      "loss": 0.7883,
+      "loss": 0.8873,
       "step": 10039
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4665593576397145e-05,
-      "loss": 0.7587,
+      "loss": 0.9379,
       "step": 10040
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4664551737293742e-05,
-      "loss": 0.8442,
+      "loss": 0.8783,
       "step": 10041
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4663509833476301e-05,
-      "loss": 0.7458,
+      "loss": 0.8445,
       "step": 10042
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4662467864959278e-05,
-      "loss": 0.9286,
+      "loss": 0.9982,
       "step": 10043
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4661425831757128e-05,
-      "loss": 0.8766,
+      "loss": 0.9117,
       "step": 10044
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4660383733884309e-05,
-      "loss": 0.7274,
+      "loss": 0.821,
       "step": 10045
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4659341571355277e-05,
-      "loss": 0.6538,
+      "loss": 0.7979,
       "step": 10046
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4658299344184491e-05,
-      "loss": 0.7495,
+      "loss": 0.8487,
       "step": 10047
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4657257052386413e-05,
-      "loss": 0.7931,
+      "loss": 0.8351,
       "step": 10048
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4656214695975499e-05,
-      "loss": 0.8053,
+      "loss": 0.8336,
       "step": 10049
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4655172274966214e-05,
-      "loss": 0.7097,
+      "loss": 0.8589,
       "step": 10050
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.465412978937302e-05,
-      "loss": 0.7151,
+      "loss": 0.8649,
       "step": 10051
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4653087239210378e-05,
-      "loss": 0.7704,
+      "loss": 0.8181,
       "step": 10052
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.465204462449275e-05,
-      "loss": 0.678,
+      "loss": 0.8633,
       "step": 10053
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4651001945234602e-05,
-      "loss": 0.7581,
+      "loss": 0.8379,
       "step": 10054
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4649959201450403e-05,
-      "loss": 0.6886,
+      "loss": 0.9151,
       "step": 10055
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.464891639315462e-05,
-      "loss": 0.7189,
+      "loss": 0.8386,
       "step": 10056
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4647873520361713e-05,
-      "loss": 0.697,
+      "loss": 0.7816,
       "step": 10057
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.464683058308616e-05,
-      "loss": 0.902,
+      "loss": 0.9644,
       "step": 10058
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4645787581342421e-05,
-      "loss": 0.8436,
+      "loss": 0.8946,
       "step": 10059
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4644744515144975e-05,
-      "loss": 0.7721,
+      "loss": 0.8204,
       "step": 10060
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4643701384508288e-05,
-      "loss": 0.7946,
+      "loss": 0.8476,
       "step": 10061
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.464265818944683e-05,
-      "loss": 0.7079,
+      "loss": 0.8706,
       "step": 10062
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4641614929975081e-05,
-      "loss": 0.7388,
+      "loss": 0.8937,
       "step": 10063
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4640571606107508e-05,
-      "loss": 0.7537,
+      "loss": 0.8287,
       "step": 10064
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4639528217858588e-05,
-      "loss": 0.7039,
+      "loss": 0.8058,
       "step": 10065
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4638484765242797e-05,
-      "loss": 0.8126,
+      "loss": 0.8961,
       "step": 10066
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.463744124827461e-05,
-      "loss": 0.8109,
+      "loss": 0.8529,
       "step": 10067
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4636397666968508e-05,
-      "loss": 0.9052,
+      "loss": 0.9107,
       "step": 10068
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4635354021338964e-05,
-      "loss": 0.7607,
+      "loss": 0.9019,
       "step": 10069
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4634310311400457e-05,
-      "loss": 0.8847,
+      "loss": 0.9037,
       "step": 10070
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4633266537167476e-05,
-      "loss": 0.7629,
+      "loss": 0.841,
       "step": 10071
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4632222698654495e-05,
-      "loss": 0.7867,
+      "loss": 0.8425,
       "step": 10072
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4631178795875994e-05,
-      "loss": 0.7063,
+      "loss": 0.7935,
       "step": 10073
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4630134828846456e-05,
-      "loss": 0.7973,
+      "loss": 0.9538,
       "step": 10074
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4629090797580368e-05,
-      "loss": 0.8411,
+      "loss": 0.9623,
       "step": 10075
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4628046702092214e-05,
-      "loss": 0.6305,
+      "loss": 0.7644,
       "step": 10076
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.462700254239648e-05,
-      "loss": 0.6005,
+      "loss": 0.7506,
       "step": 10077
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4625958318507646e-05,
-      "loss": 0.6794,
+      "loss": 0.8474,
       "step": 10078
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4624914030440207e-05,
-      "loss": 0.8563,
+      "loss": 0.9506,
       "step": 10079
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4623869678208648e-05,
-      "loss": 0.8368,
+      "loss": 0.8914,
       "step": 10080
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4622825261827456e-05,
-      "loss": 0.8167,
+      "loss": 0.9097,
       "step": 10081
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4621780781311122e-05,
-      "loss": 0.7733,
+      "loss": 0.8619,
       "step": 10082
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.462073623667414e-05,
-      "loss": 0.8048,
+      "loss": 0.8555,
       "step": 10083
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4619691627930994e-05,
-      "loss": 0.6656,
+      "loss": 0.8746,
       "step": 10084
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4618646955096186e-05,
-      "loss": 0.9577,
+      "loss": 1.0331,
       "step": 10085
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.46176022181842e-05,
-      "loss": 0.6287,
+      "loss": 0.7966,
       "step": 10086
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4616557417209538e-05,
-      "loss": 0.8077,
+      "loss": 0.9412,
       "step": 10087
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4615512552186693e-05,
-      "loss": 0.681,
+      "loss": 0.8619,
       "step": 10088
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4614467623130155e-05,
-      "loss": 0.7458,
+      "loss": 0.8469,
       "step": 10089
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4613422630054433e-05,
-      "loss": 0.6872,
+      "loss": 0.793,
       "step": 10090
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.461237757297401e-05,
-      "loss": 0.7956,
+      "loss": 0.8795,
       "step": 10091
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4611332451903396e-05,
-      "loss": 0.8085,
+      "loss": 0.9211,
       "step": 10092
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4610287266857089e-05,
-      "loss": 0.812,
+      "loss": 0.8793,
       "step": 10093
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4609242017849583e-05,
-      "loss": 0.7783,
+      "loss": 0.8774,
       "step": 10094
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4608196704895385e-05,
-      "loss": 0.8849,
+      "loss": 0.9963,
       "step": 10095
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4607151328008996e-05,
-      "loss": 0.8581,
+      "loss": 0.947,
       "step": 10096
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.460610588720492e-05,
-      "loss": 0.813,
+      "loss": 0.9053,
       "step": 10097
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.460506038249766e-05,
-      "loss": 0.8052,
+      "loss": 0.911,
       "step": 10098
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.460401481390172e-05,
-      "loss": 0.8841,
+      "loss": 0.9655,
       "step": 10099
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4602969181431607e-05,
-      "loss": 0.8715,
+      "loss": 0.9444,
       "step": 10100
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4601923485101828e-05,
-      "loss": 0.7204,
+      "loss": 0.8413,
       "step": 10101
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4600877724926888e-05,
-      "loss": 0.6493,
+      "loss": 0.7741,
       "step": 10102
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.45998319009213e-05,
-      "loss": 0.6757,
+      "loss": 0.8281,
       "step": 10103
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4598786013099568e-05,
-      "loss": 0.8609,
+      "loss": 0.9113,
       "step": 10104
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4597740061476207e-05,
-      "loss": 0.7872,
+      "loss": 0.9312,
       "step": 10105
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4596694046065725e-05,
-      "loss": 0.8193,
+      "loss": 0.924,
       "step": 10106
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4595647966882636e-05,
-      "loss": 0.8098,
+      "loss": 0.876,
       "step": 10107
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4594601823941447e-05,
-      "loss": 0.6457,
+      "loss": 0.8176,
       "step": 10108
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4593555617256685e-05,
-      "loss": 0.76,
+      "loss": 0.7624,
       "step": 10109
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.459250934684285e-05,
-      "loss": 0.6702,
+      "loss": 0.8016,
       "step": 10110
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4591463012714466e-05,
-      "loss": 0.9273,
+      "loss": 0.9826,
       "step": 10111
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4590416614886049e-05,
-      "loss": 0.8464,
+      "loss": 0.8907,
       "step": 10112
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.458937015337211e-05,
-      "loss": 0.8038,
+      "loss": 0.92,
       "step": 10113
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4588323628187176e-05,
-      "loss": 0.746,
+      "loss": 0.9227,
       "step": 10114
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4587277039345757e-05,
-      "loss": 0.7466,
+      "loss": 0.8228,
       "step": 10115
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4586230386862381e-05,
-      "loss": 0.9383,
+      "loss": 0.965,
       "step": 10116
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4585183670751563e-05,
-      "loss": 0.8783,
+      "loss": 0.9625,
       "step": 10117
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4584136891027827e-05,
-      "loss": 0.7775,
+      "loss": 0.8949,
       "step": 10118
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4583090047705699e-05,
-      "loss": 0.7361,
+      "loss": 0.8672,
       "step": 10119
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4582043140799698e-05,
-      "loss": 0.8001,
+      "loss": 0.9421,
       "step": 10120
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4580996170324347e-05,
-      "loss": 0.8083,
+      "loss": 0.9053,
       "step": 10121
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4579949136294176e-05,
-      "loss": 0.7113,
+      "loss": 0.8123,
       "step": 10122
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4578902038723708e-05,
-      "loss": 0.632,
+      "loss": 0.8743,
       "step": 10123
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4577854877627471e-05,
-      "loss": 0.7188,
+      "loss": 0.8077,
       "step": 10124
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4576807653019993e-05,
-      "loss": 0.8458,
+      "loss": 0.9234,
       "step": 10125
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4575760364915801e-05,
-      "loss": 0.7715,
+      "loss": 0.9327,
       "step": 10126
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4574713013329427e-05,
-      "loss": 0.7728,
+      "loss": 0.8163,
       "step": 10127
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4573665598275403e-05,
-      "loss": 0.7919,
+      "loss": 0.8548,
       "step": 10128
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4572618119768256e-05,
-      "loss": 0.7253,
+      "loss": 0.775,
       "step": 10129
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.457157057782252e-05,
-      "loss": 0.7427,
+      "loss": 0.8701,
       "step": 10130
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4570522972452727e-05,
-      "loss": 0.7182,
+      "loss": 0.8859,
       "step": 10131
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4569475303673416e-05,
-      "loss": 0.8295,
+      "loss": 0.7833,
       "step": 10132
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4568427571499118e-05,
-      "loss": 0.8553,
+      "loss": 0.8937,
       "step": 10133
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4567379775944365e-05,
-      "loss": 0.7952,
+      "loss": 0.8675,
       "step": 10134
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4566331917023701e-05,
-      "loss": 0.7499,
+      "loss": 0.8586,
       "step": 10135
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.456528399475166e-05,
-      "loss": 0.9308,
+      "loss": 0.9482,
       "step": 10136
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4564236009142781e-05,
-      "loss": 0.9012,
+      "loss": 0.9764,
       "step": 10137
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4563187960211602e-05,
-      "loss": 0.7557,
+      "loss": 0.8659,
       "step": 10138
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4562139847972667e-05,
-      "loss": 0.5959,
+      "loss": 0.7819,
       "step": 10139
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4561091672440513e-05,
-      "loss": 0.777,
+      "loss": 0.9087,
       "step": 10140
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4560043433629683e-05,
-      "loss": 0.7906,
+      "loss": 0.929,
       "step": 10141
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4558995131554719e-05,
-      "loss": 0.7577,
+      "loss": 0.8858,
       "step": 10142
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4557946766230169e-05,
-      "loss": 0.8434,
+      "loss": 1.0032,
       "step": 10143
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4556898337670573e-05,
-      "loss": 0.7781,
+      "loss": 0.955,
       "step": 10144
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4555849845890477e-05,
-      "loss": 0.9177,
+      "loss": 0.9632,
       "step": 10145
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.455480129090443e-05,
-      "loss": 0.8319,
+      "loss": 0.8752,
       "step": 10146
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4553752672726975e-05,
-      "loss": 0.7539,
+      "loss": 0.8588,
       "step": 10147
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4552703991372666e-05,
-      "loss": 0.942,
+      "loss": 0.9428,
       "step": 10148
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4551655246856048e-05,
-      "loss": 0.7094,
+      "loss": 0.7804,
       "step": 10149
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4550606439191668e-05,
-      "loss": 0.7864,
+      "loss": 0.8907,
       "step": 10150
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4549557568394084e-05,
-      "loss": 0.7731,
+      "loss": 0.8168,
       "step": 10151
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4548508634477842e-05,
-      "loss": 0.7072,
+      "loss": 0.88,
       "step": 10152
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4547459637457494e-05,
-      "loss": 0.8348,
+      "loss": 0.9395,
       "step": 10153
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4546410577347599e-05,
-      "loss": 0.708,
+      "loss": 0.8304,
       "step": 10154
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4545361454162708e-05,
-      "loss": 0.9402,
+      "loss": 0.9313,
       "step": 10155
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4544312267917373e-05,
-      "loss": 0.806,
+      "loss": 0.8835,
       "step": 10156
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4543263018626155e-05,
-      "loss": 0.7354,
+      "loss": 0.8941,
       "step": 10157
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4542213706303607e-05,
-      "loss": 0.7289,
+      "loss": 0.9038,
       "step": 10158
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4541164330964291e-05,
-      "loss": 0.7348,
+      "loss": 0.89,
       "step": 10159
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.454011489262276e-05,
-      "loss": 0.8219,
+      "loss": 0.9286,
       "step": 10160
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4539065391293577e-05,
-      "loss": 0.8643,
+      "loss": 0.8857,
       "step": 10161
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4538015826991302e-05,
-      "loss": 0.7988,
+      "loss": 0.9231,
       "step": 10162
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4536966199730498e-05,
-      "loss": 0.7999,
+      "loss": 0.916,
       "step": 10163
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4535916509525723e-05,
-      "loss": 0.7644,
+      "loss": 0.8614,
       "step": 10164
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.453486675639154e-05,
-      "loss": 0.7754,
+      "loss": 0.8451,
       "step": 10165
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.453381694034252e-05,
-      "loss": 0.7761,
+      "loss": 0.8623,
       "step": 10166
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.453276706139322e-05,
-      "loss": 0.7356,
+      "loss": 0.8524,
       "step": 10167
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.453171711955821e-05,
-      "loss": 0.6787,
+      "loss": 0.8215,
       "step": 10168
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4530667114852051e-05,
-      "loss": 0.7764,
+      "loss": 0.9037,
       "step": 10169
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4529617047289318e-05,
-      "loss": 0.795,
+      "loss": 0.9384,
       "step": 10170
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4528566916884572e-05,
-      "loss": 0.6886,
+      "loss": 0.7996,
       "step": 10171
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4527516723652385e-05,
-      "loss": 0.8756,
+      "loss": 0.9929,
       "step": 10172
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.452646646760733e-05,
-      "loss": 0.7916,
+      "loss": 0.8938,
       "step": 10173
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4525416148763972e-05,
-      "loss": 0.7698,
+      "loss": 0.8618,
       "step": 10174
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4524365767136888e-05,
-      "loss": 0.7931,
+      "loss": 0.8581,
       "step": 10175
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4523315322740647e-05,
-      "loss": 0.705,
+      "loss": 0.8108,
       "step": 10176
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4522264815589822e-05,
-      "loss": 0.8339,
+      "loss": 0.9867,
       "step": 10177
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4521214245698992e-05,
-      "loss": 0.8382,
+      "loss": 0.8978,
       "step": 10178
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4520163613082728e-05,
-      "loss": 0.6967,
+      "loss": 0.7829,
       "step": 10179
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4519112917755607e-05,
-      "loss": 0.7565,
+      "loss": 0.802,
       "step": 10180
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4518062159732206e-05,
-      "loss": 0.7216,
+      "loss": 0.8325,
       "step": 10181
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4517011339027102e-05,
-      "loss": 0.7377,
+      "loss": 0.8727,
       "step": 10182
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4515960455654876e-05,
-      "loss": 0.6699,
+      "loss": 0.8304,
       "step": 10183
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4514909509630105e-05,
-      "loss": 0.8135,
+      "loss": 0.8508,
       "step": 10184
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.451385850096737e-05,
-      "loss": 0.8579,
+      "loss": 0.9081,
       "step": 10185
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4512807429681258e-05,
-      "loss": 0.6994,
+      "loss": 0.7991,
       "step": 10186
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.451175629578634e-05,
-      "loss": 0.7169,
+      "loss": 0.8782,
       "step": 10187
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4510705099297205e-05,
-      "loss": 0.821,
+      "loss": 0.9298,
       "step": 10188
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.450965384022844e-05,
-      "loss": 0.7291,
+      "loss": 0.8566,
       "step": 10189
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4508602518594622e-05,
-      "loss": 0.8703,
+      "loss": 0.9104,
       "step": 10190
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4507551134410343e-05,
-      "loss": 0.838,
+      "loss": 0.9051,
       "step": 10191
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4506499687690188e-05,
-      "loss": 0.765,
+      "loss": 0.8722,
       "step": 10192
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.450544817844874e-05,
-      "loss": 0.7354,
+      "loss": 0.9237,
       "step": 10193
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4504396606700595e-05,
-      "loss": 0.7646,
+      "loss": 0.8311,
       "step": 10194
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4503344972460335e-05,
-      "loss": 0.9315,
+      "loss": 0.9662,
       "step": 10195
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4502293275742556e-05,
-      "loss": 0.7522,
+      "loss": 0.8225,
       "step": 10196
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4501241516561843e-05,
-      "loss": 0.7638,
+      "loss": 0.8571,
       "step": 10197
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4500189694932791e-05,
-      "loss": 0.9234,
+      "loss": 0.9955,
       "step": 10198
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4499137810869995e-05,
-      "loss": 0.8813,
+      "loss": 0.8937,
       "step": 10199
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4498085864388042e-05,
-      "loss": 0.7573,
+      "loss": 0.8489,
       "step": 10200
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.449703385550153e-05,
-      "loss": 0.9218,
+      "loss": 0.9603,
       "step": 10201
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4495981784225056e-05,
-      "loss": 0.8466,
+      "loss": 0.9425,
       "step": 10202
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4494929650573211e-05,
-      "loss": 0.7055,
+      "loss": 0.8756,
       "step": 10203
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4493877454560594e-05,
-      "loss": 0.773,
+      "loss": 0.8332,
       "step": 10204
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4492825196201809e-05,
-      "loss": 0.7339,
+      "loss": 0.8676,
       "step": 10205
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4491772875511443e-05,
-      "loss": 0.8572,
+      "loss": 0.9238,
       "step": 10206
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4490720492504105e-05,
-      "loss": 0.7878,
+      "loss": 0.8999,
       "step": 10207
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.448966804719439e-05,
-      "loss": 0.7507,
+      "loss": 0.8905,
       "step": 10208
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4488615539596898e-05,
-      "loss": 0.779,
+      "loss": 0.8349,
       "step": 10209
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4487562969726238e-05,
-      "loss": 0.7421,
+      "loss": 0.8315,
       "step": 10210
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4486510337597009e-05,
-      "loss": 0.7343,
+      "loss": 0.8525,
       "step": 10211
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4485457643223811e-05,
-      "loss": 0.7453,
+      "loss": 0.8297,
       "step": 10212
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4484404886621255e-05,
-      "loss": 0.7787,
+      "loss": 0.8521,
       "step": 10213
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.448335206780394e-05,
-      "loss": 0.6621,
+      "loss": 0.8611,
       "step": 10214
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.448229918678648e-05,
-      "loss": 0.8493,
+      "loss": 0.9644,
       "step": 10215
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4481246243583477e-05,
-      "loss": 0.8247,
+      "loss": 0.915,
       "step": 10216
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.448019323820954e-05,
-      "loss": 0.761,
+      "loss": 0.9026,
       "step": 10217
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.447914017067928e-05,
-      "loss": 0.7241,
+      "loss": 0.8487,
       "step": 10218
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4478087041007305e-05,
-      "loss": 0.7219,
+      "loss": 0.894,
       "step": 10219
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4477033849208223e-05,
-      "loss": 0.7028,
+      "loss": 0.795,
       "step": 10220
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4475980595296653e-05,
-      "loss": 0.8222,
+      "loss": 0.9548,
       "step": 10221
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4474927279287201e-05,
-      "loss": 0.6769,
+      "loss": 0.8696,
       "step": 10222
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.447387390119448e-05,
-      "loss": 0.8474,
+      "loss": 0.8951,
       "step": 10223
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.447282046103311e-05,
-      "loss": 0.7181,
+      "loss": 0.9802,
       "step": 10224
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.44717669588177e-05,
-      "loss": 0.86,
+      "loss": 0.8618,
       "step": 10225
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4470713394562867e-05,
-      "loss": 0.8284,
+      "loss": 0.944,
       "step": 10226
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4469659768283233e-05,
-      "loss": 0.835,
+      "loss": 0.8479,
       "step": 10227
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4468606079993406e-05,
-      "loss": 0.8456,
+      "loss": 0.9317,
       "step": 10228
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4467552329708012e-05,
-      "loss": 0.647,
+      "loss": 0.812,
       "step": 10229
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4466498517441667e-05,
-      "loss": 0.8876,
+      "loss": 0.9213,
       "step": 10230
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4465444643208994e-05,
-      "loss": 0.827,
+      "loss": 0.9253,
       "step": 10231
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.446439070702461e-05,
-      "loss": 0.722,
+      "loss": 0.831,
       "step": 10232
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4463336708903142e-05,
-      "loss": 0.713,
+      "loss": 0.7968,
       "step": 10233
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4462282648859208e-05,
-      "loss": 0.754,
+      "loss": 0.8334,
       "step": 10234
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4461228526907434e-05,
-      "loss": 0.9127,
+      "loss": 0.9246,
       "step": 10235
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4460174343062443e-05,
-      "loss": 0.6099,
+      "loss": 0.7313,
       "step": 10236
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4459120097338861e-05,
-      "loss": 0.7153,
+      "loss": 0.809,
       "step": 10237
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4458065789751317e-05,
-      "loss": 0.7072,
+      "loss": 0.868,
       "step": 10238
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.445701142031443e-05,
-      "loss": 0.6774,
+      "loss": 0.8472,
       "step": 10239
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4455956989042839e-05,
-      "loss": 0.7187,
+      "loss": 0.9119,
       "step": 10240
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4454902495951162e-05,
-      "loss": 0.8084,
+      "loss": 0.8953,
       "step": 10241
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4453847941054037e-05,
-      "loss": 0.6798,
+      "loss": 0.7711,
       "step": 10242
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4452793324366093e-05,
-      "loss": 0.8505,
+      "loss": 0.9245,
       "step": 10243
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4451738645901957e-05,
-      "loss": 0.7188,
+      "loss": 0.8841,
       "step": 10244
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4450683905676261e-05,
-      "loss": 0.8065,
+      "loss": 0.8563,
       "step": 10245
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4449629103703642e-05,
-      "loss": 0.816,
+      "loss": 0.8999,
       "step": 10246
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4448574239998733e-05,
-      "loss": 0.7814,
+      "loss": 0.9337,
       "step": 10247
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4447519314576168e-05,
-      "loss": 0.7155,
+      "loss": 0.797,
       "step": 10248
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4446464327450582e-05,
-      "loss": 0.7262,
+      "loss": 0.8344,
       "step": 10249
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4445409278636613e-05,
-      "loss": 0.8521,
+      "loss": 0.955,
       "step": 10250
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4444354168148897e-05,
-      "loss": 0.7438,
+      "loss": 0.8025,
       "step": 10251
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.444329899600207e-05,
-      "loss": 0.8008,
+      "loss": 0.8844,
       "step": 10252
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.444224376221078e-05,
-      "loss": 0.7796,
+      "loss": 0.842,
       "step": 10253
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4441188466789658e-05,
-      "loss": 0.8345,
+      "loss": 0.964,
       "step": 10254
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4440133109753345e-05,
-      "loss": 0.7943,
+      "loss": 0.8843,
       "step": 10255
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4439077691116485e-05,
-      "loss": 0.7145,
+      "loss": 0.8083,
       "step": 10256
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4438022210893721e-05,
-      "loss": 0.8094,
+      "loss": 0.8901,
       "step": 10257
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4436966669099697e-05,
-      "loss": 0.8741,
+      "loss": 0.9281,
       "step": 10258
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4435911065749056e-05,
-      "loss": 0.7679,
+      "loss": 0.9731,
       "step": 10259
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4434855400856438e-05,
-      "loss": 0.8746,
+      "loss": 0.8776,
       "step": 10260
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4433799674436497e-05,
-      "loss": 0.8944,
+      "loss": 0.9904,
       "step": 10261
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4432743886503877e-05,
-      "loss": 0.7855,
+      "loss": 0.8988,
       "step": 10262
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4431688037073226e-05,
-      "loss": 0.6896,
+      "loss": 0.8754,
       "step": 10263
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4430632126159187e-05,
-      "loss": 0.7707,
+      "loss": 0.9215,
       "step": 10264
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4429576153776417e-05,
-      "loss": 0.7393,
+      "loss": 0.8242,
       "step": 10265
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.442852011993956e-05,
-      "loss": 0.7371,
+      "loss": 0.8912,
       "step": 10266
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4427464024663272e-05,
-      "loss": 0.7211,
+      "loss": 0.8051,
       "step": 10267
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.44264078679622e-05,
-      "loss": 0.8971,
+      "loss": 0.9563,
       "step": 10268
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4425351649851e-05,
-      "loss": 0.7246,
+      "loss": 0.8261,
       "step": 10269
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4424295370344324e-05,
-      "loss": 0.8195,
+      "loss": 0.874,
       "step": 10270
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4423239029456829e-05,
-      "loss": 0.794,
+      "loss": 0.7978,
       "step": 10271
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4422182627203167e-05,
-      "loss": 0.7908,
+      "loss": 0.8954,
       "step": 10272
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4421126163597996e-05,
-      "loss": 0.6198,
+      "loss": 0.787,
       "step": 10273
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.442006963865597e-05,
-      "loss": 0.7234,
+      "loss": 0.9329,
       "step": 10274
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4419013052391753e-05,
-      "loss": 0.8583,
+      "loss": 0.9035,
       "step": 10275
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4417956404819994e-05,
-      "loss": 0.6217,
+      "loss": 0.7401,
       "step": 10276
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4416899695955364e-05,
-      "loss": 0.8023,
+      "loss": 0.8857,
       "step": 10277
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4415842925812515e-05,
-      "loss": 0.694,
+      "loss": 0.8983,
       "step": 10278
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4414786094406111e-05,
-      "loss": 0.7575,
+      "loss": 0.8913,
       "step": 10279
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4413729201750812e-05,
-      "loss": 0.8242,
+      "loss": 0.9084,
       "step": 10280
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4412672247861285e-05,
-      "loss": 0.6785,
+      "loss": 0.7957,
       "step": 10281
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4411615232752191e-05,
-      "loss": 0.7846,
+      "loss": 0.8105,
       "step": 10282
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4410558156438194e-05,
-      "loss": 0.871,
+      "loss": 0.992,
       "step": 10283
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.440950101893396e-05,
-      "loss": 0.8844,
+      "loss": 0.9516,
       "step": 10284
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4408443820254157e-05,
-      "loss": 0.807,
+      "loss": 0.9046,
       "step": 10285
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4407386560413448e-05,
-      "loss": 0.9064,
+      "loss": 1.0567,
       "step": 10286
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4406329239426506e-05,
-      "loss": 0.7873,
+      "loss": 0.8757,
       "step": 10287
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4405271857307999e-05,
-      "loss": 0.7489,
+      "loss": 0.9391,
       "step": 10288
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.440421441407259e-05,
-      "loss": 0.8449,
+      "loss": 0.9547,
       "step": 10289
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4403156909734962e-05,
-      "loss": 0.8907,
+      "loss": 0.9623,
       "step": 10290
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4402099344309775e-05,
-      "loss": 0.7863,
+      "loss": 0.8368,
       "step": 10291
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4401041717811704e-05,
-      "loss": 0.9148,
+      "loss": 0.9252,
       "step": 10292
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4399984030255425e-05,
-      "loss": 0.7807,
+      "loss": 0.9008,
       "step": 10293
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4398926281655612e-05,
-      "loss": 0.829,
+      "loss": 0.8604,
       "step": 10294
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4397868472026934e-05,
-      "loss": 0.8102,
+      "loss": 0.8898,
       "step": 10295
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4396810601384075e-05,
-      "loss": 0.8023,
+      "loss": 0.9072,
       "step": 10296
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4395752669741705e-05,
-      "loss": 0.7501,
+      "loss": 0.8497,
       "step": 10297
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4394694677114504e-05,
-      "loss": 0.7967,
+      "loss": 0.8767,
       "step": 10298
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.439363662351715e-05,
-      "loss": 0.7762,
+      "loss": 0.9431,
       "step": 10299
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4392578508964319e-05,
-      "loss": 0.7753,
+      "loss": 0.8652,
       "step": 10300
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4391520333470699e-05,
-      "loss": 0.5935,
+      "loss": 0.7829,
       "step": 10301
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.439046209705096e-05,
-      "loss": 0.7945,
+      "loss": 0.888,
       "step": 10302
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.438940379971979e-05,
-      "loss": 0.8916,
+      "loss": 0.9532,
       "step": 10303
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.438834544149187e-05,
-      "loss": 0.8447,
+      "loss": 0.9805,
       "step": 10304
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4387287022381883e-05,
-      "loss": 0.6973,
+      "loss": 0.795,
       "step": 10305
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4386228542404517e-05,
-      "loss": 0.7287,
+      "loss": 0.8618,
       "step": 10306
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4385170001574448e-05,
-      "loss": 0.8959,
+      "loss": 0.9796,
       "step": 10307
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4384111399906369e-05,
-      "loss": 0.9237,
+      "loss": 1.0121,
       "step": 10308
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4383052737414966e-05,
-      "loss": 0.8544,
+      "loss": 0.9247,
       "step": 10309
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4381994014114924e-05,
-      "loss": 0.86,
+      "loss": 0.8785,
       "step": 10310
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.4380935230020932e-05,
-      "loss": 0.7099,
+      "loss": 0.8478,
       "step": 10311
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4379876385147679e-05,
-      "loss": 0.8111,
+      "loss": 0.9931,
       "step": 10312
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4378817479509854e-05,
-      "loss": 0.8098,
+      "loss": 0.9082,
       "step": 10313
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4377758513122153e-05,
-      "loss": 0.8925,
+      "loss": 0.924,
       "step": 10314
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4376699485999261e-05,
-      "loss": 0.8182,
+      "loss": 0.8814,
       "step": 10315
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4375640398155875e-05,
-      "loss": 0.8421,
+      "loss": 0.8651,
       "step": 10316
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4374581249606685e-05,
-      "loss": 0.8888,
+      "loss": 1.0002,
       "step": 10317
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4373522040366388e-05,
-      "loss": 0.6829,
+      "loss": 0.8286,
       "step": 10318
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4372462770449676e-05,
-      "loss": 0.7431,
+      "loss": 0.8209,
       "step": 10319
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4371403439871253e-05,
-      "loss": 0.8056,
+      "loss": 0.886,
       "step": 10320
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4370344048645804e-05,
-      "loss": 0.8561,
+      "loss": 0.9049,
       "step": 10321
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4369284596788031e-05,
-      "loss": 0.7542,
+      "loss": 0.8862,
       "step": 10322
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4368225084312637e-05,
-      "loss": 0.7493,
+      "loss": 0.8723,
       "step": 10323
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4367165511234316e-05,
-      "loss": 0.6507,
+      "loss": 0.8091,
       "step": 10324
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4366105877567768e-05,
-      "loss": 0.8731,
+      "loss": 0.979,
       "step": 10325
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.43650461833277e-05,
-      "loss": 0.8924,
+      "loss": 0.9785,
       "step": 10326
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4363986428528805e-05,
-      "loss": 0.7668,
+      "loss": 0.8568,
       "step": 10327
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4362926613185792e-05,
-      "loss": 0.8302,
+      "loss": 0.9378,
       "step": 10328
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4361866737313362e-05,
-      "loss": 0.8929,
+      "loss": 0.9993,
       "step": 10329
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.436080680092622e-05,
-      "loss": 0.8171,
+      "loss": 0.9111,
       "step": 10330
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4359746804039073e-05,
-      "loss": 0.8667,
+      "loss": 1.0072,
       "step": 10331
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.435868674666662e-05,
-      "loss": 0.9569,
+      "loss": 1.0145,
       "step": 10332
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4357626628823576e-05,
-      "loss": 0.8826,
+      "loss": 0.9622,
       "step": 10333
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4356566450524642e-05,
-      "loss": 0.9267,
+      "loss": 0.949,
       "step": 10334
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4355506211784534e-05,
-      "loss": 0.6841,
+      "loss": 0.8871,
       "step": 10335
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4354445912617954e-05,
-      "loss": 0.7736,
+      "loss": 0.9756,
       "step": 10336
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4353385553039618e-05,
-      "loss": 0.9491,
+      "loss": 0.9596,
       "step": 10337
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4352325133064229e-05,
-      "loss": 0.7927,
+      "loss": 0.8582,
       "step": 10338
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4351264652706509e-05,
-      "loss": 0.7528,
+      "loss": 0.8523,
       "step": 10339
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4350204111981164e-05,
-      "loss": 0.6068,
+      "loss": 0.7911,
       "step": 10340
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4349143510902908e-05,
-      "loss": 0.7044,
+      "loss": 0.8378,
       "step": 10341
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4348082849486456e-05,
-      "loss": 0.8247,
+      "loss": 0.9253,
       "step": 10342
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4347022127746526e-05,
-      "loss": 0.7668,
+      "loss": 0.8805,
       "step": 10343
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.434596134569783e-05,
-      "loss": 0.889,
+      "loss": 0.9365,
       "step": 10344
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4344900503355086e-05,
-      "loss": 0.8385,
+      "loss": 0.9656,
       "step": 10345
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4343839600733013e-05,
-      "loss": 1.0108,
+      "loss": 1.064,
       "step": 10346
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.434277863784633e-05,
-      "loss": 0.8111,
+      "loss": 0.8535,
       "step": 10347
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4341717614709752e-05,
-      "loss": 0.8664,
+      "loss": 0.9246,
       "step": 10348
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4340656531338004e-05,
-      "loss": 0.7691,
+      "loss": 0.9479,
       "step": 10349
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4339595387745807e-05,
-      "loss": 0.7886,
+      "loss": 0.8609,
       "step": 10350
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4338534183947876e-05,
-      "loss": 0.7835,
+      "loss": 0.9712,
       "step": 10351
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4337472919958943e-05,
-      "loss": 0.7459,
+      "loss": 0.8481,
       "step": 10352
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4336411595793724e-05,
-      "loss": 0.8193,
+      "loss": 0.8735,
       "step": 10353
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4335350211466949e-05,
-      "loss": 0.8786,
+      "loss": 0.9746,
       "step": 10354
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4334288766993343e-05,
-      "loss": 0.7253,
+      "loss": 0.8246,
       "step": 10355
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4333227262387626e-05,
-      "loss": 0.7634,
+      "loss": 0.8627,
       "step": 10356
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.433216569766453e-05,
-      "loss": 0.7465,
+      "loss": 0.8381,
       "step": 10357
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4331104072838784e-05,
-      "loss": 0.8408,
+      "loss": 0.8579,
       "step": 10358
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4330042387925111e-05,
-      "loss": 0.8064,
+      "loss": 0.9023,
       "step": 10359
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4328980642938244e-05,
-      "loss": 0.7762,
+      "loss": 0.9022,
       "step": 10360
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4327918837892911e-05,
-      "loss": 0.6689,
+      "loss": 0.8382,
       "step": 10361
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4326856972803848e-05,
-      "loss": 0.7246,
+      "loss": 0.8202,
       "step": 10362
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4325795047685782e-05,
-      "loss": 0.7238,
+      "loss": 0.7685,
       "step": 10363
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4324733062553446e-05,
-      "loss": 0.7369,
+      "loss": 0.821,
       "step": 10364
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4323671017421576e-05,
-      "loss": 0.6871,
+      "loss": 0.9033,
       "step": 10365
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4322608912304907e-05,
-      "loss": 0.8149,
+      "loss": 0.9734,
       "step": 10366
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4321546747218169e-05,
-      "loss": 0.8065,
+      "loss": 0.9063,
       "step": 10367
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4320484522176104e-05,
-      "loss": 0.7534,
+      "loss": 0.8731,
       "step": 10368
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4319422237193443e-05,
-      "loss": 0.7769,
+      "loss": 0.9189,
       "step": 10369
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.431835989228493e-05,
-      "loss": 0.7652,
+      "loss": 0.8777,
       "step": 10370
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.43172974874653e-05,
-      "loss": 0.7666,
+      "loss": 0.9179,
       "step": 10371
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.431623502274929e-05,
-      "loss": 0.6854,
+      "loss": 0.8064,
       "step": 10372
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4315172498151647e-05,
-      "loss": 0.7001,
+      "loss": 0.8577,
       "step": 10373
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4314109913687104e-05,
-      "loss": 0.7053,
+      "loss": 0.8204,
       "step": 10374
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4313047269370408e-05,
-      "loss": 0.8081,
+      "loss": 0.8976,
       "step": 10375
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4311984565216304e-05,
-      "loss": 0.8095,
+      "loss": 0.9017,
       "step": 10376
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.431092180123953e-05,
-      "loss": 0.8617,
+      "loss": 0.9301,
       "step": 10377
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4309858977454832e-05,
-      "loss": 0.8093,
+      "loss": 0.9327,
       "step": 10378
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4308796093876955e-05,
-      "loss": 0.8601,
+      "loss": 0.9405,
       "step": 10379
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4307733150520647e-05,
-      "loss": 0.8535,
+      "loss": 0.9292,
       "step": 10380
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4306670147400652e-05,
-      "loss": 0.7987,
+      "loss": 0.8796,
       "step": 10381
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4305607084531721e-05,
-      "loss": 0.6651,
+      "loss": 0.8724,
       "step": 10382
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4304543961928597e-05,
-      "loss": 0.7854,
+      "loss": 0.8586,
       "step": 10383
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4303480779606036e-05,
-      "loss": 1.0514,
+      "loss": 1.0508,
       "step": 10384
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4302417537578783e-05,
-      "loss": 0.7432,
+      "loss": 0.885,
       "step": 10385
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4301354235861594e-05,
-      "loss": 0.7832,
+      "loss": 0.9336,
       "step": 10386
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4300290874469215e-05,
-      "loss": 0.7406,
+      "loss": 0.777,
       "step": 10387
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.42992274534164e-05,
-      "loss": 0.7639,
+      "loss": 0.8058,
       "step": 10388
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4298163972717909e-05,
-      "loss": 0.7074,
+      "loss": 0.7976,
       "step": 10389
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4297100432388487e-05,
-      "loss": 0.8433,
+      "loss": 0.8631,
       "step": 10390
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4296036832442895e-05,
-      "loss": 0.7844,
+      "loss": 0.9039,
       "step": 10391
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4294973172895888e-05,
-      "loss": 0.6752,
+      "loss": 0.8672,
       "step": 10392
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.429390945376222e-05,
-      "loss": 0.8892,
+      "loss": 0.9545,
       "step": 10393
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.429284567505665e-05,
-      "loss": 0.7286,
+      "loss": 0.7878,
       "step": 10394
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4291781836793939e-05,
-      "loss": 0.7939,
+      "loss": 0.8937,
       "step": 10395
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4290717938988845e-05,
-      "loss": 0.8213,
+      "loss": 0.9434,
       "step": 10396
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4289653981656127e-05,
-      "loss": 0.6841,
+      "loss": 0.7927,
       "step": 10397
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4288589964810546e-05,
-      "loss": 0.6769,
+      "loss": 0.8637,
       "step": 10398
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4287525888466862e-05,
-      "loss": 0.629,
+      "loss": 0.7667,
       "step": 10399
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4286461752639839e-05,
-      "loss": 0.751,
+      "loss": 0.9248,
       "step": 10400
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4285397557344243e-05,
-      "loss": 0.7483,
+      "loss": 0.8694,
       "step": 10401
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4284333302594838e-05,
-      "loss": 0.7813,
+      "loss": 0.9187,
       "step": 10402
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4283268988406385e-05,
-      "loss": 0.6071,
+      "loss": 0.8075,
       "step": 10403
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4282204614793648e-05,
-      "loss": 0.7414,
+      "loss": 0.8579,
       "step": 10404
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4281140181771403e-05,
-      "loss": 0.6909,
+      "loss": 0.857,
       "step": 10405
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.428007568935441e-05,
-      "loss": 0.7777,
+      "loss": 0.877,
       "step": 10406
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.427901113755744e-05,
-      "loss": 0.8044,
+      "loss": 0.9007,
       "step": 10407
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4277946526395261e-05,
-      "loss": 0.7969,
+      "loss": 0.8302,
       "step": 10408
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4276881855882648e-05,
-      "loss": 0.6246,
+      "loss": 0.7492,
       "step": 10409
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4275817126034358e-05,
-      "loss": 0.929,
+      "loss": 1.0098,
       "step": 10410
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.427475233686518e-05,
-      "loss": 0.6006,
+      "loss": 0.7843,
       "step": 10411
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4273687488389873e-05,
-      "loss": 0.8752,
+      "loss": 0.9267,
       "step": 10412
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4272622580623221e-05,
-      "loss": 0.8363,
+      "loss": 0.9167,
       "step": 10413
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4271557613579989e-05,
-      "loss": 0.757,
+      "loss": 0.8885,
       "step": 10414
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4270492587274957e-05,
-      "loss": 0.762,
+      "loss": 0.864,
       "step": 10415
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4269427501722902e-05,
-      "loss": 0.815,
+      "loss": 0.9936,
       "step": 10416
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4268362356938595e-05,
-      "loss": 0.8081,
+      "loss": 0.9247,
       "step": 10417
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4267297152936816e-05,
-      "loss": 0.7211,
+      "loss": 0.8626,
       "step": 10418
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4266231889732345e-05,
-      "loss": 0.8252,
+      "loss": 0.9487,
       "step": 10419
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4265166567339956e-05,
-      "loss": 0.7189,
+      "loss": 0.8489,
       "step": 10420
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4264101185774438e-05,
-      "loss": 0.7527,
+      "loss": 0.9532,
       "step": 10421
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4263035745050561e-05,
-      "loss": 0.8715,
+      "loss": 0.9816,
       "step": 10422
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4261970245183112e-05,
-      "loss": 0.8225,
+      "loss": 0.8919,
       "step": 10423
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4260904686186876e-05,
-      "loss": 0.8231,
+      "loss": 0.8866,
       "step": 10424
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4259839068076628e-05,
-      "loss": 0.7952,
+      "loss": 0.8745,
       "step": 10425
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4258773390867163e-05,
-      "loss": 0.6582,
+      "loss": 0.8438,
       "step": 10426
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4257707654573257e-05,
-      "loss": 0.8455,
+      "loss": 0.9579,
       "step": 10427
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4256641859209697e-05,
-      "loss": 0.8218,
+      "loss": 0.907,
       "step": 10428
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4255576004791271e-05,
-      "loss": 0.6546,
+      "loss": 0.7788,
       "step": 10429
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4254510091332767e-05,
-      "loss": 0.7298,
+      "loss": 0.8065,
       "step": 10430
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4253444118848968e-05,
-      "loss": 0.8816,
+      "loss": 0.9688,
       "step": 10431
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4252378087354673e-05,
-      "loss": 0.7942,
+      "loss": 0.8969,
       "step": 10432
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4251311996864662e-05,
-      "loss": 0.8492,
+      "loss": 0.8645,
       "step": 10433
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4250245847393727e-05,
-      "loss": 0.5518,
+      "loss": 0.8849,
       "step": 10434
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4249179638956668e-05,
-      "loss": 0.7494,
+      "loss": 0.8769,
       "step": 10435
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4248113371568262e-05,
-      "loss": 0.7748,
+      "loss": 0.8272,
       "step": 10436
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4247047045243314e-05,
-      "loss": 0.6427,
+      "loss": 0.7805,
       "step": 10437
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4245980659996615e-05,
-      "loss": 0.7268,
+      "loss": 0.8116,
       "step": 10438
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4244914215842955e-05,
-      "loss": 0.7212,
+      "loss": 0.8174,
       "step": 10439
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4243847712797137e-05,
-      "loss": 0.7246,
+      "loss": 0.9052,
       "step": 10440
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.424278115087395e-05,
-      "loss": 0.783,
+      "loss": 0.8183,
       "step": 10441
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4241714530088197e-05,
-      "loss": 0.8149,
+      "loss": 0.9,
       "step": 10442
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4240647850454671e-05,
-      "loss": 0.7655,
+      "loss": 0.8784,
       "step": 10443
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4239581111988172e-05,
-      "loss": 0.772,
+      "loss": 0.8776,
       "step": 10444
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4238514314703505e-05,
-      "loss": 0.706,
+      "loss": 0.8095,
       "step": 10445
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4237447458615461e-05,
-      "loss": 0.822,
+      "loss": 0.91,
       "step": 10446
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4236380543738847e-05,
-      "loss": 0.7229,
+      "loss": 0.8683,
       "step": 10447
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4235313570088466e-05,
-      "loss": 0.7932,
+      "loss": 0.8895,
       "step": 10448
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4234246537679114e-05,
-      "loss": 0.754,
+      "loss": 0.9182,
       "step": 10449
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.42331794465256e-05,
-      "loss": 0.8016,
+      "loss": 0.9445,
       "step": 10450
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4232112296642729e-05,
-      "loss": 0.9924,
+      "loss": 1.0475,
       "step": 10451
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4231045088045303e-05,
-      "loss": 0.8008,
+      "loss": 0.8733,
       "step": 10452
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.422997782074813e-05,
-      "loss": 0.7481,
+      "loss": 0.8327,
       "step": 10453
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.422891049476602e-05,
-      "loss": 0.7903,
+      "loss": 0.8038,
       "step": 10454
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4227843110113772e-05,
-      "loss": 0.9417,
+      "loss": 0.9925,
       "step": 10455
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.42267756668062e-05,
-      "loss": 0.8164,
+      "loss": 0.9704,
       "step": 10456
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4225708164858114e-05,
-      "loss": 0.868,
+      "loss": 0.926,
       "step": 10457
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4224640604284321e-05,
-      "loss": 0.7836,
+      "loss": 0.8917,
       "step": 10458
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4223572985099637e-05,
-      "loss": 0.8427,
+      "loss": 0.9067,
       "step": 10459
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4222505307318866e-05,
-      "loss": 0.9142,
+      "loss": 0.9514,
       "step": 10460
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4221437570956826e-05,
-      "loss": 0.6329,
+      "loss": 0.8675,
       "step": 10461
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.422036977602833e-05,
-      "loss": 0.7936,
+      "loss": 0.8658,
       "step": 10462
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4219301922548192e-05,
-      "loss": 0.8145,
+      "loss": 0.9088,
       "step": 10463
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4218234010531227e-05,
-      "loss": 0.6045,
+      "loss": 0.7369,
       "step": 10464
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4217166039992248e-05,
-      "loss": 0.6965,
+      "loss": 0.8819,
       "step": 10465
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4216098010946073e-05,
-      "loss": 0.7752,
+      "loss": 0.9176,
       "step": 10466
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4215029923407522e-05,
-      "loss": 0.594,
+      "loss": 0.7759,
       "step": 10467
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4213961777391407e-05,
-      "loss": 0.7881,
+      "loss": 0.8953,
       "step": 10468
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4212893572912557e-05,
-      "loss": 0.6796,
+      "loss": 0.7634,
       "step": 10469
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4211825309985783e-05,
-      "loss": 0.8817,
+      "loss": 0.9329,
       "step": 10470
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4210756988625906e-05,
-      "loss": 0.7423,
+      "loss": 0.8096,
       "step": 10471
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4209688608847754e-05,
-      "loss": 0.8167,
+      "loss": 0.8793,
       "step": 10472
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4208620170666146e-05,
-      "loss": 0.596,
+      "loss": 0.8397,
       "step": 10473
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4207551674095903e-05,
-      "loss": 0.7755,
+      "loss": 0.947,
       "step": 10474
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.420648311915185e-05,
-      "loss": 0.8097,
+      "loss": 0.8987,
       "step": 10475
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.420541450584881e-05,
-      "loss": 0.7135,
+      "loss": 0.85,
       "step": 10476
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4204345834201613e-05,
-      "loss": 0.7143,
+      "loss": 0.8746,
       "step": 10477
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4203277104225084e-05,
-      "loss": 0.8691,
+      "loss": 0.9471,
       "step": 10478
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4202208315934044e-05,
-      "loss": 0.6659,
+      "loss": 0.7686,
       "step": 10479
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.420113946934333e-05,
-      "loss": 0.8661,
+      "loss": 0.8596,
       "step": 10480
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4200070564467766e-05,
-      "loss": 0.6793,
+      "loss": 0.8679,
       "step": 10481
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.419900160132218e-05,
-      "loss": 0.7075,
+      "loss": 0.8454,
       "step": 10482
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4197932579921408e-05,
-      "loss": 0.8124,
+      "loss": 0.8666,
       "step": 10483
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4196863500280271e-05,
-      "loss": 0.6625,
+      "loss": 0.892,
       "step": 10484
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4195794362413615e-05,
-      "loss": 0.9181,
+      "loss": 0.981,
       "step": 10485
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4194725166336261e-05,
-      "loss": 0.7253,
+      "loss": 0.8151,
       "step": 10486
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4193655912063048e-05,
-      "loss": 0.8433,
+      "loss": 0.9254,
       "step": 10487
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4192586599608808e-05,
-      "loss": 0.8137,
+      "loss": 0.9192,
       "step": 10488
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.419151722898838e-05,
-      "loss": 0.8151,
+      "loss": 0.8912,
       "step": 10489
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4190447800216593e-05,
-      "loss": 0.6454,
+      "loss": 0.8449,
       "step": 10490
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4189378313308293e-05,
-      "loss": 0.8211,
+      "loss": 0.8328,
       "step": 10491
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4188308768278312e-05,
-      "loss": 0.7759,
+      "loss": 0.8626,
       "step": 10492
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.418723916514149e-05,
-      "loss": 0.8849,
+      "loss": 0.8878,
       "step": 10493
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4186169503912663e-05,
-      "loss": 0.7719,
+      "loss": 0.8772,
       "step": 10494
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4185099784606673e-05,
-      "loss": 1.0071,
+      "loss": 1.0222,
       "step": 10495
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4184030007238363e-05,
-      "loss": 0.804,
+      "loss": 0.8886,
       "step": 10496
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4182960171822572e-05,
-      "loss": 0.7588,
+      "loss": 0.8616,
       "step": 10497
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4181890278374144e-05,
-      "loss": 0.7905,
+      "loss": 0.839,
       "step": 10498
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4180820326907922e-05,
-      "loss": 0.7398,
+      "loss": 0.9305,
       "step": 10499
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4179750317438749e-05,
-      "loss": 0.7205,
+      "loss": 0.8703,
       "step": 10500
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4178680249981472e-05,
-      "loss": 0.6758,
+      "loss": 0.7947,
       "step": 10501
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4177610124550935e-05,
-      "loss": 0.8164,
+      "loss": 0.9315,
       "step": 10502
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4176539941161985e-05,
-      "loss": 0.726,
+      "loss": 0.8365,
       "step": 10503
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4175469699829467e-05,
-      "loss": 0.7507,
+      "loss": 0.862,
       "step": 10504
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4174399400568233e-05,
-      "loss": 0.7113,
+      "loss": 0.8291,
       "step": 10505
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.417332904339313e-05,
-      "loss": 0.716,
+      "loss": 0.8401,
       "step": 10506
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4172258628319006e-05,
-      "loss": 0.8032,
+      "loss": 0.8273,
       "step": 10507
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4171188155360713e-05,
-      "loss": 0.9063,
+      "loss": 0.8652,
       "step": 10508
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4170117624533104e-05,
-      "loss": 0.7509,
+      "loss": 0.9227,
       "step": 10509
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4169047035851032e-05,
-      "loss": 0.6778,
+      "loss": 0.8086,
       "step": 10510
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4167976389329345e-05,
-      "loss": 0.704,
+      "loss": 0.7944,
       "step": 10511
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4166905684982898e-05,
-      "loss": 0.8236,
+      "loss": 0.9523,
       "step": 10512
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4165834922826547e-05,
-      "loss": 0.7686,
+      "loss": 0.8888,
       "step": 10513
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.416476410287515e-05,
-      "loss": 0.8879,
+      "loss": 0.9307,
       "step": 10514
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4163693225143558e-05,
-      "loss": 0.8994,
+      "loss": 0.9847,
       "step": 10515
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4162622289646629e-05,
-      "loss": 0.8371,
+      "loss": 0.8774,
       "step": 10516
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4161551296399223e-05,
-      "loss": 0.761,
+      "loss": 0.8555,
       "step": 10517
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4160480245416197e-05,
-      "loss": 0.8744,
+      "loss": 0.9864,
       "step": 10518
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.415940913671241e-05,
-      "loss": 0.7376,
+      "loss": 0.8649,
       "step": 10519
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4158337970302722e-05,
-      "loss": 0.8703,
+      "loss": 0.8847,
       "step": 10520
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4157266746201996e-05,
-      "loss": 0.8539,
+      "loss": 0.9797,
       "step": 10521
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4156195464425092e-05,
-      "loss": 0.8421,
+      "loss": 0.9019,
       "step": 10522
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4155124124986874e-05,
-      "loss": 0.8602,
+      "loss": 0.9485,
       "step": 10523
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4154052727902201e-05,
-      "loss": 0.8391,
+      "loss": 0.8588,
       "step": 10524
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4152981273185942e-05,
-      "loss": 0.7683,
+      "loss": 0.8984,
       "step": 10525
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4151909760852963e-05,
-      "loss": 0.7697,
+      "loss": 0.8712,
       "step": 10526
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4150838190918122e-05,
-      "loss": 0.7737,
+      "loss": 0.9077,
       "step": 10527
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4149766563396292e-05,
-      "loss": 0.9236,
+      "loss": 1.0495,
       "step": 10528
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.414869487830234e-05,
-      "loss": 0.6688,
+      "loss": 0.8206,
       "step": 10529
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4147623135651132e-05,
-      "loss": 0.6562,
+      "loss": 0.8663,
       "step": 10530
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.414655133545754e-05,
-      "loss": 0.6428,
+      "loss": 0.7755,
       "step": 10531
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4145479477736426e-05,
-      "loss": 0.8547,
+      "loss": 0.8702,
       "step": 10532
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4144407562502668e-05,
-      "loss": 0.8375,
+      "loss": 0.9674,
       "step": 10533
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.414333558977114e-05,
-      "loss": 0.8609,
+      "loss": 0.886,
       "step": 10534
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4142263559556705e-05,
-      "loss": 0.805,
+      "loss": 0.8735,
       "step": 10535
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.414119147187424e-05,
-      "loss": 0.8063,
+      "loss": 0.9836,
       "step": 10536
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.414011932673862e-05,
-      "loss": 0.6227,
+      "loss": 0.7491,
       "step": 10537
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4139047124164715e-05,
-      "loss": 0.92,
+      "loss": 0.9496,
       "step": 10538
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.413797486416741e-05,
-      "loss": 0.8101,
+      "loss": 0.8753,
       "step": 10539
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4136902546761569e-05,
-      "loss": 0.8159,
+      "loss": 0.8237,
       "step": 10540
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4135830171962077e-05,
-      "loss": 0.7433,
+      "loss": 0.9238,
       "step": 10541
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.413475773978381e-05,
-      "loss": 0.7926,
+      "loss": 0.8458,
       "step": 10542
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4133685250241644e-05,
-      "loss": 0.8797,
+      "loss": 0.9383,
       "step": 10543
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4132612703350464e-05,
-      "loss": 0.6443,
+      "loss": 0.849,
       "step": 10544
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4131540099125143e-05,
-      "loss": 0.672,
+      "loss": 0.8447,
       "step": 10545
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4130467437580564e-05,
-      "loss": 0.7549,
+      "loss": 0.8267,
       "step": 10546
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4129394718731612e-05,
-      "loss": 0.8333,
+      "loss": 0.8679,
       "step": 10547
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4128321942593165e-05,
-      "loss": 0.8463,
+      "loss": 0.9734,
       "step": 10548
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.412724910918011e-05,
-      "loss": 0.5597,
+      "loss": 0.7924,
       "step": 10549
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4126176218507332e-05,
-      "loss": 0.8359,
+      "loss": 0.8775,
       "step": 10550
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4125103270589708e-05,
-      "loss": 0.824,
+      "loss": 0.8429,
       "step": 10551
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4124030265442132e-05,
-      "loss": 0.8744,
+      "loss": 0.9268,
       "step": 10552
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4122957203079489e-05,
-      "loss": 0.9162,
+      "loss": 0.9517,
       "step": 10553
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.412188408351666e-05,
-      "loss": 0.7429,
+      "loss": 0.8469,
       "step": 10554
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4120810906768541e-05,
-      "loss": 0.6854,
+      "loss": 0.849,
       "step": 10555
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4119737672850013e-05,
-      "loss": 0.6889,
+      "loss": 0.8279,
       "step": 10556
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4118664381775973e-05,
-      "loss": 0.8043,
+      "loss": 0.9389,
       "step": 10557
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.411759103356131e-05,
-      "loss": 0.7257,
+      "loss": 0.892,
       "step": 10558
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4116517628220909e-05,
-      "loss": 0.7666,
+      "loss": 0.8854,
       "step": 10559
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4115444165769671e-05,
-      "loss": 0.7583,
+      "loss": 0.8717,
       "step": 10560
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4114370646222484e-05,
-      "loss": 0.7219,
+      "loss": 0.9259,
       "step": 10561
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4113297069594239e-05,
-      "loss": 0.7806,
+      "loss": 0.9124,
       "step": 10562
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4112223435899834e-05,
-      "loss": 0.6353,
+      "loss": 0.7861,
       "step": 10563
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4111149745154164e-05,
-      "loss": 0.7846,
+      "loss": 0.9144,
       "step": 10564
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4110075997372124e-05,
-      "loss": 0.8061,
+      "loss": 0.8625,
       "step": 10565
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4109002192568611e-05,
-      "loss": 0.7159,
+      "loss": 0.887,
       "step": 10566
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.410792833075852e-05,
-      "loss": 0.8036,
+      "loss": 0.893,
       "step": 10567
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4106854411956757e-05,
-      "loss": 0.757,
+      "loss": 0.9149,
       "step": 10568
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4105780436178212e-05,
-      "loss": 0.7996,
+      "loss": 0.899,
       "step": 10569
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4104706403437792e-05,
-      "loss": 0.7527,
+      "loss": 0.957,
       "step": 10570
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4103632313750391e-05,
-      "loss": 0.7593,
+      "loss": 0.8717,
       "step": 10571
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4102558167130915e-05,
-      "loss": 0.9126,
+      "loss": 1.0169,
       "step": 10572
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4101483963594266e-05,
-      "loss": 0.7598,
+      "loss": 0.8679,
       "step": 10573
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4100409703155347e-05,
-      "loss": 0.8463,
+      "loss": 0.9709,
       "step": 10574
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4099335385829059e-05,
-      "loss": 0.8374,
+      "loss": 0.9692,
       "step": 10575
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.409826101163031e-05,
-      "loss": 0.7647,
+      "loss": 0.9133,
       "step": 10576
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4097186580574003e-05,
-      "loss": 0.678,
+      "loss": 0.8342,
       "step": 10577
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4096112092675045e-05,
-      "loss": 0.8174,
+      "loss": 0.9316,
       "step": 10578
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4095037547948345e-05,
-      "loss": 0.8079,
+      "loss": 0.8931,
       "step": 10579
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4093962946408808e-05,
-      "loss": 0.7978,
+      "loss": 0.976,
       "step": 10580
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4092888288071344e-05,
-      "loss": 0.7338,
+      "loss": 0.8671,
       "step": 10581
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4091813572950863e-05,
-      "loss": 0.804,
+      "loss": 0.8998,
       "step": 10582
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4090738801062271e-05,
-      "loss": 0.8465,
+      "loss": 0.8887,
       "step": 10583
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4089663972420487e-05,
-      "loss": 0.6823,
+      "loss": 0.8024,
       "step": 10584
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4088589087040414e-05,
-      "loss": 0.7348,
+      "loss": 0.9045,
       "step": 10585
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.4087514144936968e-05,
-      "loss": 0.7178,
+      "loss": 0.8692,
       "step": 10586
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4086439146125068e-05,
-      "loss": 0.7824,
+      "loss": 0.8378,
       "step": 10587
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.408536409061962e-05,
-      "loss": 0.7941,
+      "loss": 0.7944,
       "step": 10588
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.408428897843554e-05,
-      "loss": 0.734,
+      "loss": 0.8523,
       "step": 10589
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4083213809587748e-05,
-      "loss": 0.7872,
+      "loss": 0.9155,
       "step": 10590
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4082138584091155e-05,
-      "loss": 0.765,
+      "loss": 0.8627,
       "step": 10591
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.408106330196068e-05,
-      "loss": 0.6737,
+      "loss": 0.8173,
       "step": 10592
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4079987963211246e-05,
-      "loss": 0.626,
+      "loss": 0.8655,
       "step": 10593
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4078912567857764e-05,
-      "loss": 0.698,
+      "loss": 0.7462,
       "step": 10594
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4077837115915158e-05,
-      "loss": 0.828,
+      "loss": 0.9037,
       "step": 10595
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4076761607398348e-05,
-      "loss": 0.8504,
+      "loss": 0.8921,
       "step": 10596
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4075686042322257e-05,
-      "loss": 0.8708,
+      "loss": 0.9215,
       "step": 10597
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4074610420701805e-05,
-      "loss": 0.7543,
+      "loss": 0.9103,
       "step": 10598
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4073534742551909e-05,
-      "loss": 0.749,
+      "loss": 0.8898,
       "step": 10599
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4072459007887502e-05,
-      "loss": 0.7304,
+      "loss": 0.8208,
       "step": 10600
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4071383216723504e-05,
-      "loss": 0.7696,
+      "loss": 0.931,
       "step": 10601
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4070307369074841e-05,
-      "loss": 0.6694,
+      "loss": 0.8408,
       "step": 10602
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4069231464956436e-05,
-      "loss": 0.8288,
+      "loss": 0.8386,
       "step": 10603
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4068155504383223e-05,
-      "loss": 0.6992,
+      "loss": 0.845,
       "step": 10604
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4067079487370119e-05,
-      "loss": 0.7288,
+      "loss": 0.8374,
       "step": 10605
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4066003413932059e-05,
-      "loss": 0.6598,
+      "loss": 0.8596,
       "step": 10606
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4064927284083973e-05,
-      "loss": 0.7247,
+      "loss": 0.8221,
       "step": 10607
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4063851097840786e-05,
-      "loss": 0.7117,
+      "loss": 0.7985,
       "step": 10608
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4062774855217431e-05,
-      "loss": 0.6828,
+      "loss": 0.8085,
       "step": 10609
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.406169855622884e-05,
-      "loss": 0.8549,
+      "loss": 0.9855,
       "step": 10610
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4060622200889944e-05,
-      "loss": 0.8583,
+      "loss": 0.9085,
       "step": 10611
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4059545789215676e-05,
-      "loss": 0.7412,
+      "loss": 0.8995,
       "step": 10612
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4058469321220968e-05,
-      "loss": 0.7504,
+      "loss": 0.8156,
       "step": 10613
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4057392796920762e-05,
-      "loss": 0.812,
+      "loss": 0.8637,
       "step": 10614
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4056316216329982e-05,
-      "loss": 0.6957,
+      "loss": 0.8269,
       "step": 10615
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4055239579463571e-05,
-      "loss": 0.8478,
+      "loss": 0.942,
       "step": 10616
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4054162886336466e-05,
-      "loss": 0.9501,
+      "loss": 0.9796,
       "step": 10617
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.40530861369636e-05,
-      "loss": 0.755,
+      "loss": 0.8352,
       "step": 10618
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4052009331359919e-05,
-      "loss": 0.7538,
+      "loss": 0.912,
       "step": 10619
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4050932469540355e-05,
-      "loss": 0.8308,
+      "loss": 0.9309,
       "step": 10620
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4049855551519848e-05,
-      "loss": 0.7242,
+      "loss": 0.8688,
       "step": 10621
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4048778577313345e-05,
-      "loss": 0.7728,
+      "loss": 0.7833,
       "step": 10622
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4047701546935781e-05,
-      "loss": 0.7199,
+      "loss": 0.9787,
       "step": 10623
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4046624460402103e-05,
-      "loss": 0.7493,
+      "loss": 0.8235,
       "step": 10624
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4045547317727252e-05,
-      "loss": 0.8349,
+      "loss": 0.9464,
       "step": 10625
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4044470118926174e-05,
-      "loss": 0.8253,
+      "loss": 0.9335,
       "step": 10626
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.404339286401381e-05,
-      "loss": 0.871,
+      "loss": 0.9869,
       "step": 10627
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4042315553005104e-05,
-      "loss": 0.6731,
+      "loss": 0.8545,
       "step": 10628
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4041238185915007e-05,
-      "loss": 0.6941,
+      "loss": 0.876,
       "step": 10629
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4040160762758463e-05,
-      "loss": 0.714,
+      "loss": 0.8848,
       "step": 10630
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4039083283550421e-05,
-      "loss": 0.6863,
+      "loss": 0.8477,
       "step": 10631
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.403800574830583e-05,
-      "loss": 0.6708,
+      "loss": 0.8166,
       "step": 10632
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4036928157039638e-05,
-      "loss": 0.8765,
+      "loss": 0.9418,
       "step": 10633
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4035850509766795e-05,
-      "loss": 0.6727,
+      "loss": 0.8727,
       "step": 10634
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4034772806502253e-05,
-      "loss": 0.6857,
+      "loss": 0.8474,
       "step": 10635
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4033695047260963e-05,
-      "loss": 0.7898,
+      "loss": 0.8805,
       "step": 10636
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4032617232057874e-05,
-      "loss": 0.7826,
+      "loss": 0.9057,
       "step": 10637
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4031539360907946e-05,
-      "loss": 0.7475,
+      "loss": 0.8807,
       "step": 10638
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4030461433826127e-05,
-      "loss": 0.7313,
+      "loss": 0.8778,
       "step": 10639
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4029383450827379e-05,
-      "loss": 0.7588,
+      "loss": 0.8536,
       "step": 10640
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4028305411926648e-05,
-      "loss": 0.6646,
+      "loss": 0.8393,
       "step": 10641
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4027227317138896e-05,
-      "loss": 0.8095,
+      "loss": 0.9821,
       "step": 10642
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.402614916647908e-05,
-      "loss": 0.7672,
+      "loss": 0.8886,
       "step": 10643
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4025070959962155e-05,
-      "loss": 0.7882,
+      "loss": 0.8275,
       "step": 10644
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4023992697603084e-05,
-      "loss": 0.789,
+      "loss": 0.8523,
       "step": 10645
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.402291437941682e-05,
-      "loss": 0.8884,
+      "loss": 0.9807,
       "step": 10646
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4021836005418328e-05,
-      "loss": 0.7345,
+      "loss": 0.9525,
       "step": 10647
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.402075757562257e-05,
-      "loss": 0.6829,
+      "loss": 0.8066,
       "step": 10648
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4019679090044503e-05,
-      "loss": 0.7336,
+      "loss": 0.8323,
       "step": 10649
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.401860054869909e-05,
-      "loss": 0.8376,
+      "loss": 0.9327,
       "step": 10650
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.40175219516013e-05,
-      "loss": 0.7323,
+      "loss": 0.9162,
       "step": 10651
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4016443298766089e-05,
-      "loss": 0.8723,
+      "loss": 0.9407,
       "step": 10652
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4015364590208429e-05,
-      "loss": 0.5963,
+      "loss": 0.7915,
       "step": 10653
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4014285825943282e-05,
-      "loss": 0.7981,
+      "loss": 0.9422,
       "step": 10654
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4013207005985612e-05,
-      "loss": 0.5977,
+      "loss": 0.7315,
       "step": 10655
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.401212813035039e-05,
-      "loss": 0.8224,
+      "loss": 0.9047,
       "step": 10656
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4011049199052586e-05,
-      "loss": 0.7866,
+      "loss": 0.8951,
       "step": 10657
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4009970212107163e-05,
-      "loss": 0.8688,
+      "loss": 0.9069,
       "step": 10658
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.400889116952909e-05,
-      "loss": 0.6936,
+      "loss": 0.7931,
       "step": 10659
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4007812071333345e-05,
-      "loss": 0.7092,
+      "loss": 0.8851,
       "step": 10660
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.400673291753489e-05,
-      "loss": 0.6894,
+      "loss": 0.894,
       "step": 10661
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4005653708148702e-05,
-      "loss": 0.792,
+      "loss": 0.8659,
       "step": 10662
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4004574443189751e-05,
-      "loss": 0.6111,
+      "loss": 0.8064,
       "step": 10663
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4003495122673012e-05,
-      "loss": 0.8769,
+      "loss": 0.9435,
       "step": 10664
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4002415746613462e-05,
-      "loss": 0.6599,
+      "loss": 0.8218,
       "step": 10665
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4001336315026066e-05,
-      "loss": 0.7091,
+      "loss": 0.8152,
       "step": 10666
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.4000256827925812e-05,
-      "loss": 0.6878,
+      "loss": 0.8695,
       "step": 10667
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3999177285327665e-05,
-      "loss": 0.7756,
+      "loss": 0.7959,
       "step": 10668
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3998097687246608e-05,
-      "loss": 0.6335,
+      "loss": 0.7653,
       "step": 10669
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3997018033697621e-05,
-      "loss": 0.8652,
+      "loss": 0.9332,
       "step": 10670
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3995938324695677e-05,
-      "loss": 0.8238,
+      "loss": 0.9017,
       "step": 10671
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3994858560255759e-05,
-      "loss": 0.6349,
+      "loss": 0.7468,
       "step": 10672
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3993778740392847e-05,
-      "loss": 0.7732,
+      "loss": 0.9193,
       "step": 10673
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.399269886512192e-05,
-      "loss": 0.7566,
+      "loss": 0.8182,
       "step": 10674
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3991618934457963e-05,
-      "loss": 0.6829,
+      "loss": 0.851,
       "step": 10675
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3990538948415957e-05,
-      "loss": 0.764,
+      "loss": 0.9003,
       "step": 10676
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3989458907010882e-05,
-      "loss": 0.779,
+      "loss": 0.7699,
       "step": 10677
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3988378810257727e-05,
-      "loss": 0.791,
+      "loss": 0.9177,
       "step": 10678
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3987298658171473e-05,
-      "loss": 0.7544,
+      "loss": 0.9308,
       "step": 10679
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.398621845076711e-05,
-      "loss": 0.7928,
+      "loss": 0.8827,
       "step": 10680
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3985138188059622e-05,
-      "loss": 0.6701,
+      "loss": 0.8399,
       "step": 10681
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3984057870063991e-05,
-      "loss": 0.8134,
+      "loss": 0.9026,
       "step": 10682
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3982977496795215e-05,
-      "loss": 0.6273,
+      "loss": 0.7747,
       "step": 10683
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3981897068268278e-05,
-      "loss": 0.7235,
+      "loss": 0.8321,
       "step": 10684
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3980816584498166e-05,
-      "loss": 0.9583,
+      "loss": 1.0556,
       "step": 10685
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3979736045499872e-05,
-      "loss": 0.7649,
+      "loss": 0.8268,
       "step": 10686
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3978655451288386e-05,
-      "loss": 0.7519,
+      "loss": 0.8239,
       "step": 10687
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3977574801878702e-05,
-      "loss": 0.863,
+      "loss": 0.9228,
       "step": 10688
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3976494097285811e-05,
-      "loss": 0.7096,
+      "loss": 0.8384,
       "step": 10689
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3975413337524705e-05,
-      "loss": 0.6654,
+      "loss": 0.8351,
       "step": 10690
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.397433252261038e-05,
-      "loss": 0.769,
+      "loss": 0.9437,
       "step": 10691
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3973251652557831e-05,
-      "loss": 0.8436,
+      "loss": 0.8936,
       "step": 10692
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.397217072738205e-05,
-      "loss": 0.9242,
+      "loss": 0.9893,
       "step": 10693
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3971089747098039e-05,
-      "loss": 0.7396,
+      "loss": 0.8502,
       "step": 10694
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3970008711720791e-05,
-      "loss": 0.822,
+      "loss": 1.0065,
       "step": 10695
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3968927621265305e-05,
-      "loss": 0.651,
+      "loss": 0.784,
       "step": 10696
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3967846475746578e-05,
-      "loss": 0.7603,
+      "loss": 0.8259,
       "step": 10697
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3966765275179613e-05,
-      "loss": 0.7919,
+      "loss": 0.9478,
       "step": 10698
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3965684019579406e-05,
-      "loss": 0.8688,
+      "loss": 0.9684,
       "step": 10699
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3964602708960962e-05,
-      "loss": 0.8488,
+      "loss": 0.907,
       "step": 10700
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3963521343339278e-05,
-      "loss": 0.6702,
+      "loss": 0.7992,
       "step": 10701
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.396243992272936e-05,
-      "loss": 0.798,
+      "loss": 0.8798,
       "step": 10702
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3961358447146211e-05,
-      "loss": 0.8543,
+      "loss": 0.926,
       "step": 10703
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3960276916604834e-05,
-      "loss": 0.7457,
+      "loss": 0.8548,
       "step": 10704
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3959195331120232e-05,
-      "loss": 0.7948,
+      "loss": 0.8949,
       "step": 10705
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3958113690707413e-05,
-      "loss": 0.7277,
+      "loss": 0.9391,
       "step": 10706
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3957031995381383e-05,
-      "loss": 0.8378,
+      "loss": 0.9087,
       "step": 10707
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3955950245157145e-05,
-      "loss": 0.7442,
+      "loss": 0.8958,
       "step": 10708
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3954868440049713e-05,
-      "loss": 0.8447,
+      "loss": 0.9289,
       "step": 10709
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3953786580074094e-05,
-      "loss": 0.8247,
+      "loss": 0.9376,
       "step": 10710
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3952704665245293e-05,
-      "loss": 0.8152,
+      "loss": 0.9146,
       "step": 10711
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3951622695578324e-05,
-      "loss": 0.7482,
+      "loss": 0.8852,
       "step": 10712
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3950540671088198e-05,
-      "loss": 0.7233,
+      "loss": 0.804,
       "step": 10713
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3949458591789922e-05,
-      "loss": 0.7082,
+      "loss": 0.8975,
       "step": 10714
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3948376457698513e-05,
-      "loss": 0.7481,
+      "loss": 0.8285,
       "step": 10715
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3947294268828985e-05,
-      "loss": 0.7684,
+      "loss": 0.8625,
       "step": 10716
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3946212025196347e-05,
-      "loss": 0.7652,
+      "loss": 0.8721,
       "step": 10717
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3945129726815618e-05,
-      "loss": 0.7477,
+      "loss": 0.8761,
       "step": 10718
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.394404737370181e-05,
-      "loss": 0.6377,
+      "loss": 0.8228,
       "step": 10719
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.394296496586994e-05,
-      "loss": 0.6686,
+      "loss": 0.8163,
       "step": 10720
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3941882503335026e-05,
-      "loss": 0.9723,
+      "loss": 1.0334,
       "step": 10721
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3940799986112087e-05,
-      "loss": 0.8706,
+      "loss": 0.8893,
       "step": 10722
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3939717414216138e-05,
-      "loss": 0.8583,
+      "loss": 0.9533,
       "step": 10723
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.39386347876622e-05,
-      "loss": 0.7898,
+      "loss": 0.9822,
       "step": 10724
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3937552106465291e-05,
-      "loss": 0.6622,
+      "loss": 0.7114,
       "step": 10725
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3936469370640435e-05,
-      "loss": 0.6296,
+      "loss": 0.7542,
       "step": 10726
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.393538658020265e-05,
-      "loss": 0.8536,
+      "loss": 0.9454,
       "step": 10727
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.393430373516696e-05,
-      "loss": 0.5195,
+      "loss": 0.6847,
       "step": 10728
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.393322083554839e-05,
-      "loss": 0.7752,
+      "loss": 0.8846,
       "step": 10729
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3932137881361957e-05,
-      "loss": 0.7369,
+      "loss": 0.8651,
       "step": 10730
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3931054872622696e-05,
-      "loss": 0.7525,
+      "loss": 0.8454,
       "step": 10731
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3929971809345622e-05,
-      "loss": 0.7811,
+      "loss": 0.8913,
       "step": 10732
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3928888691545766e-05,
-      "loss": 0.7836,
+      "loss": 0.9136,
       "step": 10733
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3927805519238154e-05,
-      "loss": 0.7451,
+      "loss": 0.9363,
       "step": 10734
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.392672229243781e-05,
-      "loss": 0.6985,
+      "loss": 0.8754,
       "step": 10735
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3925639011159768e-05,
-      "loss": 0.7791,
+      "loss": 0.8916,
       "step": 10736
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3924555675419055e-05,
-      "loss": 0.7726,
+      "loss": 0.889,
       "step": 10737
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.39234722852307e-05,
-      "loss": 0.8111,
+      "loss": 0.926,
       "step": 10738
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3922388840609733e-05,
-      "loss": 0.7451,
+      "loss": 0.8764,
       "step": 10739
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3921305341571188e-05,
-      "loss": 0.5962,
+      "loss": 0.7786,
       "step": 10740
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3920221788130095e-05,
-      "loss": 0.7428,
+      "loss": 0.9537,
       "step": 10741
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3919138180301486e-05,
-      "loss": 0.7666,
+      "loss": 0.8611,
       "step": 10742
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3918054518100393e-05,
-      "loss": 0.727,
+      "loss": 0.8204,
       "step": 10743
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3916970801541854e-05,
-      "loss": 0.954,
+      "loss": 1.0309,
       "step": 10744
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3915887030640905e-05,
-      "loss": 0.925,
+      "loss": 0.9211,
       "step": 10745
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3914803205412575e-05,
-      "loss": 0.5982,
+      "loss": 0.7539,
       "step": 10746
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3913719325871907e-05,
-      "loss": 0.7957,
+      "loss": 0.9385,
       "step": 10747
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3912635392033938e-05,
-      "loss": 0.6977,
+      "loss": 0.8663,
       "step": 10748
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3911551403913702e-05,
-      "loss": 0.6939,
+      "loss": 0.8047,
       "step": 10749
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3910467361526238e-05,
-      "loss": 0.7241,
+      "loss": 0.8677,
       "step": 10750
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.390938326488659e-05,
-      "loss": 0.7732,
+      "loss": 0.9207,
       "step": 10751
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3908299114009797e-05,
-      "loss": 0.726,
+      "loss": 0.8802,
       "step": 10752
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3907214908910897e-05,
-      "loss": 0.7159,
+      "loss": 0.8802,
       "step": 10753
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3906130649604933e-05,
-      "loss": 0.7409,
+      "loss": 0.8917,
       "step": 10754
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3905046336106952e-05,
-      "loss": 0.7621,
+      "loss": 0.7665,
       "step": 10755
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3903961968431992e-05,
-      "loss": 0.7802,
+      "loss": 0.9475,
       "step": 10756
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3902877546595097e-05,
-      "loss": 0.6511,
+      "loss": 0.858,
       "step": 10757
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3901793070611316e-05,
-      "loss": 0.8127,
+      "loss": 0.853,
       "step": 10758
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.390070854049569e-05,
-      "loss": 0.8292,
+      "loss": 0.9021,
       "step": 10759
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3899623956263271e-05,
-      "loss": 0.7993,
+      "loss": 0.8883,
       "step": 10760
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3898539317929101e-05,
-      "loss": 0.687,
+      "loss": 0.8669,
       "step": 10761
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3897454625508224e-05,
-      "loss": 0.6356,
+      "loss": 0.8561,
       "step": 10762
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3896369879015698e-05,
-      "loss": 0.7835,
+      "loss": 0.8574,
       "step": 10763
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3895285078466569e-05,
-      "loss": 0.7985,
+      "loss": 0.8966,
       "step": 10764
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3894200223875885e-05,
-      "loss": 0.7264,
+      "loss": 0.9045,
       "step": 10765
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3893115315258699e-05,
-      "loss": 0.7621,
+      "loss": 0.8399,
       "step": 10766
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3892030352630059e-05,
-      "loss": 0.6081,
+      "loss": 0.7722,
       "step": 10767
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3890945336005022e-05,
-      "loss": 0.8566,
+      "loss": 0.934,
       "step": 10768
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.388986026539864e-05,
-      "loss": 0.7554,
+      "loss": 0.9119,
       "step": 10769
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3888775140825963e-05,
-      "loss": 0.6785,
+      "loss": 0.8735,
       "step": 10770
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3887689962302052e-05,
-      "loss": 0.7516,
+      "loss": 0.7764,
       "step": 10771
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3886604729841957e-05,
-      "loss": 0.821,
+      "loss": 0.8639,
       "step": 10772
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3885519443460733e-05,
-      "loss": 0.7589,
+      "loss": 0.9357,
       "step": 10773
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3884434103173443e-05,
-      "loss": 0.6772,
+      "loss": 0.8581,
       "step": 10774
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.388334870899514e-05,
-      "loss": 0.8439,
+      "loss": 0.8922,
       "step": 10775
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3882263260940883e-05,
-      "loss": 0.6025,
+      "loss": 0.8931,
       "step": 10776
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3881177759025732e-05,
-      "loss": 0.8019,
+      "loss": 0.9268,
       "step": 10777
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3880092203264744e-05,
-      "loss": 0.7812,
+      "loss": 0.8743,
       "step": 10778
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3879006593672986e-05,
-      "loss": 0.8341,
+      "loss": 0.9492,
       "step": 10779
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3877920930265511e-05,
-      "loss": 0.7284,
+      "loss": 0.8483,
       "step": 10780
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3876835213057389e-05,
-      "loss": 0.7971,
+      "loss": 0.8745,
       "step": 10781
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3875749442063677e-05,
-      "loss": 0.7487,
+      "loss": 0.8589,
       "step": 10782
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3874663617299441e-05,
-      "loss": 0.898,
+      "loss": 0.9623,
       "step": 10783
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3873577738779744e-05,
-      "loss": 0.7331,
+      "loss": 0.8575,
       "step": 10784
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3872491806519653e-05,
-      "loss": 0.7046,
+      "loss": 0.8421,
       "step": 10785
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3871405820534229e-05,
-      "loss": 0.8032,
+      "loss": 0.9484,
       "step": 10786
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3870319780838548e-05,
-      "loss": 0.6715,
+      "loss": 0.7654,
       "step": 10787
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3869233687447667e-05,
-      "loss": 0.7557,
+      "loss": 0.8612,
       "step": 10788
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.386814754037666e-05,
-      "loss": 0.7049,
+      "loss": 0.8435,
       "step": 10789
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3867061339640593e-05,
-      "loss": 0.7733,
+      "loss": 0.8329,
       "step": 10790
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3865975085254538e-05,
-      "loss": 0.7819,
+      "loss": 0.8879,
       "step": 10791
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.386488877723356e-05,
-      "loss": 0.8356,
+      "loss": 0.9463,
       "step": 10792
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3863802415592738e-05,
-      "loss": 0.7054,
+      "loss": 0.8609,
       "step": 10793
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3862716000347137e-05,
-      "loss": 0.7187,
+      "loss": 0.8814,
       "step": 10794
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3861629531511833e-05,
-      "loss": 0.744,
+      "loss": 0.9129,
       "step": 10795
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3860543009101899e-05,
-      "loss": 0.9808,
+      "loss": 0.946,
       "step": 10796
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3859456433132406e-05,
-      "loss": 0.7193,
+      "loss": 0.8972,
       "step": 10797
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3858369803618432e-05,
-      "loss": 0.8224,
+      "loss": 0.932,
       "step": 10798
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.385728312057505e-05,
-      "loss": 0.8595,
+      "loss": 0.9428,
       "step": 10799
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3856196384017342e-05,
-      "loss": 0.7494,
+      "loss": 0.9064,
       "step": 10800
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3855109593960379e-05,
-      "loss": 0.8758,
+      "loss": 0.8906,
       "step": 10801
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3854022750419236e-05,
-      "loss": 0.823,
+      "loss": 0.9546,
       "step": 10802
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3852935853408998e-05,
-      "loss": 0.8156,
+      "loss": 0.9202,
       "step": 10803
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3851848902944743e-05,
-      "loss": 0.6227,
+      "loss": 0.7699,
       "step": 10804
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3850761899041546e-05,
-      "loss": 0.817,
+      "loss": 0.9047,
       "step": 10805
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3849674841714493e-05,
-      "loss": 0.7708,
+      "loss": 0.8681,
       "step": 10806
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3848587730978662e-05,
-      "loss": 0.7174,
+      "loss": 0.8235,
       "step": 10807
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3847500566849142e-05,
-      "loss": 0.7514,
+      "loss": 0.9541,
       "step": 10808
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3846413349341006e-05,
-      "loss": 0.7178,
+      "loss": 0.7832,
       "step": 10809
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3845326078469341e-05,
-      "loss": 0.7337,
+      "loss": 0.8846,
       "step": 10810
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3844238754249235e-05,
-      "loss": 0.6315,
+      "loss": 0.8522,
       "step": 10811
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3843151376695771e-05,
-      "loss": 0.7734,
+      "loss": 0.8562,
       "step": 10812
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3842063945824031e-05,
-      "loss": 0.7999,
+      "loss": 0.9001,
       "step": 10813
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3840976461649109e-05,
-      "loss": 0.7804,
+      "loss": 0.8416,
       "step": 10814
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3839888924186087e-05,
-      "loss": 0.789,
+      "loss": 0.8123,
       "step": 10815
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3838801333450054e-05,
-      "loss": 0.7423,
+      "loss": 0.8826,
       "step": 10816
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3837713689456098e-05,
-      "loss": 0.9456,
+      "loss": 0.9828,
       "step": 10817
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.383662599221931e-05,
-      "loss": 0.7549,
+      "loss": 0.92,
       "step": 10818
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3835538241754783e-05,
-      "loss": 0.8152,
+      "loss": 0.9157,
       "step": 10819
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3834450438077603e-05,
-      "loss": 0.7389,
+      "loss": 0.8793,
       "step": 10820
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3833362581202861e-05,
-      "loss": 0.8136,
+      "loss": 0.91,
       "step": 10821
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3832274671145653e-05,
-      "loss": 0.7642,
+      "loss": 0.8944,
       "step": 10822
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3831186707921071e-05,
-      "loss": 0.9071,
+      "loss": 0.9352,
       "step": 10823
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3830098691544211e-05,
-      "loss": 0.8252,
+      "loss": 0.9208,
       "step": 10824
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3829010622030166e-05,
-      "loss": 0.7528,
+      "loss": 0.8642,
       "step": 10825
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.382792249939403e-05,
-      "loss": 0.7653,
+      "loss": 0.8757,
       "step": 10826
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3826834323650899e-05,
-      "loss": 0.8215,
+      "loss": 0.9545,
       "step": 10827
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3825746094815875e-05,
-      "loss": 0.8142,
+      "loss": 0.8766,
       "step": 10828
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3824657812904047e-05,
-      "loss": 0.7826,
+      "loss": 0.8933,
       "step": 10829
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3823569477930524e-05,
-      "loss": 0.8384,
+      "loss": 0.9407,
       "step": 10830
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3822481089910396e-05,
-      "loss": 0.8183,
+      "loss": 0.8981,
       "step": 10831
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3821392648858765e-05,
-      "loss": 0.807,
+      "loss": 0.895,
       "step": 10832
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3820304154790736e-05,
-      "loss": 0.8396,
+      "loss": 0.926,
       "step": 10833
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3819215607721404e-05,
-      "loss": 0.8529,
+      "loss": 0.9782,
       "step": 10834
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3818127007665876e-05,
-      "loss": 0.7311,
+      "loss": 0.8732,
       "step": 10835
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3817038354639253e-05,
-      "loss": 0.6726,
+      "loss": 0.8083,
       "step": 10836
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3815949648656642e-05,
-      "loss": 0.7101,
+      "loss": 0.9329,
       "step": 10837
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.381486088973314e-05,
-      "loss": 0.7241,
+      "loss": 0.8859,
       "step": 10838
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3813772077883854e-05,
-      "loss": 0.9321,
+      "loss": 0.919,
       "step": 10839
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3812683213123894e-05,
-      "loss": 0.6858,
+      "loss": 0.8884,
       "step": 10840
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3811594295468365e-05,
-      "loss": 0.6972,
+      "loss": 0.768,
       "step": 10841
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3810505324932369e-05,
-      "loss": 0.8679,
+      "loss": 0.928,
       "step": 10842
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3809416301531023e-05,
-      "loss": 0.827,
+      "loss": 0.9294,
       "step": 10843
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3808327225279427e-05,
-      "loss": 0.7468,
+      "loss": 0.8584,
       "step": 10844
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3807238096192695e-05,
-      "loss": 0.9142,
+      "loss": 0.9338,
       "step": 10845
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3806148914285938e-05,
-      "loss": 0.7599,
+      "loss": 0.8391,
       "step": 10846
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3805059679574261e-05,
-      "loss": 0.7354,
+      "loss": 0.8691,
       "step": 10847
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3803970392072783e-05,
-      "loss": 0.5892,
+      "loss": 0.7505,
       "step": 10848
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3802881051796613e-05,
-      "loss": 0.8793,
+      "loss": 1.0079,
       "step": 10849
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3801791658760861e-05,
-      "loss": 0.7666,
+      "loss": 0.8503,
       "step": 10850
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3800702212980646e-05,
-      "loss": 0.6583,
+      "loss": 0.8348,
       "step": 10851
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.379961271447108e-05,
-      "loss": 0.8364,
+      "loss": 0.9305,
       "step": 10852
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3798523163247279e-05,
-      "loss": 0.9262,
+      "loss": 1.0049,
       "step": 10853
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3797433559324357e-05,
-      "loss": 0.7439,
+      "loss": 0.8876,
       "step": 10854
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3796343902717435e-05,
-      "loss": 0.8743,
+      "loss": 0.9695,
       "step": 10855
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3795254193441628e-05,
-      "loss": 0.8468,
+      "loss": 0.9092,
       "step": 10856
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3794164431512052e-05,
-      "loss": 0.6176,
+      "loss": 0.7672,
       "step": 10857
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3793074616943824e-05,
-      "loss": 0.6356,
+      "loss": 0.8228,
       "step": 10858
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3791984749752073e-05,
-      "loss": 0.7509,
+      "loss": 0.8767,
       "step": 10859
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.379089482995191e-05,
-      "loss": 0.7786,
+      "loss": 0.8648,
       "step": 10860
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.3789804857558462e-05,
-      "loss": 0.6896,
+      "loss": 0.8364,
       "step": 10861
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3788714832586848e-05,
-      "loss": 0.7116,
+      "loss": 0.7973,
       "step": 10862
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3787624755052192e-05,
-      "loss": 0.7862,
+      "loss": 0.8628,
       "step": 10863
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3786534624969617e-05,
-      "loss": 0.7405,
+      "loss": 0.8164,
       "step": 10864
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3785444442354243e-05,
-      "loss": 0.6949,
+      "loss": 0.8201,
       "step": 10865
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.37843542072212e-05,
-      "loss": 0.844,
+      "loss": 0.8814,
       "step": 10866
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3783263919585611e-05,
-      "loss": 0.8709,
+      "loss": 0.8638,
       "step": 10867
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3782173579462606e-05,
-      "loss": 0.7507,
+      "loss": 0.8747,
       "step": 10868
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3781083186867306e-05,
-      "loss": 0.674,
+      "loss": 0.8343,
       "step": 10869
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3779992741814843e-05,
-      "loss": 0.7733,
+      "loss": 0.8659,
       "step": 10870
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3778902244320343e-05,
-      "loss": 0.7378,
+      "loss": 0.8728,
       "step": 10871
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3777811694398935e-05,
-      "loss": 0.6847,
+      "loss": 0.7896,
       "step": 10872
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.377672109206575e-05,
-      "loss": 0.8319,
+      "loss": 0.9069,
       "step": 10873
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.377563043733592e-05,
-      "loss": 0.7769,
+      "loss": 0.8259,
       "step": 10874
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3774539730224575e-05,
-      "loss": 0.7208,
+      "loss": 0.8104,
       "step": 10875
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3773448970746845e-05,
-      "loss": 0.6108,
+      "loss": 0.8083,
       "step": 10876
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3772358158917864e-05,
-      "loss": 0.7443,
+      "loss": 0.8803,
       "step": 10877
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3771267294752766e-05,
-      "loss": 0.7044,
+      "loss": 0.8896,
       "step": 10878
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3770176378266684e-05,
-      "loss": 0.9586,
+      "loss": 0.9624,
       "step": 10879
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3769085409474757e-05,
-      "loss": 0.7215,
+      "loss": 0.8606,
       "step": 10880
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3767994388392116e-05,
-      "loss": 0.7958,
+      "loss": 0.9074,
       "step": 10881
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3766903315033896e-05,
-      "loss": 0.7876,
+      "loss": 0.9664,
       "step": 10882
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3765812189415241e-05,
-      "loss": 0.8871,
+      "loss": 0.9809,
       "step": 10883
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3764721011551286e-05,
-      "loss": 0.7939,
+      "loss": 0.8796,
       "step": 10884
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3763629781457163e-05,
-      "loss": 0.841,
+      "loss": 0.9438,
       "step": 10885
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3762538499148021e-05,
-      "loss": 0.744,
+      "loss": 0.8391,
       "step": 10886
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3761447164638995e-05,
-      "loss": 0.7624,
+      "loss": 0.8201,
       "step": 10887
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3760355777945225e-05,
-      "loss": 0.7464,
+      "loss": 0.8509,
       "step": 10888
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3759264339081855e-05,
-      "loss": 0.6954,
+      "loss": 0.8206,
       "step": 10889
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3758172848064023e-05,
-      "loss": 0.8525,
+      "loss": 0.893,
       "step": 10890
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.375708130490688e-05,
-      "loss": 0.7314,
+      "loss": 0.9016,
       "step": 10891
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3755989709625563e-05,
-      "loss": 0.6725,
+      "loss": 0.8818,
       "step": 10892
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3754898062235216e-05,
-      "loss": 0.8287,
+      "loss": 0.9488,
       "step": 10893
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3753806362750987e-05,
-      "loss": 0.7125,
+      "loss": 0.8668,
       "step": 10894
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.375271461118802e-05,
-      "loss": 0.6878,
+      "loss": 0.7967,
       "step": 10895
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3751622807561465e-05,
-      "loss": 0.7727,
+      "loss": 0.8795,
       "step": 10896
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3750530951886463e-05,
-      "loss": 0.8444,
+      "loss": 0.8975,
       "step": 10897
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3749439044178166e-05,
-      "loss": 0.9519,
+      "loss": 0.9866,
       "step": 10898
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3748347084451723e-05,
-      "loss": 0.8455,
+      "loss": 0.926,
       "step": 10899
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3747255072722281e-05,
-      "loss": 0.6983,
+      "loss": 0.8349,
       "step": 10900
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3746163009004991e-05,
-      "loss": 0.762,
+      "loss": 0.8578,
       "step": 10901
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3745070893315007e-05,
-      "loss": 0.7325,
+      "loss": 0.889,
       "step": 10902
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3743978725667477e-05,
-      "loss": 0.6465,
+      "loss": 0.8442,
       "step": 10903
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3742886506077553e-05,
-      "loss": 0.8155,
+      "loss": 0.9116,
       "step": 10904
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.374179423456039e-05,
-      "loss": 0.7999,
+      "loss": 0.9582,
       "step": 10905
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.374070191113114e-05,
-      "loss": 0.7146,
+      "loss": 0.9274,
       "step": 10906
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.373960953580496e-05,
-      "loss": 0.7736,
+      "loss": 0.9321,
       "step": 10907
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3738517108597002e-05,
-      "loss": 0.7716,
+      "loss": 0.836,
       "step": 10908
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3737424629522424e-05,
-      "loss": 0.7734,
+      "loss": 0.8946,
       "step": 10909
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3736332098596383e-05,
-      "loss": 0.6954,
+      "loss": 0.9249,
       "step": 10910
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3735239515834034e-05,
-      "loss": 0.8617,
+      "loss": 0.9257,
       "step": 10911
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3734146881250536e-05,
-      "loss": 0.8139,
+      "loss": 0.8605,
       "step": 10912
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3733054194861049e-05,
-      "loss": 0.7751,
+      "loss": 0.8889,
       "step": 10913
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3731961456680734e-05,
-      "loss": 0.7364,
+      "loss": 0.8707,
       "step": 10914
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3730868666724746e-05,
-      "loss": 0.7704,
+      "loss": 0.9158,
       "step": 10915
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.372977582500825e-05,
-      "loss": 0.7212,
+      "loss": 0.8242,
       "step": 10916
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3728682931546402e-05,
-      "loss": 0.7268,
+      "loss": 0.8352,
       "step": 10917
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3727589986354372e-05,
-      "loss": 0.791,
+      "loss": 0.8238,
       "step": 10918
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.372649698944732e-05,
-      "loss": 0.7923,
+      "loss": 0.9059,
       "step": 10919
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.372540394084041e-05,
-      "loss": 0.6525,
+      "loss": 0.8685,
       "step": 10920
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3724310840548804e-05,
-      "loss": 0.7502,
+      "loss": 0.8581,
       "step": 10921
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.372321768858767e-05,
-      "loss": 0.7029,
+      "loss": 0.7534,
       "step": 10922
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3722124484972173e-05,
-      "loss": 0.8811,
+      "loss": 0.8995,
       "step": 10923
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3721031229717481e-05,
-      "loss": 0.7364,
+      "loss": 0.8845,
       "step": 10924
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3719937922838757e-05,
-      "loss": 0.5901,
+      "loss": 0.7701,
       "step": 10925
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3718844564351175e-05,
-      "loss": 0.6966,
+      "loss": 0.7904,
       "step": 10926
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3717751154269901e-05,
-      "loss": 0.7348,
+      "loss": 0.9071,
       "step": 10927
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3716657692610102e-05,
-      "loss": 0.7255,
+      "loss": 0.8188,
       "step": 10928
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3715564179386953e-05,
-      "loss": 0.7403,
+      "loss": 0.7622,
       "step": 10929
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3714470614615623e-05,
-      "loss": 0.9129,
+      "loss": 0.9768,
       "step": 10930
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3713376998311281e-05,
-      "loss": 0.7324,
+      "loss": 0.8347,
       "step": 10931
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3712283330489104e-05,
-      "loss": 0.8292,
+      "loss": 0.9235,
       "step": 10932
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3711189611164263e-05,
-      "loss": 0.6648,
+      "loss": 0.801,
       "step": 10933
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3710095840351932e-05,
-      "loss": 0.7819,
+      "loss": 0.8864,
       "step": 10934
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3709002018067283e-05,
-      "loss": 0.8198,
+      "loss": 0.8796,
       "step": 10935
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3707908144325495e-05,
-      "loss": 0.7406,
+      "loss": 0.8618,
       "step": 10936
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3706814219141742e-05,
-      "loss": 0.7123,
+      "loss": 0.9286,
       "step": 10937
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.37057202425312e-05,
-      "loss": 0.8218,
+      "loss": 0.9725,
       "step": 10938
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.370462621450905e-05,
-      "loss": 0.8389,
+      "loss": 0.9208,
       "step": 10939
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3703532135090465e-05,
-      "loss": 0.775,
+      "loss": 0.8381,
       "step": 10940
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3702438004290625e-05,
-      "loss": 0.7619,
+      "loss": 0.8503,
       "step": 10941
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3701343822124713e-05,
-      "loss": 0.6899,
+      "loss": 0.8562,
       "step": 10942
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3700249588607908e-05,
-      "loss": 0.7388,
+      "loss": 0.8647,
       "step": 10943
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3699155303755388e-05,
-      "loss": 0.6745,
+      "loss": 0.8058,
       "step": 10944
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3698060967582337e-05,
-      "loss": 1.0019,
+      "loss": 1.0153,
       "step": 10945
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3696966580103936e-05,
-      "loss": 0.6544,
+      "loss": 0.8674,
       "step": 10946
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.369587214133537e-05,
-      "loss": 0.7628,
+      "loss": 0.8204,
       "step": 10947
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3694777651291822e-05,
-      "loss": 0.7887,
+      "loss": 0.8247,
       "step": 10948
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3693683109988474e-05,
-      "loss": 0.6817,
+      "loss": 0.8347,
       "step": 10949
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3692588517440517e-05,
-      "loss": 0.7432,
+      "loss": 0.8527,
       "step": 10950
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3691493873663133e-05,
-      "loss": 0.6669,
+      "loss": 0.7136,
       "step": 10951
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3690399178671508e-05,
-      "loss": 0.8235,
+      "loss": 0.9258,
       "step": 10952
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3689304432480832e-05,
-      "loss": 0.7317,
+      "loss": 0.8092,
       "step": 10953
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3688209635106288e-05,
-      "loss": 0.7408,
+      "loss": 0.8832,
       "step": 10954
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3687114786563072e-05,
-      "loss": 0.7939,
+      "loss": 0.8513,
       "step": 10955
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3686019886866367e-05,
-      "loss": 0.7444,
+      "loss": 0.9202,
       "step": 10956
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3684924936031365e-05,
-      "loss": 0.5668,
+      "loss": 0.8715,
       "step": 10957
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.368382993407326e-05,
-      "loss": 0.7555,
+      "loss": 0.8159,
       "step": 10958
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3682734881007242e-05,
-      "loss": 0.8061,
+      "loss": 0.9559,
       "step": 10959
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3681639776848502e-05,
-      "loss": 0.831,
+      "loss": 0.9576,
       "step": 10960
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3680544621612231e-05,
-      "loss": 0.9128,
+      "loss": 0.9256,
       "step": 10961
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3679449415313626e-05,
-      "loss": 0.7866,
+      "loss": 0.8792,
       "step": 10962
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3678354157967884e-05,
-      "loss": 0.849,
+      "loss": 0.9318,
       "step": 10963
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3677258849590196e-05,
-      "loss": 0.8119,
+      "loss": 0.9283,
       "step": 10964
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3676163490195754e-05,
-      "loss": 0.8606,
+      "loss": 0.9335,
       "step": 10965
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3675068079799765e-05,
-      "loss": 0.8612,
+      "loss": 0.964,
       "step": 10966
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3673972618417421e-05,
-      "loss": 0.8468,
+      "loss": 0.9451,
       "step": 10967
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3672877106063915e-05,
-      "loss": 0.8296,
+      "loss": 0.903,
       "step": 10968
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3671781542754455e-05,
-      "loss": 0.6859,
+      "loss": 0.8433,
       "step": 10969
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3670685928504232e-05,
-      "loss": 0.9232,
+      "loss": 0.9749,
       "step": 10970
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3669590263328453e-05,
-      "loss": 0.7246,
+      "loss": 0.8375,
       "step": 10971
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3668494547242315e-05,
-      "loss": 0.8218,
+      "loss": 0.8917,
       "step": 10972
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3667398780261017e-05,
-      "loss": 0.731,
+      "loss": 0.8337,
       "step": 10973
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3666302962399767e-05,
-      "loss": 0.7726,
+      "loss": 0.9047,
       "step": 10974
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3665207093673765e-05,
-      "loss": 0.7517,
+      "loss": 0.8694,
       "step": 10975
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3664111174098214e-05,
-      "loss": 0.7914,
+      "loss": 0.9089,
       "step": 10976
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3663015203688319e-05,
-      "loss": 0.8317,
+      "loss": 0.9393,
       "step": 10977
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3661919182459285e-05,
-      "loss": 0.7723,
+      "loss": 0.8875,
       "step": 10978
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3660823110426318e-05,
-      "loss": 0.7475,
+      "loss": 0.8989,
       "step": 10979
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3659726987604626e-05,
-      "loss": 0.695,
+      "loss": 0.8199,
       "step": 10980
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3658630814009413e-05,
-      "loss": 0.7351,
+      "loss": 0.8743,
       "step": 10981
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3657534589655888e-05,
-      "loss": 0.6667,
+      "loss": 0.7792,
       "step": 10982
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3656438314559262e-05,
-      "loss": 0.7471,
+      "loss": 0.874,
       "step": 10983
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.365534198873474e-05,
-      "loss": 0.8337,
+      "loss": 0.9674,
       "step": 10984
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3654245612197534e-05,
-      "loss": 0.8984,
+      "loss": 0.9221,
       "step": 10985
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3653149184962857e-05,
-      "loss": 0.7479,
+      "loss": 0.8647,
       "step": 10986
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3652052707045916e-05,
-      "loss": 0.7803,
+      "loss": 0.8666,
       "step": 10987
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3650956178461927e-05,
-      "loss": 0.7109,
+      "loss": 0.8185,
       "step": 10988
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3649859599226099e-05,
-      "loss": 0.6365,
+      "loss": 0.7714,
       "step": 10989
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3648762969353651e-05,
-      "loss": 0.7602,
+      "loss": 0.9182,
       "step": 10990
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3647666288859794e-05,
-      "loss": 0.7764,
+      "loss": 0.8802,
       "step": 10991
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3646569557759736e-05,
-      "loss": 0.6906,
+      "loss": 0.7839,
       "step": 10992
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3645472776068707e-05,
-      "loss": 0.7927,
+      "loss": 0.8805,
       "step": 10993
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.364437594380191e-05,
-      "loss": 0.8009,
+      "loss": 0.9438,
       "step": 10994
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3643279060974571e-05,
-      "loss": 0.8713,
+      "loss": 0.9642,
       "step": 10995
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3642182127601902e-05,
-      "loss": 0.7224,
+      "loss": 0.8618,
       "step": 10996
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3641085143699122e-05,
-      "loss": 0.8059,
+      "loss": 0.9424,
       "step": 10997
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3639988109281455e-05,
-      "loss": 0.7162,
+      "loss": 0.8729,
       "step": 10998
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3638891024364118e-05,
-      "loss": 0.9105,
+      "loss": 0.9421,
       "step": 10999
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3637793888962328e-05,
-      "loss": 0.7851,
+      "loss": 0.9041,
       "step": 11000
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.363669670309131e-05,
-      "loss": 0.7666,
+      "loss": 0.8821,
       "step": 11001
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3635599466766285e-05,
-      "loss": 0.8304,
+      "loss": 0.8028,
       "step": 11002
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3634502180002475e-05,
-      "loss": 0.7979,
+      "loss": 0.8748,
       "step": 11003
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3633404842815106e-05,
-      "loss": 0.8349,
+      "loss": 0.8983,
       "step": 11004
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3632307455219398e-05,
-      "loss": 0.8234,
+      "loss": 0.8871,
       "step": 11005
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3631210017230578e-05,
-      "loss": 0.7184,
+      "loss": 0.8629,
       "step": 11006
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3630112528863872e-05,
-      "loss": 0.7021,
+      "loss": 0.7736,
       "step": 11007
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3629014990134504e-05,
-      "loss": 0.8512,
+      "loss": 0.8929,
       "step": 11008
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3627917401057702e-05,
-      "loss": 0.9222,
+      "loss": 0.9672,
       "step": 11009
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3626819761648696e-05,
-      "loss": 0.7904,
+      "loss": 0.8793,
       "step": 11010
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3625722071922712e-05,
-      "loss": 0.748,
+      "loss": 0.8974,
       "step": 11011
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3624624331894975e-05,
-      "loss": 0.8401,
+      "loss": 0.9226,
       "step": 11012
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3623526541580719e-05,
-      "loss": 0.7309,
+      "loss": 0.8433,
       "step": 11013
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3622428700995172e-05,
-      "loss": 0.7112,
+      "loss": 0.8867,
       "step": 11014
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3621330810153568e-05,
-      "loss": 0.6983,
+      "loss": 0.813,
       "step": 11015
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3620232869071137e-05,
-      "loss": 0.7567,
+      "loss": 0.888,
       "step": 11016
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3619134877763112e-05,
-      "loss": 0.7807,
+      "loss": 0.8632,
       "step": 11017
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3618036836244723e-05,
-      "loss": 0.6507,
+      "loss": 0.7361,
       "step": 11018
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3616938744531207e-05,
-      "loss": 0.8588,
+      "loss": 0.8855,
       "step": 11019
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.36158406026378e-05,
-      "loss": 0.7982,
+      "loss": 0.9178,
       "step": 11020
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3614742410579731e-05,
-      "loss": 0.712,
+      "loss": 0.8293,
       "step": 11021
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3613644168372244e-05,
-      "loss": 0.7718,
+      "loss": 0.8367,
       "step": 11022
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.361254587603057e-05,
-      "loss": 0.7187,
+      "loss": 0.914,
       "step": 11023
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3611447533569947e-05,
-      "loss": 0.6577,
+      "loss": 0.8201,
       "step": 11024
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3610349141005612e-05,
-      "loss": 0.751,
+      "loss": 0.9227,
       "step": 11025
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3609250698352807e-05,
-      "loss": 0.771,
+      "loss": 0.9098,
       "step": 11026
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3608152205626769e-05,
-      "loss": 0.7286,
+      "loss": 0.8655,
       "step": 11027
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3607053662842738e-05,
-      "loss": 0.7921,
+      "loss": 0.8613,
       "step": 11028
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3605955070015955e-05,
-      "loss": 0.8555,
+      "loss": 0.9189,
       "step": 11029
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3604856427161663e-05,
-      "loss": 0.8654,
+      "loss": 0.9527,
       "step": 11030
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3603757734295102e-05,
-      "loss": 0.8225,
+      "loss": 0.8846,
       "step": 11031
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3602658991431514e-05,
-      "loss": 0.75,
+      "loss": 0.9037,
       "step": 11032
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3601560198586145e-05,
-      "loss": 0.706,
+      "loss": 0.8537,
       "step": 11033
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3600461355774237e-05,
-      "loss": 0.7105,
+      "loss": 0.8567,
       "step": 11034
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3599362463011038e-05,
-      "loss": 0.738,
+      "loss": 0.8226,
       "step": 11035
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3598263520311794e-05,
-      "loss": 0.7245,
+      "loss": 0.8161,
       "step": 11036
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3597164527691744e-05,
-      "loss": 0.7401,
+      "loss": 0.9529,
       "step": 11037
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3596065485166141e-05,
-      "loss": 0.7926,
+      "loss": 0.7983,
       "step": 11038
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3594966392750233e-05,
-      "loss": 0.8481,
+      "loss": 0.8767,
       "step": 11039
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3593867250459267e-05,
-      "loss": 0.7613,
+      "loss": 0.8791,
       "step": 11040
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3592768058308492e-05,
-      "loss": 0.78,
+      "loss": 0.9266,
       "step": 11041
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3591668816313158e-05,
-      "loss": 0.6805,
+      "loss": 0.822,
       "step": 11042
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.359056952448851e-05,
-      "loss": 0.8327,
+      "loss": 0.8854,
       "step": 11043
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.358947018284981e-05,
-      "loss": 0.7695,
+      "loss": 0.8523,
       "step": 11044
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.35883707914123e-05,
-      "loss": 0.6815,
+      "loss": 0.8119,
       "step": 11045
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.358727135019124e-05,
-      "loss": 0.6948,
+      "loss": 0.8392,
       "step": 11046
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.358617185920188e-05,
-      "loss": 0.8343,
+      "loss": 0.944,
       "step": 11047
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3585072318459472e-05,
-      "loss": 0.8292,
+      "loss": 0.8895,
       "step": 11048
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3583972727979274e-05,
-      "loss": 0.7521,
+      "loss": 0.8204,
       "step": 11049
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3582873087776536e-05,
-      "loss": 0.6944,
+      "loss": 0.8089,
       "step": 11050
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3581773397866517e-05,
-      "loss": 0.6618,
+      "loss": 0.9139,
       "step": 11051
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3580673658264479e-05,
-      "loss": 0.6199,
+      "loss": 0.8035,
       "step": 11052
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3579573868985669e-05,
-      "loss": 0.8272,
+      "loss": 0.9703,
       "step": 11053
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3578474030045352e-05,
-      "loss": 0.7335,
+      "loss": 0.9126,
       "step": 11054
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3577374141458785e-05,
-      "loss": 0.7545,
+      "loss": 0.8476,
       "step": 11055
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3576274203241227e-05,
-      "loss": 0.7938,
+      "loss": 0.8772,
       "step": 11056
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.357517421540794e-05,
-      "loss": 0.6809,
+      "loss": 0.7967,
       "step": 11057
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3574074177974184e-05,
-      "loss": 0.6409,
+      "loss": 0.8196,
       "step": 11058
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3572974090955217e-05,
-      "loss": 0.7639,
+      "loss": 0.9311,
       "step": 11059
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3571873954366305e-05,
-      "loss": 0.6912,
+      "loss": 0.8646,
       "step": 11060
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.357077376822271e-05,
-      "loss": 0.7754,
+      "loss": 0.8469,
       "step": 11061
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3569673532539696e-05,
-      "loss": 0.9255,
+      "loss": 0.9304,
       "step": 11062
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3568573247332527e-05,
-      "loss": 0.7147,
+      "loss": 0.8529,
       "step": 11063
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3567472912616464e-05,
-      "loss": 0.7848,
+      "loss": 0.8877,
       "step": 11064
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3566372528406777e-05,
-      "loss": 0.8517,
+      "loss": 0.8722,
       "step": 11065
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3565272094718733e-05,
-      "loss": 0.9254,
+      "loss": 0.9503,
       "step": 11066
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3564171611567599e-05,
-      "loss": 0.7706,
+      "loss": 0.8754,
       "step": 11067
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3563071078968637e-05,
-      "loss": 0.7178,
+      "loss": 0.8595,
       "step": 11068
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3561970496937118e-05,
-      "loss": 0.8019,
+      "loss": 0.8963,
       "step": 11069
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3560869865488316e-05,
-      "loss": 0.7766,
+      "loss": 0.8771,
       "step": 11070
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3559769184637494e-05,
-      "loss": 0.7708,
+      "loss": 0.857,
       "step": 11071
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3558668454399925e-05,
-      "loss": 0.9118,
+      "loss": 1.0165,
       "step": 11072
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3557567674790883e-05,
-      "loss": 0.7279,
+      "loss": 0.8633,
       "step": 11073
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3556466845825637e-05,
-      "loss": 0.8861,
+      "loss": 0.9698,
       "step": 11074
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3555365967519455e-05,
-      "loss": 0.8247,
+      "loss": 0.9008,
       "step": 11075
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.355426503988762e-05,
-      "loss": 0.7874,
+      "loss": 0.8698,
       "step": 11076
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3553164062945396e-05,
-      "loss": 0.6524,
+      "loss": 0.7242,
       "step": 11077
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3552063036708066e-05,
-      "loss": 0.6983,
+      "loss": 0.8563,
       "step": 11078
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3550961961190897e-05,
-      "loss": 0.8321,
+      "loss": 0.8876,
       "step": 11079
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3549860836409171e-05,
-      "loss": 0.8766,
+      "loss": 0.9913,
       "step": 11080
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3548759662378163e-05,
-      "loss": 0.6886,
+      "loss": 0.8918,
       "step": 11081
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3547658439113151e-05,
-      "loss": 0.7586,
+      "loss": 0.8287,
       "step": 11082
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3546557166629408e-05,
-      "loss": 0.7218,
+      "loss": 0.8372,
       "step": 11083
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3545455844942219e-05,
-      "loss": 0.7668,
+      "loss": 0.8362,
       "step": 11084
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.354435447406686e-05,
-      "loss": 0.82,
+      "loss": 0.9146,
       "step": 11085
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3543253054018612e-05,
-      "loss": 0.8591,
+      "loss": 0.9374,
       "step": 11086
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3542151584812755e-05,
-      "loss": 0.8901,
+      "loss": 0.9248,
       "step": 11087
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3541050066464568e-05,
-      "loss": 0.7455,
+      "loss": 0.9015,
       "step": 11088
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3539948498989337e-05,
-      "loss": 0.7732,
+      "loss": 0.7932,
       "step": 11089
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3538846882402342e-05,
-      "loss": 0.9702,
+      "loss": 0.9568,
       "step": 11090
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3537745216718869e-05,
-      "loss": 0.7626,
+      "loss": 0.8332,
       "step": 11091
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.35366435019542e-05,
-      "loss": 0.9521,
+      "loss": 0.9683,
       "step": 11092
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3535541738123621e-05,
-      "loss": 0.9241,
+      "loss": 1.0245,
       "step": 11093
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3534439925242415e-05,
-      "loss": 0.7883,
+      "loss": 0.86,
       "step": 11094
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.353333806332587e-05,
-      "loss": 0.7183,
+      "loss": 0.7967,
       "step": 11095
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3532236152389273e-05,
-      "loss": 0.8168,
+      "loss": 0.9083,
       "step": 11096
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3531134192447913e-05,
-      "loss": 0.7898,
+      "loss": 0.928,
       "step": 11097
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3530032183517074e-05,
-      "loss": 1.0153,
+      "loss": 1.0234,
       "step": 11098
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3528930125612047e-05,
-      "loss": 0.7874,
+      "loss": 0.8829,
       "step": 11099
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3527828018748122e-05,
-      "loss": 0.6604,
+      "loss": 0.7549,
       "step": 11100
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3526725862940586e-05,
-      "loss": 0.7161,
+      "loss": 0.8236,
       "step": 11101
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3525623658204737e-05,
-      "loss": 0.8165,
+      "loss": 0.8692,
       "step": 11102
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.352452140455586e-05,
-      "loss": 0.6654,
+      "loss": 0.854,
       "step": 11103
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3523419102009247e-05,
-      "loss": 0.689,
+      "loss": 0.8342,
       "step": 11104
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3522316750580197e-05,
-      "loss": 0.7999,
+      "loss": 0.9211,
       "step": 11105
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3521214350283996e-05,
-      "loss": 0.7428,
+      "loss": 0.8073,
       "step": 11106
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3520111901135943e-05,
-      "loss": 0.8076,
+      "loss": 0.9212,
       "step": 11107
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3519009403151335e-05,
-      "loss": 0.761,
+      "loss": 0.8544,
       "step": 11108
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3517906856345459e-05,
-      "loss": 0.8377,
+      "loss": 0.9173,
       "step": 11109
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3516804260733621e-05,
-      "loss": 0.6738,
+      "loss": 0.7841,
       "step": 11110
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3515701616331111e-05,
-      "loss": 0.871,
+      "loss": 1.006,
       "step": 11111
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3514598923153229e-05,
-      "loss": 0.7578,
+      "loss": 0.7776,
       "step": 11112
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3513496181215273e-05,
-      "loss": 0.8466,
+      "loss": 0.9105,
       "step": 11113
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3512393390532545e-05,
-      "loss": 0.8413,
+      "loss": 0.9498,
       "step": 11114
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.351129055112034e-05,
-      "loss": 0.8675,
+      "loss": 0.9127,
       "step": 11115
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3510187662993963e-05,
-      "loss": 0.7308,
+      "loss": 0.8792,
       "step": 11116
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3509084726168708e-05,
-      "loss": 0.6884,
+      "loss": 0.8114,
       "step": 11117
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3507981740659886e-05,
-      "loss": 0.7925,
+      "loss": 0.9377,
       "step": 11118
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.350687870648279e-05,
-      "loss": 0.8532,
+      "loss": 0.8732,
       "step": 11119
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3505775623652731e-05,
-      "loss": 0.6822,
+      "loss": 0.8194,
       "step": 11120
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.350467249218501e-05,
-      "loss": 0.6751,
+      "loss": 0.789,
       "step": 11121
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3503569312094928e-05,
-      "loss": 0.7327,
+      "loss": 0.8477,
       "step": 11122
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3502466083397794e-05,
-      "loss": 0.6836,
+      "loss": 0.8399,
       "step": 11123
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3501362806108913e-05,
-      "loss": 0.6955,
+      "loss": 0.8363,
       "step": 11124
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.350025948024359e-05,
-      "loss": 0.7372,
+      "loss": 0.829,
       "step": 11125
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3499156105817133e-05,
-      "loss": 0.732,
+      "loss": 0.867,
       "step": 11126
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3498052682844849e-05,
-      "loss": 0.6641,
+      "loss": 0.8266,
       "step": 11127
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3496949211342047e-05,
-      "loss": 0.7875,
+      "loss": 0.8644,
       "step": 11128
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.349584569132404e-05,
-      "loss": 0.7442,
+      "loss": 0.8877,
       "step": 11129
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.349474212280613e-05,
-      "loss": 0.725,
+      "loss": 0.8381,
       "step": 11130
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.349363850580363e-05,
-      "loss": 0.7181,
+      "loss": 0.8559,
       "step": 11131
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3492534840331858e-05,
-      "loss": 0.8085,
+      "loss": 0.9135,
       "step": 11132
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3491431126406114e-05,
-      "loss": 0.8969,
+      "loss": 1.0158,
       "step": 11133
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3490327364041722e-05,
-      "loss": 0.7023,
+      "loss": 0.8069,
       "step": 11134
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3489223553253987e-05,
-      "loss": 0.9421,
+      "loss": 1.0027,
       "step": 11135
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.3488119694058225e-05,
-      "loss": 1.0086,
+      "loss": 1.008,
       "step": 11136
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3487015786469755e-05,
-      "loss": 0.6806,
+      "loss": 0.9033,
       "step": 11137
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3485911830503886e-05,
-      "loss": 0.9131,
+      "loss": 0.9448,
       "step": 11138
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3484807826175936e-05,
-      "loss": 0.8547,
+      "loss": 0.9673,
       "step": 11139
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3483703773501225e-05,
-      "loss": 0.7965,
+      "loss": 0.8719,
       "step": 11140
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3482599672495062e-05,
-      "loss": 0.678,
+      "loss": 0.8116,
       "step": 11141
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3481495523172772e-05,
-      "loss": 0.7922,
+      "loss": 0.8896,
       "step": 11142
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.348039132554967e-05,
-      "loss": 0.8254,
+      "loss": 0.9332,
       "step": 11143
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.347928707964108e-05,
-      "loss": 0.7661,
+      "loss": 0.9723,
       "step": 11144
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3478182785462317e-05,
-      "loss": 0.8117,
+      "loss": 0.9396,
       "step": 11145
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.34770784430287e-05,
-      "loss": 0.7266,
+      "loss": 0.8237,
       "step": 11146
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3475974052355554e-05,
-      "loss": 0.7636,
+      "loss": 0.844,
       "step": 11147
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3474869613458202e-05,
-      "loss": 0.7703,
+      "loss": 0.8516,
       "step": 11148
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3473765126351958e-05,
-      "loss": 0.6887,
+      "loss": 0.8059,
       "step": 11149
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3472660591052158e-05,
-      "loss": 0.7065,
+      "loss": 0.8547,
       "step": 11150
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3471556007574117e-05,
-      "loss": 0.8477,
+      "loss": 0.8669,
       "step": 11151
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3470451375933163e-05,
-      "loss": 0.8202,
+      "loss": 0.8476,
       "step": 11152
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3469346696144618e-05,
-      "loss": 0.9011,
+      "loss": 0.8911,
       "step": 11153
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3468241968223814e-05,
-      "loss": 0.8095,
+      "loss": 0.8945,
       "step": 11154
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3467137192186068e-05,
-      "loss": 0.771,
+      "loss": 0.8428,
       "step": 11155
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3466032368046717e-05,
-      "loss": 0.7176,
+      "loss": 0.8876,
       "step": 11156
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3464927495821086e-05,
-      "loss": 0.7928,
+      "loss": 0.9528,
       "step": 11157
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3463822575524495e-05,
-      "loss": 0.8647,
+      "loss": 0.8933,
       "step": 11158
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3462717607172287e-05,
-      "loss": 0.6931,
+      "loss": 0.9385,
       "step": 11159
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.346161259077978e-05,
-      "loss": 0.824,
+      "loss": 0.8732,
       "step": 11160
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3460507526362311e-05,
-      "loss": 0.8456,
+      "loss": 0.9257,
       "step": 11161
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3459402413935214e-05,
-      "loss": 0.7652,
+      "loss": 0.9632,
       "step": 11162
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3458297253513815e-05,
-      "loss": 0.7525,
+      "loss": 0.8483,
       "step": 11163
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3457192045113448e-05,
-      "loss": 0.5802,
+      "loss": 0.8067,
       "step": 11164
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3456086788749443e-05,
-      "loss": 0.7945,
+      "loss": 0.8706,
       "step": 11165
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.345498148443714e-05,
-      "loss": 0.7775,
+      "loss": 0.8631,
       "step": 11166
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3453876132191873e-05,
-      "loss": 0.7067,
+      "loss": 0.8542,
       "step": 11167
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.345277073202897e-05,
-      "loss": 0.6403,
+      "loss": 0.76,
       "step": 11168
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3451665283963777e-05,
-      "loss": 0.6572,
+      "loss": 0.7492,
       "step": 11169
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3450559788011623e-05,
-      "loss": 0.7013,
+      "loss": 0.8896,
       "step": 11170
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3449454244187847e-05,
-      "loss": 0.8308,
+      "loss": 0.9146,
       "step": 11171
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.344834865250779e-05,
-      "loss": 0.7908,
+      "loss": 0.8336,
       "step": 11172
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3447243012986784e-05,
-      "loss": 0.7538,
+      "loss": 0.7795,
       "step": 11173
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3446137325640176e-05,
-      "loss": 0.7979,
+      "loss": 0.9495,
       "step": 11174
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3445031590483302e-05,
-      "loss": 0.8788,
+      "loss": 0.9218,
       "step": 11175
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.34439258075315e-05,
-      "loss": 0.7166,
+      "loss": 0.7575,
       "step": 11176
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3442819976800114e-05,
-      "loss": 0.8177,
+      "loss": 0.9029,
       "step": 11177
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3441714098304488e-05,
-      "loss": 0.8131,
+      "loss": 0.9079,
       "step": 11178
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3440608172059959e-05,
-      "loss": 0.8264,
+      "loss": 0.9241,
       "step": 11179
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3439502198081875e-05,
-      "loss": 0.8423,
+      "loss": 0.9209,
       "step": 11180
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.343839617638558e-05,
-      "loss": 0.7421,
+      "loss": 0.9055,
       "step": 11181
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3437290106986414e-05,
-      "loss": 0.8346,
+      "loss": 0.8809,
       "step": 11182
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3436183989899725e-05,
-      "loss": 0.8246,
+      "loss": 0.9351,
       "step": 11183
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3435077825140859e-05,
-      "loss": 0.7438,
+      "loss": 0.8492,
       "step": 11184
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3433971612725161e-05,
-      "loss": 0.724,
+      "loss": 0.8717,
       "step": 11185
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.343286535266798e-05,
-      "loss": 0.7856,
+      "loss": 0.8612,
       "step": 11186
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.343175904498466e-05,
-      "loss": 0.8049,
+      "loss": 0.9436,
       "step": 11187
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3430652689690555e-05,
-      "loss": 0.8097,
+      "loss": 0.8438,
       "step": 11188
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3429546286801009e-05,
-      "loss": 0.9189,
+      "loss": 1.0028,
       "step": 11189
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3428439836331377e-05,
-      "loss": 0.8225,
+      "loss": 0.8999,
       "step": 11190
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3427333338297008e-05,
-      "loss": 0.6991,
+      "loss": 0.8132,
       "step": 11191
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3426226792713246e-05,
-      "loss": 0.7272,
+      "loss": 0.8991,
       "step": 11192
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3425120199595455e-05,
-      "loss": 0.6236,
+      "loss": 0.8488,
       "step": 11193
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3424013558958978e-05,
-      "loss": 0.7761,
+      "loss": 0.8955,
       "step": 11194
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3422906870819169e-05,
-      "loss": 0.7842,
+      "loss": 0.9308,
       "step": 11195
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3421800135191386e-05,
-      "loss": 0.6521,
+      "loss": 0.7871,
       "step": 11196
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3420693352090979e-05,
-      "loss": 0.7254,
+      "loss": 0.7794,
       "step": 11197
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3419586521533306e-05,
-      "loss": 0.9069,
+      "loss": 0.8869,
       "step": 11198
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3418479643533722e-05,
-      "loss": 0.7527,
+      "loss": 0.8721,
       "step": 11199
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3417372718107582e-05,
-      "loss": 0.6564,
+      "loss": 0.8398,
       "step": 11200
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3416265745270248e-05,
-      "loss": 0.754,
+      "loss": 0.886,
       "step": 11201
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3415158725037072e-05,
-      "loss": 0.7063,
+      "loss": 0.8537,
       "step": 11202
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3414051657423411e-05,
-      "loss": 0.6541,
+      "loss": 0.8199,
       "step": 11203
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.341294454244463e-05,
-      "loss": 0.8649,
+      "loss": 0.9292,
       "step": 11204
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3411837380116085e-05,
-      "loss": 0.871,
+      "loss": 0.9913,
       "step": 11205
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3410730170453137e-05,
-      "loss": 0.6729,
+      "loss": 0.7771,
       "step": 11206
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3409622913471147e-05,
-      "loss": 0.739,
+      "loss": 0.8171,
       "step": 11207
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3408515609185477e-05,
-      "loss": 0.6602,
+      "loss": 0.8284,
       "step": 11208
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.340740825761149e-05,
-      "loss": 0.7418,
+      "loss": 0.8335,
       "step": 11209
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3406300858764547e-05,
-      "loss": 0.7238,
+      "loss": 0.8199,
       "step": 11210
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3405193412660011e-05,
-      "loss": 0.9609,
+      "loss": 0.964,
       "step": 11211
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.340408591931325e-05,
-      "loss": 0.695,
+      "loss": 0.8053,
       "step": 11212
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3402978378739625e-05,
-      "loss": 0.7978,
+      "loss": 0.8486,
       "step": 11213
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3401870790954505e-05,
-      "loss": 0.8401,
+      "loss": 0.8782,
       "step": 11214
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3400763155973253e-05,
-      "loss": 0.8934,
+      "loss": 0.9054,
       "step": 11215
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3399655473811237e-05,
-      "loss": 0.771,
+      "loss": 0.8923,
       "step": 11216
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3398547744483824e-05,
-      "loss": 0.7997,
+      "loss": 0.9359,
       "step": 11217
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3397439968006387e-05,
-      "loss": 0.7621,
+      "loss": 0.9088,
       "step": 11218
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3396332144394286e-05,
-      "loss": 0.7161,
+      "loss": 0.8446,
       "step": 11219
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3395224273662902e-05,
-      "loss": 0.7565,
+      "loss": 0.9272,
       "step": 11220
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.339411635582759e-05,
-      "loss": 0.8441,
+      "loss": 0.896,
       "step": 11221
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3393008390903734e-05,
-      "loss": 0.7947,
+      "loss": 0.9611,
       "step": 11222
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3391900378906702e-05,
-      "loss": 0.5402,
+      "loss": 0.7451,
       "step": 11223
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3390792319851859e-05,
-      "loss": 0.753,
+      "loss": 0.8765,
       "step": 11224
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.338968421375459e-05,
-      "loss": 0.7433,
+      "loss": 0.855,
       "step": 11225
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3388576060630258e-05,
-      "loss": 0.7043,
+      "loss": 0.843,
       "step": 11226
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.338746786049424e-05,
-      "loss": 0.8403,
+      "loss": 0.9144,
       "step": 11227
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3386359613361915e-05,
-      "loss": 0.7797,
+      "loss": 0.8669,
       "step": 11228
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3385251319248654e-05,
-      "loss": 0.7343,
+      "loss": 0.8976,
       "step": 11229
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3384142978169832e-05,
-      "loss": 0.7304,
+      "loss": 0.9593,
       "step": 11230
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.338303459014083e-05,
-      "loss": 0.7127,
+      "loss": 0.8691,
       "step": 11231
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3381926155177021e-05,
-      "loss": 0.6536,
+      "loss": 0.8922,
       "step": 11232
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3380817673293785e-05,
-      "loss": 0.7757,
+      "loss": 0.8295,
       "step": 11233
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3379709144506504e-05,
-      "loss": 0.7003,
+      "loss": 0.8478,
       "step": 11234
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3378600568830548e-05,
-      "loss": 0.6673,
+      "loss": 0.8745,
       "step": 11235
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3377491946281306e-05,
-      "loss": 0.8009,
+      "loss": 0.8824,
       "step": 11236
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3376383276874157e-05,
-      "loss": 0.6578,
+      "loss": 0.83,
       "step": 11237
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3375274560624476e-05,
-      "loss": 0.6898,
+      "loss": 0.9146,
       "step": 11238
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3374165797547654e-05,
-      "loss": 0.6922,
+      "loss": 0.8947,
       "step": 11239
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3373056987659065e-05,
-      "loss": 0.6863,
+      "loss": 0.8635,
       "step": 11240
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.33719481309741e-05,
-      "loss": 0.7608,
+      "loss": 0.903,
       "step": 11241
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3370839227508136e-05,
-      "loss": 0.7565,
+      "loss": 0.8569,
       "step": 11242
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3369730277276561e-05,
-      "loss": 0.8039,
+      "loss": 0.9316,
       "step": 11243
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.336862128029476e-05,
-      "loss": 0.8005,
+      "loss": 0.8751,
       "step": 11244
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3367512236578115e-05,
-      "loss": 0.8795,
+      "loss": 0.922,
       "step": 11245
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.336640314614202e-05,
-      "loss": 0.7465,
+      "loss": 0.8605,
       "step": 11246
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3365294009001857e-05,
-      "loss": 0.809,
+      "loss": 0.9474,
       "step": 11247
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3364184825173012e-05,
-      "loss": 0.6323,
+      "loss": 0.7852,
       "step": 11248
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3363075594670877e-05,
-      "loss": 0.7247,
+      "loss": 0.8164,
       "step": 11249
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3361966317510841e-05,
-      "loss": 0.8102,
+      "loss": 0.8622,
       "step": 11250
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3360856993708289e-05,
-      "loss": 0.8465,
+      "loss": 0.8672,
       "step": 11251
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3359747623278619e-05,
-      "loss": 0.805,
+      "loss": 0.9122,
       "step": 11252
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3358638206237213e-05,
-      "loss": 0.7142,
+      "loss": 0.8432,
       "step": 11253
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.335752874259947e-05,
-      "loss": 0.8397,
+      "loss": 0.9691,
       "step": 11254
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.335641923238078e-05,
-      "loss": 0.7244,
+      "loss": 0.908,
       "step": 11255
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3355309675596533e-05,
-      "loss": 0.7552,
+      "loss": 0.8117,
       "step": 11256
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3354200072262124e-05,
-      "loss": 0.7669,
+      "loss": 0.8832,
       "step": 11257
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.335309042239295e-05,
-      "loss": 0.6823,
+      "loss": 0.8379,
       "step": 11258
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3351980726004406e-05,
-      "loss": 0.7392,
+      "loss": 0.823,
       "step": 11259
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3350870983111883e-05,
-      "loss": 0.7862,
+      "loss": 0.8666,
       "step": 11260
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.334976119373078e-05,
-      "loss": 0.7286,
+      "loss": 0.8482,
       "step": 11261
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3348651357876491e-05,
-      "loss": 0.7639,
+      "loss": 0.8747,
       "step": 11262
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3347541475564418e-05,
-      "loss": 0.8055,
+      "loss": 0.9211,
       "step": 11263
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3346431546809955e-05,
-      "loss": 0.9296,
+      "loss": 0.9955,
       "step": 11264
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3345321571628502e-05,
-      "loss": 0.8411,
+      "loss": 0.9134,
       "step": 11265
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3344211550035463e-05,
-      "loss": 0.6747,
+      "loss": 0.7888,
       "step": 11266
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3343101482046227e-05,
-      "loss": 0.7431,
+      "loss": 0.8322,
       "step": 11267
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3341991367676203e-05,
-      "loss": 0.7649,
+      "loss": 0.8225,
       "step": 11268
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3340881206940794e-05,
-      "loss": 0.6644,
+      "loss": 0.812,
       "step": 11269
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3339770999855394e-05,
-      "loss": 0.8202,
+      "loss": 0.979,
       "step": 11270
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3338660746435412e-05,
-      "loss": 0.7436,
+      "loss": 0.9086,
       "step": 11271
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3337550446696248e-05,
-      "loss": 0.7941,
+      "loss": 0.9449,
       "step": 11272
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3336440100653309e-05,
-      "loss": 0.8084,
+      "loss": 0.9168,
       "step": 11273
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3335329708321995e-05,
-      "loss": 0.7763,
+      "loss": 0.7866,
       "step": 11274
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3334219269717714e-05,
-      "loss": 0.866,
+      "loss": 0.9014,
       "step": 11275
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3333108784855875e-05,
-      "loss": 0.71,
+      "loss": 0.8952,
       "step": 11276
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3331998253751877e-05,
-      "loss": 0.8551,
+      "loss": 0.9776,
       "step": 11277
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.333088767642113e-05,
-      "loss": 0.7171,
+      "loss": 0.8558,
       "step": 11278
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3329777052879047e-05,
-      "loss": 0.7846,
+      "loss": 0.8169,
       "step": 11279
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3328666383141027e-05,
-      "loss": 0.7008,
+      "loss": 0.8498,
       "step": 11280
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3327555667222485e-05,
-      "loss": 0.7941,
+      "loss": 0.9354,
       "step": 11281
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3326444905138832e-05,
-      "loss": 0.802,
+      "loss": 0.9025,
       "step": 11282
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.332533409690547e-05,
-      "loss": 0.7208,
+      "loss": 0.8633,
       "step": 11283
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.332422324253782e-05,
-      "loss": 0.761,
+      "loss": 0.9066,
       "step": 11284
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3323112342051288e-05,
-      "loss": 0.739,
+      "loss": 0.8352,
       "step": 11285
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3322001395461284e-05,
-      "loss": 0.7524,
+      "loss": 0.9277,
       "step": 11286
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3320890402783229e-05,
-      "loss": 0.7787,
+      "loss": 0.8837,
       "step": 11287
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3319779364032526e-05,
-      "loss": 0.7192,
+      "loss": 0.8843,
       "step": 11288
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3318668279224598e-05,
-      "loss": 0.7676,
+      "loss": 0.9044,
       "step": 11289
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3317557148374856e-05,
-      "loss": 0.758,
+      "loss": 0.8809,
       "step": 11290
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3316445971498714e-05,
-      "loss": 0.6705,
+      "loss": 0.8134,
       "step": 11291
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3315334748611592e-05,
-      "loss": 0.7743,
+      "loss": 0.8197,
       "step": 11292
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3314223479728905e-05,
-      "loss": 0.6737,
+      "loss": 0.8167,
       "step": 11293
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3313112164866065e-05,
-      "loss": 0.8199,
+      "loss": 0.8815,
       "step": 11294
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3312000804038496e-05,
-      "loss": 0.6345,
+      "loss": 0.7325,
       "step": 11295
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3310889397261617e-05,
-      "loss": 0.9592,
+      "loss": 0.9013,
       "step": 11296
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3309777944550845e-05,
-      "loss": 0.8774,
+      "loss": 0.8723,
       "step": 11297
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.33086664459216e-05,
-      "loss": 0.7078,
+      "loss": 0.8704,
       "step": 11298
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.33075549013893e-05,
-      "loss": 0.8264,
+      "loss": 0.9344,
       "step": 11299
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.330644331096937e-05,
-      "loss": 0.7282,
+      "loss": 0.9381,
       "step": 11300
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.330533167467723e-05,
-      "loss": 0.7749,
+      "loss": 0.88,
       "step": 11301
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3304219992528302e-05,
-      "loss": 0.7653,
+      "loss": 0.8837,
       "step": 11302
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3303108264538012e-05,
-      "loss": 0.6657,
+      "loss": 0.8484,
       "step": 11303
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3301996490721778e-05,
-      "loss": 0.7494,
+      "loss": 0.8924,
       "step": 11304
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3300884671095032e-05,
-      "loss": 0.8553,
+      "loss": 0.945,
       "step": 11305
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3299772805673192e-05,
-      "loss": 0.874,
+      "loss": 0.9902,
       "step": 11306
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3298660894471684e-05,
-      "loss": 0.712,
+      "loss": 0.8368,
       "step": 11307
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.329754893750594e-05,
-      "loss": 0.7785,
+      "loss": 0.8555,
       "step": 11308
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3296436934791382e-05,
-      "loss": 0.7195,
+      "loss": 0.8168,
       "step": 11309
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3295324886343437e-05,
-      "loss": 0.7296,
+      "loss": 0.8245,
       "step": 11310
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3294212792177535e-05,
-      "loss": 0.6873,
+      "loss": 0.8559,
       "step": 11311
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3293100652309105e-05,
-      "loss": 0.69,
+      "loss": 0.8932,
       "step": 11312
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3291988466753577e-05,
-      "loss": 0.8202,
+      "loss": 0.9059,
       "step": 11313
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3290876235526377e-05,
-      "loss": 0.9603,
+      "loss": 1.0078,
       "step": 11314
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3289763958642938e-05,
-      "loss": 0.7545,
+      "loss": 0.818,
       "step": 11315
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3288651636118697e-05,
-      "loss": 0.6592,
+      "loss": 0.787,
       "step": 11316
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3287539267969076e-05,
-      "loss": 0.7925,
+      "loss": 0.8757,
       "step": 11317
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3286426854209513e-05,
-      "loss": 0.904,
+      "loss": 0.875,
       "step": 11318
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3285314394855441e-05,
-      "loss": 0.6945,
+      "loss": 0.878,
       "step": 11319
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3284201889922293e-05,
-      "loss": 0.8313,
+      "loss": 0.9046,
       "step": 11320
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3283089339425504e-05,
-      "loss": 0.8528,
+      "loss": 0.9116,
       "step": 11321
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3281976743380508e-05,
-      "loss": 0.757,
+      "loss": 0.8168,
       "step": 11322
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3280864101802737e-05,
-      "loss": 0.8632,
+      "loss": 0.9619,
       "step": 11323
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3279751414707636e-05,
-      "loss": 0.8659,
+      "loss": 0.9,
       "step": 11324
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.327863868211064e-05,
-      "loss": 0.6599,
+      "loss": 0.808,
       "step": 11325
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3277525904027178e-05,
-      "loss": 0.8494,
+      "loss": 0.8726,
       "step": 11326
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3276413080472697e-05,
-      "loss": 0.8134,
+      "loss": 0.8767,
       "step": 11327
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3275300211462632e-05,
-      "loss": 0.7989,
+      "loss": 0.9297,
       "step": 11328
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3274187297012423e-05,
-      "loss": 0.7596,
+      "loss": 0.9063,
       "step": 11329
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3273074337137515e-05,
-      "loss": 0.6994,
+      "loss": 0.8924,
       "step": 11330
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.327196133185334e-05,
-      "loss": 0.7238,
+      "loss": 0.8955,
       "step": 11331
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3270848281175348e-05,
-      "loss": 0.8676,
+      "loss": 0.9783,
       "step": 11332
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3269735185118973e-05,
-      "loss": 0.7907,
+      "loss": 0.9636,
       "step": 11333
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3268622043699662e-05,
-      "loss": 0.8258,
+      "loss": 0.8451,
       "step": 11334
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3267508856932863e-05,
-      "loss": 0.6421,
+      "loss": 0.7905,
       "step": 11335
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3266395624834008e-05,
-      "loss": 0.8498,
+      "loss": 0.885,
       "step": 11336
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3265282347418551e-05,
-      "loss": 0.7729,
+      "loss": 0.8751,
       "step": 11337
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3264169024701933e-05,
-      "loss": 0.6087,
+      "loss": 0.839,
       "step": 11338
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3263055656699601e-05,
-      "loss": 0.7535,
+      "loss": 0.81,
       "step": 11339
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3261942243427004e-05,
-      "loss": 0.8951,
+      "loss": 1.0111,
       "step": 11340
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3260828784899584e-05,
-      "loss": 0.7531,
+      "loss": 0.7846,
       "step": 11341
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.325971528113279e-05,
-      "loss": 0.8797,
+      "loss": 0.9954,
       "step": 11342
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3258601732142076e-05,
-      "loss": 0.9323,
+      "loss": 0.9856,
       "step": 11343
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.325748813794288e-05,
-      "loss": 0.7201,
+      "loss": 0.869,
       "step": 11344
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3256374498550661e-05,
-      "loss": 0.7143,
+      "loss": 0.8008,
       "step": 11345
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3255260813980868e-05,
-      "loss": 0.8268,
+      "loss": 0.9253,
       "step": 11346
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3254147084248947e-05,
-      "loss": 0.755,
+      "loss": 0.8876,
       "step": 11347
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3253033309370356e-05,
-      "loss": 0.8169,
+      "loss": 0.8951,
       "step": 11348
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.325191948936054e-05,
-      "loss": 0.7271,
+      "loss": 0.8,
       "step": 11349
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3250805624234953e-05,
-      "loss": 0.7839,
+      "loss": 0.8144,
       "step": 11350
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3249691714009054e-05,
-      "loss": 0.765,
+      "loss": 0.8832,
       "step": 11351
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3248577758698293e-05,
-      "loss": 0.6145,
+      "loss": 0.7726,
       "step": 11352
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3247463758318122e-05,
-      "loss": 0.7808,
+      "loss": 0.853,
       "step": 11353
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3246349712884004e-05,
-      "loss": 0.8107,
+      "loss": 0.9498,
       "step": 11354
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3245235622411386e-05,
-      "loss": 0.7484,
+      "loss": 0.857,
       "step": 11355
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.324412148691573e-05,
-      "loss": 0.6932,
+      "loss": 0.8673,
       "step": 11356
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3243007306412487e-05,
-      "loss": 0.8387,
+      "loss": 0.8776,
       "step": 11357
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.324189308091712e-05,
-      "loss": 0.6641,
+      "loss": 0.9034,
       "step": 11358
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.324077881044509e-05,
-      "loss": 0.826,
+      "loss": 0.9431,
       "step": 11359
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3239664495011848e-05,
-      "loss": 0.8231,
+      "loss": 0.8792,
       "step": 11360
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3238550134632858e-05,
-      "loss": 0.7261,
+      "loss": 0.8161,
       "step": 11361
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3237435729323584e-05,
-      "loss": 0.737,
+      "loss": 0.8643,
       "step": 11362
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3236321279099479e-05,
-      "loss": 0.819,
+      "loss": 0.9174,
       "step": 11363
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.323520678397601e-05,
-      "loss": 0.8458,
+      "loss": 0.9236,
       "step": 11364
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3234092243968636e-05,
-      "loss": 0.7647,
+      "loss": 0.8747,
       "step": 11365
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3232977659092818e-05,
-      "loss": 0.6639,
+      "loss": 0.8548,
       "step": 11366
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3231863029364025e-05,
-      "loss": 0.8782,
+      "loss": 0.9732,
       "step": 11367
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3230748354797717e-05,
-      "loss": 0.7126,
+      "loss": 0.831,
       "step": 11368
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3229633635409358e-05,
-      "loss": 0.9979,
+      "loss": 1.0064,
       "step": 11369
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3228518871214418e-05,
-      "loss": 0.8177,
+      "loss": 0.9044,
       "step": 11370
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3227404062228355e-05,
-      "loss": 0.7513,
+      "loss": 0.8881,
       "step": 11371
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3226289208466645e-05,
-      "loss": 0.7788,
+      "loss": 0.8978,
       "step": 11372
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3225174309944748e-05,
-      "loss": 0.8887,
+      "loss": 0.9511,
       "step": 11373
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3224059366678132e-05,
-      "loss": 0.836,
+      "loss": 1.0127,
       "step": 11374
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3222944378682268e-05,
-      "loss": 0.8235,
+      "loss": 0.9635,
       "step": 11375
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.322182934597262e-05,
-      "loss": 0.7757,
+      "loss": 0.8813,
       "step": 11376
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3220714268564664e-05,
-      "loss": 0.8595,
+      "loss": 0.9493,
       "step": 11377
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3219599146473869e-05,
-      "loss": 0.9333,
+      "loss": 0.9668,
       "step": 11378
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3218483979715699e-05,
-      "loss": 0.6619,
+      "loss": 0.8045,
       "step": 11379
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3217368768305633e-05,
-      "loss": 0.7997,
+      "loss": 0.938,
       "step": 11380
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3216253512259144e-05,
-      "loss": 0.6741,
+      "loss": 0.7755,
       "step": 11381
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3215138211591694e-05,
-      "loss": 0.7099,
+      "loss": 0.9062,
       "step": 11382
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3214022866318765e-05,
-      "loss": 0.7296,
+      "loss": 0.9452,
       "step": 11383
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3212907476455832e-05,
-      "loss": 0.7472,
+      "loss": 0.8665,
       "step": 11384
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3211792042018361e-05,
-      "loss": 0.6542,
+      "loss": 0.808,
       "step": 11385
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3210676563021836e-05,
-      "loss": 0.8241,
+      "loss": 0.8416,
       "step": 11386
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3209561039481727e-05,
-      "loss": 0.8647,
+      "loss": 0.9355,
       "step": 11387
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3208445471413514e-05,
-      "loss": 0.7827,
+      "loss": 0.9039,
       "step": 11388
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3207329858832671e-05,
-      "loss": 0.6907,
+      "loss": 0.813,
       "step": 11389
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3206214201754678e-05,
-      "loss": 0.7965,
+      "loss": 0.9177,
       "step": 11390
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3205098500195013e-05,
-      "loss": 0.7913,
+      "loss": 0.8629,
       "step": 11391
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.320398275416915e-05,
-      "loss": 0.7572,
+      "loss": 0.9275,
       "step": 11392
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3202866963692578e-05,
-      "loss": 0.7464,
+      "loss": 0.8949,
       "step": 11393
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3201751128780767e-05,
-      "loss": 0.7979,
+      "loss": 0.9169,
       "step": 11394
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3200635249449203e-05,
-      "loss": 0.8525,
+      "loss": 0.8873,
       "step": 11395
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3199519325713364e-05,
-      "loss": 0.6832,
+      "loss": 0.8855,
       "step": 11396
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3198403357588739e-05,
-      "loss": 0.7823,
+      "loss": 0.8766,
       "step": 11397
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3197287345090799e-05,
-      "loss": 0.6498,
+      "loss": 0.8243,
       "step": 11398
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3196171288235036e-05,
-      "loss": 0.8127,
+      "loss": 0.9256,
       "step": 11399
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.319505518703693e-05,
-      "loss": 0.8239,
+      "loss": 0.8749,
       "step": 11400
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.319393904151197e-05,
-      "loss": 0.6853,
+      "loss": 0.8463,
       "step": 11401
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3192822851675633e-05,
-      "loss": 0.7697,
+      "loss": 0.8329,
       "step": 11402
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3191706617543411e-05,
-      "loss": 0.8431,
+      "loss": 0.9775,
       "step": 11403
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3190590339130789e-05,
-      "loss": 0.7149,
+      "loss": 0.8434,
       "step": 11404
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3189474016453253e-05,
-      "loss": 0.6778,
+      "loss": 0.8413,
       "step": 11405
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3188357649526288e-05,
-      "loss": 0.8275,
+      "loss": 0.8737,
       "step": 11406
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3187241238365387e-05,
-      "loss": 0.9805,
+      "loss": 1.0674,
       "step": 11407
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3186124782986035e-05,
-      "loss": 0.5784,
+      "loss": 0.7127,
       "step": 11408
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.318500828340372e-05,
-      "loss": 0.7086,
+      "loss": 0.8345,
       "step": 11409
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3183891739633936e-05,
-      "loss": 0.8906,
+      "loss": 0.959,
       "step": 11410
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.3182775151692171e-05,
-      "loss": 0.7703,
+      "loss": 0.8578,
       "step": 11411
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3181658519593918e-05,
-      "loss": 0.7211,
+      "loss": 0.8414,
       "step": 11412
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3180541843354668e-05,
-      "loss": 0.8075,
+      "loss": 0.851,
       "step": 11413
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3179425122989909e-05,
-      "loss": 0.6962,
+      "loss": 0.898,
       "step": 11414
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3178308358515139e-05,
-      "loss": 0.8695,
+      "loss": 0.947,
       "step": 11415
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.317719154994585e-05,
-      "loss": 0.7429,
+      "loss": 0.865,
       "step": 11416
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3176074697297536e-05,
-      "loss": 0.7256,
+      "loss": 0.7884,
       "step": 11417
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3174957800585693e-05,
-      "loss": 0.8777,
+      "loss": 0.9535,
       "step": 11418
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3173840859825813e-05,
-      "loss": 0.8802,
+      "loss": 0.9211,
       "step": 11419
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3172723875033395e-05,
-      "loss": 0.7773,
+      "loss": 0.8698,
       "step": 11420
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3171606846223935e-05,
-      "loss": 0.8,
+      "loss": 0.8613,
       "step": 11421
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3170489773412928e-05,
-      "loss": 0.8019,
+      "loss": 0.9035,
       "step": 11422
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3169372656615878e-05,
-      "loss": 0.7746,
+      "loss": 0.8898,
       "step": 11423
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3168255495848277e-05,
-      "loss": 0.7166,
+      "loss": 0.8016,
       "step": 11424
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3167138291125626e-05,
-      "loss": 0.7092,
+      "loss": 0.8153,
       "step": 11425
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3166021042463423e-05,
-      "loss": 0.7849,
+      "loss": 0.9388,
       "step": 11426
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3164903749877172e-05,
-      "loss": 0.9132,
+      "loss": 1.0704,
       "step": 11427
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.316378641338237e-05,
-      "loss": 0.7329,
+      "loss": 0.8325,
       "step": 11428
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3162669032994524e-05,
-      "loss": 0.7503,
+      "loss": 0.8274,
       "step": 11429
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.316155160872913e-05,
-      "loss": 0.8331,
+      "loss": 0.891,
       "step": 11430
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3160434140601694e-05,
-      "loss": 0.8455,
+      "loss": 0.8668,
       "step": 11431
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.315931662862772e-05,
-      "loss": 0.6791,
+      "loss": 0.8681,
       "step": 11432
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3158199072822707e-05,
-      "loss": 0.6701,
+      "loss": 0.7924,
       "step": 11433
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3157081473202165e-05,
-      "loss": 0.7411,
+      "loss": 0.9507,
       "step": 11434
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3155963829781593e-05,
-      "loss": 0.7648,
+      "loss": 0.9077,
       "step": 11435
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3154846142576505e-05,
-      "loss": 0.7437,
+      "loss": 0.8878,
       "step": 11436
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3153728411602402e-05,
-      "loss": 0.9462,
+      "loss": 0.9561,
       "step": 11437
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.315261063687479e-05,
-      "loss": 0.7421,
+      "loss": 0.8227,
       "step": 11438
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3151492818409179e-05,
-      "loss": 0.7758,
+      "loss": 0.913,
       "step": 11439
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3150374956221077e-05,
-      "loss": 0.7287,
+      "loss": 0.894,
       "step": 11440
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3149257050325992e-05,
-      "loss": 0.6102,
+      "loss": 0.8162,
       "step": 11441
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3148139100739435e-05,
-      "loss": 0.7373,
+      "loss": 0.8231,
       "step": 11442
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3147021107476911e-05,
-      "loss": 0.704,
+      "loss": 0.848,
       "step": 11443
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3145903070553938e-05,
-      "loss": 0.7007,
+      "loss": 0.7612,
       "step": 11444
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3144784989986023e-05,
-      "loss": 0.7618,
+      "loss": 0.8234,
       "step": 11445
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3143666865788675e-05,
-      "loss": 0.9799,
+      "loss": 0.9731,
       "step": 11446
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3142548697977413e-05,
-      "loss": 0.7546,
+      "loss": 0.8836,
       "step": 11447
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3141430486567746e-05,
-      "loss": 0.8205,
+      "loss": 0.9643,
       "step": 11448
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3140312231575186e-05,
-      "loss": 0.696,
+      "loss": 0.791,
       "step": 11449
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3139193933015254e-05,
-      "loss": 0.7819,
+      "loss": 0.8852,
       "step": 11450
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3138075590903454e-05,
-      "loss": 0.9302,
+      "loss": 0.9303,
       "step": 11451
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3136957205255311e-05,
-      "loss": 0.7259,
+      "loss": 0.8764,
       "step": 11452
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3135838776086338e-05,
-      "loss": 0.9183,
+      "loss": 0.9641,
       "step": 11453
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3134720303412048e-05,
-      "loss": 0.7205,
+      "loss": 0.8733,
       "step": 11454
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3133601787247964e-05,
-      "loss": 0.5832,
+      "loss": 0.7895,
       "step": 11455
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3132483227609603e-05,
-      "loss": 0.7759,
+      "loss": 0.8748,
       "step": 11456
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3131364624512477e-05,
-      "loss": 0.7257,
+      "loss": 0.8773,
       "step": 11457
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3130245977972111e-05,
-      "loss": 0.7875,
+      "loss": 0.9057,
       "step": 11458
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3129127288004024e-05,
-      "loss": 0.7703,
+      "loss": 0.8803,
       "step": 11459
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3128008554623737e-05,
-      "loss": 0.7429,
+      "loss": 0.845,
       "step": 11460
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3126889777846767e-05,
-      "loss": 0.8223,
+      "loss": 0.945,
       "step": 11461
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3125770957688637e-05,
-      "loss": 0.646,
+      "loss": 0.8441,
       "step": 11462
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3124652094164873e-05,
-      "loss": 0.7215,
+      "loss": 0.9052,
       "step": 11463
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3123533187290995e-05,
-      "loss": 0.6065,
+      "loss": 0.8481,
       "step": 11464
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3122414237082521e-05,
-      "loss": 0.7136,
+      "loss": 0.8815,
       "step": 11465
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3121295243554983e-05,
-      "loss": 0.7537,
+      "loss": 0.895,
       "step": 11466
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3120176206723903e-05,
-      "loss": 0.6958,
+      "loss": 0.7481,
       "step": 11467
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3119057126604802e-05,
-      "loss": 0.6897,
+      "loss": 0.8452,
       "step": 11468
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3117938003213214e-05,
-      "loss": 0.882,
+      "loss": 0.9545,
       "step": 11469
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3116818836564656e-05,
-      "loss": 0.609,
+      "loss": 0.7418,
       "step": 11470
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3115699626674659e-05,
-      "loss": 0.6688,
+      "loss": 0.7416,
       "step": 11471
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.311458037355875e-05,
-      "loss": 0.8552,
+      "loss": 0.984,
       "step": 11472
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3113461077232456e-05,
-      "loss": 0.8386,
+      "loss": 0.9153,
       "step": 11473
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.311234173771131e-05,
-      "loss": 0.8039,
+      "loss": 0.9316,
       "step": 11474
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3111222355010835e-05,
-      "loss": 0.67,
+      "loss": 0.8094,
       "step": 11475
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3110102929146566e-05,
-      "loss": 0.7182,
+      "loss": 0.8369,
       "step": 11476
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3108983460134033e-05,
-      "loss": 0.7472,
+      "loss": 0.7412,
       "step": 11477
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3107863947988764e-05,
-      "loss": 0.7596,
+      "loss": 0.8239,
       "step": 11478
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3106744392726292e-05,
-      "loss": 0.7955,
+      "loss": 0.8711,
       "step": 11479
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3105624794362151e-05,
-      "loss": 0.7108,
+      "loss": 0.8396,
       "step": 11480
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3104505152911871e-05,
-      "loss": 0.7582,
+      "loss": 0.8495,
       "step": 11481
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3103385468390986e-05,
-      "loss": 0.7668,
+      "loss": 0.8963,
       "step": 11482
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.310226574081503e-05,
-      "loss": 0.728,
+      "loss": 0.9089,
       "step": 11483
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3101145970199541e-05,
-      "loss": 0.6497,
+      "loss": 0.7498,
       "step": 11484
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.310002615656005e-05,
-      "loss": 0.6323,
+      "loss": 0.7798,
       "step": 11485
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3098906299912094e-05,
-      "loss": 0.8061,
+      "loss": 0.8918,
       "step": 11486
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3097786400271213e-05,
-      "loss": 0.7515,
+      "loss": 0.7579,
       "step": 11487
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3096666457652938e-05,
-      "loss": 0.8061,
+      "loss": 0.9537,
       "step": 11488
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3095546472072811e-05,
-      "loss": 0.6716,
+      "loss": 0.7982,
       "step": 11489
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3094426443546369e-05,
-      "loss": 0.7526,
+      "loss": 0.8584,
       "step": 11490
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3093306372089148e-05,
-      "loss": 0.7403,
+      "loss": 0.8415,
       "step": 11491
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3092186257716693e-05,
-      "loss": 0.6481,
+      "loss": 0.813,
       "step": 11492
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3091066100444537e-05,
-      "loss": 0.8327,
+      "loss": 0.9058,
       "step": 11493
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3089945900288226e-05,
-      "loss": 0.8794,
+      "loss": 1.0341,
       "step": 11494
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3088825657263302e-05,
-      "loss": 0.7966,
+      "loss": 0.8651,
       "step": 11495
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3087705371385304e-05,
-      "loss": 0.6701,
+      "loss": 0.8103,
       "step": 11496
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3086585042669773e-05,
-      "loss": 0.7408,
+      "loss": 0.8746,
       "step": 11497
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3085464671132253e-05,
-      "loss": 0.6552,
+      "loss": 0.8222,
       "step": 11498
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.308434425678829e-05,
-      "loss": 0.7677,
+      "loss": 0.895,
       "step": 11499
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3083223799653429e-05,
-      "loss": 0.7516,
+      "loss": 0.8908,
       "step": 11500
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3082103299743208e-05,
-      "loss": 0.8469,
+      "loss": 0.946,
       "step": 11501
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3080982757073178e-05,
-      "loss": 0.8195,
+      "loss": 0.8964,
       "step": 11502
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3079862171658883e-05,
-      "loss": 0.7876,
+      "loss": 0.8632,
       "step": 11503
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3078741543515873e-05,
-      "loss": 0.7959,
+      "loss": 0.9273,
       "step": 11504
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3077620872659688e-05,
-      "loss": 0.8752,
+      "loss": 0.9308,
       "step": 11505
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3076500159105885e-05,
-      "loss": 0.7818,
+      "loss": 0.8587,
       "step": 11506
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3075379402870006e-05,
-      "loss": 0.7601,
+      "loss": 0.8794,
       "step": 11507
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.30742586039676e-05,
-      "loss": 0.7248,
+      "loss": 0.8532,
       "step": 11508
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3073137762414217e-05,
-      "loss": 0.7055,
+      "loss": 0.8599,
       "step": 11509
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3072016878225408e-05,
-      "loss": 0.9072,
+      "loss": 1.0002,
       "step": 11510
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3070895951416724e-05,
-      "loss": 0.8263,
+      "loss": 0.9626,
       "step": 11511
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3069774982003716e-05,
-      "loss": 0.6064,
+      "loss": 0.7972,
       "step": 11512
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3068653970001935e-05,
-      "loss": 0.6981,
+      "loss": 0.7964,
       "step": 11513
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3067532915426933e-05,
-      "loss": 0.7002,
+      "loss": 0.8562,
       "step": 11514
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3066411818294264e-05,
-      "loss": 0.675,
+      "loss": 0.8764,
       "step": 11515
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3065290678619485e-05,
-      "loss": 0.8468,
+      "loss": 0.9098,
       "step": 11516
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3064169496418144e-05,
-      "loss": 0.8269,
+      "loss": 0.8617,
       "step": 11517
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3063048271705799e-05,
-      "loss": 0.7758,
+      "loss": 0.9124,
       "step": 11518
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3061927004498006e-05,
-      "loss": 0.7031,
+      "loss": 0.8061,
       "step": 11519
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.306080569481032e-05,
-      "loss": 0.8435,
+      "loss": 0.9216,
       "step": 11520
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3059684342658294e-05,
-      "loss": 0.6523,
+      "loss": 0.8227,
       "step": 11521
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3058562948057495e-05,
-      "loss": 0.8352,
+      "loss": 0.8464,
       "step": 11522
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3057441511023471e-05,
-      "loss": 0.7608,
+      "loss": 0.8774,
       "step": 11523
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3056320031571783e-05,
-      "loss": 0.7379,
+      "loss": 0.8557,
       "step": 11524
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3055198509717995e-05,
-      "loss": 0.6665,
+      "loss": 0.863,
       "step": 11525
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3054076945477658e-05,
-      "loss": 0.7541,
+      "loss": 0.9371,
       "step": 11526
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3052955338866343e-05,
-      "loss": 0.7364,
+      "loss": 0.8232,
       "step": 11527
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.30518336898996e-05,
-      "loss": 0.7775,
+      "loss": 0.9415,
       "step": 11528
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3050711998592993e-05,
-      "loss": 0.966,
+      "loss": 0.988,
       "step": 11529
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3049590264962086e-05,
-      "loss": 0.7906,
+      "loss": 0.877,
       "step": 11530
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.304846848902244e-05,
-      "loss": 0.6382,
+      "loss": 0.8248,
       "step": 11531
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3047346670789622e-05,
-      "loss": 0.6533,
+      "loss": 0.7871,
       "step": 11532
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3046224810279193e-05,
-      "loss": 0.8528,
+      "loss": 0.9076,
       "step": 11533
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3045102907506712e-05,
-      "loss": 0.7245,
+      "loss": 0.8781,
       "step": 11534
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3043980962487752e-05,
-      "loss": 0.7096,
+      "loss": 0.7988,
       "step": 11535
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3042858975237877e-05,
-      "loss": 0.6794,
+      "loss": 0.8094,
       "step": 11536
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3041736945772648e-05,
-      "loss": 0.6371,
+      "loss": 0.8283,
       "step": 11537
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3040614874107636e-05,
-      "loss": 0.7865,
+      "loss": 0.8445,
       "step": 11538
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3039492760258405e-05,
-      "loss": 0.9147,
+      "loss": 0.967,
       "step": 11539
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3038370604240524e-05,
-      "loss": 0.7545,
+      "loss": 0.8249,
       "step": 11540
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3037248406069564e-05,
-      "loss": 0.7945,
+      "loss": 0.9034,
       "step": 11541
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.303612616576109e-05,
-      "loss": 0.7326,
+      "loss": 0.8253,
       "step": 11542
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3035003883330675e-05,
-      "loss": 0.8789,
+      "loss": 0.9154,
       "step": 11543
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3033881558793887e-05,
-      "loss": 0.8263,
+      "loss": 0.9407,
       "step": 11544
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3032759192166293e-05,
-      "loss": 0.9235,
+      "loss": 0.9713,
       "step": 11545
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3031636783463473e-05,
-      "loss": 0.8848,
+      "loss": 0.9561,
       "step": 11546
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3030514332700993e-05,
-      "loss": 0.8335,
+      "loss": 0.8833,
       "step": 11547
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3029391839894426e-05,
-      "loss": 0.8829,
+      "loss": 0.9482,
       "step": 11548
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3028269305059345e-05,
-      "loss": 0.7706,
+      "loss": 0.9066,
       "step": 11549
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3027146728211321e-05,
-      "loss": 0.7475,
+      "loss": 0.913,
       "step": 11550
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3026024109365936e-05,
-      "loss": 0.6599,
+      "loss": 0.7618,
       "step": 11551
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3024901448538758e-05,
-      "loss": 0.7759,
+      "loss": 0.8292,
       "step": 11552
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3023778745745361e-05,
-      "loss": 0.7487,
+      "loss": 0.8614,
       "step": 11553
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3022656001001328e-05,
-      "loss": 0.8958,
+      "loss": 0.9472,
       "step": 11554
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3021533214322228e-05,
-      "loss": 0.6791,
+      "loss": 0.7767,
       "step": 11555
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3020410385723645e-05,
-      "loss": 0.8792,
+      "loss": 0.9376,
       "step": 11556
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3019287515221155e-05,
-      "loss": 0.7246,
+      "loss": 0.9004,
       "step": 11557
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.301816460283033e-05,
-      "loss": 0.8071,
+      "loss": 0.9194,
       "step": 11558
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3017041648566754e-05,
-      "loss": 0.8461,
+      "loss": 0.8623,
       "step": 11559
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.301591865244601e-05,
-      "loss": 0.6898,
+      "loss": 0.7795,
       "step": 11560
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3014795614483671e-05,
-      "loss": 0.8497,
+      "loss": 0.921,
       "step": 11561
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3013672534695318e-05,
-      "loss": 0.8468,
+      "loss": 0.9796,
       "step": 11562
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3012549413096539e-05,
-      "loss": 0.7733,
+      "loss": 0.8658,
       "step": 11563
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3011426249702907e-05,
-      "loss": 0.8235,
+      "loss": 0.8549,
       "step": 11564
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3010303044530012e-05,
-      "loss": 0.7655,
+      "loss": 0.8873,
       "step": 11565
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.300917979759343e-05,
-      "loss": 0.8228,
+      "loss": 0.8358,
       "step": 11566
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3008056508908751e-05,
-      "loss": 0.8276,
+      "loss": 0.8942,
       "step": 11567
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3006933178491554e-05,
-      "loss": 0.9339,
+      "loss": 1.0084,
       "step": 11568
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3005809806357424e-05,
-      "loss": 0.6819,
+      "loss": 0.8402,
       "step": 11569
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.300468639252195e-05,
-      "loss": 0.7108,
+      "loss": 0.8781,
       "step": 11570
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3003562937000714e-05,
-      "loss": 0.612,
+      "loss": 0.8097,
       "step": 11571
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3002439439809304e-05,
-      "loss": 0.7517,
+      "loss": 0.8286,
       "step": 11572
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3001315900963308e-05,
-      "loss": 0.7545,
+      "loss": 0.8434,
       "step": 11573
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.3000192320478308e-05,
-      "loss": 0.8534,
+      "loss": 0.9168,
       "step": 11574
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2999068698369901e-05,
-      "loss": 0.7486,
+      "loss": 0.869,
       "step": 11575
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2997945034653669e-05,
-      "loss": 0.7269,
+      "loss": 0.8864,
       "step": 11576
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2996821329345204e-05,
-      "loss": 0.68,
+      "loss": 0.8671,
       "step": 11577
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2995697582460094e-05,
-      "loss": 0.8505,
+      "loss": 0.9454,
       "step": 11578
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2994573794013932e-05,
-      "loss": 0.7809,
+      "loss": 0.9287,
       "step": 11579
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2993449964022304e-05,
-      "loss": 0.9135,
+      "loss": 0.9286,
       "step": 11580
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.299232609250081e-05,
-      "loss": 0.694,
+      "loss": 0.8227,
       "step": 11581
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.299120217946503e-05,
-      "loss": 0.8131,
+      "loss": 0.8748,
       "step": 11582
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.299007822493057e-05,
-      "loss": 0.7606,
+      "loss": 0.9254,
       "step": 11583
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2988954228913019e-05,
-      "loss": 0.7718,
+      "loss": 0.9197,
       "step": 11584
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2987830191427962e-05,
-      "loss": 0.7826,
+      "loss": 0.8789,
       "step": 11585
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2986706112491004e-05,
-      "loss": 0.8295,
+      "loss": 0.9034,
       "step": 11586
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2985581992117732e-05,
-      "loss": 0.7757,
+      "loss": 0.8711,
       "step": 11587
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2984457830323752e-05,
-      "loss": 0.7243,
+      "loss": 0.7918,
       "step": 11588
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.298333362712465e-05,
-      "loss": 0.6882,
+      "loss": 0.844,
       "step": 11589
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2982209382536026e-05,
-      "loss": 0.7721,
+      "loss": 0.8735,
       "step": 11590
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.298108509657348e-05,
-      "loss": 0.6631,
+      "loss": 0.7473,
       "step": 11591
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2979960769252609e-05,
-      "loss": 0.7505,
+      "loss": 0.8288,
       "step": 11592
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2978836400589008e-05,
-      "loss": 0.7406,
+      "loss": 0.8356,
       "step": 11593
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.297771199059828e-05,
-      "loss": 0.7896,
+      "loss": 0.9274,
       "step": 11594
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2976587539296021e-05,
-      "loss": 0.6631,
+      "loss": 0.8303,
       "step": 11595
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2975463046697835e-05,
-      "loss": 0.6665,
+      "loss": 0.8765,
       "step": 11596
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.297433851281932e-05,
-      "loss": 0.7625,
+      "loss": 0.8669,
       "step": 11597
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2973213937676076e-05,
-      "loss": 0.7055,
+      "loss": 0.9298,
       "step": 11598
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2972089321283712e-05,
-      "loss": 0.7694,
+      "loss": 0.9066,
       "step": 11599
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2970964663657824e-05,
-      "loss": 0.697,
+      "loss": 0.8421,
       "step": 11600
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2969839964814014e-05,
-      "loss": 0.713,
+      "loss": 0.8701,
       "step": 11601
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.296871522476789e-05,
-      "loss": 0.6867,
+      "loss": 0.8395,
       "step": 11602
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2967590443535056e-05,
-      "loss": 0.6822,
+      "loss": 0.7725,
       "step": 11603
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2966465621131113e-05,
-      "loss": 0.7812,
+      "loss": 0.8888,
       "step": 11604
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2965340757571669e-05,
-      "loss": 0.8678,
+      "loss": 0.9535,
       "step": 11605
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2964215852872326e-05,
-      "loss": 0.7363,
+      "loss": 0.8714,
       "step": 11606
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2963090907048698e-05,
-      "loss": 0.781,
+      "loss": 0.9453,
       "step": 11607
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2961965920116386e-05,
-      "loss": 0.8275,
+      "loss": 0.9081,
       "step": 11608
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2960840892090998e-05,
-      "loss": 0.7564,
+      "loss": 0.9137,
       "step": 11609
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2959715822988148e-05,
-      "loss": 0.721,
+      "loss": 0.8034,
       "step": 11610
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2958590712823437e-05,
-      "loss": 0.7807,
+      "loss": 0.9419,
       "step": 11611
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2957465561612477e-05,
-      "loss": 0.9539,
+      "loss": 0.8838,
       "step": 11612
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.295634036937088e-05,
-      "loss": 0.7476,
+      "loss": 0.8133,
       "step": 11613
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2955215136114253e-05,
-      "loss": 0.7417,
+      "loss": 0.8619,
       "step": 11614
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.295408986185821e-05,
-      "loss": 0.7513,
+      "loss": 0.8358,
       "step": 11615
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2952964546618363e-05,
-      "loss": 0.6561,
+      "loss": 0.847,
       "step": 11616
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2951839190410318e-05,
-      "loss": 0.8884,
+      "loss": 1.0031,
       "step": 11617
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2950713793249696e-05,
-      "loss": 0.7937,
+      "loss": 0.82,
       "step": 11618
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2949588355152107e-05,
-      "loss": 0.7427,
+      "loss": 0.875,
       "step": 11619
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2948462876133161e-05,
-      "loss": 0.6623,
+      "loss": 0.8589,
       "step": 11620
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.294733735620848e-05,
-      "loss": 0.6779,
+      "loss": 0.8961,
       "step": 11621
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2946211795393675e-05,
-      "loss": 0.8571,
+      "loss": 0.9023,
       "step": 11622
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2945086193704359e-05,
-      "loss": 0.7593,
+      "loss": 0.8412,
       "step": 11623
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2943960551156152e-05,
-      "loss": 0.7048,
+      "loss": 0.8825,
       "step": 11624
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2942834867764669e-05,
-      "loss": 0.8395,
+      "loss": 0.9653,
       "step": 11625
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2941709143545526e-05,
-      "loss": 0.7044,
+      "loss": 0.8295,
       "step": 11626
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2940583378514345e-05,
-      "loss": 0.8404,
+      "loss": 0.8488,
       "step": 11627
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.293945757268674e-05,
-      "loss": 0.8742,
+      "loss": 0.9646,
       "step": 11628
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2938331726078332e-05,
-      "loss": 0.8383,
+      "loss": 0.8802,
       "step": 11629
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2937205838704739e-05,
-      "loss": 0.8415,
+      "loss": 0.9268,
       "step": 11630
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2936079910581585e-05,
-      "loss": 0.7381,
+      "loss": 0.8389,
       "step": 11631
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2934953941724485e-05,
-      "loss": 0.7816,
+      "loss": 0.8979,
       "step": 11632
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2933827932149066e-05,
-      "loss": 0.7663,
+      "loss": 0.91,
       "step": 11633
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2932701881870947e-05,
-      "loss": 0.8032,
+      "loss": 0.9217,
       "step": 11634
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.293157579090575e-05,
-      "loss": 0.7565,
+      "loss": 0.8166,
       "step": 11635
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2930449659269095e-05,
-      "loss": 0.8502,
+      "loss": 0.8761,
       "step": 11636
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2929323486976612e-05,
-      "loss": 0.7286,
+      "loss": 0.804,
       "step": 11637
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2928197274043921e-05,
-      "loss": 0.8016,
+      "loss": 0.8429,
       "step": 11638
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2927071020486648e-05,
-      "loss": 0.5566,
+      "loss": 0.7815,
       "step": 11639
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2925944726320416e-05,
-      "loss": 0.8341,
+      "loss": 0.8444,
       "step": 11640
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.292481839156085e-05,
-      "loss": 0.7995,
+      "loss": 0.8856,
       "step": 11641
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2923692016223587e-05,
-      "loss": 0.7004,
+      "loss": 0.8724,
       "step": 11642
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.292256560032424e-05,
-      "loss": 0.8443,
+      "loss": 0.8803,
       "step": 11643
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2921439143878442e-05,
-      "loss": 0.7915,
+      "loss": 0.8624,
       "step": 11644
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2920312646901822e-05,
-      "loss": 0.7153,
+      "loss": 0.831,
       "step": 11645
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2919186109410004e-05,
-      "loss": 0.8159,
+      "loss": 0.8207,
       "step": 11646
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2918059531418626e-05,
-      "loss": 0.7993,
+      "loss": 0.9022,
       "step": 11647
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.291693291294331e-05,
-      "loss": 0.7134,
+      "loss": 0.8625,
       "step": 11648
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2915806253999686e-05,
-      "loss": 0.7969,
+      "loss": 0.8828,
       "step": 11649
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2914679554603389e-05,
-      "loss": 0.8811,
+      "loss": 0.9707,
       "step": 11650
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2913552814770047e-05,
-      "loss": 0.675,
+      "loss": 0.8236,
       "step": 11651
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2912426034515295e-05,
-      "loss": 0.7271,
+      "loss": 0.8959,
       "step": 11652
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2911299213854764e-05,
-      "loss": 0.6881,
+      "loss": 0.8459,
       "step": 11653
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2910172352804087e-05,
-      "loss": 0.7138,
+      "loss": 0.8111,
       "step": 11654
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2909045451378897e-05,
-      "loss": 0.8858,
+      "loss": 0.9723,
       "step": 11655
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.290791850959483e-05,
-      "loss": 0.6428,
+      "loss": 0.8816,
       "step": 11656
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2906791527467517e-05,
-      "loss": 0.7747,
+      "loss": 0.8351,
       "step": 11657
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2905664505012599e-05,
-      "loss": 0.8173,
+      "loss": 1.019,
       "step": 11658
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2904537442245706e-05,
-      "loss": 0.8144,
+      "loss": 0.9473,
       "step": 11659
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2903410339182477e-05,
-      "loss": 0.6535,
+      "loss": 0.7576,
       "step": 11660
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2902283195838554e-05,
-      "loss": 0.7951,
+      "loss": 0.931,
       "step": 11661
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2901156012229562e-05,
-      "loss": 0.6425,
+      "loss": 0.7799,
       "step": 11662
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.290002878837115e-05,
-      "loss": 0.714,
+      "loss": 0.8651,
       "step": 11663
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2898901524278955e-05,
-      "loss": 0.7714,
+      "loss": 0.8723,
       "step": 11664
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2897774219968611e-05,
-      "loss": 0.7958,
+      "loss": 0.8094,
       "step": 11665
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2896646875455763e-05,
-      "loss": 0.8743,
+      "loss": 0.883,
       "step": 11666
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.289551949075605e-05,
-      "loss": 0.6428,
+      "loss": 0.7518,
       "step": 11667
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.289439206588511e-05,
-      "loss": 0.7946,
+      "loss": 0.858,
       "step": 11668
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2893264600858588e-05,
-      "loss": 0.8234,
+      "loss": 0.9592,
       "step": 11669
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2892137095692124e-05,
-      "loss": 0.9229,
+      "loss": 1.0599,
       "step": 11670
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2891009550401364e-05,
-      "loss": 0.7104,
+      "loss": 0.8271,
       "step": 11671
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2889881965001945e-05,
-      "loss": 0.7876,
+      "loss": 0.9172,
       "step": 11672
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2888754339509513e-05,
-      "loss": 0.8411,
+      "loss": 0.9409,
       "step": 11673
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2887626673939718e-05,
-      "loss": 0.864,
+      "loss": 0.8503,
       "step": 11674
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2886498968308196e-05,
-      "loss": 0.8565,
+      "loss": 0.8984,
       "step": 11675
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2885371222630595e-05,
-      "loss": 0.796,
+      "loss": 0.8993,
       "step": 11676
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2884243436922566e-05,
-      "loss": 0.765,
+      "loss": 0.8735,
       "step": 11677
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2883115611199752e-05,
-      "loss": 0.9863,
+      "loss": 0.976,
       "step": 11678
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2881987745477796e-05,
-      "loss": 0.6495,
+      "loss": 0.8174,
       "step": 11679
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2880859839772352e-05,
-      "loss": 0.8373,
+      "loss": 0.902,
       "step": 11680
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2879731894099061e-05,
-      "loss": 0.6109,
+      "loss": 0.7136,
       "step": 11681
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2878603908473579e-05,
-      "loss": 0.7288,
+      "loss": 0.8646,
       "step": 11682
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2877475882911553e-05,
-      "loss": 0.8199,
+      "loss": 0.9098,
       "step": 11683
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.287634781742863e-05,
-      "loss": 0.7565,
+      "loss": 0.8305,
       "step": 11684
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.2875219712040461e-05,
-      "loss": 0.7461,
+      "loss": 0.8304,
       "step": 11685
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.28740915667627e-05,
-      "loss": 0.8225,
+      "loss": 0.9613,
       "step": 11686
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2872963381610995e-05,
-      "loss": 0.6951,
+      "loss": 0.8666,
       "step": 11687
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2871835156601e-05,
-      "loss": 0.8066,
+      "loss": 0.9351,
       "step": 11688
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2870706891748364e-05,
-      "loss": 0.7632,
+      "loss": 0.9382,
       "step": 11689
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2869578587068747e-05,
-      "loss": 0.7817,
+      "loss": 0.8676,
       "step": 11690
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2868450242577797e-05,
-      "loss": 0.8453,
+      "loss": 0.907,
       "step": 11691
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.286732185829117e-05,
-      "loss": 0.5663,
+      "loss": 0.7417,
       "step": 11692
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2866193434224518e-05,
-      "loss": 0.8609,
+      "loss": 0.9208,
       "step": 11693
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2865064970393502e-05,
-      "loss": 0.695,
+      "loss": 0.8656,
       "step": 11694
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2863936466813771e-05,
-      "loss": 0.8506,
+      "loss": 0.9487,
       "step": 11695
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2862807923500989e-05,
-      "loss": 0.6786,
+      "loss": 0.9002,
       "step": 11696
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2861679340470806e-05,
-      "loss": 0.744,
+      "loss": 0.881,
       "step": 11697
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2860550717738882e-05,
-      "loss": 0.7853,
+      "loss": 0.9076,
       "step": 11698
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2859422055320878e-05,
-      "loss": 0.8049,
+      "loss": 0.8994,
       "step": 11699
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2858293353232449e-05,
-      "loss": 0.7796,
+      "loss": 0.9038,
       "step": 11700
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2857164611489257e-05,
-      "loss": 0.9106,
+      "loss": 0.9246,
       "step": 11701
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2856035830106955e-05,
-      "loss": 0.811,
+      "loss": 0.8958,
       "step": 11702
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2854907009101212e-05,
-      "loss": 0.8025,
+      "loss": 0.8324,
       "step": 11703
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2853778148487683e-05,
-      "loss": 0.7272,
+      "loss": 0.8267,
       "step": 11704
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.285264924828203e-05,
-      "loss": 0.7177,
+      "loss": 0.9049,
       "step": 11705
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2851520308499918e-05,
-      "loss": 0.7081,
+      "loss": 0.8636,
       "step": 11706
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2850391329157009e-05,
-      "loss": 0.8663,
+      "loss": 0.9584,
       "step": 11707
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.284926231026896e-05,
-      "loss": 0.8656,
+      "loss": 0.9296,
       "step": 11708
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2848133251851442e-05,
-      "loss": 0.8208,
+      "loss": 0.9141,
       "step": 11709
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2847004153920115e-05,
-      "loss": 0.7538,
+      "loss": 0.8507,
       "step": 11710
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2845875016490646e-05,
-      "loss": 0.8221,
+      "loss": 0.9401,
       "step": 11711
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.28447458395787e-05,
-      "loss": 0.6572,
+      "loss": 0.7751,
       "step": 11712
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2843616623199938e-05,
-      "loss": 0.7971,
+      "loss": 0.8072,
       "step": 11713
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2842487367370033e-05,
-      "loss": 0.7276,
+      "loss": 0.8428,
       "step": 11714
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2841358072104647e-05,
-      "loss": 0.8647,
+      "loss": 0.9476,
       "step": 11715
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.284022873741945e-05,
-      "loss": 0.7169,
+      "loss": 0.8415,
       "step": 11716
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2839099363330108e-05,
-      "loss": 0.7774,
+      "loss": 0.8756,
       "step": 11717
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2837969949852291e-05,
-      "loss": 0.7549,
+      "loss": 0.9011,
       "step": 11718
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.283684049700167e-05,
-      "loss": 0.8155,
+      "loss": 0.8988,
       "step": 11719
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.283571100479391e-05,
-      "loss": 0.8027,
+      "loss": 0.884,
       "step": 11720
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2834581473244682e-05,
-      "loss": 0.8035,
+      "loss": 0.9586,
       "step": 11721
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.283345190236966e-05,
-      "loss": 0.8056,
+      "loss": 0.8779,
       "step": 11722
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.283232229218451e-05,
-      "loss": 0.7398,
+      "loss": 0.8544,
       "step": 11723
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2831192642704909e-05,
-      "loss": 0.7753,
+      "loss": 0.85,
       "step": 11724
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2830062953946527e-05,
-      "loss": 0.8293,
+      "loss": 0.852,
       "step": 11725
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2828933225925036e-05,
-      "loss": 0.7901,
+      "loss": 0.8772,
       "step": 11726
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2827803458656112e-05,
-      "loss": 0.6099,
+      "loss": 0.6656,
       "step": 11727
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2826673652155426e-05,
-      "loss": 0.8232,
+      "loss": 0.8967,
       "step": 11728
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2825543806438651e-05,
-      "loss": 0.8207,
+      "loss": 0.8903,
       "step": 11729
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2824413921521468e-05,
-      "loss": 0.9232,
+      "loss": 1.0086,
       "step": 11730
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2823283997419549e-05,
-      "loss": 0.758,
+      "loss": 0.8522,
       "step": 11731
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2822154034148567e-05,
-      "loss": 0.7515,
+      "loss": 0.8207,
       "step": 11732
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2821024031724205e-05,
-      "loss": 0.8057,
+      "loss": 0.8953,
       "step": 11733
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2819893990162138e-05,
-      "loss": 0.7128,
+      "loss": 0.9139,
       "step": 11734
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2818763909478041e-05,
-      "loss": 0.7043,
+      "loss": 0.8178,
       "step": 11735
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2817633789687593e-05,
-      "loss": 0.7156,
+      "loss": 0.8141,
       "step": 11736
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2816503630806474e-05,
-      "loss": 0.7626,
+      "loss": 0.7622,
       "step": 11737
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2815373432850367e-05,
-      "loss": 0.8225,
+      "loss": 0.9093,
       "step": 11738
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2814243195834945e-05,
-      "loss": 0.9101,
+      "loss": 0.998,
       "step": 11739
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2813112919775892e-05,
-      "loss": 0.6325,
+      "loss": 0.7905,
       "step": 11740
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2811982604688889e-05,
-      "loss": 0.78,
+      "loss": 0.8756,
       "step": 11741
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2810852250589614e-05,
-      "loss": 0.7445,
+      "loss": 0.9226,
       "step": 11742
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2809721857493756e-05,
-      "loss": 0.7658,
+      "loss": 0.8651,
       "step": 11743
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2808591425416992e-05,
-      "loss": 0.8239,
+      "loss": 0.88,
       "step": 11744
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2807460954375006e-05,
-      "loss": 0.6038,
+      "loss": 0.7831,
       "step": 11745
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2806330444383486e-05,
-      "loss": 0.6342,
+      "loss": 0.8213,
       "step": 11746
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2805199895458108e-05,
-      "loss": 0.7018,
+      "loss": 0.8319,
       "step": 11747
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2804069307614564e-05,
-      "loss": 0.8152,
+      "loss": 0.9554,
       "step": 11748
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2802938680868537e-05,
-      "loss": 0.8434,
+      "loss": 0.9095,
       "step": 11749
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2801808015235712e-05,
-      "loss": 0.8,
+      "loss": 0.9145,
       "step": 11750
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2800677310731777e-05,
-      "loss": 0.8159,
+      "loss": 0.8386,
       "step": 11751
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2799546567372418e-05,
-      "loss": 0.6351,
+      "loss": 0.7499,
       "step": 11752
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.279841578517332e-05,
-      "loss": 0.8496,
+      "loss": 0.9558,
       "step": 11753
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2797284964150177e-05,
-      "loss": 0.8021,
+      "loss": 0.8462,
       "step": 11754
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2796154104318672e-05,
-      "loss": 0.6296,
+      "loss": 0.8201,
       "step": 11755
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2795023205694494e-05,
-      "loss": 0.7857,
+      "loss": 0.9228,
       "step": 11756
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.279389226829334e-05,
-      "loss": 0.5913,
+      "loss": 0.7384,
       "step": 11757
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2792761292130891e-05,
-      "loss": 0.7343,
+      "loss": 0.9719,
       "step": 11758
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2791630277222842e-05,
-      "loss": 0.8139,
+      "loss": 0.8888,
       "step": 11759
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2790499223584885e-05,
-      "loss": 0.7133,
+      "loss": 0.7933,
       "step": 11760
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2789368131232707e-05,
-      "loss": 0.7999,
+      "loss": 0.8524,
       "step": 11761
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2788237000182007e-05,
-      "loss": 0.6906,
+      "loss": 0.8136,
       "step": 11762
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2787105830448474e-05,
-      "loss": 0.7298,
+      "loss": 0.7937,
       "step": 11763
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2785974622047801e-05,
-      "loss": 0.7144,
+      "loss": 0.8763,
       "step": 11764
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2784843374995683e-05,
-      "loss": 0.867,
+      "loss": 0.8987,
       "step": 11765
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2783712089307815e-05,
-      "loss": 0.677,
+      "loss": 0.8046,
       "step": 11766
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2782580764999892e-05,
-      "loss": 0.716,
+      "loss": 0.8434,
       "step": 11767
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2781449402087609e-05,
-      "loss": 0.9208,
+      "loss": 0.9819,
       "step": 11768
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2780318000586658e-05,
-      "loss": 0.781,
+      "loss": 0.8881,
       "step": 11769
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2779186560512745e-05,
-      "loss": 0.7808,
+      "loss": 0.8018,
       "step": 11770
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.277805508188156e-05,
-      "loss": 0.8335,
+      "loss": 0.8874,
       "step": 11771
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2776923564708798e-05,
-      "loss": 0.6376,
+      "loss": 0.7789,
       "step": 11772
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2775792009010166e-05,
-      "loss": 0.6602,
+      "loss": 0.8159,
       "step": 11773
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2774660414801359e-05,
-      "loss": 0.8351,
+      "loss": 0.9083,
       "step": 11774
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.277352878209807e-05,
-      "loss": 0.7643,
+      "loss": 0.9114,
       "step": 11775
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.277239711091601e-05,
-      "loss": 0.821,
+      "loss": 0.8804,
       "step": 11776
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2771265401270871e-05,
-      "loss": 0.6782,
+      "loss": 0.833,
       "step": 11777
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2770133653178358e-05,
-      "loss": 0.757,
+      "loss": 0.8396,
       "step": 11778
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.276900186665417e-05,
-      "loss": 0.8308,
+      "loss": 0.9139,
       "step": 11779
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2767870041714008e-05,
-      "loss": 0.7184,
+      "loss": 0.7915,
       "step": 11780
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2766738178373578e-05,
-      "loss": 0.8223,
+      "loss": 0.9229,
       "step": 11781
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.276560627664858e-05,
-      "loss": 0.6086,
+      "loss": 0.8014,
       "step": 11782
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2764474336554721e-05,
-      "loss": 0.65,
+      "loss": 0.8106,
       "step": 11783
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2763342358107702e-05,
-      "loss": 0.739,
+      "loss": 0.8811,
       "step": 11784
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2762210341323227e-05,
-      "loss": 0.7824,
+      "loss": 0.9367,
       "step": 11785
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2761078286217005e-05,
-      "loss": 0.7073,
+      "loss": 0.8425,
       "step": 11786
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.275994619280474e-05,
-      "loss": 0.8456,
+      "loss": 0.9681,
       "step": 11787
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2758814061102132e-05,
-      "loss": 0.7118,
+      "loss": 0.9047,
       "step": 11788
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2757681891124898e-05,
-      "loss": 0.8084,
+      "loss": 0.8449,
       "step": 11789
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.275654968288874e-05,
-      "loss": 0.6907,
+      "loss": 0.8428,
       "step": 11790
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2755417436409363e-05,
-      "loss": 0.7542,
+      "loss": 0.8629,
       "step": 11791
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2754285151702481e-05,
-      "loss": 0.7427,
+      "loss": 0.7795,
       "step": 11792
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2753152828783799e-05,
-      "loss": 0.7977,
+      "loss": 0.9498,
       "step": 11793
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.275202046766903e-05,
-      "loss": 0.78,
+      "loss": 0.8969,
       "step": 11794
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2750888068373883e-05,
-      "loss": 0.7103,
+      "loss": 0.7383,
       "step": 11795
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2749755630914063e-05,
-      "loss": 0.9016,
+      "loss": 0.939,
       "step": 11796
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2748623155305286e-05,
-      "loss": 0.8565,
+      "loss": 0.8385,
       "step": 11797
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2747490641563263e-05,
-      "loss": 0.8968,
+      "loss": 0.9162,
       "step": 11798
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2746358089703705e-05,
-      "loss": 0.751,
+      "loss": 0.7814,
       "step": 11799
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2745225499742327e-05,
-      "loss": 0.7443,
+      "loss": 0.776,
       "step": 11800
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2744092871694837e-05,
-      "loss": 0.7119,
+      "loss": 0.8266,
       "step": 11801
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2742960205576953e-05,
-      "loss": 0.6197,
+      "loss": 0.7695,
       "step": 11802
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.274182750140439e-05,
-      "loss": 0.7151,
+      "loss": 0.8987,
       "step": 11803
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2740694759192858e-05,
-      "loss": 0.7663,
+      "loss": 0.9381,
       "step": 11804
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2739561978958075e-05,
-      "loss": 0.7253,
+      "loss": 0.7978,
       "step": 11805
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2738429160715758e-05,
-      "loss": 0.8091,
+      "loss": 0.9517,
       "step": 11806
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.273729630448162e-05,
-      "loss": 0.8617,
+      "loss": 0.9249,
       "step": 11807
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2736163410271381e-05,
-      "loss": 0.732,
+      "loss": 0.8748,
       "step": 11808
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2735030478100754e-05,
-      "loss": 0.6812,
+      "loss": 0.8338,
       "step": 11809
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2733897507985462e-05,
-      "loss": 0.7906,
+      "loss": 0.905,
       "step": 11810
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2732764499941222e-05,
-      "loss": 0.7379,
+      "loss": 0.8671,
       "step": 11811
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.273163145398375e-05,
-      "loss": 0.6626,
+      "loss": 0.8677,
       "step": 11812
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2730498370128767e-05,
-      "loss": 0.8694,
+      "loss": 0.8761,
       "step": 11813
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2729365248391996e-05,
-      "loss": 0.8003,
+      "loss": 0.9311,
       "step": 11814
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2728232088789153e-05,
-      "loss": 0.7568,
+      "loss": 0.8644,
       "step": 11815
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.272709889133596e-05,
-      "loss": 0.8011,
+      "loss": 0.8909,
       "step": 11816
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2725965656048138e-05,
-      "loss": 0.744,
+      "loss": 0.8658,
       "step": 11817
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2724832382941412e-05,
-      "loss": 0.7799,
+      "loss": 0.8529,
       "step": 11818
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2723699072031503e-05,
-      "loss": 0.7402,
+      "loss": 0.8524,
       "step": 11819
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2722565723334133e-05,
-      "loss": 0.8391,
+      "loss": 0.9041,
       "step": 11820
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2721432336865028e-05,
-      "loss": 0.7449,
+      "loss": 0.8543,
       "step": 11821
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2720298912639908e-05,
-      "loss": 0.8506,
+      "loss": 0.8997,
       "step": 11822
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.27191654506745e-05,
-      "loss": 0.8268,
+      "loss": 0.9839,
       "step": 11823
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2718031950984531e-05,
-      "loss": 0.9542,
+      "loss": 0.9472,
       "step": 11824
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2716898413585724e-05,
-      "loss": 0.7087,
+      "loss": 0.8397,
       "step": 11825
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2715764838493808e-05,
-      "loss": 0.6098,
+      "loss": 0.7417,
       "step": 11826
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2714631225724507e-05,
-      "loss": 0.7875,
+      "loss": 0.791,
       "step": 11827
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2713497575293547e-05,
-      "loss": 0.7346,
+      "loss": 0.9078,
       "step": 11828
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.271236388721666e-05,
-      "loss": 0.7724,
+      "loss": 0.8591,
       "step": 11829
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2711230161509576e-05,
-      "loss": 0.7006,
+      "loss": 0.8624,
       "step": 11830
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2710096398188015e-05,
-      "loss": 0.7636,
+      "loss": 0.9544,
       "step": 11831
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2708962597267713e-05,
-      "loss": 0.8511,
+      "loss": 0.9388,
       "step": 11832
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.27078287587644e-05,
-      "loss": 0.6772,
+      "loss": 0.8156,
       "step": 11833
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2706694882693804e-05,
-      "loss": 0.7262,
+      "loss": 0.8222,
       "step": 11834
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.270556096907166e-05,
-      "loss": 0.8182,
+      "loss": 0.9014,
       "step": 11835
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.270442701791369e-05,
-      "loss": 0.6608,
+      "loss": 0.7894,
       "step": 11836
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2703293029235635e-05,
-      "loss": 0.7388,
+      "loss": 0.835,
       "step": 11837
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2702159003053226e-05,
-      "loss": 0.8053,
+      "loss": 0.9526,
       "step": 11838
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2701024939382191e-05,
-      "loss": 0.6006,
+      "loss": 0.7749,
       "step": 11839
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2699890838238271e-05,
-      "loss": 0.7789,
+      "loss": 0.9167,
       "step": 11840
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2698756699637196e-05,
-      "loss": 0.7906,
+      "loss": 0.8949,
       "step": 11841
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2697622523594699e-05,
-      "loss": 0.8002,
+      "loss": 0.857,
       "step": 11842
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2696488310126518e-05,
-      "loss": 0.8517,
+      "loss": 0.902,
       "step": 11843
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2695354059248386e-05,
-      "loss": 0.7148,
+      "loss": 0.8809,
       "step": 11844
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2694219770976041e-05,
-      "loss": 0.7443,
+      "loss": 0.8516,
       "step": 11845
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2693085445325223e-05,
-      "loss": 0.7753,
+      "loss": 0.8786,
       "step": 11846
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2691951082311662e-05,
-      "loss": 0.8291,
+      "loss": 0.8709,
       "step": 11847
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2690816681951099e-05,
-      "loss": 0.7992,
+      "loss": 0.9043,
       "step": 11848
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2689682244259274e-05,
-      "loss": 0.6471,
+      "loss": 0.8161,
       "step": 11849
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2688547769251922e-05,
-      "loss": 0.7855,
+      "loss": 0.8266,
       "step": 11850
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2687413256944785e-05,
-      "loss": 0.62,
+      "loss": 0.7661,
       "step": 11851
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2686278707353603e-05,
-      "loss": 0.5881,
+      "loss": 0.7289,
       "step": 11852
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2685144120494117e-05,
-      "loss": 0.7837,
+      "loss": 0.8763,
       "step": 11853
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2684009496382062e-05,
-      "loss": 0.707,
+      "loss": 0.8216,
       "step": 11854
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2682874835033185e-05,
-      "loss": 0.7909,
+      "loss": 0.9107,
       "step": 11855
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2681740136463227e-05,
-      "loss": 0.8872,
+      "loss": 0.9659,
       "step": 11856
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2680605400687927e-05,
-      "loss": 0.6852,
+      "loss": 0.8336,
       "step": 11857
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2679470627723035e-05,
-      "loss": 0.6817,
+      "loss": 0.8377,
       "step": 11858
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2678335817584286e-05,
-      "loss": 0.7685,
+      "loss": 0.836,
       "step": 11859
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2677200970287427e-05,
-      "loss": 0.8116,
+      "loss": 0.8773,
       "step": 11860
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2676066085848205e-05,
-      "loss": 0.8035,
+      "loss": 0.8641,
       "step": 11861
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2674931164282363e-05,
-      "loss": 0.7698,
+      "loss": 0.9238,
       "step": 11862
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2673796205605645e-05,
-      "loss": 0.7537,
+      "loss": 0.8813,
       "step": 11863
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2672661209833796e-05,
-      "loss": 0.7267,
+      "loss": 0.826,
       "step": 11864
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2671526176982567e-05,
-      "loss": 0.7527,
+      "loss": 0.9117,
       "step": 11865
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2670391107067706e-05,
-      "loss": 0.819,
+      "loss": 0.915,
       "step": 11866
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2669256000104953e-05,
-      "loss": 0.7104,
+      "loss": 0.8203,
       "step": 11867
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2668120856110062e-05,
-      "loss": 0.8074,
+      "loss": 0.901,
       "step": 11868
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2666985675098777e-05,
-      "loss": 0.7721,
+      "loss": 0.8383,
       "step": 11869
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2665850457086854e-05,
-      "loss": 0.5731,
+      "loss": 0.6731,
       "step": 11870
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2664715202090034e-05,
-      "loss": 0.7984,
+      "loss": 0.9235,
       "step": 11871
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2663579910124077e-05,
-      "loss": 0.8116,
+      "loss": 0.889,
       "step": 11872
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2662444581204723e-05,
-      "loss": 0.6297,
+      "loss": 0.8354,
       "step": 11873
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.266130921534773e-05,
-      "loss": 0.7567,
+      "loss": 0.8285,
       "step": 11874
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2660173812568848e-05,
-      "loss": 0.7853,
+      "loss": 0.8436,
       "step": 11875
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2659038372883825e-05,
-      "loss": 0.8419,
+      "loss": 0.9302,
       "step": 11876
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.265790289630842e-05,
-      "loss": 0.6818,
+      "loss": 0.8028,
       "step": 11877
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2656767382858386e-05,
-      "loss": 0.7126,
+      "loss": 0.8045,
       "step": 11878
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.265563183254947e-05,
-      "loss": 0.7659,
+      "loss": 0.8269,
       "step": 11879
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2654496245397432e-05,
-      "loss": 0.8026,
+      "loss": 0.8799,
       "step": 11880
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2653360621418023e-05,
-      "loss": 0.7899,
+      "loss": 0.8722,
       "step": 11881
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2652224960627e-05,
-      "loss": 0.7395,
+      "loss": 0.8153,
       "step": 11882
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2651089263040122e-05,
-      "loss": 0.8579,
+      "loss": 1.0024,
       "step": 11883
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2649953528673138e-05,
-      "loss": 0.8118,
+      "loss": 0.8448,
       "step": 11884
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2648817757541812e-05,
-      "loss": 0.758,
+      "loss": 0.8036,
       "step": 11885
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2647681949661896e-05,
-      "loss": 0.8752,
+      "loss": 0.8928,
       "step": 11886
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2646546105049149e-05,
-      "loss": 0.7558,
+      "loss": 0.8273,
       "step": 11887
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.264541022371933e-05,
-      "loss": 0.7094,
+      "loss": 0.8125,
       "step": 11888
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2644274305688198e-05,
-      "loss": 0.6969,
+      "loss": 0.8148,
       "step": 11889
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2643138350971512e-05,
-      "loss": 0.7461,
+      "loss": 0.8562,
       "step": 11890
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2642002359585033e-05,
-      "loss": 0.7034,
+      "loss": 0.9065,
       "step": 11891
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2640866331544516e-05,
-      "loss": 0.7533,
+      "loss": 0.8271,
       "step": 11892
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2639730266865728e-05,
-      "loss": 0.6544,
+      "loss": 0.7683,
       "step": 11893
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.263859416556443e-05,
-      "loss": 0.8381,
+      "loss": 0.8831,
       "step": 11894
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2637458027656377e-05,
-      "loss": 0.8116,
+      "loss": 0.9046,
       "step": 11895
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2636321853157338e-05,
-      "loss": 0.8024,
+      "loss": 0.8113,
       "step": 11896
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2635185642083073e-05,
-      "loss": 0.8565,
+      "loss": 0.9802,
       "step": 11897
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2634049394449346e-05,
-      "loss": 0.7799,
+      "loss": 0.8895,
       "step": 11898
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2632913110271922e-05,
-      "loss": 0.8166,
+      "loss": 0.8962,
       "step": 11899
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2631776789566564e-05,
-      "loss": 0.849,
+      "loss": 0.9108,
       "step": 11900
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2630640432349037e-05,
-      "loss": 0.6525,
+      "loss": 0.8245,
       "step": 11901
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2629504038635108e-05,
-      "loss": 0.7166,
+      "loss": 0.8473,
       "step": 11902
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2628367608440538e-05,
-      "loss": 0.7162,
+      "loss": 0.8746,
       "step": 11903
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2627231141781098e-05,
-      "loss": 0.6219,
+      "loss": 0.739,
       "step": 11904
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2626094638672552e-05,
-      "loss": 0.7695,
+      "loss": 0.9195,
       "step": 11905
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.262495809913067e-05,
-      "loss": 0.8337,
+      "loss": 0.9607,
       "step": 11906
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.262382152317122e-05,
-      "loss": 0.6241,
+      "loss": 0.7501,
       "step": 11907
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2622684910809966e-05,
-      "loss": 0.6808,
+      "loss": 0.7615,
       "step": 11908
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2621548262062684e-05,
-      "loss": 0.7947,
+      "loss": 0.8927,
       "step": 11909
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2620411576945139e-05,
-      "loss": 0.8191,
+      "loss": 0.8896,
       "step": 11910
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2619274855473099e-05,
-      "loss": 0.7987,
+      "loss": 0.9036,
       "step": 11911
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2618138097662337e-05,
-      "loss": 0.809,
+      "loss": 0.9023,
       "step": 11912
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2617001303528621e-05,
-      "loss": 0.858,
+      "loss": 0.9287,
       "step": 11913
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.261586447308773e-05,
-      "loss": 0.6872,
+      "loss": 0.8485,
       "step": 11914
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.261472760635543e-05,
-      "loss": 0.8674,
+      "loss": 0.951,
       "step": 11915
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.261359070334749e-05,
-      "loss": 0.7148,
+      "loss": 0.8155,
       "step": 11916
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2612453764079693e-05,
-      "loss": 0.6548,
+      "loss": 0.7433,
       "step": 11917
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2611316788567805e-05,
-      "loss": 0.7741,
+      "loss": 0.901,
       "step": 11918
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2610179776827598e-05,
-      "loss": 0.7753,
+      "loss": 0.8627,
       "step": 11919
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2609042728874852e-05,
-      "loss": 0.7068,
+      "loss": 0.8548,
       "step": 11920
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2607905644725342e-05,
-      "loss": 0.6827,
+      "loss": 0.837,
       "step": 11921
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2606768524394837e-05,
-      "loss": 0.7766,
+      "loss": 0.8978,
       "step": 11922
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2605631367899121e-05,
-      "loss": 0.7426,
+      "loss": 0.8572,
       "step": 11923
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2604494175253966e-05,
-      "loss": 0.7234,
+      "loss": 0.8275,
       "step": 11924
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2603356946475147e-05,
-      "loss": 0.7137,
+      "loss": 0.8485,
       "step": 11925
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2602219681578447e-05,
-      "loss": 0.8,
+      "loss": 0.9096,
       "step": 11926
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.260108238057964e-05,
-      "loss": 0.7025,
+      "loss": 0.8765,
       "step": 11927
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2599945043494506e-05,
-      "loss": 0.6547,
+      "loss": 0.7971,
       "step": 11928
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2598807670338821e-05,
-      "loss": 0.7615,
+      "loss": 0.9304,
       "step": 11929
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2597670261128371e-05,
-      "loss": 0.7515,
+      "loss": 0.8928,
       "step": 11930
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.259653281587893e-05,
-      "loss": 0.7437,
+      "loss": 0.8475,
       "step": 11931
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2595395334606277e-05,
-      "loss": 0.7255,
+      "loss": 0.8022,
       "step": 11932
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.25942578173262e-05,
-      "loss": 0.8032,
+      "loss": 0.8836,
       "step": 11933
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2593120264054477e-05,
-      "loss": 0.7052,
+      "loss": 0.876,
       "step": 11934
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2591982674806886e-05,
-      "loss": 0.7671,
+      "loss": 0.8717,
       "step": 11935
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2590845049599217e-05,
-      "loss": 0.7869,
+      "loss": 0.8588,
       "step": 11936
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2589707388447243e-05,
-      "loss": 0.7174,
+      "loss": 0.8496,
       "step": 11937
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2588569691366757e-05,
-      "loss": 0.8913,
+      "loss": 0.9351,
       "step": 11938
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.258743195837354e-05,
-      "loss": 0.8763,
+      "loss": 0.9152,
       "step": 11939
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2586294189483375e-05,
-      "loss": 0.6624,
+      "loss": 0.7535,
       "step": 11940
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2585156384712045e-05,
-      "loss": 0.7853,
+      "loss": 0.9481,
       "step": 11941
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2584018544075343e-05,
-      "loss": 0.7746,
+      "loss": 0.849,
       "step": 11942
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2582880667589046e-05,
-      "loss": 0.7333,
+      "loss": 0.8454,
       "step": 11943
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2581742755268947e-05,
-      "loss": 0.642,
+      "loss": 0.7845,
       "step": 11944
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2580604807130828e-05,
-      "loss": 0.8028,
+      "loss": 0.8854,
       "step": 11945
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2579466823190478e-05,
-      "loss": 0.8108,
+      "loss": 0.8928,
       "step": 11946
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2578328803463688e-05,
-      "loss": 0.7409,
+      "loss": 0.8069,
       "step": 11947
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2577190747966244e-05,
-      "loss": 0.7527,
+      "loss": 0.8524,
       "step": 11948
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2576052656713934e-05,
-      "loss": 0.7959,
+      "loss": 0.879,
       "step": 11949
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.257491452972255e-05,
-      "loss": 0.735,
+      "loss": 0.8908,
       "step": 11950
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2573776367007878e-05,
-      "loss": 0.645,
+      "loss": 0.8148,
       "step": 11951
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.257263816858571e-05,
-      "loss": 0.6577,
+      "loss": 0.8391,
       "step": 11952
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2571499934471839e-05,
-      "loss": 0.7676,
+      "loss": 0.8806,
       "step": 11953
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2570361664682055e-05,
-      "loss": 0.7385,
+      "loss": 0.8274,
       "step": 11954
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.256922335923215e-05,
-      "loss": 0.8373,
+      "loss": 0.8745,
       "step": 11955
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2568085018137914e-05,
-      "loss": 0.7185,
+      "loss": 0.8091,
       "step": 11956
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2566946641415146e-05,
-      "loss": 0.7743,
+      "loss": 0.9124,
       "step": 11957
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2565808229079633e-05,
-      "loss": 0.7036,
+      "loss": 0.7968,
       "step": 11958
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2564669781147169e-05,
-      "loss": 0.7288,
+      "loss": 0.8597,
       "step": 11959
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2563531297633553e-05,
-      "loss": 0.7307,
+      "loss": 0.8372,
       "step": 11960
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.2562392778554577e-05,
-      "loss": 0.8828,
+      "loss": 1.0147,
       "step": 11961
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2561254223926037e-05,
-      "loss": 0.641,
+      "loss": 0.7667,
       "step": 11962
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.256011563376373e-05,
-      "loss": 0.7987,
+      "loss": 0.9015,
       "step": 11963
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2558977008083449e-05,
-      "loss": 0.6651,
+      "loss": 0.7989,
       "step": 11964
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2557838346900993e-05,
-      "loss": 0.6957,
+      "loss": 0.8814,
       "step": 11965
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.255669965023216e-05,
-      "loss": 0.6976,
+      "loss": 0.8418,
       "step": 11966
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2555560918092745e-05,
-      "loss": 0.8624,
+      "loss": 0.9336,
       "step": 11967
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.255442215049855e-05,
-      "loss": 0.7003,
+      "loss": 0.8273,
       "step": 11968
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.255328334746537e-05,
-      "loss": 0.8455,
+      "loss": 0.8901,
       "step": 11969
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2552144509009008e-05,
-      "loss": 0.7676,
+      "loss": 0.8941,
       "step": 11970
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2551005635145264e-05,
-      "loss": 0.8533,
+      "loss": 0.9132,
       "step": 11971
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2549866725889933e-05,
-      "loss": 0.7235,
+      "loss": 0.8812,
       "step": 11972
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2548727781258817e-05,
-      "loss": 0.6442,
+      "loss": 0.7812,
       "step": 11973
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2547588801267722e-05,
-      "loss": 0.7184,
+      "loss": 0.8934,
       "step": 11974
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2546449785932446e-05,
-      "loss": 0.8067,
+      "loss": 0.923,
       "step": 11975
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2545310735268794e-05,
-      "loss": 0.6259,
+      "loss": 0.7797,
       "step": 11976
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2544171649292567e-05,
-      "loss": 0.7311,
+      "loss": 0.8354,
       "step": 11977
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2543032528019566e-05,
-      "loss": 0.9134,
+      "loss": 1.0479,
       "step": 11978
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2541893371465596e-05,
-      "loss": 0.6579,
+      "loss": 0.8182,
       "step": 11979
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2540754179646463e-05,
-      "loss": 0.7022,
+      "loss": 0.8487,
       "step": 11980
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2539614952577971e-05,
-      "loss": 0.8085,
+      "loss": 0.9038,
       "step": 11981
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2538475690275926e-05,
-      "loss": 0.7017,
+      "loss": 0.825,
       "step": 11982
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.253733639275613e-05,
-      "loss": 0.6149,
+      "loss": 0.7783,
       "step": 11983
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2536197060034394e-05,
-      "loss": 0.7404,
+      "loss": 0.8689,
       "step": 11984
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.253505769212652e-05,
-      "loss": 0.7477,
+      "loss": 0.8714,
       "step": 11985
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2533918289048317e-05,
-      "loss": 0.7477,
+      "loss": 0.8194,
       "step": 11986
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.25327788508156e-05,
-      "loss": 0.7993,
+      "loss": 0.8833,
       "step": 11987
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.253163937744416e-05,
-      "loss": 0.7456,
+      "loss": 0.8959,
       "step": 11988
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2530499868949822e-05,
-      "loss": 0.7398,
+      "loss": 0.8131,
       "step": 11989
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2529360325348385e-05,
-      "loss": 0.8225,
+      "loss": 0.9054,
       "step": 11990
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2528220746655663e-05,
-      "loss": 0.7431,
+      "loss": 0.8904,
       "step": 11991
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2527081132887466e-05,
-      "loss": 0.7668,
+      "loss": 0.8623,
       "step": 11992
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2525941484059606e-05,
-      "loss": 0.8069,
+      "loss": 0.8836,
       "step": 11993
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2524801800187887e-05,
-      "loss": 0.8225,
+      "loss": 0.8932,
       "step": 11994
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2523662081288128e-05,
-      "loss": 0.7817,
+      "loss": 0.8662,
       "step": 11995
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2522522327376136e-05,
-      "loss": 0.7798,
+      "loss": 0.8296,
       "step": 11996
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2521382538467727e-05,
-      "loss": 0.6769,
+      "loss": 0.8083,
       "step": 11997
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2520242714578715e-05,
-      "loss": 0.8482,
+      "loss": 0.8874,
       "step": 11998
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2519102855724907e-05,
-      "loss": 0.7757,
+      "loss": 0.9338,
       "step": 11999
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2517962961922123e-05,
-      "loss": 0.8373,
+      "loss": 0.9548,
       "step": 12000
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2516823033186178e-05,
-      "loss": 0.6937,
+      "loss": 0.7697,
       "step": 12001
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.251568306953288e-05,
-      "loss": 0.6606,
+      "loss": 0.778,
       "step": 12002
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2514543070978052e-05,
-      "loss": 0.6819,
+      "loss": 0.8212,
       "step": 12003
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2513403037537506e-05,
-      "loss": 0.7127,
+      "loss": 0.825,
       "step": 12004
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2512262969227055e-05,
-      "loss": 0.7332,
+      "loss": 0.8521,
       "step": 12005
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2511122866062526e-05,
-      "loss": 0.7202,
+      "loss": 0.8307,
       "step": 12006
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2509982728059726e-05,
-      "loss": 0.7705,
+      "loss": 0.8205,
       "step": 12007
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2508842555234479e-05,
-      "loss": 0.6602,
+      "loss": 0.8641,
       "step": 12008
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2507702347602597e-05,
-      "loss": 0.7983,
+      "loss": 0.8577,
       "step": 12009
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2506562105179907e-05,
-      "loss": 0.8862,
+      "loss": 0.9307,
       "step": 12010
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2505421827982222e-05,
-      "loss": 0.6621,
+      "loss": 0.7777,
       "step": 12011
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2504281516025365e-05,
-      "loss": 0.7815,
+      "loss": 0.878,
       "step": 12012
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2503141169325157e-05,
-      "loss": 0.9143,
+      "loss": 0.9479,
       "step": 12013
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2502000787897415e-05,
-      "loss": 0.7006,
+      "loss": 0.8295,
       "step": 12014
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.250086037175796e-05,
-      "loss": 0.7832,
+      "loss": 0.8647,
       "step": 12015
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2499719920922616e-05,
-      "loss": 0.6587,
+      "loss": 0.8105,
       "step": 12016
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2498579435407207e-05,
-      "loss": 0.6508,
+      "loss": 0.803,
       "step": 12017
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2497438915227552e-05,
-      "loss": 0.8202,
+      "loss": 0.8673,
       "step": 12018
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2496298360399475e-05,
-      "loss": 0.5802,
+      "loss": 0.7869,
       "step": 12019
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.24951577709388e-05,
-      "loss": 0.7142,
+      "loss": 0.8091,
       "step": 12020
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2494017146861353e-05,
-      "loss": 0.7595,
+      "loss": 0.8297,
       "step": 12021
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2492876488182955e-05,
-      "loss": 0.7694,
+      "loss": 0.8887,
       "step": 12022
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2491735794919433e-05,
-      "loss": 0.8571,
+      "loss": 0.928,
       "step": 12023
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2490595067086612e-05,
-      "loss": 0.7118,
+      "loss": 0.8536,
       "step": 12024
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.248945430470032e-05,
-      "loss": 0.715,
+      "loss": 0.8033,
       "step": 12025
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.248831350777638e-05,
-      "loss": 0.8781,
+      "loss": 0.9781,
       "step": 12026
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.248717267633062e-05,
-      "loss": 0.6998,
+      "loss": 0.8648,
       "step": 12027
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2486031810378867e-05,
-      "loss": 0.7376,
+      "loss": 0.8785,
       "step": 12028
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2484890909936954e-05,
-      "loss": 0.7581,
+      "loss": 0.8725,
       "step": 12029
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2483749975020702e-05,
-      "loss": 0.6323,
+      "loss": 0.7575,
       "step": 12030
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2482609005645943e-05,
-      "loss": 0.737,
+      "loss": 0.7629,
       "step": 12031
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2481468001828507e-05,
-      "loss": 0.8144,
+      "loss": 0.9532,
       "step": 12032
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2480326963584224e-05,
-      "loss": 0.7753,
+      "loss": 0.852,
       "step": 12033
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2479185890928922e-05,
-      "loss": 0.6944,
+      "loss": 0.7994,
       "step": 12034
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2478044783878433e-05,
-      "loss": 0.8375,
+      "loss": 0.8799,
       "step": 12035
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.247690364244859e-05,
-      "loss": 0.7996,
+      "loss": 0.9072,
       "step": 12036
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2475762466655224e-05,
-      "loss": 0.7452,
+      "loss": 0.8491,
       "step": 12037
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2474621256514165e-05,
-      "loss": 0.6963,
+      "loss": 0.8094,
       "step": 12038
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2473480012041246e-05,
-      "loss": 0.8603,
+      "loss": 0.9374,
       "step": 12039
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2472338733252302e-05,
-      "loss": 0.6994,
+      "loss": 0.8658,
       "step": 12040
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2471197420163169e-05,
-      "loss": 0.8052,
+      "loss": 0.9229,
       "step": 12041
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2470056072789674e-05,
-      "loss": 0.659,
+      "loss": 0.8052,
       "step": 12042
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2468914691147655e-05,
-      "loss": 0.7787,
+      "loss": 0.9077,
       "step": 12043
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2467773275252952e-05,
-      "loss": 0.7326,
+      "loss": 0.8476,
       "step": 12044
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2466631825121394e-05,
-      "loss": 0.7163,
+      "loss": 0.8157,
       "step": 12045
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.246549034076882e-05,
-      "loss": 0.8146,
+      "loss": 0.8924,
       "step": 12046
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2464348822211063e-05,
-      "loss": 0.6438,
+      "loss": 0.8281,
       "step": 12047
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2463207269463964e-05,
-      "loss": 0.7788,
+      "loss": 0.9271,
       "step": 12048
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2462065682543358e-05,
-      "loss": 0.6573,
+      "loss": 0.839,
       "step": 12049
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2460924061465085e-05,
-      "loss": 0.7017,
+      "loss": 0.9032,
       "step": 12050
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2459782406244982e-05,
-      "loss": 0.8511,
+      "loss": 0.8795,
       "step": 12051
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2458640716898887e-05,
-      "loss": 0.7013,
+      "loss": 0.8647,
       "step": 12052
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.245749899344264e-05,
-      "loss": 0.7051,
+      "loss": 0.8669,
       "step": 12053
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2456357235892085e-05,
-      "loss": 0.5882,
+      "loss": 0.7522,
       "step": 12054
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2455215444263057e-05,
-      "loss": 0.7436,
+      "loss": 0.8158,
       "step": 12055
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2454073618571396e-05,
-      "loss": 0.7218,
+      "loss": 0.8625,
       "step": 12056
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2452931758832947e-05,
-      "loss": 0.9263,
+      "loss": 0.9942,
       "step": 12057
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2451789865063548e-05,
-      "loss": 0.5964,
+      "loss": 0.6951,
       "step": 12058
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2450647937279045e-05,
-      "loss": 0.7661,
+      "loss": 0.8068,
       "step": 12059
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2449505975495282e-05,
-      "loss": 0.6246,
+      "loss": 0.8536,
       "step": 12060
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2448363979728093e-05,
-      "loss": 0.8283,
+      "loss": 0.8675,
       "step": 12061
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2447221949993332e-05,
-      "loss": 0.7399,
+      "loss": 0.7993,
       "step": 12062
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2446079886306836e-05,
-      "loss": 0.9083,
+      "loss": 0.9954,
       "step": 12063
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2444937788684455e-05,
-      "loss": 0.7582,
+      "loss": 0.8693,
       "step": 12064
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.244379565714203e-05,
-      "loss": 0.8634,
+      "loss": 0.8879,
       "step": 12065
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2442653491695404e-05,
-      "loss": 0.7831,
+      "loss": 0.8115,
       "step": 12066
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2441511292360431e-05,
-      "loss": 0.6633,
+      "loss": 0.8828,
       "step": 12067
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.244036905915295e-05,
-      "loss": 0.8647,
+      "loss": 0.9498,
       "step": 12068
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2439226792088812e-05,
-      "loss": 0.7311,
+      "loss": 0.822,
       "step": 12069
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2438084491183866e-05,
-      "loss": 0.6665,
+      "loss": 0.846,
       "step": 12070
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.243694215645395e-05,
-      "loss": 0.7529,
+      "loss": 0.8377,
       "step": 12071
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2435799787914924e-05,
-      "loss": 0.7565,
+      "loss": 0.913,
       "step": 12072
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2434657385582634e-05,
-      "loss": 0.7637,
+      "loss": 0.8813,
       "step": 12073
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2433514949472923e-05,
-      "loss": 0.6014,
+      "loss": 0.7305,
       "step": 12074
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2432372479601644e-05,
-      "loss": 0.8108,
+      "loss": 0.9716,
       "step": 12075
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2431229975984652e-05,
-      "loss": 0.7244,
+      "loss": 0.8164,
       "step": 12076
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2430087438637788e-05,
-      "loss": 0.8061,
+      "loss": 0.8541,
       "step": 12077
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2428944867576912e-05,
-      "loss": 0.9983,
+      "loss": 0.9473,
       "step": 12078
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2427802262817871e-05,
-      "loss": 0.8044,
+      "loss": 0.8995,
       "step": 12079
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.242665962437652e-05,
-      "loss": 0.87,
+      "loss": 0.9172,
       "step": 12080
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.242551695226871e-05,
-      "loss": 0.7393,
+      "loss": 0.8958,
       "step": 12081
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.242437424651029e-05,
-      "loss": 0.6775,
+      "loss": 0.7333,
       "step": 12082
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2423231507117124e-05,
-      "loss": 0.6776,
+      "loss": 0.844,
       "step": 12083
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2422088734105054e-05,
-      "loss": 0.6407,
+      "loss": 0.7628,
       "step": 12084
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.242094592748994e-05,
-      "loss": 0.7752,
+      "loss": 0.8408,
       "step": 12085
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2419803087287639e-05,
-      "loss": 0.6922,
+      "loss": 0.826,
       "step": 12086
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2418660213513998e-05,
-      "loss": 0.7286,
+      "loss": 0.9172,
       "step": 12087
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2417517306184881e-05,
-      "loss": 0.8675,
+      "loss": 0.916,
       "step": 12088
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2416374365316141e-05,
-      "loss": 0.8479,
+      "loss": 0.9211,
       "step": 12089
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2415231390923634e-05,
-      "loss": 0.9165,
+      "loss": 0.9929,
       "step": 12090
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2414088383023219e-05,
-      "loss": 0.806,
+      "loss": 0.8773,
       "step": 12091
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2412945341630753e-05,
-      "loss": 0.8021,
+      "loss": 0.8759,
       "step": 12092
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2411802266762094e-05,
-      "loss": 0.8945,
+      "loss": 0.9356,
       "step": 12093
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2410659158433102e-05,
-      "loss": 0.8329,
+      "loss": 0.8687,
       "step": 12094
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2409516016659631e-05,
-      "loss": 0.7313,
+      "loss": 0.8401,
       "step": 12095
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2408372841457547e-05,
-      "loss": 0.7947,
+      "loss": 0.9069,
       "step": 12096
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2407229632842706e-05,
-      "loss": 0.785,
+      "loss": 0.9238,
       "step": 12097
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.240608639083097e-05,
-      "loss": 0.7938,
+      "loss": 0.8489,
       "step": 12098
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2404943115438197e-05,
-      "loss": 0.6685,
+      "loss": 0.8952,
       "step": 12099
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2403799806680252e-05,
-      "loss": 0.6766,
+      "loss": 0.8325,
       "step": 12100
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2402656464572993e-05,
-      "loss": 0.7496,
+      "loss": 0.8297,
       "step": 12101
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2401513089132288e-05,
-      "loss": 0.7267,
+      "loss": 0.8509,
       "step": 12102
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2400369680373996e-05,
-      "loss": 0.8235,
+      "loss": 0.8967,
       "step": 12103
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.239922623831398e-05,
-      "loss": 0.694,
+      "loss": 0.8082,
       "step": 12104
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2398082762968102e-05,
-      "loss": 0.8066,
+      "loss": 0.9322,
       "step": 12105
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2396939254352227e-05,
-      "loss": 0.6659,
+      "loss": 0.7508,
       "step": 12106
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2395795712482223e-05,
-      "loss": 0.8499,
+      "loss": 0.9258,
       "step": 12107
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2394652137373952e-05,
-      "loss": 0.827,
+      "loss": 0.8802,
       "step": 12108
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2393508529043282e-05,
-      "loss": 0.6712,
+      "loss": 0.7802,
       "step": 12109
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2392364887506074e-05,
-      "loss": 0.6802,
+      "loss": 0.7788,
       "step": 12110
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.23912212127782e-05,
-      "loss": 0.5414,
+      "loss": 0.7612,
       "step": 12111
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2390077504875522e-05,
-      "loss": 0.727,
+      "loss": 0.8875,
       "step": 12112
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2388933763813913e-05,
-      "loss": 0.724,
+      "loss": 0.832,
       "step": 12113
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2387789989609233e-05,
-      "loss": 0.8159,
+      "loss": 0.8393,
       "step": 12114
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.238664618227736e-05,
-      "loss": 0.8478,
+      "loss": 0.8845,
       "step": 12115
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2385502341834155e-05,
-      "loss": 0.8558,
+      "loss": 0.9144,
       "step": 12116
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2384358468295488e-05,
-      "loss": 0.8019,
+      "loss": 0.9227,
       "step": 12117
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2383214561677233e-05,
-      "loss": 0.715,
+      "loss": 0.799,
       "step": 12118
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2382070621995252e-05,
-      "loss": 0.7965,
+      "loss": 0.8431,
       "step": 12119
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2380926649265426e-05,
-      "loss": 0.7206,
+      "loss": 0.8282,
       "step": 12120
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2379782643503622e-05,
-      "loss": 0.6781,
+      "loss": 0.8086,
       "step": 12121
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2378638604725708e-05,
-      "loss": 0.841,
+      "loss": 0.9258,
       "step": 12122
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2377494532947559e-05,
-      "loss": 0.8412,
+      "loss": 0.9649,
       "step": 12123
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2376350428185042e-05,
-      "loss": 0.7067,
+      "loss": 0.8105,
       "step": 12124
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.237520629045404e-05,
-      "loss": 0.8106,
+      "loss": 0.9207,
       "step": 12125
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2374062119770418e-05,
-      "loss": 0.8061,
+      "loss": 0.915,
       "step": 12126
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2372917916150053e-05,
-      "loss": 0.6701,
+      "loss": 0.8765,
       "step": 12127
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2371773679608818e-05,
-      "loss": 0.6893,
+      "loss": 0.8064,
       "step": 12128
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2370629410162592e-05,
-      "loss": 0.798,
+      "loss": 0.8998,
       "step": 12129
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2369485107827241e-05,
-      "loss": 0.772,
+      "loss": 0.8408,
       "step": 12130
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.236834077261865e-05,
-      "loss": 0.9156,
+      "loss": 0.9797,
       "step": 12131
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.236719640455269e-05,
-      "loss": 0.6634,
+      "loss": 0.7762,
       "step": 12132
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2366052003645237e-05,
-      "loss": 0.7329,
+      "loss": 0.8299,
       "step": 12133
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.236490756991217e-05,
-      "loss": 0.7849,
+      "loss": 0.8987,
       "step": 12134
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2363763103369365e-05,
-      "loss": 0.5864,
+      "loss": 0.7382,
       "step": 12135
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2362618604032703e-05,
-      "loss": 0.7736,
+      "loss": 0.8772,
       "step": 12136
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.236147407191806e-05,
-      "loss": 0.8131,
+      "loss": 0.8709,
       "step": 12137
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.236032950704131e-05,
-      "loss": 0.8466,
+      "loss": 0.8854,
       "step": 12138
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2359184909418343e-05,
-      "loss": 0.7272,
+      "loss": 0.8839,
       "step": 12139
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.235804027906503e-05,
-      "loss": 0.9083,
+      "loss": 1.0087,
       "step": 12140
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2356895615997255e-05,
-      "loss": 0.7705,
+      "loss": 0.8503,
       "step": 12141
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2355750920230898e-05,
-      "loss": 0.7585,
+      "loss": 0.8758,
       "step": 12142
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2354606191781837e-05,
-      "loss": 0.8265,
+      "loss": 0.973,
       "step": 12143
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2353461430665956e-05,
-      "loss": 0.7636,
+      "loss": 0.8677,
       "step": 12144
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2352316636899139e-05,
-      "loss": 0.6786,
+      "loss": 0.8268,
       "step": 12145
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2351171810497262e-05,
-      "loss": 0.7843,
+      "loss": 0.8653,
       "step": 12146
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2350026951476214e-05,
-      "loss": 0.7562,
+      "loss": 0.9035,
       "step": 12147
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2348882059851879e-05,
-      "loss": 0.727,
+      "loss": 0.8368,
       "step": 12148
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2347737135640134e-05,
-      "loss": 0.7557,
+      "loss": 0.8424,
       "step": 12149
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2346592178856871e-05,
-      "loss": 0.6498,
+      "loss": 0.811,
       "step": 12150
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2345447189517968e-05,
-      "loss": 0.8349,
+      "loss": 0.9051,
       "step": 12151
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2344302167639315e-05,
-      "loss": 0.6258,
+      "loss": 0.7801,
       "step": 12152
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2343157113236795e-05,
-      "loss": 0.7643,
+      "loss": 0.85,
       "step": 12153
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2342012026326291e-05,
-      "loss": 0.7168,
+      "loss": 0.8559,
       "step": 12154
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2340866906923697e-05,
-      "loss": 0.6372,
+      "loss": 0.8766,
       "step": 12155
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2339721755044895e-05,
-      "loss": 0.674,
+      "loss": 0.8014,
       "step": 12156
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2338576570705774e-05,
-      "loss": 0.7744,
+      "loss": 0.8557,
       "step": 12157
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.233743135392222e-05,
-      "loss": 0.6988,
+      "loss": 0.8321,
       "step": 12158
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2336286104710124e-05,
-      "loss": 0.7825,
+      "loss": 0.8983,
       "step": 12159
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2335140823085372e-05,
-      "loss": 0.865,
+      "loss": 0.9329,
       "step": 12160
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2333995509063854e-05,
-      "loss": 0.8525,
+      "loss": 0.9601,
       "step": 12161
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2332850162661458e-05,
-      "loss": 0.835,
+      "loss": 0.8923,
       "step": 12162
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.233170478389408e-05,
-      "loss": 0.8077,
+      "loss": 0.8975,
       "step": 12163
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2330559372777602e-05,
-      "loss": 0.8742,
+      "loss": 0.8735,
       "step": 12164
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2329413929327922e-05,
-      "loss": 0.7608,
+      "loss": 0.8292,
       "step": 12165
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.232826845356093e-05,
-      "loss": 0.6659,
+      "loss": 0.782,
       "step": 12166
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2327122945492514e-05,
-      "loss": 0.7768,
+      "loss": 0.8909,
       "step": 12167
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.232597740513857e-05,
-      "loss": 0.7923,
+      "loss": 0.8271,
       "step": 12168
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.232483183251499e-05,
-      "loss": 0.8445,
+      "loss": 0.9408,
       "step": 12169
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2323686227637664e-05,
-      "loss": 0.6198,
+      "loss": 0.7877,
       "step": 12170
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.232254059052249e-05,
-      "loss": 0.8219,
+      "loss": 0.9209,
       "step": 12171
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2321394921185363e-05,
-      "loss": 0.8276,
+      "loss": 0.934,
       "step": 12172
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.232024921964217e-05,
-      "loss": 0.792,
+      "loss": 0.9418,
       "step": 12173
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2319103485908816e-05,
-      "loss": 0.8227,
+      "loss": 0.8668,
       "step": 12174
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.231795772000119e-05,
-      "loss": 0.8482,
+      "loss": 0.9364,
       "step": 12175
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.231681192193519e-05,
-      "loss": 0.8222,
+      "loss": 0.9134,
       "step": 12176
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.231566609172671e-05,
-      "loss": 0.6217,
+      "loss": 0.8408,
       "step": 12177
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.231452022939165e-05,
-      "loss": 0.8629,
+      "loss": 0.9525,
       "step": 12178
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2313374334945906e-05,
-      "loss": 0.8291,
+      "loss": 0.888,
       "step": 12179
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2312228408405374e-05,
-      "loss": 0.8078,
+      "loss": 0.8593,
       "step": 12180
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2311082449785954e-05,
-      "loss": 0.731,
+      "loss": 0.9179,
       "step": 12181
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2309936459103544e-05,
-      "loss": 0.8554,
+      "loss": 0.9355,
       "step": 12182
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2308790436374043e-05,
-      "loss": 0.8646,
+      "loss": 0.8971,
       "step": 12183
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2307644381613351e-05,
-      "loss": 0.7538,
+      "loss": 0.8242,
       "step": 12184
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2306498294837366e-05,
-      "loss": 0.7805,
+      "loss": 0.8268,
       "step": 12185
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.230535217606199e-05,
-      "loss": 0.8028,
+      "loss": 0.8947,
       "step": 12186
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2304206025303125e-05,
-      "loss": 0.7778,
+      "loss": 0.9016,
       "step": 12187
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.230305984257667e-05,
-      "loss": 0.7695,
+      "loss": 0.8557,
       "step": 12188
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2301913627898524e-05,
-      "loss": 0.8748,
+      "loss": 0.9852,
       "step": 12189
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2300767381284597e-05,
-      "loss": 0.8003,
+      "loss": 0.8817,
       "step": 12190
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2299621102750783e-05,
-      "loss": 0.7608,
+      "loss": 0.8124,
       "step": 12191
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.229847479231299e-05,
-      "loss": 0.8512,
+      "loss": 0.9012,
       "step": 12192
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2297328449987122e-05,
-      "loss": 0.6622,
+      "loss": 0.7923,
       "step": 12193
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2296182075789081e-05,
-      "loss": 0.7997,
+      "loss": 0.8762,
       "step": 12194
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2295035669734771e-05,
-      "loss": 0.675,
+      "loss": 0.8466,
       "step": 12195
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.22938892318401e-05,
-      "loss": 0.7004,
+      "loss": 0.773,
       "step": 12196
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2292742762120966e-05,
-      "loss": 0.7909,
+      "loss": 0.8582,
       "step": 12197
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2291596260593284e-05,
-      "loss": 0.7253,
+      "loss": 0.8431,
       "step": 12198
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.229044972727295e-05,
-      "loss": 0.7941,
+      "loss": 0.8995,
       "step": 12199
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2289303162175878e-05,
-      "loss": 0.8286,
+      "loss": 0.8813,
       "step": 12200
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2288156565317975e-05,
-      "loss": 0.8464,
+      "loss": 0.9581,
       "step": 12201
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2287009936715141e-05,
-      "loss": 0.7809,
+      "loss": 0.8947,
       "step": 12202
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2285863276383294e-05,
-      "loss": 0.6495,
+      "loss": 0.8151,
       "step": 12203
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2284716584338335e-05,
-      "loss": 0.5819,
+      "loss": 0.6942,
       "step": 12204
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2283569860596175e-05,
-      "loss": 0.728,
+      "loss": 0.8698,
       "step": 12205
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2282423105172722e-05,
-      "loss": 0.6806,
+      "loss": 0.8157,
       "step": 12206
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2281276318083887e-05,
-      "loss": 0.7254,
+      "loss": 0.8348,
       "step": 12207
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2280129499345581e-05,
-      "loss": 0.7002,
+      "loss": 0.7804,
       "step": 12208
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2278982648973715e-05,
-      "loss": 0.7079,
+      "loss": 0.8286,
       "step": 12209
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2277835766984194e-05,
-      "loss": 0.8702,
+      "loss": 0.8904,
       "step": 12210
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2276688853392938e-05,
-      "loss": 0.7847,
+      "loss": 0.8691,
       "step": 12211
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2275541908215853e-05,
-      "loss": 0.6977,
+      "loss": 0.8566,
       "step": 12212
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.227439493146885e-05,
-      "loss": 0.7128,
+      "loss": 0.8175,
       "step": 12213
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2273247923167846e-05,
-      "loss": 0.7973,
+      "loss": 0.9188,
       "step": 12214
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2272100883328752e-05,
-      "loss": 0.8008,
+      "loss": 0.94,
       "step": 12215
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2270953811967484e-05,
-      "loss": 0.7789,
+      "loss": 0.8575,
       "step": 12216
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2269806709099954e-05,
-      "loss": 0.7453,
+      "loss": 0.9173,
       "step": 12217
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2268659574742073e-05,
-      "loss": 0.7145,
+      "loss": 0.8231,
       "step": 12218
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2267512408909762e-05,
-      "loss": 0.7489,
+      "loss": 0.897,
       "step": 12219
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2266365211618933e-05,
-      "loss": 0.6233,
+      "loss": 0.7872,
       "step": 12220
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.22652179828855e-05,
-      "loss": 0.7687,
+      "loss": 0.9193,
       "step": 12221
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2264070722725383e-05,
-      "loss": 0.8058,
+      "loss": 0.8209,
       "step": 12222
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2262923431154497e-05,
-      "loss": 0.718,
+      "loss": 0.8585,
       "step": 12223
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2261776108188758e-05,
-      "loss": 0.7724,
+      "loss": 0.9026,
       "step": 12224
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2260628753844089e-05,
-      "loss": 0.7077,
+      "loss": 0.7627,
       "step": 12225
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2259481368136397e-05,
-      "loss": 0.719,
+      "loss": 0.8831,
       "step": 12226
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2258333951081612e-05,
-      "loss": 0.7309,
+      "loss": 0.8611,
       "step": 12227
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2257186502695647e-05,
-      "loss": 0.7468,
+      "loss": 0.879,
       "step": 12228
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2256039022994418e-05,
-      "loss": 0.6604,
+      "loss": 0.8159,
       "step": 12229
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.225489151199385e-05,
-      "loss": 0.5856,
+      "loss": 0.7555,
       "step": 12230
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2253743969709864e-05,
-      "loss": 0.7875,
+      "loss": 0.9405,
       "step": 12231
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2252596396158374e-05,
-      "loss": 0.6847,
+      "loss": 0.787,
       "step": 12232
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2251448791355309e-05,
-      "loss": 0.7983,
+      "loss": 0.9717,
       "step": 12233
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2250301155316582e-05,
-      "loss": 0.6119,
+      "loss": 0.7263,
       "step": 12234
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2249153488058124e-05,
-      "loss": 0.6465,
+      "loss": 0.8094,
       "step": 12235
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.2248005789595852e-05,
-      "loss": 0.7006,
+      "loss": 0.772,
       "step": 12236
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2246858059945687e-05,
-      "loss": 0.7299,
+      "loss": 0.7907,
       "step": 12237
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2245710299123557e-05,
-      "loss": 0.7668,
+      "loss": 0.8206,
       "step": 12238
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2244562507145379e-05,
-      "loss": 0.5933,
+      "loss": 0.7846,
       "step": 12239
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2243414684027081e-05,
-      "loss": 0.8847,
+      "loss": 0.9818,
       "step": 12240
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.224226682978459e-05,
-      "loss": 0.6728,
+      "loss": 0.8234,
       "step": 12241
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2241118944433828e-05,
-      "loss": 0.7931,
+      "loss": 0.8615,
       "step": 12242
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2239971027990717e-05,
-      "loss": 0.7962,
+      "loss": 0.8586,
       "step": 12243
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2238823080471191e-05,
-      "loss": 0.8654,
+      "loss": 1.022,
       "step": 12244
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2237675101891167e-05,
-      "loss": 0.7174,
+      "loss": 0.8858,
       "step": 12245
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2236527092266578e-05,
-      "loss": 0.6049,
+      "loss": 0.7622,
       "step": 12246
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2235379051613347e-05,
-      "loss": 0.7961,
+      "loss": 0.8152,
       "step": 12247
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2234230979947405e-05,
-      "loss": 0.8625,
+      "loss": 0.966,
       "step": 12248
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2233082877284679e-05,
-      "loss": 0.7032,
+      "loss": 0.8998,
       "step": 12249
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2231934743641094e-05,
-      "loss": 0.7442,
+      "loss": 0.8902,
       "step": 12250
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2230786579032583e-05,
-      "loss": 0.8629,
+      "loss": 0.8616,
       "step": 12251
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2229638383475073e-05,
-      "loss": 0.6659,
+      "loss": 0.7601,
       "step": 12252
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2228490156984491e-05,
-      "loss": 0.7359,
+      "loss": 0.7767,
       "step": 12253
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2227341899576774e-05,
-      "loss": 0.767,
+      "loss": 0.8683,
       "step": 12254
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2226193611267848e-05,
-      "loss": 0.8592,
+      "loss": 0.9213,
       "step": 12255
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2225045292073642e-05,
-      "loss": 0.7442,
+      "loss": 0.8729,
       "step": 12256
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2223896942010092e-05,
-      "loss": 0.9196,
+      "loss": 0.9312,
       "step": 12257
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2222748561093122e-05,
-      "loss": 0.7193,
+      "loss": 0.8039,
       "step": 12258
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2221600149338675e-05,
-      "loss": 0.7517,
+      "loss": 0.9047,
       "step": 12259
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2220451706762677e-05,
-      "loss": 0.7739,
+      "loss": 0.8127,
       "step": 12260
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2219303233381058e-05,
-      "loss": 0.8433,
+      "loss": 0.88,
       "step": 12261
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2218154729209762e-05,
-      "loss": 0.8801,
+      "loss": 0.9286,
       "step": 12262
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2217006194264709e-05,
-      "loss": 0.7024,
+      "loss": 0.8103,
       "step": 12263
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2215857628561844e-05,
-      "loss": 0.8104,
+      "loss": 0.8611,
       "step": 12264
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.22147090321171e-05,
-      "loss": 0.7097,
+      "loss": 0.8229,
       "step": 12265
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2213560404946408e-05,
-      "loss": 0.7792,
+      "loss": 0.8693,
       "step": 12266
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2212411747065707e-05,
-      "loss": 0.8086,
+      "loss": 0.9576,
       "step": 12267
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2211263058490932e-05,
-      "loss": 0.8761,
+      "loss": 0.9048,
       "step": 12268
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.221011433923802e-05,
-      "loss": 0.7816,
+      "loss": 0.8946,
       "step": 12269
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2208965589322906e-05,
-      "loss": 0.6589,
+      "loss": 0.8732,
       "step": 12270
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.220781680876153e-05,
-      "loss": 0.8508,
+      "loss": 0.9243,
       "step": 12271
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2206667997569826e-05,
-      "loss": 0.7637,
+      "loss": 0.8758,
       "step": 12272
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2205519155763737e-05,
-      "loss": 0.7498,
+      "loss": 0.8947,
       "step": 12273
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2204370283359196e-05,
-      "loss": 0.7971,
+      "loss": 0.8716,
       "step": 12274
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.220322138037215e-05,
-      "loss": 0.7648,
+      "loss": 0.9221,
       "step": 12275
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.220207244681853e-05,
-      "loss": 0.7144,
+      "loss": 0.8284,
       "step": 12276
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2200923482714278e-05,
-      "loss": 0.7014,
+      "loss": 0.7993,
       "step": 12277
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2199774488075336e-05,
-      "loss": 0.7907,
+      "loss": 0.8741,
       "step": 12278
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2198625462917643e-05,
-      "loss": 0.7943,
+      "loss": 0.878,
       "step": 12279
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2197476407257145e-05,
-      "loss": 0.7734,
+      "loss": 0.8109,
       "step": 12280
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2196327321109778e-05,
-      "loss": 0.8232,
+      "loss": 0.9062,
       "step": 12281
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2195178204491483e-05,
-      "loss": 0.7277,
+      "loss": 0.8217,
       "step": 12282
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2194029057418208e-05,
-      "loss": 0.8091,
+      "loss": 0.8654,
       "step": 12283
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2192879879905892e-05,
-      "loss": 0.9435,
+      "loss": 0.9571,
       "step": 12284
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2191730671970476e-05,
-      "loss": 0.8631,
+      "loss": 0.9602,
       "step": 12285
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.219058143362791e-05,
-      "loss": 0.6769,
+      "loss": 0.8129,
       "step": 12286
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2189432164894135e-05,
-      "loss": 0.8183,
+      "loss": 0.8817,
       "step": 12287
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2188282865785093e-05,
-      "loss": 0.7218,
+      "loss": 0.8584,
       "step": 12288
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2187133536316733e-05,
-      "loss": 0.7747,
+      "loss": 0.8513,
       "step": 12289
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2185984176504994e-05,
-      "loss": 0.7302,
+      "loss": 0.8654,
       "step": 12290
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.218483478636583e-05,
-      "loss": 0.8054,
+      "loss": 0.8541,
       "step": 12291
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2183685365915183e-05,
-      "loss": 0.8648,
+      "loss": 0.906,
       "step": 12292
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2182535915169002e-05,
-      "loss": 0.7196,
+      "loss": 0.8645,
       "step": 12293
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.218138643414323e-05,
-      "loss": 0.7597,
+      "loss": 0.886,
       "step": 12294
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2180236922853812e-05,
-      "loss": 0.828,
+      "loss": 0.9646,
       "step": 12295
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2179087381316705e-05,
-      "loss": 0.7186,
+      "loss": 0.9338,
       "step": 12296
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.217793780954785e-05,
-      "loss": 0.6994,
+      "loss": 0.8872,
       "step": 12297
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.21767882075632e-05,
-      "loss": 0.7892,
+      "loss": 0.8792,
       "step": 12298
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2175638575378701e-05,
-      "loss": 0.5851,
+      "loss": 0.8242,
       "step": 12299
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2174488913010302e-05,
-      "loss": 0.8463,
+      "loss": 0.9088,
       "step": 12300
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2173339220473956e-05,
-      "loss": 0.6859,
+      "loss": 0.8245,
       "step": 12301
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2172189497785612e-05,
-      "loss": 0.6575,
+      "loss": 0.7906,
       "step": 12302
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.217103974496122e-05,
-      "loss": 0.772,
+      "loss": 0.8928,
       "step": 12303
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2169889962016731e-05,
-      "loss": 0.6967,
+      "loss": 0.8725,
       "step": 12304
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2168740148968101e-05,
-      "loss": 0.8849,
+      "loss": 0.8788,
       "step": 12305
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2167590305831274e-05,
-      "loss": 0.8762,
+      "loss": 0.9381,
       "step": 12306
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.216644043262221e-05,
-      "loss": 0.6801,
+      "loss": 0.898,
       "step": 12307
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2165290529356858e-05,
-      "loss": 0.9032,
+      "loss": 0.9591,
       "step": 12308
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.216414059605117e-05,
-      "loss": 0.8278,
+      "loss": 0.9634,
       "step": 12309
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2162990632721106e-05,
-      "loss": 0.623,
+      "loss": 0.8183,
       "step": 12310
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2161840639382614e-05,
-      "loss": 0.663,
+      "loss": 0.8202,
       "step": 12311
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.216069061605165e-05,
-      "loss": 0.9396,
+      "loss": 0.9536,
       "step": 12312
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2159540562744169e-05,
-      "loss": 0.6342,
+      "loss": 0.7322,
       "step": 12313
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2158390479476125e-05,
-      "loss": 0.7315,
+      "loss": 0.8918,
       "step": 12314
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.215724036626348e-05,
-      "loss": 0.7499,
+      "loss": 0.849,
       "step": 12315
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2156090223122183e-05,
-      "loss": 0.7842,
+      "loss": 0.8901,
       "step": 12316
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2154940050068191e-05,
-      "loss": 0.5675,
+      "loss": 0.7792,
       "step": 12317
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2153789847117468e-05,
-      "loss": 0.8384,
+      "loss": 0.8911,
       "step": 12318
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2152639614285962e-05,
-      "loss": 0.8903,
+      "loss": 0.9138,
       "step": 12319
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2151489351589638e-05,
-      "loss": 0.696,
+      "loss": 0.837,
       "step": 12320
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2150339059044452e-05,
-      "loss": 0.805,
+      "loss": 0.9286,
       "step": 12321
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.214918873666636e-05,
-      "loss": 0.8155,
+      "loss": 0.7986,
       "step": 12322
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2148038384471328e-05,
-      "loss": 0.776,
+      "loss": 0.9245,
       "step": 12323
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2146888002475307e-05,
-      "loss": 0.7472,
+      "loss": 0.8714,
       "step": 12324
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2145737590694263e-05,
-      "loss": 0.7654,
+      "loss": 0.8668,
       "step": 12325
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2144587149144154e-05,
-      "loss": 0.6994,
+      "loss": 0.8354,
       "step": 12326
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2143436677840943e-05,
-      "loss": 0.7983,
+      "loss": 0.8503,
       "step": 12327
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2142286176800586e-05,
-      "loss": 0.91,
+      "loss": 0.9699,
       "step": 12328
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2141135646039048e-05,
-      "loss": 0.6992,
+      "loss": 0.836,
       "step": 12329
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2139985085572293e-05,
-      "loss": 0.6758,
+      "loss": 0.838,
       "step": 12330
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2138834495416279e-05,
-      "loss": 0.8549,
+      "loss": 0.9434,
       "step": 12331
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2137683875586976e-05,
-      "loss": 0.7192,
+      "loss": 0.8391,
       "step": 12332
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2136533226100336e-05,
-      "loss": 0.6717,
+      "loss": 0.8449,
       "step": 12333
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2135382546972332e-05,
-      "loss": 0.7303,
+      "loss": 0.8138,
       "step": 12334
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2134231838218922e-05,
-      "loss": 0.7552,
+      "loss": 0.9475,
       "step": 12335
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2133081099856076e-05,
-      "loss": 0.7384,
+      "loss": 0.8607,
       "step": 12336
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2131930331899757e-05,
-      "loss": 0.6542,
+      "loss": 0.849,
       "step": 12337
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2130779534365925e-05,
-      "loss": 0.7498,
+      "loss": 0.913,
       "step": 12338
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2129628707270556e-05,
-      "loss": 0.7856,
+      "loss": 0.9203,
       "step": 12339
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2128477850629608e-05,
-      "loss": 0.7693,
+      "loss": 0.8868,
       "step": 12340
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2127326964459047e-05,
-      "loss": 0.9123,
+      "loss": 0.9771,
       "step": 12341
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2126176048774844e-05,
-      "loss": 0.8504,
+      "loss": 0.9111,
       "step": 12342
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2125025103592966e-05,
-      "loss": 0.7872,
+      "loss": 0.9059,
       "step": 12343
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.212387412892938e-05,
-      "loss": 0.7926,
+      "loss": 0.9005,
       "step": 12344
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2122723124800054e-05,
-      "loss": 0.7095,
+      "loss": 0.8434,
       "step": 12345
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2121572091220955e-05,
-      "loss": 0.7051,
+      "loss": 0.884,
       "step": 12346
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2120421028208053e-05,
-      "loss": 0.6391,
+      "loss": 0.8475,
       "step": 12347
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.211926993577732e-05,
-      "loss": 0.7812,
+      "loss": 0.913,
       "step": 12348
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2118118813944721e-05,
-      "loss": 0.6587,
+      "loss": 0.7923,
       "step": 12349
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.211696766272623e-05,
-      "loss": 0.927,
+      "loss": 0.9828,
       "step": 12350
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2115816482137817e-05,
-      "loss": 0.8612,
+      "loss": 0.9218,
       "step": 12351
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2114665272195452e-05,
-      "loss": 0.7582,
+      "loss": 0.8339,
       "step": 12352
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2113514032915106e-05,
-      "loss": 0.8616,
+      "loss": 1.0063,
       "step": 12353
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.211236276431275e-05,
-      "loss": 0.8001,
+      "loss": 0.8236,
       "step": 12354
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2111211466404361e-05,
-      "loss": 0.8781,
+      "loss": 0.9031,
       "step": 12355
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2110060139205908e-05,
-      "loss": 0.8688,
+      "loss": 1.002,
       "step": 12356
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.210890878273336e-05,
-      "loss": 0.797,
+      "loss": 0.9249,
       "step": 12357
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.21077573970027e-05,
-      "loss": 0.7683,
+      "loss": 0.9127,
       "step": 12358
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2106605982029893e-05,
-      "loss": 0.913,
+      "loss": 0.9653,
       "step": 12359
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2105454537830919e-05,
-      "loss": 0.735,
+      "loss": 0.77,
       "step": 12360
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.210430306442175e-05,
-      "loss": 0.8028,
+      "loss": 0.8732,
       "step": 12361
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2103151561818359e-05,
-      "loss": 0.7452,
+      "loss": 0.8231,
       "step": 12362
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2102000030036727e-05,
-      "loss": 0.935,
+      "loss": 0.9897,
       "step": 12363
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2100848469092827e-05,
-      "loss": 0.7422,
+      "loss": 0.8551,
       "step": 12364
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2099696879002633e-05,
-      "loss": 0.8497,
+      "loss": 0.9283,
       "step": 12365
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2098545259782126e-05,
-      "loss": 0.5585,
+      "loss": 0.7695,
       "step": 12366
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2097393611447282e-05,
-      "loss": 0.8282,
+      "loss": 0.9113,
       "step": 12367
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2096241934014074e-05,
-      "loss": 0.8301,
+      "loss": 0.9737,
       "step": 12368
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2095090227498486e-05,
-      "loss": 0.6272,
+      "loss": 0.8306,
       "step": 12369
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2093938491916493e-05,
-      "loss": 0.7118,
+      "loss": 0.9004,
       "step": 12370
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2092786727284077e-05,
-      "loss": 0.8275,
+      "loss": 0.908,
       "step": 12371
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2091634933617212e-05,
-      "loss": 0.7516,
+      "loss": 0.8419,
       "step": 12372
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2090483110931877e-05,
-      "loss": 0.8302,
+      "loss": 0.9506,
       "step": 12373
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2089331259244057e-05,
-      "loss": 0.7505,
+      "loss": 0.9088,
       "step": 12374
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2088179378569729e-05,
-      "loss": 0.7702,
+      "loss": 0.8376,
       "step": 12375
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2087027468924877e-05,
-      "loss": 0.6063,
+      "loss": 0.8792,
       "step": 12376
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2085875530325482e-05,
-      "loss": 0.8385,
+      "loss": 0.9121,
       "step": 12377
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2084723562787519e-05,
-      "loss": 0.7063,
+      "loss": 0.882,
       "step": 12378
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2083571566326977e-05,
-      "loss": 0.9021,
+      "loss": 0.9569,
       "step": 12379
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2082419540959835e-05,
-      "loss": 0.831,
+      "loss": 0.8715,
       "step": 12380
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2081267486702076e-05,
-      "loss": 0.7802,
+      "loss": 0.837,
       "step": 12381
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2080115403569684e-05,
-      "loss": 0.5373,
+      "loss": 0.727,
       "step": 12382
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2078963291578642e-05,
-      "loss": 0.7466,
+      "loss": 0.8826,
       "step": 12383
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2077811150744934e-05,
-      "loss": 0.8827,
+      "loss": 0.8919,
       "step": 12384
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2076658981084543e-05,
-      "loss": 0.8206,
+      "loss": 0.8945,
       "step": 12385
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2075506782613457e-05,
-      "loss": 0.6251,
+      "loss": 0.8578,
       "step": 12386
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2074354555347658e-05,
-      "loss": 0.6111,
+      "loss": 0.7321,
       "step": 12387
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2073202299303134e-05,
-      "loss": 0.8343,
+      "loss": 0.8834,
       "step": 12388
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2072050014495868e-05,
-      "loss": 0.8596,
+      "loss": 0.9296,
       "step": 12389
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2070897700941851e-05,
-      "loss": 0.8839,
+      "loss": 0.9703,
       "step": 12390
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2069745358657064e-05,
-      "loss": 0.6817,
+      "loss": 0.9007,
       "step": 12391
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2068592987657497e-05,
-      "loss": 0.6619,
+      "loss": 0.8732,
       "step": 12392
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2067440587959138e-05,
-      "loss": 0.7334,
+      "loss": 0.8402,
       "step": 12393
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2066288159577972e-05,
-      "loss": 0.799,
+      "loss": 0.8878,
       "step": 12394
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.206513570252999e-05,
-      "loss": 0.7374,
+      "loss": 0.8721,
       "step": 12395
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2063983216831183e-05,
-      "loss": 0.8133,
+      "loss": 0.8542,
       "step": 12396
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2062830702497535e-05,
-      "loss": 0.9828,
+      "loss": 0.9731,
       "step": 12397
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2061678159545039e-05,
-      "loss": 0.7378,
+      "loss": 0.8881,
       "step": 12398
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2060525587989681e-05,
-      "loss": 0.7266,
+      "loss": 0.7994,
       "step": 12399
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2059372987847455e-05,
-      "loss": 0.7433,
+      "loss": 0.8406,
       "step": 12400
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2058220359134354e-05,
-      "loss": 0.7316,
+      "loss": 0.8808,
       "step": 12401
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.205706770186636e-05,
-      "loss": 0.7926,
+      "loss": 0.8944,
       "step": 12402
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2055915016059474e-05,
-      "loss": 0.7452,
+      "loss": 0.8625,
       "step": 12403
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2054762301729685e-05,
-      "loss": 0.9549,
+      "loss": 0.9918,
       "step": 12404
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2053609558892983e-05,
-      "loss": 0.7877,
+      "loss": 0.8592,
       "step": 12405
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2052456787565361e-05,
-      "loss": 0.7628,
+      "loss": 0.879,
       "step": 12406
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2051303987762815e-05,
-      "loss": 0.6858,
+      "loss": 0.838,
       "step": 12407
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2050151159501335e-05,
-      "loss": 0.7563,
+      "loss": 0.8974,
       "step": 12408
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.204899830279692e-05,
-      "loss": 0.8028,
+      "loss": 0.8759,
       "step": 12409
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2047845417665556e-05,
-      "loss": 0.8632,
+      "loss": 0.9481,
       "step": 12410
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2046692504123245e-05,
-      "loss": 0.8075,
+      "loss": 0.8873,
       "step": 12411
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2045539562185979e-05,
-      "loss": 0.6731,
+      "loss": 0.8346,
       "step": 12412
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2044386591869752e-05,
-      "loss": 0.8654,
+      "loss": 0.8781,
       "step": 12413
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2043233593190565e-05,
-      "loss": 0.6816,
+      "loss": 0.8221,
       "step": 12414
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2042080566164408e-05,
-      "loss": 0.7487,
+      "loss": 0.7838,
       "step": 12415
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.204092751080728e-05,
-      "loss": 0.7169,
+      "loss": 0.8721,
       "step": 12416
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.203977442713518e-05,
-      "loss": 0.7591,
+      "loss": 0.8831,
       "step": 12417
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2038621315164103e-05,
-      "loss": 0.7145,
+      "loss": 0.8798,
       "step": 12418
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2037468174910049e-05,
-      "loss": 0.688,
+      "loss": 0.8231,
       "step": 12419
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2036315006389015e-05,
-      "loss": 0.8887,
+      "loss": 0.9683,
       "step": 12420
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2035161809616998e-05,
-      "loss": 0.7214,
+      "loss": 0.8466,
       "step": 12421
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.203400858461e-05,
-      "loss": 0.6683,
+      "loss": 0.8433,
       "step": 12422
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2032855331384017e-05,
-      "loss": 0.7225,
+      "loss": 0.889,
       "step": 12423
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.203170204995505e-05,
-      "loss": 0.6593,
+      "loss": 0.768,
       "step": 12424
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2030548740339102e-05,
-      "loss": 0.7908,
+      "loss": 0.8875,
       "step": 12425
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2029395402552171e-05,
-      "loss": 0.7,
+      "loss": 0.8508,
       "step": 12426
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2028242036610257e-05,
-      "loss": 0.6194,
+      "loss": 0.7922,
       "step": 12427
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2027088642529366e-05,
-      "loss": 0.7568,
+      "loss": 0.8533,
       "step": 12428
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.202593522032549e-05,
-      "loss": 0.6788,
+      "loss": 0.8299,
       "step": 12429
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2024781770014642e-05,
-      "loss": 0.7316,
+      "loss": 0.8853,
       "step": 12430
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.202362829161282e-05,
-      "loss": 0.6368,
+      "loss": 0.8269,
       "step": 12431
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.202247478513602e-05,
-      "loss": 0.663,
+      "loss": 0.8097,
       "step": 12432
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2021321250600258e-05,
-      "loss": 0.8307,
+      "loss": 0.9881,
       "step": 12433
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.202016768802153e-05,
-      "loss": 0.7247,
+      "loss": 0.8448,
       "step": 12434
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2019014097415843e-05,
-      "loss": 0.842,
+      "loss": 0.9512,
       "step": 12435
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2017860478799196e-05,
-      "loss": 0.6719,
+      "loss": 0.8121,
       "step": 12436
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2016706832187602e-05,
-      "loss": 0.5426,
+      "loss": 0.734,
       "step": 12437
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.201555315759706e-05,
-      "loss": 0.6918,
+      "loss": 0.8019,
       "step": 12438
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2014399455043577e-05,
-      "loss": 0.6307,
+      "loss": 0.7822,
       "step": 12439
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.201324572454316e-05,
-      "loss": 0.706,
+      "loss": 0.8284,
       "step": 12440
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2012091966111816e-05,
-      "loss": 0.7629,
+      "loss": 0.8917,
       "step": 12441
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2010938179765553e-05,
-      "loss": 0.7237,
+      "loss": 0.8401,
       "step": 12442
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2009784365520371e-05,
-      "loss": 0.6801,
+      "loss": 0.776,
       "step": 12443
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2008630523392286e-05,
-      "loss": 0.8242,
+      "loss": 0.8453,
       "step": 12444
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2007476653397298e-05,
-      "loss": 0.7333,
+      "loss": 0.9269,
       "step": 12445
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2006322755551425e-05,
-      "loss": 0.8697,
+      "loss": 0.9942,
       "step": 12446
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2005168829870668e-05,
-      "loss": 0.7397,
+      "loss": 0.8193,
       "step": 12447
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.200401487637104e-05,
-      "loss": 0.6942,
+      "loss": 0.7982,
       "step": 12448
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2002860895068549e-05,
-      "loss": 0.716,
+      "loss": 0.8398,
       "step": 12449
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.20017068859792e-05,
-      "loss": 0.7392,
+      "loss": 0.8301,
       "step": 12450
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.2000552849119012e-05,
-      "loss": 0.6917,
+      "loss": 0.791,
       "step": 12451
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1999398784503993e-05,
-      "loss": 0.7329,
+      "loss": 0.8669,
       "step": 12452
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1998244692150149e-05,
-      "loss": 0.9307,
+      "loss": 0.9779,
       "step": 12453
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1997090572073498e-05,
-      "loss": 0.7233,
+      "loss": 0.9199,
       "step": 12454
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1995936424290047e-05,
-      "loss": 0.7318,
+      "loss": 0.8877,
       "step": 12455
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1994782248815808e-05,
-      "loss": 0.8311,
+      "loss": 0.8496,
       "step": 12456
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1993628045666799e-05,
-      "loss": 0.6757,
+      "loss": 0.799,
       "step": 12457
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.199247381485903e-05,
-      "loss": 0.6968,
+      "loss": 0.7808,
       "step": 12458
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.199131955640851e-05,
-      "loss": 0.6804,
+      "loss": 0.906,
       "step": 12459
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.199016527033126e-05,
-      "loss": 0.7885,
+      "loss": 0.8916,
       "step": 12460
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1989010956643287e-05,
-      "loss": 0.8918,
+      "loss": 0.9458,
       "step": 12461
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.198785661536061e-05,
-      "loss": 0.7966,
+      "loss": 0.9302,
       "step": 12462
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1986702246499246e-05,
-      "loss": 0.851,
+      "loss": 0.878,
       "step": 12463
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1985547850075206e-05,
-      "loss": 0.8347,
+      "loss": 0.9612,
       "step": 12464
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1984393426104506e-05,
-      "loss": 0.8279,
+      "loss": 0.9122,
       "step": 12465
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1983238974603161e-05,
-      "loss": 0.8474,
+      "loss": 0.8572,
       "step": 12466
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1982084495587195e-05,
-      "loss": 0.6867,
+      "loss": 0.7843,
       "step": 12467
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1980929989072614e-05,
-      "loss": 0.7355,
+      "loss": 0.8972,
       "step": 12468
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1979775455075437e-05,
-      "loss": 0.8178,
+      "loss": 0.8476,
       "step": 12469
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.197862089361169e-05,
-      "loss": 0.7965,
+      "loss": 0.8298,
       "step": 12470
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1977466304697382e-05,
-      "loss": 0.7035,
+      "loss": 0.7535,
       "step": 12471
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1976311688348537e-05,
-      "loss": 0.6514,
+      "loss": 0.8076,
       "step": 12472
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.197515704458117e-05,
-      "loss": 0.7258,
+      "loss": 0.8194,
       "step": 12473
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.19740023734113e-05,
-      "loss": 0.8275,
+      "loss": 0.9055,
       "step": 12474
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1972847674854951e-05,
-      "loss": 0.6793,
+      "loss": 0.8105,
       "step": 12475
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1971692948928137e-05,
-      "loss": 0.8881,
+      "loss": 0.959,
       "step": 12476
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1970538195646881e-05,
-      "loss": 0.7195,
+      "loss": 0.8678,
       "step": 12477
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1969383415027207e-05,
-      "loss": 0.7922,
+      "loss": 0.8929,
       "step": 12478
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1968228607085129e-05,
-      "loss": 0.7737,
+      "loss": 0.8946,
       "step": 12479
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.196707377183667e-05,
-      "loss": 0.7702,
+      "loss": 0.9146,
       "step": 12480
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1965918909297857e-05,
-      "loss": 0.6892,
+      "loss": 0.7875,
       "step": 12481
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1964764019484706e-05,
-      "loss": 0.7771,
+      "loss": 0.8627,
       "step": 12482
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1963609102413242e-05,
-      "loss": 0.7538,
+      "loss": 0.8335,
       "step": 12483
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1962454158099491e-05,
-      "loss": 0.7796,
+      "loss": 0.8262,
       "step": 12484
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.196129918655947e-05,
-      "loss": 0.7854,
+      "loss": 0.8902,
       "step": 12485
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1960144187809208e-05,
-      "loss": 0.7254,
+      "loss": 0.8333,
       "step": 12486
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1958989161864725e-05,
-      "loss": 0.9415,
+      "loss": 1.0172,
       "step": 12487
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1957834108742046e-05,
-      "loss": 0.7036,
+      "loss": 0.8375,
       "step": 12488
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1956679028457196e-05,
-      "loss": 0.6712,
+      "loss": 0.776,
       "step": 12489
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.19555239210262e-05,
-      "loss": 0.7541,
+      "loss": 0.8503,
       "step": 12490
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1954368786465087e-05,
-      "loss": 0.7597,
+      "loss": 0.8922,
       "step": 12491
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1953213624789879e-05,
-      "loss": 0.7345,
+      "loss": 0.8572,
       "step": 12492
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.19520584360166e-05,
-      "loss": 0.7595,
+      "loss": 0.9455,
       "step": 12493
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1950903220161286e-05,
-      "loss": 0.8537,
+      "loss": 0.9567,
       "step": 12494
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1949747977239953e-05,
-      "loss": 0.8016,
+      "loss": 0.8316,
       "step": 12495
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1948592707268634e-05,
-      "loss": 0.735,
+      "loss": 0.8647,
       "step": 12496
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1947437410263355e-05,
-      "loss": 0.6441,
+      "loss": 0.8185,
       "step": 12497
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1946282086240148e-05,
-      "loss": 0.8397,
+      "loss": 0.94,
       "step": 12498
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1945126735215037e-05,
-      "loss": 0.8142,
+      "loss": 0.853,
       "step": 12499
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1943971357204052e-05,
-      "loss": 0.707,
+      "loss": 0.839,
       "step": 12500
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.194281595222322e-05,
-      "loss": 0.7437,
+      "loss": 0.8687,
       "step": 12501
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1941660520288579e-05,
-      "loss": 0.7426,
+      "loss": 0.8615,
       "step": 12502
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.194050506141615e-05,
-      "loss": 0.8624,
+      "loss": 0.8762,
       "step": 12503
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1939349575621966e-05,
-      "loss": 0.7368,
+      "loss": 0.8357,
       "step": 12504
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.193819406292206e-05,
-      "loss": 0.709,
+      "loss": 0.759,
       "step": 12505
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1937038523332457e-05,
-      "loss": 0.7805,
+      "loss": 0.855,
       "step": 12506
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1935882956869198e-05,
-      "loss": 0.7993,
+      "loss": 0.8858,
       "step": 12507
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1934727363548308e-05,
-      "loss": 0.77,
+      "loss": 0.8109,
       "step": 12508
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.193357174338582e-05,
-      "loss": 0.884,
+      "loss": 0.9647,
       "step": 12509
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1932416096397767e-05,
-      "loss": 0.8402,
+      "loss": 0.972,
       "step": 12510
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.1931260422600183e-05,
-      "loss": 0.7958,
+      "loss": 0.7969,
       "step": 12511
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.19301047220091e-05,
-      "loss": 0.801,
+      "loss": 0.9307,
       "step": 12512
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1928948994640554e-05,
-      "loss": 0.8358,
+      "loss": 0.8734,
       "step": 12513
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.192779324051058e-05,
-      "loss": 0.7447,
+      "loss": 0.8923,
       "step": 12514
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1926637459635203e-05,
-      "loss": 0.7097,
+      "loss": 0.8487,
       "step": 12515
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.192548165203047e-05,
-      "loss": 0.7661,
+      "loss": 0.8994,
       "step": 12516
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1924325817712408e-05,
-      "loss": 0.8292,
+      "loss": 0.9171,
       "step": 12517
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1923169956697058e-05,
-      "loss": 0.807,
+      "loss": 0.9349,
       "step": 12518
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1922014069000454e-05,
-      "loss": 0.7268,
+      "loss": 0.8495,
       "step": 12519
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1920858154638627e-05,
-      "loss": 0.7759,
+      "loss": 0.899,
       "step": 12520
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1919702213627624e-05,
-      "loss": 0.6706,
+      "loss": 0.6946,
       "step": 12521
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1918546245983476e-05,
-      "loss": 0.7518,
+      "loss": 0.8839,
       "step": 12522
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1917390251722218e-05,
-      "loss": 0.772,
+      "loss": 0.9187,
       "step": 12523
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1916234230859893e-05,
-      "loss": 0.724,
+      "loss": 0.8503,
       "step": 12524
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1915078183412534e-05,
-      "loss": 0.7486,
+      "loss": 0.8991,
       "step": 12525
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1913922109396185e-05,
-      "loss": 0.7498,
+      "loss": 0.8075,
       "step": 12526
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1912766008826882e-05,
-      "loss": 0.7172,
+      "loss": 0.8501,
       "step": 12527
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1911609881720663e-05,
-      "loss": 0.7385,
+      "loss": 0.9105,
       "step": 12528
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1910453728093569e-05,
-      "loss": 0.7356,
+      "loss": 0.8395,
       "step": 12529
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1909297547961641e-05,
-      "loss": 0.7558,
+      "loss": 0.8569,
       "step": 12530
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1908141341340918e-05,
-      "loss": 0.6656,
+      "loss": 0.8653,
       "step": 12531
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1906985108247441e-05,
-      "loss": 0.7299,
+      "loss": 0.8699,
       "step": 12532
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1905828848697252e-05,
-      "loss": 0.9033,
+      "loss": 0.8935,
       "step": 12533
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1904672562706392e-05,
-      "loss": 0.7769,
+      "loss": 0.9197,
       "step": 12534
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1903516250290903e-05,
-      "loss": 0.9427,
+      "loss": 1.0199,
       "step": 12535
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1902359911466826e-05,
-      "loss": 0.8252,
+      "loss": 0.9045,
       "step": 12536
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1901203546250205e-05,
-      "loss": 0.7612,
+      "loss": 0.8934,
       "step": 12537
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1900047154657081e-05,
-      "loss": 0.8429,
+      "loss": 0.8918,
       "step": 12538
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1898890736703499e-05,
-      "loss": 0.7439,
+      "loss": 0.9046,
       "step": 12539
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1897734292405503e-05,
-      "loss": 0.711,
+      "loss": 0.8889,
       "step": 12540
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1896577821779135e-05,
-      "loss": 0.6093,
+      "loss": 0.7623,
       "step": 12541
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.189542132484044e-05,
-      "loss": 0.8112,
+      "loss": 0.8731,
       "step": 12542
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1894264801605466e-05,
-      "loss": 0.7834,
+      "loss": 0.8623,
       "step": 12543
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1893108252090253e-05,
-      "loss": 0.7701,
+      "loss": 0.8939,
       "step": 12544
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1891951676310851e-05,
-      "loss": 0.5558,
+      "loss": 0.7419,
       "step": 12545
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1890795074283301e-05,
-      "loss": 0.7268,
+      "loss": 0.8461,
       "step": 12546
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1889638446023655e-05,
-      "loss": 0.6678,
+      "loss": 0.7959,
       "step": 12547
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1888481791547957e-05,
-      "loss": 0.7998,
+      "loss": 0.9126,
       "step": 12548
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.188732511087225e-05,
-      "loss": 0.767,
+      "loss": 0.8815,
       "step": 12549
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1886168404012589e-05,
-      "loss": 0.786,
+      "loss": 0.9072,
       "step": 12550
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1885011670985014e-05,
-      "loss": 0.849,
+      "loss": 0.9426,
       "step": 12551
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1883854911805576e-05,
-      "loss": 0.6191,
+      "loss": 0.8346,
       "step": 12552
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1882698126490327e-05,
-      "loss": 0.7858,
+      "loss": 0.9183,
       "step": 12553
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1881541315055314e-05,
-      "loss": 0.7798,
+      "loss": 0.8585,
       "step": 12554
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.188038447751658e-05,
-      "loss": 0.6614,
+      "loss": 0.7996,
       "step": 12555
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1879227613890181e-05,
-      "loss": 0.9522,
+      "loss": 0.9649,
       "step": 12556
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1878070724192165e-05,
-      "loss": 0.7095,
+      "loss": 0.8429,
       "step": 12557
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1876913808438584e-05,
-      "loss": 0.6923,
+      "loss": 0.8509,
       "step": 12558
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1875756866645487e-05,
-      "loss": 0.7585,
+      "loss": 0.8918,
       "step": 12559
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1874599898828923e-05,
-      "loss": 0.7977,
+      "loss": 0.8268,
       "step": 12560
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1873442905004946e-05,
-      "loss": 0.7257,
+      "loss": 0.7563,
       "step": 12561
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1872285885189607e-05,
-      "loss": 0.8044,
+      "loss": 0.848,
       "step": 12562
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1871128839398958e-05,
-      "loss": 0.7513,
+      "loss": 0.7976,
       "step": 12563
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.186997176764905e-05,
-      "loss": 0.7083,
+      "loss": 0.8036,
       "step": 12564
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1868814669955936e-05,
-      "loss": 0.7548,
+      "loss": 0.8171,
       "step": 12565
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1867657546335674e-05,
-      "loss": 0.6979,
+      "loss": 0.8709,
       "step": 12566
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.186650039680431e-05,
-      "loss": 0.7469,
+      "loss": 0.8425,
       "step": 12567
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1865343221377903e-05,
-      "loss": 0.7496,
+      "loss": 0.8767,
       "step": 12568
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1864186020072503e-05,
-      "loss": 0.5706,
+      "loss": 0.7941,
       "step": 12569
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.186302879290417e-05,
-      "loss": 0.8042,
+      "loss": 0.8952,
       "step": 12570
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1861871539888951e-05,
-      "loss": 0.7045,
+      "loss": 0.8684,
       "step": 12571
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1860714261042911e-05,
-      "loss": 0.7049,
+      "loss": 0.8032,
       "step": 12572
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1859556956382097e-05,
-      "loss": 0.7653,
+      "loss": 0.8457,
       "step": 12573
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1858399625922573e-05,
-      "loss": 0.8211,
+      "loss": 0.9226,
       "step": 12574
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1857242269680386e-05,
-      "loss": 0.8545,
+      "loss": 0.9339,
       "step": 12575
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1856084887671599e-05,
-      "loss": 0.7082,
+      "loss": 0.8259,
       "step": 12576
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.185492747991227e-05,
-      "loss": 0.7997,
+      "loss": 0.94,
       "step": 12577
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1853770046418454e-05,
-      "loss": 0.7164,
+      "loss": 0.8035,
       "step": 12578
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1852612587206204e-05,
-      "loss": 0.7738,
+      "loss": 0.9101,
       "step": 12579
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1851455102291588e-05,
-      "loss": 0.6509,
+      "loss": 0.8138,
       "step": 12580
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1850297591690658e-05,
-      "loss": 0.7192,
+      "loss": 0.8624,
       "step": 12581
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1849140055419474e-05,
-      "loss": 0.7894,
+      "loss": 0.8756,
       "step": 12582
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1847982493494097e-05,
-      "loss": 0.7331,
+      "loss": 0.8422,
       "step": 12583
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.184682490593058e-05,
-      "loss": 0.9187,
+      "loss": 0.8846,
       "step": 12584
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.184566729274499e-05,
-      "loss": 0.8034,
+      "loss": 0.9001,
       "step": 12585
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1844509653953389e-05,
-      "loss": 0.7229,
+      "loss": 0.7892,
       "step": 12586
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1843351989571828e-05,
-      "loss": 0.8046,
+      "loss": 0.8135,
       "step": 12587
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1842194299616377e-05,
-      "loss": 0.6279,
+      "loss": 0.8028,
       "step": 12588
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.184103658410309e-05,
-      "loss": 0.6925,
+      "loss": 0.8823,
       "step": 12589
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1839878843048036e-05,
-      "loss": 0.7537,
+      "loss": 0.8003,
       "step": 12590
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1838721076467273e-05,
-      "loss": 0.8337,
+      "loss": 0.9047,
       "step": 12591
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.183756328437686e-05,
-      "loss": 0.7598,
+      "loss": 0.9162,
       "step": 12592
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1836405466792869e-05,
-      "loss": 0.8106,
+      "loss": 0.9008,
       "step": 12593
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1835247623731356e-05,
-      "loss": 0.6413,
+      "loss": 0.8194,
       "step": 12594
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1834089755208385e-05,
-      "loss": 0.7024,
+      "loss": 0.8353,
       "step": 12595
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1832931861240022e-05,
-      "loss": 0.8346,
+      "loss": 0.854,
       "step": 12596
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1831773941842331e-05,
-      "loss": 0.7551,
+      "loss": 0.8152,
       "step": 12597
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1830615997031372e-05,
-      "loss": 0.6252,
+      "loss": 0.7522,
       "step": 12598
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1829458026823215e-05,
-      "loss": 0.7538,
+      "loss": 0.8913,
       "step": 12599
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1828300031233926e-05,
-      "loss": 0.704,
+      "loss": 0.8074,
       "step": 12600
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1827142010279567e-05,
-      "loss": 0.7138,
+      "loss": 0.8365,
       "step": 12601
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1825983963976202e-05,
-      "loss": 0.7116,
+      "loss": 0.821,
       "step": 12602
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1824825892339902e-05,
-      "loss": 0.7017,
+      "loss": 0.9278,
       "step": 12603
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1823667795386733e-05,
-      "loss": 0.7983,
+      "loss": 0.8622,
       "step": 12604
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1822509673132759e-05,
-      "loss": 0.7581,
+      "loss": 0.8361,
       "step": 12605
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.182135152559405e-05,
-      "loss": 0.6405,
+      "loss": 0.7935,
       "step": 12606
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1820193352786673e-05,
-      "loss": 0.7129,
+      "loss": 0.8793,
       "step": 12607
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1819035154726694e-05,
-      "loss": 0.632,
+      "loss": 0.7974,
       "step": 12608
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1817876931430185e-05,
-      "loss": 0.6749,
+      "loss": 0.7682,
       "step": 12609
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1816718682913213e-05,
-      "loss": 0.6601,
+      "loss": 0.7637,
       "step": 12610
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1815560409191843e-05,
-      "loss": 0.6495,
+      "loss": 0.848,
       "step": 12611
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1814402110282155e-05,
-      "loss": 0.7533,
+      "loss": 0.927,
       "step": 12612
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1813243786200208e-05,
-      "loss": 0.7957,
+      "loss": 0.8589,
       "step": 12613
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1812085436962075e-05,
-      "loss": 0.7962,
+      "loss": 0.9033,
       "step": 12614
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.181092706258383e-05,
-      "loss": 0.7124,
+      "loss": 0.8986,
       "step": 12615
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.180976866308154e-05,
-      "loss": 0.7873,
+      "loss": 0.903,
       "step": 12616
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1808610238471279e-05,
-      "loss": 0.8579,
+      "loss": 0.9712,
       "step": 12617
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1807451788769116e-05,
-      "loss": 0.7568,
+      "loss": 0.813,
       "step": 12618
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1806293313991124e-05,
-      "loss": 0.8086,
+      "loss": 0.8333,
       "step": 12619
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1805134814153376e-05,
-      "loss": 0.7603,
+      "loss": 0.8398,
       "step": 12620
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1803976289271941e-05,
-      "loss": 0.9281,
+      "loss": 0.9472,
       "step": 12621
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1802817739362898e-05,
-      "loss": 0.8093,
+      "loss": 0.8327,
       "step": 12622
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1801659164442314e-05,
-      "loss": 0.7688,
+      "loss": 0.8718,
       "step": 12623
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1800500564526265e-05,
-      "loss": 0.6968,
+      "loss": 0.9056,
       "step": 12624
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1799341939630829e-05,
-      "loss": 0.7784,
+      "loss": 0.859,
       "step": 12625
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1798183289772074e-05,
-      "loss": 0.7771,
+      "loss": 0.8712,
       "step": 12626
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1797024614966075e-05,
-      "loss": 0.7083,
+      "loss": 0.7893,
       "step": 12627
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1795865915228912e-05,
-      "loss": 0.7759,
+      "loss": 0.8172,
       "step": 12628
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1794707190576655e-05,
-      "loss": 0.8544,
+      "loss": 0.906,
       "step": 12629
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1793548441025384e-05,
-      "loss": 0.7687,
+      "loss": 0.835,
       "step": 12630
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1792389666591173e-05,
-      "loss": 0.7162,
+      "loss": 0.7945,
       "step": 12631
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1791230867290098e-05,
-      "loss": 0.7593,
+      "loss": 0.8439,
       "step": 12632
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1790072043138237e-05,
-      "loss": 0.7969,
+      "loss": 0.8375,
       "step": 12633
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1788913194151665e-05,
-      "loss": 0.722,
+      "loss": 0.7888,
       "step": 12634
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1787754320346461e-05,
-      "loss": 0.8839,
+      "loss": 0.9568,
       "step": 12635
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1786595421738702e-05,
-      "loss": 0.8597,
+      "loss": 0.9036,
       "step": 12636
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1785436498344466e-05,
-      "loss": 0.7487,
+      "loss": 0.9177,
       "step": 12637
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1784277550179834e-05,
-      "loss": 0.7532,
+      "loss": 0.8215,
       "step": 12638
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1783118577260883e-05,
-      "loss": 0.7717,
+      "loss": 0.8398,
       "step": 12639
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1781959579603688e-05,
-      "loss": 0.6927,
+      "loss": 0.7627,
       "step": 12640
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1780800557224335e-05,
-      "loss": 0.7831,
+      "loss": 0.9681,
       "step": 12641
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.17796415101389e-05,
-      "loss": 0.7403,
+      "loss": 0.8327,
       "step": 12642
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1778482438363464e-05,
-      "loss": 0.9033,
+      "loss": 0.8822,
       "step": 12643
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1777323341914107e-05,
-      "loss": 0.8132,
+      "loss": 0.9755,
       "step": 12644
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.177616422080691e-05,
-      "loss": 0.7954,
+      "loss": 0.89,
       "step": 12645
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1775005075057955e-05,
-      "loss": 0.6445,
+      "loss": 0.8059,
       "step": 12646
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1773845904683323e-05,
-      "loss": 0.662,
+      "loss": 0.8165,
       "step": 12647
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1772686709699094e-05,
-      "loss": 0.8648,
+      "loss": 0.9135,
       "step": 12648
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1771527490121356e-05,
-      "loss": 0.6996,
+      "loss": 0.8285,
       "step": 12649
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1770368245966187e-05,
-      "loss": 0.8356,
+      "loss": 0.9212,
       "step": 12650
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1769208977249666e-05,
-      "loss": 0.7273,
+      "loss": 0.7979,
       "step": 12651
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1768049683987884e-05,
-      "loss": 0.7145,
+      "loss": 0.8122,
       "step": 12652
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.176689036619692e-05,
-      "loss": 0.7485,
+      "loss": 0.8622,
       "step": 12653
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.176573102389286e-05,
-      "loss": 0.7546,
+      "loss": 0.8021,
       "step": 12654
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1764571657091787e-05,
-      "loss": 0.5854,
+      "loss": 0.7663,
       "step": 12655
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1763412265809784e-05,
-      "loss": 0.6957,
+      "loss": 0.8204,
       "step": 12656
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1762252850062943e-05,
-      "loss": 0.6822,
+      "loss": 0.7844,
       "step": 12657
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1761093409867341e-05,
-      "loss": 0.7233,
+      "loss": 0.8805,
       "step": 12658
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1759933945239066e-05,
-      "loss": 0.8762,
+      "loss": 0.9996,
       "step": 12659
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1758774456194204e-05,
-      "loss": 0.8016,
+      "loss": 0.9187,
       "step": 12660
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1757614942748843e-05,
-      "loss": 0.7319,
+      "loss": 0.8486,
       "step": 12661
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1756455404919068e-05,
-      "loss": 0.7591,
+      "loss": 0.9211,
       "step": 12662
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1755295842720966e-05,
-      "loss": 0.7669,
+      "loss": 0.921,
       "step": 12663
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1754136256170626e-05,
-      "loss": 0.9045,
+      "loss": 0.9449,
       "step": 12664
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1752976645284133e-05,
-      "loss": 0.6593,
+      "loss": 0.8194,
       "step": 12665
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1751817010077578e-05,
-      "loss": 0.7487,
+      "loss": 0.8876,
       "step": 12666
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1750657350567043e-05,
-      "loss": 0.6164,
+      "loss": 0.7828,
       "step": 12667
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1749497666768625e-05,
-      "loss": 0.7754,
+      "loss": 0.8682,
       "step": 12668
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1748337958698409e-05,
-      "loss": 0.6954,
+      "loss": 0.7453,
       "step": 12669
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1747178226372482e-05,
-      "loss": 0.6982,
+      "loss": 0.8327,
       "step": 12670
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1746018469806938e-05,
-      "loss": 0.6483,
+      "loss": 0.8221,
       "step": 12671
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1744858689017862e-05,
-      "loss": 0.6808,
+      "loss": 0.808,
       "step": 12672
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.174369888402135e-05,
-      "loss": 0.7987,
+      "loss": 0.9061,
       "step": 12673
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.174253905483349e-05,
-      "loss": 0.6821,
+      "loss": 0.779,
       "step": 12674
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1741379201470373e-05,
-      "loss": 0.7502,
+      "loss": 0.8442,
       "step": 12675
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.174021932394809e-05,
-      "loss": 0.7914,
+      "loss": 0.9004,
       "step": 12676
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1739059422282735e-05,
-      "loss": 0.6993,
+      "loss": 0.8728,
       "step": 12677
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1737899496490397e-05,
-      "loss": 0.6922,
+      "loss": 0.8279,
       "step": 12678
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1736739546587168e-05,
-      "loss": 0.7691,
+      "loss": 0.8007,
       "step": 12679
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.173557957258914e-05,
-      "loss": 0.7589,
+      "loss": 0.8741,
       "step": 12680
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.173441957451241e-05,
-      "loss": 0.8029,
+      "loss": 0.8714,
       "step": 12681
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1733259552373071e-05,
-      "loss": 0.7206,
+      "loss": 0.8235,
       "step": 12682
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.173209950618721e-05,
-      "loss": 0.8169,
+      "loss": 0.8841,
       "step": 12683
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1730939435970932e-05,
-      "loss": 0.7958,
+      "loss": 0.8538,
       "step": 12684
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.172977934174032e-05,
-      "loss": 0.82,
+      "loss": 0.938,
       "step": 12685
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1728619223511476e-05,
-      "loss": 0.794,
+      "loss": 0.912,
       "step": 12686
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1727459081300494e-05,
-      "loss": 0.8208,
+      "loss": 0.9255,
       "step": 12687
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1726298915123465e-05,
-      "loss": 0.8069,
+      "loss": 0.8813,
       "step": 12688
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.172513872499649e-05,
-      "loss": 0.7644,
+      "loss": 0.8322,
       "step": 12689
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1723978510935666e-05,
-      "loss": 0.9136,
+      "loss": 0.8505,
       "step": 12690
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1722818272957079e-05,
-      "loss": 0.8672,
+      "loss": 0.9562,
       "step": 12691
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1721658011076838e-05,
-      "loss": 0.8314,
+      "loss": 0.9576,
       "step": 12692
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1720497725311036e-05,
-      "loss": 0.7418,
+      "loss": 0.8386,
       "step": 12693
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1719337415675765e-05,
-      "loss": 0.676,
+      "loss": 0.8269,
       "step": 12694
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1718177082187127e-05,
-      "loss": 0.777,
+      "loss": 0.9156,
       "step": 12695
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1717016724861223e-05,
-      "loss": 0.798,
+      "loss": 0.8679,
       "step": 12696
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1715856343714148e-05,
-      "loss": 0.6844,
+      "loss": 0.7879,
       "step": 12697
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1714695938761997e-05,
-      "loss": 0.724,
+      "loss": 0.8078,
       "step": 12698
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1713535510020875e-05,
-      "loss": 0.7142,
+      "loss": 0.8496,
       "step": 12699
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.171237505750688e-05,
-      "loss": 0.7913,
+      "loss": 0.8607,
       "step": 12700
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1711214581236106e-05,
-      "loss": 0.8219,
+      "loss": 0.8784,
       "step": 12701
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1710054081224662e-05,
-      "loss": 0.6758,
+      "loss": 0.8899,
       "step": 12702
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1708893557488642e-05,
-      "loss": 0.7977,
+      "loss": 0.9498,
       "step": 12703
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1707733010044149e-05,
-      "loss": 0.5879,
+      "loss": 0.7058,
       "step": 12704
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1706572438907285e-05,
-      "loss": 0.7621,
+      "loss": 0.9269,
       "step": 12705
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.170541184409415e-05,
-      "loss": 0.7219,
+      "loss": 0.821,
       "step": 12706
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1704251225620842e-05,
-      "loss": 0.8487,
+      "loss": 0.9771,
       "step": 12707
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1703090583503466e-05,
-      "loss": 0.7901,
+      "loss": 0.9344,
       "step": 12708
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1701929917758128e-05,
-      "loss": 0.9622,
+      "loss": 1.0118,
       "step": 12709
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1700769228400925e-05,
-      "loss": 0.6827,
+      "loss": 0.7975,
       "step": 12710
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1699608515447962e-05,
-      "loss": 0.7442,
+      "loss": 0.7979,
       "step": 12711
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1698447778915344e-05,
-      "loss": 0.6908,
+      "loss": 0.8429,
       "step": 12712
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1697287018819173e-05,
-      "loss": 0.7593,
+      "loss": 0.8365,
       "step": 12713
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1696126235175551e-05,
-      "loss": 0.7236,
+      "loss": 0.8405,
       "step": 12714
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1694965428000585e-05,
-      "loss": 0.761,
+      "loss": 0.8475,
       "step": 12715
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.169380459731038e-05,
-      "loss": 0.8088,
+      "loss": 0.9028,
       "step": 12716
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1692643743121036e-05,
-      "loss": 0.7967,
+      "loss": 0.8741,
       "step": 12717
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1691482865448662e-05,
-      "loss": 0.8347,
+      "loss": 0.9057,
       "step": 12718
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1690321964309366e-05,
-      "loss": 0.7421,
+      "loss": 0.8618,
       "step": 12719
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1689161039719248e-05,
-      "loss": 0.7452,
+      "loss": 0.8114,
       "step": 12720
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1688000091694417e-05,
-      "loss": 0.8045,
+      "loss": 0.932,
       "step": 12721
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1686839120250984e-05,
-      "loss": 0.7062,
+      "loss": 0.797,
       "step": 12722
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1685678125405046e-05,
-      "loss": 0.6683,
+      "loss": 0.7963,
       "step": 12723
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1684517107172718e-05,
-      "loss": 0.7701,
+      "loss": 0.8118,
       "step": 12724
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1683356065570107e-05,
-      "loss": 0.723,
+      "loss": 0.8402,
       "step": 12725
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1682195000613314e-05,
-      "loss": 0.8141,
+      "loss": 0.846,
       "step": 12726
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1681033912318458e-05,
-      "loss": 0.709,
+      "loss": 0.9025,
       "step": 12727
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1679872800701637e-05,
-      "loss": 0.8527,
+      "loss": 0.8803,
       "step": 12728
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1678711665778968e-05,
-      "loss": 0.6801,
+      "loss": 0.8586,
       "step": 12729
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1677550507566555e-05,
-      "loss": 0.7388,
+      "loss": 0.8017,
       "step": 12730
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1676389326080506e-05,
-      "loss": 0.7244,
+      "loss": 0.7857,
       "step": 12731
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1675228121336937e-05,
-      "loss": 0.8168,
+      "loss": 0.8503,
       "step": 12732
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1674066893351954e-05,
-      "loss": 0.7901,
+      "loss": 0.8728,
       "step": 12733
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.167290564214167e-05,
-      "loss": 0.5557,
+      "loss": 0.7272,
       "step": 12734
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.167174436772219e-05,
-      "loss": 0.7054,
+      "loss": 0.849,
       "step": 12735
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1670583070109632e-05,
-      "loss": 0.7374,
+      "loss": 0.8961,
       "step": 12736
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1669421749320102e-05,
-      "loss": 0.8747,
+      "loss": 0.951,
       "step": 12737
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1668260405369716e-05,
-      "loss": 0.6931,
+      "loss": 0.8773,
       "step": 12738
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1667099038274579e-05,
-      "loss": 0.6761,
+      "loss": 0.7586,
       "step": 12739
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1665937648050814e-05,
-      "loss": 0.7746,
+      "loss": 0.8612,
       "step": 12740
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1664776234714525e-05,
-      "loss": 0.804,
+      "loss": 0.9252,
       "step": 12741
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1663614798281827e-05,
-      "loss": 0.7292,
+      "loss": 0.7915,
       "step": 12742
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1662453338768835e-05,
-      "loss": 0.782,
+      "loss": 0.8801,
       "step": 12743
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1661291856191659e-05,
-      "loss": 0.7921,
+      "loss": 0.9154,
       "step": 12744
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1660130350566419e-05,
-      "loss": 0.7742,
+      "loss": 0.7832,
       "step": 12745
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1658968821909227e-05,
-      "loss": 0.5983,
+      "loss": 0.8041,
       "step": 12746
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1657807270236192e-05,
-      "loss": 0.8592,
+      "loss": 0.8471,
       "step": 12747
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1656645695563435e-05,
-      "loss": 0.7378,
+      "loss": 0.8448,
       "step": 12748
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1655484097907073e-05,
-      "loss": 0.7019,
+      "loss": 0.7939,
       "step": 12749
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1654322477283213e-05,
-      "loss": 0.7763,
+      "loss": 0.8053,
       "step": 12750
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1653160833707977e-05,
-      "loss": 0.6459,
+      "loss": 0.7948,
       "step": 12751
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1651999167197484e-05,
-      "loss": 0.7671,
+      "loss": 0.7895,
       "step": 12752
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1650837477767843e-05,
-      "loss": 0.6572,
+      "loss": 0.7704,
       "step": 12753
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1649675765435173e-05,
-      "loss": 0.6552,
+      "loss": 0.8508,
       "step": 12754
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1648514030215593e-05,
-      "loss": 0.7446,
+      "loss": 0.8566,
       "step": 12755
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1647352272125221e-05,
-      "loss": 0.7804,
+      "loss": 0.7979,
       "step": 12756
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1646190491180172e-05,
-      "loss": 0.7263,
+      "loss": 0.8768,
       "step": 12757
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1645028687396567e-05,
-      "loss": 0.7135,
+      "loss": 0.8924,
       "step": 12758
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1643866860790523e-05,
-      "loss": 0.7094,
+      "loss": 0.8929,
       "step": 12759
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1642705011378156e-05,
-      "loss": 0.7424,
+      "loss": 0.8655,
       "step": 12760
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.164154313917559e-05,
-      "loss": 0.802,
+      "loss": 0.8954,
       "step": 12761
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1640381244198944e-05,
-      "loss": 0.733,
+      "loss": 0.786,
       "step": 12762
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1639219326464332e-05,
-      "loss": 0.8908,
+      "loss": 0.9366,
       "step": 12763
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.163805738598788e-05,
-      "loss": 0.7873,
+      "loss": 0.8846,
       "step": 12764
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1636895422785706e-05,
-      "loss": 0.7753,
+      "loss": 0.8763,
       "step": 12765
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1635733436873929e-05,
-      "loss": 0.7931,
+      "loss": 0.9048,
       "step": 12766
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.163457142826867e-05,
-      "loss": 0.7339,
+      "loss": 0.8774,
       "step": 12767
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1633409396986052e-05,
-      "loss": 0.6609,
+      "loss": 0.785,
       "step": 12768
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1632247343042198e-05,
-      "loss": 0.7767,
+      "loss": 0.8587,
       "step": 12769
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1631085266453228e-05,
-      "loss": 0.6911,
+      "loss": 0.8401,
       "step": 12770
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1629923167235263e-05,
-      "loss": 0.7303,
+      "loss": 0.8262,
       "step": 12771
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1628761045404432e-05,
-      "loss": 0.8103,
+      "loss": 0.9024,
       "step": 12772
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1627598900976843e-05,
-      "loss": 0.7827,
+      "loss": 0.9039,
       "step": 12773
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1626436733968636e-05,
-      "loss": 0.6467,
+      "loss": 0.8551,
       "step": 12774
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1625274544395923e-05,
-      "loss": 0.6933,
+      "loss": 0.8153,
       "step": 12775
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1624112332274832e-05,
-      "loss": 0.7259,
+      "loss": 0.8832,
       "step": 12776
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1622950097621488e-05,
-      "loss": 0.7045,
+      "loss": 0.872,
       "step": 12777
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1621787840452015e-05,
-      "loss": 0.7341,
+      "loss": 0.8698,
       "step": 12778
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1620625560782532e-05,
-      "loss": 0.7999,
+      "loss": 0.8653,
       "step": 12779
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1619463258629174e-05,
-      "loss": 0.66,
+      "loss": 0.7217,
       "step": 12780
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1618300934008059e-05,
-      "loss": 0.7582,
+      "loss": 0.8655,
       "step": 12781
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1617138586935313e-05,
-      "loss": 0.684,
+      "loss": 0.8536,
       "step": 12782
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1615976217427065e-05,
-      "loss": 0.6922,
+      "loss": 0.8162,
       "step": 12783
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1614813825499438e-05,
-      "loss": 0.7983,
+      "loss": 0.8574,
       "step": 12784
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1613651411168562e-05,
-      "loss": 0.7413,
+      "loss": 0.8626,
       "step": 12785
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.1612488974450565e-05,
-      "loss": 0.7473,
+      "loss": 0.8958,
       "step": 12786
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1611326515361567e-05,
-      "loss": 0.7704,
+      "loss": 0.924,
       "step": 12787
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.16101640339177e-05,
-      "loss": 0.8855,
+      "loss": 0.9058,
       "step": 12788
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1609001530135095e-05,
-      "loss": 0.7276,
+      "loss": 0.9056,
       "step": 12789
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1607839004029872e-05,
-      "loss": 0.8012,
+      "loss": 0.8521,
       "step": 12790
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1606676455618171e-05,
-      "loss": 0.6191,
+      "loss": 0.7138,
       "step": 12791
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1605513884916106e-05,
-      "loss": 0.6972,
+      "loss": 0.8867,
       "step": 12792
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.160435129193982e-05,
-      "loss": 0.6174,
+      "loss": 0.7537,
       "step": 12793
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1603188676705433e-05,
-      "loss": 0.7297,
+      "loss": 0.8297,
       "step": 12794
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1602026039229077e-05,
-      "loss": 0.8211,
+      "loss": 0.8915,
       "step": 12795
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1600863379526883e-05,
-      "loss": 0.7815,
+      "loss": 0.9053,
       "step": 12796
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1599700697614982e-05,
-      "loss": 0.6993,
+      "loss": 0.8657,
       "step": 12797
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1598537993509501e-05,
-      "loss": 0.7505,
+      "loss": 0.8704,
       "step": 12798
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1597375267226579e-05,
-      "loss": 0.7234,
+      "loss": 0.728,
       "step": 12799
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1596212518782336e-05,
-      "loss": 0.8724,
+      "loss": 0.9145,
       "step": 12800
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.159504974819291e-05,
-      "loss": 0.8012,
+      "loss": 0.8573,
       "step": 12801
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1593886955474434e-05,
-      "loss": 0.7582,
+      "loss": 0.806,
       "step": 12802
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1592724140643035e-05,
-      "loss": 0.8455,
+      "loss": 0.8609,
       "step": 12803
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1591561303714852e-05,
-      "loss": 0.7744,
+      "loss": 0.8037,
       "step": 12804
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.159039844470601e-05,
-      "loss": 0.7586,
+      "loss": 0.847,
       "step": 12805
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1589235563632648e-05,
-      "loss": 0.9265,
+      "loss": 0.9117,
       "step": 12806
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1588072660510899e-05,
-      "loss": 0.7315,
+      "loss": 0.8086,
       "step": 12807
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1586909735356893e-05,
-      "loss": 0.7273,
+      "loss": 0.8214,
       "step": 12808
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1585746788186765e-05,
-      "loss": 0.6916,
+      "loss": 0.9328,
       "step": 12809
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1584583819016653e-05,
-      "loss": 0.7165,
+      "loss": 0.8707,
       "step": 12810
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1583420827862689e-05,
-      "loss": 0.7364,
+      "loss": 0.8197,
       "step": 12811
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1582257814741005e-05,
-      "loss": 0.78,
+      "loss": 0.8421,
       "step": 12812
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1581094779667741e-05,
-      "loss": 0.8174,
+      "loss": 0.8421,
       "step": 12813
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1579931722659028e-05,
-      "loss": 0.7715,
+      "loss": 0.8876,
       "step": 12814
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1578768643731007e-05,
-      "loss": 0.7421,
+      "loss": 0.8424,
       "step": 12815
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1577605542899806e-05,
-      "loss": 0.8392,
+      "loss": 0.9073,
       "step": 12816
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.157644242018157e-05,
-      "loss": 0.7255,
+      "loss": 0.8281,
       "step": 12817
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1575279275592432e-05,
-      "loss": 0.7779,
+      "loss": 0.8381,
       "step": 12818
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1574116109148527e-05,
-      "loss": 0.6323,
+      "loss": 0.7657,
       "step": 12819
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1572952920865997e-05,
-      "loss": 0.834,
+      "loss": 0.9295,
       "step": 12820
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1571789710760977e-05,
-      "loss": 0.7861,
+      "loss": 0.9808,
       "step": 12821
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1570626478849602e-05,
-      "loss": 0.8132,
+      "loss": 0.917,
       "step": 12822
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1569463225148014e-05,
-      "loss": 0.737,
+      "loss": 0.8231,
       "step": 12823
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1568299949672352e-05,
-      "loss": 0.8024,
+      "loss": 0.8923,
       "step": 12824
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1567136652438752e-05,
-      "loss": 0.8376,
+      "loss": 0.9488,
       "step": 12825
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1565973333463356e-05,
-      "loss": 0.7121,
+      "loss": 0.807,
       "step": 12826
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.15648099927623e-05,
-      "loss": 0.6493,
+      "loss": 0.7645,
       "step": 12827
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1563646630351726e-05,
-      "loss": 0.8085,
+      "loss": 0.9032,
       "step": 12828
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1562483246247775e-05,
-      "loss": 0.8041,
+      "loss": 0.8808,
       "step": 12829
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1561319840466584e-05,
-      "loss": 0.5896,
+      "loss": 0.7834,
       "step": 12830
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.15601564130243e-05,
-      "loss": 0.7146,
+      "loss": 0.8226,
       "step": 12831
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1558992963937052e-05,
-      "loss": 0.8031,
+      "loss": 0.8966,
       "step": 12832
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1557829493220994e-05,
-      "loss": 0.8323,
+      "loss": 0.946,
       "step": 12833
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.155666600089226e-05,
-      "loss": 0.753,
+      "loss": 0.8716,
       "step": 12834
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1555502486966992e-05,
-      "loss": 0.8397,
+      "loss": 0.9185,
       "step": 12835
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1554338951461337e-05,
-      "loss": 0.7634,
+      "loss": 0.9333,
       "step": 12836
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1553175394391433e-05,
-      "loss": 0.6478,
+      "loss": 0.7762,
       "step": 12837
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1552011815773426e-05,
-      "loss": 0.7533,
+      "loss": 0.79,
       "step": 12838
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1550848215623454e-05,
-      "loss": 0.7368,
+      "loss": 0.8712,
       "step": 12839
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1549684593957665e-05,
-      "loss": 0.6346,
+      "loss": 0.7524,
       "step": 12840
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.15485209507922e-05,
-      "loss": 0.7811,
+      "loss": 0.9008,
       "step": 12841
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1547357286143206e-05,
-      "loss": 0.8776,
+      "loss": 0.9388,
       "step": 12842
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1546193600026822e-05,
-      "loss": 0.8586,
+      "loss": 0.9064,
       "step": 12843
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1545029892459199e-05,
-      "loss": 0.7766,
+      "loss": 0.9299,
       "step": 12844
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1543866163456476e-05,
-      "loss": 0.7941,
+      "loss": 0.9164,
       "step": 12845
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1542702413034801e-05,
-      "loss": 0.6464,
+      "loss": 0.7731,
       "step": 12846
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.154153864121032e-05,
-      "loss": 0.764,
+      "loss": 0.8672,
       "step": 12847
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1540374847999176e-05,
-      "loss": 0.7574,
+      "loss": 0.8739,
       "step": 12848
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1539211033417516e-05,
-      "loss": 0.8025,
+      "loss": 0.8784,
       "step": 12849
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.153804719748149e-05,
-      "loss": 0.6398,
+      "loss": 0.8048,
       "step": 12850
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1536883340207239e-05,
-      "loss": 0.8677,
+      "loss": 0.8556,
       "step": 12851
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1535719461610912e-05,
-      "loss": 0.8116,
+      "loss": 0.9293,
       "step": 12852
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1534555561708659e-05,
-      "loss": 0.6882,
+      "loss": 0.7757,
       "step": 12853
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.153339164051662e-05,
-      "loss": 0.9331,
+      "loss": 1.0143,
       "step": 12854
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1532227698050951e-05,
-      "loss": 0.7053,
+      "loss": 0.8484,
       "step": 12855
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1531063734327793e-05,
-      "loss": 0.6915,
+      "loss": 0.8618,
       "step": 12856
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1529899749363303e-05,
-      "loss": 0.8265,
+      "loss": 0.8537,
       "step": 12857
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1528735743173622e-05,
-      "loss": 0.7385,
+      "loss": 0.8833,
       "step": 12858
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.15275717157749e-05,
-      "loss": 0.7807,
+      "loss": 0.8776,
       "step": 12859
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.152640766718329e-05,
-      "loss": 0.8062,
+      "loss": 0.8979,
       "step": 12860
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1525243597414938e-05,
-      "loss": 0.813,
+      "loss": 0.8675,
       "step": 12861
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1524079506485996e-05,
-      "loss": 0.7838,
+      "loss": 0.9001,
       "step": 12862
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1522915394412615e-05,
-      "loss": 0.6566,
+      "loss": 0.7838,
       "step": 12863
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1521751261210942e-05,
-      "loss": 0.8031,
+      "loss": 0.9153,
       "step": 12864
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1520587106897127e-05,
-      "loss": 0.6836,
+      "loss": 0.7925,
       "step": 12865
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1519422931487328e-05,
-      "loss": 0.7483,
+      "loss": 0.8483,
       "step": 12866
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1518258734997688e-05,
-      "loss": 0.7321,
+      "loss": 0.8109,
       "step": 12867
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1517094517444364e-05,
-      "loss": 0.8097,
+      "loss": 0.9298,
       "step": 12868
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1515930278843509e-05,
-      "loss": 0.7237,
+      "loss": 0.8564,
       "step": 12869
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1514766019211265e-05,
-      "loss": 0.8024,
+      "loss": 0.9078,
       "step": 12870
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1513601738563796e-05,
-      "loss": 0.7096,
+      "loss": 0.8064,
       "step": 12871
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1512437436917248e-05,
-      "loss": 0.6961,
+      "loss": 0.8569,
       "step": 12872
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.151127311428778e-05,
-      "loss": 0.7487,
+      "loss": 0.8848,
       "step": 12873
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.151010877069154e-05,
-      "loss": 0.8049,
+      "loss": 0.7818,
       "step": 12874
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1508944406144681e-05,
-      "loss": 0.8277,
+      "loss": 0.8912,
       "step": 12875
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1507780020663364e-05,
-      "loss": 0.7813,
+      "loss": 0.8136,
       "step": 12876
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1506615614263737e-05,
-      "loss": 0.7326,
+      "loss": 0.812,
       "step": 12877
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1505451186961952e-05,
-      "loss": 0.6894,
+      "loss": 0.7979,
       "step": 12878
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.150428673877417e-05,
-      "loss": 0.8486,
+      "loss": 0.8382,
       "step": 12879
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1503122269716546e-05,
-      "loss": 0.6416,
+      "loss": 0.7496,
       "step": 12880
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1501957779805231e-05,
-      "loss": 0.662,
+      "loss": 0.7869,
       "step": 12881
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1500793269056385e-05,
-      "loss": 0.92,
+      "loss": 0.9218,
       "step": 12882
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1499628737486157e-05,
-      "loss": 0.75,
+      "loss": 0.8815,
       "step": 12883
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.149846418511071e-05,
-      "loss": 0.8519,
+      "loss": 0.9886,
       "step": 12884
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1497299611946201e-05,
-      "loss": 0.5916,
+      "loss": 0.7696,
       "step": 12885
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1496135018008783e-05,
-      "loss": 0.6051,
+      "loss": 0.7543,
       "step": 12886
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1494970403314618e-05,
-      "loss": 0.6586,
+      "loss": 0.7686,
       "step": 12887
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1493805767879853e-05,
-      "loss": 0.7295,
+      "loss": 0.8571,
       "step": 12888
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1492641111720657e-05,
-      "loss": 0.7741,
+      "loss": 0.8773,
       "step": 12889
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1491476434853182e-05,
-      "loss": 0.8386,
+      "loss": 0.9129,
       "step": 12890
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1490311737293587e-05,
-      "loss": 0.8073,
+      "loss": 0.9167,
       "step": 12891
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1489147019058031e-05,
-      "loss": 0.6878,
+      "loss": 0.8819,
       "step": 12892
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1487982280162674e-05,
-      "loss": 0.7045,
+      "loss": 0.8163,
       "step": 12893
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1486817520623673e-05,
-      "loss": 0.7658,
+      "loss": 0.8388,
       "step": 12894
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.148565274045719e-05,
-      "loss": 0.6757,
+      "loss": 0.7779,
       "step": 12895
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1484487939679383e-05,
-      "loss": 0.7582,
+      "loss": 0.8422,
       "step": 12896
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1483323118306408e-05,
-      "loss": 0.852,
+      "loss": 0.9144,
       "step": 12897
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1482158276354433e-05,
-      "loss": 0.6588,
+      "loss": 0.8269,
       "step": 12898
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1480993413839613e-05,
-      "loss": 0.7745,
+      "loss": 0.7693,
       "step": 12899
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1479828530778112e-05,
-      "loss": 0.7759,
+      "loss": 0.8967,
       "step": 12900
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.147866362718609e-05,
-      "loss": 0.7334,
+      "loss": 0.8418,
       "step": 12901
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1477498703079706e-05,
-      "loss": 0.8699,
+      "loss": 0.8835,
       "step": 12902
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1476333758475127e-05,
-      "loss": 0.7764,
+      "loss": 0.8866,
       "step": 12903
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1475168793388512e-05,
-      "loss": 0.6601,
+      "loss": 0.7717,
       "step": 12904
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1474003807836019e-05,
-      "loss": 0.7902,
+      "loss": 0.8597,
       "step": 12905
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.147283880183382e-05,
-      "loss": 0.5818,
+      "loss": 0.7884,
       "step": 12906
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1471673775398067e-05,
-      "loss": 0.866,
+      "loss": 0.9429,
       "step": 12907
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1470508728544932e-05,
-      "loss": 0.7437,
+      "loss": 0.7608,
       "step": 12908
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1469343661290573e-05,
-      "loss": 0.6859,
+      "loss": 0.8914,
       "step": 12909
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1468178573651154e-05,
-      "loss": 0.7729,
+      "loss": 0.8041,
       "step": 12910
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1467013465642842e-05,
-      "loss": 0.6979,
+      "loss": 0.8123,
       "step": 12911
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1465848337281797e-05,
-      "loss": 0.758,
+      "loss": 0.8383,
       "step": 12912
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1464683188584189e-05,
-      "loss": 0.7722,
+      "loss": 0.7682,
       "step": 12913
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1463518019566182e-05,
-      "loss": 0.7047,
+      "loss": 0.7855,
       "step": 12914
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1462352830243935e-05,
-      "loss": 0.7912,
+      "loss": 0.8986,
       "step": 12915
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1461187620633618e-05,
-      "loss": 0.7533,
+      "loss": 0.8414,
       "step": 12916
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1460022390751397e-05,
-      "loss": 0.8715,
+      "loss": 0.9441,
       "step": 12917
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1458857140613434e-05,
-      "loss": 0.7511,
+      "loss": 0.9163,
       "step": 12918
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1457691870235903e-05,
-      "loss": 0.849,
+      "loss": 0.8913,
       "step": 12919
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1456526579634963e-05,
-      "loss": 0.7845,
+      "loss": 0.933,
       "step": 12920
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.145536126882678e-05,
-      "loss": 0.7243,
+      "loss": 0.8598,
       "step": 12921
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.145419593782753e-05,
-      "loss": 0.811,
+      "loss": 0.9176,
       "step": 12922
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1453030586653372e-05,
-      "loss": 0.776,
+      "loss": 0.8402,
       "step": 12923
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1451865215320477e-05,
-      "loss": 0.7409,
+      "loss": 0.8529,
       "step": 12924
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1450699823845011e-05,
-      "loss": 0.6254,
+      "loss": 0.8171,
       "step": 12925
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1449534412243143e-05,
-      "loss": 0.6416,
+      "loss": 0.7818,
       "step": 12926
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1448368980531043e-05,
-      "loss": 0.8496,
+      "loss": 0.9789,
       "step": 12927
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1447203528724877e-05,
-      "loss": 0.73,
+      "loss": 0.8557,
       "step": 12928
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1446038056840818e-05,
-      "loss": 0.6769,
+      "loss": 0.784,
       "step": 12929
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.144487256489503e-05,
-      "loss": 0.7604,
+      "loss": 0.9041,
       "step": 12930
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1443707052903683e-05,
-      "loss": 0.9366,
+      "loss": 0.9493,
       "step": 12931
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1442541520882955e-05,
-      "loss": 0.8775,
+      "loss": 0.9064,
       "step": 12932
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1441375968849006e-05,
-      "loss": 0.5878,
+      "loss": 0.8306,
       "step": 12933
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.144021039681801e-05,
-      "loss": 0.7598,
+      "loss": 0.858,
       "step": 12934
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1439044804806141e-05,
-      "loss": 0.8063,
+      "loss": 0.9254,
       "step": 12935
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1437879192829566e-05,
-      "loss": 0.766,
+      "loss": 0.8623,
       "step": 12936
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1436713560904456e-05,
-      "loss": 0.8248,
+      "loss": 0.9215,
       "step": 12937
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1435547909046985e-05,
-      "loss": 0.6523,
+      "loss": 0.8416,
       "step": 12938
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.143438223727332e-05,
-      "loss": 0.6914,
+      "loss": 0.8619,
       "step": 12939
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1433216545599642e-05,
-      "loss": 0.7235,
+      "loss": 0.8471,
       "step": 12940
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1432050834042115e-05,
-      "loss": 0.7349,
+      "loss": 0.9106,
       "step": 12941
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1430885102616913e-05,
-      "loss": 0.6615,
+      "loss": 0.8009,
       "step": 12942
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1429719351340214e-05,
-      "loss": 0.6882,
+      "loss": 0.7959,
       "step": 12943
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1428553580228186e-05,
-      "loss": 0.7271,
+      "loss": 0.851,
       "step": 12944
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1427387789297004e-05,
-      "loss": 0.5929,
+      "loss": 0.7926,
       "step": 12945
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1426221978562844e-05,
-      "loss": 0.8893,
+      "loss": 0.9182,
       "step": 12946
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1425056148041874e-05,
-      "loss": 0.7107,
+      "loss": 0.8605,
       "step": 12947
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.142389029775027e-05,
-      "loss": 0.7103,
+      "loss": 0.8508,
       "step": 12948
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1422724427704215e-05,
-      "loss": 0.8385,
+      "loss": 0.8903,
       "step": 12949
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.142155853791987e-05,
-      "loss": 0.7644,
+      "loss": 0.8745,
       "step": 12950
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1420392628413421e-05,
-      "loss": 0.7918,
+      "loss": 0.8735,
       "step": 12951
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1419226699201041e-05,
-      "loss": 0.9491,
+      "loss": 0.9675,
       "step": 12952
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.14180607502989e-05,
-      "loss": 0.6421,
+      "loss": 0.8239,
       "step": 12953
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1416894781723183e-05,
-      "loss": 0.8725,
+      "loss": 0.9369,
       "step": 12954
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1415728793490055e-05,
-      "loss": 0.7859,
+      "loss": 0.8875,
       "step": 12955
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1414562785615703e-05,
-      "loss": 0.7669,
+      "loss": 0.869,
       "step": 12956
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.14133967581163e-05,
-      "loss": 0.5993,
+      "loss": 0.7328,
       "step": 12957
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1412230711008019e-05,
-      "loss": 0.7727,
+      "loss": 0.8749,
       "step": 12958
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1411064644307044e-05,
-      "loss": 0.7617,
+      "loss": 0.8561,
       "step": 12959
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1409898558029548e-05,
-      "loss": 0.6973,
+      "loss": 0.8107,
       "step": 12960
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1408732452191709e-05,
-      "loss": 0.7369,
+      "loss": 0.7929,
       "step": 12961
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1407566326809707e-05,
-      "loss": 0.8232,
+      "loss": 0.9075,
       "step": 12962
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1406400181899719e-05,
-      "loss": 0.7807,
+      "loss": 0.8958,
       "step": 12963
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1405234017477924e-05,
-      "loss": 0.8794,
+      "loss": 0.92,
       "step": 12964
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1404067833560501e-05,
-      "loss": 0.7387,
+      "loss": 0.8897,
       "step": 12965
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1402901630163627e-05,
-      "loss": 0.6887,
+      "loss": 0.8562,
       "step": 12966
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1401735407303484e-05,
-      "loss": 0.6708,
+      "loss": 0.7372,
       "step": 12967
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1400569164996254e-05,
-      "loss": 0.6592,
+      "loss": 0.8252,
       "step": 12968
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.139940290325811e-05,
-      "loss": 0.7454,
+      "loss": 0.9044,
       "step": 12969
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.139823662210524e-05,
-      "loss": 0.7233,
+      "loss": 0.7852,
       "step": 12970
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1397070321553816e-05,
-      "loss": 0.6709,
+      "loss": 0.7905,
       "step": 12971
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1395904001620026e-05,
-      "loss": 0.7672,
+      "loss": 0.8876,
       "step": 12972
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.139473766232005e-05,
-      "loss": 0.8428,
+      "loss": 0.9584,
       "step": 12973
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1393571303670064e-05,
-      "loss": 0.7841,
+      "loss": 0.8881,
       "step": 12974
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1392404925686255e-05,
-      "loss": 0.7942,
+      "loss": 0.9351,
       "step": 12975
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1391238528384807e-05,
-      "loss": 0.8204,
+      "loss": 0.859,
       "step": 12976
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1390072111781894e-05,
-      "loss": 0.649,
+      "loss": 0.7415,
       "step": 12977
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1388905675893705e-05,
-      "loss": 1.0042,
+      "loss": 1.0375,
       "step": 12978
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.138773922073642e-05,
-      "loss": 0.705,
+      "loss": 0.8106,
       "step": 12979
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1386572746326219e-05,
-      "loss": 0.7048,
+      "loss": 0.8273,
       "step": 12980
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1385406252679292e-05,
-      "loss": 0.7122,
+      "loss": 0.8579,
       "step": 12981
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1384239739811815e-05,
-      "loss": 0.7614,
+      "loss": 0.8454,
       "step": 12982
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1383073207739983e-05,
-      "loss": 0.6537,
+      "loss": 0.8211,
       "step": 12983
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1381906656479967e-05,
-      "loss": 0.7551,
+      "loss": 0.8694,
       "step": 12984
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1380740086047958e-05,
-      "loss": 0.7444,
+      "loss": 0.8915,
       "step": 12985
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1379573496460139e-05,
-      "loss": 0.8125,
+      "loss": 0.8878,
       "step": 12986
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1378406887732693e-05,
-      "loss": 0.7265,
+      "loss": 0.8254,
       "step": 12987
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1377240259881812e-05,
-      "loss": 0.6897,
+      "loss": 0.8538,
       "step": 12988
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1376073612923675e-05,
-      "loss": 0.7642,
+      "loss": 0.9018,
       "step": 12989
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1374906946874467e-05,
-      "loss": 0.5273,
+      "loss": 0.796,
       "step": 12990
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1373740261750377e-05,
-      "loss": 0.8256,
+      "loss": 0.8822,
       "step": 12991
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.137257355756759e-05,
-      "loss": 0.7397,
+      "loss": 0.8838,
       "step": 12992
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1371406834342289e-05,
-      "loss": 0.6372,
+      "loss": 0.8377,
       "step": 12993
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.137024009209067e-05,
-      "loss": 0.8517,
+      "loss": 0.8607,
       "step": 12994
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1369073330828908e-05,
-      "loss": 0.8632,
+      "loss": 0.9009,
       "step": 12995
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.13679065505732e-05,
-      "loss": 0.7959,
+      "loss": 0.8907,
       "step": 12996
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1366739751339728e-05,
-      "loss": 0.7311,
+      "loss": 0.9299,
       "step": 12997
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.136557293314468e-05,
-      "loss": 0.8295,
+      "loss": 0.9645,
       "step": 12998
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1364406096004248e-05,
-      "loss": 0.7481,
+      "loss": 0.8719,
       "step": 12999
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1363239239934617e-05,
-      "loss": 0.6627,
+      "loss": 0.797,
       "step": 13000
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1362072364951971e-05,
-      "loss": 0.8398,
+      "loss": 0.9343,
       "step": 13001
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.136090547107251e-05,
-      "loss": 0.7216,
+      "loss": 0.8976,
       "step": 13002
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1359738558312412e-05,
-      "loss": 0.8596,
+      "loss": 0.9397,
       "step": 13003
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1358571626687874e-05,
-      "loss": 0.7509,
+      "loss": 0.7229,
       "step": 13004
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.135740467621508e-05,
-      "loss": 0.8413,
+      "loss": 0.9658,
       "step": 13005
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.135623770691022e-05,
-      "loss": 0.7875,
+      "loss": 0.842,
       "step": 13006
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.135507071878949e-05,
-      "loss": 0.6953,
+      "loss": 0.9071,
       "step": 13007
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1353903711869074e-05,
-      "loss": 0.6751,
+      "loss": 0.7876,
       "step": 13008
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1352736686165164e-05,
-      "loss": 0.6905,
+      "loss": 0.8311,
       "step": 13009
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1351569641693956e-05,
-      "loss": 0.8141,
+      "loss": 0.8897,
       "step": 13010
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1350402578471632e-05,
-      "loss": 0.7561,
+      "loss": 0.8703,
       "step": 13011
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.134923549651439e-05,
-      "loss": 0.7684,
+      "loss": 0.8928,
       "step": 13012
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1348068395838423e-05,
-      "loss": 0.7808,
+      "loss": 0.8937,
       "step": 13013
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1346901276459913e-05,
-      "loss": 0.7811,
+      "loss": 0.8545,
       "step": 13014
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1345734138395063e-05,
-      "loss": 0.7314,
+      "loss": 0.813,
       "step": 13015
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1344566981660064e-05,
-      "loss": 0.7839,
+      "loss": 0.933,
       "step": 13016
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.13433998062711e-05,
-      "loss": 0.6032,
+      "loss": 0.8391,
       "step": 13017
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1342232612244373e-05,
-      "loss": 0.7045,
+      "loss": 0.8418,
       "step": 13018
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1341065399596073e-05,
-      "loss": 0.5618,
+      "loss": 0.7174,
       "step": 13019
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1339898168342392e-05,
-      "loss": 0.7574,
+      "loss": 0.9032,
       "step": 13020
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1338730918499527e-05,
-      "loss": 0.7652,
+      "loss": 0.8335,
       "step": 13021
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1337563650083667e-05,
-      "loss": 0.7924,
+      "loss": 0.9052,
       "step": 13022
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1336396363111013e-05,
-      "loss": 0.6559,
+      "loss": 0.775,
       "step": 13023
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1335229057597755e-05,
-      "loss": 0.7332,
+      "loss": 0.849,
       "step": 13024
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1334061733560082e-05,
-      "loss": 0.7779,
+      "loss": 0.9068,
       "step": 13025
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1332894391014201e-05,
-      "loss": 0.8833,
+      "loss": 0.9091,
       "step": 13026
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.13317270299763e-05,
-      "loss": 0.8085,
+      "loss": 0.9421,
       "step": 13027
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1330559650462576e-05,
-      "loss": 0.7557,
+      "loss": 0.8328,
       "step": 13028
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1329392252489226e-05,
-      "loss": 0.8147,
+      "loss": 0.8574,
       "step": 13029
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1328224836072441e-05,
-      "loss": 0.6661,
+      "loss": 0.7592,
       "step": 13030
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1327057401228426e-05,
-      "loss": 0.7973,
+      "loss": 0.9039,
       "step": 13031
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1325889947973368e-05,
-      "loss": 0.7829,
+      "loss": 0.8555,
       "step": 13032
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1324722476323469e-05,
-      "loss": 0.8085,
+      "loss": 0.8954,
       "step": 13033
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1323554986294927e-05,
-      "loss": 0.7516,
+      "loss": 0.8301,
       "step": 13034
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1322387477903937e-05,
-      "loss": 0.6836,
+      "loss": 0.7977,
       "step": 13035
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1321219951166695e-05,
-      "loss": 0.7333,
+      "loss": 0.8625,
       "step": 13036
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1320052406099402e-05,
-      "loss": 0.8994,
+      "loss": 0.9809,
       "step": 13037
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1318884842718253e-05,
-      "loss": 0.9205,
+      "loss": 0.955,
       "step": 13038
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1317717261039452e-05,
-      "loss": 0.7536,
+      "loss": 0.8437,
       "step": 13039
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1316549661079193e-05,
-      "loss": 0.7872,
+      "loss": 0.8729,
       "step": 13040
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1315382042853673e-05,
-      "loss": 0.7025,
+      "loss": 0.7863,
       "step": 13041
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1314214406379096e-05,
-      "loss": 0.7308,
+      "loss": 0.8809,
       "step": 13042
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1313046751671655e-05,
-      "loss": 0.7078,
+      "loss": 0.8408,
       "step": 13043
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1311879078747555e-05,
-      "loss": 0.6229,
+      "loss": 0.8344,
       "step": 13044
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1310711387622997e-05,
-      "loss": 0.7158,
+      "loss": 0.8285,
       "step": 13045
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1309543678314174e-05,
-      "loss": 0.9352,
+      "loss": 1.0545,
       "step": 13046
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1308375950837294e-05,
-      "loss": 0.7471,
+      "loss": 0.8597,
       "step": 13047
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1307208205208554e-05,
-      "loss": 0.7731,
+      "loss": 0.8592,
       "step": 13048
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.130604044144415e-05,
-      "loss": 0.8094,
+      "loss": 0.877,
       "step": 13049
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1304872659560291e-05,
-      "loss": 0.6325,
+      "loss": 0.7633,
       "step": 13050
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1303704859573178e-05,
-      "loss": 0.7285,
+      "loss": 0.8556,
       "step": 13051
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1302537041499004e-05,
-      "loss": 0.6143,
+      "loss": 0.8103,
       "step": 13052
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.130136920535398e-05,
-      "loss": 0.6463,
+      "loss": 0.8382,
       "step": 13053
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1300201351154303e-05,
-      "loss": 0.7466,
+      "loss": 0.8478,
       "step": 13054
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1299033478916179e-05,
-      "loss": 0.668,
+      "loss": 0.8533,
       "step": 13055
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1297865588655808e-05,
-      "loss": 0.7415,
+      "loss": 0.876,
       "step": 13056
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1296697680389393e-05,
-      "loss": 0.7424,
+      "loss": 0.8031,
       "step": 13057
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1295529754133136e-05,
-      "loss": 0.7552,
+      "loss": 0.8634,
       "step": 13058
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1294361809903244e-05,
-      "loss": 0.6765,
+      "loss": 0.8622,
       "step": 13059
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.129319384771592e-05,
-      "loss": 0.7355,
+      "loss": 0.87,
       "step": 13060
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.1292025867587364e-05,
-      "loss": 0.816,
+      "loss": 0.8883,
       "step": 13061
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1290857869533783e-05,
-      "loss": 0.7652,
+      "loss": 0.8774,
       "step": 13062
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1289689853571379e-05,
-      "loss": 0.7026,
+      "loss": 0.8336,
       "step": 13063
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1288521819716358e-05,
-      "loss": 0.7576,
+      "loss": 0.8492,
       "step": 13064
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1287353767984925e-05,
-      "loss": 0.8016,
+      "loss": 0.9505,
       "step": 13065
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1286185698393286e-05,
-      "loss": 0.7229,
+      "loss": 0.8405,
       "step": 13066
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1285017610957644e-05,
-      "loss": 0.8042,
+      "loss": 0.9377,
       "step": 13067
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1283849505694208e-05,
-      "loss": 0.8709,
+      "loss": 0.958,
       "step": 13068
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1282681382619183e-05,
-      "loss": 0.7557,
+      "loss": 0.8571,
       "step": 13069
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1281513241748767e-05,
-      "loss": 0.7346,
+      "loss": 0.821,
       "step": 13070
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.128034508309918e-05,
-      "loss": 0.6732,
+      "loss": 0.7838,
       "step": 13071
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.127917690668662e-05,
-      "loss": 0.7506,
+      "loss": 0.9037,
       "step": 13072
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1278008712527291e-05,
-      "loss": 0.7586,
+      "loss": 0.8674,
       "step": 13073
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1276840500637409e-05,
-      "loss": 0.7805,
+      "loss": 0.8792,
       "step": 13074
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1275672271033175e-05,
-      "loss": 0.7217,
+      "loss": 0.8258,
       "step": 13075
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1274504023730796e-05,
-      "loss": 0.7256,
+      "loss": 0.876,
       "step": 13076
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1273335758746486e-05,
-      "loss": 0.7098,
+      "loss": 0.8788,
       "step": 13077
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1272167476096447e-05,
-      "loss": 0.7283,
+      "loss": 0.8618,
       "step": 13078
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.127099917579689e-05,
-      "loss": 0.6893,
+      "loss": 0.7653,
       "step": 13079
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1269830857864024e-05,
-      "loss": 0.7725,
+      "loss": 0.8545,
       "step": 13080
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1268662522314053e-05,
-      "loss": 0.6573,
+      "loss": 0.8293,
       "step": 13081
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1267494169163193e-05,
-      "loss": 0.5796,
+      "loss": 0.719,
       "step": 13082
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1266325798427646e-05,
-      "loss": 0.6946,
+      "loss": 0.9186,
       "step": 13083
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1265157410123628e-05,
-      "loss": 0.7636,
+      "loss": 0.8673,
       "step": 13084
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1263989004267346e-05,
-      "loss": 0.6922,
+      "loss": 0.8447,
       "step": 13085
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1262820580875007e-05,
-      "loss": 0.7655,
+      "loss": 0.805,
       "step": 13086
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.126165213996283e-05,
-      "loss": 0.7543,
+      "loss": 0.8797,
       "step": 13087
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1260483681547014e-05,
-      "loss": 0.6689,
+      "loss": 0.8363,
       "step": 13088
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1259315205643775e-05,
-      "loss": 0.7659,
+      "loss": 0.8978,
       "step": 13089
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1258146712269328e-05,
-      "loss": 0.5536,
+      "loss": 0.7266,
       "step": 13090
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1256978201439879e-05,
-      "loss": 0.9058,
+      "loss": 0.9004,
       "step": 13091
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.125580967317164e-05,
-      "loss": 0.7816,
+      "loss": 0.9664,
       "step": 13092
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1254641127480825e-05,
-      "loss": 0.7324,
+      "loss": 0.8195,
       "step": 13093
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1253472564383643e-05,
-      "loss": 0.7332,
+      "loss": 0.7989,
       "step": 13094
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.125230398389631e-05,
-      "loss": 0.7719,
+      "loss": 0.9479,
       "step": 13095
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1251135386035032e-05,
-      "loss": 0.6596,
+      "loss": 0.7561,
       "step": 13096
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1249966770816028e-05,
-      "loss": 0.7361,
+      "loss": 0.8962,
       "step": 13097
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.124879813825551e-05,
-      "loss": 0.7268,
+      "loss": 0.862,
       "step": 13098
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1247629488369687e-05,
-      "loss": 0.6648,
+      "loss": 0.8735,
       "step": 13099
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1246460821174778e-05,
-      "loss": 0.8566,
+      "loss": 0.8633,
       "step": 13100
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1245292136686991e-05,
-      "loss": 0.8253,
+      "loss": 0.9275,
       "step": 13101
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1244123434922542e-05,
-      "loss": 0.7605,
+      "loss": 0.9092,
       "step": 13102
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1242954715897649e-05,
-      "loss": 0.8685,
+      "loss": 0.8809,
       "step": 13103
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1241785979628522e-05,
-      "loss": 0.8214,
+      "loss": 0.956,
       "step": 13104
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1240617226131371e-05,
-      "loss": 0.7294,
+      "loss": 0.7713,
       "step": 13105
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1239448455422422e-05,
-      "loss": 0.7827,
+      "loss": 0.9451,
       "step": 13106
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1238279667517883e-05,
-      "loss": 0.7294,
+      "loss": 0.8591,
       "step": 13107
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1237110862433968e-05,
-      "loss": 0.6394,
+      "loss": 0.7519,
       "step": 13108
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1235942040186897e-05,
-      "loss": 0.8786,
+      "loss": 1.0652,
       "step": 13109
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1234773200792883e-05,
-      "loss": 0.6339,
+      "loss": 0.8308,
       "step": 13110
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1233604344268145e-05,
-      "loss": 0.7005,
+      "loss": 0.8078,
       "step": 13111
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1232435470628895e-05,
-      "loss": 0.7956,
+      "loss": 0.9285,
       "step": 13112
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1231266579891352e-05,
-      "loss": 0.7089,
+      "loss": 0.7977,
       "step": 13113
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1230097672071732e-05,
-      "loss": 0.6317,
+      "loss": 0.8297,
       "step": 13114
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1228928747186254e-05,
-      "loss": 0.7295,
+      "loss": 0.8327,
       "step": 13115
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1227759805251128e-05,
-      "loss": 0.6955,
+      "loss": 0.87,
       "step": 13116
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1226590846282584e-05,
-      "loss": 0.7108,
+      "loss": 0.8528,
       "step": 13117
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1225421870296826e-05,
-      "loss": 0.8085,
+      "loss": 0.8486,
       "step": 13118
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1224252877310083e-05,
-      "loss": 0.7384,
+      "loss": 0.8628,
       "step": 13119
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1223083867338564e-05,
-      "loss": 0.671,
+      "loss": 0.7653,
       "step": 13120
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1221914840398494e-05,
-      "loss": 0.7238,
+      "loss": 0.8232,
       "step": 13121
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1220745796506089e-05,
-      "loss": 0.6774,
+      "loss": 0.8258,
       "step": 13122
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1219576735677569e-05,
-      "loss": 0.7618,
+      "loss": 0.8726,
       "step": 13123
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1218407657929149e-05,
-      "loss": 0.8473,
+      "loss": 0.9632,
       "step": 13124
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1217238563277053e-05,
-      "loss": 0.677,
+      "loss": 0.8063,
       "step": 13125
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1216069451737498e-05,
-      "loss": 0.7944,
+      "loss": 0.8865,
       "step": 13126
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1214900323326708e-05,
-      "loss": 0.9465,
+      "loss": 0.9735,
       "step": 13127
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1213731178060898e-05,
-      "loss": 0.7936,
+      "loss": 0.8941,
       "step": 13128
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.121256201595629e-05,
-      "loss": 0.8266,
+      "loss": 0.9682,
       "step": 13129
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1211392837029104e-05,
-      "loss": 0.7948,
+      "loss": 0.9076,
       "step": 13130
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.121022364129556e-05,
-      "loss": 0.6178,
+      "loss": 0.7927,
       "step": 13131
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1209054428771881e-05,
-      "loss": 0.8149,
+      "loss": 0.8933,
       "step": 13132
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.120788519947429e-05,
-      "loss": 0.8037,
+      "loss": 0.9585,
       "step": 13133
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1206715953419003e-05,
-      "loss": 0.8228,
+      "loss": 0.8425,
       "step": 13134
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1205546690622244e-05,
-      "loss": 0.8449,
+      "loss": 0.8693,
       "step": 13135
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1204377411100237e-05,
-      "loss": 0.7848,
+      "loss": 0.8915,
       "step": 13136
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1203208114869199e-05,
-      "loss": 0.7822,
+      "loss": 0.9095,
       "step": 13137
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.120203880194536e-05,
-      "loss": 0.7807,
+      "loss": 0.8472,
       "step": 13138
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1200869472344933e-05,
-      "loss": 0.6509,
+      "loss": 0.8253,
       "step": 13139
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.119970012608415e-05,
-      "loss": 0.731,
+      "loss": 0.8341,
       "step": 13140
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.119853076317923e-05,
-      "loss": 0.6971,
+      "loss": 0.8528,
       "step": 13141
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1197361383646393e-05,
-      "loss": 0.7843,
+      "loss": 0.8409,
       "step": 13142
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1196191987501866e-05,
-      "loss": 0.6517,
+      "loss": 0.77,
       "step": 13143
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1195022574761874e-05,
-      "loss": 0.5897,
+      "loss": 0.6737,
       "step": 13144
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1193853145442638e-05,
-      "loss": 0.9235,
+      "loss": 0.9759,
       "step": 13145
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1192683699560385e-05,
-      "loss": 0.7091,
+      "loss": 0.8474,
       "step": 13146
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1191514237131339e-05,
-      "loss": 0.8215,
+      "loss": 0.9331,
       "step": 13147
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.119034475817172e-05,
-      "loss": 0.7086,
+      "loss": 0.8588,
       "step": 13148
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1189175262697758e-05,
-      "loss": 0.8862,
+      "loss": 0.9668,
       "step": 13149
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1188005750725674e-05,
-      "loss": 0.8102,
+      "loss": 0.9123,
       "step": 13150
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1186836222271699e-05,
-      "loss": 0.7461,
+      "loss": 0.9346,
       "step": 13151
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1185666677352054e-05,
-      "loss": 0.7321,
+      "loss": 0.8262,
       "step": 13152
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1184497115982966e-05,
-      "loss": 0.6511,
+      "loss": 0.853,
       "step": 13153
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1183327538180661e-05,
-      "loss": 0.6296,
+      "loss": 0.7449,
       "step": 13154
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1182157943961367e-05,
-      "loss": 0.7381,
+      "loss": 0.8364,
       "step": 13155
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.118098833334131e-05,
-      "loss": 0.6387,
+      "loss": 0.8332,
       "step": 13156
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1179818706336714e-05,
-      "loss": 0.8018,
+      "loss": 0.832,
       "step": 13157
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1178649062963805e-05,
-      "loss": 0.7418,
+      "loss": 0.8692,
       "step": 13158
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1177479403238813e-05,
-      "loss": 0.7744,
+      "loss": 0.8334,
       "step": 13159
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1176309727177968e-05,
-      "loss": 0.6809,
+      "loss": 0.8297,
       "step": 13160
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1175140034797491e-05,
-      "loss": 0.6887,
+      "loss": 0.8738,
       "step": 13161
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1173970326113617e-05,
-      "loss": 0.7492,
+      "loss": 0.852,
       "step": 13162
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1172800601142568e-05,
-      "loss": 0.7303,
+      "loss": 0.8517,
       "step": 13163
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1171630859900572e-05,
-      "loss": 0.7255,
+      "loss": 0.7983,
       "step": 13164
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1170461102403865e-05,
-      "loss": 0.822,
+      "loss": 0.9332,
       "step": 13165
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1169291328668667e-05,
-      "loss": 0.9021,
+      "loss": 0.947,
       "step": 13166
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1168121538711214e-05,
-      "loss": 0.7576,
+      "loss": 0.8753,
       "step": 13167
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1166951732547731e-05,
-      "loss": 0.8748,
+      "loss": 0.9266,
       "step": 13168
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1165781910194446e-05,
-      "loss": 0.6772,
+      "loss": 0.8746,
       "step": 13169
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1164612071667596e-05,
-      "loss": 0.7293,
+      "loss": 0.9067,
       "step": 13170
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1163442216983402e-05,
-      "loss": 0.6615,
+      "loss": 0.789,
       "step": 13171
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1162272346158098e-05,
-      "loss": 0.6532,
+      "loss": 0.7867,
       "step": 13172
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1161102459207916e-05,
-      "loss": 0.9255,
+      "loss": 0.9063,
       "step": 13173
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1159932556149084e-05,
-      "loss": 0.8178,
+      "loss": 0.9092,
       "step": 13174
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1158762636997835e-05,
-      "loss": 0.8504,
+      "loss": 1.0043,
       "step": 13175
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1157592701770396e-05,
-      "loss": 0.8163,
+      "loss": 0.9237,
       "step": 13176
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1156422750483003e-05,
-      "loss": 0.7752,
+      "loss": 0.8749,
       "step": 13177
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1155252783151884e-05,
-      "loss": 0.7526,
+      "loss": 0.8418,
       "step": 13178
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.115408279979327e-05,
-      "loss": 0.7343,
+      "loss": 0.8532,
       "step": 13179
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1152912800423396e-05,
-      "loss": 0.7685,
+      "loss": 0.8644,
       "step": 13180
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1151742785058493e-05,
-      "loss": 0.7578,
+      "loss": 0.8798,
       "step": 13181
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1150572753714792e-05,
-      "loss": 0.8453,
+      "loss": 0.9092,
       "step": 13182
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1149402706408528e-05,
-      "loss": 0.8089,
+      "loss": 0.9347,
       "step": 13183
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1148232643155932e-05,
-      "loss": 0.6929,
+      "loss": 0.8238,
       "step": 13184
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1147062563973235e-05,
-      "loss": 0.7227,
+      "loss": 0.7606,
       "step": 13185
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1145892468876673e-05,
-      "loss": 0.7042,
+      "loss": 0.8329,
       "step": 13186
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1144722357882483e-05,
-      "loss": 0.747,
+      "loss": 0.7586,
       "step": 13187
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.114355223100689e-05,
-      "loss": 0.7327,
+      "loss": 0.8415,
       "step": 13188
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1142382088266133e-05,
-      "loss": 0.7118,
+      "loss": 0.7921,
       "step": 13189
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1141211929676447e-05,
-      "loss": 0.8509,
+      "loss": 0.9323,
       "step": 13190
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.114004175525406e-05,
-      "loss": 0.865,
+      "loss": 0.8782,
       "step": 13191
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1138871565015216e-05,
-      "loss": 0.7621,
+      "loss": 0.8685,
       "step": 13192
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1137701358976142e-05,
-      "loss": 0.7283,
+      "loss": 0.8711,
       "step": 13193
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1136531137153082e-05,
-      "loss": 0.7305,
+      "loss": 0.856,
       "step": 13194
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.113536089956226e-05,
-      "loss": 0.8052,
+      "loss": 0.8518,
       "step": 13195
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1134190646219916e-05,
-      "loss": 0.7801,
+      "loss": 0.8417,
       "step": 13196
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1133020377142287e-05,
-      "loss": 0.7629,
+      "loss": 0.916,
       "step": 13197
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1131850092345606e-05,
-      "loss": 0.7273,
+      "loss": 0.8376,
       "step": 13198
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1130679791846112e-05,
-      "loss": 0.8024,
+      "loss": 0.854,
       "step": 13199
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1129509475660042e-05,
-      "loss": 0.7483,
+      "loss": 0.8722,
       "step": 13200
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1128339143803628e-05,
-      "loss": 0.5746,
+      "loss": 0.8014,
       "step": 13201
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.112716879629311e-05,
-      "loss": 0.6301,
+      "loss": 0.7096,
       "step": 13202
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1125998433144727e-05,
-      "loss": 0.7672,
+      "loss": 0.8021,
       "step": 13203
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.112482805437471e-05,
-      "loss": 0.8697,
+      "loss": 0.9572,
       "step": 13204
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1123657659999301e-05,
-      "loss": 0.766,
+      "loss": 0.7872,
       "step": 13205
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1122487250034738e-05,
-      "loss": 0.7353,
+      "loss": 0.8601,
       "step": 13206
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1121316824497253e-05,
-      "loss": 0.8702,
+      "loss": 0.9181,
       "step": 13207
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.112014638340309e-05,
-      "loss": 0.8444,
+      "loss": 0.9799,
       "step": 13208
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1118975926768483e-05,
-      "loss": 0.794,
+      "loss": 0.8188,
       "step": 13209
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1117805454609675e-05,
-      "loss": 0.9085,
+      "loss": 0.9857,
       "step": 13210
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1116634966942903e-05,
-      "loss": 0.8346,
+      "loss": 0.928,
       "step": 13211
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1115464463784402e-05,
-      "loss": 0.8819,
+      "loss": 1.0335,
       "step": 13212
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.111429394515042e-05,
-      "loss": 0.7843,
+      "loss": 0.8781,
       "step": 13213
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1113123411057185e-05,
-      "loss": 0.8891,
+      "loss": 0.9678,
       "step": 13214
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1111952861520944e-05,
-      "loss": 0.7421,
+      "loss": 0.9089,
       "step": 13215
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1110782296557935e-05,
-      "loss": 0.7335,
+      "loss": 0.8688,
       "step": 13216
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1109611716184397e-05,
-      "loss": 0.7207,
+      "loss": 0.88,
       "step": 13217
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1108441120416571e-05,
-      "loss": 0.8967,
+      "loss": 0.9415,
       "step": 13218
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1107270509270698e-05,
-      "loss": 0.782,
+      "loss": 0.878,
       "step": 13219
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1106099882763015e-05,
-      "loss": 0.7203,
+      "loss": 0.797,
       "step": 13220
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1104929240909767e-05,
-      "loss": 0.7347,
+      "loss": 0.8593,
       "step": 13221
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1103758583727193e-05,
-      "loss": 0.5972,
+      "loss": 0.7889,
       "step": 13222
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1102587911231535e-05,
-      "loss": 0.7617,
+      "loss": 0.8715,
       "step": 13223
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1101417223439035e-05,
-      "loss": 0.8004,
+      "loss": 0.9086,
       "step": 13224
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1100246520365935e-05,
-      "loss": 0.8144,
+      "loss": 0.9541,
       "step": 13225
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1099075802028472e-05,
-      "loss": 0.8055,
+      "loss": 0.8696,
       "step": 13226
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1097905068442894e-05,
-      "loss": 0.777,
+      "loss": 0.8782,
       "step": 13227
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.109673431962544e-05,
-      "loss": 0.7028,
+      "loss": 0.8945,
       "step": 13228
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1095563555592352e-05,
-      "loss": 0.7172,
+      "loss": 0.8119,
       "step": 13229
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1094392776359877e-05,
-      "loss": 0.8953,
+      "loss": 1.0026,
       "step": 13230
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1093221981944254e-05,
-      "loss": 0.7081,
+      "loss": 0.772,
       "step": 13231
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.109205117236173e-05,
-      "loss": 0.6918,
+      "loss": 0.8156,
       "step": 13232
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1090880347628538e-05,
-      "loss": 0.6803,
+      "loss": 0.7606,
       "step": 13233
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1089709507760935e-05,
-      "loss": 0.8818,
+      "loss": 0.9407,
       "step": 13234
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1088538652775158e-05,
-      "loss": 0.7391,
+      "loss": 0.8398,
       "step": 13235
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1087367782687449e-05,
-      "loss": 0.7839,
+      "loss": 0.862,
       "step": 13236
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1086196897514057e-05,
-      "loss": 0.6695,
+      "loss": 0.8655,
       "step": 13237
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1085025997271222e-05,
-      "loss": 0.6595,
+      "loss": 0.8464,
       "step": 13238
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1083855081975192e-05,
-      "loss": 0.5949,
+      "loss": 0.7052,
       "step": 13239
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.108268415164221e-05,
-      "loss": 0.7978,
+      "loss": 0.8533,
       "step": 13240
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1081513206288521e-05,
-      "loss": 0.6066,
+      "loss": 0.8097,
       "step": 13241
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1080342245930373e-05,
-      "loss": 0.7611,
+      "loss": 0.865,
       "step": 13242
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1079171270584009e-05,
-      "loss": 0.7381,
+      "loss": 0.8709,
       "step": 13243
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1078000280265673e-05,
-      "loss": 0.6959,
+      "loss": 0.8327,
       "step": 13244
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1076829274991614e-05,
-      "loss": 0.6206,
+      "loss": 0.7929,
       "step": 13245
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1075658254778076e-05,
-      "loss": 0.7758,
+      "loss": 0.8651,
       "step": 13246
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1074487219641306e-05,
-      "loss": 0.6479,
+      "loss": 0.7714,
       "step": 13247
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.107331616959755e-05,
-      "loss": 0.7401,
+      "loss": 0.8314,
       "step": 13248
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1072145104663052e-05,
-      "loss": 0.8396,
+      "loss": 0.95,
       "step": 13249
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1070974024854066e-05,
-      "loss": 0.7537,
+      "loss": 0.9054,
       "step": 13250
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1069802930186837e-05,
-      "loss": 0.8102,
+      "loss": 0.8452,
       "step": 13251
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1068631820677606e-05,
-      "loss": 0.6647,
+      "loss": 0.8067,
       "step": 13252
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1067460696342627e-05,
-      "loss": 0.8766,
+      "loss": 0.9564,
       "step": 13253
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1066289557198142e-05,
-      "loss": 0.8491,
+      "loss": 0.8675,
       "step": 13254
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1065118403260407e-05,
-      "loss": 0.8205,
+      "loss": 0.9252,
       "step": 13255
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1063947234545662e-05,
-      "loss": 0.7962,
+      "loss": 0.8406,
       "step": 13256
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1062776051070159e-05,
-      "loss": 0.6945,
+      "loss": 0.7858,
       "step": 13257
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1061604852850148e-05,
-      "loss": 0.8049,
+      "loss": 0.872,
       "step": 13258
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1060433639901874e-05,
-      "loss": 0.6898,
+      "loss": 0.8196,
       "step": 13259
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1059262412241586e-05,
-      "loss": 0.7186,
+      "loss": 0.8099,
       "step": 13260
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1058091169885538e-05,
-      "loss": 0.7442,
+      "loss": 0.8755,
       "step": 13261
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1056919912849977e-05,
-      "loss": 0.8465,
+      "loss": 0.8666,
       "step": 13262
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.105574864115115e-05,
-      "loss": 0.7877,
+      "loss": 0.7599,
       "step": 13263
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.105457735480531e-05,
-      "loss": 0.7984,
+      "loss": 0.9293,
       "step": 13264
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1053406053828701e-05,
-      "loss": 0.7378,
+      "loss": 0.8694,
       "step": 13265
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1052234738237581e-05,
-      "loss": 0.8148,
+      "loss": 0.9201,
       "step": 13266
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1051063408048197e-05,
-      "loss": 0.7562,
+      "loss": 0.9103,
       "step": 13267
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1049892063276799e-05,
-      "loss": 0.7171,
+      "loss": 0.9069,
       "step": 13268
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1048720703939639e-05,
-      "loss": 0.6758,
+      "loss": 0.7644,
       "step": 13269
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1047549330052967e-05,
-      "loss": 0.6894,
+      "loss": 0.8088,
       "step": 13270
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1046377941633034e-05,
-      "loss": 0.7307,
+      "loss": 0.8886,
       "step": 13271
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1045206538696094e-05,
-      "loss": 0.6491,
+      "loss": 0.8171,
       "step": 13272
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1044035121258391e-05,
-      "loss": 0.6655,
+      "loss": 0.8121,
       "step": 13273
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1042863689336188e-05,
-      "loss": 0.7361,
+      "loss": 0.7771,
       "step": 13274
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1041692242945727e-05,
-      "loss": 0.8121,
+      "loss": 0.8507,
       "step": 13275
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1040520782103264e-05,
-      "loss": 0.7243,
+      "loss": 0.8603,
       "step": 13276
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1039349306825053e-05,
-      "loss": 0.6947,
+      "loss": 0.8593,
       "step": 13277
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1038177817127345e-05,
-      "loss": 0.8709,
+      "loss": 0.8741,
       "step": 13278
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.103700631302639e-05,
-      "loss": 0.7575,
+      "loss": 0.8051,
       "step": 13279
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1035834794538446e-05,
-      "loss": 0.6811,
+      "loss": 0.8042,
       "step": 13280
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1034663261679763e-05,
-      "loss": 0.6409,
+      "loss": 0.7672,
       "step": 13281
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1033491714466598e-05,
-      "loss": 0.7355,
+      "loss": 0.8265,
       "step": 13282
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.10323201529152e-05,
-      "loss": 0.7142,
+      "loss": 0.8392,
       "step": 13283
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1031148577041822e-05,
-      "loss": 0.6916,
+      "loss": 0.8415,
       "step": 13284
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1029976986862722e-05,
-      "loss": 0.8355,
+      "loss": 0.8693,
       "step": 13285
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1028805382394156e-05,
-      "loss": 0.6851,
+      "loss": 0.7284,
       "step": 13286
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.102763376365237e-05,
-      "loss": 0.8171,
+      "loss": 0.8717,
       "step": 13287
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1026462130653627e-05,
-      "loss": 0.7712,
+      "loss": 0.9371,
       "step": 13288
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.102529048341418e-05,
-      "loss": 0.648,
+      "loss": 0.824,
       "step": 13289
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1024118821950278e-05,
-      "loss": 0.7119,
+      "loss": 0.8672,
       "step": 13290
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1022947146278185e-05,
-      "loss": 0.7764,
+      "loss": 0.8709,
       "step": 13291
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1021775456414146e-05,
-      "loss": 0.8115,
+      "loss": 0.9066,
       "step": 13292
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1020603752374426e-05,
-      "loss": 0.7833,
+      "loss": 0.8745,
       "step": 13293
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1019432034175275e-05,
-      "loss": 0.822,
+      "loss": 0.9666,
       "step": 13294
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1018260301832952e-05,
-      "loss": 0.7788,
+      "loss": 0.8608,
       "step": 13295
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1017088555363713e-05,
-      "loss": 0.7552,
+      "loss": 0.8437,
       "step": 13296
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1015916794783809e-05,
-      "loss": 0.7755,
+      "loss": 0.8731,
       "step": 13297
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1014745020109505e-05,
-      "loss": 0.7172,
+      "loss": 0.84,
       "step": 13298
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1013573231357053e-05,
-      "loss": 0.6826,
+      "loss": 0.8225,
       "step": 13299
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1012401428542709e-05,
-      "loss": 0.7998,
+      "loss": 0.8857,
       "step": 13300
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1011229611682734e-05,
-      "loss": 0.6616,
+      "loss": 0.7924,
       "step": 13301
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1010057780793382e-05,
-      "loss": 0.7299,
+      "loss": 0.855,
       "step": 13302
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.100888593589091e-05,
-      "loss": 0.7393,
+      "loss": 0.8317,
       "step": 13303
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1007714076991578e-05,
-      "loss": 0.7614,
+      "loss": 0.8549,
       "step": 13304
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1006542204111644e-05,
-      "loss": 0.6784,
+      "loss": 0.8601,
       "step": 13305
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1005370317267364e-05,
-      "loss": 0.7185,
+      "loss": 0.8732,
       "step": 13306
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1004198416474999e-05,
-      "loss": 0.7577,
+      "loss": 0.8432,
       "step": 13307
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.10030265017508e-05,
-      "loss": 0.7478,
+      "loss": 0.8688,
       "step": 13308
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.100185457311104e-05,
-      "loss": 0.712,
+      "loss": 0.833,
       "step": 13309
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.1000682630571962e-05,
-      "loss": 0.8367,
+      "loss": 0.8941,
       "step": 13310
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0999510674149836e-05,
-      "loss": 0.7606,
+      "loss": 0.9331,
       "step": 13311
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0998338703860918e-05,
-      "loss": 0.7454,
+      "loss": 0.8394,
       "step": 13312
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0997166719721463e-05,
-      "loss": 0.7694,
+      "loss": 0.909,
       "step": 13313
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0995994721747738e-05,
-      "loss": 0.7141,
+      "loss": 0.8516,
       "step": 13314
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0994822709956e-05,
-      "loss": 0.8488,
+      "loss": 0.9175,
       "step": 13315
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0993650684362505e-05,
-      "loss": 0.6566,
+      "loss": 0.8069,
       "step": 13316
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0992478644983517e-05,
-      "loss": 0.7956,
+      "loss": 0.8469,
       "step": 13317
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0991306591835297e-05,
-      "loss": 0.8226,
+      "loss": 0.8895,
       "step": 13318
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0990134524934107e-05,
-      "loss": 0.7726,
+      "loss": 0.9193,
       "step": 13319
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0988962444296203e-05,
-      "loss": 0.7407,
+      "loss": 0.8647,
       "step": 13320
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0987790349937846e-05,
-      "loss": 0.915,
+      "loss": 0.9511,
       "step": 13321
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0986618241875301e-05,
-      "loss": 0.6624,
+      "loss": 0.8711,
       "step": 13322
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.098544612012483e-05,
-      "loss": 0.8118,
+      "loss": 0.8562,
       "step": 13323
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.098427398470269e-05,
-      "loss": 0.707,
+      "loss": 0.8456,
       "step": 13324
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0983101835625146e-05,
-      "loss": 0.8671,
+      "loss": 0.9005,
       "step": 13325
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0981929672908457e-05,
-      "loss": 0.6529,
+      "loss": 0.8552,
       "step": 13326
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0980757496568888e-05,
-      "loss": 0.7742,
+      "loss": 0.8579,
       "step": 13327
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0979585306622704e-05,
-      "loss": 0.6596,
+      "loss": 0.7627,
       "step": 13328
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0978413103086158e-05,
-      "loss": 0.7896,
+      "loss": 0.8751,
       "step": 13329
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0977240885975522e-05,
-      "loss": 0.6486,
+      "loss": 0.7578,
       "step": 13330
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0976068655307056e-05,
-      "loss": 0.899,
+      "loss": 0.9772,
       "step": 13331
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0974896411097019e-05,
-      "loss": 0.653,
+      "loss": 0.9175,
       "step": 13332
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0973724153361678e-05,
-      "loss": 0.8122,
+      "loss": 0.8954,
       "step": 13333
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0972551882117297e-05,
-      "loss": 0.8418,
+      "loss": 0.9163,
       "step": 13334
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.097137959738014e-05,
-      "loss": 0.7662,
+      "loss": 0.9321,
       "step": 13335
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.0970207299166466e-05,
-      "loss": 0.7636,
+      "loss": 0.9118,
       "step": 13336
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0969034987492542e-05,
-      "loss": 0.8528,
+      "loss": 0.9499,
       "step": 13337
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0967862662374635e-05,
-      "loss": 0.7227,
+      "loss": 0.7393,
       "step": 13338
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0966690323829007e-05,
-      "loss": 0.6594,
+      "loss": 0.7778,
       "step": 13339
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0965517971871919e-05,
-      "loss": 0.7797,
+      "loss": 0.8823,
       "step": 13340
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0964345606519642e-05,
-      "loss": 0.7184,
+      "loss": 0.899,
       "step": 13341
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0963173227788437e-05,
-      "loss": 0.6512,
+      "loss": 0.7555,
       "step": 13342
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.096200083569457e-05,
-      "loss": 0.7369,
+      "loss": 0.8277,
       "step": 13343
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0960828430254307e-05,
-      "loss": 0.7669,
+      "loss": 0.8314,
       "step": 13344
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0959656011483913e-05,
-      "loss": 0.625,
+      "loss": 0.7512,
       "step": 13345
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0958483579399652e-05,
-      "loss": 0.7573,
+      "loss": 0.8692,
       "step": 13346
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0957311134017795e-05,
-      "loss": 0.7883,
+      "loss": 0.8857,
       "step": 13347
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0956138675354599e-05,
-      "loss": 0.6815,
+      "loss": 0.8762,
       "step": 13348
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0954966203426338e-05,
-      "loss": 0.7437,
+      "loss": 0.823,
       "step": 13349
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0953793718249278e-05,
-      "loss": 0.824,
+      "loss": 0.8971,
       "step": 13350
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0952621219839678e-05,
-      "loss": 0.8331,
+      "loss": 0.8087,
       "step": 13351
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0951448708213815e-05,
-      "loss": 0.8307,
+      "loss": 0.9054,
       "step": 13352
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0950276183387948e-05,
-      "loss": 0.7797,
+      "loss": 0.874,
       "step": 13353
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0949103645378348e-05,
-      "loss": 0.4996,
+      "loss": 0.7014,
       "step": 13354
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0947931094201282e-05,
-      "loss": 0.7102,
+      "loss": 0.8936,
       "step": 13355
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0946758529873015e-05,
-      "loss": 0.7047,
+      "loss": 0.8089,
       "step": 13356
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.094558595240982e-05,
-      "loss": 0.6741,
+      "loss": 0.7603,
       "step": 13357
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.094441336182796e-05,
-      "loss": 0.725,
+      "loss": 0.8196,
       "step": 13358
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0943240758143705e-05,
-      "loss": 0.7744,
+      "loss": 0.8968,
       "step": 13359
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0942068141373323e-05,
-      "loss": 0.6823,
+      "loss": 0.7474,
       "step": 13360
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0940895511533084e-05,
-      "loss": 0.794,
+      "loss": 0.8696,
       "step": 13361
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0939722868639248e-05,
-      "loss": 0.7548,
+      "loss": 0.864,
       "step": 13362
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0938550212708097e-05,
-      "loss": 0.6782,
+      "loss": 0.7965,
       "step": 13363
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0937377543755889e-05,
-      "loss": 0.8381,
+      "loss": 0.9294,
       "step": 13364
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0936204861798903e-05,
-      "loss": 0.7117,
+      "loss": 0.8018,
       "step": 13365
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0935032166853399e-05,
-      "loss": 0.7324,
+      "loss": 0.8881,
       "step": 13366
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0933859458935651e-05,
-      "loss": 0.8049,
+      "loss": 0.9308,
       "step": 13367
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0932686738061928e-05,
-      "loss": 0.7094,
+      "loss": 0.771,
       "step": 13368
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.09315140042485e-05,
-      "loss": 0.7357,
+      "loss": 0.8457,
       "step": 13369
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0930341257511636e-05,
-      "loss": 0.7705,
+      "loss": 0.8242,
       "step": 13370
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.092916849786761e-05,
-      "loss": 0.7258,
+      "loss": 0.9321,
       "step": 13371
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0927995725332686e-05,
-      "loss": 0.6965,
+      "loss": 0.8667,
       "step": 13372
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.092682293992314e-05,
-      "loss": 0.8448,
+      "loss": 0.9076,
       "step": 13373
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.092565014165524e-05,
-      "loss": 0.8835,
+      "loss": 0.9479,
       "step": 13374
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0924477330545255e-05,
-      "loss": 0.7602,
+      "loss": 0.8346,
       "step": 13375
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0923304506609462e-05,
-      "loss": 0.6984,
+      "loss": 0.7972,
       "step": 13376
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0922131669864127e-05,
-      "loss": 0.704,
+      "loss": 0.8011,
       "step": 13377
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0920958820325525e-05,
-      "loss": 0.7728,
+      "loss": 0.8486,
       "step": 13378
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0919785958009926e-05,
-      "loss": 0.865,
+      "loss": 0.9514,
       "step": 13379
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.09186130829336e-05,
-      "loss": 0.7279,
+      "loss": 0.8547,
       "step": 13380
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0917440195112822e-05,
-      "loss": 0.6178,
+      "loss": 0.7136,
       "step": 13381
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0916267294563865e-05,
-      "loss": 0.7872,
+      "loss": 0.9508,
       "step": 13382
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0915094381302994e-05,
-      "loss": 0.8026,
+      "loss": 0.8269,
       "step": 13383
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0913921455346493e-05,
-      "loss": 0.7383,
+      "loss": 0.8124,
       "step": 13384
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0912748516710626e-05,
-      "loss": 0.6855,
+      "loss": 0.8902,
       "step": 13385
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0911575565411669e-05,
-      "loss": 0.6751,
+      "loss": 0.8088,
       "step": 13386
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0910402601465895e-05,
-      "loss": 0.7663,
+      "loss": 0.8106,
       "step": 13387
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0909229624889573e-05,
-      "loss": 0.7954,
+      "loss": 0.9195,
       "step": 13388
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0908056635698982e-05,
-      "loss": 0.7076,
+      "loss": 0.86,
       "step": 13389
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0906883633910395e-05,
-      "loss": 0.7667,
+      "loss": 0.8601,
       "step": 13390
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.090571061954008e-05,
-      "loss": 0.7103,
+      "loss": 0.8013,
       "step": 13391
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0904537592604321e-05,
-      "loss": 0.7492,
+      "loss": 0.7988,
       "step": 13392
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.090336455311938e-05,
-      "loss": 0.8019,
+      "loss": 0.9084,
       "step": 13393
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0902191501101541e-05,
-      "loss": 0.7202,
+      "loss": 0.8821,
       "step": 13394
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0901018436567077e-05,
-      "loss": 0.7361,
+      "loss": 0.8491,
       "step": 13395
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0899845359532257e-05,
-      "loss": 0.7153,
+      "loss": 0.8419,
       "step": 13396
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0898672270013361e-05,
-      "loss": 0.7227,
+      "loss": 0.8315,
       "step": 13397
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0897499168026662e-05,
-      "loss": 0.7072,
+      "loss": 0.7961,
       "step": 13398
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0896326053588436e-05,
-      "loss": 0.7766,
+      "loss": 0.8972,
       "step": 13399
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0895152926714955e-05,
-      "loss": 0.7222,
+      "loss": 0.9148,
       "step": 13400
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0893979787422502e-05,
-      "loss": 0.8939,
+      "loss": 0.9727,
       "step": 13401
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0892806635727344e-05,
-      "loss": 0.6977,
+      "loss": 0.8114,
       "step": 13402
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0891633471645761e-05,
-      "loss": 0.6691,
+      "loss": 0.8574,
       "step": 13403
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.089046029519403e-05,
-      "loss": 0.8486,
+      "loss": 0.9243,
       "step": 13404
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0889287106388426e-05,
-      "loss": 0.7661,
+      "loss": 0.8195,
       "step": 13405
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0888113905245226e-05,
-      "loss": 0.7112,
+      "loss": 0.8548,
       "step": 13406
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0886940691780704e-05,
-      "loss": 0.7326,
+      "loss": 0.8302,
       "step": 13407
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0885767466011139e-05,
-      "loss": 0.8047,
+      "loss": 0.9481,
       "step": 13408
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0884594227952805e-05,
-      "loss": 0.7428,
+      "loss": 0.8377,
       "step": 13409
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0883420977621983e-05,
-      "loss": 0.8365,
+      "loss": 0.9268,
       "step": 13410
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0882247715034949e-05,
-      "loss": 0.7631,
+      "loss": 0.8501,
       "step": 13411
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0881074440207977e-05,
-      "loss": 0.6739,
+      "loss": 0.7501,
       "step": 13412
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0879901153157348e-05,
-      "loss": 0.7936,
+      "loss": 0.8106,
       "step": 13413
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0878727853899342e-05,
-      "loss": 0.6986,
+      "loss": 0.8907,
       "step": 13414
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.087755454245023e-05,
-      "loss": 0.7303,
+      "loss": 0.8388,
       "step": 13415
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0876381218826298e-05,
-      "loss": 0.7917,
+      "loss": 0.8374,
       "step": 13416
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0875207883043817e-05,
-      "loss": 0.7489,
+      "loss": 0.8267,
       "step": 13417
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0874034535119067e-05,
-      "loss": 0.7824,
+      "loss": 0.8378,
       "step": 13418
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.087286117506833e-05,
-      "loss": 0.7563,
+      "loss": 0.8462,
       "step": 13419
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0871687802907881e-05,
-      "loss": 0.6904,
+      "loss": 0.8027,
       "step": 13420
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0870514418654003e-05,
-      "loss": 0.7846,
+      "loss": 0.8956,
       "step": 13421
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0869341022322972e-05,
-      "loss": 0.7935,
+      "loss": 0.8314,
       "step": 13422
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0868167613931066e-05,
-      "loss": 0.888,
+      "loss": 1.0006,
       "step": 13423
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0866994193494571e-05,
-      "loss": 0.732,
+      "loss": 0.8606,
       "step": 13424
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0865820761029755e-05,
-      "loss": 0.7944,
+      "loss": 0.9172,
       "step": 13425
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.086464731655291e-05,
-      "loss": 0.8263,
+      "loss": 0.9108,
       "step": 13426
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0863473860080307e-05,
-      "loss": 0.8345,
+      "loss": 0.8829,
       "step": 13427
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0862300391628229e-05,
-      "loss": 0.7222,
+      "loss": 0.7979,
       "step": 13428
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0861126911212958e-05,
-      "loss": 0.8654,
+      "loss": 0.8995,
       "step": 13429
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0859953418850773e-05,
-      "loss": 0.7681,
+      "loss": 0.8694,
       "step": 13430
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0858779914557952e-05,
-      "loss": 0.8181,
+      "loss": 0.8774,
       "step": 13431
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0857606398350781e-05,
-      "loss": 0.7647,
+      "loss": 0.8116,
       "step": 13432
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0856432870245537e-05,
-      "loss": 0.8848,
+      "loss": 0.9055,
       "step": 13433
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.08552593302585e-05,
-      "loss": 0.7551,
+      "loss": 0.8822,
       "step": 13434
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0854085778405956e-05,
-      "loss": 0.6861,
+      "loss": 0.819,
       "step": 13435
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0852912214704184e-05,
-      "loss": 0.7419,
+      "loss": 0.8317,
       "step": 13436
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0851738639169462e-05,
-      "loss": 0.6542,
+      "loss": 0.749,
       "step": 13437
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0850565051818079e-05,
-      "loss": 0.7276,
+      "loss": 0.7876,
       "step": 13438
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.084939145266631e-05,
-      "loss": 0.5838,
+      "loss": 0.7629,
       "step": 13439
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.084821784173044e-05,
-      "loss": 0.6801,
+      "loss": 0.8109,
       "step": 13440
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0847044219026752e-05,
-      "loss": 0.6332,
+      "loss": 0.8275,
       "step": 13441
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0845870584571525e-05,
-      "loss": 0.6918,
+      "loss": 0.8031,
       "step": 13442
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0844696938381048e-05,
-      "loss": 0.8376,
+      "loss": 0.8886,
       "step": 13443
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0843523280471594e-05,
-      "loss": 0.6769,
+      "loss": 0.8745,
       "step": 13444
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0842349610859455e-05,
-      "loss": 0.7347,
+      "loss": 0.8955,
       "step": 13445
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.084117592956091e-05,
-      "loss": 0.8935,
+      "loss": 0.9308,
       "step": 13446
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.084000223659224e-05,
-      "loss": 0.91,
+      "loss": 0.9281,
       "step": 13447
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0838828531969734e-05,
-      "loss": 0.7528,
+      "loss": 0.7948,
       "step": 13448
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0837654815709668e-05,
-      "loss": 0.8554,
+      "loss": 0.9013,
       "step": 13449
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0836481087828335e-05,
-      "loss": 0.791,
+      "loss": 0.8622,
       "step": 13450
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.083530734834201e-05,
-      "loss": 0.7792,
+      "loss": 0.875,
       "step": 13451
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.083413359726698e-05,
-      "loss": 0.6124,
+      "loss": 0.836,
       "step": 13452
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0832959834619534e-05,
-      "loss": 0.7765,
+      "loss": 0.9205,
       "step": 13453
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.083178606041595e-05,
-      "loss": 0.7489,
+      "loss": 0.8416,
       "step": 13454
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0830612274672514e-05,
-      "loss": 0.6494,
+      "loss": 0.8111,
       "step": 13455
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0829438477405512e-05,
-      "loss": 0.6946,
+      "loss": 0.7676,
       "step": 13456
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0828264668631228e-05,
-      "loss": 0.8221,
+      "loss": 0.9622,
       "step": 13457
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0827090848365944e-05,
-      "loss": 0.823,
+      "loss": 0.8309,
       "step": 13458
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0825917016625953e-05,
-      "loss": 0.6184,
+      "loss": 0.74,
       "step": 13459
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.082474317342753e-05,
-      "loss": 0.6979,
+      "loss": 0.8517,
       "step": 13460
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0823569318786971e-05,
-      "loss": 0.6338,
+      "loss": 0.8059,
       "step": 13461
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0822395452720555e-05,
-      "loss": 0.7039,
+      "loss": 0.9124,
       "step": 13462
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0821221575244567e-05,
-      "loss": 0.8219,
+      "loss": 0.9532,
       "step": 13463
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0820047686375295e-05,
-      "loss": 0.7461,
+      "loss": 0.8489,
       "step": 13464
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0818873786129022e-05,
-      "loss": 0.6597,
+      "loss": 0.8086,
       "step": 13465
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0817699874522043e-05,
-      "loss": 0.7553,
+      "loss": 0.8777,
       "step": 13466
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0816525951570635e-05,
-      "loss": 0.6482,
+      "loss": 0.7794,
       "step": 13467
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0815352017291087e-05,
-      "loss": 0.6561,
+      "loss": 0.8679,
       "step": 13468
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0814178071699688e-05,
-      "loss": 0.6848,
+      "loss": 0.8596,
       "step": 13469
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0813004114812726e-05,
-      "loss": 0.7237,
+      "loss": 0.8536,
       "step": 13470
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.081183014664648e-05,
-      "loss": 0.8247,
+      "loss": 0.9078,
       "step": 13471
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0810656167217246e-05,
-      "loss": 0.7671,
+      "loss": 0.8365,
       "step": 13472
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0809482176541308e-05,
-      "loss": 0.755,
+      "loss": 0.9309,
       "step": 13473
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0808308174634951e-05,
-      "loss": 0.6998,
+      "loss": 0.8156,
       "step": 13474
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0807134161514468e-05,
-      "loss": 0.764,
+      "loss": 0.8576,
       "step": 13475
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.080596013719614e-05,
-      "loss": 0.653,
+      "loss": 0.8331,
       "step": 13476
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0804786101696263e-05,
-      "loss": 0.7346,
+      "loss": 0.8476,
       "step": 13477
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0803612055031118e-05,
-      "loss": 0.7075,
+      "loss": 0.8367,
       "step": 13478
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0802437997216994e-05,
-      "loss": 0.8409,
+      "loss": 0.97,
       "step": 13479
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0801263928270185e-05,
-      "loss": 0.7736,
+      "loss": 0.8748,
       "step": 13480
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0800089848206976e-05,
-      "loss": 0.8006,
+      "loss": 0.8609,
       "step": 13481
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0798915757043656e-05,
-      "loss": 0.6558,
+      "loss": 0.7993,
       "step": 13482
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.079774165479651e-05,
-      "loss": 0.6721,
+      "loss": 0.7938,
       "step": 13483
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0796567541481831e-05,
-      "loss": 0.7052,
+      "loss": 0.8726,
       "step": 13484
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0795393417115909e-05,
-      "loss": 0.6515,
+      "loss": 0.8434,
       "step": 13485
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0794219281715033e-05,
-      "loss": 0.762,
+      "loss": 0.8459,
       "step": 13486
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0793045135295488e-05,
-      "loss": 0.7897,
+      "loss": 0.9229,
       "step": 13487
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0791870977873568e-05,
-      "loss": 0.8402,
+      "loss": 0.9156,
       "step": 13488
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0790696809465565e-05,
-      "loss": 0.8895,
+      "loss": 0.9185,
       "step": 13489
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.078952263008776e-05,
-      "loss": 0.7145,
+      "loss": 0.8367,
       "step": 13490
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0788348439756451e-05,
-      "loss": 0.7138,
+      "loss": 0.7847,
       "step": 13491
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0787174238487925e-05,
-      "loss": 0.7483,
+      "loss": 0.8898,
       "step": 13492
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0786000026298475e-05,
-      "loss": 0.6437,
+      "loss": 0.7628,
       "step": 13493
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.078482580320439e-05,
-      "loss": 0.791,
+      "loss": 0.8633,
       "step": 13494
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.078365156922196e-05,
-      "loss": 0.7089,
+      "loss": 0.8358,
       "step": 13495
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0782477324367477e-05,
-      "loss": 0.8482,
+      "loss": 0.9338,
       "step": 13496
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0781303068657232e-05,
-      "loss": 0.6172,
+      "loss": 0.8054,
       "step": 13497
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0780128802107513e-05,
-      "loss": 0.8246,
+      "loss": 0.8523,
       "step": 13498
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0778954524734616e-05,
-      "loss": 0.7439,
+      "loss": 0.8685,
       "step": 13499
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0777780236554827e-05,
-      "loss": 0.7115,
+      "loss": 0.8673,
       "step": 13500
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0776605937584444e-05,
-      "loss": 0.7873,
+      "loss": 0.8713,
       "step": 13501
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0775431627839753e-05,
-      "loss": 0.6811,
+      "loss": 0.7823,
       "step": 13502
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.077425730733705e-05,
-      "loss": 0.6072,
+      "loss": 0.7555,
       "step": 13503
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0773082976092626e-05,
-      "loss": 0.6499,
+      "loss": 0.7827,
       "step": 13504
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0771908634122772e-05,
-      "loss": 0.6881,
+      "loss": 0.8136,
       "step": 13505
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.077073428144378e-05,
-      "loss": 0.8731,
+      "loss": 0.9218,
       "step": 13506
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0769559918071944e-05,
-      "loss": 0.8228,
+      "loss": 0.8484,
       "step": 13507
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0768385544023554e-05,
-      "loss": 0.7473,
+      "loss": 0.7775,
       "step": 13508
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0767211159314908e-05,
-      "loss": 0.6695,
+      "loss": 0.785,
       "step": 13509
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0766036763962293e-05,
-      "loss": 0.7671,
+      "loss": 0.8254,
       "step": 13510
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0764862357982007e-05,
-      "loss": 0.8445,
+      "loss": 0.8779,
       "step": 13511
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.076368794139034e-05,
-      "loss": 0.6458,
+      "loss": 0.8078,
       "step": 13512
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0762513514203588e-05,
-      "loss": 0.8056,
+      "loss": 0.8902,
       "step": 13513
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0761339076438039e-05,
-      "loss": 0.6424,
+      "loss": 0.7278,
       "step": 13514
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0760164628109993e-05,
-      "loss": 0.757,
+      "loss": 0.7981,
       "step": 13515
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0758990169235742e-05,
-      "loss": 0.7454,
+      "loss": 0.8913,
       "step": 13516
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0757815699831575e-05,
-      "loss": 0.8458,
+      "loss": 0.943,
       "step": 13517
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0756641219913795e-05,
-      "loss": 0.7444,
+      "loss": 0.8509,
       "step": 13518
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0755466729498689e-05,
-      "loss": 0.6317,
+      "loss": 0.7312,
       "step": 13519
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0754292228602559e-05,
-      "loss": 0.7338,
+      "loss": 0.7703,
       "step": 13520
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0753117717241688e-05,
-      "loss": 0.7359,
+      "loss": 0.8266,
       "step": 13521
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0751943195432382e-05,
-      "loss": 0.7179,
+      "loss": 0.8107,
       "step": 13522
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0750768663190929e-05,
-      "loss": 0.702,
+      "loss": 0.8009,
       "step": 13523
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0749594120533623e-05,
-      "loss": 0.6755,
+      "loss": 0.8227,
       "step": 13524
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0748419567476766e-05,
-      "loss": 0.7815,
+      "loss": 0.9128,
       "step": 13525
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.074724500403665e-05,
-      "loss": 0.7724,
+      "loss": 0.8594,
       "step": 13526
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0746070430229567e-05,
-      "loss": 0.6743,
+      "loss": 0.7975,
       "step": 13527
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.074489584607182e-05,
-      "loss": 0.7532,
+      "loss": 0.8732,
       "step": 13528
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0743721251579694e-05,
-      "loss": 0.7115,
+      "loss": 0.8227,
       "step": 13529
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0742546646769493e-05,
-      "loss": 0.7724,
+      "loss": 0.8705,
       "step": 13530
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0741372031657513e-05,
-      "loss": 0.6711,
+      "loss": 0.8667,
       "step": 13531
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0740197406260045e-05,
-      "loss": 0.7775,
+      "loss": 0.8807,
       "step": 13532
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0739022770593387e-05,
-      "loss": 0.7666,
+      "loss": 0.8543,
       "step": 13533
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0737848124673841e-05,
-      "loss": 0.7931,
+      "loss": 0.9136,
       "step": 13534
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0736673468517695e-05,
-      "loss": 0.8575,
+      "loss": 0.9367,
       "step": 13535
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0735498802141253e-05,
-      "loss": 0.7573,
+      "loss": 0.8758,
       "step": 13536
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0734324125560809e-05,
-      "loss": 0.5892,
+      "loss": 0.856,
       "step": 13537
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0733149438792657e-05,
-      "loss": 0.7869,
+      "loss": 0.8424,
       "step": 13538
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0731974741853102e-05,
-      "loss": 0.8161,
+      "loss": 0.8696,
       "step": 13539
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0730800034758428e-05,
-      "loss": 0.7519,
+      "loss": 0.9278,
       "step": 13540
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0729625317524945e-05,
-      "loss": 0.7048,
+      "loss": 0.8214,
       "step": 13541
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0728450590168948e-05,
-      "loss": 0.7731,
+      "loss": 0.8345,
       "step": 13542
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.072727585270673e-05,
-      "loss": 0.8348,
+      "loss": 0.8989,
       "step": 13543
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.072610110515459e-05,
-      "loss": 0.8773,
+      "loss": 0.8863,
       "step": 13544
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.072492634752883e-05,
-      "loss": 0.8314,
+      "loss": 0.8869,
       "step": 13545
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0723751579845743e-05,
-      "loss": 0.5928,
+      "loss": 0.808,
       "step": 13546
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0722576802121631e-05,
-      "loss": 0.6544,
+      "loss": 0.7453,
       "step": 13547
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0721402014372788e-05,
-      "loss": 0.7077,
+      "loss": 0.8826,
       "step": 13548
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.072022721661552e-05,
-      "loss": 0.7366,
+      "loss": 0.8814,
       "step": 13549
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.071905240886612e-05,
-      "loss": 0.6324,
+      "loss": 0.7552,
       "step": 13550
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0717877591140885e-05,
-      "loss": 0.7461,
+      "loss": 0.8573,
       "step": 13551
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0716702763456122e-05,
-      "loss": 0.6581,
+      "loss": 0.7766,
       "step": 13552
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0715527925828124e-05,
-      "loss": 0.6852,
+      "loss": 0.9,
       "step": 13553
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0714353078273188e-05,
-      "loss": 0.771,
+      "loss": 0.9064,
       "step": 13554
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.071317822080762e-05,
-      "loss": 0.8116,
+      "loss": 0.9014,
       "step": 13555
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0712003353447716e-05,
-      "loss": 0.8303,
+      "loss": 0.9159,
       "step": 13556
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.071082847620977e-05,
-      "loss": 0.6672,
+      "loss": 0.8434,
       "step": 13557
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0709653589110097e-05,
-      "loss": 0.701,
+      "loss": 0.896,
       "step": 13558
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.070847869216498e-05,
-      "loss": 0.7434,
+      "loss": 0.8275,
       "step": 13559
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0707303785390727e-05,
-      "loss": 0.8063,
+      "loss": 0.8876,
       "step": 13560
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.070612886880364e-05,
-      "loss": 0.6476,
+      "loss": 0.946,
       "step": 13561
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0704953942420016e-05,
-      "loss": 0.8321,
+      "loss": 0.9143,
       "step": 13562
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0703779006256157e-05,
-      "loss": 0.719,
+      "loss": 0.8033,
       "step": 13563
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0702604060328359e-05,
-      "loss": 0.8918,
+      "loss": 0.8605,
       "step": 13564
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.070142910465293e-05,
-      "loss": 0.7588,
+      "loss": 0.8244,
       "step": 13565
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0700254139246166e-05,
-      "loss": 0.7832,
+      "loss": 0.914,
       "step": 13566
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0699079164124368e-05,
-      "loss": 0.8866,
+      "loss": 0.9663,
       "step": 13567
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0697904179303841e-05,
-      "loss": 0.7012,
+      "loss": 0.8192,
       "step": 13568
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0696729184800884e-05,
-      "loss": 0.6505,
+      "loss": 0.8011,
       "step": 13569
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0695554180631798e-05,
-      "loss": 0.792,
+      "loss": 0.8542,
       "step": 13570
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0694379166812884e-05,
-      "loss": 0.7525,
+      "loss": 0.8038,
       "step": 13571
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0693204143360445e-05,
-      "loss": 0.7374,
+      "loss": 0.7875,
       "step": 13572
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.069202911029078e-05,
-      "loss": 0.7169,
+      "loss": 0.817,
       "step": 13573
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0690854067620194e-05,
-      "loss": 0.7586,
+      "loss": 0.8773,
       "step": 13574
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0689679015364986e-05,
-      "loss": 0.9418,
+      "loss": 1.0025,
       "step": 13575
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0688503953541465e-05,
-      "loss": 0.8164,
+      "loss": 0.8812,
       "step": 13576
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0687328882165928e-05,
-      "loss": 0.8087,
+      "loss": 0.8495,
       "step": 13577
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0686153801254673e-05,
-      "loss": 0.8136,
+      "loss": 0.8452,
       "step": 13578
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0684978710824012e-05,
-      "loss": 0.6349,
+      "loss": 0.7921,
       "step": 13579
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.068380361089024e-05,
-      "loss": 0.708,
+      "loss": 0.7676,
       "step": 13580
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0682628501469664e-05,
-      "loss": 0.7348,
+      "loss": 0.8409,
       "step": 13581
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0681453382578588e-05,
-      "loss": 0.6615,
+      "loss": 0.8004,
       "step": 13582
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0680278254233312e-05,
-      "loss": 0.7641,
+      "loss": 0.8441,
       "step": 13583
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.067910311645014e-05,
-      "loss": 0.8018,
+      "loss": 0.8875,
       "step": 13584
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0677927969245377e-05,
-      "loss": 0.791,
+      "loss": 0.8785,
       "step": 13585
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0676752812635323e-05,
-      "loss": 0.7641,
+      "loss": 0.8202,
       "step": 13586
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0675577646636286e-05,
-      "loss": 0.8446,
+      "loss": 0.8995,
       "step": 13587
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0674402471264567e-05,
-      "loss": 0.6928,
+      "loss": 0.8041,
       "step": 13588
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0673227286536469e-05,
-      "loss": 0.8754,
+      "loss": 0.9389,
       "step": 13589
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.06720520924683e-05,
-      "loss": 0.7708,
+      "loss": 0.889,
       "step": 13590
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0670876889076358e-05,
-      "loss": 0.7167,
+      "loss": 0.8726,
       "step": 13591
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0669701676376955e-05,
-      "loss": 0.867,
+      "loss": 0.8974,
       "step": 13592
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0668526454386392e-05,
-      "loss": 0.6606,
+      "loss": 0.8286,
       "step": 13593
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.066735122312097e-05,
-      "loss": 0.7736,
+      "loss": 0.8612,
       "step": 13594
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0666175982596997e-05,
-      "loss": 0.6548,
+      "loss": 0.8111,
       "step": 13595
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0665000732830778e-05,
-      "loss": 0.7515,
+      "loss": 0.8666,
       "step": 13596
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0663825473838621e-05,
-      "loss": 0.6047,
+      "loss": 0.7737,
       "step": 13597
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0662650205636825e-05,
-      "loss": 0.8097,
+      "loss": 0.8879,
       "step": 13598
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0661474928241692e-05,
-      "loss": 0.7595,
+      "loss": 0.9178,
       "step": 13599
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.066029964166954e-05,
-      "loss": 0.809,
+      "loss": 0.8566,
       "step": 13600
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0659124345936665e-05,
-      "loss": 0.8228,
+      "loss": 0.9479,
       "step": 13601
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0657949041059374e-05,
-      "loss": 0.8113,
+      "loss": 0.9008,
       "step": 13602
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0656773727053975e-05,
-      "loss": 0.7353,
+      "loss": 0.8759,
       "step": 13603
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.065559840393677e-05,
-      "loss": 0.7721,
+      "loss": 0.8804,
       "step": 13604
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0654423071724069e-05,
-      "loss": 0.7457,
+      "loss": 0.7878,
       "step": 13605
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0653247730432177e-05,
-      "loss": 0.7096,
+      "loss": 0.832,
       "step": 13606
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0652072380077398e-05,
-      "loss": 0.8469,
+      "loss": 0.9309,
       "step": 13607
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.065089702067604e-05,
-      "loss": 0.6385,
+      "loss": 0.7966,
       "step": 13608
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.064972165224441e-05,
-      "loss": 0.7846,
+      "loss": 0.9167,
       "step": 13609
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.0648546274798814e-05,
-      "loss": 0.8595,
+      "loss": 0.9554,
       "step": 13610
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.064737088835556e-05,
-      "loss": 0.9566,
+      "loss": 0.9572,
       "step": 13611
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0646195492930954e-05,
-      "loss": 0.7018,
+      "loss": 0.7954,
       "step": 13612
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0645020088541296e-05,
-      "loss": 0.8211,
+      "loss": 0.9235,
       "step": 13613
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0643844675202905e-05,
-      "loss": 0.7308,
+      "loss": 0.8299,
       "step": 13614
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0642669252932082e-05,
-      "loss": 0.7417,
+      "loss": 0.8061,
       "step": 13615
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0641493821745135e-05,
-      "loss": 0.6443,
+      "loss": 0.7541,
       "step": 13616
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0640318381658372e-05,
-      "loss": 0.7961,
+      "loss": 0.9236,
       "step": 13617
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0639142932688098e-05,
-      "loss": 0.7534,
+      "loss": 0.8006,
       "step": 13618
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0637967474850623e-05,
-      "loss": 0.6734,
+      "loss": 0.7752,
       "step": 13619
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0636792008162253e-05,
-      "loss": 0.9323,
+      "loss": 0.9182,
       "step": 13620
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0635616532639302e-05,
-      "loss": 0.7206,
+      "loss": 0.779,
       "step": 13621
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.063444104829807e-05,
-      "loss": 0.8325,
+      "loss": 0.9711,
       "step": 13622
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0633265555154868e-05,
-      "loss": 0.7686,
+      "loss": 0.8471,
       "step": 13623
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0632090053226007e-05,
-      "loss": 0.7978,
+      "loss": 0.9143,
       "step": 13624
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0630914542527793e-05,
-      "loss": 0.7296,
+      "loss": 0.7937,
       "step": 13625
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0629739023076533e-05,
-      "loss": 0.6893,
+      "loss": 0.8279,
       "step": 13626
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.062856349488854e-05,
-      "loss": 0.7847,
+      "loss": 0.8411,
       "step": 13627
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.062738795798012e-05,
-      "loss": 0.6713,
+      "loss": 0.78,
       "step": 13628
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0626212412367578e-05,
-      "loss": 0.5851,
+      "loss": 0.8074,
       "step": 13629
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0625036858067233e-05,
-      "loss": 0.7463,
+      "loss": 0.8696,
       "step": 13630
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0623861295095383e-05,
-      "loss": 0.7813,
+      "loss": 0.8527,
       "step": 13631
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0622685723468346e-05,
-      "loss": 0.7565,
+      "loss": 0.8665,
       "step": 13632
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.062151014320243e-05,
-      "loss": 0.7299,
+      "loss": 0.8794,
       "step": 13633
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0620334554313938e-05,
-      "loss": 0.7374,
+      "loss": 0.9134,
       "step": 13634
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0619158956819191e-05,
-      "loss": 0.7,
+      "loss": 0.8596,
       "step": 13635
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0617983350734484e-05,
-      "loss": 0.752,
+      "loss": 0.8684,
       "step": 13636
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.061680773607614e-05,
-      "loss": 0.6896,
+      "loss": 0.8354,
       "step": 13637
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0615632112860464e-05,
-      "loss": 0.6826,
+      "loss": 0.7692,
       "step": 13638
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0614456481103761e-05,
-      "loss": 0.6447,
+      "loss": 0.8152,
       "step": 13639
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0613280840822352e-05,
-      "loss": 0.708,
+      "loss": 0.924,
       "step": 13640
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0612105192032539e-05,
-      "loss": 0.746,
+      "loss": 0.8445,
       "step": 13641
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.061092953475063e-05,
-      "loss": 0.7438,
+      "loss": 0.8683,
       "step": 13642
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0609753868992947e-05,
-      "loss": 0.7504,
+      "loss": 0.8532,
       "step": 13643
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0608578194775794e-05,
-      "loss": 0.6981,
+      "loss": 0.8023,
       "step": 13644
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.060740251211548e-05,
-      "loss": 0.7545,
+      "loss": 0.8713,
       "step": 13645
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0606226821028319e-05,
-      "loss": 0.7523,
+      "loss": 0.8063,
       "step": 13646
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0605051121530616e-05,
-      "loss": 0.7963,
+      "loss": 0.89,
       "step": 13647
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0603875413638694e-05,
-      "loss": 0.7347,
+      "loss": 0.8676,
       "step": 13648
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0602699697368854e-05,
-      "loss": 0.7588,
+      "loss": 0.8714,
       "step": 13649
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.060152397273741e-05,
-      "loss": 0.6827,
+      "loss": 0.8091,
       "step": 13650
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0600348239760679e-05,
-      "loss": 0.6874,
+      "loss": 0.8294,
       "step": 13651
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0599172498454964e-05,
-      "loss": 0.6859,
+      "loss": 0.8605,
       "step": 13652
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0597996748836579e-05,
-      "loss": 0.7384,
+      "loss": 0.905,
       "step": 13653
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0596820990921844e-05,
-      "loss": 0.6595,
+      "loss": 0.8488,
       "step": 13654
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0595645224727059e-05,
-      "loss": 0.7247,
+      "loss": 0.8022,
       "step": 13655
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0594469450268542e-05,
-      "loss": 0.7434,
+      "loss": 0.8502,
       "step": 13656
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.059329366756261e-05,
-      "loss": 0.71,
+      "loss": 0.8122,
       "step": 13657
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0592117876625562e-05,
-      "loss": 0.7742,
+      "loss": 0.8592,
       "step": 13658
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0590942077473725e-05,
-      "loss": 0.7219,
+      "loss": 0.8624,
       "step": 13659
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0589766270123403e-05,
-      "loss": 0.7568,
+      "loss": 0.8717,
       "step": 13660
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.058859045459091e-05,
-      "loss": 0.7654,
+      "loss": 0.8593,
       "step": 13661
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0587414630892561e-05,
-      "loss": 0.7977,
+      "loss": 0.8994,
       "step": 13662
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0586238799044666e-05,
-      "loss": 0.708,
+      "loss": 0.8856,
       "step": 13663
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0585062959063538e-05,
-      "loss": 0.9195,
+      "loss": 0.9444,
       "step": 13664
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0583887110965497e-05,
-      "loss": 0.7663,
+      "loss": 0.8532,
       "step": 13665
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0582711254766844e-05,
-      "loss": 0.6397,
+      "loss": 0.8021,
       "step": 13666
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0581535390483904e-05,
-      "loss": 0.7909,
+      "loss": 0.8795,
       "step": 13667
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0580359518132985e-05,
-      "loss": 0.603,
+      "loss": 0.8189,
       "step": 13668
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0579183637730398e-05,
-      "loss": 0.8804,
+      "loss": 0.906,
       "step": 13669
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0578007749292462e-05,
-      "loss": 0.7745,
+      "loss": 0.7785,
       "step": 13670
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.057683185283549e-05,
-      "loss": 0.8412,
+      "loss": 0.8664,
       "step": 13671
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0575655948375792e-05,
-      "loss": 0.6171,
+      "loss": 0.8107,
       "step": 13672
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0574480035929688e-05,
-      "loss": 0.7953,
+      "loss": 0.856,
       "step": 13673
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0573304115513483e-05,
-      "loss": 0.9064,
+      "loss": 0.9286,
       "step": 13674
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0572128187143502e-05,
-      "loss": 0.7412,
+      "loss": 0.8238,
       "step": 13675
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.057095225083605e-05,
-      "loss": 0.909,
+      "loss": 0.9904,
       "step": 13676
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0569776306607449e-05,
-      "loss": 0.6253,
+      "loss": 0.7276,
       "step": 13677
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0568600354474007e-05,
-      "loss": 0.7511,
+      "loss": 0.7971,
       "step": 13678
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0567424394452045e-05,
-      "loss": 0.7095,
+      "loss": 0.7142,
       "step": 13679
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0566248426557873e-05,
-      "loss": 0.6371,
+      "loss": 0.7848,
       "step": 13680
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0565072450807807e-05,
-      "loss": 0.8194,
+      "loss": 0.9171,
       "step": 13681
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0563896467218162e-05,
-      "loss": 0.8461,
+      "loss": 0.8969,
       "step": 13682
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0562720475805255e-05,
-      "loss": 0.7605,
+      "loss": 0.8774,
       "step": 13683
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0561544476585401e-05,
-      "loss": 0.7142,
+      "loss": 0.8805,
       "step": 13684
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.056036846957491e-05,
-      "loss": 0.8526,
+      "loss": 1.0175,
       "step": 13685
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0559192454790106e-05,
-      "loss": 0.7112,
+      "loss": 0.8066,
       "step": 13686
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0558016432247296e-05,
-      "loss": 0.6023,
+      "loss": 0.7749,
       "step": 13687
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0556840401962802e-05,
-      "loss": 0.6943,
+      "loss": 0.7883,
       "step": 13688
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.055566436395294e-05,
-      "loss": 0.7805,
+      "loss": 0.8027,
       "step": 13689
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0554488318234018e-05,
-      "loss": 0.7218,
+      "loss": 0.8281,
       "step": 13690
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.055331226482236e-05,
-      "loss": 0.7844,
+      "loss": 0.8818,
       "step": 13691
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.055213620373428e-05,
-      "loss": 0.7552,
+      "loss": 0.7672,
       "step": 13692
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0550960134986094e-05,
-      "loss": 0.6064,
+      "loss": 0.7625,
       "step": 13693
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0549784058594117e-05,
-      "loss": 0.7383,
+      "loss": 0.8276,
       "step": 13694
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0548607974574666e-05,
-      "loss": 0.8263,
+      "loss": 0.9646,
       "step": 13695
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0547431882944057e-05,
-      "loss": 0.6949,
+      "loss": 0.8665,
       "step": 13696
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0546255783718608e-05,
-      "loss": 0.7066,
+      "loss": 0.8747,
       "step": 13697
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0545079676914636e-05,
-      "loss": 0.7353,
+      "loss": 0.814,
       "step": 13698
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0543903562548457e-05,
-      "loss": 0.6886,
+      "loss": 0.8383,
       "step": 13699
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0542727440636387e-05,
-      "loss": 0.6523,
+      "loss": 0.8762,
       "step": 13700
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0541551311194741e-05,
-      "loss": 0.7588,
+      "loss": 0.8782,
       "step": 13701
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0540375174239842e-05,
-      "loss": 0.78,
+      "loss": 0.933,
       "step": 13702
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0539199029788004e-05,
-      "loss": 0.8347,
+      "loss": 0.9417,
       "step": 13703
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0538022877855543e-05,
-      "loss": 0.8203,
+      "loss": 0.9141,
       "step": 13704
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.053684671845878e-05,
-      "loss": 0.9559,
+      "loss": 0.9796,
       "step": 13705
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0535670551614028e-05,
-      "loss": 0.7999,
+      "loss": 0.9304,
       "step": 13706
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.053449437733761e-05,
-      "loss": 0.7271,
+      "loss": 0.8746,
       "step": 13707
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.053331819564584e-05,
-      "loss": 0.73,
+      "loss": 0.8513,
       "step": 13708
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0532142006555032e-05,
-      "loss": 0.6824,
+      "loss": 0.8676,
       "step": 13709
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0530965810081515e-05,
-      "loss": 0.8057,
+      "loss": 0.8571,
       "step": 13710
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0529789606241598e-05,
-      "loss": 0.709,
+      "loss": 0.8035,
       "step": 13711
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0528613395051603e-05,
-      "loss": 0.7884,
+      "loss": 0.876,
       "step": 13712
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0527437176527848e-05,
-      "loss": 0.7758,
+      "loss": 0.8758,
       "step": 13713
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.052626095068665e-05,
-      "loss": 0.7707,
+      "loss": 0.9062,
       "step": 13714
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0525084717544325e-05,
-      "loss": 0.7767,
+      "loss": 0.8237,
       "step": 13715
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0523908477117197e-05,
-      "loss": 0.7191,
+      "loss": 0.9274,
       "step": 13716
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0522732229421582e-05,
-      "loss": 0.7697,
+      "loss": 0.8657,
       "step": 13717
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0521555974473799e-05,
-      "loss": 0.7457,
+      "loss": 0.83,
       "step": 13718
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0520379712290166e-05,
-      "loss": 0.7519,
+      "loss": 0.8625,
       "step": 13719
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0519203442887003e-05,
-      "loss": 0.8645,
+      "loss": 0.8618,
       "step": 13720
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0518027166280632e-05,
-      "loss": 0.6538,
+      "loss": 0.7694,
       "step": 13721
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0516850882487364e-05,
-      "loss": 0.8356,
+      "loss": 0.9351,
       "step": 13722
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0515674591523526e-05,
-      "loss": 0.7064,
+      "loss": 0.8663,
       "step": 13723
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0514498293405435e-05,
-      "loss": 0.774,
+      "loss": 0.8664,
       "step": 13724
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0513321988149408e-05,
-      "loss": 0.8053,
+      "loss": 0.8848,
       "step": 13725
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0512145675771768e-05,
-      "loss": 0.6929,
+      "loss": 0.83,
       "step": 13726
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0510969356288836e-05,
-      "loss": 0.7619,
+      "loss": 0.906,
       "step": 13727
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0509793029716924e-05,
-      "loss": 0.7016,
+      "loss": 0.8263,
       "step": 13728
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.050861669607236e-05,
-      "loss": 0.8738,
+      "loss": 0.9221,
       "step": 13729
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0507440355371463e-05,
-      "loss": 0.8594,
+      "loss": 0.9743,
       "step": 13730
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0506264007630548e-05,
-      "loss": 0.8488,
+      "loss": 0.967,
       "step": 13731
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.050508765286594e-05,
-      "loss": 0.763,
+      "loss": 0.9507,
       "step": 13732
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0503911291093956e-05,
-      "loss": 0.7974,
+      "loss": 0.8434,
       "step": 13733
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.050273492233092e-05,
-      "loss": 0.7719,
+      "loss": 0.9045,
       "step": 13734
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0501558546593144e-05,
-      "loss": 0.7219,
+      "loss": 0.8421,
       "step": 13735
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.050038216389696e-05,
-      "loss": 0.6701,
+      "loss": 0.7716,
       "step": 13736
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0499205774258685e-05,
-      "loss": 0.7575,
+      "loss": 0.8887,
       "step": 13737
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0498029377694633e-05,
-      "loss": 0.6932,
+      "loss": 0.8487,
       "step": 13738
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0496852974221134e-05,
-      "loss": 0.652,
+      "loss": 0.8251,
       "step": 13739
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0495676563854505e-05,
-      "loss": 0.8589,
+      "loss": 0.9176,
       "step": 13740
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0494500146611064e-05,
-      "loss": 0.6334,
+      "loss": 0.8028,
       "step": 13741
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0493323722507138e-05,
-      "loss": 0.715,
+      "loss": 0.8257,
       "step": 13742
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0492147291559045e-05,
-      "loss": 0.7345,
+      "loss": 0.8535,
       "step": 13743
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0490970853783105e-05,
-      "loss": 0.7467,
+      "loss": 0.8693,
       "step": 13744
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0489794409195641e-05,
-      "loss": 0.7504,
+      "loss": 0.9034,
       "step": 13745
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0488617957812974e-05,
-      "loss": 0.6921,
+      "loss": 0.9152,
       "step": 13746
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0487441499651428e-05,
-      "loss": 0.7174,
+      "loss": 0.8555,
       "step": 13747
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0486265034727323e-05,
-      "loss": 0.7342,
+      "loss": 0.8587,
       "step": 13748
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0485088563056977e-05,
-      "loss": 0.6227,
+      "loss": 0.7617,
       "step": 13749
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0483912084656722e-05,
-      "loss": 0.6562,
+      "loss": 0.8229,
       "step": 13750
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0482735599542867e-05,
-      "loss": 0.7182,
+      "loss": 0.841,
       "step": 13751
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0481559107731745e-05,
-      "loss": 0.722,
+      "loss": 0.8851,
       "step": 13752
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0480382609239672e-05,
-      "loss": 0.7381,
+      "loss": 0.8343,
       "step": 13753
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0479206104082969e-05,
-      "loss": 0.8065,
+      "loss": 0.8576,
       "step": 13754
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0478029592277966e-05,
-      "loss": 0.8547,
+      "loss": 0.8881,
       "step": 13755
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0476853073840977e-05,
-      "loss": 0.7556,
+      "loss": 0.8986,
       "step": 13756
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.047567654878833e-05,
-      "loss": 0.7166,
+      "loss": 0.9043,
       "step": 13757
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0474500017136345e-05,
-      "loss": 0.8696,
+      "loss": 0.8745,
       "step": 13758
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0473323478901346e-05,
-      "loss": 0.7227,
+      "loss": 0.8366,
       "step": 13759
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0472146934099654e-05,
-      "loss": 0.7283,
+      "loss": 0.8329,
       "step": 13760
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0470970382747594e-05,
-      "loss": 0.7159,
+      "loss": 0.8439,
       "step": 13761
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0469793824861487e-05,
-      "loss": 0.8076,
+      "loss": 0.9033,
       "step": 13762
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0468617260457659e-05,
-      "loss": 0.7472,
+      "loss": 0.8749,
       "step": 13763
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0467440689552433e-05,
-      "loss": 0.6476,
+      "loss": 0.727,
       "step": 13764
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0466264112162126e-05,
-      "loss": 0.7459,
+      "loss": 0.8134,
       "step": 13765
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.046508752830307e-05,
-      "loss": 0.7366,
+      "loss": 0.7372,
       "step": 13766
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0463910937991584e-05,
-      "loss": 0.7209,
+      "loss": 0.7964,
       "step": 13767
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.046273434124399e-05,
-      "loss": 0.7484,
+      "loss": 0.8322,
       "step": 13768
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0461557738076615e-05,
-      "loss": 0.7741,
+      "loss": 0.9101,
       "step": 13769
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0460381128505781e-05,
-      "loss": 0.6811,
+      "loss": 0.7904,
       "step": 13770
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0459204512547811e-05,
-      "loss": 0.7311,
+      "loss": 0.8229,
       "step": 13771
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0458027890219032e-05,
-      "loss": 0.713,
+      "loss": 0.8373,
       "step": 13772
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0456851261535763e-05,
-      "loss": 0.744,
+      "loss": 0.8602,
       "step": 13773
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0455674626514334e-05,
-      "loss": 0.7411,
+      "loss": 0.8118,
       "step": 13774
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0454497985171064e-05,
-      "loss": 0.7283,
+      "loss": 0.789,
       "step": 13775
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.045332133752228e-05,
-      "loss": 0.6363,
+      "loss": 0.844,
       "step": 13776
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0452144683584306e-05,
-      "loss": 0.8062,
+      "loss": 0.8673,
       "step": 13777
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0450968023373464e-05,
-      "loss": 0.7279,
+      "loss": 0.9364,
       "step": 13778
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0449791356906082e-05,
-      "loss": 0.7893,
+      "loss": 0.8396,
       "step": 13779
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0448614684198483e-05,
-      "loss": 0.7627,
+      "loss": 0.8505,
       "step": 13780
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0447438005266991e-05,
-      "loss": 0.6223,
+      "loss": 0.8047,
       "step": 13781
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0446261320127932e-05,
-      "loss": 0.8158,
+      "loss": 0.8485,
       "step": 13782
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0445084628797633e-05,
-      "loss": 0.7749,
+      "loss": 0.9075,
       "step": 13783
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0443907931292411e-05,
-      "loss": 0.7256,
+      "loss": 0.8165,
       "step": 13784
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0442731227628602e-05,
-      "loss": 0.6763,
+      "loss": 0.8332,
       "step": 13785
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0441554517822521e-05,
-      "loss": 0.8843,
+      "loss": 0.8833,
       "step": 13786
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.04403778018905e-05,
-      "loss": 0.677,
+      "loss": 0.7947,
       "step": 13787
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0439201079848864e-05,
-      "loss": 0.8023,
+      "loss": 0.8896,
       "step": 13788
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.043802435171393e-05,
-      "loss": 0.7528,
+      "loss": 0.8961,
       "step": 13789
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0436847617502035e-05,
-      "loss": 0.7913,
+      "loss": 0.8546,
       "step": 13790
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0435670877229496e-05,
-      "loss": 0.6451,
+      "loss": 0.7902,
       "step": 13791
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0434494130912642e-05,
-      "loss": 0.8155,
+      "loss": 0.8803,
       "step": 13792
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.04333173785678e-05,
-      "loss": 0.8628,
+      "loss": 0.8912,
       "step": 13793
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0432140620211292e-05,
-      "loss": 0.8016,
+      "loss": 0.8772,
       "step": 13794
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0430963855859448e-05,
-      "loss": 0.9167,
+      "loss": 0.9257,
       "step": 13795
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.042978708552859e-05,
-      "loss": 0.6981,
+      "loss": 0.8044,
       "step": 13796
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0428610309235047e-05,
-      "loss": 0.8488,
+      "loss": 0.889,
       "step": 13797
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0427433526995146e-05,
-      "loss": 0.6829,
+      "loss": 0.755,
       "step": 13798
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0426256738825208e-05,
-      "loss": 0.7566,
+      "loss": 0.9097,
       "step": 13799
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0425079944741563e-05,
-      "loss": 0.8534,
+      "loss": 0.8572,
       "step": 13800
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0423903144760538e-05,
-      "loss": 0.8104,
+      "loss": 0.8387,
       "step": 13801
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0422726338898459e-05,
-      "loss": 0.61,
+      "loss": 0.7337,
       "step": 13802
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.042154952717165e-05,
-      "loss": 0.7799,
+      "loss": 0.8682,
       "step": 13803
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0420372709596442e-05,
-      "loss": 0.6949,
+      "loss": 0.8039,
       "step": 13804
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0419195886189156e-05,
-      "loss": 0.6271,
+      "loss": 0.8034,
       "step": 13805
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0418019056966125e-05,
-      "loss": 0.6767,
+      "loss": 0.853,
       "step": 13806
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.041684222194367e-05,
-      "loss": 0.8132,
+      "loss": 0.7848,
       "step": 13807
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0415665381138123e-05,
-      "loss": 0.6384,
+      "loss": 0.7674,
       "step": 13808
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0414488534565807e-05,
-      "loss": 0.7107,
+      "loss": 0.7762,
       "step": 13809
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.041331168224305e-05,
-      "loss": 0.6846,
+      "loss": 0.8367,
       "step": 13810
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0412134824186181e-05,
-      "loss": 0.6319,
+      "loss": 0.7733,
       "step": 13811
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0410957960411527e-05,
-      "loss": 0.8377,
+      "loss": 0.8683,
       "step": 13812
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0409781090935412e-05,
-      "loss": 0.7259,
+      "loss": 0.858,
       "step": 13813
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0408604215774167e-05,
-      "loss": 0.6904,
+      "loss": 0.7984,
       "step": 13814
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.040742733494412e-05,
-      "loss": 0.7306,
+      "loss": 0.7541,
       "step": 13815
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0406250448461593e-05,
-      "loss": 0.8101,
+      "loss": 0.905,
       "step": 13816
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.040507355634292e-05,
-      "loss": 0.7402,
+      "loss": 0.8458,
       "step": 13817
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0403896658604425e-05,
-      "loss": 0.6889,
+      "loss": 0.7825,
       "step": 13818
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0402719755262436e-05,
-      "loss": 0.8759,
+      "loss": 0.906,
       "step": 13819
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0401542846333286e-05,
-      "loss": 0.7326,
+      "loss": 0.8586,
       "step": 13820
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0400365931833294e-05,
-      "loss": 0.657,
+      "loss": 0.7046,
       "step": 13821
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0399189011778795e-05,
-      "loss": 0.7734,
+      "loss": 0.8439,
       "step": 13822
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0398012086186116e-05,
-      "loss": 0.6936,
+      "loss": 0.8148,
       "step": 13823
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0396835155071581e-05,
-      "loss": 0.7003,
+      "loss": 0.8372,
       "step": 13824
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0395658218451521e-05,
-      "loss": 0.7426,
+      "loss": 0.7857,
       "step": 13825
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0394481276342268e-05,
-      "loss": 0.6854,
+      "loss": 0.8349,
       "step": 13826
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0393304328760145e-05,
-      "loss": 0.7283,
+      "loss": 0.7969,
       "step": 13827
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0392127375721483e-05,
-      "loss": 0.7725,
+      "loss": 0.8494,
       "step": 13828
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0390950417242608e-05,
-      "loss": 0.8257,
+      "loss": 0.9148,
       "step": 13829
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0389773453339856e-05,
-      "loss": 0.7425,
+      "loss": 0.7575,
       "step": 13830
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0388596484029546e-05,
-      "loss": 0.7555,
+      "loss": 0.7635,
       "step": 13831
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0387419509328011e-05,
-      "loss": 0.6416,
+      "loss": 0.8204,
       "step": 13832
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0386242529251582e-05,
-      "loss": 0.7461,
+      "loss": 0.9046,
       "step": 13833
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0385065543816583e-05,
-      "loss": 0.6287,
+      "loss": 0.7884,
       "step": 13834
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0383888553039347e-05,
-      "loss": 0.8656,
+      "loss": 0.9204,
       "step": 13835
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0382711556936204e-05,
-      "loss": 0.7645,
+      "loss": 0.814,
       "step": 13836
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0381534555523479e-05,
-      "loss": 0.8517,
+      "loss": 0.8933,
       "step": 13837
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0380357548817506e-05,
-      "loss": 0.6696,
+      "loss": 0.7607,
       "step": 13838
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.037918053683461e-05,
-      "loss": 0.5767,
+      "loss": 0.7575,
       "step": 13839
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.037800351959112e-05,
-      "loss": 0.7862,
+      "loss": 0.8691,
       "step": 13840
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0376826497103369e-05,
-      "loss": 0.7732,
+      "loss": 0.9083,
       "step": 13841
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0375649469387685e-05,
-      "loss": 0.6531,
+      "loss": 0.8735,
       "step": 13842
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.03744724364604e-05,
-      "loss": 0.8069,
+      "loss": 0.9404,
       "step": 13843
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0373295398337839e-05,
-      "loss": 0.7342,
+      "loss": 0.8001,
       "step": 13844
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0372118355036334e-05,
-      "loss": 0.7681,
+      "loss": 0.8696,
       "step": 13845
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0370941306572215e-05,
-      "loss": 0.7598,
+      "loss": 0.8657,
       "step": 13846
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.036976425296181e-05,
-      "loss": 0.6749,
+      "loss": 0.7493,
       "step": 13847
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0368587194221455e-05,
-      "loss": 0.7671,
+      "loss": 0.8604,
       "step": 13848
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0367410130367472e-05,
-      "loss": 0.7926,
+      "loss": 0.9167,
       "step": 13849
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0366233061416195e-05,
-      "loss": 0.7123,
+      "loss": 0.8552,
       "step": 13850
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0365055987383956e-05,
-      "loss": 0.6411,
+      "loss": 0.8745,
       "step": 13851
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0363878908287082e-05,
-      "loss": 0.6221,
+      "loss": 0.7074,
       "step": 13852
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0362701824141903e-05,
-      "loss": 0.7225,
+      "loss": 0.7685,
       "step": 13853
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0361524734964752e-05,
-      "loss": 0.686,
+      "loss": 0.8219,
       "step": 13854
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.036034764077196e-05,
-      "loss": 0.6269,
+      "loss": 0.8142,
       "step": 13855
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0359170541579853e-05,
-      "loss": 0.7259,
+      "loss": 0.8564,
       "step": 13856
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0357993437404765e-05,
-      "loss": 0.9142,
+      "loss": 1.0143,
       "step": 13857
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0356816328263024e-05,
-      "loss": 0.8681,
+      "loss": 0.8586,
       "step": 13858
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0355639214170966e-05,
-      "loss": 0.8094,
+      "loss": 0.9274,
       "step": 13859
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0354462095144917e-05,
-      "loss": 0.6813,
+      "loss": 0.8569,
       "step": 13860
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.035328497120121e-05,
-      "loss": 0.6866,
+      "loss": 0.8017,
       "step": 13861
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0352107842356176e-05,
-      "loss": 0.7175,
+      "loss": 0.7888,
       "step": 13862
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0350930708626144e-05,
-      "loss": 0.5855,
+      "loss": 0.7557,
       "step": 13863
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0349753570027446e-05,
-      "loss": 0.7653,
+      "loss": 0.9265,
       "step": 13864
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0348576426576416e-05,
-      "loss": 0.6964,
+      "loss": 0.8936,
       "step": 13865
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0347399278289379e-05,
-      "loss": 0.6867,
+      "loss": 0.855,
       "step": 13866
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0346222125182672e-05,
-      "loss": 0.6628,
+      "loss": 0.8117,
       "step": 13867
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0345044967272625e-05,
-      "loss": 0.7118,
+      "loss": 0.9235,
       "step": 13868
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0343867804575567e-05,
-      "loss": 0.8056,
+      "loss": 0.8594,
       "step": 13869
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0342690637107831e-05,
-      "loss": 0.6165,
+      "loss": 0.6674,
       "step": 13870
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.034151346488575e-05,
-      "loss": 0.7513,
+      "loss": 0.771,
       "step": 13871
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0340336287925652e-05,
-      "loss": 0.7608,
+      "loss": 0.8758,
       "step": 13872
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0339159106243875e-05,
-      "loss": 0.7216,
+      "loss": 0.8152,
       "step": 13873
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0337981919856742e-05,
-      "loss": 0.8589,
+      "loss": 0.8623,
       "step": 13874
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0336804728780592e-05,
-      "loss": 0.7068,
+      "loss": 0.8781,
       "step": 13875
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0335627533031754e-05,
-      "loss": 0.5674,
+      "loss": 0.7811,
       "step": 13876
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0334450332626558e-05,
-      "loss": 0.7817,
+      "loss": 0.8904,
       "step": 13877
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.033327312758134e-05,
-      "loss": 0.7148,
+      "loss": 0.757,
       "step": 13878
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.033209591791243e-05,
-      "loss": 0.7594,
+      "loss": 0.7825,
       "step": 13879
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0330918703636158e-05,
-      "loss": 0.8137,
+      "loss": 0.9042,
       "step": 13880
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0329741484768862e-05,
-      "loss": 0.8322,
+      "loss": 0.9128,
       "step": 13881
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0328564261326872e-05,
-      "loss": 0.8184,
+      "loss": 0.8513,
       "step": 13882
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0327387033326515e-05,
-      "loss": 0.652,
+      "loss": 0.8215,
       "step": 13883
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0326209800784133e-05,
-      "loss": 0.829,
+      "loss": 0.9351,
       "step": 13884
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.0325032563716047e-05,
-      "loss": 0.7697,
+      "loss": 0.868,
       "step": 13885
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0323855322138598e-05,
-      "loss": 0.674,
+      "loss": 0.8679,
       "step": 13886
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0322678076068116e-05,
-      "loss": 0.7866,
+      "loss": 0.8357,
       "step": 13887
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0321500825520932e-05,
-      "loss": 0.7626,
+      "loss": 0.9065,
       "step": 13888
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.032032357051338e-05,
-      "loss": 0.6836,
+      "loss": 0.8252,
       "step": 13889
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0319146311061791e-05,
-      "loss": 0.7115,
+      "loss": 0.8748,
       "step": 13890
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0317969047182505e-05,
-      "loss": 0.722,
+      "loss": 0.8467,
       "step": 13891
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0316791778891848e-05,
-      "loss": 0.9101,
+      "loss": 0.9015,
       "step": 13892
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0315614506206152e-05,
-      "loss": 0.7439,
+      "loss": 0.8999,
       "step": 13893
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0314437229141755e-05,
-      "loss": 0.7331,
+      "loss": 0.8533,
       "step": 13894
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0313259947714988e-05,
-      "loss": 0.7695,
+      "loss": 0.8417,
       "step": 13895
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.031208266194218e-05,
-      "loss": 0.7548,
+      "loss": 0.8354,
       "step": 13896
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0310905371839673e-05,
-      "loss": 0.8891,
+      "loss": 0.9963,
       "step": 13897
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.030972807742379e-05,
-      "loss": 0.6314,
+      "loss": 0.799,
       "step": 13898
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0308550778710873e-05,
-      "loss": 0.5944,
+      "loss": 0.6978,
       "step": 13899
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.030737347571725e-05,
-      "loss": 0.7363,
+      "loss": 0.9119,
       "step": 13900
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0306196168459255e-05,
-      "loss": 0.6943,
+      "loss": 0.7994,
       "step": 13901
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0305018856953226e-05,
-      "loss": 0.7488,
+      "loss": 0.8343,
       "step": 13902
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0303841541215491e-05,
-      "loss": 0.6939,
+      "loss": 0.8596,
       "step": 13903
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0302664221262389e-05,
-      "loss": 0.7396,
+      "loss": 0.8209,
       "step": 13904
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0301486897110246e-05,
-      "loss": 0.8499,
+      "loss": 0.9152,
       "step": 13905
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0300309568775402e-05,
-      "loss": 0.7555,
+      "loss": 0.8552,
       "step": 13906
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0299132236274188e-05,
-      "loss": 0.5947,
+      "loss": 0.7402,
       "step": 13907
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0297954899622939e-05,
-      "loss": 0.6983,
+      "loss": 0.8334,
       "step": 13908
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0296777558837988e-05,
-      "loss": 0.6807,
+      "loss": 0.8045,
       "step": 13909
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.029560021393567e-05,
-      "loss": 0.7049,
+      "loss": 0.7995,
       "step": 13910
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0294422864932318e-05,
-      "loss": 0.8225,
+      "loss": 0.9075,
       "step": 13911
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0293245511844268e-05,
-      "loss": 0.7897,
+      "loss": 0.9293,
       "step": 13912
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.029206815468785e-05,
-      "loss": 0.7717,
+      "loss": 0.8094,
       "step": 13913
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0290890793479402e-05,
-      "loss": 0.7558,
+      "loss": 0.8178,
       "step": 13914
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0289713428235257e-05,
-      "loss": 0.8154,
+      "loss": 0.9301,
       "step": 13915
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.028853605897175e-05,
-      "loss": 0.7003,
+      "loss": 0.8038,
       "step": 13916
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0287358685705213e-05,
-      "loss": 0.6688,
+      "loss": 0.79,
       "step": 13917
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0286181308451984e-05,
-      "loss": 0.7142,
+      "loss": 0.7952,
       "step": 13918
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0285003927228394e-05,
-      "loss": 0.6306,
+      "loss": 0.7117,
       "step": 13919
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0283826542050778e-05,
-      "loss": 0.7416,
+      "loss": 0.8345,
       "step": 13920
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0282649152935473e-05,
-      "loss": 0.7523,
+      "loss": 0.8352,
       "step": 13921
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0281471759898814e-05,
-      "loss": 0.7075,
+      "loss": 0.8454,
       "step": 13922
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.028029436295713e-05,
-      "loss": 0.6455,
+      "loss": 0.812,
       "step": 13923
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0279116962126763e-05,
-      "loss": 0.6332,
+      "loss": 0.7779,
       "step": 13924
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.027793955742404e-05,
-      "loss": 0.752,
+      "loss": 0.8176,
       "step": 13925
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0276762148865303e-05,
-      "loss": 0.8848,
+      "loss": 0.8919,
       "step": 13926
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0275584736466882e-05,
-      "loss": 0.7659,
+      "loss": 0.8388,
       "step": 13927
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0274407320245112e-05,
-      "loss": 0.7215,
+      "loss": 0.8489,
       "step": 13928
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0273229900216335e-05,
-      "loss": 0.7001,
+      "loss": 0.81,
       "step": 13929
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0272052476396876e-05,
-      "loss": 0.75,
+      "loss": 0.8263,
       "step": 13930
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0270875048803077e-05,
-      "loss": 0.8496,
+      "loss": 0.9269,
       "step": 13931
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0269697617451271e-05,
-      "loss": 0.835,
+      "loss": 0.9408,
       "step": 13932
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0268520182357792e-05,
-      "loss": 0.7194,
+      "loss": 0.8582,
       "step": 13933
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.026734274353898e-05,
-      "loss": 0.762,
+      "loss": 0.8814,
       "step": 13934
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0266165301011163e-05,
-      "loss": 0.7858,
+      "loss": 0.8018,
       "step": 13935
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.026498785479068e-05,
-      "loss": 0.6079,
+      "loss": 0.7228,
       "step": 13936
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.026381040489387e-05,
-      "loss": 0.7358,
+      "loss": 0.7956,
       "step": 13937
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0262632951337064e-05,
-      "loss": 0.6883,
+      "loss": 0.856,
       "step": 13938
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0261455494136594e-05,
-      "loss": 0.6667,
+      "loss": 0.7876,
       "step": 13939
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0260278033308804e-05,
-      "loss": 0.8034,
+      "loss": 0.8302,
       "step": 13940
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0259100568870026e-05,
-      "loss": 0.8813,
+      "loss": 0.9787,
       "step": 13941
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0257923100836596e-05,
-      "loss": 0.7988,
+      "loss": 0.8814,
       "step": 13942
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0256745629224846e-05,
-      "loss": 0.7647,
+      "loss": 0.9153,
       "step": 13943
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0255568154051115e-05,
-      "loss": 0.6363,
+      "loss": 0.7691,
       "step": 13944
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.025439067533174e-05,
-      "loss": 0.8133,
+      "loss": 0.9082,
       "step": 13945
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0253213193083052e-05,
-      "loss": 0.7177,
+      "loss": 0.8253,
       "step": 13946
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0252035707321393e-05,
-      "loss": 0.9934,
+      "loss": 0.9872,
       "step": 13947
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0250858218063098e-05,
-      "loss": 0.6051,
+      "loss": 0.7851,
       "step": 13948
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0249680725324496e-05,
-      "loss": 0.6561,
+      "loss": 0.7522,
       "step": 13949
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0248503229121933e-05,
-      "loss": 0.6512,
+      "loss": 0.7432,
       "step": 13950
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0247325729471738e-05,
-      "loss": 0.6454,
+      "loss": 0.792,
       "step": 13951
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.024614822639025e-05,
-      "loss": 0.774,
+      "loss": 0.802,
       "step": 13952
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0244970719893804e-05,
-      "loss": 0.7336,
+      "loss": 0.9244,
       "step": 13953
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0243793209998739e-05,
-      "loss": 0.7721,
+      "loss": 0.8849,
       "step": 13954
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0242615696721386e-05,
-      "loss": 0.8786,
+      "loss": 0.9581,
       "step": 13955
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0241438180078085e-05,
-      "loss": 0.7498,
+      "loss": 0.8643,
       "step": 13956
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0240260660085171e-05,
-      "loss": 0.7853,
+      "loss": 0.8764,
       "step": 13957
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0239083136758982e-05,
-      "loss": 0.9007,
+      "loss": 0.957,
       "step": 13958
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0237905610115858e-05,
-      "loss": 0.8777,
+      "loss": 0.86,
       "step": 13959
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0236728080172127e-05,
-      "loss": 0.714,
+      "loss": 0.857,
       "step": 13960
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0235550546944132e-05,
-      "loss": 0.8165,
+      "loss": 0.9075,
       "step": 13961
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0234373010448205e-05,
-      "loss": 0.7035,
+      "loss": 0.8347,
       "step": 13962
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0233195470700685e-05,
-      "loss": 0.5874,
+      "loss": 0.8016,
       "step": 13963
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.023201792771791e-05,
-      "loss": 0.7517,
+      "loss": 0.8242,
       "step": 13964
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0230840381516214e-05,
-      "loss": 0.8082,
+      "loss": 0.8378,
       "step": 13965
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0229662832111937e-05,
-      "loss": 0.7798,
+      "loss": 0.8548,
       "step": 13966
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0228485279521414e-05,
-      "loss": 0.6301,
+      "loss": 0.8568,
       "step": 13967
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.022730772376098e-05,
-      "loss": 0.5327,
+      "loss": 0.7125,
       "step": 13968
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0226130164846974e-05,
-      "loss": 0.7292,
+      "loss": 0.8965,
       "step": 13969
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0224952602795737e-05,
-      "loss": 0.8081,
+      "loss": 0.8689,
       "step": 13970
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0223775037623596e-05,
-      "loss": 0.6746,
+      "loss": 0.8098,
       "step": 13971
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0222597469346898e-05,
-      "loss": 0.7088,
+      "loss": 0.8516,
       "step": 13972
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0221419897981973e-05,
-      "loss": 0.8596,
+      "loss": 0.8346,
       "step": 13973
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0220242323545162e-05,
-      "loss": 0.8089,
+      "loss": 0.8558,
       "step": 13974
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0219064746052803e-05,
-      "loss": 0.5852,
+      "loss": 0.7337,
       "step": 13975
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0217887165521228e-05,
-      "loss": 0.7051,
+      "loss": 0.7659,
       "step": 13976
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0216709581966781e-05,
-      "loss": 0.8095,
+      "loss": 0.9186,
       "step": 13977
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0215531995405797e-05,
-      "loss": 0.7901,
+      "loss": 0.9064,
       "step": 13978
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.021435440585461e-05,
-      "loss": 0.9125,
+      "loss": 0.8959,
       "step": 13979
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0213176813329559e-05,
-      "loss": 0.792,
+      "loss": 0.8484,
       "step": 13980
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0211999217846984e-05,
-      "loss": 0.7596,
+      "loss": 0.8988,
       "step": 13981
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0210821619423219e-05,
-      "loss": 0.8502,
+      "loss": 0.9,
       "step": 13982
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0209644018074604e-05,
-      "loss": 0.6636,
+      "loss": 0.8163,
       "step": 13983
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0208466413817475e-05,
-      "loss": 0.6812,
+      "loss": 0.8854,
       "step": 13984
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0207288806668172e-05,
-      "loss": 0.7669,
+      "loss": 0.8728,
       "step": 13985
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.020611119664303e-05,
-      "loss": 0.6788,
+      "loss": 0.8339,
       "step": 13986
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0204933583758387e-05,
-      "loss": 0.7642,
+      "loss": 0.8557,
       "step": 13987
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0203755968030581e-05,
-      "loss": 0.7491,
+      "loss": 0.782,
       "step": 13988
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0202578349475951e-05,
-      "loss": 0.7826,
+      "loss": 0.9215,
       "step": 13989
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0201400728110833e-05,
-      "loss": 0.7324,
+      "loss": 0.8396,
       "step": 13990
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0200223103951567e-05,
-      "loss": 0.6806,
+      "loss": 0.8117,
       "step": 13991
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0199045477014488e-05,
-      "loss": 0.6465,
+      "loss": 0.8263,
       "step": 13992
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0197867847315937e-05,
-      "loss": 0.7305,
+      "loss": 0.8656,
       "step": 13993
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.019669021487225e-05,
-      "loss": 0.798,
+      "loss": 0.9031,
       "step": 13994
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0195512579699764e-05,
-      "loss": 0.7646,
+      "loss": 0.9565,
       "step": 13995
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0194334941814819e-05,
-      "loss": 0.7972,
+      "loss": 0.8719,
       "step": 13996
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.019315730123375e-05,
-      "loss": 0.6791,
+      "loss": 0.7874,
       "step": 13997
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.01919796579729e-05,
-      "loss": 0.6411,
+      "loss": 0.6925,
       "step": 13998
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0190802012048611e-05,
-      "loss": 0.8001,
+      "loss": 0.8543,
       "step": 13999
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0189624363477207e-05,
-      "loss": 0.7549,
+      "loss": 0.9225,
       "step": 14000
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0188446712275035e-05,
-      "loss": 0.6016,
+      "loss": 0.7972,
       "step": 14001
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0187269058458431e-05,
-      "loss": 0.7079,
+      "loss": 0.8649,
       "step": 14002
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0186091402043739e-05,
-      "loss": 0.803,
+      "loss": 0.8867,
       "step": 14003
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0184913743047288e-05,
-      "loss": 0.8393,
+      "loss": 0.8976,
       "step": 14004
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0183736081485422e-05,
-      "loss": 0.8866,
+      "loss": 0.9214,
       "step": 14005
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0182558417374482e-05,
-      "loss": 0.5567,
+      "loss": 0.7819,
       "step": 14006
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0181380750730801e-05,
-      "loss": 0.7762,
+      "loss": 0.903,
       "step": 14007
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0180203081570719e-05,
-      "loss": 0.7989,
+      "loss": 0.8656,
       "step": 14008
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0179025409910575e-05,
-      "loss": 0.7581,
+      "loss": 0.8113,
       "step": 14009
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0177847735766709e-05,
-      "loss": 0.684,
+      "loss": 0.8619,
       "step": 14010
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0176670059155456e-05,
-      "loss": 0.6659,
+      "loss": 0.8212,
       "step": 14011
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0175492380093158e-05,
-      "loss": 0.5227,
+      "loss": 0.7103,
       "step": 14012
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0174314698596148e-05,
-      "loss": 0.7997,
+      "loss": 0.8466,
       "step": 14013
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0173137014680774e-05,
-      "loss": 0.683,
+      "loss": 0.8279,
       "step": 14014
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0171959328363367e-05,
-      "loss": 0.7717,
+      "loss": 0.9063,
       "step": 14015
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.017078163966027e-05,
-      "loss": 0.6206,
+      "loss": 0.7536,
       "step": 14016
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0169603948587818e-05,
-      "loss": 0.7017,
+      "loss": 0.7844,
       "step": 14017
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0168426255162353e-05,
-      "loss": 0.7947,
+      "loss": 0.9353,
       "step": 14018
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0167248559400216e-05,
-      "loss": 0.7087,
+      "loss": 0.8318,
       "step": 14019
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0166070861317738e-05,
-      "loss": 0.6965,
+      "loss": 0.8566,
       "step": 14020
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0164893160931263e-05,
-      "loss": 0.7586,
+      "loss": 0.7909,
       "step": 14021
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.016371545825713e-05,
-      "loss": 0.7857,
+      "loss": 0.8333,
       "step": 14022
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0162537753311678e-05,
-      "loss": 0.7808,
+      "loss": 0.8656,
       "step": 14023
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0161360046111243e-05,
-      "loss": 0.7196,
+      "loss": 0.8515,
       "step": 14024
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0160182336672171e-05,
-      "loss": 0.8337,
+      "loss": 0.8521,
       "step": 14025
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0159004625010794e-05,
-      "loss": 0.7909,
+      "loss": 0.8466,
       "step": 14026
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0157826911143449e-05,
-      "loss": 0.7105,
+      "loss": 0.8643,
       "step": 14027
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0156649195086485e-05,
-      "loss": 0.6896,
+      "loss": 0.8604,
       "step": 14028
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0155471476856234e-05,
-      "loss": 0.6279,
+      "loss": 0.8347,
       "step": 14029
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0154293756469036e-05,
-      "loss": 0.6915,
+      "loss": 0.7618,
       "step": 14030
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0153116033941233e-05,
-      "loss": 0.8119,
+      "loss": 0.9541,
       "step": 14031
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.015193830928916e-05,
-      "loss": 0.8544,
+      "loss": 0.898,
       "step": 14032
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.015076058252916e-05,
-      "loss": 0.7782,
+      "loss": 0.8724,
       "step": 14033
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0149582853677571e-05,
-      "loss": 0.7925,
+      "loss": 0.8606,
       "step": 14034
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.014840512275073e-05,
-      "loss": 0.7631,
+      "loss": 0.8941,
       "step": 14035
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.014722738976498e-05,
-      "loss": 0.7162,
+      "loss": 0.8529,
       "step": 14036
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0146049654736661e-05,
-      "loss": 0.691,
+      "loss": 0.8314,
       "step": 14037
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.014487191768211e-05,
-      "loss": 0.8073,
+      "loss": 0.8752,
       "step": 14038
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0143694178617664e-05,
-      "loss": 0.6224,
+      "loss": 0.7832,
       "step": 14039
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0142516437559664e-05,
-      "loss": 0.7504,
+      "loss": 0.8925,
       "step": 14040
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0141338694524455e-05,
-      "loss": 0.7654,
+      "loss": 0.8604,
       "step": 14041
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0140160949528366e-05,
-      "loss": 0.7955,
+      "loss": 0.9326,
       "step": 14042
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0138983202587749e-05,
-      "loss": 0.8691,
+      "loss": 0.9248,
       "step": 14043
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0137805453718934e-05,
-      "loss": 0.7601,
+      "loss": 0.9147,
       "step": 14044
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0136627702938265e-05,
-      "loss": 0.6918,
+      "loss": 0.7943,
       "step": 14045
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.013544995026208e-05,
-      "loss": 0.6715,
+      "loss": 0.8537,
       "step": 14046
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0134272195706717e-05,
-      "loss": 0.728,
+      "loss": 0.8985,
       "step": 14047
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.013309443928852e-05,
-      "loss": 0.6998,
+      "loss": 0.7982,
       "step": 14048
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0131916681023828e-05,
-      "loss": 0.6678,
+      "loss": 0.7717,
       "step": 14049
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0130738920928977e-05,
-      "loss": 0.536,
+      "loss": 0.7476,
       "step": 14050
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0129561159020309e-05,
-      "loss": 0.8643,
+      "loss": 0.9241,
       "step": 14051
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0128383395314166e-05,
-      "loss": 0.7236,
+      "loss": 0.8371,
       "step": 14052
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0127205629826885e-05,
-      "loss": 0.7078,
+      "loss": 0.8771,
       "step": 14053
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0126027862574804e-05,
-      "loss": 0.7685,
+      "loss": 0.8568,
       "step": 14054
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0124850093574266e-05,
-      "loss": 0.8008,
+      "loss": 0.9153,
       "step": 14055
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0123672322841611e-05,
-      "loss": 0.7837,
+      "loss": 0.8555,
       "step": 14056
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0122494550393178e-05,
-      "loss": 0.6541,
+      "loss": 0.7788,
       "step": 14057
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0121316776245304e-05,
-      "loss": 0.9002,
+      "loss": 0.972,
       "step": 14058
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0120139000414338e-05,
-      "loss": 0.8756,
+      "loss": 0.9431,
       "step": 14059
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.011896122291661e-05,
-      "loss": 0.7268,
+      "loss": 0.8111,
       "step": 14060
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0117783443768462e-05,
-      "loss": 0.6563,
+      "loss": 0.8031,
       "step": 14061
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0116605662986238e-05,
-      "loss": 0.7679,
+      "loss": 0.8798,
       "step": 14062
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0115427880586277e-05,
-      "loss": 0.9053,
+      "loss": 0.9323,
       "step": 14063
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0114250096584914e-05,
-      "loss": 0.8878,
+      "loss": 0.9728,
       "step": 14064
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0113072310998497e-05,
-      "loss": 0.6966,
+      "loss": 0.8002,
       "step": 14065
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0111894523843361e-05,
-      "loss": 0.7319,
+      "loss": 0.8632,
       "step": 14066
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0110716735135846e-05,
-      "loss": 0.6496,
+      "loss": 0.7957,
       "step": 14067
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0109538944892294e-05,
-      "loss": 0.6998,
+      "loss": 0.9044,
       "step": 14068
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0108361153129045e-05,
-      "loss": 0.7886,
+      "loss": 0.8823,
       "step": 14069
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0107183359862438e-05,
-      "loss": 0.8053,
+      "loss": 0.9071,
       "step": 14070
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0106005565108817e-05,
-      "loss": 0.7107,
+      "loss": 0.8285,
       "step": 14071
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0104827768884516e-05,
-      "loss": 0.7141,
+      "loss": 0.8567,
       "step": 14072
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0103649971205878e-05,
-      "loss": 0.7285,
+      "loss": 0.884,
       "step": 14073
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0102472172089246e-05,
-      "loss": 0.7574,
+      "loss": 0.8538,
       "step": 14074
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0101294371550958e-05,
-      "loss": 0.6211,
+      "loss": 0.8151,
       "step": 14075
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0100116569607355e-05,
-      "loss": 0.8233,
+      "loss": 0.8925,
       "step": 14076
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0098938766274773e-05,
-      "loss": 0.7894,
+      "loss": 0.9202,
       "step": 14077
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0097760961569558e-05,
-      "loss": 0.809,
+      "loss": 0.8097,
       "step": 14078
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0096583155508049e-05,
-      "loss": 0.7192,
+      "loss": 0.8003,
       "step": 14079
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0095405348106584e-05,
-      "loss": 0.7895,
+      "loss": 0.7871,
       "step": 14080
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0094227539381507e-05,
-      "loss": 0.7221,
+      "loss": 0.8971,
       "step": 14081
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0093049729349156e-05,
-      "loss": 0.7312,
+      "loss": 0.8494,
       "step": 14082
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.009187191802587e-05,
-      "loss": 0.7339,
+      "loss": 0.8548,
       "step": 14083
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0090694105427994e-05,
-      "loss": 0.8806,
+      "loss": 0.9164,
       "step": 14084
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0089516291571864e-05,
-      "loss": 0.7989,
+      "loss": 0.9266,
       "step": 14085
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0088338476473824e-05,
-      "loss": 0.8593,
+      "loss": 0.9586,
       "step": 14086
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0087160660150213e-05,
-      "loss": 0.7683,
+      "loss": 0.841,
       "step": 14087
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.008598284261737e-05,
-      "loss": 0.7442,
+      "loss": 0.868,
       "step": 14088
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0084805023891639e-05,
-      "loss": 0.6713,
+      "loss": 0.7215,
       "step": 14089
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0083627203989359e-05,
-      "loss": 0.7548,
+      "loss": 0.8816,
       "step": 14090
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0082449382926866e-05,
-      "loss": 0.7009,
+      "loss": 0.7988,
       "step": 14091
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0081271560720509e-05,
-      "loss": 0.805,
+      "loss": 0.9135,
       "step": 14092
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0080093737386626e-05,
-      "loss": 0.708,
+      "loss": 0.7996,
       "step": 14093
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0078915912941554e-05,
-      "loss": 0.8077,
+      "loss": 0.9247,
       "step": 14094
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0077738087401634e-05,
-      "loss": 0.6641,
+      "loss": 0.8113,
       "step": 14095
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0076560260783208e-05,
-      "loss": 0.7554,
+      "loss": 0.8775,
       "step": 14096
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0075382433102619e-05,
-      "loss": 0.7325,
+      "loss": 0.8543,
       "step": 14097
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0074204604376206e-05,
-      "loss": 0.6176,
+      "loss": 0.7909,
       "step": 14098
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0073026774620307e-05,
-      "loss": 0.7477,
+      "loss": 0.8054,
       "step": 14099
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0071848943851268e-05,
-      "loss": 0.6675,
+      "loss": 0.8353,
       "step": 14100
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0070671112085421e-05,
-      "loss": 0.6916,
+      "loss": 0.7192,
       "step": 14101
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.006949327933912e-05,
-      "loss": 0.7197,
+      "loss": 0.8905,
       "step": 14102
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0068315445628696e-05,
-      "loss": 0.8605,
+      "loss": 0.8923,
       "step": 14103
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.006713761097049e-05,
-      "loss": 0.6363,
+      "loss": 0.785,
       "step": 14104
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0065959775380847e-05,
-      "loss": 0.82,
+      "loss": 0.9113,
       "step": 14105
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0064781938876105e-05,
-      "loss": 0.7027,
+      "loss": 0.7729,
       "step": 14106
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0063604101472605e-05,
-      "loss": 0.5807,
+      "loss": 0.7444,
       "step": 14107
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0062426263186689e-05,
-      "loss": 0.7583,
+      "loss": 0.8521,
       "step": 14108
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0061248424034698e-05,
-      "loss": 0.5732,
+      "loss": 0.6809,
       "step": 14109
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0060070584032968e-05,
-      "loss": 0.964,
+      "loss": 0.9662,
       "step": 14110
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0058892743197847e-05,
-      "loss": 0.7814,
+      "loss": 0.877,
       "step": 14111
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0057714901545672e-05,
-      "loss": 0.6436,
+      "loss": 0.755,
       "step": 14112
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0056537059092787e-05,
-      "loss": 0.6907,
+      "loss": 0.8109,
       "step": 14113
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.005535921585553e-05,
-      "loss": 0.7938,
+      "loss": 0.855,
       "step": 14114
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.005418137185024e-05,
-      "loss": 0.7549,
+      "loss": 0.8195,
       "step": 14115
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0053003527093263e-05,
-      "loss": 0.7991,
+      "loss": 0.8847,
       "step": 14116
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0051825681600936e-05,
-      "loss": 0.9279,
+      "loss": 0.9666,
       "step": 14117
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.00506478353896e-05,
-      "loss": 0.7908,
+      "loss": 0.8752,
       "step": 14118
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0049469988475599e-05,
-      "loss": 0.6792,
+      "loss": 0.8647,
       "step": 14119
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.004829214087527e-05,
-      "loss": 0.7698,
+      "loss": 0.8122,
       "step": 14120
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0047114292604957e-05,
-      "loss": 0.7272,
+      "loss": 0.8553,
       "step": 14121
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0045936443681001e-05,
-      "loss": 0.7474,
+      "loss": 0.7372,
       "step": 14122
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0044758594119742e-05,
-      "loss": 0.6334,
+      "loss": 0.7476,
       "step": 14123
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.004358074393752e-05,
-      "loss": 0.7708,
+      "loss": 0.8521,
       "step": 14124
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0042402893150679e-05,
-      "loss": 0.7513,
+      "loss": 0.8074,
       "step": 14125
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0041225041775553e-05,
-      "loss": 0.709,
+      "loss": 0.9086,
       "step": 14126
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0040047189828492e-05,
-      "loss": 0.7306,
+      "loss": 0.8628,
       "step": 14127
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0038869337325831e-05,
-      "loss": 0.6207,
+      "loss": 0.7294,
       "step": 14128
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0037691484283916e-05,
-      "loss": 0.6838,
+      "loss": 0.7995,
       "step": 14129
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0036513630719084e-05,
-      "loss": 0.7932,
+      "loss": 0.9076,
       "step": 14130
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0035335776647677e-05,
-      "loss": 0.7488,
+      "loss": 0.8957,
       "step": 14131
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0034157922086037e-05,
-      "loss": 0.7317,
+      "loss": 0.8452,
       "step": 14132
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0032980067050504e-05,
-      "loss": 0.6973,
+      "loss": 0.833,
       "step": 14133
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.003180221155742e-05,
-      "loss": 0.7451,
+      "loss": 0.8108,
       "step": 14134
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0030624355623125e-05,
-      "loss": 0.8,
+      "loss": 0.9045,
       "step": 14135
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0029446499263957e-05,
-      "loss": 0.858,
+      "loss": 0.96,
       "step": 14136
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0028268642496264e-05,
-      "loss": 0.6881,
+      "loss": 0.8832,
       "step": 14137
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0027090785336384e-05,
-      "loss": 0.7595,
+      "loss": 0.8465,
       "step": 14138
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0025912927800655e-05,
-      "loss": 0.7468,
+      "loss": 0.8022,
       "step": 14139
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0024735069905424e-05,
-      "loss": 0.7892,
+      "loss": 0.854,
       "step": 14140
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0023557211667026e-05,
-      "loss": 0.8367,
+      "loss": 0.9225,
       "step": 14141
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0022379353101808e-05,
-      "loss": 0.8009,
+      "loss": 0.9131,
       "step": 14142
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0021201494226107e-05,
-      "loss": 0.65,
+      "loss": 0.805,
       "step": 14143
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0020023635056264e-05,
-      "loss": 0.7288,
+      "loss": 0.8804,
       "step": 14144
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0018845775608623e-05,
-      "loss": 0.6404,
+      "loss": 0.8139,
       "step": 14145
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0017667915899525e-05,
-      "loss": 0.858,
+      "loss": 0.9303,
       "step": 14146
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0016490055945307e-05,
-      "loss": 0.7558,
+      "loss": 0.9183,
       "step": 14147
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0015312195762314e-05,
-      "loss": 0.7163,
+      "loss": 0.7504,
       "step": 14148
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0014134335366887e-05,
-      "loss": 0.7985,
+      "loss": 0.9587,
       "step": 14149
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0012956474775364e-05,
-      "loss": 0.7888,
+      "loss": 0.8772,
       "step": 14150
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0011778614004091e-05,
-      "loss": 0.7858,
+      "loss": 0.8705,
       "step": 14151
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0010600753069406e-05,
-      "loss": 0.6819,
+      "loss": 0.7537,
       "step": 14152
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.000942289198765e-05,
-      "loss": 0.8626,
+      "loss": 0.9039,
       "step": 14153
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0008245030775165e-05,
-      "loss": 0.6058,
+      "loss": 0.8636,
       "step": 14154
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.000706716944829e-05,
-      "loss": 0.701,
+      "loss": 0.823,
       "step": 14155
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0005889308023369e-05,
-      "loss": 0.7176,
+      "loss": 0.8794,
       "step": 14156
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0004711446516741e-05,
-      "loss": 0.8416,
+      "loss": 0.9254,
       "step": 14157
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0003533584944752e-05,
-      "loss": 0.7274,
+      "loss": 0.8092,
       "step": 14158
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.0002355723323737e-05,
-      "loss": 0.7265,
+      "loss": 0.8747,
       "step": 14159
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.000117786167004e-05,
-      "loss": 0.6717,
+      "loss": 0.834,
       "step": 14160
     },
     {
       "epoch": 0.52,
       "learning_rate": 1e-05,
-      "loss": 0.8412,
+      "loss": 0.9048,
       "step": 14161
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.998822138329964e-06,
-      "loss": 0.6908,
+      "loss": 0.793,
       "step": 14162
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.997644276676266e-06,
-      "loss": 0.7767,
+      "loss": 0.7789,
       "step": 14163
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.996466415055253e-06,
-      "loss": 0.624,
+      "loss": 0.7658,
       "step": 14164
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.99528855348326e-06,
-      "loss": 0.6998,
+      "loss": 0.8438,
       "step": 14165
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.994110691976635e-06,
-      "loss": 0.7729,
+      "loss": 0.8807,
       "step": 14166
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.992932830551714e-06,
-      "loss": 0.8152,
+      "loss": 0.8757,
       "step": 14167
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.99175496922484e-06,
-      "loss": 0.6748,
+      "loss": 0.7774,
       "step": 14168
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.990577108012353e-06,
-      "loss": 0.9202,
+      "loss": 1.0109,
       "step": 14169
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.9893992469306e-06,
-      "loss": 0.8245,
+      "loss": 0.8796,
       "step": 14170
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.98822138599591e-06,
-      "loss": 0.8244,
+      "loss": 0.8695,
       "step": 14171
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.98704352522464e-06,
-      "loss": 0.7875,
+      "loss": 0.8798,
       "step": 14172
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.985865664633116e-06,
-      "loss": 0.7239,
+      "loss": 0.8907,
       "step": 14173
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.984687804237689e-06,
-      "loss": 0.6582,
+      "loss": 0.7864,
       "step": 14174
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.983509944054693e-06,
-      "loss": 0.8458,
+      "loss": 0.9996,
       "step": 14175
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.982332084100478e-06,
-      "loss": 0.7343,
+      "loss": 0.8623,
       "step": 14176
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.981154224391377e-06,
-      "loss": 0.767,
+      "loss": 0.8938,
       "step": 14177
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.979976364943739e-06,
-      "loss": 0.8027,
+      "loss": 0.8208,
       "step": 14178
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.978798505773894e-06,
-      "loss": 0.7421,
+      "loss": 0.8203,
       "step": 14179
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.977620646898195e-06,
-      "loss": 0.7745,
+      "loss": 0.8852,
       "step": 14180
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.976442788332976e-06,
-      "loss": 0.6069,
+      "loss": 0.8083,
       "step": 14181
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.97526493009458e-06,
-      "loss": 0.5979,
+      "loss": 0.7758,
       "step": 14182
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.97408707219935e-06,
-      "loss": 0.7319,
+      "loss": 0.8638,
       "step": 14183
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.97290921466362e-06,
-      "loss": 0.6765,
+      "loss": 0.7942,
       "step": 14184
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.971731357503741e-06,
-      "loss": 0.6642,
+      "loss": 0.8123,
       "step": 14185
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.970553500736045e-06,
-      "loss": 0.6398,
+      "loss": 0.7817,
       "step": 14186
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.96937564437688e-06,
-      "loss": 0.7832,
+      "loss": 0.9161,
       "step": 14187
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.968197788442585e-06,
-      "loss": 0.7347,
+      "loss": 0.8186,
       "step": 14188
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.9670199329495e-06,
-      "loss": 0.6187,
+      "loss": 0.7604,
       "step": 14189
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.965842077913963e-06,
-      "loss": 0.7928,
+      "loss": 0.8965,
       "step": 14190
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.964664223352327e-06,
-      "loss": 0.7926,
+      "loss": 0.8271,
       "step": 14191
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.963486369280919e-06,
-      "loss": 0.8241,
+      "loss": 0.9373,
       "step": 14192
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.962308515716086e-06,
-      "loss": 0.8514,
+      "loss": 0.9503,
       "step": 14193
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.961130662674169e-06,
-      "loss": 0.7676,
+      "loss": 0.8887,
       "step": 14194
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.95995281017151e-06,
-      "loss": 0.6834,
+      "loss": 0.8366,
       "step": 14195
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.958774958224447e-06,
-      "loss": 0.8381,
+      "loss": 0.9226,
       "step": 14196
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.957597106849325e-06,
-      "loss": 0.8714,
+      "loss": 0.9155,
       "step": 14197
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.95641925606248e-06,
-      "loss": 0.7278,
+      "loss": 0.8791,
       "step": 14198
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.955241405880261e-06,
-      "loss": 0.7369,
+      "loss": 0.81,
       "step": 14199
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.954063556319004e-06,
-      "loss": 0.6999,
+      "loss": 0.8634,
       "step": 14200
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.952885707395044e-06,
-      "loss": 0.7041,
+      "loss": 0.8175,
       "step": 14201
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.951707859124733e-06,
-      "loss": 0.7472,
+      "loss": 0.8911,
       "step": 14202
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.950530011524404e-06,
-      "loss": 0.7684,
+      "loss": 0.8831,
       "step": 14203
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.949352164610405e-06,
-      "loss": 0.7068,
+      "loss": 0.8178,
       "step": 14204
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.948174318399068e-06,
-      "loss": 0.63,
+      "loss": 0.7998,
       "step": 14205
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.946996472906742e-06,
-      "loss": 0.7141,
+      "loss": 0.7881,
       "step": 14206
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.945818628149762e-06,
-      "loss": 0.7281,
+      "loss": 0.8379,
       "step": 14207
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.944640784144474e-06,
-      "loss": 0.7829,
+      "loss": 0.9078,
       "step": 14208
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.943462940907213e-06,
-      "loss": 0.6819,
+      "loss": 0.7674,
       "step": 14209
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.942285098454331e-06,
-      "loss": 0.668,
+      "loss": 0.8133,
       "step": 14210
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.941107256802153e-06,
-      "loss": 0.765,
+      "loss": 0.8427,
       "step": 14211
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.939929415967034e-06,
-      "loss": 0.7234,
+      "loss": 0.8227,
       "step": 14212
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.938751575965306e-06,
-      "loss": 0.8771,
+      "loss": 0.8876,
       "step": 14213
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.937573736813314e-06,
-      "loss": 0.7757,
+      "loss": 0.8691,
       "step": 14214
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.936395898527397e-06,
-      "loss": 0.6657,
+      "loss": 0.7568,
       "step": 14215
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.935218061123898e-06,
-      "loss": 0.7385,
+      "loss": 0.931,
       "step": 14216
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.934040224619154e-06,
-      "loss": 0.8016,
+      "loss": 0.8582,
       "step": 14217
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.932862389029514e-06,
-      "loss": 0.7891,
+      "loss": 0.8553,
       "step": 14218
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.931684554371309e-06,
-      "loss": 0.7735,
+      "loss": 0.9301,
       "step": 14219
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.930506720660884e-06,
-      "loss": 0.8276,
+      "loss": 0.8729,
       "step": 14220
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.92932888791458e-06,
-      "loss": 0.7543,
+      "loss": 0.8509,
       "step": 14221
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.928151056148735e-06,
-      "loss": 0.7086,
+      "loss": 0.837,
       "step": 14222
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.926973225379698e-06,
-      "loss": 0.652,
+      "loss": 0.7903,
       "step": 14223
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.925795395623798e-06,
-      "loss": 0.5715,
+      "loss": 0.7131,
       "step": 14224
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.924617566897386e-06,
-      "loss": 0.7593,
+      "loss": 0.8699,
       "step": 14225
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.923439739216796e-06,
-      "loss": 0.6822,
+      "loss": 0.8962,
       "step": 14226
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.922261912598371e-06,
-      "loss": 0.7848,
+      "loss": 0.8469,
       "step": 14227
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.92108408705845e-06,
-      "loss": 0.6979,
+      "loss": 0.7991,
       "step": 14228
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.919906262613378e-06,
-      "loss": 0.6682,
+      "loss": 0.7775,
       "step": 14229
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.91872843927949e-06,
-      "loss": 0.6026,
+      "loss": 0.7901,
       "step": 14230
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.917550617073136e-06,
-      "loss": 0.8548,
+      "loss": 0.9346,
       "step": 14231
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.916372796010644e-06,
-      "loss": 0.711,
+      "loss": 0.8089,
       "step": 14232
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.915194976108363e-06,
-      "loss": 0.844,
+      "loss": 0.9397,
       "step": 14233
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.91401715738263e-06,
-      "loss": 0.664,
+      "loss": 0.7684,
       "step": 14234
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.912839339849789e-06,
-      "loss": 0.7898,
+      "loss": 0.8653,
       "step": 14235
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.911661523526176e-06,
-      "loss": 0.7454,
+      "loss": 0.8219,
       "step": 14236
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.91048370842814e-06,
-      "loss": 0.766,
+      "loss": 0.8592,
       "step": 14237
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.909305894572011e-06,
-      "loss": 0.6623,
+      "loss": 0.7645,
       "step": 14238
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.908128081974133e-06,
-      "loss": 0.8237,
+      "loss": 0.8745,
       "step": 14239
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.906950270650849e-06,
-      "loss": 0.7966,
+      "loss": 0.941,
       "step": 14240
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.905772460618496e-06,
-      "loss": 0.8309,
+      "loss": 0.8835,
       "step": 14241
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.904594651893423e-06,
-      "loss": 0.6791,
+      "loss": 0.8792,
       "step": 14242
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.903416844491955e-06,
-      "loss": 0.7902,
+      "loss": 0.8891,
       "step": 14243
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.902239038430447e-06,
-      "loss": 0.7889,
+      "loss": 0.8481,
       "step": 14244
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.90106123372523e-06,
-      "loss": 0.6101,
+      "loss": 0.7717,
       "step": 14245
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.899883430392652e-06,
-      "loss": 0.7698,
+      "loss": 0.8853,
       "step": 14246
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.898705628449044e-06,
-      "loss": 0.6724,
+      "loss": 0.7367,
       "step": 14247
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.897527827910757e-06,
-      "loss": 0.8296,
+      "loss": 0.9476,
       "step": 14248
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.896350028794122e-06,
-      "loss": 0.8826,
+      "loss": 0.8946,
       "step": 14249
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.895172231115489e-06,
-      "loss": 0.7645,
+      "loss": 0.9321,
       "step": 14250
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.893994434891185e-06,
-      "loss": 0.6445,
+      "loss": 0.8409,
       "step": 14251
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.892816640137564e-06,
-      "loss": 0.7676,
+      "loss": 0.8265,
       "step": 14252
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.891638846870955e-06,
-      "loss": 0.7173,
+      "loss": 0.8709,
       "step": 14253
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.890461055107707e-06,
-      "loss": 0.6639,
+      "loss": 0.8091,
       "step": 14254
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.889283264864154e-06,
-      "loss": 0.6837,
+      "loss": 0.8565,
       "step": 14255
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.88810547615664e-06,
-      "loss": 0.8419,
+      "loss": 0.9105,
       "step": 14256
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.886927689001508e-06,
-      "loss": 0.6974,
+      "loss": 0.8052,
       "step": 14257
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.885749903415089e-06,
-      "loss": 0.6502,
+      "loss": 0.816,
       "step": 14258
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.884572119413728e-06,
-      "loss": 0.6026,
+      "loss": 0.8582,
       "step": 14259
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.883394337013764e-06,
-      "loss": 0.7809,
+      "loss": 0.8916,
       "step": 14260
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.882216556231541e-06,
-      "loss": 0.7713,
+      "loss": 0.9055,
       "step": 14261
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.881038777083394e-06,
-      "loss": 0.6404,
+      "loss": 0.7613,
       "step": 14262
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.879860999585669e-06,
-      "loss": 0.7625,
+      "loss": 0.9291,
       "step": 14263
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.878683223754698e-06,
-      "loss": 0.7357,
+      "loss": 0.8491,
       "step": 14264
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.877505449606827e-06,
-      "loss": 0.6174,
+      "loss": 0.8116,
       "step": 14265
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.876327677158392e-06,
-      "loss": 0.7514,
+      "loss": 0.9183,
       "step": 14266
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.875149906425737e-06,
-      "loss": 0.6297,
+      "loss": 0.7289,
       "step": 14267
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.873972137425198e-06,
-      "loss": 0.6531,
+      "loss": 0.7708,
       "step": 14268
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.872794370173119e-06,
-      "loss": 0.701,
+      "loss": 0.9043,
       "step": 14269
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.871616604685836e-06,
-      "loss": 0.909,
+      "loss": 0.9253,
       "step": 14270
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.870438840979694e-06,
-      "loss": 0.8228,
+      "loss": 0.9071,
       "step": 14271
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.869261079071026e-06,
-      "loss": 0.7373,
+      "loss": 0.8451,
       "step": 14272
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.868083318976175e-06,
-      "loss": 0.7464,
+      "loss": 0.8351,
       "step": 14273
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.86690556071148e-06,
-      "loss": 0.7612,
+      "loss": 0.873,
       "step": 14274
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.865727804293284e-06,
-      "loss": 0.7454,
+      "loss": 0.8472,
       "step": 14275
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.864550049737926e-06,
-      "loss": 0.6509,
+      "loss": 0.7395,
       "step": 14276
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.86337229706174e-06,
-      "loss": 0.809,
+      "loss": 0.9107,
       "step": 14277
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.86219454628107e-06,
-      "loss": 0.6665,
+      "loss": 0.8207,
       "step": 14278
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.861016797412254e-06,
-      "loss": 0.7963,
+      "loss": 0.8976,
       "step": 14279
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.859839050471636e-06,
-      "loss": 0.7526,
+      "loss": 0.7812,
       "step": 14280
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.85866130547555e-06,
-      "loss": 0.6798,
+      "loss": 0.8175,
       "step": 14281
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.857483562440341e-06,
-      "loss": 0.6727,
+      "loss": 0.7967,
       "step": 14282
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.85630582138234e-06,
-      "loss": 0.8487,
+      "loss": 0.927,
       "step": 14283
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.855128082317897e-06,
-      "loss": 0.676,
+      "loss": 0.8537,
       "step": 14284
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.853950345263342e-06,
-      "loss": 0.7329,
+      "loss": 0.8674,
       "step": 14285
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.852772610235022e-06,
-      "loss": 0.6148,
+      "loss": 0.8251,
       "step": 14286
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.851594877249271e-06,
-      "loss": 0.7347,
+      "loss": 0.8059,
       "step": 14287
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.850417146322432e-06,
-      "loss": 0.8244,
+      "loss": 0.8795,
       "step": 14288
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.849239417470841e-06,
-      "loss": 0.8127,
+      "loss": 0.8394,
       "step": 14289
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.848061690710843e-06,
-      "loss": 0.6764,
+      "loss": 0.8756,
       "step": 14290
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.84688396605877e-06,
-      "loss": 0.8145,
+      "loss": 0.9135,
       "step": 14291
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.845706243530967e-06,
-      "loss": 0.8181,
+      "loss": 0.8553,
       "step": 14292
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.844528523143767e-06,
-      "loss": 0.7417,
+      "loss": 0.8979,
       "step": 14293
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.843350804913517e-06,
-      "loss": 0.6987,
+      "loss": 0.8457,
       "step": 14294
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.84217308885655e-06,
-      "loss": 0.8281,
+      "loss": 1.047,
       "step": 14295
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.84099537498921e-06,
-      "loss": 0.6894,
+      "loss": 0.7733,
       "step": 14296
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.839817663327835e-06,
-      "loss": 0.7372,
+      "loss": 0.8675,
       "step": 14297
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.838639953888758e-06,
-      "loss": 0.7049,
+      "loss": 0.8188,
       "step": 14298
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.837462246688327e-06,
-      "loss": 0.737,
+      "loss": 0.8539,
       "step": 14299
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.836284541742871e-06,
-      "loss": 0.8241,
+      "loss": 0.8769,
       "step": 14300
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.83510683906874e-06,
-      "loss": 0.6738,
+      "loss": 0.8068,
       "step": 14301
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.833929138682263e-06,
-      "loss": 0.7763,
+      "loss": 0.8488,
       "step": 14302
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.832751440599791e-06,
-      "loss": 0.7634,
+      "loss": 0.9073,
       "step": 14303
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.831573744837648e-06,
-      "loss": 0.7295,
+      "loss": 0.8203,
       "step": 14304
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.830396051412185e-06,
-      "loss": 0.5677,
+      "loss": 0.8047,
       "step": 14305
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.829218360339732e-06,
-      "loss": 0.7232,
+      "loss": 0.8838,
       "step": 14306
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.828040671636635e-06,
-      "loss": 0.8462,
+      "loss": 0.9109,
       "step": 14307
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.826862985319228e-06,
-      "loss": 0.7723,
+      "loss": 0.9029,
       "step": 14308
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.825685301403854e-06,
-      "loss": 0.6966,
+      "loss": 0.8222,
       "step": 14309
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.824507619906844e-06,
-      "loss": 0.6684,
+      "loss": 0.8096,
       "step": 14310
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.823329940844549e-06,
-      "loss": 0.7494,
+      "loss": 0.8402,
       "step": 14311
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.822152264233294e-06,
-      "loss": 0.7399,
+      "loss": 0.7746,
       "step": 14312
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.820974590089428e-06,
-      "loss": 0.6874,
+      "loss": 0.7606,
       "step": 14313
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.819796918429283e-06,
-      "loss": 0.812,
+      "loss": 0.9063,
       "step": 14314
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.8186192492692e-06,
-      "loss": 0.6821,
+      "loss": 0.8111,
       "step": 14315
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.817441582625523e-06,
-      "loss": 0.8185,
+      "loss": 0.9102,
       "step": 14316
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.81626391851458e-06,
-      "loss": 0.7815,
+      "loss": 0.808,
       "step": 14317
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.815086256952715e-06,
-      "loss": 0.7769,
+      "loss": 0.8427,
       "step": 14318
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.813908597956265e-06,
-      "loss": 0.7274,
+      "loss": 0.8584,
       "step": 14319
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.812730941541572e-06,
-      "loss": 0.72,
+      "loss": 0.8144,
       "step": 14320
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.811553287724967e-06,
-      "loss": 0.7984,
+      "loss": 0.9451,
       "step": 14321
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.810375636522798e-06,
-      "loss": 0.751,
+      "loss": 0.8178,
       "step": 14322
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.809197987951394e-06,
-      "loss": 0.7542,
+      "loss": 0.8228,
       "step": 14323
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.808020342027101e-06,
-      "loss": 0.8462,
+      "loss": 0.8524,
       "step": 14324
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.80684269876625e-06,
-      "loss": 0.6917,
+      "loss": 0.8101,
       "step": 14325
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.805665058185183e-06,
-      "loss": 0.7156,
+      "loss": 0.7964,
       "step": 14326
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.804487420300237e-06,
-      "loss": 0.6689,
+      "loss": 0.8211,
       "step": 14327
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.803309785127752e-06,
-      "loss": 0.6402,
+      "loss": 0.8182,
       "step": 14328
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.802132152684063e-06,
-      "loss": 0.5732,
+      "loss": 0.7988,
       "step": 14329
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.800954522985515e-06,
-      "loss": 0.7752,
+      "loss": 0.8327,
       "step": 14330
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.799776896048435e-06,
-      "loss": 0.7608,
+      "loss": 0.91,
       "step": 14331
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.798599271889169e-06,
-      "loss": 0.6784,
+      "loss": 0.7967,
       "step": 14332
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.79742165052405e-06,
-      "loss": 0.7663,
+      "loss": 0.9201,
       "step": 14333
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.79624403196942e-06,
-      "loss": 0.8355,
+      "loss": 0.9023,
       "step": 14334
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.795066416241618e-06,
-      "loss": 0.9277,
+      "loss": 1.0043,
       "step": 14335
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.793888803356974e-06,
-      "loss": 0.582,
+      "loss": 0.8629,
       "step": 14336
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.792711193331833e-06,
-      "loss": 0.6831,
+      "loss": 0.8486,
       "step": 14337
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.791533586182527e-06,
-      "loss": 0.698,
+      "loss": 0.7461,
       "step": 14338
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.7903559819254e-06,
-      "loss": 0.7157,
+      "loss": 0.8358,
       "step": 14339
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.789178380576783e-06,
-      "loss": 0.586,
+      "loss": 0.7348,
       "step": 14340
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.78800078215302e-06,
-      "loss": 0.7211,
+      "loss": 0.8362,
       "step": 14341
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.786823186670443e-06,
-      "loss": 0.7476,
+      "loss": 0.8608,
       "step": 14342
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.785645594145396e-06,
-      "loss": 0.8627,
+      "loss": 0.9027,
       "step": 14343
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.784468004594207e-06,
-      "loss": 0.6625,
+      "loss": 0.8055,
       "step": 14344
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.783290418033222e-06,
-      "loss": 0.5666,
+      "loss": 0.7317,
       "step": 14345
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.782112834478772e-06,
-      "loss": 0.8158,
+      "loss": 0.8345,
       "step": 14346
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.7809352539472e-06,
-      "loss": 0.7228,
+      "loss": 0.8646,
       "step": 14347
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.779757676454838e-06,
-      "loss": 0.6321,
+      "loss": 0.7632,
       "step": 14348
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.77858010201803e-06,
-      "loss": 0.7806,
+      "loss": 0.884,
       "step": 14349
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.777402530653103e-06,
-      "loss": 0.7895,
+      "loss": 0.8618,
       "step": 14350
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.776224962376407e-06,
-      "loss": 0.7147,
+      "loss": 0.7975,
       "step": 14351
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.775047397204267e-06,
-      "loss": 0.6611,
+      "loss": 0.7985,
       "step": 14352
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.773869835153029e-06,
-      "loss": 0.6575,
+      "loss": 0.7442,
       "step": 14353
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.772692276239025e-06,
-      "loss": 0.7263,
+      "loss": 0.8632,
       "step": 14354
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.77151472047859e-06,
-      "loss": 0.6977,
+      "loss": 0.782,
       "step": 14355
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.77033716788807e-06,
-      "loss": 0.6813,
+      "loss": 0.8653,
       "step": 14356
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.769159618483789e-06,
-      "loss": 0.6169,
+      "loss": 0.7407,
       "step": 14357
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.767982072282095e-06,
-      "loss": 0.7405,
+      "loss": 0.8552,
       "step": 14358
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.766804529299317e-06,
-      "loss": 0.7327,
+      "loss": 0.9048,
       "step": 14359
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.7656269895518e-06,
-      "loss": 0.7737,
+      "loss": 0.9485,
       "step": 14360
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.76444945305587e-06,
-      "loss": 0.6953,
+      "loss": 0.8265,
       "step": 14361
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.763271919827878e-06,
-      "loss": 0.7215,
+      "loss": 0.8041,
       "step": 14362
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.762094389884146e-06,
-      "loss": 0.7521,
+      "loss": 0.887,
       "step": 14363
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.76091686324102e-06,
-      "loss": 0.7778,
+      "loss": 0.903,
       "step": 14364
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.75973933991483e-06,
-      "loss": 0.6487,
+      "loss": 0.778,
       "step": 14365
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.758561819921918e-06,
-      "loss": 0.8143,
+      "loss": 0.9488,
       "step": 14366
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.757384303278616e-06,
-      "loss": 0.6171,
+      "loss": 0.7409,
       "step": 14367
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.756206790001265e-06,
-      "loss": 0.7243,
+      "loss": 0.9152,
       "step": 14368
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.755029280106196e-06,
-      "loss": 0.8694,
+      "loss": 0.902,
       "step": 14369
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.753851773609754e-06,
-      "loss": 0.7811,
+      "loss": 0.8507,
       "step": 14370
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.752674270528264e-06,
-      "loss": 0.5945,
+      "loss": 0.8107,
       "step": 14371
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.75149677087807e-06,
-      "loss": 0.8454,
+      "loss": 0.8748,
       "step": 14372
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.750319274675506e-06,
-      "loss": 0.8163,
+      "loss": 0.8411,
       "step": 14373
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.749141781936907e-06,
-      "loss": 0.7047,
+      "loss": 0.775,
       "step": 14374
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.747964292678612e-06,
-      "loss": 0.8698,
+      "loss": 0.9274,
       "step": 14375
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.74678680691695e-06,
-      "loss": 0.7643,
+      "loss": 0.7739,
       "step": 14376
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.745609324668266e-06,
-      "loss": 0.6668,
+      "loss": 0.7626,
       "step": 14377
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.744431845948888e-06,
-      "loss": 0.7885,
+      "loss": 0.8576,
       "step": 14378
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.743254370775159e-06,
-      "loss": 0.6596,
+      "loss": 0.8165,
       "step": 14379
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.74207689916341e-06,
-      "loss": 0.6671,
+      "loss": 0.8095,
       "step": 14380
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.74089943112998e-06,
-      "loss": 0.7269,
+      "loss": 0.7934,
       "step": 14381
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.739721966691196e-06,
-      "loss": 0.7747,
+      "loss": 0.8899,
       "step": 14382
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.73854450586341e-06,
-      "loss": 0.7563,
+      "loss": 0.8827,
       "step": 14383
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.73736704866294e-06,
-      "loss": 0.8019,
+      "loss": 0.8922,
       "step": 14384
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.736189595106133e-06,
-      "loss": 0.6139,
+      "loss": 0.7494,
       "step": 14385
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.73501214520932e-06,
-      "loss": 0.8288,
+      "loss": 0.8805,
       "step": 14386
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.733834698988838e-06,
-      "loss": 0.9232,
+      "loss": 0.9724,
       "step": 14387
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.732657256461022e-06,
-      "loss": 0.7399,
+      "loss": 0.8453,
       "step": 14388
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.731479817642211e-06,
-      "loss": 0.7821,
+      "loss": 0.9274,
       "step": 14389
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.730302382548728e-06,
-      "loss": 0.643,
+      "loss": 0.7387,
       "step": 14390
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.729124951196926e-06,
-      "loss": 0.8333,
+      "loss": 0.9731,
       "step": 14391
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.727947523603127e-06,
-      "loss": 0.8102,
+      "loss": 0.9409,
       "step": 14392
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.726770099783669e-06,
-      "loss": 0.7113,
+      "loss": 0.8222,
       "step": 14393
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.725592679754891e-06,
-      "loss": 0.7839,
+      "loss": 0.8636,
       "step": 14394
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.724415263533121e-06,
-      "loss": 0.7402,
+      "loss": 0.7943,
       "step": 14395
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.723237851134704e-06,
-      "loss": 0.8331,
+      "loss": 0.8879,
       "step": 14396
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.722060442575964e-06,
-      "loss": 0.6645,
+      "loss": 0.8505,
       "step": 14397
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.720883037873244e-06,
-      "loss": 0.7564,
+      "loss": 0.8305,
       "step": 14398
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.719705637042873e-06,
-      "loss": 0.7732,
+      "loss": 0.9109,
       "step": 14399
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.718528240101191e-06,
-      "loss": 0.6855,
+      "loss": 0.7677,
       "step": 14400
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.717350847064527e-06,
-      "loss": 0.8806,
+      "loss": 0.9492,
       "step": 14401
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.716173457949225e-06,
-      "loss": 0.6886,
+      "loss": 0.8629,
       "step": 14402
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.71499607277161e-06,
-      "loss": 0.7641,
+      "loss": 0.8955,
       "step": 14403
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.71381869154802e-06,
-      "loss": 0.7969,
+      "loss": 0.9118,
       "step": 14404
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.712641314294788e-06,
-      "loss": 0.8078,
+      "loss": 0.8769,
       "step": 14405
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.711463941028253e-06,
-      "loss": 0.772,
+      "loss": 0.8762,
       "step": 14406
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.710286571764743e-06,
-      "loss": 0.6956,
+      "loss": 0.8351,
       "step": 14407
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.709109206520601e-06,
-      "loss": 0.7654,
+      "loss": 0.9106,
       "step": 14408
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.70793184531215e-06,
-      "loss": 0.873,
+      "loss": 0.9234,
       "step": 14409
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.706754488155736e-06,
-      "loss": 0.7192,
+      "loss": 0.8066,
       "step": 14410
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.705577135067687e-06,
-      "loss": 0.7994,
+      "loss": 0.9164,
       "step": 14411
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.704399786064333e-06,
-      "loss": 0.6313,
+      "loss": 0.8393,
       "step": 14412
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.703222441162015e-06,
-      "loss": 0.6921,
+      "loss": 0.736,
       "step": 14413
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.702045100377063e-06,
-      "loss": 0.7471,
+      "loss": 0.8296,
       "step": 14414
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.700867763725817e-06,
-      "loss": 0.6019,
+      "loss": 0.7252,
       "step": 14415
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.699690431224603e-06,
-      "loss": 0.7373,
+      "loss": 0.8337,
       "step": 14416
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.698513102889759e-06,
-      "loss": 0.7871,
+      "loss": 0.8225,
       "step": 14417
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.697335778737616e-06,
-      "loss": 0.8027,
+      "loss": 0.8943,
       "step": 14418
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.696158458784512e-06,
-      "loss": 0.7192,
+      "loss": 0.7655,
       "step": 14419
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.694981143046776e-06,
-      "loss": 0.7617,
+      "loss": 0.8196,
       "step": 14420
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.693803831540747e-06,
-      "loss": 0.697,
+      "loss": 0.8028,
       "step": 14421
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.69262652428275e-06,
-      "loss": 0.7291,
+      "loss": 0.7996,
       "step": 14422
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.69144922128913e-06,
-      "loss": 0.777,
+      "loss": 0.8812,
       "step": 14423
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.690271922576211e-06,
-      "loss": 0.6669,
+      "loss": 0.8095,
       "step": 14424
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.689094628160332e-06,
-      "loss": 0.7042,
+      "loss": 0.8048,
       "step": 14425
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.687917338057821e-06,
-      "loss": 0.6313,
+      "loss": 0.7455,
       "step": 14426
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.686740052285016e-06,
-      "loss": 0.6581,
+      "loss": 0.8102,
       "step": 14427
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.685562770858246e-06,
-      "loss": 0.8406,
+      "loss": 0.9104,
       "step": 14428
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.684385493793851e-06,
-      "loss": 0.7746,
+      "loss": 0.8395,
       "step": 14429
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.683208221108157e-06,
-      "loss": 0.7775,
+      "loss": 0.8157,
       "step": 14430
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.682030952817498e-06,
-      "loss": 0.6923,
+      "loss": 0.8192,
       "step": 14431
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.68085368893821e-06,
-      "loss": 0.8057,
+      "loss": 0.8432,
       "step": 14432
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.679676429486624e-06,
-      "loss": 0.7163,
+      "loss": 0.8152,
       "step": 14433
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.678499174479075e-06,
-      "loss": 0.704,
+      "loss": 0.8372,
       "step": 14434
     },
     {
       "epoch": 0.52,
       "learning_rate": 9.677321923931888e-06,
-      "loss": 0.6879,
+      "loss": 0.8317,
       "step": 14435
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.676144677861407e-06,
-      "loss": 0.633,
+      "loss": 0.8061,
       "step": 14436
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.674967436283957e-06,
-      "loss": 0.8932,
+      "loss": 0.9591,
       "step": 14437
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.673790199215874e-06,
-      "loss": 0.5903,
+      "loss": 0.7886,
       "step": 14438
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.672612966673485e-06,
-      "loss": 0.6852,
+      "loss": 0.8428,
       "step": 14439
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.671435738673131e-06,
-      "loss": 0.7694,
+      "loss": 0.9095,
       "step": 14440
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.670258515231137e-06,
-      "loss": 0.7497,
+      "loss": 0.852,
       "step": 14441
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.669081296363843e-06,
-      "loss": 0.6382,
+      "loss": 0.8086,
       "step": 14442
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.667904082087573e-06,
-      "loss": 0.699,
+      "loss": 0.8703,
       "step": 14443
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.666726872418663e-06,
-      "loss": 0.7589,
+      "loss": 0.8512,
       "step": 14444
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.665549667373442e-06,
-      "loss": 0.8343,
+      "loss": 0.9654,
       "step": 14445
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.66437246696825e-06,
-      "loss": 0.5667,
+      "loss": 0.757,
       "step": 14446
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.66319527121941e-06,
-      "loss": 0.6548,
+      "loss": 0.848,
       "step": 14447
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.662018080143262e-06,
-      "loss": 0.7374,
+      "loss": 0.7893,
       "step": 14448
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.660840893756132e-06,
-      "loss": 0.6816,
+      "loss": 0.7648,
       "step": 14449
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.659663712074351e-06,
-      "loss": 0.7583,
+      "loss": 0.8367,
       "step": 14450
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.658486535114254e-06,
-      "loss": 0.7358,
+      "loss": 0.8008,
       "step": 14451
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.65730936289217e-06,
-      "loss": 0.7793,
+      "loss": 0.882,
       "step": 14452
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.65613219542444e-06,
-      "loss": 0.88,
+      "loss": 0.9059,
       "step": 14453
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.654955032727378e-06,
-      "loss": 0.8652,
+      "loss": 0.956,
       "step": 14454
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.653777874817333e-06,
-      "loss": 0.8106,
+      "loss": 0.8905,
       "step": 14455
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.652600721710623e-06,
-      "loss": 0.579,
+      "loss": 0.7454,
       "step": 14456
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.651423573423589e-06,
-      "loss": 0.8753,
+      "loss": 0.9648,
       "step": 14457
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.650246429972556e-06,
-      "loss": 0.7887,
+      "loss": 0.884,
       "step": 14458
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.64906929137386e-06,
-      "loss": 0.7115,
+      "loss": 0.7669,
       "step": 14459
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.647892157643825e-06,
-      "loss": 0.9063,
+      "loss": 0.9662,
       "step": 14460
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.646715028798793e-06,
-      "loss": 0.7413,
+      "loss": 0.8083,
       "step": 14461
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.645537904855083e-06,
-      "loss": 0.6895,
+      "loss": 0.8234,
       "step": 14462
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.644360785829036e-06,
-      "loss": 0.8633,
+      "loss": 0.904,
       "step": 14463
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.643183671736976e-06,
-      "loss": 0.7261,
+      "loss": 0.8789,
       "step": 14464
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.642006562595238e-06,
-      "loss": 0.7696,
+      "loss": 0.892,
       "step": 14465
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.640829458420149e-06,
-      "loss": 0.7356,
+      "loss": 0.864,
       "step": 14466
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.639652359228044e-06,
-      "loss": 0.7214,
+      "loss": 0.814,
       "step": 14467
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.638475265035253e-06,
-      "loss": 0.7823,
+      "loss": 0.8567,
       "step": 14468
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.6372981758581e-06,
-      "loss": 0.7275,
+      "loss": 0.7996,
       "step": 14469
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.636121091712923e-06,
-      "loss": 0.8247,
+      "loss": 0.9292,
       "step": 14470
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.634944012616048e-06,
-      "loss": 0.7532,
+      "loss": 0.9425,
       "step": 14471
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.633766938583807e-06,
-      "loss": 0.8173,
+      "loss": 0.9035,
       "step": 14472
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.63258986963253e-06,
-      "loss": 0.8161,
+      "loss": 0.8177,
       "step": 14473
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.63141280577855e-06,
-      "loss": 0.7585,
+      "loss": 0.8335,
       "step": 14474
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.630235747038191e-06,
-      "loss": 0.8003,
+      "loss": 0.8744,
       "step": 14475
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.629058693427789e-06,
-      "loss": 0.6674,
+      "loss": 0.7795,
       "step": 14476
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.627881644963667e-06,
-      "loss": 0.7811,
+      "loss": 0.8699,
       "step": 14477
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.626704601662164e-06,
-      "loss": 0.6364,
+      "loss": 0.7289,
       "step": 14478
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.625527563539602e-06,
-      "loss": 0.7198,
+      "loss": 0.8517,
       "step": 14479
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.624350530612318e-06,
-      "loss": 0.8569,
+      "loss": 0.929,
       "step": 14480
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.62317350289663e-06,
-      "loss": 0.7439,
+      "loss": 0.8101,
       "step": 14481
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.621996480408883e-06,
-      "loss": 0.6889,
+      "loss": 0.9014,
       "step": 14482
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.620819463165393e-06,
-      "loss": 0.7669,
+      "loss": 0.8992,
       "step": 14483
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.619642451182498e-06,
-      "loss": 0.7564,
+      "loss": 0.8597,
       "step": 14484
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.618465444476521e-06,
-      "loss": 0.6434,
+      "loss": 0.8895,
       "step": 14485
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.617288443063798e-06,
-      "loss": 0.8283,
+      "loss": 0.8694,
       "step": 14486
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.616111446960656e-06,
-      "loss": 0.7393,
+      "loss": 0.8791,
       "step": 14487
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.61493445618342e-06,
-      "loss": 0.7226,
+      "loss": 0.8088,
       "step": 14488
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.613757470748423e-06,
-      "loss": 0.7711,
+      "loss": 0.8319,
       "step": 14489
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.612580490671992e-06,
-      "loss": 0.6981,
+      "loss": 0.7939,
       "step": 14490
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.61140351597046e-06,
-      "loss": 0.7595,
+      "loss": 0.9117,
       "step": 14491
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.610226546660147e-06,
-      "loss": 0.8082,
+      "loss": 0.9095,
       "step": 14492
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.609049582757395e-06,
-      "loss": 0.8589,
+      "loss": 0.9459,
       "step": 14493
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.607872624278519e-06,
-      "loss": 0.7477,
+      "loss": 0.8315,
       "step": 14494
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.606695671239858e-06,
-      "loss": 0.8223,
+      "loss": 0.8427,
       "step": 14495
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.605518723657736e-06,
-      "loss": 0.758,
+      "loss": 0.8534,
       "step": 14496
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.60434178154848e-06,
-      "loss": 0.8437,
+      "loss": 0.947,
       "step": 14497
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.60316484492842e-06,
-      "loss": 0.7788,
+      "loss": 0.9008,
       "step": 14498
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.601987913813889e-06,
-      "loss": 0.8347,
+      "loss": 0.9073,
       "step": 14499
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.600810988221207e-06,
-      "loss": 0.8322,
+      "loss": 0.8647,
       "step": 14500
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.59963406816671e-06,
-      "loss": 0.7274,
+      "loss": 0.9024,
       "step": 14501
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.598457153666719e-06,
-      "loss": 0.7824,
+      "loss": 0.8617,
       "step": 14502
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.597280244737565e-06,
-      "loss": 0.8131,
+      "loss": 0.8611,
       "step": 14503
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.596103341395577e-06,
-      "loss": 0.79,
+      "loss": 0.9133,
       "step": 14504
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.594926443657084e-06,
-      "loss": 0.8661,
+      "loss": 0.9709,
       "step": 14505
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.593749551538412e-06,
-      "loss": 0.7849,
+      "loss": 0.843,
       "step": 14506
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.592572665055884e-06,
-      "loss": 0.8458,
+      "loss": 0.8837,
       "step": 14507
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.591395784225838e-06,
-      "loss": 0.7159,
+      "loss": 0.8246,
       "step": 14508
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.590218909064591e-06,
-      "loss": 0.7557,
+      "loss": 0.877,
       "step": 14509
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.589042039588478e-06,
-      "loss": 0.7298,
+      "loss": 0.881,
       "step": 14510
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.58786517581382e-06,
-      "loss": 0.8164,
+      "loss": 0.9357,
       "step": 14511
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.586688317756953e-06,
-      "loss": 0.6723,
+      "loss": 0.781,
       "step": 14512
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.585511465434195e-06,
-      "loss": 0.6741,
+      "loss": 0.8707,
       "step": 14513
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.584334618861882e-06,
-      "loss": 0.7772,
+      "loss": 0.9586,
       "step": 14514
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.583157778056333e-06,
-      "loss": 0.7352,
+      "loss": 0.9018,
       "step": 14515
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.581980943033879e-06,
-      "loss": 0.8539,
+      "loss": 0.8838,
       "step": 14516
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.580804113810844e-06,
-      "loss": 0.6949,
+      "loss": 0.8472,
       "step": 14517
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.57962729040356e-06,
-      "loss": 0.4943,
+      "loss": 0.7307,
       "step": 14518
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.57845047282835e-06,
-      "loss": 0.7372,
+      "loss": 0.8444,
       "step": 14519
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.577273661101545e-06,
-      "loss": 0.7483,
+      "loss": 0.864,
       "step": 14520
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.57609685523946e-06,
-      "loss": 0.7649,
+      "loss": 0.8917,
       "step": 14521
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.574920055258439e-06,
-      "loss": 0.6987,
+      "loss": 0.8117,
       "step": 14522
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.573743261174793e-06,
-      "loss": 0.7705,
+      "loss": 0.9225,
       "step": 14523
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.572566473004857e-06,
-      "loss": 0.7283,
+      "loss": 0.818,
       "step": 14524
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.571389690764953e-06,
-      "loss": 0.6663,
+      "loss": 0.7912,
       "step": 14525
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.570212914471411e-06,
-      "loss": 0.6727,
+      "loss": 0.8237,
       "step": 14526
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.569036144140558e-06,
-      "loss": 0.6789,
+      "loss": 0.8806,
       "step": 14527
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.567859379788711e-06,
-      "loss": 0.7281,
+      "loss": 0.8971,
       "step": 14528
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.566682621432205e-06,
-      "loss": 0.7975,
+      "loss": 0.8568,
       "step": 14529
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.56550586908736e-06,
-      "loss": 0.834,
+      "loss": 0.9259,
       "step": 14530
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.564329122770507e-06,
-      "loss": 0.7294,
+      "loss": 0.8036,
       "step": 14531
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.563152382497967e-06,
-      "loss": 0.7231,
+      "loss": 0.7963,
       "step": 14532
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.561975648286074e-06,
-      "loss": 0.7117,
+      "loss": 0.7777,
       "step": 14533
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.56079892015114e-06,
-      "loss": 0.7876,
+      "loss": 0.8607,
       "step": 14534
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.559622198109503e-06,
-      "loss": 0.6175,
+      "loss": 0.7491,
       "step": 14535
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.558445482177479e-06,
-      "loss": 0.6937,
+      "loss": 0.8048,
       "step": 14536
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.5572687723714e-06,
-      "loss": 0.814,
+      "loss": 0.8788,
       "step": 14537
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.556092068707587e-06,
-      "loss": 0.6885,
+      "loss": 0.7682,
       "step": 14538
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.55491537120237e-06,
-      "loss": 0.7319,
+      "loss": 0.7946,
       "step": 14539
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.553738679872066e-06,
-      "loss": 0.6104,
+      "loss": 0.7756,
       "step": 14540
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.552561994733012e-06,
-      "loss": 0.7082,
+      "loss": 0.8154,
       "step": 14541
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.551385315801518e-06,
-      "loss": 0.7151,
+      "loss": 0.8321,
       "step": 14542
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.550208643093921e-06,
-      "loss": 0.6845,
+      "loss": 0.7988,
       "step": 14543
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.549031976626538e-06,
-      "loss": 0.6627,
+      "loss": 0.782,
       "step": 14544
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.547855316415697e-06,
-      "loss": 0.8194,
+      "loss": 0.8889,
       "step": 14545
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.546678662477727e-06,
-      "loss": 0.6102,
+      "loss": 0.7639,
       "step": 14546
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.54550201482894e-06,
-      "loss": 0.8088,
+      "loss": 0.8958,
       "step": 14547
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.54432537348567e-06,
-      "loss": 0.7955,
+      "loss": 0.8908,
       "step": 14548
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.54314873846424e-06,
-      "loss": 0.6466,
+      "loss": 0.8914,
       "step": 14549
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.541972109780972e-06,
-      "loss": 0.7697,
+      "loss": 0.856,
       "step": 14550
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.54079548745219e-06,
-      "loss": 0.7194,
+      "loss": 0.8365,
       "step": 14551
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.539618871494226e-06,
-      "loss": 0.7095,
+      "loss": 0.8015,
       "step": 14552
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.538442261923387e-06,
-      "loss": 0.6572,
+      "loss": 0.7489,
       "step": 14553
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.537265658756014e-06,
-      "loss": 0.6334,
+      "loss": 0.8298,
       "step": 14554
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.53608906200842e-06,
-      "loss": 0.7072,
+      "loss": 0.8533,
       "step": 14555
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.534912471696933e-06,
-      "loss": 0.6842,
+      "loss": 0.8733,
       "step": 14556
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.533735887837874e-06,
-      "loss": 0.7322,
+      "loss": 0.8436,
       "step": 14557
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.53255931044757e-06,
-      "loss": 0.8076,
+      "loss": 0.8779,
       "step": 14558
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.531382739542341e-06,
-      "loss": 0.805,
+      "loss": 0.8797,
       "step": 14559
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.530206175138516e-06,
-      "loss": 0.8535,
+      "loss": 0.9133,
       "step": 14560
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.529029617252406e-06,
-      "loss": 0.708,
+      "loss": 0.7991,
       "step": 14561
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.52785306590035e-06,
-      "loss": 0.5833,
+      "loss": 0.7363,
       "step": 14562
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.526676521098656e-06,
-      "loss": 0.6068,
+      "loss": 0.8167,
       "step": 14563
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.525499982863658e-06,
-      "loss": 0.5873,
+      "loss": 0.8355,
       "step": 14564
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.524323451211677e-06,
-      "loss": 0.796,
+      "loss": 0.9086,
       "step": 14565
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.523146926159024e-06,
-      "loss": 0.6551,
+      "loss": 0.7686,
       "step": 14566
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.52197040772204e-06,
-      "loss": 0.7706,
+      "loss": 0.8854,
       "step": 14567
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.520793895917033e-06,
-      "loss": 0.7848,
+      "loss": 0.8336,
       "step": 14568
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.519617390760333e-06,
-      "loss": 0.7148,
+      "loss": 0.8129,
       "step": 14569
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.518440892268259e-06,
-      "loss": 0.7252,
+      "loss": 0.7799,
       "step": 14570
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.517264400457136e-06,
-      "loss": 0.7225,
+      "loss": 0.8625,
       "step": 14571
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.516087915343281e-06,
-      "loss": 0.6705,
+      "loss": 0.7942,
       "step": 14572
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.514911436943024e-06,
-      "loss": 0.6883,
+      "loss": 0.805,
       "step": 14573
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.51373496527268e-06,
-      "loss": 0.8366,
+      "loss": 0.9205,
       "step": 14574
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.512558500348575e-06,
-      "loss": 0.758,
+      "loss": 0.8387,
       "step": 14575
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.511382042187026e-06,
-      "loss": 0.7122,
+      "loss": 0.8004,
       "step": 14576
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.51020559080436e-06,
-      "loss": 0.7041,
+      "loss": 0.7312,
       "step": 14577
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.509029146216897e-06,
-      "loss": 0.7256,
+      "loss": 0.8767,
       "step": 14578
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.507852708440957e-06,
-      "loss": 0.8334,
+      "loss": 0.9874,
       "step": 14579
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.506676277492862e-06,
-      "loss": 0.6409,
+      "loss": 0.85,
       "step": 14580
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.505499853388938e-06,
-      "loss": 0.7842,
+      "loss": 0.9225,
       "step": 14581
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.504323436145497e-06,
-      "loss": 0.8251,
+      "loss": 0.861,
       "step": 14582
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.503147025778867e-06,
-      "loss": 0.7806,
+      "loss": 0.8605,
       "step": 14583
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.501970622305368e-06,
-      "loss": 0.6013,
+      "loss": 0.7254,
       "step": 14584
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.500794225741318e-06,
-      "loss": 0.7733,
+      "loss": 0.869,
       "step": 14585
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.499617836103044e-06,
-      "loss": 0.6813,
+      "loss": 0.9072,
       "step": 14586
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.498441453406858e-06,
-      "loss": 0.7245,
+      "loss": 0.8461,
       "step": 14587
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.497265077669087e-06,
-      "loss": 0.8004,
+      "loss": 0.885,
       "step": 14588
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.496088708906047e-06,
-      "loss": 0.7318,
+      "loss": 0.8205,
       "step": 14589
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.494912347134064e-06,
-      "loss": 0.6766,
+      "loss": 0.8473,
       "step": 14590
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.493735992369453e-06,
-      "loss": 0.7087,
+      "loss": 0.8028,
       "step": 14591
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.492559644628543e-06,
-      "loss": 0.706,
+      "loss": 0.8467,
       "step": 14592
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.49138330392764e-06,
-      "loss": 0.9013,
+      "loss": 0.965,
       "step": 14593
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.490206970283079e-06,
-      "loss": 0.7296,
+      "loss": 0.8337,
       "step": 14594
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.489030643711168e-06,
-      "loss": 0.7888,
+      "loss": 0.8604,
       "step": 14595
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.487854324228234e-06,
-      "loss": 0.7024,
+      "loss": 0.7782,
       "step": 14596
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.486678011850592e-06,
-      "loss": 0.6484,
+      "loss": 0.8413,
       "step": 14597
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.485501706594568e-06,
-      "loss": 0.8728,
+      "loss": 0.9643,
       "step": 14598
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.484325408476475e-06,
-      "loss": 0.829,
+      "loss": 0.9058,
       "step": 14599
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.48314911751264e-06,
-      "loss": 0.6944,
+      "loss": 0.8633,
       "step": 14600
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.481972833719373e-06,
-      "loss": 0.6891,
+      "loss": 0.8339,
       "step": 14601
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.480796557112998e-06,
-      "loss": 0.7588,
+      "loss": 0.8495,
       "step": 14602
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.479620287709838e-06,
-      "loss": 0.6928,
+      "loss": 0.7998,
       "step": 14603
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.478444025526203e-06,
-      "loss": 0.807,
+      "loss": 0.8794,
       "step": 14604
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.477267770578423e-06,
-      "loss": 0.7581,
+      "loss": 0.8514,
       "step": 14605
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.476091522882804e-06,
-      "loss": 0.6473,
+      "loss": 0.7916,
       "step": 14606
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.474915282455678e-06,
-      "loss": 0.7978,
+      "loss": 0.8998,
       "step": 14607
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.473739049313354e-06,
-      "loss": 0.6835,
+      "loss": 0.9227,
       "step": 14608
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.472562823472157e-06,
-      "loss": 0.7715,
+      "loss": 0.8604,
       "step": 14609
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.471386604948399e-06,
-      "loss": 0.7426,
+      "loss": 0.8179,
       "step": 14610
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.470210393758404e-06,
-      "loss": 0.6065,
+      "loss": 0.7927,
       "step": 14611
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.469034189918485e-06,
-      "loss": 0.7674,
+      "loss": 0.8442,
       "step": 14612
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.46785799344497e-06,
-      "loss": 0.8553,
+      "loss": 0.886,
       "step": 14613
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.466681804354164e-06,
-      "loss": 0.6928,
+      "loss": 0.809,
       "step": 14614
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.465505622662395e-06,
-      "loss": 0.711,
+      "loss": 0.7768,
       "step": 14615
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.464329448385972e-06,
-      "loss": 0.716,
+      "loss": 0.8386,
       "step": 14616
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.463153281541223e-06,
-      "loss": 0.6462,
+      "loss": 0.8528,
       "step": 14617
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.461977122144457e-06,
-      "loss": 0.7175,
+      "loss": 0.8093,
       "step": 14618
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.460800970212001e-06,
-      "loss": 0.7099,
+      "loss": 0.8524,
       "step": 14619
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.45962482576016e-06,
-      "loss": 0.7875,
+      "loss": 0.8874,
       "step": 14620
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.458448688805262e-06,
-      "loss": 0.7259,
+      "loss": 0.8524,
       "step": 14621
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.45727255936362e-06,
-      "loss": 0.7323,
+      "loss": 0.833,
       "step": 14622
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.456096437451548e-06,
-      "loss": 0.7114,
+      "loss": 0.7982,
       "step": 14623
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.45492032308537e-06,
-      "loss": 0.855,
+      "loss": 0.9495,
       "step": 14624
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.453744216281394e-06,
-      "loss": 0.5926,
+      "loss": 0.8451,
       "step": 14625
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.452568117055948e-06,
-      "loss": 0.6508,
+      "loss": 0.8067,
       "step": 14626
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.451392025425338e-06,
-      "loss": 0.7357,
+      "loss": 0.8925,
       "step": 14627
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.450215941405886e-06,
-      "loss": 0.7342,
+      "loss": 0.8072,
       "step": 14628
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.449039865013909e-06,
-      "loss": 0.6721,
+      "loss": 0.7574,
       "step": 14629
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.447863796265721e-06,
-      "loss": 0.7734,
+      "loss": 0.869,
       "step": 14630
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.44668773517764e-06,
-      "loss": 0.6853,
+      "loss": 0.8676,
       "step": 14631
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.445511681765985e-06,
-      "loss": 0.6229,
+      "loss": 0.7822,
       "step": 14632
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.444335636047062e-06,
-      "loss": 0.7032,
+      "loss": 0.7868,
       "step": 14633
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.4431595980372e-06,
-      "loss": 0.7012,
+      "loss": 0.85,
       "step": 14634
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.441983567752705e-06,
-      "loss": 0.8991,
+      "loss": 0.9291,
       "step": 14635
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.440807545209897e-06,
-      "loss": 0.6466,
+      "loss": 0.8147,
       "step": 14636
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.43963153042509e-06,
-      "loss": 0.715,
+      "loss": 0.8599,
       "step": 14637
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.438455523414602e-06,
-      "loss": 0.836,
+      "loss": 0.9667,
       "step": 14638
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.437279524194745e-06,
-      "loss": 0.6141,
+      "loss": 0.727,
       "step": 14639
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.436103532781841e-06,
-      "loss": 0.676,
+      "loss": 0.7871,
       "step": 14640
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.434927549192198e-06,
-      "loss": 0.8758,
+      "loss": 0.9261,
       "step": 14641
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.43375157344213e-06,
-      "loss": 0.8301,
+      "loss": 0.9322,
       "step": 14642
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.432575605547959e-06,
-      "loss": 0.7062,
+      "loss": 0.7755,
       "step": 14643
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.431399645525994e-06,
-      "loss": 0.9045,
+      "loss": 0.9433,
       "step": 14644
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.430223693392558e-06,
-      "loss": 0.6528,
+      "loss": 0.7401,
       "step": 14645
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.429047749163953e-06,
-      "loss": 0.6996,
+      "loss": 0.8126,
       "step": 14646
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.427871812856503e-06,
-      "loss": 0.6986,
+      "loss": 0.8751,
       "step": 14647
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.426695884486519e-06,
-      "loss": 0.7485,
+      "loss": 0.8625,
       "step": 14648
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.425519964070317e-06,
-      "loss": 0.7571,
+      "loss": 0.8163,
       "step": 14649
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.42434405162421e-06,
-      "loss": 0.7007,
+      "loss": 0.854,
       "step": 14650
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.423168147164513e-06,
-      "loss": 0.7508,
+      "loss": 0.7757,
       "step": 14651
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.421992250707538e-06,
-      "loss": 0.7527,
+      "loss": 0.8581,
       "step": 14652
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.420816362269605e-06,
-      "loss": 0.7447,
+      "loss": 0.832,
       "step": 14653
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.419640481867018e-06,
-      "loss": 0.839,
+      "loss": 0.9024,
       "step": 14654
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.4184646095161e-06,
-      "loss": 0.7148,
+      "loss": 0.8077,
       "step": 14655
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.417288745233156e-06,
-      "loss": 0.6511,
+      "loss": 0.8536,
       "step": 14656
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.416112889034506e-06,
-      "loss": 0.8094,
+      "loss": 0.8601,
       "step": 14657
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.414937040936462e-06,
-      "loss": 0.6307,
+      "loss": 0.8603,
       "step": 14658
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.413761200955337e-06,
-      "loss": 0.7213,
+      "loss": 0.8095,
       "step": 14659
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.412585369107444e-06,
-      "loss": 0.7164,
+      "loss": 0.8283,
       "step": 14660
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.411409545409094e-06,
-      "loss": 0.7255,
+      "loss": 0.8672,
       "step": 14661
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.4102337298766e-06,
-      "loss": 0.6565,
+      "loss": 0.78,
       "step": 14662
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.409057922526278e-06,
-      "loss": 0.8014,
+      "loss": 0.8525,
       "step": 14663
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.407882123374441e-06,
-      "loss": 0.764,
+      "loss": 0.8156,
       "step": 14664
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.406706332437395e-06,
-      "loss": 0.8032,
+      "loss": 0.9487,
       "step": 14665
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.405530549731461e-06,
-      "loss": 0.7274,
+      "loss": 0.8049,
       "step": 14666
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.404354775272943e-06,
-      "loss": 0.6978,
+      "loss": 0.8568,
       "step": 14667
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.403179009078161e-06,
-      "loss": 0.6157,
+      "loss": 0.764,
       "step": 14668
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.402003251163421e-06,
-      "loss": 0.7012,
+      "loss": 0.7733,
       "step": 14669
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.400827501545039e-06,
-      "loss": 0.8601,
+      "loss": 0.9555,
       "step": 14670
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.399651760239323e-06,
-      "loss": 0.5936,
+      "loss": 0.7433,
       "step": 14671
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.398476027262591e-06,
-      "loss": 0.7143,
+      "loss": 0.86,
       "step": 14672
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.397300302631149e-06,
-      "loss": 0.5919,
+      "loss": 0.7722,
       "step": 14673
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.39612458636131e-06,
-      "loss": 0.699,
+      "loss": 0.889,
       "step": 14674
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.394948878469383e-06,
-      "loss": 0.6964,
+      "loss": 0.8045,
       "step": 14675
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.393773178971686e-06,
-      "loss": 0.7843,
+      "loss": 0.8214,
       "step": 14676
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.392597487884522e-06,
-      "loss": 0.9069,
+      "loss": 0.946,
       "step": 14677
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.39142180522421e-06,
-      "loss": 0.6871,
+      "loss": 0.8347,
       "step": 14678
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.390246131007057e-06,
-      "loss": 0.7967,
+      "loss": 0.8364,
       "step": 14679
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.389070465249371e-06,
-      "loss": 0.8621,
+      "loss": 0.9437,
       "step": 14680
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.387894807967466e-06,
-      "loss": 0.7705,
+      "loss": 0.8981,
       "step": 14681
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.386719159177652e-06,
-      "loss": 0.697,
+      "loss": 0.892,
       "step": 14682
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.38554351889624e-06,
-      "loss": 0.7047,
+      "loss": 0.8265,
       "step": 14683
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.38436788713954e-06,
-      "loss": 0.7397,
+      "loss": 0.8108,
       "step": 14684
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.383192263923865e-06,
-      "loss": 0.6223,
+      "loss": 0.7865,
       "step": 14685
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.382016649265518e-06,
-      "loss": 0.7202,
+      "loss": 0.8551,
       "step": 14686
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.380841043180815e-06,
-      "loss": 0.7007,
+      "loss": 0.8673,
       "step": 14687
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.379665445686062e-06,
-      "loss": 0.6804,
+      "loss": 0.8096,
       "step": 14688
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.378489856797574e-06,
-      "loss": 0.6944,
+      "loss": 0.8525,
       "step": 14689
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.377314276531653e-06,
-      "loss": 0.6106,
+      "loss": 0.7505,
       "step": 14690
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.376138704904619e-06,
-      "loss": 0.5638,
+      "loss": 0.7359,
       "step": 14691
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.374963141932769e-06,
-      "loss": 0.6436,
+      "loss": 0.8285,
       "step": 14692
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.373787587632424e-06,
-      "loss": 0.9178,
+      "loss": 0.8883,
       "step": 14693
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.372612042019883e-06,
-      "loss": 0.7719,
+      "loss": 0.8903,
       "step": 14694
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.371436505111464e-06,
-      "loss": 0.7769,
+      "loss": 0.9138,
       "step": 14695
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.370260976923469e-06,
-      "loss": 0.8903,
+      "loss": 0.9841,
       "step": 14696
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.36908545747221e-06,
-      "loss": 0.7105,
+      "loss": 0.8491,
       "step": 14697
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.367909946774e-06,
-      "loss": 0.7315,
+      "loss": 0.8583,
       "step": 14698
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.366734444845135e-06,
-      "loss": 0.8085,
+      "loss": 0.9134,
       "step": 14699
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.365558951701936e-06,
-      "loss": 0.7835,
+      "loss": 0.8188,
       "step": 14700
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.364383467360703e-06,
-      "loss": 0.7815,
+      "loss": 0.8887,
       "step": 14701
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.363207991837749e-06,
-      "loss": 0.7871,
+      "loss": 0.8803,
       "step": 14702
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.362032525149379e-06,
-      "loss": 0.8286,
+      "loss": 0.9204,
       "step": 14703
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.360857067311909e-06,
-      "loss": 0.7828,
+      "loss": 0.874,
       "step": 14704
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.35968161834163e-06,
-      "loss": 0.8265,
+      "loss": 0.8807,
       "step": 14705
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.358506178254868e-06,
-      "loss": 0.7819,
+      "loss": 0.8215,
       "step": 14706
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.357330747067921e-06,
-      "loss": 0.6549,
+      "loss": 0.781,
       "step": 14707
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.356155324797097e-06,
-      "loss": 0.7095,
+      "loss": 0.847,
       "step": 14708
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.354979911458704e-06,
-      "loss": 0.8262,
+      "loss": 0.9096,
       "step": 14709
     },
     {
       "epoch": 0.53,
       "learning_rate": 9.353804507069051e-06,
-      "loss": 0.9127,
+      "loss": 0.9695,
       "step": 14710
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.352629111644442e-06,
-      "loss": 0.74,
+      "loss": 0.8481,
       "step": 14711
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.35145372520119e-06,
-      "loss": 0.7518,
+      "loss": 0.8374,
       "step": 14712
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.350278347755591e-06,
-      "loss": 0.7049,
+      "loss": 0.8592,
       "step": 14713
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.349102979323961e-06,
-      "loss": 0.6882,
+      "loss": 0.7773,
       "step": 14714
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.347927619922604e-06,
-      "loss": 0.765,
+      "loss": 0.8461,
       "step": 14715
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.346752269567826e-06,
-      "loss": 0.7279,
+      "loss": 0.8109,
       "step": 14716
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.345576928275936e-06,
-      "loss": 0.7082,
+      "loss": 0.8253,
       "step": 14717
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.344401596063234e-06,
-      "loss": 0.6618,
+      "loss": 0.7668,
       "step": 14718
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.343226272946032e-06,
-      "loss": 0.7298,
+      "loss": 0.8517,
       "step": 14719
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.34205095894063e-06,
-      "loss": 0.5529,
+      "loss": 0.7378,
       "step": 14720
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.34087565406334e-06,
-      "loss": 0.6266,
+      "loss": 0.8515,
       "step": 14721
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.339700358330464e-06,
-      "loss": 0.6759,
+      "loss": 0.8288,
       "step": 14722
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.33852507175831e-06,
-      "loss": 0.7552,
+      "loss": 0.8782,
       "step": 14723
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.33734979436318e-06,
-      "loss": 0.6407,
+      "loss": 0.7926,
       "step": 14724
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.336174526161386e-06,
-      "loss": 0.7353,
+      "loss": 0.8511,
       "step": 14725
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.334999267169223e-06,
-      "loss": 0.6822,
+      "loss": 0.773,
       "step": 14726
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.333824017403004e-06,
-      "loss": 0.7679,
+      "loss": 0.9017,
       "step": 14727
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.332648776879031e-06,
-      "loss": 0.8269,
+      "loss": 0.9286,
       "step": 14728
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.331473545613611e-06,
-      "loss": 0.7201,
+      "loss": 0.7505,
       "step": 14729
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.330298323623045e-06,
-      "loss": 0.7603,
+      "loss": 0.8532,
       "step": 14730
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.329123110923643e-06,
-      "loss": 0.595,
+      "loss": 0.8059,
       "step": 14731
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.3279479075317e-06,
-      "loss": 0.7901,
+      "loss": 0.8495,
       "step": 14732
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.326772713463534e-06,
-      "loss": 0.6923,
+      "loss": 0.8425,
       "step": 14733
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.325597528735436e-06,
-      "loss": 0.7354,
+      "loss": 0.7704,
       "step": 14734
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.324422353363718e-06,
-      "loss": 0.7855,
+      "loss": 0.9333,
       "step": 14735
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.323247187364682e-06,
-      "loss": 0.807,
+      "loss": 0.8355,
       "step": 14736
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.322072030754626e-06,
-      "loss": 0.6472,
+      "loss": 0.8101,
       "step": 14737
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.320896883549864e-06,
-      "loss": 0.7769,
+      "loss": 0.91,
       "step": 14738
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.319721745766693e-06,
-      "loss": 0.5763,
+      "loss": 0.7238,
       "step": 14739
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.318546617421417e-06,
-      "loss": 0.6365,
+      "loss": 0.8036,
       "step": 14740
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.317371498530337e-06,
-      "loss": 0.784,
+      "loss": 0.8022,
       "step": 14741
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.316196389109763e-06,
-      "loss": 0.7084,
+      "loss": 0.8304,
       "step": 14742
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.31502128917599e-06,
-      "loss": 0.629,
+      "loss": 0.8055,
       "step": 14743
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.31384619874533e-06,
-      "loss": 0.6018,
+      "loss": 0.7843,
       "step": 14744
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.312671117834076e-06,
-      "loss": 0.7907,
+      "loss": 0.8416,
       "step": 14745
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.311496046458539e-06,
-      "loss": 0.6481,
+      "loss": 0.8256,
       "step": 14746
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.310320984635014e-06,
-      "loss": 0.7024,
+      "loss": 0.7636,
       "step": 14747
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.309145932379808e-06,
-      "loss": 0.8203,
+      "loss": 0.8882,
       "step": 14748
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.307970889709221e-06,
-      "loss": 0.7174,
+      "loss": 0.834,
       "step": 14749
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.306795856639558e-06,
-      "loss": 0.7244,
+      "loss": 0.8667,
       "step": 14750
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.305620833187117e-06,
-      "loss": 0.6959,
+      "loss": 0.7965,
       "step": 14751
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.304445819368206e-06,
-      "loss": 0.891,
+      "loss": 0.8826,
       "step": 14752
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.303270815199117e-06,
-      "loss": 0.8478,
+      "loss": 0.9125,
       "step": 14753
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.30209582069616e-06,
-      "loss": 0.7415,
+      "loss": 0.7845,
       "step": 14754
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.300920835875633e-06,
-      "loss": 0.7226,
+      "loss": 0.8547,
       "step": 14755
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.299745860753836e-06,
-      "loss": 0.7319,
+      "loss": 0.851,
       "step": 14756
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.298570895347075e-06,
-      "loss": 0.7726,
+      "loss": 0.9398,
       "step": 14757
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.297395939671645e-06,
-      "loss": 0.9026,
+      "loss": 0.8924,
       "step": 14758
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.29622099374385e-06,
-      "loss": 0.8023,
+      "loss": 0.9027,
       "step": 14759
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.295046057579988e-06,
-      "loss": 0.7294,
+      "loss": 0.8333,
       "step": 14760
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.293871131196364e-06,
-      "loss": 0.8031,
+      "loss": 0.8673,
       "step": 14761
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.292696214609274e-06,
-      "loss": 0.8701,
+      "loss": 0.8781,
       "step": 14762
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.291521307835026e-06,
-      "loss": 0.7156,
+      "loss": 0.8418,
       "step": 14763
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.290346410889907e-06,
-      "loss": 0.7417,
+      "loss": 0.8843,
       "step": 14764
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.289171523790231e-06,
-      "loss": 0.6204,
+      "loss": 0.8134,
       "step": 14765
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.287996646552289e-06,
-      "loss": 0.8017,
+      "loss": 0.8372,
       "step": 14766
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.286821779192382e-06,
-      "loss": 0.7559,
+      "loss": 0.8567,
       "step": 14767
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.285646921726813e-06,
-      "loss": 0.6984,
+      "loss": 0.8144,
       "step": 14768
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.284472074171879e-06,
-      "loss": 0.9246,
+      "loss": 0.9052,
       "step": 14769
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.28329723654388e-06,
-      "loss": 0.9175,
+      "loss": 0.9499,
       "step": 14770
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.282122408859116e-06,
-      "loss": 0.7345,
+      "loss": 0.8351,
       "step": 14771
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.28094759113388e-06,
-      "loss": 0.7408,
+      "loss": 0.8081,
       "step": 14772
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.279772783384483e-06,
-      "loss": 0.7653,
+      "loss": 0.7953,
       "step": 14773
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.278597985627212e-06,
-      "loss": 0.7488,
+      "loss": 0.8005,
       "step": 14774
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.277423197878372e-06,
-      "loss": 0.8565,
+      "loss": 0.9103,
       "step": 14775
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.276248420154263e-06,
-      "loss": 0.7647,
+      "loss": 0.8988,
       "step": 14776
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.275073652471173e-06,
-      "loss": 0.8151,
+      "loss": 0.8764,
       "step": 14777
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.273898894845415e-06,
-      "loss": 0.7187,
+      "loss": 0.7633,
       "step": 14778
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.272724147293276e-06,
-      "loss": 0.72,
+      "loss": 0.7933,
       "step": 14779
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.271549409831057e-06,
-      "loss": 0.6872,
+      "loss": 0.8211,
       "step": 14780
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.270374682475056e-06,
-      "loss": 0.7402,
+      "loss": 0.7803,
       "step": 14781
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.269199965241574e-06,
-      "loss": 0.7438,
+      "loss": 0.8021,
       "step": 14782
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.268025258146903e-06,
-      "loss": 0.8161,
+      "loss": 0.8843,
       "step": 14783
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.266850561207345e-06,
-      "loss": 0.7545,
+      "loss": 0.8508,
       "step": 14784
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.265675874439195e-06,
-      "loss": 0.5961,
+      "loss": 0.7783,
       "step": 14785
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.264501197858748e-06,
-      "loss": 0.8779,
+      "loss": 0.8889,
       "step": 14786
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.263326531482304e-06,
-      "loss": 0.8593,
+      "loss": 0.9412,
       "step": 14787
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.26215187532616e-06,
-      "loss": 0.6231,
+      "loss": 0.6979,
       "step": 14788
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.26097722940661e-06,
-      "loss": 0.8023,
+      "loss": 0.8439,
       "step": 14789
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.25980259373996e-06,
-      "loss": 0.7672,
+      "loss": 0.8975,
       "step": 14790
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.258627968342489e-06,
-      "loss": 0.8093,
+      "loss": 0.8865,
       "step": 14791
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.25745335323051e-06,
-      "loss": 0.7551,
+      "loss": 0.8589,
       "step": 14792
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.256278748420307e-06,
-      "loss": 0.572,
+      "loss": 0.7794,
       "step": 14793
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.255104153928185e-06,
-      "loss": 0.718,
+      "loss": 0.8296,
       "step": 14794
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.253929569770436e-06,
-      "loss": 0.7718,
+      "loss": 0.8771,
       "step": 14795
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.252754995963352e-06,
-      "loss": 0.7814,
+      "loss": 0.8885,
       "step": 14796
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.251580432523237e-06,
-      "loss": 0.6847,
+      "loss": 0.7467,
       "step": 14797
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.250405879466378e-06,
-      "loss": 0.7209,
+      "loss": 0.8534,
       "step": 14798
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.249231336809076e-06,
-      "loss": 0.7237,
+      "loss": 0.829,
       "step": 14799
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.248056804567621e-06,
-      "loss": 0.7631,
+      "loss": 0.8333,
       "step": 14800
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.246882282758314e-06,
-      "loss": 0.8897,
+      "loss": 0.9512,
       "step": 14801
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.245707771397445e-06,
-      "loss": 0.6969,
+      "loss": 0.9034,
       "step": 14802
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.244533270501313e-06,
-      "loss": 0.7892,
+      "loss": 0.8699,
       "step": 14803
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.243358780086205e-06,
-      "loss": 0.7569,
+      "loss": 0.771,
       "step": 14804
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.242184300168426e-06,
-      "loss": 0.8246,
+      "loss": 0.9053,
       "step": 14805
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.241009830764261e-06,
-      "loss": 0.709,
+      "loss": 0.8644,
       "step": 14806
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.23983537189001e-06,
-      "loss": 0.7009,
+      "loss": 0.8218,
       "step": 14807
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.238660923561963e-06,
-      "loss": 0.8551,
+      "loss": 0.9525,
       "step": 14808
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.237486485796417e-06,
-      "loss": 0.7169,
+      "loss": 0.7908,
       "step": 14809
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.236312058609662e-06,
-      "loss": 0.853,
+      "loss": 0.8308,
       "step": 14810
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.235137642017998e-06,
-      "loss": 0.7468,
+      "loss": 0.8394,
       "step": 14811
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.233963236037708e-06,
-      "loss": 0.8107,
+      "loss": 0.8854,
       "step": 14812
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.232788840685096e-06,
-      "loss": 0.7779,
+      "loss": 0.8115,
       "step": 14813
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.231614455976448e-06,
-      "loss": 0.6892,
+      "loss": 0.8318,
       "step": 14814
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.230440081928059e-06,
-      "loss": 0.6229,
+      "loss": 0.8058,
       "step": 14815
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.229265718556226e-06,
-      "loss": 0.7542,
+      "loss": 0.8649,
       "step": 14816
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.228091365877231e-06,
-      "loss": 0.77,
+      "loss": 0.9157,
       "step": 14817
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.226917023907379e-06,
-      "loss": 0.8346,
+      "loss": 0.8568,
       "step": 14818
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.225742692662953e-06,
-      "loss": 0.7699,
+      "loss": 0.7878,
       "step": 14819
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.22456837216025e-06,
-      "loss": 0.7007,
+      "loss": 0.8607,
       "step": 14820
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.223394062415558e-06,
-      "loss": 0.6831,
+      "loss": 0.8123,
       "step": 14821
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.222219763445174e-06,
-      "loss": 0.6959,
+      "loss": 0.8506,
       "step": 14822
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.221045475265386e-06,
-      "loss": 0.7397,
+      "loss": 0.8193,
       "step": 14823
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.21987119789249e-06,
-      "loss": 0.8062,
+      "loss": 0.8601,
       "step": 14824
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.218696931342773e-06,
-      "loss": 0.6351,
+      "loss": 0.8632,
       "step": 14825
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.217522675632527e-06,
-      "loss": 0.6578,
+      "loss": 0.6825,
       "step": 14826
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.216348430778041e-06,
-      "loss": 0.8494,
+      "loss": 0.8997,
       "step": 14827
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.215174196795612e-06,
-      "loss": 0.7934,
+      "loss": 0.9189,
       "step": 14828
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.213999973701525e-06,
-      "loss": 0.8403,
+      "loss": 0.9126,
       "step": 14829
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.212825761512077e-06,
-      "loss": 0.8681,
+      "loss": 0.9138,
       "step": 14830
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.211651560243549e-06,
-      "loss": 0.7902,
+      "loss": 0.8327,
       "step": 14831
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.210477369912244e-06,
-      "loss": 0.7273,
+      "loss": 0.8522,
       "step": 14832
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.209303190534442e-06,
-      "loss": 0.5905,
+      "loss": 0.7057,
       "step": 14833
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.208129022126435e-06,
-      "loss": 0.7511,
+      "loss": 0.8371,
       "step": 14834
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.206954864704519e-06,
-      "loss": 0.7203,
+      "loss": 0.8286,
       "step": 14835
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.20578071828497e-06,
-      "loss": 0.6498,
+      "loss": 0.7128,
       "step": 14836
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.204606582884096e-06,
-      "loss": 0.7746,
+      "loss": 0.9267,
       "step": 14837
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.203432458518172e-06,
-      "loss": 0.6523,
+      "loss": 0.7339,
       "step": 14838
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.202258345203494e-06,
-      "loss": 0.7771,
+      "loss": 0.9008,
       "step": 14839
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.201084242956348e-06,
-      "loss": 0.6947,
+      "loss": 0.8144,
       "step": 14840
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.199910151793026e-06,
-      "loss": 0.6879,
+      "loss": 0.8312,
       "step": 14841
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.198736071729815e-06,
-      "loss": 0.5987,
+      "loss": 0.8269,
       "step": 14842
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.197562002783009e-06,
-      "loss": 0.7712,
+      "loss": 0.8746,
       "step": 14843
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.196387944968882e-06,
-      "loss": 0.5726,
+      "loss": 0.7353,
       "step": 14844
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.195213898303742e-06,
-      "loss": 0.8038,
+      "loss": 0.8816,
       "step": 14845
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.19403986280386e-06,
-      "loss": 0.7871,
+      "loss": 0.8755,
       "step": 14846
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.192865838485535e-06,
-      "loss": 0.7596,
+      "loss": 0.888,
       "step": 14847
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.191691825365049e-06,
-      "loss": 0.7743,
+      "loss": 0.8976,
       "step": 14848
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.190517823458695e-06,
-      "loss": 0.8648,
+      "loss": 0.9312,
       "step": 14849
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.189343832782752e-06,
-      "loss": 0.709,
+      "loss": 0.8105,
       "step": 14850
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.188169853353522e-06,
-      "loss": 0.7078,
+      "loss": 0.8288,
       "step": 14851
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.18699588518728e-06,
-      "loss": 0.7479,
+      "loss": 0.9097,
       "step": 14852
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.185821928300314e-06,
-      "loss": 0.7164,
+      "loss": 0.8148,
       "step": 14853
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.184647982708915e-06,
-      "loss": 0.8061,
+      "loss": 0.8268,
       "step": 14854
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.183474048429368e-06,
-      "loss": 0.7213,
+      "loss": 0.8143,
       "step": 14855
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.182300125477963e-06,
-      "loss": 0.7211,
+      "loss": 0.8651,
       "step": 14856
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.18112621387098e-06,
-      "loss": 0.7398,
+      "loss": 0.7914,
       "step": 14857
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.17995231362471e-06,
-      "loss": 0.5957,
+      "loss": 0.7545,
       "step": 14858
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.178778424755436e-06,
-      "loss": 0.627,
+      "loss": 0.7915,
       "step": 14859
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.17760454727945e-06,
-      "loss": 0.7538,
+      "loss": 0.7833,
       "step": 14860
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.17643068121303e-06,
-      "loss": 0.7981,
+      "loss": 0.8514,
       "step": 14861
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.175256826572471e-06,
-      "loss": 0.7389,
+      "loss": 0.8065,
       "step": 14862
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.174082983374049e-06,
-      "loss": 0.7268,
+      "loss": 0.8269,
       "step": 14863
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.172909151634057e-06,
-      "loss": 0.7052,
+      "loss": 0.8753,
       "step": 14864
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.171735331368774e-06,
-      "loss": 0.7045,
+      "loss": 0.8767,
       "step": 14865
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.170561522594492e-06,
-      "loss": 0.6767,
+      "loss": 0.7002,
       "step": 14866
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.169387725327488e-06,
-      "loss": 0.6777,
+      "loss": 0.7726,
       "step": 14867
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.168213939584052e-06,
-      "loss": 0.7758,
+      "loss": 0.8511,
       "step": 14868
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.167040165380466e-06,
-      "loss": 0.798,
+      "loss": 0.9335,
       "step": 14869
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.165866402733021e-06,
-      "loss": 0.6388,
+      "loss": 0.8017,
       "step": 14870
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.164692651657993e-06,
-      "loss": 0.7602,
+      "loss": 0.8506,
       "step": 14871
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.163518912171669e-06,
-      "loss": 0.6992,
+      "loss": 0.8783,
       "step": 14872
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.162345184290334e-06,
-      "loss": 0.8319,
+      "loss": 0.9106,
       "step": 14873
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.16117146803027e-06,
-      "loss": 0.6275,
+      "loss": 0.7632,
       "step": 14874
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.159997763407765e-06,
-      "loss": 0.69,
+      "loss": 0.8794,
       "step": 14875
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.158824070439092e-06,
-      "loss": 0.7747,
+      "loss": 0.8366,
       "step": 14876
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.15765038914055e-06,
-      "loss": 0.7272,
+      "loss": 0.7763,
       "step": 14877
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.156476719528409e-06,
-      "loss": 0.854,
+      "loss": 0.9277,
       "step": 14878
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.155303061618957e-06,
-      "loss": 0.8232,
+      "loss": 0.8776,
       "step": 14879
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.154129415428477e-06,
-      "loss": 0.6703,
+      "loss": 0.7794,
       "step": 14880
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.152955780973251e-06,
-      "loss": 0.8246,
+      "loss": 0.9172,
       "step": 14881
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.151782158269562e-06,
-      "loss": 0.6593,
+      "loss": 0.7389,
       "step": 14882
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.150608547333695e-06,
-      "loss": 0.6177,
+      "loss": 0.8126,
       "step": 14883
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.149434948181924e-06,
-      "loss": 0.7917,
+      "loss": 0.8977,
       "step": 14884
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.14826136083054e-06,
-      "loss": 0.6898,
+      "loss": 0.7451,
       "step": 14885
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.147087785295818e-06,
-      "loss": 0.6712,
+      "loss": 0.8333,
       "step": 14886
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.145914221594046e-06,
-      "loss": 0.6538,
+      "loss": 0.8132,
       "step": 14887
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.1447406697415e-06,
-      "loss": 0.782,
+      "loss": 0.8232,
       "step": 14888
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.143567129754465e-06,
-      "loss": 0.7199,
+      "loss": 0.7864,
       "step": 14889
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.142393601649224e-06,
-      "loss": 0.756,
+      "loss": 0.8346,
       "step": 14890
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.141220085442051e-06,
-      "loss": 0.7133,
+      "loss": 0.7918,
       "step": 14891
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.140046581149232e-06,
-      "loss": 0.6716,
+      "loss": 0.827,
       "step": 14892
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.138873088787045e-06,
-      "loss": 0.7609,
+      "loss": 0.8673,
       "step": 14893
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.137699608371775e-06,
-      "loss": 0.6856,
+      "loss": 0.8425,
       "step": 14894
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.136526139919696e-06,
-      "loss": 0.7296,
+      "loss": 0.852,
       "step": 14895
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.135352683447096e-06,
-      "loss": 0.7151,
+      "loss": 0.815,
       "step": 14896
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.134179238970247e-06,
-      "loss": 0.8197,
+      "loss": 0.8538,
       "step": 14897
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.133005806505435e-06,
-      "loss": 0.755,
+      "loss": 0.8265,
       "step": 14898
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.131832386068934e-06,
-      "loss": 0.7356,
+      "loss": 0.8162,
       "step": 14899
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.130658977677031e-06,
-      "loss": 0.7745,
+      "loss": 0.8995,
       "step": 14900
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.129485581345997e-06,
-      "loss": 0.7091,
+      "loss": 0.8339,
       "step": 14901
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.128312197092122e-06,
-      "loss": 0.6899,
+      "loss": 0.7967,
       "step": 14902
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.127138824931669e-06,
-      "loss": 0.776,
+      "loss": 0.9559,
       "step": 14903
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.125965464880936e-06,
-      "loss": 0.7403,
+      "loss": 0.8309,
       "step": 14904
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.124792116956186e-06,
-      "loss": 0.7217,
+      "loss": 0.8682,
       "step": 14905
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.123618781173706e-06,
-      "loss": 0.6894,
+      "loss": 0.7574,
       "step": 14906
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.12244545754977e-06,
-      "loss": 0.7544,
+      "loss": 0.8655,
       "step": 14907
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.121272146100661e-06,
-      "loss": 0.8502,
+      "loss": 0.954,
       "step": 14908
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.120098846842655e-06,
-      "loss": 0.6734,
+      "loss": 0.7887,
       "step": 14909
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.118925559792025e-06,
-      "loss": 0.7068,
+      "loss": 0.8367,
       "step": 14910
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.117752284965056e-06,
-      "loss": 0.6886,
+      "loss": 0.7442,
       "step": 14911
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.11657902237802e-06,
-      "loss": 0.8226,
+      "loss": 0.9479,
       "step": 14912
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.115405772047199e-06,
-      "loss": 0.8788,
+      "loss": 0.8875,
       "step": 14913
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.114232533988864e-06,
-      "loss": 0.6636,
+      "loss": 0.8344,
       "step": 14914
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.113059308219301e-06,
-      "loss": 0.7207,
+      "loss": 0.8013,
       "step": 14915
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.111886094754776e-06,
-      "loss": 0.5662,
+      "loss": 0.7993,
       "step": 14916
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.110712893611577e-06,
-      "loss": 0.714,
+      "loss": 0.8786,
       "step": 14917
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.109539704805971e-06,
-      "loss": 0.643,
+      "loss": 0.7603,
       "step": 14918
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.10836652835424e-06,
-      "loss": 0.7047,
+      "loss": 0.8101,
       "step": 14919
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.107193364272657e-06,
-      "loss": 0.6547,
+      "loss": 0.8192,
       "step": 14920
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.106020212577501e-06,
-      "loss": 0.6959,
+      "loss": 0.8115,
       "step": 14921
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.104847073285043e-06,
-      "loss": 0.6766,
+      "loss": 0.8195,
       "step": 14922
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.103673946411569e-06,
-      "loss": 0.649,
+      "loss": 0.7801,
       "step": 14923
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.10250083197334e-06,
-      "loss": 0.6155,
+      "loss": 0.8704,
       "step": 14924
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.101327729986642e-06,
-      "loss": 0.6885,
+      "loss": 0.8145,
       "step": 14925
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.100154640467745e-06,
-      "loss": 0.7746,
+      "loss": 0.8683,
       "step": 14926
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.098981563432927e-06,
-      "loss": 0.7609,
+      "loss": 0.8973,
       "step": 14927
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.097808498898462e-06,
-      "loss": 0.8547,
+      "loss": 0.8779,
       "step": 14928
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.096635446880623e-06,
-      "loss": 0.6472,
+      "loss": 0.7846,
       "step": 14929
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.095462407395686e-06,
-      "loss": 0.5703,
+      "loss": 0.7194,
       "step": 14930
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.094289380459922e-06,
-      "loss": 0.8205,
+      "loss": 0.8626,
       "step": 14931
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.09311636608961e-06,
-      "loss": 0.7453,
+      "loss": 0.8358,
       "step": 14932
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.09194336430102e-06,
-      "loss": 0.6882,
+      "loss": 0.7793,
       "step": 14933
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.090770375110432e-06,
-      "loss": 0.7131,
+      "loss": 0.8497,
       "step": 14934
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.08959739853411e-06,
-      "loss": 0.637,
+      "loss": 0.8284,
       "step": 14935
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.088424434588336e-06,
-      "loss": 0.9162,
+      "loss": 0.9453,
       "step": 14936
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.087251483289376e-06,
-      "loss": 0.7556,
+      "loss": 0.9165,
       "step": 14937
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.08607854465351e-06,
-      "loss": 0.7437,
+      "loss": 0.8923,
       "step": 14938
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.084905618697006e-06,
-      "loss": 0.7802,
+      "loss": 0.8659,
       "step": 14939
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.083732705436138e-06,
-      "loss": 0.7528,
+      "loss": 0.8232,
       "step": 14940
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.082559804887178e-06,
-      "loss": 0.5656,
+      "loss": 0.6812,
       "step": 14941
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.081386917066403e-06,
-      "loss": 0.7857,
+      "loss": 0.8593,
       "step": 14942
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.080214041990074e-06,
-      "loss": 0.6895,
+      "loss": 0.8228,
       "step": 14943
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.079041179674478e-06,
-      "loss": 0.7452,
+      "loss": 0.8345,
       "step": 14944
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.077868330135874e-06,
-      "loss": 0.7756,
+      "loss": 0.8708,
       "step": 14945
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.076695493390541e-06,
-      "loss": 0.7299,
+      "loss": 0.8461,
       "step": 14946
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.07552266945475e-06,
-      "loss": 0.7431,
+      "loss": 0.8154,
       "step": 14947
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.074349858344763e-06,
-      "loss": 0.6333,
+      "loss": 0.8155,
       "step": 14948
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.073177060076866e-06,
-      "loss": 0.746,
+      "loss": 0.7766,
       "step": 14949
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.072004274667318e-06,
-      "loss": 0.6955,
+      "loss": 0.8369,
       "step": 14950
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.070831502132396e-06,
-      "loss": 0.7033,
+      "loss": 0.8534,
       "step": 14951
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.069658742488366e-06,
-      "loss": 0.8355,
+      "loss": 0.9014,
       "step": 14952
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.068485995751504e-06,
-      "loss": 0.7429,
+      "loss": 0.9072,
       "step": 14953
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.067313261938074e-06,
-      "loss": 0.9083,
+      "loss": 0.9162,
       "step": 14954
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.066140541064354e-06,
-      "loss": 0.6067,
+      "loss": 0.7945,
       "step": 14955
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.064967833146605e-06,
-      "loss": 0.6345,
+      "loss": 0.7403,
       "step": 14956
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.063795138201102e-06,
-      "loss": 0.6404,
+      "loss": 0.8127,
       "step": 14957
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.06262245624411e-06,
-      "loss": 0.7366,
+      "loss": 0.8232,
       "step": 14958
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.061449787291907e-06,
-      "loss": 0.765,
+      "loss": 0.8967,
       "step": 14959
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.06027713136075e-06,
-      "loss": 0.776,
+      "loss": 0.8974,
       "step": 14960
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.05910448846692e-06,
-      "loss": 0.7825,
+      "loss": 0.8437,
       "step": 14961
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.057931858626677e-06,
-      "loss": 0.7815,
+      "loss": 0.8734,
       "step": 14962
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.056759241856298e-06,
-      "loss": 0.9156,
+      "loss": 0.985,
       "step": 14963
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.055586638172041e-06,
-      "loss": 0.5806,
+      "loss": 0.6602,
       "step": 14964
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.054414047590183e-06,
-      "loss": 0.6944,
+      "loss": 0.8443,
       "step": 14965
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.053241470126986e-06,
-      "loss": 0.79,
+      "loss": 0.9029,
       "step": 14966
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.05206890579872e-06,
-      "loss": 0.5666,
+      "loss": 0.7154,
       "step": 14967
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.050896354621657e-06,
-      "loss": 0.8207,
+      "loss": 0.815,
       "step": 14968
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.049723816612056e-06,
-      "loss": 0.7584,
+      "loss": 0.7528,
       "step": 14969
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.04855129178619e-06,
-      "loss": 0.6414,
+      "loss": 0.7616,
       "step": 14970
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.047378780160324e-06,
-      "loss": 0.7494,
+      "loss": 0.9435,
       "step": 14971
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.046206281750727e-06,
-      "loss": 0.7835,
+      "loss": 0.8207,
       "step": 14972
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.045033796573664e-06,
-      "loss": 0.7823,
+      "loss": 0.8931,
       "step": 14973
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.043861324645406e-06,
-      "loss": 0.7712,
+      "loss": 0.8254,
       "step": 14974
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.042688865982207e-06,
-      "loss": 0.8474,
+      "loss": 0.9376,
       "step": 14975
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.041516420600351e-06,
-      "loss": 0.7019,
+      "loss": 0.7706,
       "step": 14976
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.040343988516089e-06,
-      "loss": 0.621,
+      "loss": 0.7903,
       "step": 14977
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.039171569745694e-06,
-      "loss": 0.745,
+      "loss": 0.8818,
       "step": 14978
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.037999164305431e-06,
-      "loss": 0.6134,
+      "loss": 0.8142,
       "step": 14979
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.036826772211565e-06,
-      "loss": 0.7684,
+      "loss": 0.8587,
       "step": 14980
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.035654393480358e-06,
-      "loss": 0.7371,
+      "loss": 0.7595,
       "step": 14981
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.034482028128082e-06,
-      "loss": 0.7305,
+      "loss": 0.8495,
       "step": 14982
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.033309676170994e-06,
-      "loss": 0.7477,
+      "loss": 0.8636,
       "step": 14983
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.032137337625367e-06,
-      "loss": 0.7357,
+      "loss": 0.8797,
       "step": 14984
     },
     {
       "epoch": 0.54,
       "learning_rate": 9.030965012507461e-06,
-      "loss": 0.7174,
+      "loss": 0.8419,
       "step": 14985
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.029792700833537e-06,
-      "loss": 0.8382,
+      "loss": 0.8574,
       "step": 14986
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.028620402619868e-06,
-      "loss": 0.6782,
+      "loss": 0.8118,
       "step": 14987
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.027448117882704e-06,
-      "loss": 0.6875,
+      "loss": 0.8036,
       "step": 14988
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.026275846638327e-06,
-      "loss": 0.7268,
+      "loss": 0.8162,
       "step": 14989
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.025103588902985e-06,
-      "loss": 0.7033,
+      "loss": 0.8101,
       "step": 14990
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.02393134469295e-06,
-      "loss": 0.7337,
+      "loss": 0.8585,
       "step": 14991
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.02275911402448e-06,
-      "loss": 0.8842,
+      "loss": 0.9028,
       "step": 14992
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.021586896913844e-06,
-      "loss": 0.8017,
+      "loss": 0.9018,
       "step": 14993
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.020414693377299e-06,
-      "loss": 0.777,
+      "loss": 0.8907,
       "step": 14994
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.019242503431113e-06,
-      "loss": 0.7195,
+      "loss": 0.7389,
       "step": 14995
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.018070327091545e-06,
-      "loss": 0.7885,
+      "loss": 0.8459,
       "step": 14996
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.016898164374858e-06,
-      "loss": 0.59,
+      "loss": 0.8044,
       "step": 14997
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.015726015297312e-06,
-      "loss": 0.6694,
+      "loss": 0.7228,
       "step": 14998
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.014553879875172e-06,
-      "loss": 0.6402,
+      "loss": 0.7817,
       "step": 14999
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.013381758124699e-06,
-      "loss": 0.7891,
+      "loss": 0.882,
       "step": 15000
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.012209650062157e-06,
-      "loss": 0.6789,
+      "loss": 0.821,
       "step": 15001
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.011037555703799e-06,
-      "loss": 0.635,
+      "loss": 0.7802,
       "step": 15002
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.009865475065897e-06,
-      "loss": 0.6787,
+      "loss": 0.8531,
       "step": 15003
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.008693408164704e-06,
-      "loss": 0.6789,
+      "loss": 0.823,
       "step": 15004
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.007521355016486e-06,
-      "loss": 0.7785,
+      "loss": 0.8701,
       "step": 15005
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.006349315637501e-06,
-      "loss": 0.7428,
+      "loss": 0.8234,
       "step": 15006
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.005177290044004e-06,
-      "loss": 0.8071,
+      "loss": 0.8815,
       "step": 15007
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.004005278252267e-06,
-      "loss": 0.6276,
+      "loss": 0.7911,
       "step": 15008
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.00283328027854e-06,
-      "loss": 0.6026,
+      "loss": 0.7022,
       "step": 15009
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.001661296139089e-06,
-      "loss": 0.5619,
+      "loss": 0.7624,
       "step": 15010
     },
     {
       "epoch": 0.55,
       "learning_rate": 9.000489325850168e-06,
-      "loss": 0.6265,
+      "loss": 0.7467,
       "step": 15011
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.99931736942804e-06,
-      "loss": 0.6611,
+      "loss": 0.8104,
       "step": 15012
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.998145426888963e-06,
-      "loss": 0.6481,
+      "loss": 0.7913,
       "step": 15013
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.996973498249201e-06,
-      "loss": 0.6885,
+      "loss": 0.8467,
       "step": 15014
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.995801583525003e-06,
-      "loss": 0.7717,
+      "loss": 0.8175,
       "step": 15015
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.99462968273264e-06,
-      "loss": 0.6819,
+      "loss": 0.8196,
       "step": 15016
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.993457795888358e-06,
-      "loss": 0.8135,
+      "loss": 0.941,
       "step": 15017
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.992285923008424e-06,
-      "loss": 0.6618,
+      "loss": 0.7357,
       "step": 15018
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.99111406410909e-06,
-      "loss": 0.7622,
+      "loss": 0.8542,
       "step": 15019
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.98994221920662e-06,
-      "loss": 0.6808,
+      "loss": 0.7421,
       "step": 15020
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.988770388317266e-06,
-      "loss": 0.7259,
+      "loss": 0.8323,
       "step": 15021
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.987598571457293e-06,
-      "loss": 0.769,
+      "loss": 0.8121,
       "step": 15022
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.986426768642949e-06,
-      "loss": 0.7872,
+      "loss": 0.8829,
       "step": 15023
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.985254979890498e-06,
-      "loss": 0.7863,
+      "loss": 0.8784,
       "step": 15024
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.984083205216193e-06,
-      "loss": 0.6547,
+      "loss": 0.8036,
       "step": 15025
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.982911444636292e-06,
-      "loss": 0.7229,
+      "loss": 0.8387,
       "step": 15026
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.981739698167053e-06,
-      "loss": 0.5633,
+      "loss": 0.7077,
       "step": 15027
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.980567965824729e-06,
-      "loss": 0.7251,
+      "loss": 0.8302,
       "step": 15028
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.979396247625579e-06,
-      "loss": 0.8284,
+      "loss": 0.8751,
       "step": 15029
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.978224543585858e-06,
-      "loss": 0.6572,
+      "loss": 0.8654,
       "step": 15030
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.977052853721822e-06,
-      "loss": 0.6973,
+      "loss": 0.8227,
       "step": 15031
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.975881178049723e-06,
-      "loss": 0.6766,
+      "loss": 0.8898,
       "step": 15032
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.974709516585825e-06,
-      "loss": 0.7361,
+      "loss": 0.8156,
       "step": 15033
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.973537869346373e-06,
-      "loss": 0.7369,
+      "loss": 0.8076,
       "step": 15034
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.972366236347631e-06,
-      "loss": 0.7497,
+      "loss": 0.8829,
       "step": 15035
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.971194617605847e-06,
-      "loss": 0.7578,
+      "loss": 0.851,
       "step": 15036
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.97002301313728e-06,
-      "loss": 0.7471,
+      "loss": 0.8912,
       "step": 15037
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.968851422958178e-06,
-      "loss": 0.7347,
+      "loss": 0.8367,
       "step": 15038
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.967679847084804e-06,
-      "loss": 0.8722,
+      "loss": 0.9633,
       "step": 15039
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.966508285533404e-06,
-      "loss": 0.7639,
+      "loss": 0.9158,
       "step": 15040
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.96533673832024e-06,
-      "loss": 0.7427,
+      "loss": 0.855,
       "step": 15041
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.964165205461554e-06,
-      "loss": 0.6323,
+      "loss": 0.7734,
       "step": 15042
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.962993686973612e-06,
-      "loss": 0.7891,
+      "loss": 0.9064,
       "step": 15043
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.96182218287266e-06,
-      "loss": 0.5916,
+      "loss": 0.7663,
       "step": 15044
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.96065069317495e-06,
-      "loss": 0.7163,
+      "loss": 0.781,
       "step": 15045
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.959479217896741e-06,
-      "loss": 0.6718,
+      "loss": 0.7852,
       "step": 15046
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.958307757054275e-06,
-      "loss": 0.6487,
+      "loss": 0.7892,
       "step": 15047
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.957136310663819e-06,
-      "loss": 0.7059,
+      "loss": 0.8434,
       "step": 15048
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.95596487874161e-06,
-      "loss": 0.6493,
+      "loss": 0.823,
       "step": 15049
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.954793461303911e-06,
-      "loss": 0.779,
+      "loss": 0.8418,
       "step": 15050
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.953622058366967e-06,
-      "loss": 0.6499,
+      "loss": 0.7521,
       "step": 15051
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.952450669947034e-06,
-      "loss": 0.7081,
+      "loss": 0.7797,
       "step": 15052
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.951279296060361e-06,
-      "loss": 0.6835,
+      "loss": 0.7848,
       "step": 15053
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.950107936723203e-06,
-      "loss": 0.7262,
+      "loss": 0.781,
       "step": 15054
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.948936591951803e-06,
-      "loss": 0.744,
+      "loss": 0.7945,
       "step": 15055
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.94776526176242e-06,
-      "loss": 0.725,
+      "loss": 0.9557,
       "step": 15056
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.946593946171299e-06,
-      "loss": 0.7103,
+      "loss": 0.7993,
       "step": 15057
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.945422645194695e-06,
-      "loss": 0.5793,
+      "loss": 0.7554,
       "step": 15058
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.944251358848852e-06,
-      "loss": 0.8366,
+      "loss": 0.9092,
       "step": 15059
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.943080087150025e-06,
-      "loss": 0.7858,
+      "loss": 0.9364,
       "step": 15060
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.941908830114462e-06,
-      "loss": 0.6938,
+      "loss": 0.8977,
       "step": 15061
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.940737587758415e-06,
-      "loss": 0.7073,
+      "loss": 0.8594,
       "step": 15062
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.939566360098131e-06,
-      "loss": 0.8233,
+      "loss": 0.8446,
       "step": 15063
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.938395147149857e-06,
-      "loss": 0.7437,
+      "loss": 0.8348,
       "step": 15064
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.937223948929845e-06,
-      "loss": 0.8187,
+      "loss": 0.8322,
       "step": 15065
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.936052765454341e-06,
-      "loss": 0.8629,
+      "loss": 0.9035,
       "step": 15066
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.9348815967396e-06,
-      "loss": 0.7682,
+      "loss": 0.8591,
       "step": 15067
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.93371044280186e-06,
-      "loss": 0.7174,
+      "loss": 0.8192,
       "step": 15068
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.932539303657378e-06,
-      "loss": 0.6381,
+      "loss": 0.7708,
       "step": 15069
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.931368179322396e-06,
-      "loss": 0.7956,
+      "loss": 0.8751,
       "step": 15070
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.930197069813167e-06,
-      "loss": 0.8212,
+      "loss": 0.8867,
       "step": 15071
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.929025975145934e-06,
-      "loss": 0.7902,
+      "loss": 0.8778,
       "step": 15072
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.92785489533695e-06,
-      "loss": 0.6685,
+      "loss": 0.7486,
       "step": 15073
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.92668383040245e-06,
-      "loss": 0.7559,
+      "loss": 0.8166,
       "step": 15074
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.925512780358699e-06,
-      "loss": 0.7267,
+      "loss": 0.873,
       "step": 15075
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.924341745221926e-06,
-      "loss": 0.7378,
+      "loss": 0.7467,
       "step": 15076
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.923170725008389e-06,
-      "loss": 0.7189,
+      "loss": 0.7739,
       "step": 15077
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.921999719734328e-06,
-      "loss": 0.6262,
+      "loss": 0.8127,
       "step": 15078
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.920828729415993e-06,
-      "loss": 0.7038,
+      "loss": 0.7505,
       "step": 15079
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.919657754069627e-06,
-      "loss": 0.7759,
+      "loss": 0.887,
       "step": 15080
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.91848679371148e-06,
-      "loss": 0.7076,
+      "loss": 0.8675,
       "step": 15081
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.917315848357794e-06,
-      "loss": 0.7824,
+      "loss": 0.8117,
       "step": 15082
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.916144918024811e-06,
-      "loss": 0.6771,
+      "loss": 0.7369,
       "step": 15083
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.914974002728782e-06,
-      "loss": 0.6478,
+      "loss": 0.8871,
       "step": 15084
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.913803102485946e-06,
-      "loss": 0.6969,
+      "loss": 0.7472,
       "step": 15085
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.912632217312556e-06,
-      "loss": 0.6196,
+      "loss": 0.7328,
       "step": 15086
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.911461347224846e-06,
-      "loss": 0.5807,
+      "loss": 0.7563,
       "step": 15087
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.91029049223907e-06,
-      "loss": 0.7032,
+      "loss": 0.8264,
       "step": 15088
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.909119652371463e-06,
-      "loss": 0.845,
+      "loss": 0.9196,
       "step": 15089
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.907948827638276e-06,
-      "loss": 0.8581,
+      "loss": 0.9178,
       "step": 15090
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.906778018055748e-06,
-      "loss": 0.768,
+      "loss": 0.8531,
       "step": 15091
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.905607223640124e-06,
-      "loss": 0.6489,
+      "loss": 0.8619,
       "step": 15092
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.904436444407648e-06,
-      "loss": 0.7715,
+      "loss": 0.8856,
       "step": 15093
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.903265680374564e-06,
-      "loss": 0.6944,
+      "loss": 0.8,
       "step": 15094
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.902094931557107e-06,
-      "loss": 0.7314,
+      "loss": 0.9008,
       "step": 15095
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.900924197971531e-06,
-      "loss": 0.6865,
+      "loss": 0.8311,
       "step": 15096
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.899753479634067e-06,
-      "loss": 0.8202,
+      "loss": 0.9046,
       "step": 15097
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.898582776560967e-06,
-      "loss": 0.7918,
+      "loss": 0.8329,
       "step": 15098
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.897412088768465e-06,
-      "loss": 0.7745,
+      "loss": 0.8727,
       "step": 15099
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.89624141627281e-06,
-      "loss": 0.7696,
+      "loss": 0.8768,
       "step": 15100
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.895070759090238e-06,
-      "loss": 0.7796,
+      "loss": 0.8696,
       "step": 15101
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.893900117236988e-06,
-      "loss": 0.7862,
+      "loss": 0.8533,
       "step": 15102
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.892729490729307e-06,
-      "loss": 0.7171,
+      "loss": 0.8484,
       "step": 15103
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.891558879583432e-06,
-      "loss": 0.7074,
+      "loss": 0.797,
       "step": 15104
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.890388283815607e-06,
-      "loss": 0.7611,
+      "loss": 0.8049,
       "step": 15105
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.889217703442066e-06,
-      "loss": 0.7508,
+      "loss": 0.8119,
       "step": 15106
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.88804713847906e-06,
-      "loss": 0.8498,
+      "loss": 0.907,
       "step": 15107
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.886876588942817e-06,
-      "loss": 0.8561,
+      "loss": 0.8636,
       "step": 15108
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.885706054849585e-06,
-      "loss": 0.7341,
+      "loss": 0.8611,
       "step": 15109
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.884535536215598e-06,
-      "loss": 0.7941,
+      "loss": 0.9261,
       "step": 15110
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.8833650330571e-06,
-      "loss": 0.8439,
+      "loss": 0.8865,
       "step": 15111
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.882194545390327e-06,
-      "loss": 0.7908,
+      "loss": 0.8553,
       "step": 15112
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.88102407323152e-06,
-      "loss": 0.7698,
+      "loss": 0.8451,
       "step": 15113
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.879853616596912e-06,
-      "loss": 0.7918,
+      "loss": 0.8814,
       "step": 15114
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.878683175502752e-06,
-      "loss": 0.7925,
+      "loss": 0.8097,
       "step": 15115
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.877512749965267e-06,
-      "loss": 0.7311,
+      "loss": 0.9003,
       "step": 15116
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.876342340000702e-06,
-      "loss": 0.6777,
+      "loss": 0.8102,
       "step": 15117
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.875171945625292e-06,
-      "loss": 0.5675,
+      "loss": 0.7367,
       "step": 15118
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.874001566855276e-06,
-      "loss": 0.5892,
+      "loss": 0.7961,
       "step": 15119
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.872831203706893e-06,
-      "loss": 0.686,
+      "loss": 0.7728,
       "step": 15120
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.871660856196374e-06,
-      "loss": 0.7293,
+      "loss": 0.8287,
       "step": 15121
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.870490524339963e-06,
-      "loss": 0.8707,
+      "loss": 0.8933,
       "step": 15122
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.86932020815389e-06,
-      "loss": 0.6628,
+      "loss": 0.7328,
       "step": 15123
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.868149907654397e-06,
-      "loss": 0.7762,
+      "loss": 0.892,
       "step": 15124
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.866979622857715e-06,
-      "loss": 0.6211,
+      "loss": 0.7544,
       "step": 15125
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.865809353780089e-06,
-      "loss": 0.7105,
+      "loss": 0.8014,
       "step": 15126
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.864639100437741e-06,
-      "loss": 0.7587,
+      "loss": 0.8282,
       "step": 15127
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.863468862846924e-06,
-      "loss": 0.7496,
+      "loss": 0.8747,
       "step": 15128
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.862298641023858e-06,
-      "loss": 0.7489,
+      "loss": 0.8972,
       "step": 15129
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.861128434984785e-06,
-      "loss": 0.6315,
+      "loss": 0.7434,
       "step": 15130
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.859958244745938e-06,
-      "loss": 0.703,
+      "loss": 0.8183,
       "step": 15131
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.858788070323556e-06,
-      "loss": 0.786,
+      "loss": 0.8085,
       "step": 15132
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.857617911733868e-06,
-      "loss": 0.6683,
+      "loss": 0.7419,
       "step": 15133
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.856447768993114e-06,
-      "loss": 0.679,
+      "loss": 0.9127,
       "step": 15134
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.85527764211752e-06,
-      "loss": 0.6323,
+      "loss": 0.7841,
       "step": 15135
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.854107531123329e-06,
-      "loss": 0.7443,
+      "loss": 0.8354,
       "step": 15136
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.852937436026766e-06,
-      "loss": 0.6145,
+      "loss": 0.8079,
       "step": 15137
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.851767356844071e-06,
-      "loss": 0.7364,
+      "loss": 0.8595,
       "step": 15138
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.850597293591477e-06,
-      "loss": 0.9304,
+      "loss": 0.997,
       "step": 15139
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.849427246285212e-06,
-      "loss": 0.9982,
+      "loss": 1.0415,
       "step": 15140
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.84825721494151e-06,
-      "loss": 0.5733,
+      "loss": 0.7815,
       "step": 15141
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.847087199576607e-06,
-      "loss": 0.6855,
+      "loss": 0.7928,
       "step": 15142
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.845917200206734e-06,
-      "loss": 0.6756,
+      "loss": 0.8112,
       "step": 15143
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.84474721684812e-06,
-      "loss": 0.6138,
+      "loss": 0.7593,
       "step": 15144
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.843577249517004e-06,
-      "loss": 0.7144,
+      "loss": 0.784,
       "step": 15145
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.842407298229605e-06,
-      "loss": 0.7906,
+      "loss": 0.9028,
       "step": 15146
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.84123736300217e-06,
-      "loss": 0.6532,
+      "loss": 0.7666,
       "step": 15147
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.840067443850917e-06,
-      "loss": 0.6545,
+      "loss": 0.798,
       "step": 15148
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.838897540792086e-06,
-      "loss": 0.692,
+      "loss": 0.8061,
       "step": 15149
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.837727653841904e-06,
-      "loss": 0.802,
+      "loss": 0.8472,
       "step": 15150
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.836557783016601e-06,
-      "loss": 0.5205,
+      "loss": 0.6954,
       "step": 15151
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.835387928332406e-06,
-      "loss": 0.7016,
+      "loss": 0.788,
       "step": 15152
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.834218089805557e-06,
-      "loss": 0.672,
+      "loss": 0.7504,
       "step": 15153
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.833048267452269e-06,
-      "loss": 0.7208,
+      "loss": 0.847,
       "step": 15154
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.831878461288789e-06,
-      "loss": 0.6827,
+      "loss": 0.8487,
       "step": 15155
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.830708671331335e-06,
-      "loss": 0.732,
+      "loss": 0.8243,
       "step": 15156
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.829538897596138e-06,
-      "loss": 0.8288,
+      "loss": 0.9259,
       "step": 15157
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.828369140099433e-06,
-      "loss": 0.7759,
+      "loss": 0.7663,
       "step": 15158
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.827199398857435e-06,
-      "loss": 0.7576,
+      "loss": 0.8544,
       "step": 15159
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.82602967388639e-06,
-      "loss": 0.7864,
+      "loss": 0.8513,
       "step": 15160
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.82485996520251e-06,
-      "loss": 0.6915,
+      "loss": 0.8389,
       "step": 15161
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.823690272822037e-06,
-      "loss": 0.6562,
+      "loss": 0.8178,
       "step": 15162
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.822520596761188e-06,
-      "loss": 0.749,
+      "loss": 0.8283,
       "step": 15163
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.821350937036198e-06,
-      "loss": 0.6039,
+      "loss": 0.8431,
       "step": 15164
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.820181293663289e-06,
-      "loss": 0.8376,
+      "loss": 0.9184,
       "step": 15165
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.819011666658696e-06,
-      "loss": 0.8012,
+      "loss": 0.9076,
       "step": 15166
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.817842056038635e-06,
-      "loss": 0.7902,
+      "loss": 0.8589,
       "step": 15167
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.81667246181934e-06,
-      "loss": 0.7341,
+      "loss": 0.8496,
       "step": 15168
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.815502884017036e-06,
-      "loss": 0.7049,
+      "loss": 0.8655,
       "step": 15169
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.814333322647948e-06,
-      "loss": 0.8239,
+      "loss": 0.9203,
       "step": 15170
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.813163777728303e-06,
-      "loss": 0.8379,
+      "loss": 0.9203,
       "step": 15171
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.811994249274329e-06,
-      "loss": 0.8465,
+      "loss": 0.8966,
       "step": 15172
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.810824737302244e-06,
-      "loss": 0.7988,
+      "loss": 0.8387,
       "step": 15173
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.809655241828284e-06,
-      "loss": 0.6222,
+      "loss": 0.7694,
       "step": 15174
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.808485762868666e-06,
-      "loss": 0.6633,
+      "loss": 0.7949,
       "step": 15175
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.807316300439618e-06,
-      "loss": 0.8077,
+      "loss": 0.918,
       "step": 15176
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.806146854557364e-06,
-      "loss": 0.7642,
+      "loss": 0.8652,
       "step": 15177
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.804977425238127e-06,
-      "loss": 0.717,
+      "loss": 0.8895,
       "step": 15178
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.803808012498138e-06,
-      "loss": 0.8116,
+      "loss": 0.8307,
       "step": 15179
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.802638616353612e-06,
-      "loss": 0.6988,
+      "loss": 0.8277,
       "step": 15180
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.801469236820777e-06,
-      "loss": 0.8551,
+      "loss": 0.8696,
       "step": 15181
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.800299873915854e-06,
-      "loss": 0.6483,
+      "loss": 0.7362,
       "step": 15182
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.79913052765507e-06,
-      "loss": 0.7427,
+      "loss": 0.8786,
       "step": 15183
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.797961198054642e-06,
-      "loss": 0.7334,
+      "loss": 0.7757,
       "step": 15184
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.796791885130804e-06,
-      "loss": 0.8238,
+      "loss": 0.8477,
       "step": 15185
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.795622588899764e-06,
-      "loss": 0.8552,
+      "loss": 0.9434,
       "step": 15186
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.794453309377759e-06,
-      "loss": 0.6606,
+      "loss": 0.8239,
       "step": 15187
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.793284046581e-06,
-      "loss": 0.6653,
+      "loss": 0.7813,
       "step": 15188
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.792114800525714e-06,
-      "loss": 0.8963,
+      "loss": 0.9662,
       "step": 15189
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.790945571228119e-06,
-      "loss": 0.6637,
+      "loss": 0.7979,
       "step": 15190
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.789776358704441e-06,
-      "loss": 0.7112,
+      "loss": 0.7658,
       "step": 15191
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.788607162970897e-06,
-      "loss": 0.7489,
+      "loss": 0.8011,
       "step": 15192
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.787437984043714e-06,
-      "loss": 0.8082,
+      "loss": 0.9026,
       "step": 15193
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.786268821939105e-06,
-      "loss": 0.5737,
+      "loss": 0.7437,
       "step": 15194
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.785099676673296e-06,
-      "loss": 0.7961,
+      "loss": 0.8617,
       "step": 15195
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.783930548262503e-06,
-      "loss": 0.5174,
+      "loss": 0.6901,
       "step": 15196
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.782761436722949e-06,
-      "loss": 0.6184,
+      "loss": 0.7722,
       "step": 15197
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.781592342070856e-06,
-      "loss": 0.7732,
+      "loss": 0.9019,
       "step": 15198
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.780423264322434e-06,
-      "loss": 0.7826,
+      "loss": 0.8119,
       "step": 15199
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.779254203493916e-06,
-      "loss": 0.7403,
+      "loss": 0.7677,
       "step": 15200
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.77808515960151e-06,
-      "loss": 0.7748,
+      "loss": 0.8849,
       "step": 15201
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.776916132661439e-06,
-      "loss": 0.7933,
+      "loss": 0.898,
       "step": 15202
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.775747122689922e-06,
-      "loss": 0.6688,
+      "loss": 0.7777,
       "step": 15203
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.774578129703176e-06,
-      "loss": 0.7294,
+      "loss": 0.8384,
       "step": 15204
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.77340915371742e-06,
-      "loss": 0.5485,
+      "loss": 0.7717,
       "step": 15205
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.772240194748875e-06,
-      "loss": 0.7227,
+      "loss": 0.8677,
       "step": 15206
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.77107125281375e-06,
-      "loss": 0.6731,
+      "loss": 0.8016,
       "step": 15207
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.76990232792827e-06,
-      "loss": 0.7257,
+      "loss": 0.838,
       "step": 15208
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.76873342010865e-06,
-      "loss": 0.7322,
+      "loss": 0.8905,
       "step": 15209
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.767564529371106e-06,
-      "loss": 0.6981,
+      "loss": 0.872,
       "step": 15210
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.766395655731856e-06,
-      "loss": 0.8063,
+      "loss": 0.8633,
       "step": 15211
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.765226799207119e-06,
-      "loss": 0.7638,
+      "loss": 0.8632,
       "step": 15212
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.764057959813103e-06,
-      "loss": 0.739,
+      "loss": 0.8705,
       "step": 15213
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.762889137566033e-06,
-      "loss": 0.8292,
+      "loss": 0.9138,
       "step": 15214
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.761720332482122e-06,
-      "loss": 0.8348,
+      "loss": 0.8497,
       "step": 15215
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.76055154457758e-06,
-      "loss": 0.7219,
+      "loss": 0.8759,
       "step": 15216
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.759382773868634e-06,
-      "loss": 0.7659,
+      "loss": 0.8673,
       "step": 15217
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.758214020371481e-06,
-      "loss": 0.8083,
+      "loss": 0.8895,
       "step": 15218
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.757045284102356e-06,
-      "loss": 0.7134,
+      "loss": 0.7834,
       "step": 15219
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.75587656507746e-06,
-      "loss": 0.8395,
+      "loss": 0.9163,
       "step": 15220
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.754707863313012e-06,
-      "loss": 0.7362,
+      "loss": 0.8046,
       "step": 15221
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.753539178825225e-06,
-      "loss": 0.8344,
+      "loss": 0.8525,
       "step": 15222
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.752370511630315e-06,
-      "loss": 0.6897,
+      "loss": 0.7994,
       "step": 15223
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.751201861744491e-06,
-      "loss": 0.7167,
+      "loss": 0.781,
       "step": 15224
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.750033229183976e-06,
-      "loss": 0.613,
+      "loss": 0.804,
       "step": 15225
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.748864613964968e-06,
-      "loss": 0.656,
+      "loss": 0.8195,
       "step": 15226
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.747696016103696e-06,
-      "loss": 0.6703,
+      "loss": 0.8245,
       "step": 15227
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.746527435616358e-06,
-      "loss": 0.832,
+      "loss": 0.8827,
       "step": 15228
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.745358872519179e-06,
-      "loss": 0.8356,
+      "loss": 0.8862,
       "step": 15229
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.744190326828362e-06,
-      "loss": 0.7968,
+      "loss": 0.9066,
       "step": 15230
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.743021798560123e-06,
-      "loss": 0.6222,
+      "loss": 0.8009,
       "step": 15231
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.741853287730673e-06,
-      "loss": 0.6368,
+      "loss": 0.8435,
       "step": 15232
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.740684794356226e-06,
-      "loss": 0.654,
+      "loss": 0.8464,
       "step": 15233
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.739516318452989e-06,
-      "loss": 0.7135,
+      "loss": 0.8148,
       "step": 15234
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.738347860037176e-06,
-      "loss": 0.6348,
+      "loss": 0.7435,
       "step": 15235
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.737179419124995e-06,
-      "loss": 0.8637,
+      "loss": 0.913,
       "step": 15236
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.736010995732657e-06,
-      "loss": 0.8447,
+      "loss": 0.8928,
       "step": 15237
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.734842589876377e-06,
-      "loss": 0.7826,
+      "loss": 0.8148,
       "step": 15238
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.733674201572357e-06,
-      "loss": 0.7147,
+      "loss": 0.8212,
       "step": 15239
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.732505830836814e-06,
-      "loss": 0.6209,
+      "loss": 0.7187,
       "step": 15240
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.73133747768595e-06,
-      "loss": 0.6944,
+      "loss": 0.8517,
       "step": 15241
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.730169142135981e-06,
-      "loss": 0.6816,
+      "loss": 0.8204,
       "step": 15242
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.729000824203111e-06,
-      "loss": 0.8752,
+      "loss": 0.9616,
       "step": 15243
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.727832523903556e-06,
-      "loss": 0.6532,
+      "loss": 0.7823,
       "step": 15244
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.726664241253514e-06,
-      "loss": 0.7782,
+      "loss": 0.8595,
       "step": 15245
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.725495976269205e-06,
-      "loss": 0.7108,
+      "loss": 0.8254,
       "step": 15246
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.724327728966827e-06,
-      "loss": 0.7181,
+      "loss": 0.8157,
       "step": 15247
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.723159499362595e-06,
-      "loss": 0.7398,
+      "loss": 0.8547,
       "step": 15248
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.721991287472709e-06,
-      "loss": 0.7162,
+      "loss": 0.7991,
       "step": 15249
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.720823093313384e-06,
-      "loss": 0.7578,
+      "loss": 0.8976,
       "step": 15250
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.719654916900822e-06,
-      "loss": 0.761,
+      "loss": 0.8685,
       "step": 15251
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.718486758251234e-06,
-      "loss": 0.6467,
+      "loss": 0.7702,
       "step": 15252
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.717318617380819e-06,
-      "loss": 0.7539,
+      "loss": 0.8725,
       "step": 15253
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.716150494305795e-06,
-      "loss": 0.699,
+      "loss": 0.8536,
       "step": 15254
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.714982389042358e-06,
-      "loss": 0.6059,
+      "loss": 0.769,
       "step": 15255
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.713814301606716e-06,
-      "loss": 0.9311,
+      "loss": 0.9587,
       "step": 15256
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.71264623201508e-06,
-      "loss": 0.5672,
+      "loss": 0.7777,
       "step": 15257
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.711478180283644e-06,
-      "loss": 0.7236,
+      "loss": 0.8311,
       "step": 15258
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.710310146428626e-06,
-      "loss": 0.5795,
+      "loss": 0.769,
       "step": 15259
     },
     {
       "epoch": 0.55,
       "learning_rate": 8.70914213046622e-06,
-      "loss": 0.6854,
+      "loss": 0.8169,
       "step": 15260
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.70797413241264e-06,
-      "loss": 0.7604,
+      "loss": 0.8584,
       "step": 15261
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.706806152284084e-06,
-      "loss": 0.779,
+      "loss": 0.8148,
       "step": 15262
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.705638190096757e-06,
-      "loss": 0.6972,
+      "loss": 0.7999,
       "step": 15263
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.704470245866863e-06,
-      "loss": 0.6463,
+      "loss": 0.7731,
       "step": 15264
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.70330231961061e-06,
-      "loss": 0.6657,
+      "loss": 0.7637,
       "step": 15265
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.702134411344194e-06,
-      "loss": 0.7216,
+      "loss": 0.8572,
       "step": 15266
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.700966521083824e-06,
-      "loss": 0.7261,
+      "loss": 0.8784,
       "step": 15267
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.699798648845698e-06,
-      "loss": 0.8505,
+      "loss": 0.9276,
       "step": 15268
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.698630794646023e-06,
-      "loss": 0.6601,
+      "loss": 0.7625,
       "step": 15269
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.697462958500998e-06,
-      "loss": 0.7177,
+      "loss": 0.7718,
       "step": 15270
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.696295140426827e-06,
-      "loss": 0.7069,
+      "loss": 0.8285,
       "step": 15271
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.695127340439709e-06,
-      "loss": 0.7106,
+      "loss": 0.8418,
       "step": 15272
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.693959558555853e-06,
-      "loss": 0.6943,
+      "loss": 0.8292,
       "step": 15273
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.692791794791453e-06,
-      "loss": 0.6547,
+      "loss": 0.8441,
       "step": 15274
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.69162404916271e-06,
-      "loss": 0.7053,
+      "loss": 0.8548,
       "step": 15275
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.690456321685828e-06,
-      "loss": 0.7891,
+      "loss": 0.9685,
       "step": 15276
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.689288612377006e-06,
-      "loss": 0.7759,
+      "loss": 0.8112,
       "step": 15277
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.688120921252448e-06,
-      "loss": 0.7495,
+      "loss": 0.8427,
       "step": 15278
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.686953248328347e-06,
-      "loss": 0.7709,
+      "loss": 0.8447,
       "step": 15279
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.68578559362091e-06,
-      "loss": 0.7449,
+      "loss": 0.8476,
       "step": 15280
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.684617957146329e-06,
-      "loss": 0.8831,
+      "loss": 0.9338,
       "step": 15281
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.68345033892081e-06,
-      "loss": 0.8028,
+      "loss": 0.8494,
       "step": 15282
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.682282738960548e-06,
-      "loss": 0.7098,
+      "loss": 0.7901,
       "step": 15283
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.681115157281748e-06,
-      "loss": 0.7675,
+      "loss": 0.9152,
       "step": 15284
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.679947593900597e-06,
-      "loss": 0.6983,
+      "loss": 0.8324,
       "step": 15285
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.678780048833309e-06,
-      "loss": 0.7119,
+      "loss": 0.8483,
       "step": 15286
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.677612522096066e-06,
-      "loss": 0.6599,
+      "loss": 0.8023,
       "step": 15287
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.676445013705076e-06,
-      "loss": 0.7751,
+      "loss": 0.8541,
       "step": 15288
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.675277523676532e-06,
-      "loss": 0.6958,
+      "loss": 0.83,
       "step": 15289
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.674110052026633e-06,
-      "loss": 0.8191,
+      "loss": 0.8616,
       "step": 15290
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.672942598771576e-06,
-      "loss": 0.6744,
+      "loss": 0.911,
       "step": 15291
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.671775163927562e-06,
-      "loss": 0.6648,
+      "loss": 0.7546,
       "step": 15292
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.67060774751078e-06,
-      "loss": 0.6652,
+      "loss": 0.805,
       "step": 15293
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.669440349537427e-06,
-      "loss": 0.7783,
+      "loss": 0.8464,
       "step": 15294
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.668272970023703e-06,
-      "loss": 0.6738,
+      "loss": 0.8124,
       "step": 15295
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.6671056089858e-06,
-      "loss": 0.7819,
+      "loss": 0.8722,
       "step": 15296
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.665938266439921e-06,
-      "loss": 0.9354,
+      "loss": 1.0023,
       "step": 15297
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.66477094240225e-06,
-      "loss": 0.7719,
+      "loss": 0.8715,
       "step": 15298
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.663603636888994e-06,
-      "loss": 0.8138,
+      "loss": 0.9075,
       "step": 15299
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.662436349916335e-06,
-      "loss": 0.6863,
+      "loss": 0.7861,
       "step": 15300
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.661269081500477e-06,
-      "loss": 0.8348,
+      "loss": 0.8807,
       "step": 15301
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.660101831657608e-06,
-      "loss": 0.6629,
+      "loss": 0.7855,
       "step": 15302
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.65893460040393e-06,
-      "loss": 0.662,
+      "loss": 0.7727,
       "step": 15303
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.657767387755627e-06,
-      "loss": 0.7055,
+      "loss": 0.8033,
       "step": 15304
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.656600193728903e-06,
-      "loss": 0.9067,
+      "loss": 0.9788,
       "step": 15305
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.65543301833994e-06,
-      "loss": 0.7567,
+      "loss": 0.7927,
       "step": 15306
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.654265861604938e-06,
-      "loss": 0.7851,
+      "loss": 0.8988,
       "step": 15307
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.653098723540087e-06,
-      "loss": 0.5597,
+      "loss": 0.7411,
       "step": 15308
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.65193160416158e-06,
-      "loss": 0.7375,
+      "loss": 0.7829,
       "step": 15309
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.65076450348561e-06,
-      "loss": 0.6905,
+      "loss": 0.8499,
       "step": 15310
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.64959742152837e-06,
-      "loss": 0.7262,
+      "loss": 0.7912,
       "step": 15311
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.64843035830605e-06,
-      "loss": 0.6894,
+      "loss": 0.7794,
       "step": 15312
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.647263313834838e-06,
-      "loss": 0.723,
+      "loss": 0.835,
       "step": 15313
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.64609628813093e-06,
-      "loss": 0.7675,
+      "loss": 0.8536,
       "step": 15314
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.644929281210514e-06,
-      "loss": 0.6894,
+      "loss": 0.8687,
       "step": 15315
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.643762293089783e-06,
-      "loss": 0.6966,
+      "loss": 0.8091,
       "step": 15316
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.642595323784924e-06,
-      "loss": 0.793,
+      "loss": 0.9277,
       "step": 15317
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.641428373312133e-06,
-      "loss": 0.7571,
+      "loss": 0.8783,
       "step": 15318
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.640261441687591e-06,
-      "loss": 0.7197,
+      "loss": 0.846,
       "step": 15319
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.639094528927495e-06,
-      "loss": 0.7867,
+      "loss": 0.8698,
       "step": 15320
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.637927635048029e-06,
-      "loss": 0.7184,
+      "loss": 0.8062,
       "step": 15321
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.636760760065387e-06,
-      "loss": 0.6869,
+      "loss": 0.8411,
       "step": 15322
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.635593903995754e-06,
-      "loss": 0.6969,
+      "loss": 0.8189,
       "step": 15323
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.634427066855322e-06,
-      "loss": 0.7495,
+      "loss": 0.8479,
       "step": 15324
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.633260248660272e-06,
-      "loss": 0.7516,
+      "loss": 0.8884,
       "step": 15325
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.632093449426804e-06,
-      "loss": 0.7162,
+      "loss": 0.7865,
       "step": 15326
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.630926669171092e-06,
-      "loss": 0.6629,
+      "loss": 0.8337,
       "step": 15327
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.629759907909333e-06,
-      "loss": 0.7882,
+      "loss": 0.804,
       "step": 15328
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.62859316565771e-06,
-      "loss": 0.6838,
+      "loss": 0.7447,
       "step": 15329
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.627426442432413e-06,
-      "loss": 0.671,
+      "loss": 0.825,
       "step": 15330
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.62625973824963e-06,
-      "loss": 0.6419,
+      "loss": 0.8015,
       "step": 15331
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.625093053125538e-06,
-      "loss": 0.5649,
+      "loss": 0.6802,
       "step": 15332
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.623926387076332e-06,
-      "loss": 0.7492,
+      "loss": 0.7652,
       "step": 15333
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.62275974011819e-06,
-      "loss": 0.5943,
+      "loss": 0.7373,
       "step": 15334
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.621593112267309e-06,
-      "loss": 0.7032,
+      "loss": 0.8429,
       "step": 15335
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.620426503539863e-06,
-      "loss": 0.6064,
+      "loss": 0.7631,
       "step": 15336
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.619259913952047e-06,
-      "loss": 0.6603,
+      "loss": 0.7835,
       "step": 15337
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.618093343520037e-06,
-      "loss": 0.726,
+      "loss": 0.8833,
       "step": 15338
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.616926792260022e-06,
-      "loss": 0.7606,
+      "loss": 0.8666,
       "step": 15339
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.615760260188184e-06,
-      "loss": 0.6716,
+      "loss": 0.7869,
       "step": 15340
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.614593747320711e-06,
-      "loss": 0.8433,
+      "loss": 0.899,
       "step": 15341
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.613427253673783e-06,
-      "loss": 0.7359,
+      "loss": 0.8445,
       "step": 15342
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.612260779263584e-06,
-      "loss": 0.7944,
+      "loss": 0.8801,
       "step": 15343
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.611094324106296e-06,
-      "loss": 0.6812,
+      "loss": 0.8324,
       "step": 15344
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.60992788821811e-06,
-      "loss": 0.7524,
+      "loss": 0.8348,
       "step": 15345
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.608761471615197e-06,
-      "loss": 0.7183,
+      "loss": 0.828,
       "step": 15346
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.607595074313746e-06,
-      "loss": 0.701,
+      "loss": 0.7642,
       "step": 15347
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.606428696329936e-06,
-      "loss": 0.9453,
+      "loss": 0.9506,
       "step": 15348
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.605262337679954e-06,
-      "loss": 0.7072,
+      "loss": 0.8107,
       "step": 15349
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.604095998379979e-06,
-      "loss": 0.6941,
+      "loss": 0.8478,
       "step": 15350
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.602929678446189e-06,
-      "loss": 0.6738,
+      "loss": 0.7594,
       "step": 15351
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.601763377894767e-06,
-      "loss": 0.6662,
+      "loss": 0.8722,
       "step": 15352
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.600597096741893e-06,
-      "loss": 0.8036,
+      "loss": 0.9228,
       "step": 15353
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.599430835003751e-06,
-      "loss": 0.6881,
+      "loss": 0.8059,
       "step": 15354
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.598264592696518e-06,
-      "loss": 0.71,
+      "loss": 0.8886,
       "step": 15355
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.597098369836378e-06,
-      "loss": 0.7482,
+      "loss": 0.8771,
       "step": 15356
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.595932166439502e-06,
-      "loss": 0.7901,
+      "loss": 0.9044,
       "step": 15357
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.594765982522081e-06,
-      "loss": 0.7967,
+      "loss": 0.8669,
       "step": 15358
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.593599818100285e-06,
-      "loss": 0.5861,
+      "loss": 0.8197,
       "step": 15359
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.592433673190297e-06,
-      "loss": 0.8276,
+      "loss": 0.8381,
       "step": 15360
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.591267547808293e-06,
-      "loss": 0.7655,
+      "loss": 0.8341,
       "step": 15361
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.590101441970456e-06,
-      "loss": 0.7637,
+      "loss": 0.9115,
       "step": 15362
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.588935355692957e-06,
-      "loss": 0.7188,
+      "loss": 0.826,
       "step": 15363
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.587769288991983e-06,
-      "loss": 0.703,
+      "loss": 0.886,
       "step": 15364
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.5866032418837e-06,
-      "loss": 0.7979,
+      "loss": 0.9217,
       "step": 15365
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.585437214384299e-06,
-      "loss": 0.6491,
+      "loss": 0.8321,
       "step": 15366
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.584271206509944e-06,
-      "loss": 0.8533,
+      "loss": 0.8895,
       "step": 15367
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.583105218276822e-06,
-      "loss": 0.7587,
+      "loss": 0.8594,
       "step": 15368
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.581939249701105e-06,
-      "loss": 0.706,
+      "loss": 0.758,
       "step": 15369
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.58077330079896e-06,
-      "loss": 0.7021,
+      "loss": 0.8107,
       "step": 15370
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.579607371586582e-06,
-      "loss": 0.8957,
+      "loss": 0.8659,
       "step": 15371
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.578441462080132e-06,
-      "loss": 0.8182,
+      "loss": 0.9254,
       "step": 15372
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.577275572295791e-06,
-      "loss": 0.8354,
+      "loss": 0.9397,
       "step": 15373
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.576109702249731e-06,
-      "loss": 0.8529,
+      "loss": 0.9277,
       "step": 15374
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.574943851958131e-06,
-      "loss": 0.8306,
+      "loss": 0.8881,
       "step": 15375
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.57377802143716e-06,
-      "loss": 0.7523,
+      "loss": 0.8896,
       "step": 15376
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.572612210703e-06,
-      "loss": 0.6937,
+      "loss": 0.7847,
       "step": 15377
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.571446419771815e-06,
-      "loss": 0.8045,
+      "loss": 0.8599,
       "step": 15378
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.57028064865979e-06,
-      "loss": 0.8463,
+      "loss": 0.8469,
       "step": 15379
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.569114897383087e-06,
-      "loss": 0.7459,
+      "loss": 0.8547,
       "step": 15380
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.567949165957888e-06,
-      "loss": 0.8036,
+      "loss": 0.9163,
       "step": 15381
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.56678345440036e-06,
-      "loss": 0.7088,
+      "loss": 0.7604,
       "step": 15382
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.565617762726682e-06,
-      "loss": 0.6081,
+      "loss": 0.785,
       "step": 15383
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.564452090953017e-06,
-      "loss": 0.8516,
+      "loss": 0.9326,
       "step": 15384
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.563286439095549e-06,
-      "loss": 0.7602,
+      "loss": 0.8695,
       "step": 15385
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.562120807170437e-06,
-      "loss": 0.8349,
+      "loss": 0.8402,
       "step": 15386
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.560955195193862e-06,
-      "loss": 0.7563,
+      "loss": 0.7808,
       "step": 15387
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.559789603181993e-06,
-      "loss": 0.6851,
+      "loss": 0.8792,
       "step": 15388
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.558624031150997e-06,
-      "loss": 0.8727,
+      "loss": 0.9556,
       "step": 15389
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.557458479117052e-06,
-      "loss": 0.8061,
+      "loss": 0.877,
       "step": 15390
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.556292947096318e-06,
-      "loss": 0.7084,
+      "loss": 0.8299,
       "step": 15391
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.555127435104975e-06,
-      "loss": 0.8518,
+      "loss": 0.9097,
       "step": 15392
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.553961943159186e-06,
-      "loss": 0.7219,
+      "loss": 0.8176,
       "step": 15393
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.552796471275125e-06,
-      "loss": 0.635,
+      "loss": 0.7973,
       "step": 15394
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.551631019468958e-06,
-      "loss": 0.7413,
+      "loss": 0.8151,
       "step": 15395
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.550465587756862e-06,
-      "loss": 0.8069,
+      "loss": 0.8685,
       "step": 15396
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.54930017615499e-06,
-      "loss": 0.7481,
+      "loss": 0.8228,
       "step": 15397
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.548134784679527e-06,
-      "loss": 0.8263,
+      "loss": 0.9076,
       "step": 15398
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.54696941334663e-06,
-      "loss": 0.6734,
+      "loss": 0.8568,
       "step": 15399
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.545804062172472e-06,
-      "loss": 0.6941,
+      "loss": 0.8479,
       "step": 15400
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.544638731173218e-06,
-      "loss": 0.6291,
+      "loss": 0.8162,
       "step": 15401
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.54347342036504e-06,
-      "loss": 0.8303,
+      "loss": 0.8735,
       "step": 15402
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.542308129764097e-06,
-      "loss": 0.7607,
+      "loss": 0.8635,
       "step": 15403
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.541142859386567e-06,
-      "loss": 0.9422,
+      "loss": 0.9991,
       "step": 15404
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.539977609248605e-06,
-      "loss": 0.8277,
+      "loss": 0.8986,
       "step": 15405
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.538812379366385e-06,
-      "loss": 0.7887,
+      "loss": 0.844,
       "step": 15406
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.53764716975607e-06,
-      "loss": 0.7428,
+      "loss": 0.8167,
       "step": 15407
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.536481980433821e-06,
-      "loss": 0.8176,
+      "loss": 0.866,
       "step": 15408
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.535316811415815e-06,
-      "loss": 0.7873,
+      "loss": 0.8775,
       "step": 15409
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.534151662718205e-06,
-      "loss": 0.6699,
+      "loss": 0.8379,
       "step": 15410
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.532986534357163e-06,
-      "loss": 0.6837,
+      "loss": 0.8354,
       "step": 15411
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.53182142634885e-06,
-      "loss": 0.7639,
+      "loss": 0.8749,
       "step": 15412
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.530656338709433e-06,
-      "loss": 0.6702,
+      "loss": 0.8153,
       "step": 15413
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.529491271455073e-06,
-      "loss": 0.6416,
+      "loss": 0.8052,
       "step": 15414
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.528326224601937e-06,
-      "loss": 0.7402,
+      "loss": 0.8192,
       "step": 15415
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.527161198166184e-06,
-      "loss": 0.7141,
+      "loss": 0.8866,
       "step": 15416
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.525996192163985e-06,
-      "loss": 0.7121,
+      "loss": 0.8154,
       "step": 15417
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.524831206611492e-06,
-      "loss": 0.7402,
+      "loss": 0.8214,
       "step": 15418
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.523666241524877e-06,
-      "loss": 0.7019,
+      "loss": 0.8962,
       "step": 15419
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.522501296920294e-06,
-      "loss": 0.7375,
+      "loss": 0.7941,
       "step": 15420
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.521336372813913e-06,
-      "loss": 0.7657,
+      "loss": 0.8713,
       "step": 15421
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.520171469221888e-06,
-      "loss": 0.656,
+      "loss": 0.8048,
       "step": 15422
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.51900658616039e-06,
-      "loss": 0.6707,
+      "loss": 0.8366,
       "step": 15423
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.517841723645569e-06,
-      "loss": 0.866,
+      "loss": 0.9316,
       "step": 15424
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.516676881693595e-06,
-      "loss": 0.7847,
+      "loss": 0.8518,
       "step": 15425
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.515512060320622e-06,
-      "loss": 0.6693,
+      "loss": 0.7679,
       "step": 15426
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.514347259542814e-06,
-      "loss": 0.6795,
+      "loss": 0.8031,
       "step": 15427
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.513182479376332e-06,
-      "loss": 0.699,
+      "loss": 0.8213,
       "step": 15428
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.512017719837329e-06,
-      "loss": 0.7694,
+      "loss": 0.8735,
       "step": 15429
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.510852980941974e-06,
-      "loss": 0.6621,
+      "loss": 0.8006,
       "step": 15430
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.509688262706416e-06,
-      "loss": 0.6323,
+      "loss": 0.7616,
       "step": 15431
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.508523565146823e-06,
-      "loss": 0.8203,
+      "loss": 0.8386,
       "step": 15432
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.507358888279346e-06,
-      "loss": 0.6743,
+      "loss": 0.7848,
       "step": 15433
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.506194232120148e-06,
-      "loss": 0.7798,
+      "loss": 0.9064,
       "step": 15434
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.505029596685385e-06,
-      "loss": 0.781,
+      "loss": 0.8269,
       "step": 15435
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.50386498199122e-06,
-      "loss": 0.8784,
+      "loss": 0.9403,
       "step": 15436
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.502700388053797e-06,
-      "loss": 0.811,
+      "loss": 0.8803,
       "step": 15437
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.501535814889291e-06,
-      "loss": 0.7014,
+      "loss": 0.7997,
       "step": 15438
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.500371262513843e-06,
-      "loss": 0.5583,
+      "loss": 0.7171,
       "step": 15439
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.499206730943619e-06,
-      "loss": 0.8433,
+      "loss": 0.9146,
       "step": 15440
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.49804222019477e-06,
-      "loss": 0.6762,
+      "loss": 0.8421,
       "step": 15441
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.496877730283456e-06,
-      "loss": 0.6269,
+      "loss": 0.7789,
       "step": 15442
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.495713261225828e-06,
-      "loss": 0.7937,
+      "loss": 0.8994,
       "step": 15443
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.49454881303805e-06,
-      "loss": 0.5829,
+      "loss": 0.6865,
       "step": 15444
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.49338438573627e-06,
-      "loss": 0.7944,
+      "loss": 0.8959,
       "step": 15445
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.49221997933664e-06,
-      "loss": 0.8448,
+      "loss": 0.9433,
       "step": 15446
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.491055593855322e-06,
-      "loss": 0.7243,
+      "loss": 0.8566,
       "step": 15447
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.489891229308462e-06,
-      "loss": 0.6727,
+      "loss": 0.7859,
       "step": 15448
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.488726885712226e-06,
-      "loss": 0.6914,
+      "loss": 0.7798,
       "step": 15449
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.487562563082754e-06,
-      "loss": 0.8605,
+      "loss": 0.9175,
       "step": 15450
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.486398261436208e-06,
-      "loss": 0.7122,
+      "loss": 0.8432,
       "step": 15451
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.485233980788738e-06,
-      "loss": 0.6719,
+      "loss": 0.7677,
       "step": 15452
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.484069721156498e-06,
-      "loss": 0.6218,
+      "loss": 0.7035,
       "step": 15453
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.482905482555637e-06,
-      "loss": 0.6687,
+      "loss": 0.7815,
       "step": 15454
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.481741265002315e-06,
-      "loss": 0.7181,
+      "loss": 0.8649,
       "step": 15455
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.480577068512673e-06,
-      "loss": 0.7498,
+      "loss": 0.8182,
       "step": 15456
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.479412893102874e-06,
-      "loss": 0.6475,
+      "loss": 0.7522,
       "step": 15457
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.478248738789061e-06,
-      "loss": 0.7025,
+      "loss": 0.831,
       "step": 15458
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.477084605587388e-06,
-      "loss": 0.7264,
+      "loss": 0.777,
       "step": 15459
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.475920493514003e-06,
-      "loss": 0.6246,
+      "loss": 0.8389,
       "step": 15460
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.474756402585063e-06,
-      "loss": 0.7436,
+      "loss": 0.8963,
       "step": 15461
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.47359233281671e-06,
-      "loss": 0.7863,
+      "loss": 0.9128,
       "step": 15462
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.472428284225102e-06,
-      "loss": 0.7988,
+      "loss": 0.9113,
       "step": 15463
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.471264256826384e-06,
-      "loss": 0.7289,
+      "loss": 0.8904,
       "step": 15464
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.4701002506367e-06,
-      "loss": 0.6293,
+      "loss": 0.8128,
       "step": 15465
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.468936265672209e-06,
-      "loss": 0.6304,
+      "loss": 0.8498,
       "step": 15466
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.467772301949054e-06,
-      "loss": 0.7466,
+      "loss": 0.8659,
       "step": 15467
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.466608359483386e-06,
-      "loss": 0.6715,
+      "loss": 0.8159,
       "step": 15468
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.465444438291346e-06,
-      "loss": 0.7814,
+      "loss": 0.8962,
       "step": 15469
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.464280538389093e-06,
-      "loss": 0.8321,
+      "loss": 0.9179,
       "step": 15470
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.463116659792764e-06,
-      "loss": 0.77,
+      "loss": 0.8661,
       "step": 15471
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.461952802518515e-06,
-      "loss": 0.7806,
+      "loss": 0.833,
       "step": 15472
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.460788966582485e-06,
-      "loss": 0.6919,
+      "loss": 0.8444,
       "step": 15473
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.459625152000826e-06,
-      "loss": 0.5881,
+      "loss": 0.784,
       "step": 15474
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.45846135878968e-06,
-      "loss": 0.7863,
+      "loss": 0.846,
       "step": 15475
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.457297586965202e-06,
-      "loss": 0.6925,
+      "loss": 0.8382,
       "step": 15476
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.456133836543527e-06,
-      "loss": 0.7422,
+      "loss": 0.8641,
       "step": 15477
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.454970107540805e-06,
-      "loss": 0.6247,
+      "loss": 0.7968,
       "step": 15478
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.453806399973178e-06,
-      "loss": 0.6941,
+      "loss": 0.8195,
       "step": 15479
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.452642713856797e-06,
-      "loss": 0.6837,
+      "loss": 0.8056,
       "step": 15480
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.451479049207801e-06,
-      "loss": 0.7602,
+      "loss": 0.8377,
       "step": 15481
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.450315406042339e-06,
-      "loss": 0.7278,
+      "loss": 0.8287,
       "step": 15482
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.449151784376546e-06,
-      "loss": 0.8967,
+      "loss": 0.9445,
       "step": 15483
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.44798818422658e-06,
-      "loss": 0.7872,
+      "loss": 0.8443,
       "step": 15484
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.44682460560857e-06,
-      "loss": 0.7055,
+      "loss": 0.804,
       "step": 15485
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.445661048538665e-06,
-      "loss": 0.7509,
+      "loss": 0.8672,
       "step": 15486
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.44449751303301e-06,
-      "loss": 0.7821,
+      "loss": 0.8582,
       "step": 15487
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.443333999107743e-06,
-      "loss": 0.8705,
+      "loss": 0.8923,
       "step": 15488
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.442170506779011e-06,
-      "loss": 0.7156,
+      "loss": 0.8341,
       "step": 15489
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.44100703606295e-06,
-      "loss": 0.7079,
+      "loss": 0.8423,
       "step": 15490
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.439843586975707e-06,
-      "loss": 0.7072,
+      "loss": 0.7751,
       "step": 15491
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.438680159533417e-06,
-      "loss": 0.6781,
+      "loss": 0.8289,
       "step": 15492
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.437516753752228e-06,
-      "loss": 0.6766,
+      "loss": 0.8544,
       "step": 15493
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.436353369648274e-06,
-      "loss": 0.7304,
+      "loss": 0.7493,
       "step": 15494
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.435190007237703e-06,
-      "loss": 0.7723,
+      "loss": 0.8539,
       "step": 15495
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.434026666536644e-06,
-      "loss": 0.6789,
+      "loss": 0.7967,
       "step": 15496
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.432863347561251e-06,
-      "loss": 0.6877,
+      "loss": 0.8185,
       "step": 15497
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.43170005032765e-06,
-      "loss": 0.6286,
+      "loss": 0.8099,
       "step": 15498
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.430536774851988e-06,
-      "loss": 0.6428,
+      "loss": 0.8005,
       "step": 15499
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.4293735211504e-06,
-      "loss": 0.7567,
+      "loss": 0.8846,
       "step": 15500
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.428210289239026e-06,
-      "loss": 0.8002,
+      "loss": 0.853,
       "step": 15501
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.427047079134003e-06,
-      "loss": 0.7642,
+      "loss": 0.8517,
       "step": 15502
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.425883890851476e-06,
-      "loss": 0.7647,
+      "loss": 0.7983,
       "step": 15503
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.424720724407573e-06,
-      "loss": 0.7563,
+      "loss": 0.8752,
       "step": 15504
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.423557579818433e-06,
-      "loss": 0.9055,
+      "loss": 0.8872,
       "step": 15505
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.422394457100198e-06,
-      "loss": 0.6847,
+      "loss": 0.8279,
       "step": 15506
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.421231356268998e-06,
-      "loss": 0.6808,
+      "loss": 0.7659,
       "step": 15507
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.420068277340977e-06,
-      "loss": 0.6571,
+      "loss": 0.7685,
       "step": 15508
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.418905220332262e-06,
-      "loss": 0.7239,
+      "loss": 0.8595,
       "step": 15509
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.417742185259e-06,
-      "loss": 0.777,
+      "loss": 0.9414,
       "step": 15510
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.416579172137314e-06,
-      "loss": 0.8224,
+      "loss": 0.8718,
       "step": 15511
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.41541618098335e-06,
-      "loss": 0.6338,
+      "loss": 0.836,
       "step": 15512
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.414253211813234e-06,
-      "loss": 0.8083,
+      "loss": 0.8546,
       "step": 15513
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.413090264643108e-06,
-      "loss": 0.6443,
+      "loss": 0.7787,
       "step": 15514
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.411927339489101e-06,
-      "loss": 0.6666,
+      "loss": 0.7903,
       "step": 15515
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.410764436367355e-06,
-      "loss": 0.8119,
+      "loss": 0.8785,
       "step": 15516
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.409601555293991e-06,
-      "loss": 0.6607,
+      "loss": 0.8302,
       "step": 15517
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.408438696285153e-06,
-      "loss": 0.6867,
+      "loss": 0.8202,
       "step": 15518
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.407275859356967e-06,
-      "loss": 0.8064,
+      "loss": 0.8275,
       "step": 15519
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.406113044525569e-06,
-      "loss": 0.6234,
+      "loss": 0.7888,
       "step": 15520
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.40495025180709e-06,
-      "loss": 0.632,
+      "loss": 0.793,
       "step": 15521
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.403787481217668e-06,
-      "loss": 0.7288,
+      "loss": 0.8529,
       "step": 15522
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.402624732773428e-06,
-      "loss": 0.6675,
+      "loss": 0.765,
       "step": 15523
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.4014620064905e-06,
-      "loss": 0.6621,
+      "loss": 0.733,
       "step": 15524
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.400299302385021e-06,
-      "loss": 0.6538,
+      "loss": 0.787,
       "step": 15525
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.399136620473119e-06,
-      "loss": 0.695,
+      "loss": 0.8207,
       "step": 15526
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.397973960770928e-06,
-      "loss": 0.7005,
+      "loss": 0.7588,
       "step": 15527
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.39681132329457e-06,
-      "loss": 0.6977,
+      "loss": 0.8274,
       "step": 15528
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.395648708060187e-06,
-      "loss": 0.6611,
+      "loss": 0.8458,
       "step": 15529
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.394486115083896e-06,
-      "loss": 0.6917,
+      "loss": 0.8069,
       "step": 15530
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.393323544381835e-06,
-      "loss": 0.6538,
+      "loss": 0.8212,
       "step": 15531
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.392160995970128e-06,
-      "loss": 0.7131,
+      "loss": 0.8156,
       "step": 15532
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.390998469864908e-06,
-      "loss": 0.7815,
+      "loss": 0.8867,
       "step": 15533
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.3898359660823e-06,
-      "loss": 0.7684,
+      "loss": 0.9328,
       "step": 15534
     },
     {
       "epoch": 0.56,
       "learning_rate": 8.388673484638436e-06,
-      "loss": 0.6921,
+      "loss": 0.7867,
       "step": 15535
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.387511025549437e-06,
-      "loss": 0.7725,
+      "loss": 0.8123,
       "step": 15536
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.38634858883144e-06,
-      "loss": 0.7679,
+      "loss": 0.8454,
       "step": 15537
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.385186174500562e-06,
-      "loss": 0.6313,
+      "loss": 0.759,
       "step": 15538
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.384023782572938e-06,
-      "loss": 0.7244,
+      "loss": 0.8463,
       "step": 15539
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.382861413064687e-06,
-      "loss": 0.7971,
+      "loss": 0.9273,
       "step": 15540
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.381699065991944e-06,
-      "loss": 0.8129,
+      "loss": 0.9164,
       "step": 15541
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.380536741370831e-06,
-      "loss": 0.7659,
+      "loss": 0.7827,
       "step": 15542
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.37937443921747e-06,
-      "loss": 0.6906,
+      "loss": 0.8246,
       "step": 15543
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.378212159547992e-06,
-      "loss": 0.7482,
+      "loss": 0.8277,
       "step": 15544
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.377049902378514e-06,
-      "loss": 0.7232,
+      "loss": 0.8469,
       "step": 15545
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.375887667725171e-06,
-      "loss": 0.6504,
+      "loss": 0.8117,
       "step": 15546
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.374725455604079e-06,
-      "loss": 0.4687,
+      "loss": 0.6818,
       "step": 15547
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.37356326603137e-06,
-      "loss": 0.6908,
+      "loss": 0.7709,
       "step": 15548
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.372401099023158e-06,
-      "loss": 0.7139,
+      "loss": 0.8743,
       "step": 15549
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.371238954595575e-06,
-      "loss": 0.6509,
+      "loss": 0.7535,
       "step": 15550
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.370076832764737e-06,
-      "loss": 0.6464,
+      "loss": 0.8299,
       "step": 15551
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.368914733546774e-06,
-      "loss": 0.6618,
+      "loss": 0.7758,
       "step": 15552
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.367752656957802e-06,
-      "loss": 0.8028,
+      "loss": 0.8869,
       "step": 15553
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.36659060301395e-06,
-      "loss": 0.7366,
+      "loss": 0.8059,
       "step": 15554
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.36542857173133e-06,
-      "loss": 0.6814,
+      "loss": 0.8206,
       "step": 15555
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.364266563126076e-06,
-      "loss": 0.7296,
+      "loss": 0.8343,
       "step": 15556
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.363104577214297e-06,
-      "loss": 0.6984,
+      "loss": 0.833,
       "step": 15557
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.361942614012124e-06,
-      "loss": 0.7056,
+      "loss": 0.8108,
       "step": 15558
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.36078067353567e-06,
-      "loss": 0.6796,
+      "loss": 0.8184,
       "step": 15559
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.35961875580106e-06,
-      "loss": 0.7399,
+      "loss": 0.8193,
       "step": 15560
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.358456860824413e-06,
-      "loss": 0.6194,
+      "loss": 0.7446,
       "step": 15561
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.357294988621845e-06,
-      "loss": 0.5809,
+      "loss": 0.7226,
       "step": 15562
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.356133139209482e-06,
-      "loss": 0.7445,
+      "loss": 0.8426,
       "step": 15563
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.354971312603436e-06,
-      "loss": 0.7487,
+      "loss": 0.88,
       "step": 15564
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.353809508819831e-06,
-      "loss": 0.7609,
+      "loss": 0.8428,
       "step": 15565
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.35264772787478e-06,
-      "loss": 0.745,
+      "loss": 0.8395,
       "step": 15566
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.351485969784412e-06,
-      "loss": 0.6826,
+      "loss": 0.7743,
       "step": 15567
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.350324234564829e-06,
-      "loss": 0.6602,
+      "loss": 0.8256,
       "step": 15568
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.349162522232164e-06,
-      "loss": 0.8107,
+      "loss": 0.8777,
       "step": 15569
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.348000832802521e-06,
-      "loss": 0.7723,
+      "loss": 0.8828,
       "step": 15570
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.346839166292025e-06,
-      "loss": 0.6716,
+      "loss": 0.7869,
       "step": 15571
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.345677522716789e-06,
-      "loss": 0.777,
+      "loss": 0.9069,
       "step": 15572
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.344515902092932e-06,
-      "loss": 0.7596,
+      "loss": 0.8855,
       "step": 15573
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.343354304436563e-06,
-      "loss": 0.5763,
+      "loss": 0.7755,
       "step": 15574
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.34219272976381e-06,
-      "loss": 0.7788,
+      "loss": 0.834,
       "step": 15575
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.341031178090778e-06,
-      "loss": 0.6724,
+      "loss": 0.8456,
       "step": 15576
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.339869649433582e-06,
-      "loss": 0.7848,
+      "loss": 0.8873,
       "step": 15577
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.338708143808341e-06,
-      "loss": 0.7321,
+      "loss": 0.8751,
       "step": 15578
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.337546661231168e-06,
-      "loss": 0.8017,
+      "loss": 0.8574,
       "step": 15579
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.336385201718178e-06,
-      "loss": 0.7502,
+      "loss": 0.8229,
       "step": 15580
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.335223765285478e-06,
-      "loss": 0.6918,
+      "loss": 0.8484,
       "step": 15581
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.334062351949193e-06,
-      "loss": 0.7811,
+      "loss": 0.8997,
       "step": 15582
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.332900961725423e-06,
-      "loss": 0.6704,
+      "loss": 0.849,
       "step": 15583
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.33173959463029e-06,
-      "loss": 0.7176,
+      "loss": 0.8937,
       "step": 15584
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.330578250679901e-06,
-      "loss": 0.6356,
+      "loss": 0.7399,
       "step": 15585
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.329416929890373e-06,
-      "loss": 0.6655,
+      "loss": 0.7815,
       "step": 15586
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.328255632277811e-06,
-      "loss": 0.805,
+      "loss": 0.8933,
       "step": 15587
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.327094357858336e-06,
-      "loss": 0.8493,
+      "loss": 0.8931,
       "step": 15588
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.325933106648048e-06,
-      "loss": 0.7115,
+      "loss": 0.8332,
       "step": 15589
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.324771878663064e-06,
-      "loss": 0.6626,
+      "loss": 0.7139,
       "step": 15590
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.323610673919494e-06,
-      "loss": 0.6712,
+      "loss": 0.8741,
       "step": 15591
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.322449492433448e-06,
-      "loss": 0.7713,
+      "loss": 0.8387,
       "step": 15592
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.321288334221034e-06,
-      "loss": 0.8248,
+      "loss": 0.8281,
       "step": 15593
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.320127199298366e-06,
-      "loss": 0.6789,
+      "loss": 0.7869,
       "step": 15594
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.318966087681544e-06,
-      "loss": 0.7157,
+      "loss": 0.8191,
       "step": 15595
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.317804999386688e-06,
-      "loss": 0.6921,
+      "loss": 0.8465,
       "step": 15596
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.316643934429897e-06,
-      "loss": 0.591,
+      "loss": 0.7447,
       "step": 15597
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.315482892827285e-06,
-      "loss": 0.742,
+      "loss": 0.8369,
       "step": 15598
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.314321874594959e-06,
-      "loss": 0.837,
+      "loss": 0.9417,
       "step": 15599
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.31316087974902e-06,
-      "loss": 0.6907,
+      "loss": 0.7882,
       "step": 15600
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.311999908305588e-06,
-      "loss": 0.7653,
+      "loss": 0.8206,
       "step": 15601
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.310838960280757e-06,
-      "loss": 0.8259,
+      "loss": 0.8983,
       "step": 15602
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.30967803569064e-06,
-      "loss": 0.7552,
+      "loss": 0.8415,
       "step": 15603
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.308517134551341e-06,
-      "loss": 0.7772,
+      "loss": 0.8161,
       "step": 15604
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.30735625687897e-06,
-      "loss": 0.6921,
+      "loss": 0.8345,
       "step": 15605
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.306195402689624e-06,
-      "loss": 0.8199,
+      "loss": 0.8607,
       "step": 15606
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.30503457199942e-06,
-      "loss": 0.8228,
+      "loss": 0.8553,
       "step": 15607
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.30387376482445e-06,
-      "loss": 0.7327,
+      "loss": 0.8168,
       "step": 15608
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.302712981180832e-06,
-      "loss": 0.7686,
+      "loss": 0.8595,
       "step": 15609
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.301552221084658e-06,
-      "loss": 0.6369,
+      "loss": 0.8264,
       "step": 15610
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.30039148455204e-06,
-      "loss": 0.7281,
+      "loss": 0.8833,
       "step": 15611
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.299230771599077e-06,
-      "loss": 0.8048,
+      "loss": 0.8237,
       "step": 15612
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.298070082241874e-06,
-      "loss": 0.721,
+      "loss": 0.8077,
       "step": 15613
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.296909416496532e-06,
-      "loss": 0.647,
+      "loss": 0.7917,
       "step": 15614
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.295748774379163e-06,
-      "loss": 0.7713,
+      "loss": 0.812,
       "step": 15615
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.294588155905854e-06,
-      "loss": 0.8141,
+      "loss": 0.8187,
       "step": 15616
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.293427561092718e-06,
-      "loss": 0.6584,
+      "loss": 0.8546,
       "step": 15617
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.292266989955853e-06,
-      "loss": 0.7371,
+      "loss": 0.8699,
       "step": 15618
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.29110644251136e-06,
-      "loss": 0.6725,
+      "loss": 0.8701,
       "step": 15619
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.289945918775343e-06,
-      "loss": 0.6493,
+      "loss": 0.8373,
       "step": 15620
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.288785418763895e-06,
-      "loss": 0.812,
+      "loss": 0.8717,
       "step": 15621
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.287624942493126e-06,
-      "loss": 0.734,
+      "loss": 0.8241,
       "step": 15622
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.286464489979128e-06,
-      "loss": 0.8038,
+      "loss": 0.8815,
       "step": 15623
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.285304061238006e-06,
-      "loss": 0.7034,
+      "loss": 0.7493,
       "step": 15624
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.284143656285858e-06,
-      "loss": 0.6118,
+      "loss": 0.763,
       "step": 15625
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.28298327513878e-06,
-      "loss": 0.713,
+      "loss": 0.8329,
       "step": 15626
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.281822917812873e-06,
-      "loss": 0.7817,
+      "loss": 0.8727,
       "step": 15627
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.28066258432424e-06,
-      "loss": 0.6877,
+      "loss": 0.7982,
       "step": 15628
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.279502274688968e-06,
-      "loss": 0.6596,
+      "loss": 0.7887,
       "step": 15629
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.278341988923164e-06,
-      "loss": 0.7053,
+      "loss": 0.8293,
       "step": 15630
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.277181727042921e-06,
-      "loss": 0.7523,
+      "loss": 0.816,
       "step": 15631
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.27602148906434e-06,
-      "loss": 0.8454,
+      "loss": 0.9863,
       "step": 15632
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.27486127500351e-06,
-      "loss": 0.7225,
+      "loss": 0.8415,
       "step": 15633
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.273701084876538e-06,
-      "loss": 0.745,
+      "loss": 0.9003,
       "step": 15634
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.272540918699507e-06,
-      "loss": 0.7443,
+      "loss": 0.7804,
       "step": 15635
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.271380776488527e-06,
-      "loss": 0.6076,
+      "loss": 0.7857,
       "step": 15636
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.270220658259684e-06,
-      "loss": 0.677,
+      "loss": 0.873,
       "step": 15637
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.269060564029073e-06,
-      "loss": 0.6737,
+      "loss": 0.867,
       "step": 15638
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.267900493812794e-06,
-      "loss": 0.6572,
+      "loss": 0.8137,
       "step": 15639
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.266740447626932e-06,
-      "loss": 0.7564,
+      "loss": 0.878,
       "step": 15640
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.265580425487593e-06,
-      "loss": 0.7835,
+      "loss": 0.8107,
       "step": 15641
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.264420427410861e-06,
-      "loss": 0.7956,
+      "loss": 0.8727,
       "step": 15642
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.263260453412837e-06,
-      "loss": 0.7567,
+      "loss": 0.83,
       "step": 15643
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.262100503509606e-06,
-      "loss": 0.6095,
+      "loss": 0.7872,
       "step": 15644
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.260940577717268e-06,
-      "loss": 0.7339,
+      "loss": 0.8049,
       "step": 15645
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.25978067605191e-06,
-      "loss": 0.7546,
+      "loss": 0.7933,
       "step": 15646
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.258620798529629e-06,
-      "loss": 0.7058,
+      "loss": 0.8352,
       "step": 15647
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.25746094516651e-06,
-      "loss": 0.7216,
+      "loss": 0.8034,
       "step": 15648
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.256301115978652e-06,
-      "loss": 0.7999,
+      "loss": 0.8393,
       "step": 15649
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.255141310982138e-06,
-      "loss": 0.6318,
+      "loss": 0.8495,
       "step": 15650
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.253981530193064e-06,
-      "loss": 0.643,
+      "loss": 0.7535,
       "step": 15651
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.25282177362752e-06,
-      "loss": 0.713,
+      "loss": 0.8689,
       "step": 15652
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.251662041301595e-06,
-      "loss": 0.7824,
+      "loss": 0.9288,
       "step": 15653
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.250502333231376e-06,
-      "loss": 0.8278,
+      "loss": 0.8068,
       "step": 15654
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.24934264943296e-06,
-      "loss": 0.9778,
+      "loss": 0.9654,
       "step": 15655
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.248182989922429e-06,
-      "loss": 0.7268,
+      "loss": 0.8466,
       "step": 15656
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.24702335471587e-06,
-      "loss": 0.7132,
+      "loss": 0.7689,
       "step": 15657
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.24586374382938e-06,
-      "loss": 0.7702,
+      "loss": 0.8694,
       "step": 15658
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.244704157279036e-06,
-      "loss": 0.8373,
+      "loss": 0.8622,
       "step": 15659
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.243544595080937e-06,
-      "loss": 0.5494,
+      "loss": 0.7344,
       "step": 15660
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.242385057251162e-06,
-      "loss": 0.7815,
+      "loss": 0.8337,
       "step": 15661
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.2412255438058e-06,
-      "loss": 0.6803,
+      "loss": 0.7514,
       "step": 15662
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.240066054760937e-06,
-      "loss": 0.7227,
+      "loss": 0.856,
       "step": 15663
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.238906590132662e-06,
-      "loss": 0.6095,
+      "loss": 0.8473,
       "step": 15664
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.237747149937059e-06,
-      "loss": 0.7517,
+      "loss": 0.8569,
       "step": 15665
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.236587734190218e-06,
-      "loss": 0.9652,
+      "loss": 0.9518,
       "step": 15666
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.235428342908213e-06,
-      "loss": 0.7738,
+      "loss": 0.8315,
       "step": 15667
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.234268976107142e-06,
-      "loss": 0.7652,
+      "loss": 0.7959,
       "step": 15668
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.233109633803081e-06,
-      "loss": 0.7452,
+      "loss": 0.9099,
       "step": 15669
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.231950316012118e-06,
-      "loss": 0.664,
+      "loss": 0.7605,
       "step": 15670
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.230791022750336e-06,
-      "loss": 0.5615,
+      "loss": 0.819,
       "step": 15671
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.229631754033818e-06,
-      "loss": 0.6195,
+      "loss": 0.7637,
       "step": 15672
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.228472509878646e-06,
-      "loss": 0.7979,
+      "loss": 0.8591,
       "step": 15673
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.227313290300907e-06,
-      "loss": 0.6423,
+      "loss": 0.7865,
       "step": 15674
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.226154095316682e-06,
-      "loss": 0.6498,
+      "loss": 0.7958,
       "step": 15675
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.224994924942049e-06,
-      "loss": 0.762,
+      "loss": 0.7989,
       "step": 15676
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.223835779193094e-06,
-      "loss": 0.6322,
+      "loss": 0.8257,
       "step": 15677
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.222676658085897e-06,
-      "loss": 0.8532,
+      "loss": 0.8768,
       "step": 15678
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.221517561636543e-06,
-      "loss": 0.599,
+      "loss": 0.8081,
       "step": 15679
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.220358489861102e-06,
-      "loss": 0.7055,
+      "loss": 0.8136,
       "step": 15680
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.21919944277567e-06,
-      "loss": 0.7818,
+      "loss": 0.8224,
       "step": 15681
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.218040420396315e-06,
-      "loss": 0.8124,
+      "loss": 0.8972,
       "step": 15682
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.216881422739123e-06,
-      "loss": 0.7608,
+      "loss": 0.8381,
       "step": 15683
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.215722449820167e-06,
-      "loss": 0.7974,
+      "loss": 0.8613,
       "step": 15684
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.214563501655535e-06,
-      "loss": 0.7105,
+      "loss": 0.8476,
       "step": 15685
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.213404578261298e-06,
-      "loss": 0.6582,
+      "loss": 0.8231,
       "step": 15686
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.212245679653544e-06,
-      "loss": 0.7455,
+      "loss": 0.8755,
       "step": 15687
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.211086805848337e-06,
-      "loss": 0.6876,
+      "loss": 0.8021,
       "step": 15688
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.209927956861766e-06,
-      "loss": 0.7069,
+      "loss": 0.7783,
       "step": 15689
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.208769132709904e-06,
-      "loss": 0.7289,
+      "loss": 0.882,
       "step": 15690
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.207610333408829e-06,
-      "loss": 0.7006,
+      "loss": 0.8252,
       "step": 15691
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.206451558974616e-06,
-      "loss": 0.5854,
+      "loss": 0.7655,
       "step": 15692
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.205292809423348e-06,
-      "loss": 0.8071,
+      "loss": 0.8499,
       "step": 15693
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.204134084771094e-06,
-      "loss": 0.7318,
+      "loss": 0.8666,
       "step": 15694
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.202975385033928e-06,
-      "loss": 0.8176,
+      "loss": 0.8799,
       "step": 15695
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.201816710227933e-06,
-      "loss": 0.7338,
+      "loss": 0.8699,
       "step": 15696
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.200658060369176e-06,
-      "loss": 0.7573,
+      "loss": 0.8562,
       "step": 15697
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.199499435473739e-06,
-      "loss": 0.7114,
+      "loss": 0.9368,
       "step": 15698
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.198340835557687e-06,
-      "loss": 0.6911,
+      "loss": 0.8651,
       "step": 15699
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.197182260637107e-06,
-      "loss": 0.7188,
+      "loss": 0.8208,
       "step": 15700
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.196023710728062e-06,
-      "loss": 0.7417,
+      "loss": 0.7695,
       "step": 15701
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.19486518584663e-06,
-      "loss": 0.6757,
+      "loss": 0.7311,
       "step": 15702
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.193706686008878e-06,
-      "loss": 0.7369,
+      "loss": 0.858,
       "step": 15703
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.192548211230886e-06,
-      "loss": 0.7494,
+      "loss": 0.8198,
       "step": 15704
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.191389761528723e-06,
-      "loss": 0.8292,
+      "loss": 0.9861,
       "step": 15705
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.190231336918464e-06,
-      "loss": 0.7327,
+      "loss": 0.8698,
       "step": 15706
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.18907293741617e-06,
-      "loss": 0.7577,
+      "loss": 0.8254,
       "step": 15707
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.187914563037928e-06,
-      "loss": 0.6367,
+      "loss": 0.8067,
       "step": 15708
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.186756213799795e-06,
-      "loss": 0.7558,
+      "loss": 0.8906,
       "step": 15709
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.18559788971785e-06,
-      "loss": 0.7875,
+      "loss": 0.8744,
       "step": 15710
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.184439590808155e-06,
-      "loss": 0.6906,
+      "loss": 0.7916,
       "step": 15711
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.18328131708679e-06,
-      "loss": 0.8701,
+      "loss": 0.8575,
       "step": 15712
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.182123068569815e-06,
-      "loss": 0.7186,
+      "loss": 0.8394,
       "step": 15713
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.180964845273309e-06,
-      "loss": 0.6678,
+      "loss": 0.8168,
       "step": 15714
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.179806647213332e-06,
-      "loss": 0.8061,
+      "loss": 0.9212,
       "step": 15715
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.178648474405952e-06,
-      "loss": 0.6049,
+      "loss": 0.7259,
       "step": 15716
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.177490326867245e-06,
-      "loss": 0.6515,
+      "loss": 0.7983,
       "step": 15717
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.176332204613269e-06,
-      "loss": 0.6717,
+      "loss": 0.8184,
       "step": 15718
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.175174107660103e-06,
-      "loss": 0.5564,
+      "loss": 0.6881,
       "step": 15719
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.174016036023802e-06,
-      "loss": 0.7747,
+      "loss": 0.8663,
       "step": 15720
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.17285798972044e-06,
-      "loss": 0.7395,
+      "loss": 0.8492,
       "step": 15721
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.171699968766079e-06,
-      "loss": 0.9248,
+      "loss": 0.9517,
       "step": 15722
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.170541973176788e-06,
-      "loss": 0.6239,
+      "loss": 0.7744,
       "step": 15723
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.169384002968628e-06,
-      "loss": 0.7785,
+      "loss": 0.8592,
       "step": 15724
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.168226058157674e-06,
-      "loss": 0.7528,
+      "loss": 0.8101,
       "step": 15725
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.16706813875998e-06,
-      "loss": 0.7027,
+      "loss": 0.7928,
       "step": 15726
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.165910244791619e-06,
-      "loss": 0.5766,
+      "loss": 0.7638,
       "step": 15727
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.164752376268646e-06,
-      "loss": 0.895,
+      "loss": 0.9373,
       "step": 15728
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.163594533207133e-06,
-      "loss": 0.6557,
+      "loss": 0.7636,
       "step": 15729
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.16243671562314e-06,
-      "loss": 0.6393,
+      "loss": 0.7522,
       "step": 15730
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.161278923532729e-06,
-      "loss": 0.8043,
+      "loss": 0.8528,
       "step": 15731
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.160121156951965e-06,
-      "loss": 0.7329,
+      "loss": 0.8675,
       "step": 15732
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.158963415896912e-06,
-      "loss": 0.6494,
+      "loss": 0.8041,
       "step": 15733
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.157805700383628e-06,
-      "loss": 0.7556,
+      "loss": 0.8399,
       "step": 15734
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.156648010428175e-06,
-      "loss": 0.7573,
+      "loss": 0.7976,
       "step": 15735
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.155490346046618e-06,
-      "loss": 0.7001,
+      "loss": 0.8528,
       "step": 15736
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.154332707255011e-06,
-      "loss": 0.84,
+      "loss": 0.9492,
       "step": 15737
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.153175094069424e-06,
-      "loss": 0.7205,
+      "loss": 0.7901,
       "step": 15738
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.152017506505906e-06,
-      "loss": 0.7231,
+      "loss": 0.8286,
       "step": 15739
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.15085994458053e-06,
-      "loss": 0.6988,
+      "loss": 0.7926,
       "step": 15740
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.149702408309345e-06,
-      "loss": 0.8972,
+      "loss": 0.9796,
       "step": 15741
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.148544897708415e-06,
-      "loss": 0.672,
+      "loss": 0.7935,
       "step": 15742
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.147387412793794e-06,
-      "loss": 0.6253,
+      "loss": 0.7273,
       "step": 15743
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.14622995358155e-06,
-      "loss": 0.67,
+      "loss": 0.8259,
       "step": 15744
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.14507252008773e-06,
-      "loss": 0.4654,
+      "loss": 0.7429,
       "step": 15745
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.143915112328403e-06,
-      "loss": 0.7099,
+      "loss": 0.8248,
       "step": 15746
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.142757730319612e-06,
-      "loss": 0.7106,
+      "loss": 0.8868,
       "step": 15747
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.14160037407743e-06,
-      "loss": 0.6604,
+      "loss": 0.8386,
       "step": 15748
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.140443043617902e-06,
-      "loss": 0.7582,
+      "loss": 0.8564,
       "step": 15749
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.13928573895709e-06,
-      "loss": 0.763,
+      "loss": 0.8086,
       "step": 15750
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.138128460111047e-06,
-      "loss": 0.585,
+      "loss": 0.7534,
       "step": 15751
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.136971207095833e-06,
-      "loss": 0.7134,
+      "loss": 0.8546,
       "step": 15752
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.135813979927502e-06,
-      "loss": 0.7221,
+      "loss": 0.8147,
       "step": 15753
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.1346567786221e-06,
-      "loss": 0.6804,
+      "loss": 0.781,
       "step": 15754
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.133499603195694e-06,
-      "loss": 0.7684,
+      "loss": 0.7988,
       "step": 15755
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.13234245366433e-06,
-      "loss": 0.7496,
+      "loss": 0.8385,
       "step": 15756
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.131185330044065e-06,
-      "loss": 0.6365,
+      "loss": 0.7905,
       "step": 15757
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.130028232350951e-06,
-      "loss": 0.6879,
+      "loss": 0.7751,
       "step": 15758
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.128871160601047e-06,
-      "loss": 0.6383,
+      "loss": 0.7894,
       "step": 15759
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.127714114810395e-06,
-      "loss": 0.7766,
+      "loss": 0.811,
       "step": 15760
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.126557094995059e-06,
-      "loss": 0.6982,
+      "loss": 0.7654,
       "step": 15761
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.125400101171079e-06,
-      "loss": 0.7602,
+      "loss": 0.8338,
       "step": 15762
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.124243133354516e-06,
-      "loss": 0.751,
+      "loss": 0.9018,
       "step": 15763
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.123086191561417e-06,
-      "loss": 0.7864,
+      "loss": 0.8788,
       "step": 15764
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.121929275807838e-06,
-      "loss": 0.6837,
+      "loss": 0.8082,
       "step": 15765
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.120772386109818e-06,
-      "loss": 0.6863,
+      "loss": 0.8198,
       "step": 15766
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.119615522483424e-06,
-      "loss": 0.7738,
+      "loss": 0.8546,
       "step": 15767
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.11845868494469e-06,
-      "loss": 0.6199,
+      "loss": 0.8067,
       "step": 15768
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.117301873509675e-06,
-      "loss": 0.8456,
+      "loss": 0.8694,
       "step": 15769
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.116145088194424e-06,
-      "loss": 0.7185,
+      "loss": 0.8759,
       "step": 15770
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.114988329014988e-06,
-      "loss": 0.7629,
+      "loss": 0.8801,
       "step": 15771
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.113831595987418e-06,
-      "loss": 0.7849,
+      "loss": 0.8489,
       "step": 15772
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.112674889127752e-06,
-      "loss": 0.5278,
+      "loss": 0.6914,
       "step": 15773
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.111518208452048e-06,
-      "loss": 0.7007,
+      "loss": 0.877,
       "step": 15774
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.110361553976347e-06,
-      "loss": 0.7626,
+      "loss": 0.8153,
       "step": 15775
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.1092049257167e-06,
-      "loss": 0.7508,
+      "loss": 0.8461,
       "step": 15776
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.108048323689152e-06,
-      "loss": 0.8699,
+      "loss": 0.92,
       "step": 15777
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.10689174790975e-06,
-      "loss": 0.7954,
+      "loss": 0.9225,
       "step": 15778
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.105735198394535e-06,
-      "loss": 0.7161,
+      "loss": 0.8466,
       "step": 15779
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.104578675159564e-06,
-      "loss": 0.751,
+      "loss": 0.8076,
       "step": 15780
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.103422178220868e-06,
-      "loss": 0.704,
+      "loss": 0.9024,
       "step": 15781
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.102265707594502e-06,
-      "loss": 0.7574,
+      "loss": 0.833,
       "step": 15782
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.101109263296503e-06,
-      "loss": 0.6945,
+      "loss": 0.7531,
       "step": 15783
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.099952845342922e-06,
-      "loss": 0.7507,
+      "loss": 0.8635,
       "step": 15784
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.098796453749797e-06,
-      "loss": 0.6416,
+      "loss": 0.7962,
       "step": 15785
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.09764008853318e-06,
-      "loss": 0.6999,
+      "loss": 0.8659,
       "step": 15786
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.0964837497091e-06,
-      "loss": 0.7591,
+      "loss": 0.9001,
       "step": 15787
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.09532743729361e-06,
-      "loss": 0.7684,
+      "loss": 0.8319,
       "step": 15788
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.09417115130275e-06,
-      "loss": 0.6352,
+      "loss": 0.8581,
       "step": 15789
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.09301489175256e-06,
-      "loss": 0.6689,
+      "loss": 0.8918,
       "step": 15790
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.091858658659086e-06,
-      "loss": 0.7243,
+      "loss": 0.8534,
       "step": 15791
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.090702452038362e-06,
-      "loss": 0.7288,
+      "loss": 0.8113,
       "step": 15792
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.089546271906436e-06,
-      "loss": 0.6342,
+      "loss": 0.7839,
       "step": 15793
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.088390118279342e-06,
-      "loss": 0.554,
+      "loss": 0.6983,
       "step": 15794
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.087233991173123e-06,
-      "loss": 0.6184,
+      "loss": 0.7483,
       "step": 15795
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.086077890603818e-06,
-      "loss": 0.7581,
+      "loss": 0.8324,
       "step": 15796
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.084921816587468e-06,
-      "loss": 0.7081,
+      "loss": 0.8263,
       "step": 15797
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.08376576914011e-06,
-      "loss": 0.7221,
+      "loss": 0.8436,
       "step": 15798
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.082609748277787e-06,
-      "loss": 0.7505,
+      "loss": 0.8127,
       "step": 15799
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.081453754016528e-06,
-      "loss": 0.7624,
+      "loss": 0.8476,
       "step": 15800
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.08029778637238e-06,
-      "loss": 0.6883,
+      "loss": 0.8091,
       "step": 15801
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.079141845361373e-06,
-      "loss": 0.6806,
+      "loss": 0.8189,
       "step": 15802
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.07798593099955e-06,
-      "loss": 0.6985,
+      "loss": 0.849,
       "step": 15803
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.076830043302944e-06,
-      "loss": 0.6462,
+      "loss": 0.7714,
       "step": 15804
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.075674182287596e-06,
-      "loss": 0.7607,
+      "loss": 0.8606,
       "step": 15805
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.074518347969532e-06,
-      "loss": 0.637,
+      "loss": 0.784,
       "step": 15806
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.0733625403648e-06,
-      "loss": 0.8011,
+      "loss": 0.8816,
       "step": 15807
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.072206759489424e-06,
-      "loss": 0.517,
+      "loss": 0.6527,
       "step": 15808
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.071051005359447e-06,
-      "loss": 0.6368,
+      "loss": 0.714,
       "step": 15809
     },
     {
       "epoch": 0.57,
       "learning_rate": 8.069895277990904e-06,
-      "loss": 0.8973,
+      "loss": 0.8705,
       "step": 15810
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.068739577399818e-06,
-      "loss": 0.7543,
+      "loss": 0.8074,
       "step": 15811
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.067583903602238e-06,
-      "loss": 0.7551,
+      "loss": 0.8626,
       "step": 15812
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.066428256614184e-06,
-      "loss": 0.7721,
+      "loss": 0.8138,
       "step": 15813
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.065272636451697e-06,
-      "loss": 0.7412,
+      "loss": 0.8109,
       "step": 15814
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.064117043130805e-06,
-      "loss": 0.7432,
+      "loss": 0.8638,
       "step": 15815
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.062961476667544e-06,
-      "loss": 0.8071,
+      "loss": 0.913,
       "step": 15816
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.061805937077943e-06,
-      "loss": 0.7387,
+      "loss": 0.8596,
       "step": 15817
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.060650424378039e-06,
-      "loss": 0.8047,
+      "loss": 0.9561,
       "step": 15818
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.059494938583851e-06,
-      "loss": 0.6628,
+      "loss": 0.7916,
       "step": 15819
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.058339479711425e-06,
-      "loss": 0.6599,
+      "loss": 0.7729,
       "step": 15820
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.057184047776779e-06,
-      "loss": 0.6664,
+      "loss": 0.811,
       "step": 15821
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.056028642795952e-06,
-      "loss": 0.6872,
+      "loss": 0.7785,
       "step": 15822
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.054873264784965e-06,
-      "loss": 0.8127,
+      "loss": 0.8777,
       "step": 15823
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.053717913759853e-06,
-      "loss": 0.536,
+      "loss": 0.6885,
       "step": 15824
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.052562589736645e-06,
-      "loss": 0.7066,
+      "loss": 0.863,
       "step": 15825
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.051407292731371e-06,
-      "loss": 0.7144,
+      "loss": 0.8123,
       "step": 15826
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.05025202276005e-06,
-      "loss": 0.6151,
+      "loss": 0.7998,
       "step": 15827
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.04909677983872e-06,
-      "loss": 0.6675,
+      "loss": 0.8351,
       "step": 15828
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.047941563983402e-06,
-      "loss": 0.75,
+      "loss": 0.7535,
       "step": 15829
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.046786375210124e-06,
-      "loss": 0.71,
+      "loss": 0.7768,
       "step": 15830
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.045631213534918e-06,
-      "loss": 0.8191,
+      "loss": 0.8615,
       "step": 15831
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.044476078973803e-06,
-      "loss": 0.7607,
+      "loss": 0.8286,
       "step": 15832
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.043320971542809e-06,
-      "loss": 0.744,
+      "loss": 0.8105,
       "step": 15833
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.042165891257959e-06,
-      "loss": 0.6633,
+      "loss": 0.8162,
       "step": 15834
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.04101083813528e-06,
-      "loss": 0.7739,
+      "loss": 0.8308,
       "step": 15835
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.039855812190795e-06,
-      "loss": 0.837,
+      "loss": 0.9589,
       "step": 15836
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.038700813440533e-06,
-      "loss": 0.6759,
+      "loss": 0.8061,
       "step": 15837
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.03754584190051e-06,
-      "loss": 0.6385,
+      "loss": 0.8519,
       "step": 15838
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.03639089758676e-06,
-      "loss": 0.6885,
+      "loss": 0.755,
       "step": 15839
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.035235980515295e-06,
-      "loss": 0.7002,
+      "loss": 0.8141,
       "step": 15840
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.034081090702146e-06,
-      "loss": 0.7775,
+      "loss": 0.8773,
       "step": 15841
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.032926228163329e-06,
-      "loss": 0.7129,
+      "loss": 0.8207,
       "step": 15842
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.031771392914873e-06,
-      "loss": 0.6344,
+      "loss": 0.8326,
       "step": 15843
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.030616584972794e-06,
-      "loss": 0.7477,
+      "loss": 0.8459,
       "step": 15844
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.02946180435312e-06,
-      "loss": 0.7831,
+      "loss": 0.8693,
       "step": 15845
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.028307051071862e-06,
-      "loss": 0.7326,
+      "loss": 0.7848,
       "step": 15846
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.027152325145052e-06,
-      "loss": 0.7422,
+      "loss": 0.9043,
       "step": 15847
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.025997626588702e-06,
-      "loss": 0.752,
+      "loss": 0.8276,
       "step": 15848
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.024842955418832e-06,
-      "loss": 0.7467,
+      "loss": 0.8405,
       "step": 15849
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.02368831165147e-06,
-      "loss": 0.7629,
+      "loss": 0.9669,
       "step": 15850
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.02253369530262e-06,
-      "loss": 0.5919,
+      "loss": 0.7594,
       "step": 15851
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.021379106388316e-06,
-      "loss": 0.7722,
+      "loss": 0.809,
       "step": 15852
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.020224544924566e-06,
-      "loss": 0.7775,
+      "loss": 0.8769,
       "step": 15853
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.019070010927393e-06,
-      "loss": 0.7148,
+      "loss": 0.7354,
       "step": 15854
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.01791550441281e-06,
-      "loss": 0.743,
+      "loss": 0.795,
       "step": 15855
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.01676102539684e-06,
-      "loss": 0.7628,
+      "loss": 0.7835,
       "step": 15856
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.015606573895496e-06,
-      "loss": 0.6112,
+      "loss": 0.6978,
       "step": 15857
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.014452149924799e-06,
-      "loss": 0.8092,
+      "loss": 0.9063,
       "step": 15858
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.013297753500756e-06,
-      "loss": 0.7951,
+      "loss": 0.9761,
       "step": 15859
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.012143384639391e-06,
-      "loss": 0.69,
+      "loss": 0.804,
       "step": 15860
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.010989043356715e-06,
-      "loss": 0.8011,
+      "loss": 0.9201,
       "step": 15861
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.009834729668744e-06,
-      "loss": 0.6229,
+      "loss": 0.7458,
       "step": 15862
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.00868044359149e-06,
-      "loss": 0.6075,
+      "loss": 0.7621,
       "step": 15863
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.007526185140974e-06,
-      "loss": 0.8004,
+      "loss": 0.8405,
       "step": 15864
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.006371954333201e-06,
-      "loss": 0.8552,
+      "loss": 0.8765,
       "step": 15865
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.005217751184194e-06,
-      "loss": 0.7423,
+      "loss": 0.895,
       "step": 15866
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.004063575709958e-06,
-      "loss": 0.7578,
+      "loss": 0.8929,
       "step": 15867
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.002909427926505e-06,
-      "loss": 0.647,
+      "loss": 0.7994,
       "step": 15868
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.001755307849854e-06,
-      "loss": 0.706,
+      "loss": 0.7712,
       "step": 15869
     },
     {
       "epoch": 0.58,
       "learning_rate": 8.00060121549601e-06,
-      "loss": 0.6198,
+      "loss": 0.7796,
       "step": 15870
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.999447150880992e-06,
-      "loss": 0.6958,
+      "loss": 0.826,
       "step": 15871
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.998293114020803e-06,
-      "loss": 0.8259,
+      "loss": 0.9195,
       "step": 15872
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.997139104931458e-06,
-      "loss": 0.7493,
+      "loss": 0.8414,
       "step": 15873
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.995985123628963e-06,
-      "loss": 0.6708,
+      "loss": 0.8078,
       "step": 15874
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.994831170129335e-06,
-      "loss": 0.7618,
+      "loss": 0.9054,
       "step": 15875
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.993677244448577e-06,
-      "loss": 0.7368,
+      "loss": 0.8399,
       "step": 15876
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.992523346602704e-06,
-      "loss": 0.6578,
+      "loss": 0.7764,
       "step": 15877
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.991369476607716e-06,
-      "loss": 0.7759,
+      "loss": 0.8349,
       "step": 15878
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.990215634479632e-06,
-      "loss": 0.734,
+      "loss": 0.8689,
       "step": 15879
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.98906182023445e-06,
-      "loss": 0.9674,
+      "loss": 0.9549,
       "step": 15880
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.987908033888186e-06,
-      "loss": 0.8008,
+      "loss": 0.8202,
       "step": 15881
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.986754275456841e-06,
-      "loss": 0.7159,
+      "loss": 0.8271,
       "step": 15882
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.985600544956424e-06,
-      "loss": 0.8376,
+      "loss": 0.9326,
       "step": 15883
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.984446842402942e-06,
-      "loss": 0.7771,
+      "loss": 0.8146,
       "step": 15884
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.983293167812402e-06,
-      "loss": 0.7119,
+      "loss": 0.8264,
       "step": 15885
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.982139521200809e-06,
-      "loss": 0.7181,
+      "loss": 0.8182,
       "step": 15886
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.980985902584162e-06,
-      "loss": 0.7705,
+      "loss": 0.8354,
       "step": 15887
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.979832311978474e-06,
-      "loss": 0.7962,
+      "loss": 0.8273,
       "step": 15888
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.978678749399744e-06,
-      "loss": 0.7264,
+      "loss": 0.7908,
       "step": 15889
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.977525214863984e-06,
-      "loss": 0.5775,
+      "loss": 0.7427,
       "step": 15890
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.976371708387185e-06,
-      "loss": 0.7617,
+      "loss": 0.9008,
       "step": 15891
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.975218229985365e-06,
-      "loss": 0.7221,
+      "loss": 0.8567,
       "step": 15892
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.974064779674513e-06,
-      "loss": 0.6043,
+      "loss": 0.808,
       "step": 15893
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.972911357470641e-06,
-      "loss": 0.8596,
+      "loss": 0.9006,
       "step": 15894
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.971757963389744e-06,
-      "loss": 0.6475,
+      "loss": 0.7876,
       "step": 15895
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.970604597447832e-06,
-      "loss": 0.8226,
+      "loss": 0.8762,
       "step": 15896
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.9694512596609e-06,
-      "loss": 0.8235,
+      "loss": 0.8949,
       "step": 15897
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.968297950044952e-06,
-      "loss": 0.5919,
+      "loss": 0.7134,
       "step": 15898
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.967144668615985e-06,
-      "loss": 0.705,
+      "loss": 0.8186,
       "step": 15899
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.965991415390004e-06,
-      "loss": 0.6838,
+      "loss": 0.8129,
       "step": 15900
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.964838190383004e-06,
-      "loss": 0.8435,
+      "loss": 0.9764,
       "step": 15901
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.963684993610988e-06,
-      "loss": 0.6118,
+      "loss": 0.7554,
       "step": 15902
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.962531825089952e-06,
-      "loss": 0.6021,
+      "loss": 0.7874,
       "step": 15903
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.9613786848359e-06,
-      "loss": 0.6835,
+      "loss": 0.8463,
       "step": 15904
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.960225572864825e-06,
-      "loss": 0.7221,
+      "loss": 0.8503,
       "step": 15905
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.959072489192723e-06,
-      "loss": 0.7415,
+      "loss": 0.8361,
       "step": 15906
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.957919433835597e-06,
-      "loss": 0.7507,
+      "loss": 0.8994,
       "step": 15907
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.95676640680944e-06,
-      "loss": 0.7762,
+      "loss": 0.8318,
       "step": 15908
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.955613408130253e-06,
-      "loss": 0.8,
+      "loss": 0.8201,
       "step": 15909
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.954460437814024e-06,
-      "loss": 0.622,
+      "loss": 0.8367,
       "step": 15910
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.953307495876762e-06,
-      "loss": 0.7239,
+      "loss": 0.8578,
       "step": 15911
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.952154582334447e-06,
-      "loss": 0.7068,
+      "loss": 0.8549,
       "step": 15912
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.951001697203085e-06,
-      "loss": 0.7738,
+      "loss": 0.8225,
       "step": 15913
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.949848840498667e-06,
-      "loss": 0.636,
+      "loss": 0.7658,
       "step": 15914
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.948696012237188e-06,
-      "loss": 0.6471,
+      "loss": 0.8261,
       "step": 15915
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.947543212434639e-06,
-      "loss": 0.8195,
+      "loss": 0.8923,
       "step": 15916
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.946390441107022e-06,
-      "loss": 0.6712,
+      "loss": 0.8889,
       "step": 15917
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.945237698270317e-06,
-      "loss": 0.773,
+      "loss": 0.8282,
       "step": 15918
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.944084983940528e-06,
-      "loss": 0.7189,
+      "loss": 0.8271,
       "step": 15919
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.94293229813364e-06,
-      "loss": 0.7239,
+      "loss": 0.8443,
       "step": 15920
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.94177964086565e-06,
-      "loss": 0.7813,
+      "loss": 0.8772,
       "step": 15921
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.940627012152545e-06,
-      "loss": 0.8008,
+      "loss": 0.8985,
       "step": 15922
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.93947441201032e-06,
-      "loss": 0.7985,
+      "loss": 0.8568,
       "step": 15923
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.938321840454968e-06,
-      "loss": 0.6778,
+      "loss": 0.7505,
       "step": 15924
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.93716929750247e-06,
-      "loss": 0.5961,
+      "loss": 0.7048,
       "step": 15925
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.936016783168822e-06,
-      "loss": 0.7427,
+      "loss": 0.8464,
       "step": 15926
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.934864297470011e-06,
-      "loss": 0.7527,
+      "loss": 0.9039,
       "step": 15927
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.933711840422031e-06,
-      "loss": 0.7462,
+      "loss": 0.9566,
       "step": 15928
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.932559412040865e-06,
-      "loss": 0.6775,
+      "loss": 0.7874,
       "step": 15929
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.93140701234251e-06,
-      "loss": 0.7314,
+      "loss": 0.8243,
       "step": 15930
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.930254641342941e-06,
-      "loss": 0.6116,
+      "loss": 0.7628,
       "step": 15931
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.929102299058154e-06,
-      "loss": 0.8239,
+      "loss": 0.8659,
       "step": 15932
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.927949985504133e-06,
-      "loss": 0.6959,
+      "loss": 0.776,
       "step": 15933
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.926797700696869e-06,
-      "loss": 0.7454,
+      "loss": 0.8494,
       "step": 15934
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.925645444652342e-06,
-      "loss": 0.8152,
+      "loss": 0.8376,
       "step": 15935
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.924493217386545e-06,
-      "loss": 0.6198,
+      "loss": 0.7785,
       "step": 15936
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.923341018915457e-06,
-      "loss": 0.8389,
+      "loss": 0.9258,
       "step": 15937
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.922188849255071e-06,
-      "loss": 0.613,
+      "loss": 0.8099,
       "step": 15938
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.92103670842136e-06,
-      "loss": 0.6591,
+      "loss": 0.8187,
       "step": 15939
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.91988459643032e-06,
-      "loss": 0.6831,
+      "loss": 0.7859,
       "step": 15940
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.918732513297926e-06,
-      "loss": 0.61,
+      "loss": 0.735,
       "step": 15941
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.917580459040169e-06,
-      "loss": 0.6939,
+      "loss": 0.8106,
       "step": 15942
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.91642843367303e-06,
-      "loss": 0.8439,
+      "loss": 0.9569,
       "step": 15943
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.915276437212484e-06,
-      "loss": 0.7419,
+      "loss": 0.8771,
       "step": 15944
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.914124469674524e-06,
-      "loss": 0.5828,
+      "loss": 0.6983,
       "step": 15945
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.912972531075125e-06,
-      "loss": 0.6815,
+      "loss": 0.8285,
       "step": 15946
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.911820621430273e-06,
-      "loss": 0.6493,
+      "loss": 0.7669,
       "step": 15947
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.910668740755945e-06,
-      "loss": 0.7561,
+      "loss": 0.8441,
       "step": 15948
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.909516889068128e-06,
-      "loss": 0.6645,
+      "loss": 0.7915,
       "step": 15949
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.908365066382793e-06,
-      "loss": 0.7921,
+      "loss": 0.819,
       "step": 15950
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.90721327271593e-06,
-      "loss": 0.6792,
+      "loss": 0.7583,
       "step": 15951
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.906061508083508e-06,
-      "loss": 0.6402,
+      "loss": 0.7878,
       "step": 15952
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.904909772501516e-06,
-      "loss": 0.8454,
+      "loss": 0.9325,
       "step": 15953
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.903758065985927e-06,
-      "loss": 0.7539,
+      "loss": 0.8571,
       "step": 15954
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.902606388552721e-06,
-      "loss": 0.769,
+      "loss": 0.8341,
       "step": 15955
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.901454740217874e-06,
-      "loss": 0.8018,
+      "loss": 0.9031,
       "step": 15956
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.90030312099737e-06,
-      "loss": 0.8125,
+      "loss": 0.8405,
       "step": 15957
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.899151530907175e-06,
-      "loss": 0.6538,
+      "loss": 0.745,
       "step": 15958
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.897999969963276e-06,
-      "loss": 0.6553,
+      "loss": 0.8062,
       "step": 15959
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.896848438181641e-06,
-      "loss": 0.8488,
+      "loss": 0.9404,
       "step": 15960
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.895696935578255e-06,
-      "loss": 0.7711,
+      "loss": 0.8869,
       "step": 15961
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.894545462169083e-06,
-      "loss": 0.7485,
+      "loss": 0.8035,
       "step": 15962
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.893394017970109e-06,
-      "loss": 0.6989,
+      "loss": 0.8003,
       "step": 15963
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.892242602997306e-06,
-      "loss": 0.6558,
+      "loss": 0.7656,
       "step": 15964
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.891091217266641e-06,
-      "loss": 0.8006,
+      "loss": 0.9271,
       "step": 15965
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.889939860794097e-06,
-      "loss": 0.6339,
+      "loss": 0.7622,
       "step": 15966
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.88878853359564e-06,
-      "loss": 0.805,
+      "loss": 0.8873,
       "step": 15967
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.887637235687251e-06,
-      "loss": 0.7449,
+      "loss": 0.8135,
       "step": 15968
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.886485967084896e-06,
-      "loss": 0.7154,
+      "loss": 0.8279,
       "step": 15969
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.885334727804553e-06,
-      "loss": 0.6692,
+      "loss": 0.7328,
       "step": 15970
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.884183517862186e-06,
-      "loss": 0.7485,
+      "loss": 0.8231,
       "step": 15971
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.883032337273773e-06,
-      "loss": 0.8017,
+      "loss": 0.8578,
       "step": 15972
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.88188118605528e-06,
-      "loss": 0.8168,
+      "loss": 0.9087,
       "step": 15973
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.880730064222683e-06,
-      "loss": 0.6762,
+      "loss": 0.8116,
       "step": 15974
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.879578971791949e-06,
-      "loss": 0.735,
+      "loss": 0.8974,
       "step": 15975
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.87842790877905e-06,
-      "loss": 0.7402,
+      "loss": 0.8125,
       "step": 15976
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.877276875199948e-06,
-      "loss": 0.6625,
+      "loss": 0.7777,
       "step": 15977
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.876125871070623e-06,
-      "loss": 0.7341,
+      "loss": 0.7682,
       "step": 15978
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.874974896407035e-06,
-      "loss": 0.8122,
+      "loss": 0.872,
       "step": 15979
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.873823951225158e-06,
-      "loss": 0.643,
+      "loss": 0.7308,
       "step": 15980
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.872673035540955e-06,
-      "loss": 0.7472,
+      "loss": 0.8066,
       "step": 15981
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.871522149370397e-06,
-      "loss": 0.8717,
+      "loss": 0.9011,
       "step": 15982
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.87037129272945e-06,
-      "loss": 0.7192,
+      "loss": 0.8718,
       "step": 15983
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.869220465634076e-06,
-      "loss": 0.5919,
+      "loss": 0.8112,
       "step": 15984
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.86806966810025e-06,
-      "loss": 0.6494,
+      "loss": 0.7979,
       "step": 15985
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.866918900143926e-06,
-      "loss": 0.7629,
+      "loss": 0.8541,
       "step": 15986
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.86576816178108e-06,
-      "loss": 0.8385,
+      "loss": 0.9264,
       "step": 15987
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.864617453027671e-06,
-      "loss": 0.6319,
+      "loss": 0.7743,
       "step": 15988
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.863466773899669e-06,
-      "loss": 0.8018,
+      "loss": 0.818,
       "step": 15989
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.862316124413027e-06,
-      "loss": 0.7308,
+      "loss": 0.8425,
       "step": 15990
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.861165504583723e-06,
-      "loss": 0.8351,
+      "loss": 0.894,
       "step": 15991
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.860014914427709e-06,
-      "loss": 0.7784,
+      "loss": 0.9401,
       "step": 15992
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.858864353960954e-06,
-      "loss": 0.5791,
+      "loss": 0.7506,
       "step": 15993
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.857713823199415e-06,
-      "loss": 0.7264,
+      "loss": 0.8144,
       "step": 15994
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.85656332215906e-06,
-      "loss": 0.7541,
+      "loss": 0.8705,
       "step": 15995
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.855412850855846e-06,
-      "loss": 0.7494,
+      "loss": 0.8898,
       "step": 15996
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.854262409305739e-06,
-      "loss": 0.826,
+      "loss": 0.8693,
       "step": 15997
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.853111997524694e-06,
-      "loss": 0.6968,
+      "loss": 0.8604,
       "step": 15998
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.851961615528675e-06,
-      "loss": 0.7127,
+      "loss": 0.7872,
       "step": 15999
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.85081126333364e-06,
-      "loss": 0.7549,
+      "loss": 0.9079,
       "step": 16000
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.84966094095555e-06,
-      "loss": 0.683,
+      "loss": 0.7662,
       "step": 16001
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.848510648410367e-06,
-      "loss": 0.7421,
+      "loss": 0.8696,
       "step": 16002
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.847360385714041e-06,
-      "loss": 0.7416,
+      "loss": 0.8233,
       "step": 16003
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.846210152882539e-06,
-      "loss": 0.6398,
+      "loss": 0.8264,
       "step": 16004
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.84505994993181e-06,
-      "loss": 0.7065,
+      "loss": 0.8332,
       "step": 16005
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.843909776877822e-06,
-      "loss": 0.8302,
+      "loss": 0.9041,
       "step": 16006
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.842759633736522e-06,
-      "loss": 0.8057,
+      "loss": 0.8822,
       "step": 16007
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.841609520523877e-06,
-      "loss": 0.7377,
+      "loss": 0.8158,
       "step": 16008
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.840459437255833e-06,
-      "loss": 0.6361,
+      "loss": 0.7553,
       "step": 16009
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.839309383948355e-06,
-      "loss": 0.6476,
+      "loss": 0.7849,
       "step": 16010
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.83815936061739e-06,
-      "loss": 0.7065,
+      "loss": 0.8331,
       "step": 16011
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.837009367278898e-06,
-      "loss": 0.6637,
+      "loss": 0.7301,
       "step": 16012
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.83585940394883e-06,
-      "loss": 0.722,
+      "loss": 0.8118,
       "step": 16013
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.834709470643145e-06,
-      "loss": 0.6705,
+      "loss": 0.8013,
       "step": 16014
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.833559567377791e-06,
-      "loss": 0.6637,
+      "loss": 0.7859,
       "step": 16015
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.83240969416873e-06,
-      "loss": 0.7088,
+      "loss": 0.743,
       "step": 16016
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.8312598510319e-06,
-      "loss": 0.7433,
+      "loss": 0.7937,
       "step": 16017
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.83011003798327e-06,
-      "loss": 0.6932,
+      "loss": 0.7667,
       "step": 16018
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.828960255038783e-06,
-      "loss": 0.7539,
+      "loss": 0.846,
       "step": 16019
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.82781050221439e-06,
-      "loss": 0.6147,
+      "loss": 0.7899,
       "step": 16020
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.826660779526049e-06,
-      "loss": 0.6854,
+      "loss": 0.7967,
       "step": 16021
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.8255110869897e-06,
-      "loss": 0.7233,
+      "loss": 0.8266,
       "step": 16022
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.824361424621305e-06,
-      "loss": 0.6832,
+      "loss": 0.8165,
       "step": 16023
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.823211792436805e-06,
-      "loss": 0.8255,
+      "loss": 0.9301,
       "step": 16024
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.822062190452153e-06,
-      "loss": 0.922,
+      "loss": 0.9442,
       "step": 16025
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.820912618683298e-06,
-      "loss": 0.6895,
+      "loss": 0.8177,
       "step": 16026
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.81976307714619e-06,
-      "loss": 0.7228,
+      "loss": 0.8421,
       "step": 16027
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.818613565856775e-06,
-      "loss": 0.7561,
+      "loss": 0.8982,
       "step": 16028
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.817464084831003e-06,
-      "loss": 0.6009,
+      "loss": 0.7788,
       "step": 16029
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.816314634084819e-06,
-      "loss": 0.6554,
+      "loss": 0.7623,
       "step": 16030
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.815165213634171e-06,
-      "loss": 0.6064,
+      "loss": 0.7892,
       "step": 16031
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.814015823495006e-06,
-      "loss": 0.6867,
+      "loss": 0.8168,
       "step": 16032
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.81286646368327e-06,
-      "loss": 0.7959,
+      "loss": 0.8617,
       "step": 16033
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.811717134214908e-06,
-      "loss": 0.7075,
+      "loss": 0.7742,
       "step": 16034
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.810567835105867e-06,
-      "loss": 0.7568,
+      "loss": 0.8318,
       "step": 16035
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.80941856637209e-06,
-      "loss": 0.7697,
+      "loss": 0.9091,
       "step": 16036
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.808269328029526e-06,
-      "loss": 0.6512,
+      "loss": 0.8879,
       "step": 16037
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.807120120094112e-06,
-      "loss": 0.5486,
+      "loss": 0.8014,
       "step": 16038
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.805970942581795e-06,
-      "loss": 0.8074,
+      "loss": 0.8615,
       "step": 16039
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.80482179550852e-06,
-      "loss": 0.7061,
+      "loss": 0.8347,
       "step": 16040
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.803672678890225e-06,
-      "loss": 0.7508,
+      "loss": 0.8543,
       "step": 16041
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.802523592742862e-06,
-      "loss": 0.7757,
+      "loss": 0.8699,
       "step": 16042
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.801374537082359e-06,
-      "loss": 0.6996,
+      "loss": 0.8679,
       "step": 16043
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.800225511924669e-06,
-      "loss": 0.7137,
+      "loss": 0.7912,
       "step": 16044
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.799076517285725e-06,
-      "loss": 0.7386,
+      "loss": 0.8401,
       "step": 16045
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.797927553181474e-06,
-      "loss": 0.6162,
+      "loss": 0.8279,
       "step": 16046
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.796778619627853e-06,
-      "loss": 0.6851,
+      "loss": 0.7951,
       "step": 16047
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.795629716640807e-06,
-      "loss": 0.8027,
+      "loss": 0.8427,
       "step": 16048
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.794480844236265e-06,
-      "loss": 0.7962,
+      "loss": 0.8895,
       "step": 16049
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.793332002430178e-06,
-      "loss": 0.7651,
+      "loss": 0.8631,
       "step": 16050
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.792183191238472e-06,
-      "loss": 0.7894,
+      "loss": 0.8954,
       "step": 16051
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.791034410677096e-06,
-      "loss": 0.861,
+      "loss": 0.9327,
       "step": 16052
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.789885660761981e-06,
-      "loss": 0.7056,
+      "loss": 0.8024,
       "step": 16053
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.78873694150907e-06,
-      "loss": 0.7793,
+      "loss": 0.8366,
       "step": 16054
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.787588252934293e-06,
-      "loss": 0.6598,
+      "loss": 0.7139,
       "step": 16055
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.786439595053595e-06,
-      "loss": 0.6498,
+      "loss": 0.8224,
       "step": 16056
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.7852909678829e-06,
-      "loss": 0.6685,
+      "loss": 0.7341,
       "step": 16057
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.784142371438158e-06,
-      "loss": 0.6537,
+      "loss": 0.8837,
       "step": 16058
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.782993805735294e-06,
-      "loss": 0.6295,
+      "loss": 0.8398,
       "step": 16059
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.781845270790242e-06,
-      "loss": 0.6808,
+      "loss": 0.7536,
       "step": 16060
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.780696766618947e-06,
-      "loss": 0.7581,
+      "loss": 0.9239,
       "step": 16061
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.779548293237326e-06,
-      "loss": 0.6449,
+      "loss": 0.8647,
       "step": 16062
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.77839985066133e-06,
-      "loss": 0.7712,
+      "loss": 0.846,
       "step": 16063
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.77725143890688e-06,
-      "loss": 0.751,
+      "loss": 0.8032,
       "step": 16064
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.776103057989913e-06,
-      "loss": 0.7304,
+      "loss": 0.8474,
       "step": 16065
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.774954707926361e-06,
-      "loss": 0.7645,
+      "loss": 0.8744,
       "step": 16066
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.773806388732156e-06,
-      "loss": 0.7846,
+      "loss": 0.9069,
       "step": 16067
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.772658100423228e-06,
-      "loss": 0.6782,
+      "loss": 0.817,
       "step": 16068
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.77150984301551e-06,
-      "loss": 0.5547,
+      "loss": 0.7966,
       "step": 16069
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.77036161652493e-06,
-      "loss": 0.7213,
+      "loss": 0.8078,
       "step": 16070
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.76921342096742e-06,
-      "loss": 0.8184,
+      "loss": 0.922,
       "step": 16071
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.768065256358908e-06,
-      "loss": 0.7334,
+      "loss": 0.8681,
       "step": 16072
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.766917122715325e-06,
-      "loss": 0.6597,
+      "loss": 0.7798,
       "step": 16073
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.765769020052595e-06,
-      "loss": 0.6741,
+      "loss": 0.7764,
       "step": 16074
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.764620948386655e-06,
-      "loss": 0.7493,
+      "loss": 0.9065,
       "step": 16075
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.763472907733423e-06,
-      "loss": 0.6532,
+      "loss": 0.8245,
       "step": 16076
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.762324898108835e-06,
-      "loss": 0.7085,
+      "loss": 0.8236,
       "step": 16077
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.761176919528814e-06,
-      "loss": 0.7075,
+      "loss": 0.7984,
       "step": 16078
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.760028972009284e-06,
-      "loss": 0.6985,
+      "loss": 0.8251,
       "step": 16079
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.758881055566177e-06,
-      "loss": 0.8222,
+      "loss": 0.9669,
       "step": 16080
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.757733170215412e-06,
-      "loss": 0.7947,
+      "loss": 0.8795,
       "step": 16081
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.756585315972922e-06,
-      "loss": 0.7192,
+      "loss": 0.8444,
       "step": 16082
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.755437492854624e-06,
-      "loss": 0.7749,
+      "loss": 0.8882,
       "step": 16083
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.75428970087645e-06,
-      "loss": 0.7236,
+      "loss": 0.8255,
       "step": 16084
     },
     {
       "epoch": 0.58,
       "learning_rate": 7.753141940054316e-06,
-      "loss": 0.7034,
+      "loss": 0.8447,
       "step": 16085
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.751994210404153e-06,
-      "loss": 0.847,
+      "loss": 0.8754,
       "step": 16086
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.750846511941878e-06,
-      "loss": 0.7468,
+      "loss": 0.8508,
       "step": 16087
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.74969884468342e-06,
-      "loss": 0.7953,
+      "loss": 0.9003,
       "step": 16088
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.748551208644693e-06,
-      "loss": 0.8015,
+      "loss": 0.8741,
       "step": 16089
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.747403603841629e-06,
-      "loss": 0.7029,
+      "loss": 0.797,
       "step": 16090
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.746256030290139e-06,
-      "loss": 0.8109,
+      "loss": 0.9368,
       "step": 16091
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.745108488006152e-06,
-      "loss": 0.703,
+      "loss": 0.8129,
       "step": 16092
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.743960977005584e-06,
-      "loss": 0.7096,
+      "loss": 0.8091,
       "step": 16093
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.742813497304358e-06,
-      "loss": 0.6911,
+      "loss": 0.7675,
       "step": 16094
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.74166604891839e-06,
-      "loss": 0.7143,
+      "loss": 0.7916,
       "step": 16095
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.740518631863606e-06,
-      "loss": 0.6591,
+      "loss": 0.8401,
       "step": 16096
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.739371246155918e-06,
-      "loss": 0.8923,
+      "loss": 0.9252,
       "step": 16097
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.738223891811244e-06,
-      "loss": 0.81,
+      "loss": 0.8655,
       "step": 16098
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.737076568845506e-06,
-      "loss": 0.7268,
+      "loss": 0.8187,
       "step": 16099
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.735929277274618e-06,
-      "loss": 0.7161,
+      "loss": 0.8765,
       "step": 16100
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.734782017114506e-06,
-      "loss": 0.7638,
+      "loss": 0.8412,
       "step": 16101
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.73363478838107e-06,
-      "loss": 0.841,
+      "loss": 0.9599,
       "step": 16102
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.732487591090243e-06,
-      "loss": 0.8141,
+      "loss": 0.8543,
       "step": 16103
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.731340425257929e-06,
-      "loss": 0.591,
+      "loss": 0.7766,
       "step": 16104
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.73019329090005e-06,
-      "loss": 0.7428,
+      "loss": 0.8353,
       "step": 16105
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.729046188032518e-06,
-      "loss": 0.8331,
+      "loss": 0.8387,
       "step": 16106
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.72789911667125e-06,
-      "loss": 0.7657,
+      "loss": 0.8623,
       "step": 16107
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.726752076832154e-06,
-      "loss": 0.6195,
+      "loss": 0.8071,
       "step": 16108
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.725605068531154e-06,
-      "loss": 0.8982,
+      "loss": 0.9383,
       "step": 16109
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.72445809178415e-06,
-      "loss": 0.6826,
+      "loss": 0.7355,
       "step": 16110
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.723311146607066e-06,
-      "loss": 0.7231,
+      "loss": 0.8276,
       "step": 16111
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.722164233015804e-06,
-      "loss": 0.8419,
+      "loss": 0.877,
       "step": 16112
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.721017351026287e-06,
-      "loss": 0.6106,
+      "loss": 0.7329,
       "step": 16113
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.719870500654419e-06,
-      "loss": 0.7571,
+      "loss": 0.865,
       "step": 16114
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.718723681916116e-06,
-      "loss": 0.6328,
+      "loss": 0.7484,
       "step": 16115
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.717576894827283e-06,
-      "loss": 0.7058,
+      "loss": 0.8449,
       "step": 16116
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.716430139403829e-06,
-      "loss": 0.7043,
+      "loss": 0.7877,
       "step": 16117
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.71528341566167e-06,
-      "loss": 0.8187,
+      "loss": 0.9011,
       "step": 16118
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.71413672361671e-06,
-      "loss": 0.7715,
+      "loss": 0.7986,
       "step": 16119
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.712990063284864e-06,
-      "loss": 0.8273,
+      "loss": 0.8107,
       "step": 16120
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.711843434682029e-06,
-      "loss": 0.7255,
+      "loss": 0.7664,
       "step": 16121
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.710696837824127e-06,
-      "loss": 0.8437,
+      "loss": 0.8996,
       "step": 16122
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.709550272727053e-06,
-      "loss": 0.8158,
+      "loss": 0.9272,
       "step": 16123
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.708403739406722e-06,
-      "loss": 0.9056,
+      "loss": 0.9561,
       "step": 16124
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.707257237879036e-06,
-      "loss": 0.8027,
+      "loss": 0.8807,
       "step": 16125
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.706110768159905e-06,
-      "loss": 0.7647,
+      "loss": 0.9025,
       "step": 16126
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.70496433026523e-06,
-      "loss": 0.7664,
+      "loss": 0.8899,
       "step": 16127
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.703817924210922e-06,
-      "loss": 0.7273,
+      "loss": 0.9256,
       "step": 16128
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.702671550012876e-06,
-      "loss": 0.7746,
+      "loss": 0.8112,
       "step": 16129
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.701525207687011e-06,
-      "loss": 0.6481,
+      "loss": 0.8045,
       "step": 16130
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.700378897249217e-06,
-      "loss": 0.7197,
+      "loss": 0.8664,
       "step": 16131
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.699232618715406e-06,
-      "loss": 0.7708,
+      "loss": 0.879,
       "step": 16132
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.698086372101477e-06,
-      "loss": 0.6422,
+      "loss": 0.7655,
       "step": 16133
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.696940157423333e-06,
-      "loss": 0.6386,
+      "loss": 0.7106,
       "step": 16134
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.695793974696882e-06,
-      "loss": 0.7209,
+      "loss": 0.8052,
       "step": 16135
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.694647823938012e-06,
-      "loss": 0.7216,
+      "loss": 0.8147,
       "step": 16136
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.693501705162638e-06,
-      "loss": 0.6453,
+      "loss": 0.7338,
       "step": 16137
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.692355618386654e-06,
-      "loss": 0.824,
+      "loss": 0.9368,
       "step": 16138
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.69120956362596e-06,
-      "loss": 0.6317,
+      "loss": 0.8259,
       "step": 16139
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.690063540896458e-06,
-      "loss": 0.797,
+      "loss": 0.8335,
       "step": 16140
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.688917550214051e-06,
-      "loss": 0.7598,
+      "loss": 0.8643,
       "step": 16141
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.68777159159463e-06,
-      "loss": 0.7115,
+      "loss": 0.8137,
       "step": 16142
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.686625665054099e-06,
-      "loss": 0.7339,
+      "loss": 0.8011,
       "step": 16143
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.685479770608351e-06,
-      "loss": 0.771,
+      "loss": 0.8531,
       "step": 16144
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.684333908273291e-06,
-      "loss": 0.8381,
+      "loss": 0.9359,
       "step": 16145
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.683188078064811e-06,
-      "loss": 0.7811,
+      "loss": 0.9064,
       "step": 16146
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.682042279998815e-06,
-      "loss": 0.8033,
+      "loss": 0.8197,
       "step": 16147
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.680896514091184e-06,
-      "loss": 0.7648,
+      "loss": 0.8583,
       "step": 16148
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.679750780357831e-06,
-      "loss": 0.6779,
+      "loss": 0.7616,
       "step": 16149
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.67860507881464e-06,
-      "loss": 0.6114,
+      "loss": 0.7265,
       "step": 16150
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.677459409477513e-06,
-      "loss": 0.6535,
+      "loss": 0.7565,
       "step": 16151
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.676313772362338e-06,
-      "loss": 0.6566,
+      "loss": 0.7827,
       "step": 16152
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.675168167485014e-06,
-      "loss": 0.7917,
+      "loss": 0.8316,
       "step": 16153
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.674022594861435e-06,
-      "loss": 0.6804,
+      "loss": 0.7658,
       "step": 16154
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.672877054507491e-06,
-      "loss": 0.7793,
+      "loss": 0.8473,
       "step": 16155
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.671731546439075e-06,
-      "loss": 0.6212,
+      "loss": 0.7816,
       "step": 16156
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.67058607067208e-06,
-      "loss": 0.7254,
+      "loss": 0.8934,
       "step": 16157
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.6694406272224e-06,
-      "loss": 0.7413,
+      "loss": 0.8046,
       "step": 16158
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.668295216105924e-06,
-      "loss": 0.5418,
+      "loss": 0.7789,
       "step": 16159
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.667149837338546e-06,
-      "loss": 0.7118,
+      "loss": 0.7966,
       "step": 16160
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.666004490936148e-06,
-      "loss": 0.7391,
+      "loss": 0.7749,
       "step": 16161
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.664859176914633e-06,
-      "loss": 0.7104,
+      "loss": 0.8512,
       "step": 16162
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.66371389528988e-06,
-      "loss": 0.5633,
+      "loss": 0.7682,
       "step": 16163
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.662568646077783e-06,
-      "loss": 0.6108,
+      "loss": 0.7664,
       "step": 16164
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.661423429294228e-06,
-      "loss": 0.7635,
+      "loss": 0.8705,
       "step": 16165
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.660278244955106e-06,
-      "loss": 0.825,
+      "loss": 0.8709,
       "step": 16166
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.659133093076303e-06,
-      "loss": 0.7338,
+      "loss": 0.7897,
       "step": 16167
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.65798797367371e-06,
-      "loss": 0.6837,
+      "loss": 0.8366,
       "step": 16168
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.65684288676321e-06,
-      "loss": 0.6086,
+      "loss": 0.673,
       "step": 16169
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.655697832360689e-06,
-      "loss": 0.8189,
+      "loss": 0.8512,
       "step": 16170
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.654552810482034e-06,
-      "loss": 0.7907,
+      "loss": 0.9171,
       "step": 16171
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.653407821143132e-06,
-      "loss": 0.7608,
+      "loss": 0.873,
       "step": 16172
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.652262864359869e-06,
-      "loss": 0.6778,
+      "loss": 0.8602,
       "step": 16173
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.651117940148123e-06,
-      "loss": 0.6895,
+      "loss": 0.8406,
       "step": 16174
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.64997304852379e-06,
-      "loss": 0.7643,
+      "loss": 0.8357,
       "step": 16175
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.648828189502741e-06,
-      "loss": 0.6633,
+      "loss": 0.8191,
       "step": 16176
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.647683363100866e-06,
-      "loss": 0.7289,
+      "loss": 0.7892,
       "step": 16177
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.646538569334047e-06,
-      "loss": 0.7477,
+      "loss": 0.8206,
       "step": 16178
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.645393808218166e-06,
-      "loss": 0.6563,
+      "loss": 0.7544,
       "step": 16179
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.644249079769105e-06,
-      "loss": 0.8023,
+      "loss": 0.8547,
       "step": 16180
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.643104384002748e-06,
-      "loss": 0.7184,
+      "loss": 0.8433,
       "step": 16181
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.641959720934971e-06,
-      "loss": 0.634,
+      "loss": 0.8094,
       "step": 16182
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.640815090581658e-06,
-      "loss": 0.8413,
+      "loss": 0.8752,
       "step": 16183
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.639670492958688e-06,
-      "loss": 0.7943,
+      "loss": 0.9289,
       "step": 16184
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.638525928081943e-06,
-      "loss": 0.8059,
+      "loss": 0.8246,
       "step": 16185
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.637381395967298e-06,
-      "loss": 0.7497,
+      "loss": 0.8306,
       "step": 16186
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.636236896630638e-06,
-      "loss": 0.6059,
+      "loss": 0.7664,
       "step": 16187
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.63509243008783e-06,
-      "loss": 0.7779,
+      "loss": 0.8726,
       "step": 16188
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.633947996354766e-06,
-      "loss": 0.6807,
+      "loss": 0.8563,
       "step": 16189
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.632803595447313e-06,
-      "loss": 0.8037,
+      "loss": 0.8391,
       "step": 16190
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.631659227381354e-06,
-      "loss": 0.646,
+      "loss": 0.7922,
       "step": 16191
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.63051489217276e-06,
-      "loss": 0.8055,
+      "loss": 0.8575,
       "step": 16192
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.629370589837412e-06,
-      "loss": 0.6896,
+      "loss": 0.7553,
       "step": 16193
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.628226320391185e-06,
-      "loss": 0.572,
+      "loss": 0.6937,
       "step": 16194
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.627082083849951e-06,
-      "loss": 0.6505,
+      "loss": 0.7791,
       "step": 16195
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.625937880229586e-06,
-      "loss": 0.6924,
+      "loss": 0.7862,
       "step": 16196
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.624793709545964e-06,
-      "loss": 0.7715,
+      "loss": 0.8919,
       "step": 16197
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.62364957181496e-06,
-      "loss": 0.7107,
+      "loss": 0.8456,
       "step": 16198
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.622505467052446e-06,
-      "loss": 0.7443,
+      "loss": 0.8595,
       "step": 16199
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.621361395274298e-06,
-      "loss": 0.9383,
+      "loss": 0.972,
       "step": 16200
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.62021735649638e-06,
-      "loss": 0.6825,
+      "loss": 0.8171,
       "step": 16201
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.619073350734576e-06,
-      "loss": 0.6529,
+      "loss": 0.762,
       "step": 16202
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.6179293780047485e-06,
-      "loss": 0.8049,
+      "loss": 0.8752,
       "step": 16203
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.616785438322772e-06,
-      "loss": 0.6442,
+      "loss": 0.8175,
       "step": 16204
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.615641531704514e-06,
-      "loss": 0.7546,
+      "loss": 0.896,
       "step": 16205
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.614497658165848e-06,
-      "loss": 0.8005,
+      "loss": 0.8812,
       "step": 16206
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.613353817722641e-06,
-      "loss": 0.6098,
+      "loss": 0.7236,
       "step": 16207
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.61221001039077e-06,
-      "loss": 0.8088,
+      "loss": 0.9731,
       "step": 16208
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.6110662361860895e-06,
-      "loss": 0.7259,
+      "loss": 0.8032,
       "step": 16209
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.60992249512448e-06,
-      "loss": 0.7057,
+      "loss": 0.8319,
       "step": 16210
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.6087787872218025e-06,
-      "loss": 0.6127,
+      "loss": 0.7832,
       "step": 16211
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.607635112493927e-06,
-      "loss": 0.6787,
+      "loss": 0.7643,
       "step": 16212
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.606491470956724e-06,
-      "loss": 0.8098,
+      "loss": 0.81,
       "step": 16213
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.605347862626051e-06,
-      "loss": 0.6112,
+      "loss": 0.806,
       "step": 16214
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.604204287517781e-06,
-      "loss": 0.7449,
+      "loss": 0.8001,
       "step": 16215
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.6030607456477755e-06,
-      "loss": 0.7246,
+      "loss": 0.9067,
       "step": 16216
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.601917237031902e-06,
-      "loss": 0.7944,
+      "loss": 0.8349,
       "step": 16217
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.600773761686025e-06,
-      "loss": 0.7823,
+      "loss": 0.8938,
       "step": 16218
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.59963031962601e-06,
-      "loss": 0.8344,
+      "loss": 0.9088,
       "step": 16219
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.598486910867713e-06,
-      "loss": 0.7933,
+      "loss": 0.8223,
       "step": 16220
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.597343535427009e-06,
-      "loss": 0.8086,
+      "loss": 0.8792,
       "step": 16221
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.59620019331975e-06,
-      "loss": 0.7099,
+      "loss": 0.7766,
       "step": 16222
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.595056884561806e-06,
-      "loss": 0.7598,
+      "loss": 0.8484,
       "step": 16223
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.593913609169032e-06,
-      "loss": 0.6942,
+      "loss": 0.782,
       "step": 16224
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.592770367157296e-06,
-      "loss": 0.681,
+      "loss": 0.8223,
       "step": 16225
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.591627158542454e-06,
-      "loss": 0.8118,
+      "loss": 0.82,
       "step": 16226
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.590483983340371e-06,
-      "loss": 0.7569,
+      "loss": 0.8459,
       "step": 16227
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.589340841566899e-06,
-      "loss": 0.6304,
+      "loss": 0.8038,
       "step": 16228
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.588197733237908e-06,
-      "loss": 0.5955,
+      "loss": 0.7501,
       "step": 16229
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.587054658369247e-06,
-      "loss": 0.6556,
+      "loss": 0.7824,
       "step": 16230
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.585911616976782e-06,
-      "loss": 0.7503,
+      "loss": 0.8694,
       "step": 16231
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.584768609076371e-06,
-      "loss": 0.7331,
+      "loss": 0.7994,
       "step": 16232
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.583625634683861e-06,
-      "loss": 0.6373,
+      "loss": 0.787,
       "step": 16233
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.582482693815124e-06,
-      "loss": 0.6493,
+      "loss": 0.854,
       "step": 16234
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.581339786486005e-06,
-      "loss": 0.6028,
+      "loss": 0.799,
       "step": 16235
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.580196912712368e-06,
-      "loss": 0.7907,
+      "loss": 0.8602,
       "step": 16236
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.579054072510062e-06,
-      "loss": 0.6977,
+      "loss": 0.8901,
       "step": 16237
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.577911265894948e-06,
-      "loss": 0.6976,
+      "loss": 0.8383,
       "step": 16238
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.576768492882877e-06,
-      "loss": 0.7177,
+      "loss": 0.8627,
       "step": 16239
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.57562575348971e-06,
-      "loss": 0.5553,
+      "loss": 0.6892,
       "step": 16240
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.574483047731291e-06,
-      "loss": 0.6795,
+      "loss": 0.8666,
       "step": 16241
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.573340375623481e-06,
-      "loss": 0.7308,
+      "loss": 0.8879,
       "step": 16242
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.572197737182129e-06,
-      "loss": 0.7462,
+      "loss": 0.8695,
       "step": 16243
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.571055132423088e-06,
-      "loss": 0.7034,
+      "loss": 0.8328,
       "step": 16244
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.569912561362212e-06,
-      "loss": 0.7463,
+      "loss": 0.8076,
       "step": 16245
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.568770024015352e-06,
-      "loss": 0.7991,
+      "loss": 0.8463,
       "step": 16246
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.567627520398356e-06,
-      "loss": 0.7403,
+      "loss": 0.8982,
       "step": 16247
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.566485050527082e-06,
-      "loss": 0.7884,
+      "loss": 0.8468,
       "step": 16248
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.56534261441737e-06,
-      "loss": 0.6659,
+      "loss": 0.7735,
       "step": 16249
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5642002120850775e-06,
-      "loss": 0.73,
+      "loss": 0.8983,
       "step": 16250
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.563057843546051e-06,
-      "loss": 0.6893,
+      "loss": 0.7765,
       "step": 16251
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.561915508816138e-06,
-      "loss": 0.6915,
+      "loss": 0.7304,
       "step": 16252
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.560773207911192e-06,
-      "loss": 0.8544,
+      "loss": 0.9274,
       "step": 16253
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.559630940847053e-06,
-      "loss": 0.7915,
+      "loss": 0.8581,
       "step": 16254
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.558488707639573e-06,
-      "loss": 0.7048,
+      "loss": 0.8465,
       "step": 16255
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5573465083045975e-06,
-      "loss": 0.7838,
+      "loss": 0.8546,
       "step": 16256
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5562043428579745e-06,
-      "loss": 0.7076,
+      "loss": 0.8467,
       "step": 16257
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.555062211315547e-06,
-      "loss": 0.7026,
+      "loss": 0.7579,
       "step": 16258
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.553920113693167e-06,
-      "loss": 0.6072,
+      "loss": 0.7292,
       "step": 16259
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.552778050006669e-06,
-      "loss": 0.8681,
+      "loss": 0.8524,
       "step": 16260
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.55163602027191e-06,
-      "loss": 0.8087,
+      "loss": 0.7843,
       "step": 16261
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.550494024504722e-06,
-      "loss": 0.7231,
+      "loss": 0.8584,
       "step": 16262
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.549352062720956e-06,
-      "loss": 0.6407,
+      "loss": 0.7661,
       "step": 16263
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.548210134936452e-06,
-      "loss": 0.712,
+      "loss": 0.8534,
       "step": 16264
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.547068241167056e-06,
-      "loss": 0.6188,
+      "loss": 0.7315,
       "step": 16265
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.545926381428605e-06,
-      "loss": 0.8609,
+      "loss": 0.894,
       "step": 16266
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5447845557369485e-06,
-      "loss": 0.8441,
+      "loss": 0.8627,
       "step": 16267
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.543642764107918e-06,
-      "loss": 0.9026,
+      "loss": 0.8904,
       "step": 16268
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.542501006557361e-06,
-      "loss": 0.7032,
+      "loss": 0.7856,
       "step": 16269
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.541359283101116e-06,
-      "loss": 0.6741,
+      "loss": 0.7952,
       "step": 16270
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5402175937550215e-06,
-      "loss": 0.7917,
+      "loss": 0.8933,
       "step": 16271
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.539075938534921e-06,
-      "loss": 0.6813,
+      "loss": 0.7974,
       "step": 16272
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.537934317456643e-06,
-      "loss": 0.6612,
+      "loss": 0.7532,
       "step": 16273
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5367927305360415e-06,
-      "loss": 0.7925,
+      "loss": 0.872,
       "step": 16274
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.535651177788941e-06,
-      "loss": 0.6921,
+      "loss": 0.7946,
       "step": 16275
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.534509659231186e-06,
-      "loss": 0.6935,
+      "loss": 0.7998,
       "step": 16276
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.533368174878609e-06,
-      "loss": 0.5809,
+      "loss": 0.7138,
       "step": 16277
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.532226724747051e-06,
-      "loss": 0.6331,
+      "loss": 0.7062,
       "step": 16278
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.531085308852344e-06,
-      "loss": 0.7398,
+      "loss": 0.8122,
       "step": 16279
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5299439272103305e-06,
-      "loss": 0.778,
+      "loss": 0.8905,
       "step": 16280
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.528802579836835e-06,
-      "loss": 0.7149,
+      "loss": 0.8227,
       "step": 16281
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5276612667477e-06,
-      "loss": 0.6191,
+      "loss": 0.7837,
       "step": 16282
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.526519987958756e-06,
-      "loss": 0.7368,
+      "loss": 0.8622,
       "step": 16283
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.525378743485838e-06,
-      "loss": 0.7815,
+      "loss": 0.932,
       "step": 16284
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.524237533344778e-06,
-      "loss": 0.7716,
+      "loss": 0.8445,
       "step": 16285
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.523096357551414e-06,
-      "loss": 0.8726,
+      "loss": 0.902,
       "step": 16286
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.521955216121566e-06,
-      "loss": 0.6699,
+      "loss": 0.7831,
       "step": 16287
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.520814109071082e-06,
-      "loss": 0.6596,
+      "loss": 0.7799,
       "step": 16288
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.519673036415781e-06,
-      "loss": 0.7422,
+      "loss": 0.8276,
       "step": 16289
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.518531998171496e-06,
-      "loss": 0.8216,
+      "loss": 0.8633,
       "step": 16290
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.517390994354063e-06,
-      "loss": 0.6563,
+      "loss": 0.7886,
       "step": 16291
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5162500249793e-06,
-      "loss": 0.7663,
+      "loss": 0.8259,
       "step": 16292
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.515109090063052e-06,
-      "loss": 0.8457,
+      "loss": 0.892,
       "step": 16293
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5139681896211344e-06,
-      "loss": 0.7442,
+      "loss": 0.8092,
       "step": 16294
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5128273236693835e-06,
-      "loss": 0.7939,
+      "loss": 0.8966,
       "step": 16295
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.511686492223624e-06,
-      "loss": 0.7534,
+      "loss": 0.8,
       "step": 16296
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.510545695299683e-06,
-      "loss": 0.5728,
+      "loss": 0.6713,
       "step": 16297
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.509404932913389e-06,
-      "loss": 0.7374,
+      "loss": 0.8039,
       "step": 16298
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.508264205080571e-06,
-      "loss": 0.7416,
+      "loss": 0.7643,
       "step": 16299
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5071235118170445e-06,
-      "loss": 0.6767,
+      "loss": 0.8298,
       "step": 16300
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.50598285313865e-06,
-      "loss": 0.685,
+      "loss": 0.8181,
       "step": 16301
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.504842229061201e-06,
-      "loss": 0.775,
+      "loss": 0.8755,
       "step": 16302
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.503701639600527e-06,
-      "loss": 0.814,
+      "loss": 0.8662,
       "step": 16303
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.50256108477245e-06,
-      "loss": 0.7897,
+      "loss": 0.8518,
       "step": 16304
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5014205645927955e-06,
-      "loss": 0.7355,
+      "loss": 0.8397,
       "step": 16305
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.5002800790773835e-06,
-      "loss": 0.6601,
+      "loss": 0.8025,
       "step": 16306
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.499139628242044e-06,
-      "loss": 0.6468,
+      "loss": 0.8038,
       "step": 16307
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.497999212102591e-06,
-      "loss": 0.7197,
+      "loss": 0.8501,
       "step": 16308
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4968588306748465e-06,
-      "loss": 0.6417,
+      "loss": 0.8609,
       "step": 16309
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.495718483974638e-06,
-      "loss": 0.7442,
+      "loss": 0.8681,
       "step": 16310
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4945781720177794e-06,
-      "loss": 0.5967,
+      "loss": 0.7889,
       "step": 16311
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.493437894820097e-06,
-      "loss": 0.7294,
+      "loss": 0.8522,
       "step": 16312
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.492297652397405e-06,
-      "loss": 0.7652,
+      "loss": 0.8341,
       "step": 16313
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.491157444765527e-06,
-      "loss": 0.817,
+      "loss": 0.9091,
       "step": 16314
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.490017271940278e-06,
-      "loss": 0.7303,
+      "loss": 0.8509,
       "step": 16315
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.488877133937478e-06,
-      "loss": 0.7046,
+      "loss": 0.8241,
       "step": 16316
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.487737030772945e-06,
-      "loss": 0.7046,
+      "loss": 0.8569,
       "step": 16317
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.486596962462498e-06,
-      "loss": 0.6641,
+      "loss": 0.8628,
       "step": 16318
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.485456929021949e-06,
-      "loss": 0.7303,
+      "loss": 0.8878,
       "step": 16319
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4843169304671224e-06,
-      "loss": 0.7392,
+      "loss": 0.8955,
       "step": 16320
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.483176966813825e-06,
-      "loss": 0.6718,
+      "loss": 0.8107,
       "step": 16321
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.482037038077878e-06,
-      "loss": 0.741,
+      "loss": 0.7709,
       "step": 16322
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.480897144275093e-06,
-      "loss": 0.772,
+      "loss": 0.8179,
       "step": 16323
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.479757285421288e-06,
-      "loss": 0.7313,
+      "loss": 0.8091,
       "step": 16324
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4786174615322725e-06,
-      "loss": 0.8162,
+      "loss": 0.9543,
       "step": 16325
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.477477672623867e-06,
-      "loss": 0.7327,
+      "loss": 0.7979,
       "step": 16326
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.476337918711878e-06,
-      "loss": 0.7586,
+      "loss": 0.7903,
       "step": 16327
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.475198199812117e-06,
-      "loss": 0.6146,
+      "loss": 0.7372,
       "step": 16328
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.474058515940401e-06,
-      "loss": 0.7208,
+      "loss": 0.8383,
       "step": 16329
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.472918867112536e-06,
-      "loss": 0.8189,
+      "loss": 0.8083,
       "step": 16330
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.471779253344342e-06,
-      "loss": 0.6437,
+      "loss": 0.7733,
       "step": 16331
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.470639674651616e-06,
-      "loss": 0.7444,
+      "loss": 0.8915,
       "step": 16332
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.469500131050185e-06,
-      "loss": 0.7411,
+      "loss": 0.8704,
       "step": 16333
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4683606225558425e-06,
-      "loss": 0.8126,
+      "loss": 0.8492,
       "step": 16334
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.467221149184407e-06,
-      "loss": 0.8822,
+      "loss": 0.9458,
       "step": 16335
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4660817109516825e-06,
-      "loss": 0.6655,
+      "loss": 0.7929,
       "step": 16336
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.464942307873482e-06,
-      "loss": 0.6623,
+      "loss": 0.8305,
       "step": 16337
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.463802939965608e-06,
-      "loss": 0.6726,
+      "loss": 0.8259,
       "step": 16338
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.462663607243873e-06,
-      "loss": 0.7968,
+      "loss": 0.8137,
       "step": 16339
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.461524309724077e-06,
-      "loss": 0.7792,
+      "loss": 0.847,
       "step": 16340
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.460385047422032e-06,
-      "loss": 0.7137,
+      "loss": 0.8209,
       "step": 16341
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4592458203535374e-06,
-      "loss": 0.7297,
+      "loss": 0.868,
       "step": 16342
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.458106628534406e-06,
-      "loss": 0.7394,
+      "loss": 0.8492,
       "step": 16343
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.456967471980437e-06,
-      "loss": 0.5721,
+      "loss": 0.7556,
       "step": 16344
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.455828350707437e-06,
-      "loss": 0.7846,
+      "loss": 0.8393,
       "step": 16345
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.4546892647312105e-06,
-      "loss": 0.8064,
+      "loss": 0.8821,
       "step": 16346
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.453550214067556e-06,
-      "loss": 0.6982,
+      "loss": 0.8139,
       "step": 16347
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.452411198732281e-06,
-      "loss": 0.7907,
+      "loss": 0.8617,
       "step": 16348
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.451272218741184e-06,
-      "loss": 0.7707,
+      "loss": 0.8836,
       "step": 16349
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.450133274110072e-06,
-      "loss": 0.6828,
+      "loss": 0.8176,
       "step": 16350
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.448994364854741e-06,
-      "loss": 0.7733,
+      "loss": 0.8806,
       "step": 16351
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.447855490990996e-06,
-      "loss": 0.6239,
+      "loss": 0.8278,
       "step": 16352
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.446716652534631e-06,
-      "loss": 0.7819,
+      "loss": 0.8891,
       "step": 16353
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.445577849501453e-06,
-      "loss": 0.7211,
+      "loss": 0.8068,
       "step": 16354
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.444439081907256e-06,
-      "loss": 0.7654,
+      "loss": 0.8654,
       "step": 16355
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.443300349767844e-06,
-      "loss": 0.7838,
+      "loss": 0.8014,
       "step": 16356
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.442161653099008e-06,
-      "loss": 0.7572,
+      "loss": 0.8618,
       "step": 16357
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.441022991916555e-06,
-      "loss": 0.6343,
+      "loss": 0.7604,
       "step": 16358
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.439884366236272e-06,
-      "loss": 0.6236,
+      "loss": 0.776,
       "step": 16359
     },
     {
       "epoch": 0.59,
       "learning_rate": 7.438745776073966e-06,
-      "loss": 0.8306,
+      "loss": 0.9382,
       "step": 16360
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4376072214454245e-06,
-      "loss": 0.7849,
+      "loss": 0.8614,
       "step": 16361
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4364687023664485e-06,
-      "loss": 0.7321,
+      "loss": 0.8457,
       "step": 16362
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.435330218852832e-06,
-      "loss": 0.7049,
+      "loss": 0.751,
       "step": 16363
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.43419177092037e-06,
-      "loss": 0.6726,
+      "loss": 0.7488,
       "step": 16364
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.433053358584861e-06,
-      "loss": 0.7274,
+      "loss": 0.8461,
       "step": 16365
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.431914981862088e-06,
-      "loss": 0.5411,
+      "loss": 0.7545,
       "step": 16366
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.430776640767854e-06,
-      "loss": 0.7696,
+      "loss": 0.8364,
       "step": 16367
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.429638335317948e-06,
-      "loss": 0.7222,
+      "loss": 0.8338,
       "step": 16368
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.428500065528164e-06,
-      "loss": 0.8241,
+      "loss": 0.878,
       "step": 16369
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.427361831414292e-06,
-      "loss": 0.7803,
+      "loss": 0.9121,
       "step": 16370
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4262236329921275e-06,
-      "loss": 0.7441,
+      "loss": 0.8602,
       "step": 16371
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4250854702774536e-06,
-      "loss": 0.671,
+      "loss": 0.8179,
       "step": 16372
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.423947343286071e-06,
-      "loss": 0.6208,
+      "loss": 0.8047,
       "step": 16373
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.422809252033758e-06,
-      "loss": 0.8076,
+      "loss": 0.912,
       "step": 16374
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4216711965363156e-06,
-      "loss": 0.6612,
+      "loss": 0.7758,
       "step": 16375
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4205331768095225e-06,
-      "loss": 0.8089,
+      "loss": 0.9087,
       "step": 16376
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.419395192869174e-06,
-      "loss": 0.6204,
+      "loss": 0.7134,
       "step": 16377
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.418257244731056e-06,
-      "loss": 0.7987,
+      "loss": 0.8756,
       "step": 16378
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.417119332410958e-06,
-      "loss": 0.7976,
+      "loss": 0.8183,
       "step": 16379
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.415981455924661e-06,
-      "loss": 0.8157,
+      "loss": 0.9056,
       "step": 16380
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4148436152879565e-06,
-      "loss": 0.7292,
+      "loss": 0.8223,
       "step": 16381
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.413705810516627e-06,
-      "loss": 0.7917,
+      "loss": 0.8609,
       "step": 16382
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.412568041626463e-06,
-      "loss": 0.5474,
+      "loss": 0.7462,
       "step": 16383
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.411430308633247e-06,
-      "loss": 0.6632,
+      "loss": 0.7236,
       "step": 16384
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.41029261155276e-06,
-      "loss": 0.7933,
+      "loss": 0.8324,
       "step": 16385
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.40915495040079e-06,
-      "loss": 0.8403,
+      "loss": 0.894,
       "step": 16386
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.408017325193118e-06,
-      "loss": 0.7396,
+      "loss": 0.8701,
       "step": 16387
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.406879735945529e-06,
-      "loss": 0.739,
+      "loss": 0.8074,
       "step": 16388
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.405742182673803e-06,
-      "loss": 0.6238,
+      "loss": 0.738,
       "step": 16389
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.404604665393724e-06,
-      "loss": 0.6941,
+      "loss": 0.8122,
       "step": 16390
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.403467184121072e-06,
-      "loss": 0.7644,
+      "loss": 0.7801,
       "step": 16391
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4023297388716345e-06,
-      "loss": 0.6095,
+      "loss": 0.8109,
       "step": 16392
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.4011923296611795e-06,
-      "loss": 0.7229,
+      "loss": 0.8304,
       "step": 16393
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.400054956505498e-06,
-      "loss": 0.8451,
+      "loss": 0.8593,
       "step": 16394
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.398917619420361e-06,
-      "loss": 0.714,
+      "loss": 0.8217,
       "step": 16395
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.397780318421554e-06,
-      "loss": 0.7051,
+      "loss": 0.8344,
       "step": 16396
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.396643053524852e-06,
-      "loss": 0.7237,
+      "loss": 0.7661,
       "step": 16397
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.395505824746038e-06,
-      "loss": 0.7831,
+      "loss": 0.8748,
       "step": 16398
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.394368632100879e-06,
-      "loss": 0.7264,
+      "loss": 0.8584,
       "step": 16399
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.393231475605165e-06,
-      "loss": 0.6836,
+      "loss": 0.851,
       "step": 16400
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.392094355274661e-06,
-      "loss": 0.7659,
+      "loss": 0.7891,
       "step": 16401
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.39095727112515e-06,
-      "loss": 0.6609,
+      "loss": 0.7915,
       "step": 16402
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.389820223172407e-06,
-      "loss": 0.6435,
+      "loss": 0.8173,
       "step": 16403
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3886832114322e-06,
-      "loss": 0.8064,
+      "loss": 0.8871,
       "step": 16404
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.387546235920312e-06,
-      "loss": 0.645,
+      "loss": 0.7746,
       "step": 16405
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.38640929665251e-06,
-      "loss": 0.629,
+      "loss": 0.802,
       "step": 16406
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.385272393644575e-06,
-      "loss": 0.8213,
+      "loss": 0.8559,
       "step": 16407
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.384135526912272e-06,
-      "loss": 0.7322,
+      "loss": 0.8759,
       "step": 16408
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3829986964713796e-06,
-      "loss": 0.5908,
+      "loss": 0.787,
       "step": 16409
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.381861902337665e-06,
-      "loss": 0.8328,
+      "loss": 0.8691,
       "step": 16410
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.380725144526905e-06,
-      "loss": 0.8447,
+      "loss": 0.9123,
       "step": 16411
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.379588423054863e-06,
-      "loss": 0.8167,
+      "loss": 0.9127,
       "step": 16412
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.37845173793732e-06,
-      "loss": 0.7734,
+      "loss": 0.9232,
       "step": 16413
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.377315089190033e-06,
-      "loss": 0.8585,
+      "loss": 0.8613,
       "step": 16414
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.376178476828784e-06,
-      "loss": 0.7781,
+      "loss": 0.7526,
       "step": 16415
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.375041900869331e-06,
-      "loss": 0.6785,
+      "loss": 0.7967,
       "step": 16416
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3739053613274495e-06,
-      "loss": 0.7516,
+      "loss": 0.804,
       "step": 16417
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.372768858218904e-06,
-      "loss": 0.5144,
+      "loss": 0.7361,
       "step": 16418
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.371632391559467e-06,
-      "loss": 0.6237,
+      "loss": 0.8528,
       "step": 16419
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.370495961364896e-06,
-      "loss": 0.7232,
+      "loss": 0.8739,
       "step": 16420
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3693595676509665e-06,
-      "loss": 0.7199,
+      "loss": 0.8331,
       "step": 16421
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3682232104334405e-06,
-      "loss": 0.7427,
+      "loss": 0.9189,
       "step": 16422
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.36708688972808e-06,
-      "loss": 0.7055,
+      "loss": 0.7616,
       "step": 16423
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.365950605550658e-06,
-      "loss": 0.6967,
+      "loss": 0.8315,
       "step": 16424
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.364814357916931e-06,
-      "loss": 0.7655,
+      "loss": 0.9032,
       "step": 16425
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.363678146842668e-06,
-      "loss": 0.6778,
+      "loss": 0.8527,
       "step": 16426
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3625419723436265e-06,
-      "loss": 0.776,
+      "loss": 0.913,
       "step": 16427
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.361405834435576e-06,
-      "loss": 0.7171,
+      "loss": 0.8225,
       "step": 16428
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.360269733134273e-06,
-      "loss": 0.7028,
+      "loss": 0.7824,
       "step": 16429
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.359133668455488e-06,
-      "loss": 0.8787,
+      "loss": 0.919,
       "step": 16430
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.357997640414969e-06,
-      "loss": 0.6754,
+      "loss": 0.8303,
       "step": 16431
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3568616490284916e-06,
-      "loss": 0.6769,
+      "loss": 0.7818,
       "step": 16432
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.355725694311804e-06,
-      "loss": 0.7614,
+      "loss": 0.854,
       "step": 16433
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.354589776280672e-06,
-      "loss": 0.6033,
+      "loss": 0.7181,
       "step": 16434
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.353453894950853e-06,
-      "loss": 0.6686,
+      "loss": 0.7436,
       "step": 16435
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3523180503381075e-06,
-      "loss": 0.7183,
+      "loss": 0.8162,
       "step": 16436
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.35118224245819e-06,
-      "loss": 0.7555,
+      "loss": 0.8508,
       "step": 16437
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3500464713268646e-06,
-      "loss": 0.769,
+      "loss": 0.8564,
       "step": 16438
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.348910736959879e-06,
-      "loss": 0.922,
+      "loss": 0.8739,
       "step": 16439
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.347775039373002e-06,
-      "loss": 0.6122,
+      "loss": 0.7088,
       "step": 16440
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.346639378581979e-06,
-      "loss": 0.7034,
+      "loss": 0.7936,
       "step": 16441
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.345503754602572e-06,
-      "loss": 0.827,
+      "loss": 0.8938,
       "step": 16442
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.344368167450535e-06,
-      "loss": 0.8422,
+      "loss": 0.8679,
       "step": 16443
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.343232617141618e-06,
-      "loss": 0.7174,
+      "loss": 0.784,
       "step": 16444
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3420971036915835e-06,
-      "loss": 0.8194,
+      "loss": 0.8618,
       "step": 16445
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.340961627116177e-06,
-      "loss": 0.7131,
+      "loss": 0.8533,
       "step": 16446
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.339826187431156e-06,
-      "loss": 0.8093,
+      "loss": 0.8493,
       "step": 16447
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3386907846522714e-06,
-      "loss": 0.7373,
+      "loss": 0.8503,
       "step": 16448
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.337555418795279e-06,
-      "loss": 0.7739,
+      "loss": 0.8665,
       "step": 16449
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.336420089875925e-06,
-      "loss": 0.7115,
+      "loss": 0.793,
       "step": 16450
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.335284797909967e-06,
-      "loss": 0.7244,
+      "loss": 0.7611,
       "step": 16451
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.334149542913148e-06,
-      "loss": 0.7788,
+      "loss": 0.8564,
       "step": 16452
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.333014324901224e-06,
-      "loss": 0.716,
+      "loss": 0.8458,
       "step": 16453
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.33187914388994e-06,
-      "loss": 0.7751,
+      "loss": 0.8853,
       "step": 16454
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3307439998950496e-06,
-      "loss": 0.7605,
+      "loss": 0.8354,
       "step": 16455
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.329608892932296e-06,
-      "loss": 0.7778,
+      "loss": 0.832,
       "step": 16456
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.328473823017434e-06,
-      "loss": 0.8366,
+      "loss": 0.8966,
       "step": 16457
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.327338790166203e-06,
-      "loss": 0.6068,
+      "loss": 0.7232,
       "step": 16458
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.32620379439436e-06,
-      "loss": 0.8007,
+      "loss": 0.9042,
       "step": 16459
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.325068835717641e-06,
-      "loss": 0.789,
+      "loss": 0.8699,
       "step": 16460
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.323933914151799e-06,
-      "loss": 0.7791,
+      "loss": 0.8509,
       "step": 16461
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.322799029712578e-06,
-      "loss": 0.7532,
+      "loss": 0.7994,
       "step": 16462
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.3216641824157174e-06,
-      "loss": 0.875,
+      "loss": 0.8618,
       "step": 16463
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.320529372276972e-06,
-      "loss": 0.7244,
+      "loss": 0.7639,
       "step": 16464
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.319394599312075e-06,
-      "loss": 0.7423,
+      "loss": 0.8192,
       "step": 16465
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.318259863536779e-06,
-      "loss": 0.7481,
+      "loss": 0.838,
       "step": 16466
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.317125164966819e-06,
-      "loss": 0.6799,
+      "loss": 0.7673,
       "step": 16467
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.315990503617941e-06,
-      "loss": 0.7408,
+      "loss": 0.7974,
       "step": 16468
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.314855879505887e-06,
-      "loss": 0.7762,
+      "loss": 0.8826,
       "step": 16469
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.313721292646401e-06,
-      "loss": 0.6077,
+      "loss": 0.7706,
       "step": 16470
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.312586743055215e-06,
-      "loss": 0.7203,
+      "loss": 0.8181,
       "step": 16471
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.311452230748082e-06,
-      "loss": 0.7581,
+      "loss": 0.879,
       "step": 16472
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.310317755740728e-06,
-      "loss": 0.7506,
+      "loss": 0.793,
       "step": 16473
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.309183318048903e-06,
-      "loss": 0.6033,
+      "loss": 0.8264,
       "step": 16474
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.308048917688339e-06,
-      "loss": 0.6442,
+      "loss": 0.7876,
       "step": 16475
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.30691455467478e-06,
-      "loss": 0.7446,
+      "loss": 0.7841,
       "step": 16476
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.305780229023958e-06,
-      "loss": 0.7276,
+      "loss": 0.8655,
       "step": 16477
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.304645940751616e-06,
-      "loss": 0.71,
+      "loss": 0.8208,
       "step": 16478
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.303511689873484e-06,
-      "loss": 0.7709,
+      "loss": 0.8613,
       "step": 16479
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.302377476405303e-06,
-      "loss": 0.6134,
+      "loss": 0.6884,
       "step": 16480
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.301243300362808e-06,
-      "loss": 0.7887,
+      "loss": 0.8788,
       "step": 16481
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.30010916176173e-06,
-      "loss": 0.5727,
+      "loss": 0.7238,
       "step": 16482
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.298975060617812e-06,
-      "loss": 0.6988,
+      "loss": 0.8317,
       "step": 16483
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.297840996946777e-06,
-      "loss": 0.8366,
+      "loss": 0.9261,
       "step": 16484
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.296706970764369e-06,
-      "loss": 0.7422,
+      "loss": 0.8101,
       "step": 16485
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.295572982086313e-06,
-      "loss": 0.8103,
+      "loss": 0.9138,
       "step": 16486
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.294439030928347e-06,
-      "loss": 0.6515,
+      "loss": 0.7757,
       "step": 16487
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.293305117306197e-06,
-      "loss": 0.4929,
+      "loss": 0.7144,
       "step": 16488
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.292171241235603e-06,
-      "loss": 0.8231,
+      "loss": 0.885,
       "step": 16489
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.291037402732287e-06,
-      "loss": 0.8406,
+      "loss": 0.905,
       "step": 16490
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.289903601811989e-06,
-      "loss": 0.8862,
+      "loss": 0.9258,
       "step": 16491
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.288769838490428e-06,
-      "loss": 0.7009,
+      "loss": 0.7705,
       "step": 16492
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.28763611278334e-06,
-      "loss": 0.6658,
+      "loss": 0.7911,
       "step": 16493
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.286502424706453e-06,
-      "loss": 0.7573,
+      "loss": 0.8451,
       "step": 16494
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.285368774275497e-06,
-      "loss": 0.6552,
+      "loss": 0.7722,
       "step": 16495
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.284235161506193e-06,
-      "loss": 0.6718,
+      "loss": 0.804,
       "step": 16496
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.28310158641428e-06,
-      "loss": 0.7535,
+      "loss": 0.8335,
       "step": 16497
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.281968049015469e-06,
-      "loss": 0.7343,
+      "loss": 0.8756,
       "step": 16498
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.280834549325503e-06,
-      "loss": 0.8075,
+      "loss": 0.8505,
       "step": 16499
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.279701087360096e-06,
-      "loss": 0.6937,
+      "loss": 0.8641,
       "step": 16500
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.278567663134977e-06,
-      "loss": 0.8263,
+      "loss": 0.8787,
       "step": 16501
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.277434276665872e-06,
-      "loss": 0.7619,
+      "loss": 0.8847,
       "step": 16502
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.276300927968498e-06,
-      "loss": 0.7443,
+      "loss": 0.8592,
       "step": 16503
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.275167617058592e-06,
-      "loss": 0.7131,
+      "loss": 0.7821,
       "step": 16504
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.274034343951863e-06,
-      "loss": 0.8073,
+      "loss": 0.8968,
       "step": 16505
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.272901108664045e-06,
-      "loss": 0.7203,
+      "loss": 0.7777,
       "step": 16506
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.27176791121085e-06,
-      "loss": 0.819,
+      "loss": 0.9313,
       "step": 16507
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.270634751608008e-06,
-      "loss": 0.6614,
+      "loss": 0.733,
       "step": 16508
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.269501629871233e-06,
-      "loss": 0.7682,
+      "loss": 0.8777,
       "step": 16509
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2683685460162535e-06,
-      "loss": 0.7462,
+      "loss": 0.8616,
       "step": 16510
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.267235500058778e-06,
-      "loss": 0.724,
+      "loss": 0.8912,
       "step": 16511
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.266102492014539e-06,
-      "loss": 0.8331,
+      "loss": 0.9261,
       "step": 16512
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2649695218992455e-06,
-      "loss": 0.5372,
+      "loss": 0.7964,
       "step": 16513
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.263836589728622e-06,
-      "loss": 0.8147,
+      "loss": 0.8333,
       "step": 16514
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.26270369551838e-06,
-      "loss": 0.6081,
+      "loss": 0.7508,
       "step": 16515
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.261570839284245e-06,
-      "loss": 0.6408,
+      "loss": 0.8044,
       "step": 16516
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2604380210419245e-06,
-      "loss": 0.8286,
+      "loss": 0.8689,
       "step": 16517
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.259305240807147e-06,
-      "loss": 0.8773,
+      "loss": 0.9253,
       "step": 16518
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.258172498595615e-06,
-      "loss": 0.6921,
+      "loss": 0.8228,
       "step": 16519
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.257039794423048e-06,
-      "loss": 0.6567,
+      "loss": 0.8037,
       "step": 16520
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.255907128305166e-06,
-      "loss": 0.8312,
+      "loss": 0.9088,
       "step": 16521
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2547745002576765e-06,
-      "loss": 0.7243,
+      "loss": 0.8519,
       "step": 16522
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2536419102963e-06,
-      "loss": 0.6293,
+      "loss": 0.7517,
       "step": 16523
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.252509358436741e-06,
-      "loss": 0.7437,
+      "loss": 0.8059,
       "step": 16524
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.251376844694718e-06,
-      "loss": 0.7526,
+      "loss": 0.8508,
       "step": 16525
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.25024436908594e-06,
-      "loss": 0.7968,
+      "loss": 0.8762,
       "step": 16526
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2491119316261225e-06,
-      "loss": 0.7413,
+      "loss": 0.8651,
       "step": 16527
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.247979532330971e-06,
-      "loss": 0.7428,
+      "loss": 0.8535,
       "step": 16528
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.246847171216203e-06,
-      "loss": 0.9131,
+      "loss": 0.9973,
       "step": 16529
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.245714848297519e-06,
-      "loss": 0.726,
+      "loss": 0.8466,
       "step": 16530
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.24458256359064e-06,
-      "loss": 0.665,
+      "loss": 0.7571,
       "step": 16531
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.243450317111262e-06,
-      "loss": 0.7207,
+      "loss": 0.8638,
       "step": 16532
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.242318108875105e-06,
-      "loss": 0.7517,
+      "loss": 0.8352,
       "step": 16533
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2411859388978675e-06,
-      "loss": 0.6753,
+      "loss": 0.8121,
       "step": 16534
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.240053807195264e-06,
-      "loss": 0.7357,
+      "loss": 0.8361,
       "step": 16535
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.238921713782996e-06,
-      "loss": 0.7078,
+      "loss": 0.8117,
       "step": 16536
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.237789658676775e-06,
-      "loss": 0.7912,
+      "loss": 0.8491,
       "step": 16537
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2366576418923014e-06,
-      "loss": 0.7851,
+      "loss": 0.8123,
       "step": 16538
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.235525663445282e-06,
-      "loss": 0.7383,
+      "loss": 0.8436,
       "step": 16539
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.234393723351423e-06,
-      "loss": 0.6716,
+      "loss": 0.7337,
       "step": 16540
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.233261821626424e-06,
-      "loss": 0.7424,
+      "loss": 0.7871,
       "step": 16541
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.232129958285996e-06,
-      "loss": 0.6593,
+      "loss": 0.8403,
       "step": 16542
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.230998133345832e-06,
-      "loss": 0.7651,
+      "loss": 0.8478,
       "step": 16543
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.229866346821647e-06,
-      "loss": 0.8192,
+      "loss": 0.8946,
       "step": 16544
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.228734598729131e-06,
-      "loss": 0.7503,
+      "loss": 0.8057,
       "step": 16545
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.227602889083993e-06,
-      "loss": 0.755,
+      "loss": 0.8788,
       "step": 16546
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2264712179019294e-06,
-      "loss": 0.715,
+      "loss": 0.8591,
       "step": 16547
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.225339585198645e-06,
-      "loss": 0.7796,
+      "loss": 0.8694,
       "step": 16548
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.224207990989836e-06,
-      "loss": 0.7048,
+      "loss": 0.732,
       "step": 16549
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2230764352912045e-06,
-      "loss": 0.6148,
+      "loss": 0.8156,
       "step": 16550
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.221944918118444e-06,
-      "loss": 0.7186,
+      "loss": 0.8659,
       "step": 16551
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.22081343948726e-06,
-      "loss": 0.7982,
+      "loss": 0.8619,
       "step": 16552
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.219681999413341e-06,
-      "loss": 0.6531,
+      "loss": 0.8518,
       "step": 16553
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.218550597912395e-06,
-      "loss": 0.7592,
+      "loss": 0.8623,
       "step": 16554
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.217419235000109e-06,
-      "loss": 0.7639,
+      "loss": 0.8104,
       "step": 16555
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.216287910692185e-06,
-      "loss": 0.6385,
+      "loss": 0.8422,
       "step": 16556
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.21515662500432e-06,
-      "loss": 0.7307,
+      "loss": 0.8663,
       "step": 16557
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.214025377952203e-06,
-      "loss": 0.7013,
+      "loss": 0.903,
       "step": 16558
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.21289416955153e-06,
-      "loss": 0.7701,
+      "loss": 0.8775,
       "step": 16559
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.211762999817995e-06,
-      "loss": 0.8423,
+      "loss": 0.9254,
       "step": 16560
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.210631868767295e-06,
-      "loss": 0.6648,
+      "loss": 0.8149,
       "step": 16561
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.209500776415118e-06,
-      "loss": 0.6803,
+      "loss": 0.7577,
       "step": 16562
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.208369722777163e-06,
-      "loss": 0.727,
+      "loss": 0.8172,
       "step": 16563
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.207238707869113e-06,
-      "loss": 0.8304,
+      "loss": 0.9495,
       "step": 16564
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.206107731706664e-06,
-      "loss": 0.7785,
+      "loss": 0.8411,
       "step": 16565
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.204976794305506e-06,
-      "loss": 0.7217,
+      "loss": 0.8509,
       "step": 16566
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.203845895681332e-06,
-      "loss": 0.7265,
+      "loss": 0.8474,
       "step": 16567
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.202715035849826e-06,
-      "loss": 0.7329,
+      "loss": 0.9036,
       "step": 16568
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.2015842148266825e-06,
-      "loss": 0.8257,
+      "loss": 0.887,
       "step": 16569
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.200453432627583e-06,
-      "loss": 0.79,
+      "loss": 0.8652,
       "step": 16570
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.199322689268227e-06,
-      "loss": 0.7186,
+      "loss": 0.802,
       "step": 16571
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1981919847642885e-06,
-      "loss": 0.7684,
+      "loss": 0.8503,
       "step": 16572
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1970613191314655e-06,
-      "loss": 0.8543,
+      "loss": 0.9528,
       "step": 16573
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.195930692385436e-06,
-      "loss": 0.6839,
+      "loss": 0.7049,
       "step": 16574
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.194800104541892e-06,
-      "loss": 0.7186,
+      "loss": 0.8023,
       "step": 16575
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.193669555616521e-06,
-      "loss": 0.7204,
+      "loss": 0.8629,
       "step": 16576
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.192539045624996e-06,
-      "loss": 0.7227,
+      "loss": 0.8548,
       "step": 16577
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.191408574583012e-06,
-      "loss": 0.7789,
+      "loss": 0.8113,
       "step": 16578
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.190278142506247e-06,
-      "loss": 0.7049,
+      "loss": 0.8285,
       "step": 16579
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.189147749410388e-06,
-      "loss": 0.7102,
+      "loss": 0.8365,
       "step": 16580
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.188017395311115e-06,
-      "loss": 0.7164,
+      "loss": 0.8276,
       "step": 16581
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.186887080224114e-06,
-      "loss": 0.7257,
+      "loss": 0.811,
       "step": 16582
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.185756804165057e-06,
-      "loss": 0.587,
+      "loss": 0.7333,
       "step": 16583
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.184626567149639e-06,
-      "loss": 0.6456,
+      "loss": 0.7288,
       "step": 16584
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.183496369193528e-06,
-      "loss": 0.8108,
+      "loss": 0.887,
       "step": 16585
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.18236621031241e-06,
-      "loss": 0.8032,
+      "loss": 0.9184,
       "step": 16586
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.181236090521961e-06,
-      "loss": 0.6897,
+      "loss": 0.8411,
       "step": 16587
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.180106009837866e-06,
-      "loss": 0.6859,
+      "loss": 0.7998,
       "step": 16588
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.178975968275795e-06,
-      "loss": 0.8451,
+      "loss": 0.9737,
       "step": 16589
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.177845965851436e-06,
-      "loss": 0.8772,
+      "loss": 0.8985,
       "step": 16590
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.176716002580455e-06,
-      "loss": 0.8063,
+      "loss": 0.8948,
       "step": 16591
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1755860784785356e-06,
-      "loss": 0.7021,
+      "loss": 0.8156,
       "step": 16592
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.17445619356135e-06,
-      "loss": 0.6763,
+      "loss": 0.802,
       "step": 16593
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.173326347844578e-06,
-      "loss": 0.7024,
+      "loss": 0.8223,
       "step": 16594
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.172196541343894e-06,
-      "loss": 0.7834,
+      "loss": 0.8449,
       "step": 16595
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.171066774074967e-06,
-      "loss": 0.7366,
+      "loss": 0.8154,
       "step": 16596
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.169937046053478e-06,
-      "loss": 0.7049,
+      "loss": 0.8815,
       "step": 16597
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.168807357295094e-06,
-      "loss": 0.7456,
+      "loss": 0.8533,
       "step": 16598
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1676777078154925e-06,
-      "loss": 0.6243,
+      "loss": 0.781,
       "step": 16599
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1665480976303435e-06,
-      "loss": 0.6706,
+      "loss": 0.8253,
       "step": 16600
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1654185267553235e-06,
-      "loss": 0.7806,
+      "loss": 0.8782,
       "step": 16601
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.164288995206093e-06,
-      "loss": 0.7639,
+      "loss": 0.864,
       "step": 16602
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.163159502998334e-06,
-      "loss": 0.7643,
+      "loss": 0.8372,
       "step": 16603
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.16203005014771e-06,
-      "loss": 0.8195,
+      "loss": 0.811,
       "step": 16604
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.160900636669894e-06,
-      "loss": 0.6449,
+      "loss": 0.7688,
       "step": 16605
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.159771262580552e-06,
-      "loss": 0.8001,
+      "loss": 0.8327,
       "step": 16606
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.158641927895356e-06,
-      "loss": 0.7758,
+      "loss": 0.8442,
       "step": 16607
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.157512632629968e-06,
-      "loss": 0.695,
+      "loss": 0.7875,
       "step": 16608
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.156383376800065e-06,
-      "loss": 0.7999,
+      "loss": 0.8901,
       "step": 16609
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.155254160421301e-06,
-      "loss": 0.6162,
+      "loss": 0.8091,
       "step": 16610
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.154124983509356e-06,
-      "loss": 0.6513,
+      "loss": 0.8055,
       "step": 16611
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.152995846079886e-06,
-      "loss": 0.7559,
+      "loss": 0.8271,
       "step": 16612
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.15186674814856e-06,
-      "loss": 0.7067,
+      "loss": 0.8718,
       "step": 16613
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.150737689731044e-06,
-      "loss": 0.7942,
+      "loss": 0.8526,
       "step": 16614
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.149608670842995e-06,
-      "loss": 0.6937,
+      "loss": 0.8114,
       "step": 16615
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.148479691500086e-06,
-      "loss": 0.7203,
+      "loss": 0.8214,
       "step": 16616
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1473507517179715e-06,
-      "loss": 0.6744,
+      "loss": 0.8164,
       "step": 16617
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.146221851512322e-06,
-      "loss": 0.7501,
+      "loss": 0.8151,
       "step": 16618
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.145092990898792e-06,
-      "loss": 0.6359,
+      "loss": 0.8586,
       "step": 16619
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.143964169893048e-06,
-      "loss": 0.6977,
+      "loss": 0.8388,
       "step": 16620
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.142835388510747e-06,
-      "loss": 0.7791,
+      "loss": 0.8584,
       "step": 16621
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.141706646767555e-06,
-      "loss": 0.8222,
+      "loss": 0.9101,
       "step": 16622
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.140577944679124e-06,
-      "loss": 0.7747,
+      "loss": 0.9129,
       "step": 16623
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1394492822611195e-06,
-      "loss": 0.6868,
+      "loss": 0.8866,
       "step": 16624
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.138320659529195e-06,
-      "loss": 0.7637,
+      "loss": 0.7878,
       "step": 16625
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.137192076499015e-06,
-      "loss": 0.6545,
+      "loss": 0.7472,
       "step": 16626
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.136063533186229e-06,
-      "loss": 0.7297,
+      "loss": 0.8564,
       "step": 16627
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1349350296065e-06,
-      "loss": 0.6988,
+      "loss": 0.8845,
       "step": 16628
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.133806565775482e-06,
-      "loss": 0.715,
+      "loss": 0.838,
       "step": 16629
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.132678141708834e-06,
-      "loss": 0.6754,
+      "loss": 0.8365,
       "step": 16630
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.131549757422205e-06,
-      "loss": 0.7292,
+      "loss": 0.8769,
       "step": 16631
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.1304214129312565e-06,
-      "loss": 0.7587,
+      "loss": 0.8562,
       "step": 16632
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.129293108251638e-06,
-      "loss": 0.7504,
+      "loss": 0.8525,
       "step": 16633
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.128164843399004e-06,
-      "loss": 0.8041,
+      "loss": 0.8926,
       "step": 16634
     },
     {
       "epoch": 0.6,
       "learning_rate": 7.12703661838901e-06,
-      "loss": 0.6738,
+      "loss": 0.7885,
       "step": 16635
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.125908433237304e-06,
-      "loss": 0.7308,
+      "loss": 0.8355,
       "step": 16636
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.124780287959543e-06,
-      "loss": 0.5934,
+      "loss": 0.7816,
       "step": 16637
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.123652182571373e-06,
-      "loss": 0.6417,
+      "loss": 0.835,
       "step": 16638
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.1225241170884515e-06,
-      "loss": 0.6355,
+      "loss": 0.8121,
       "step": 16639
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.121396091526422e-06,
-      "loss": 0.715,
+      "loss": 0.7661,
       "step": 16640
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.1202681059009424e-06,
-      "loss": 0.7215,
+      "loss": 0.8166,
       "step": 16641
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.119140160227651e-06,
-      "loss": 0.6094,
+      "loss": 0.7419,
       "step": 16642
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.118012254522208e-06,
-      "loss": 0.6986,
+      "loss": 0.829,
       "step": 16643
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.1168843888002514e-06,
-      "loss": 0.71,
+      "loss": 0.7791,
       "step": 16644
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.115756563077436e-06,
-      "loss": 0.5955,
+      "loss": 0.7897,
       "step": 16645
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.114628777369405e-06,
-      "loss": 0.7908,
+      "loss": 0.8467,
       "step": 16646
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.113501031691806e-06,
-      "loss": 0.7978,
+      "loss": 0.8902,
       "step": 16647
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.112373326060285e-06,
-      "loss": 0.8375,
+      "loss": 0.8426,
       "step": 16648
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.111245660490489e-06,
-      "loss": 0.6947,
+      "loss": 0.8346,
       "step": 16649
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.110118034998055e-06,
-      "loss": 0.7731,
+      "loss": 0.8538,
       "step": 16650
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.108990449598641e-06,
-      "loss": 0.8217,
+      "loss": 0.9053,
       "step": 16651
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.107862904307878e-06,
-      "loss": 0.6805,
+      "loss": 0.7717,
       "step": 16652
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.106735399141413e-06,
-      "loss": 0.8228,
+      "loss": 0.9117,
       "step": 16653
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.105607934114895e-06,
-      "loss": 0.8108,
+      "loss": 0.8358,
       "step": 16654
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.104480509243952e-06,
-      "loss": 0.6198,
+      "loss": 0.7618,
       "step": 16655
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.103353124544241e-06,
-      "loss": 0.8082,
+      "loss": 0.8579,
       "step": 16656
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.102225780031391e-06,
-      "loss": 0.7982,
+      "loss": 0.876,
       "step": 16657
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.101098475721049e-06,
-      "loss": 0.6695,
+      "loss": 0.7979,
       "step": 16658
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.099971211628851e-06,
-      "loss": 0.915,
+      "loss": 0.9426,
       "step": 16659
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.098843987770439e-06,
-      "loss": 0.7467,
+      "loss": 0.8848,
       "step": 16660
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.09771680416145e-06,
-      "loss": 0.7393,
+      "loss": 0.8903,
       "step": 16661
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.096589660817526e-06,
-      "loss": 0.7745,
+      "loss": 0.8362,
       "step": 16662
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.095462557754296e-06,
-      "loss": 0.831,
+      "loss": 0.9172,
       "step": 16663
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.094335494987405e-06,
-      "loss": 0.7371,
+      "loss": 0.8695,
       "step": 16664
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.093208472532484e-06,
-      "loss": 0.6736,
+      "loss": 0.7649,
       "step": 16665
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.092081490405174e-06,
-      "loss": 0.615,
+      "loss": 0.8063,
       "step": 16666
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.090954548621104e-06,
-      "loss": 0.7653,
+      "loss": 0.7939,
       "step": 16667
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.089827647195917e-06,
-      "loss": 0.8411,
+      "loss": 0.9048,
       "step": 16668
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.088700786145237e-06,
-      "loss": 0.7381,
+      "loss": 0.7663,
       "step": 16669
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.087573965484709e-06,
-      "loss": 0.7206,
+      "loss": 0.8371,
       "step": 16670
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.086447185229954e-06,
-      "loss": 0.6313,
+      "loss": 0.7568,
       "step": 16671
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.085320445396615e-06,
-      "loss": 0.6259,
+      "loss": 0.7056,
       "step": 16672
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.084193746000321e-06,
-      "loss": 0.8474,
+      "loss": 0.9513,
       "step": 16673
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.083067087056694e-06,
-      "loss": 0.7821,
+      "loss": 0.833,
       "step": 16674
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.081940468581381e-06,
-      "loss": 0.7864,
+      "loss": 0.9145,
       "step": 16675
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0808138905899985e-06,
-      "loss": 0.6857,
+      "loss": 0.8071,
       "step": 16676
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.079687353098183e-06,
-      "loss": 0.7287,
+      "loss": 0.782,
       "step": 16677
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.078560856121562e-06,
-      "loss": 0.8008,
+      "loss": 0.8577,
       "step": 16678
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.077434399675765e-06,
-      "loss": 0.6385,
+      "loss": 0.7332,
       "step": 16679
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.076307983776416e-06,
-      "loss": 0.7814,
+      "loss": 0.8925,
       "step": 16680
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.07518160843915e-06,
-      "loss": 0.768,
+      "loss": 0.9016,
       "step": 16681
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.074055273679584e-06,
-      "loss": 0.7512,
+      "loss": 0.8898,
       "step": 16682
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.072928979513356e-06,
-      "loss": 0.7645,
+      "loss": 0.8613,
       "step": 16683
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.071802725956081e-06,
-      "loss": 0.7158,
+      "loss": 0.8261,
       "step": 16684
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.070676513023391e-06,
-      "loss": 0.6899,
+      "loss": 0.799,
       "step": 16685
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0695503407309055e-06,
-      "loss": 0.7109,
+      "loss": 0.7906,
       "step": 16686
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.068424209094254e-06,
-      "loss": 0.7674,
+      "loss": 0.8851,
       "step": 16687
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.067298118129054e-06,
-      "loss": 0.6879,
+      "loss": 0.8215,
       "step": 16688
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.066172067850937e-06,
-      "loss": 0.9355,
+      "loss": 0.9252,
       "step": 16689
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.065046058275515e-06,
-      "loss": 0.6011,
+      "loss": 0.7429,
       "step": 16690
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.063920089418418e-06,
-      "loss": 0.6487,
+      "loss": 0.7881,
       "step": 16691
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.062794161295264e-06,
-      "loss": 0.8236,
+      "loss": 0.9354,
       "step": 16692
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.06166827392167e-06,
-      "loss": 0.7141,
+      "loss": 0.8112,
       "step": 16693
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.060542427313266e-06,
-      "loss": 0.7003,
+      "loss": 0.7896,
       "step": 16694
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.05941662148566e-06,
-      "loss": 0.7232,
+      "loss": 0.8025,
       "step": 16695
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.058290856454477e-06,
-      "loss": 0.6883,
+      "loss": 0.7981,
       "step": 16696
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0571651322353355e-06,
-      "loss": 0.8598,
+      "loss": 0.8829,
       "step": 16697
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.056039448843852e-06,
-      "loss": 0.7704,
+      "loss": 0.8305,
       "step": 16698
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.054913806295643e-06,
-      "loss": 0.7129,
+      "loss": 0.8354,
       "step": 16699
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.053788204606329e-06,
-      "loss": 0.6622,
+      "loss": 0.7429,
       "step": 16700
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0526626437915205e-06,
-      "loss": 0.6947,
+      "loss": 0.859,
       "step": 16701
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.05153712386684e-06,
-      "loss": 0.6935,
+      "loss": 0.867,
       "step": 16702
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.050411644847897e-06,
-      "loss": 0.7177,
+      "loss": 0.8649,
       "step": 16703
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.049286206750307e-06,
-      "loss": 0.6879,
+      "loss": 0.7838,
       "step": 16704
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0481608095896825e-06,
-      "loss": 0.6899,
+      "loss": 0.8704,
       "step": 16705
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.047035453381641e-06,
-      "loss": 0.67,
+      "loss": 0.7786,
       "step": 16706
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0459101381417915e-06,
-      "loss": 0.7487,
+      "loss": 0.893,
       "step": 16707
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.044784863885752e-06,
-      "loss": 0.7883,
+      "loss": 0.9044,
       "step": 16708
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.043659630629121e-06,
-      "loss": 0.8646,
+      "loss": 0.9086,
       "step": 16709
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0425344383875274e-06,
-      "loss": 0.6488,
+      "loss": 0.6867,
       "step": 16710
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.041409287176569e-06,
-      "loss": 0.75,
+      "loss": 0.8297,
       "step": 16711
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.040284177011856e-06,
-      "loss": 0.6069,
+      "loss": 0.8249,
       "step": 16712
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.039159107909006e-06,
-      "loss": 0.6367,
+      "loss": 0.7091,
       "step": 16713
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.038034079883615e-06,
-      "loss": 0.7707,
+      "loss": 0.8347,
       "step": 16714
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.036909092951308e-06,
-      "loss": 0.7637,
+      "loss": 0.8322,
       "step": 16715
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.035784147127676e-06,
-      "loss": 0.6568,
+      "loss": 0.8313,
       "step": 16716
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.034659242428337e-06,
-      "loss": 0.7021,
+      "loss": 0.8151,
       "step": 16717
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0335343788688894e-06,
-      "loss": 0.66,
+      "loss": 0.7476,
       "step": 16718
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0324095564649475e-06,
-      "loss": 0.7467,
+      "loss": 0.8823,
       "step": 16719
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.03128477523211e-06,
-      "loss": 0.7232,
+      "loss": 0.9194,
       "step": 16720
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.030160035185989e-06,
-      "loss": 0.6329,
+      "loss": 0.798,
       "step": 16721
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.029035336342177e-06,
-      "loss": 0.6111,
+      "loss": 0.7283,
       "step": 16722
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.027910678716291e-06,
-      "loss": 0.6181,
+      "loss": 0.7318,
       "step": 16723
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.026786062323922e-06,
-      "loss": 0.7836,
+      "loss": 0.8736,
       "step": 16724
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0256614871806815e-06,
-      "loss": 0.6207,
+      "loss": 0.7599,
       "step": 16725
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.024536953302165e-06,
-      "loss": 0.7726,
+      "loss": 0.8338,
       "step": 16726
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.02341246070398e-06,
-      "loss": 0.8302,
+      "loss": 0.9593,
       "step": 16727
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.022288009401721e-06,
-      "loss": 0.7799,
+      "loss": 0.8118,
       "step": 16728
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.021163599410995e-06,
-      "loss": 0.7386,
+      "loss": 0.8775,
       "step": 16729
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0200392307473955e-06,
-      "loss": 0.7187,
+      "loss": 0.8141,
       "step": 16730
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.018914903426521e-06,
-      "loss": 0.6928,
+      "loss": 0.8572,
       "step": 16731
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.017790617463977e-06,
-      "loss": 0.7511,
+      "loss": 0.8399,
       "step": 16732
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.016666372875352e-06,
-      "loss": 0.6322,
+      "loss": 0.8592,
       "step": 16733
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.015542169676255e-06,
-      "loss": 0.7254,
+      "loss": 0.9161,
       "step": 16734
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.014418007882271e-06,
-      "loss": 0.6862,
+      "loss": 0.7652,
       "step": 16735
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.013293887509003e-06,
-      "loss": 0.6979,
+      "loss": 0.7713,
       "step": 16736
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.012169808572041e-06,
-      "loss": 0.5622,
+      "loss": 0.7128,
       "step": 16737
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.011045771086988e-06,
-      "loss": 0.6804,
+      "loss": 0.7579,
       "step": 16738
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.009921775069432e-06,
-      "loss": 0.7585,
+      "loss": 0.7582,
       "step": 16739
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0087978205349715e-06,
-      "loss": 0.7423,
+      "loss": 0.7817,
       "step": 16740
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.007673907499193e-06,
-      "loss": 0.731,
+      "loss": 0.854,
       "step": 16741
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.006550035977698e-06,
-      "loss": 0.6469,
+      "loss": 0.8422,
       "step": 16742
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.005426205986071e-06,
-      "loss": 0.6117,
+      "loss": 0.8051,
       "step": 16743
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.004302417539909e-06,
-      "loss": 0.7044,
+      "loss": 0.7824,
       "step": 16744
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.003178670654798e-06,
-      "loss": 0.7751,
+      "loss": 0.8445,
       "step": 16745
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.002054965346332e-06,
-      "loss": 0.6933,
+      "loss": 0.834,
       "step": 16746
     },
     {
       "epoch": 0.61,
       "learning_rate": 7.0009313016300994e-06,
-      "loss": 0.6489,
+      "loss": 0.7613,
       "step": 16747
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9998076795216926e-06,
-      "loss": 0.6279,
+      "loss": 0.8604,
       "step": 16748
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9986840990366965e-06,
-      "loss": 0.659,
+      "loss": 0.7901,
       "step": 16749
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.997560560190699e-06,
-      "loss": 0.6655,
+      "loss": 0.8418,
       "step": 16750
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.996437062999291e-06,
-      "loss": 0.8408,
+      "loss": 0.9117,
       "step": 16751
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.995313607478054e-06,
-      "loss": 0.7751,
+      "loss": 0.8314,
       "step": 16752
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9941901936425806e-06,
-      "loss": 0.8296,
+      "loss": 0.8692,
       "step": 16753
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.993066821508448e-06,
-      "loss": 0.7541,
+      "loss": 0.8117,
       "step": 16754
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9919434910912545e-06,
-      "loss": 0.8319,
+      "loss": 0.9093,
       "step": 16755
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.990820202406572e-06,
-      "loss": 0.844,
+      "loss": 0.942,
       "step": 16756
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9896969554699926e-06,
-      "loss": 0.7446,
+      "loss": 0.7271,
       "step": 16757
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9885737502970944e-06,
-      "loss": 0.754,
+      "loss": 0.7257,
       "step": 16758
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.987450586903464e-06,
-      "loss": 0.6768,
+      "loss": 0.8242,
       "step": 16759
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.986327465304681e-06,
-      "loss": 0.8028,
+      "loss": 0.9039,
       "step": 16760
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.985204385516333e-06,
-      "loss": 0.6262,
+      "loss": 0.7259,
       "step": 16761
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9840813475539925e-06,
-      "loss": 0.8296,
+      "loss": 0.9135,
       "step": 16762
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9829583514332464e-06,
-      "loss": 0.7108,
+      "loss": 0.83,
       "step": 16763
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.98183539716967e-06,
-      "loss": 0.6626,
+      "loss": 0.8174,
       "step": 16764
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.980712484778848e-06,
-      "loss": 0.7635,
+      "loss": 0.8835,
       "step": 16765
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9795896142763545e-06,
-      "loss": 0.6936,
+      "loss": 0.7882,
       "step": 16766
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.978466785677773e-06,
-      "loss": 0.7445,
+      "loss": 0.8119,
       "step": 16767
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.977343998998675e-06,
-      "loss": 0.7697,
+      "loss": 0.8223,
       "step": 16768
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.976221254254641e-06,
-      "loss": 0.8755,
+      "loss": 0.968,
       "step": 16769
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.975098551461248e-06,
-      "loss": 0.7683,
+      "loss": 0.868,
       "step": 16770
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.973975890634069e-06,
-      "loss": 0.673,
+      "loss": 0.815,
       "step": 16771
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.972853271788681e-06,
-      "loss": 0.7349,
+      "loss": 0.8068,
       "step": 16772
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.971730694940659e-06,
-      "loss": 0.6235,
+      "loss": 0.7038,
       "step": 16773
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.970608160105581e-06,
-      "loss": 0.7657,
+      "loss": 0.8825,
       "step": 16774
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.969485667299012e-06,
-      "loss": 0.6282,
+      "loss": 0.7523,
       "step": 16775
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9683632165365315e-06,
-      "loss": 0.7933,
+      "loss": 0.875,
       "step": 16776
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.967240807833707e-06,
-      "loss": 0.7787,
+      "loss": 0.8923,
       "step": 16777
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.966118441206117e-06,
-      "loss": 0.7148,
+      "loss": 0.7914,
       "step": 16778
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.964996116669327e-06,
-      "loss": 0.6687,
+      "loss": 0.845,
       "step": 16779
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.963873834238913e-06,
-      "loss": 0.7244,
+      "loss": 0.875,
       "step": 16780
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.962751593930437e-06,
-      "loss": 0.6694,
+      "loss": 0.7854,
       "step": 16781
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.961629395759478e-06,
-      "loss": 0.7589,
+      "loss": 0.8855,
       "step": 16782
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.960507239741597e-06,
-      "loss": 0.708,
+      "loss": 0.8678,
       "step": 16783
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.959385125892369e-06,
-      "loss": 0.7309,
+      "loss": 0.7879,
       "step": 16784
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.958263054227353e-06,
-      "loss": 0.8319,
+      "loss": 0.8636,
       "step": 16785
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.957141024762126e-06,
-      "loss": 0.7243,
+      "loss": 0.8497,
       "step": 16786
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.956019037512252e-06,
-      "loss": 0.7411,
+      "loss": 0.8266,
       "step": 16787
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.954897092493289e-06,
-      "loss": 0.6127,
+      "loss": 0.7538,
       "step": 16788
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.953775189720813e-06,
-      "loss": 0.6094,
+      "loss": 0.743,
       "step": 16789
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.95265332921038e-06,
-      "loss": 0.7561,
+      "loss": 0.8263,
       "step": 16790
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.951531510977562e-06,
-      "loss": 0.6667,
+      "loss": 0.7899,
       "step": 16791
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9504097350379156e-06,
-      "loss": 0.6836,
+      "loss": 0.8255,
       "step": 16792
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.949288001407013e-06,
-      "loss": 0.7937,
+      "loss": 0.862,
       "step": 16793
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.948166310100404e-06,
-      "loss": 0.6683,
+      "loss": 0.794,
       "step": 16794
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.947044661133664e-06,
-      "loss": 0.8513,
+      "loss": 0.9106,
       "step": 16795
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.945923054522343e-06,
-      "loss": 0.7368,
+      "loss": 0.791,
       "step": 16796
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9448014902820095e-06,
-      "loss": 0.7077,
+      "loss": 0.8796,
       "step": 16797
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.943679968428216e-06,
-      "loss": 0.6862,
+      "loss": 0.7572,
       "step": 16798
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.942558488976532e-06,
-      "loss": 0.7221,
+      "loss": 0.8501,
       "step": 16799
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.941437051942507e-06,
-      "loss": 0.7424,
+      "loss": 0.892,
       "step": 16800
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.940315657341707e-06,
-      "loss": 0.6642,
+      "loss": 0.8327,
       "step": 16801
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.939194305189684e-06,
-      "loss": 0.721,
+      "loss": 0.7716,
       "step": 16802
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.938072995501998e-06,
-      "loss": 0.8754,
+      "loss": 0.9318,
       "step": 16803
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.936951728294203e-06,
-      "loss": 0.7327,
+      "loss": 0.8511,
       "step": 16804
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.93583050358186e-06,
-      "loss": 0.6625,
+      "loss": 0.7251,
       "step": 16805
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.934709321380522e-06,
-      "loss": 0.7102,
+      "loss": 0.7998,
       "step": 16806
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.93358818170574e-06,
-      "loss": 0.8136,
+      "loss": 0.8706,
       "step": 16807
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.932467084573071e-06,
-      "loss": 0.5934,
+      "loss": 0.7054,
       "step": 16808
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.931346029998069e-06,
-      "loss": 0.7477,
+      "loss": 0.8237,
       "step": 16809
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.930225017996289e-06,
-      "loss": 0.6658,
+      "loss": 0.8055,
       "step": 16810
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.929104048583279e-06,
-      "loss": 0.7546,
+      "loss": 0.8235,
       "step": 16811
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.927983121774598e-06,
-      "loss": 0.6617,
+      "loss": 0.8394,
       "step": 16812
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.926862237585785e-06,
-      "loss": 0.7219,
+      "loss": 0.8415,
       "step": 16813
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.925741396032406e-06,
-      "loss": 0.6558,
+      "loss": 0.7804,
       "step": 16814
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.924620597129998e-06,
-      "loss": 0.716,
+      "loss": 0.8206,
       "step": 16815
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.923499840894119e-06,
-      "loss": 0.6777,
+      "loss": 0.7907,
       "step": 16816
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.922379127340311e-06,
-      "loss": 0.8149,
+      "loss": 0.897,
       "step": 16817
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.92125845648413e-06,
-      "loss": 0.7356,
+      "loss": 0.8092,
       "step": 16818
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.920137828341116e-06,
-      "loss": 0.6061,
+      "loss": 0.7688,
       "step": 16819
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.919017242926826e-06,
-      "loss": 0.6494,
+      "loss": 0.7412,
       "step": 16820
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.917896700256792e-06,
-      "loss": 0.7691,
+      "loss": 0.9054,
       "step": 16821
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.916776200346576e-06,
-      "loss": 0.6603,
+      "loss": 0.8148,
       "step": 16822
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9156557432117114e-06,
-      "loss": 0.7997,
+      "loss": 0.8737,
       "step": 16823
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.914535328867749e-06,
-      "loss": 0.8447,
+      "loss": 0.9165,
       "step": 16824
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.913414957330233e-06,
-      "loss": 0.7961,
+      "loss": 0.8297,
       "step": 16825
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9122946286147e-06,
-      "loss": 0.7584,
+      "loss": 0.9408,
       "step": 16826
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.911174342736703e-06,
-      "loss": 0.8447,
+      "loss": 0.9614,
       "step": 16827
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.910054099711776e-06,
-      "loss": 0.8222,
+      "loss": 0.8404,
       "step": 16828
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.908933899555466e-06,
-      "loss": 0.8922,
+      "loss": 0.8881,
       "step": 16829
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.907813742283312e-06,
-      "loss": 0.8519,
+      "loss": 0.8937,
       "step": 16830
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.906693627910855e-06,
-      "loss": 0.7491,
+      "loss": 0.8661,
       "step": 16831
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.905573556453634e-06,
-      "loss": 0.8491,
+      "loss": 0.8765,
       "step": 16832
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.904453527927194e-06,
-      "loss": 0.833,
+      "loss": 0.8075,
       "step": 16833
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.9033335423470645e-06,
-      "loss": 0.6393,
+      "loss": 0.794,
       "step": 16834
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.902213599728791e-06,
-      "loss": 0.9944,
+      "loss": 1.017,
       "step": 16835
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.901093700087906e-06,
-      "loss": 0.7797,
+      "loss": 0.8427,
       "step": 16836
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.899973843439952e-06,
-      "loss": 0.6588,
+      "loss": 0.8066,
       "step": 16837
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.898854029800461e-06,
-      "loss": 0.727,
+      "loss": 0.8405,
       "step": 16838
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8977342591849724e-06,
-      "loss": 0.7548,
+      "loss": 0.7597,
       "step": 16839
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.896614531609015e-06,
-      "loss": 0.7024,
+      "loss": 0.8371,
       "step": 16840
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.895494847088134e-06,
-      "loss": 0.7574,
+      "loss": 0.8147,
       "step": 16841
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.894375205637852e-06,
-      "loss": 0.6449,
+      "loss": 0.7747,
       "step": 16842
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.89325560727371e-06,
-      "loss": 0.7877,
+      "loss": 0.8831,
       "step": 16843
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8921360520112405e-06,
-      "loss": 0.6932,
+      "loss": 0.7874,
       "step": 16844
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.891016539865969e-06,
-      "loss": 0.6686,
+      "loss": 0.7836,
       "step": 16845
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.889897070853437e-06,
-      "loss": 0.6986,
+      "loss": 0.8221,
       "step": 16846
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8887776449891666e-06,
-      "loss": 0.7208,
+      "loss": 0.8021,
       "step": 16847
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.887658262288695e-06,
-      "loss": 0.6389,
+      "loss": 0.7832,
       "step": 16848
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.886538922767546e-06,
-      "loss": 0.7327,
+      "loss": 0.8169,
       "step": 16849
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.885419626441254e-06,
-      "loss": 0.6305,
+      "loss": 0.7631,
       "step": 16850
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.884300373325344e-06,
-      "loss": 0.8849,
+      "loss": 0.8419,
       "step": 16851
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.88318116343535e-06,
-      "loss": 0.7249,
+      "loss": 0.8313,
       "step": 16852
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.882061996786789e-06,
-      "loss": 0.6381,
+      "loss": 0.7613,
       "step": 16853
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8809428733952e-06,
-      "loss": 0.778,
+      "loss": 0.9092,
       "step": 16854
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.879823793276099e-06,
-      "loss": 0.7648,
+      "loss": 0.8137,
       "step": 16855
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.878704756445018e-06,
-      "loss": 0.5925,
+      "loss": 0.8257,
       "step": 16856
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.877585762917478e-06,
-      "loss": 0.683,
+      "loss": 0.768,
       "step": 16857
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8764668127090085e-06,
-      "loss": 0.6261,
+      "loss": 0.7062,
       "step": 16858
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.875347905835127e-06,
-      "loss": 0.6645,
+      "loss": 0.7808,
       "step": 16859
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8742290423113645e-06,
-      "loss": 0.6691,
+      "loss": 0.8643,
       "step": 16860
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.873110222153235e-06,
-      "loss": 0.7412,
+      "loss": 0.8055,
       "step": 16861
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.871991445376267e-06,
-      "loss": 0.8113,
+      "loss": 0.9188,
       "step": 16862
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.87087271199598e-06,
-      "loss": 0.6494,
+      "loss": 0.818,
       "step": 16863
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.86975402202789e-06,
-      "loss": 0.7293,
+      "loss": 0.8267,
       "step": 16864
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.868635375487529e-06,
-      "loss": 0.6193,
+      "loss": 0.7674,
       "step": 16865
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.867516772390401e-06,
-      "loss": 0.6401,
+      "loss": 0.7857,
       "step": 16866
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.86639821275204e-06,
-      "loss": 0.8032,
+      "loss": 0.8795,
       "step": 16867
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8652796965879546e-06,
-      "loss": 0.7755,
+      "loss": 0.8432,
       "step": 16868
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.864161223913666e-06,
-      "loss": 0.8676,
+      "loss": 0.8219,
       "step": 16869
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.863042794744692e-06,
-      "loss": 0.582,
+      "loss": 0.7653,
       "step": 16870
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.861924409096548e-06,
-      "loss": 0.7937,
+      "loss": 0.8755,
       "step": 16871
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.860806066984749e-06,
-      "loss": 0.7375,
+      "loss": 0.7768,
       "step": 16872
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.859687768424816e-06,
-      "loss": 0.7396,
+      "loss": 0.8313,
       "step": 16873
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.858569513432256e-06,
-      "loss": 0.6343,
+      "loss": 0.7879,
       "step": 16874
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.85745130202259e-06,
-      "loss": 0.7147,
+      "loss": 0.8042,
       "step": 16875
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.856333134211325e-06,
-      "loss": 0.7988,
+      "loss": 0.8415,
       "step": 16876
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.85521501001398e-06,
-      "loss": 0.8385,
+      "loss": 0.8938,
       "step": 16877
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.854096929446062e-06,
-      "loss": 0.8046,
+      "loss": 0.9334,
       "step": 16878
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8529788925230915e-06,
-      "loss": 0.6781,
+      "loss": 0.8164,
       "step": 16879
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.851860899260566e-06,
-      "loss": 0.8551,
+      "loss": 0.911,
       "step": 16880
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8507429496740105e-06,
-      "loss": 0.7814,
+      "loss": 0.8815,
       "step": 16881
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.849625043778927e-06,
-      "loss": 0.8086,
+      "loss": 0.905,
       "step": 16882
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8485071815908235e-06,
-      "loss": 0.6356,
+      "loss": 0.7919,
       "step": 16883
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8473893631252155e-06,
-      "loss": 0.6868,
+      "loss": 0.7876,
       "step": 16884
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.846271588397601e-06,
-      "loss": 0.7359,
+      "loss": 0.8167,
       "step": 16885
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8451538574235e-06,
-      "loss": 0.7723,
+      "loss": 0.8077,
       "step": 16886
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.844036170218409e-06,
-      "loss": 0.9099,
+      "loss": 1.0008,
       "step": 16887
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8429185267978405e-06,
-      "loss": 0.588,
+      "loss": 0.7803,
       "step": 16888
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.841800927177296e-06,
-      "loss": 0.7675,
+      "loss": 0.8086,
       "step": 16889
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.840683371372286e-06,
-      "loss": 0.7282,
+      "loss": 0.8577,
       "step": 16890
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8395658593983076e-06,
-      "loss": 0.6577,
+      "loss": 0.8495,
       "step": 16891
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.838448391270873e-06,
-      "loss": 0.6433,
+      "loss": 0.7913,
       "step": 16892
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.837330967005477e-06,
-      "loss": 0.7656,
+      "loss": 0.8865,
       "step": 16893
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.836213586617631e-06,
-      "loss": 0.728,
+      "loss": 0.7794,
       "step": 16894
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8350962501228305e-06,
-      "loss": 0.6519,
+      "loss": 0.7489,
       "step": 16895
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.833978957536579e-06,
-      "loss": 0.699,
+      "loss": 0.8212,
       "step": 16896
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.832861708874377e-06,
-      "loss": 0.6883,
+      "loss": 0.8376,
       "step": 16897
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.831744504151725e-06,
-      "loss": 0.764,
+      "loss": 0.8174,
       "step": 16898
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.830627343384123e-06,
-      "loss": 0.7876,
+      "loss": 0.8459,
       "step": 16899
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8295102265870735e-06,
-      "loss": 0.724,
+      "loss": 0.8484,
       "step": 16900
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8283931537760674e-06,
-      "loss": 0.6545,
+      "loss": 0.8516,
       "step": 16901
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.827276124966608e-06,
-      "loss": 0.6542,
+      "loss": 0.7271,
       "step": 16902
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.826159140174191e-06,
-      "loss": 0.7885,
+      "loss": 0.8379,
       "step": 16903
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.82504219941431e-06,
-      "loss": 0.6388,
+      "loss": 0.7461,
       "step": 16904
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.823925302702469e-06,
-      "loss": 0.6906,
+      "loss": 0.801,
       "step": 16905
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.8228084500541534e-06,
-      "loss": 0.6964,
+      "loss": 0.8352,
       "step": 16906
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.821691641484865e-06,
-      "loss": 0.6985,
+      "loss": 0.8457,
       "step": 16907
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.820574877010093e-06,
-      "loss": 0.7691,
+      "loss": 0.835,
       "step": 16908
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.819458156645337e-06,
-      "loss": 0.7211,
+      "loss": 0.8717,
       "step": 16909
     },
     {
       "epoch": 0.61,
       "learning_rate": 6.818341480406083e-06,
-      "loss": 0.6382,
+      "loss": 0.7832,
       "step": 16910
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.817224848307833e-06,
-      "loss": 0.7339,
+      "loss": 0.8075,
       "step": 16911
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.8161082603660645e-06,
-      "loss": 0.7299,
+      "loss": 0.8777,
       "step": 16912
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.814991716596283e-06,
-      "loss": 0.6874,
+      "loss": 0.7973,
       "step": 16913
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.813875217013968e-06,
-      "loss": 0.6999,
+      "loss": 0.8608,
       "step": 16914
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.812758761634617e-06,
-      "loss": 0.763,
+      "loss": 0.8424,
       "step": 16915
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.811642350473713e-06,
-      "loss": 0.8215,
+      "loss": 0.8481,
       "step": 16916
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.81052598354675e-06,
-      "loss": 0.774,
+      "loss": 0.8103,
       "step": 16917
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.809409660869212e-06,
-      "loss": 0.7031,
+      "loss": 0.7699,
       "step": 16918
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.808293382456592e-06,
-      "loss": 0.8173,
+      "loss": 0.9067,
       "step": 16919
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.8071771483243665e-06,
-      "loss": 0.7407,
+      "loss": 0.8317,
       "step": 16920
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.806060958488034e-06,
-      "loss": 0.8096,
+      "loss": 0.8538,
       "step": 16921
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.804944812963072e-06,
-      "loss": 0.7436,
+      "loss": 0.8783,
       "step": 16922
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.803828711764966e-06,
-      "loss": 0.7292,
+      "loss": 0.8076,
       "step": 16923
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.802712654909207e-06,
-      "loss": 0.7358,
+      "loss": 0.8197,
       "step": 16924
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.8015966424112655e-06,
-      "loss": 0.7229,
+      "loss": 0.8593,
       "step": 16925
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.8004806742866405e-06,
-      "loss": 0.6774,
+      "loss": 0.794,
       "step": 16926
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.799364750550801e-06,
-      "loss": 0.7454,
+      "loss": 0.8443,
       "step": 16927
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.798248871219237e-06,
-      "loss": 0.8704,
+      "loss": 0.9026,
       "step": 16928
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.797133036307426e-06,
-      "loss": 0.7359,
+      "loss": 0.83,
       "step": 16929
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.796017245830851e-06,
-      "loss": 0.8069,
+      "loss": 0.8862,
       "step": 16930
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.794901499804989e-06,
-      "loss": 0.7938,
+      "loss": 0.8387,
       "step": 16931
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.793785798245325e-06,
-      "loss": 0.8292,
+      "loss": 0.9525,
       "step": 16932
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.79267014116733e-06,
-      "loss": 0.7666,
+      "loss": 0.8524,
       "step": 16933
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.791554528586488e-06,
-      "loss": 0.7867,
+      "loss": 0.8733,
       "step": 16934
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.790438960518274e-06,
-      "loss": 0.6542,
+      "loss": 0.7165,
       "step": 16935
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.789323436978167e-06,
-      "loss": 0.6746,
+      "loss": 0.8281,
       "step": 16936
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.78820795798164e-06,
-      "loss": 0.8097,
+      "loss": 0.859,
       "step": 16937
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.787092523544172e-06,
-      "loss": 0.7608,
+      "loss": 0.7816,
       "step": 16938
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7859771336812354e-06,
-      "loss": 0.6943,
+      "loss": 0.8203,
       "step": 16939
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7848617884083105e-06,
-      "loss": 0.8116,
+      "loss": 0.8785,
       "step": 16940
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.783746487740863e-06,
-      "loss": 0.6388,
+      "loss": 0.7261,
       "step": 16941
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.782631231694368e-06,
-      "loss": 0.8141,
+      "loss": 0.8542,
       "step": 16942
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7815160202843035e-06,
-      "loss": 0.7816,
+      "loss": 0.8328,
       "step": 16943
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7804008535261345e-06,
-      "loss": 0.83,
+      "loss": 0.9236,
       "step": 16944
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.779285731435339e-06,
-      "loss": 0.719,
+      "loss": 0.8756,
       "step": 16945
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.778170654027381e-06,
-      "loss": 0.756,
+      "loss": 0.8274,
       "step": 16946
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.777055621317736e-06,
-      "loss": 0.7101,
+      "loss": 0.8052,
       "step": 16947
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.775940633321871e-06,
-      "loss": 0.7237,
+      "loss": 0.8113,
       "step": 16948
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.774825690055256e-06,
-      "loss": 0.6342,
+      "loss": 0.7708,
       "step": 16949
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.773710791533356e-06,
-      "loss": 0.5839,
+      "loss": 0.8068,
       "step": 16950
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.772595937771646e-06,
-      "loss": 0.7196,
+      "loss": 0.8448,
       "step": 16951
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.771481128785582e-06,
-      "loss": 0.7339,
+      "loss": 0.8881,
       "step": 16952
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.770366364590644e-06,
-      "loss": 0.6153,
+      "loss": 0.788,
       "step": 16953
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7692516452022846e-06,
-      "loss": 0.7345,
+      "loss": 0.8896,
       "step": 16954
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7681369706359774e-06,
-      "loss": 0.6441,
+      "loss": 0.8325,
       "step": 16955
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.767022340907183e-06,
-      "loss": 0.7685,
+      "loss": 0.834,
       "step": 16956
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.765907756031368e-06,
-      "loss": 0.846,
+      "loss": 0.9137,
       "step": 16957
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.764793216023993e-06,
-      "loss": 0.6959,
+      "loss": 0.9127,
       "step": 16958
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.763678720900524e-06,
-      "loss": 0.6387,
+      "loss": 0.787,
       "step": 16959
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7625642706764215e-06,
-      "loss": 0.7822,
+      "loss": 0.8605,
       "step": 16960
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.761449865367142e-06,
-      "loss": 0.6283,
+      "loss": 0.7712,
       "step": 16961
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.760335504988155e-06,
-      "loss": 0.66,
+      "loss": 0.7873,
       "step": 16962
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.759221189554913e-06,
-      "loss": 0.939,
+      "loss": 1.0111,
       "step": 16963
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.758106919082883e-06,
-      "loss": 0.8196,
+      "loss": 0.9217,
       "step": 16964
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.756992693587514e-06,
-      "loss": 0.609,
+      "loss": 0.7744,
       "step": 16965
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7558785130842775e-06,
-      "loss": 0.7467,
+      "loss": 0.8217,
       "step": 16966
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.754764377588618e-06,
-      "loss": 0.7459,
+      "loss": 0.8133,
       "step": 16967
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.753650287116e-06,
-      "loss": 0.7134,
+      "loss": 0.7791,
       "step": 16968
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.752536241681878e-06,
-      "loss": 0.688,
+      "loss": 0.8431,
       "step": 16969
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.75142224130171e-06,
-      "loss": 0.7817,
+      "loss": 0.8419,
       "step": 16970
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.750308285990946e-06,
-      "loss": 0.7207,
+      "loss": 0.8284,
       "step": 16971
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.749194375765049e-06,
-      "loss": 0.6873,
+      "loss": 0.8109,
       "step": 16972
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.748080510639463e-06,
-      "loss": 0.7247,
+      "loss": 0.8042,
       "step": 16973
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.746966690629648e-06,
-      "loss": 0.7088,
+      "loss": 0.8297,
       "step": 16974
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.745852915751054e-06,
-      "loss": 0.7249,
+      "loss": 0.927,
       "step": 16975
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.744739186019133e-06,
-      "loss": 0.7456,
+      "loss": 0.8203,
       "step": 16976
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.743625501449338e-06,
-      "loss": 0.7958,
+      "loss": 0.7992,
       "step": 16977
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.742511862057122e-06,
-      "loss": 0.6808,
+      "loss": 0.8665,
       "step": 16978
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.74139826785793e-06,
-      "loss": 0.7908,
+      "loss": 0.8218,
       "step": 16979
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.740284718867211e-06,
-      "loss": 0.6975,
+      "loss": 0.8022,
       "step": 16980
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.739171215100419e-06,
-      "loss": 0.8813,
+      "loss": 0.8841,
       "step": 16981
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7380577565729996e-06,
-      "loss": 0.8118,
+      "loss": 0.9295,
       "step": 16982
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.736944343300401e-06,
-      "loss": 0.8895,
+      "loss": 0.9433,
       "step": 16983
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7358309752980685e-06,
-      "loss": 0.6817,
+      "loss": 0.8057,
       "step": 16984
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7347176525814536e-06,
-      "loss": 0.6669,
+      "loss": 0.8724,
       "step": 16985
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.733604375165995e-06,
-      "loss": 0.7103,
+      "loss": 0.7958,
       "step": 16986
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7324911430671435e-06,
-      "loss": 0.852,
+      "loss": 0.9224,
       "step": 16987
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.731377956300338e-06,
-      "loss": 0.5898,
+      "loss": 0.7717,
       "step": 16988
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.730264814881029e-06,
-      "loss": 0.6824,
+      "loss": 0.7266,
       "step": 16989
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.729151718824654e-06,
-      "loss": 0.6407,
+      "loss": 0.777,
       "step": 16990
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.728038668146663e-06,
-      "loss": 0.6381,
+      "loss": 0.7326,
       "step": 16991
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.726925662862485e-06,
-      "loss": 0.612,
+      "loss": 0.7522,
       "step": 16992
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7258127029875775e-06,
-      "loss": 0.7885,
+      "loss": 0.9038,
       "step": 16993
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.72469978853737e-06,
-      "loss": 0.6799,
+      "loss": 0.7721,
       "step": 16994
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.723586919527306e-06,
-      "loss": 0.7075,
+      "loss": 0.7503,
       "step": 16995
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.722474095972824e-06,
-      "loss": 0.9674,
+      "loss": 0.9177,
       "step": 16996
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.721361317889366e-06,
-      "loss": 0.8181,
+      "loss": 0.8651,
       "step": 16997
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.720248585292368e-06,
-      "loss": 0.6831,
+      "loss": 0.8131,
       "step": 16998
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.719135898197264e-06,
-      "loss": 0.7326,
+      "loss": 0.8229,
       "step": 16999
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.718023256619498e-06,
-      "loss": 0.704,
+      "loss": 0.8111,
       "step": 17000
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7169106605745e-06,
-      "loss": 0.8386,
+      "loss": 0.9457,
       "step": 17001
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.715798110077711e-06,
-      "loss": 0.8017,
+      "loss": 0.884,
       "step": 17002
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.714685605144561e-06,
-      "loss": 0.6848,
+      "loss": 0.7456,
       "step": 17003
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.713573145790491e-06,
-      "loss": 0.6974,
+      "loss": 0.7918,
       "step": 17004
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.712460732030928e-06,
-      "loss": 0.8186,
+      "loss": 0.8471,
       "step": 17005
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.711348363881307e-06,
-      "loss": 0.6696,
+      "loss": 0.7867,
       "step": 17006
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.710236041357061e-06,
-      "loss": 0.6724,
+      "loss": 0.8314,
       "step": 17007
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.709123764473625e-06,
-      "loss": 0.6508,
+      "loss": 0.8224,
       "step": 17008
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.708011533246426e-06,
-      "loss": 0.7143,
+      "loss": 0.7753,
       "step": 17009
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.706899347690897e-06,
-      "loss": 0.6653,
+      "loss": 0.7818,
       "step": 17010
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.705787207822465e-06,
-      "loss": 0.7066,
+      "loss": 0.7721,
       "step": 17011
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.704675113656567e-06,
-      "loss": 0.7239,
+      "loss": 0.86,
       "step": 17012
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.703563065208622e-06,
-      "loss": 0.6454,
+      "loss": 0.7821,
       "step": 17013
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7024510624940645e-06,
-      "loss": 0.6152,
+      "loss": 0.7771,
       "step": 17014
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.701339105528317e-06,
-      "loss": 0.6505,
+      "loss": 0.7765,
       "step": 17015
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.7002271943268116e-06,
-      "loss": 0.7178,
+      "loss": 0.8434,
       "step": 17016
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.699115328904975e-06,
-      "loss": 0.6901,
+      "loss": 0.7775,
       "step": 17017
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.698003509278225e-06,
-      "loss": 0.735,
+      "loss": 0.7907,
       "step": 17018
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.696891735461993e-06,
-      "loss": 0.7408,
+      "loss": 0.8088,
       "step": 17019
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6957800074716994e-06,
-      "loss": 0.864,
+      "loss": 0.8722,
       "step": 17020
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.694668325322774e-06,
-      "loss": 0.7429,
+      "loss": 0.8008,
       "step": 17021
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.693556689030633e-06,
-      "loss": 0.7329,
+      "loss": 0.7981,
       "step": 17022
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.692445098610704e-06,
-      "loss": 0.7214,
+      "loss": 0.8242,
       "step": 17023
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.691333554078403e-06,
-      "loss": 0.7554,
+      "loss": 0.8365,
       "step": 17024
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.69022205544916e-06,
-      "loss": 0.7506,
+      "loss": 0.9388,
       "step": 17025
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.689110602738384e-06,
-      "loss": 0.6461,
+      "loss": 0.7855,
       "step": 17026
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.687999195961505e-06,
-      "loss": 0.8234,
+      "loss": 0.8974,
       "step": 17027
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.686887835133936e-06,
-      "loss": 0.662,
+      "loss": 0.7933,
       "step": 17028
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.685776520271099e-06,
-      "loss": 0.8565,
+      "loss": 0.837,
       "step": 17029
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6846652513884095e-06,
-      "loss": 0.6703,
+      "loss": 0.8047,
       "step": 17030
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.683554028501289e-06,
-      "loss": 0.6879,
+      "loss": 0.7973,
       "step": 17031
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.682442851625144e-06,
-      "loss": 0.7586,
+      "loss": 0.8701,
       "step": 17032
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.681331720775403e-06,
-      "loss": 0.7384,
+      "loss": 0.8182,
       "step": 17033
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.680220635967474e-06,
-      "loss": 0.8035,
+      "loss": 0.8571,
       "step": 17034
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.679109597216775e-06,
-      "loss": 0.7112,
+      "loss": 0.803,
       "step": 17035
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6779986045387204e-06,
-      "loss": 0.6358,
+      "loss": 0.8624,
       "step": 17036
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.676887657948715e-06,
-      "loss": 0.6294,
+      "loss": 0.7361,
       "step": 17037
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.675776757462184e-06,
-      "loss": 0.8143,
+      "loss": 0.8868,
       "step": 17038
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.674665903094533e-06,
-      "loss": 0.6333,
+      "loss": 0.6999,
       "step": 17039
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.673555094861175e-06,
-      "loss": 0.6806,
+      "loss": 0.7704,
       "step": 17040
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.672444332777517e-06,
-      "loss": 0.7284,
+      "loss": 0.8291,
       "step": 17041
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.671333616858976e-06,
-      "loss": 0.8574,
+      "loss": 0.9426,
       "step": 17042
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.670222947120957e-06,
-      "loss": 0.7323,
+      "loss": 0.8691,
       "step": 17043
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.669112323578873e-06,
-      "loss": 0.714,
+      "loss": 0.8027,
       "step": 17044
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.668001746248125e-06,
-      "loss": 0.8994,
+      "loss": 0.8826,
       "step": 17045
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.666891215144129e-06,
-      "loss": 0.7098,
+      "loss": 0.9077,
       "step": 17046
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6657807302822856e-06,
-      "loss": 0.6192,
+      "loss": 0.8122,
       "step": 17047
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.664670291678006e-06,
-      "loss": 0.759,
+      "loss": 0.8318,
       "step": 17048
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.663559899346692e-06,
-      "loss": 0.7648,
+      "loss": 0.8535,
       "step": 17049
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.662449553303753e-06,
-      "loss": 0.8051,
+      "loss": 0.9091,
       "step": 17050
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.661339253564587e-06,
-      "loss": 0.8048,
+      "loss": 0.8893,
       "step": 17051
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.660229000144609e-06,
-      "loss": 0.6248,
+      "loss": 0.7179,
       "step": 17052
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6591187930592095e-06,
-      "loss": 0.7383,
+      "loss": 0.8371,
       "step": 17053
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.658008632323798e-06,
-      "loss": 0.6943,
+      "loss": 0.7915,
       "step": 17054
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.656898517953776e-06,
-      "loss": 0.69,
+      "loss": 0.8112,
       "step": 17055
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.655788449964542e-06,
-      "loss": 0.6933,
+      "loss": 0.7769,
       "step": 17056
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.654678428371502e-06,
-      "loss": 0.697,
+      "loss": 0.8485,
       "step": 17057
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.653568453190048e-06,
-      "loss": 0.7716,
+      "loss": 0.8068,
       "step": 17058
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.652458524435587e-06,
-      "loss": 0.6007,
+      "loss": 0.7006,
       "step": 17059
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.65134864212351e-06,
-      "loss": 0.6423,
+      "loss": 0.7995,
       "step": 17060
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.650238806269225e-06,
-      "loss": 0.8107,
+      "loss": 0.8393,
       "step": 17061
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.649129016888118e-06,
-      "loss": 0.9024,
+      "loss": 0.9371,
       "step": 17062
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.648019273995599e-06,
-      "loss": 0.8083,
+      "loss": 0.9246,
       "step": 17063
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.646909577607049e-06,
-      "loss": 0.8173,
+      "loss": 0.8577,
       "step": 17064
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.645799927737877e-06,
-      "loss": 0.6107,
+      "loss": 0.7746,
       "step": 17065
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.644690324403469e-06,
-      "loss": 0.7179,
+      "loss": 0.8333,
       "step": 17066
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.643580767619225e-06,
-      "loss": 0.8044,
+      "loss": 0.9503,
       "step": 17067
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.642471257400531e-06,
-      "loss": 0.6426,
+      "loss": 0.7918,
       "step": 17068
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.641361793762788e-06,
-      "loss": 0.6471,
+      "loss": 0.8413,
       "step": 17069
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.640252376721383e-06,
-      "loss": 0.7341,
+      "loss": 0.7822,
       "step": 17070
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.639143006291713e-06,
-      "loss": 0.6179,
+      "loss": 0.7976,
       "step": 17071
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.638033682489162e-06,
-      "loss": 0.6756,
+      "loss": 0.7488,
       "step": 17072
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.636924405329126e-06,
-      "loss": 0.6683,
+      "loss": 0.8236,
       "step": 17073
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6358151748269916e-06,
-      "loss": 0.8668,
+      "loss": 0.972,
       "step": 17074
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.634705990998146e-06,
-      "loss": 0.8699,
+      "loss": 0.9043,
       "step": 17075
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.633596853857984e-06,
-      "loss": 0.5985,
+      "loss": 0.7732,
       "step": 17076
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.632487763421886e-06,
-      "loss": 0.709,
+      "loss": 0.8974,
       "step": 17077
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.631378719705245e-06,
-      "loss": 0.7931,
+      "loss": 0.9066,
       "step": 17078
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6302697227234415e-06,
-      "loss": 0.7806,
+      "loss": 0.8486,
       "step": 17079
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.629160772491867e-06,
-      "loss": 0.6238,
+      "loss": 0.7247,
       "step": 17080
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.628051869025902e-06,
-      "loss": 0.7216,
+      "loss": 0.859,
       "step": 17081
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6269430123409365e-06,
-      "loss": 0.7472,
+      "loss": 0.8378,
       "step": 17082
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.625834202452348e-06,
-      "loss": 0.7581,
+      "loss": 0.8453,
       "step": 17083
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.624725439375527e-06,
-      "loss": 0.6864,
+      "loss": 0.8342,
       "step": 17084
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.623616723125847e-06,
-      "loss": 0.7133,
+      "loss": 0.8587,
       "step": 17085
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.622508053718697e-06,
-      "loss": 0.6908,
+      "loss": 0.8213,
       "step": 17086
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.621399431169453e-06,
-      "loss": 0.8049,
+      "loss": 0.8834,
       "step": 17087
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.620290855493501e-06,
-      "loss": 0.7044,
+      "loss": 0.8309,
       "step": 17088
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.619182326706216e-06,
-      "loss": 0.722,
+      "loss": 0.8316,
       "step": 17089
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.618073844822983e-06,
-      "loss": 0.5993,
+      "loss": 0.7244,
       "step": 17090
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.6169654098591715e-06,
-      "loss": 0.7434,
+      "loss": 0.8323,
       "step": 17091
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.615857021830172e-06,
-      "loss": 0.6564,
+      "loss": 0.7784,
       "step": 17092
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.61474868075135e-06,
-      "loss": 0.8275,
+      "loss": 0.916,
       "step": 17093
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.613640386638088e-06,
-      "loss": 0.6807,
+      "loss": 0.7679,
       "step": 17094
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.612532139505764e-06,
-      "loss": 0.7038,
+      "loss": 0.8132,
       "step": 17095
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.611423939369745e-06,
-      "loss": 0.6752,
+      "loss": 0.8312,
       "step": 17096
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.610315786245416e-06,
-      "loss": 0.7498,
+      "loss": 0.8535,
       "step": 17097
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.609207680148142e-06,
-      "loss": 0.5112,
+      "loss": 0.7427,
       "step": 17098
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.608099621093304e-06,
-      "loss": 0.7114,
+      "loss": 0.8055,
       "step": 17099
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.606991609096267e-06,
-      "loss": 0.792,
+      "loss": 0.8611,
       "step": 17100
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.605883644172412e-06,
-      "loss": 0.6313,
+      "loss": 0.8076,
       "step": 17101
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.604775726337102e-06,
-      "loss": 0.6702,
+      "loss": 0.7854,
       "step": 17102
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.603667855605716e-06,
-      "loss": 0.7147,
+      "loss": 0.9126,
       "step": 17103
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.602560031993614e-06,
-      "loss": 0.6417,
+      "loss": 0.7432,
       "step": 17104
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.601452255516176e-06,
-      "loss": 0.778,
+      "loss": 0.7729,
       "step": 17105
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.600344526188765e-06,
-      "loss": 0.6921,
+      "loss": 0.7573,
       "step": 17106
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.59923684402675e-06,
-      "loss": 0.7858,
+      "loss": 0.8454,
       "step": 17107
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.598129209045497e-06,
-      "loss": 0.781,
+      "loss": 0.9327,
       "step": 17108
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5970216212603755e-06,
-      "loss": 0.7802,
+      "loss": 0.8571,
       "step": 17109
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.595914080686751e-06,
-      "loss": 0.6777,
+      "loss": 0.7286,
       "step": 17110
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5948065873399925e-06,
-      "loss": 0.6755,
+      "loss": 0.8795,
       "step": 17111
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.593699141235458e-06,
-      "loss": 0.7228,
+      "loss": 0.7956,
       "step": 17112
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.592591742388513e-06,
-      "loss": 0.7065,
+      "loss": 0.8016,
       "step": 17113
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.591484390814526e-06,
-      "loss": 0.6933,
+      "loss": 0.8196,
       "step": 17114
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.590377086528854e-06,
-      "loss": 0.5939,
+      "loss": 0.8036,
       "step": 17115
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5892698295468685e-06,
-      "loss": 0.7,
+      "loss": 0.8169,
       "step": 17116
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.588162619883919e-06,
-      "loss": 0.7271,
+      "loss": 0.8349,
       "step": 17117
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.587055457555375e-06,
-      "loss": 0.757,
+      "loss": 0.8416,
       "step": 17118
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5859483425765916e-06,
-      "loss": 0.7418,
+      "loss": 0.8416,
       "step": 17119
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.584841274962935e-06,
-      "loss": 0.703,
+      "loss": 0.7739,
       "step": 17120
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5837342547297554e-06,
-      "loss": 0.7325,
+      "loss": 0.7955,
       "step": 17121
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.582627281892422e-06,
-      "loss": 0.7241,
+      "loss": 0.7652,
       "step": 17122
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.581520356466279e-06,
-      "loss": 0.6309,
+      "loss": 0.8063,
       "step": 17123
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.580413478466699e-06,
-      "loss": 0.5361,
+      "loss": 0.7049,
       "step": 17124
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.579306647909023e-06,
-      "loss": 0.6835,
+      "loss": 0.8191,
       "step": 17125
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.578199864808619e-06,
-      "loss": 0.5851,
+      "loss": 0.7066,
       "step": 17126
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.577093129180834e-06,
-      "loss": 0.757,
+      "loss": 0.8562,
       "step": 17127
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5759864410410265e-06,
-      "loss": 0.7908,
+      "loss": 0.9457,
       "step": 17128
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.574879800404548e-06,
-      "loss": 0.6188,
+      "loss": 0.8224,
       "step": 17129
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.573773207286756e-06,
-      "loss": 0.7615,
+      "loss": 0.8469,
       "step": 17130
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.572666661702997e-06,
-      "loss": 0.6943,
+      "loss": 0.7965,
       "step": 17131
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5715601636686244e-06,
-      "loss": 0.6316,
+      "loss": 0.8084,
       "step": 17132
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.570453713198992e-06,
-      "loss": 0.7082,
+      "loss": 0.7903,
       "step": 17133
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.569347310309446e-06,
-      "loss": 0.745,
+      "loss": 0.8102,
       "step": 17134
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.568240955015344e-06,
-      "loss": 0.7617,
+      "loss": 0.8113,
       "step": 17135
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.567134647332023e-06,
-      "loss": 0.6958,
+      "loss": 0.8528,
       "step": 17136
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5660283872748445e-06,
-      "loss": 0.7943,
+      "loss": 0.8892,
       "step": 17137
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.564922174859145e-06,
-      "loss": 0.6544,
+      "loss": 0.8013,
       "step": 17138
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.563816010100279e-06,
-      "loss": 0.6164,
+      "loss": 0.813,
       "step": 17139
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.562709893013588e-06,
-      "loss": 0.811,
+      "loss": 0.8469,
       "step": 17140
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.561603823614424e-06,
-      "loss": 0.7272,
+      "loss": 0.9073,
       "step": 17141
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5604978019181245e-06,
-      "loss": 0.794,
+      "loss": 0.8643,
       "step": 17142
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5593918279400434e-06,
-      "loss": 0.6119,
+      "loss": 0.7512,
       "step": 17143
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.558285901695515e-06,
-      "loss": 0.7143,
+      "loss": 0.7441,
       "step": 17144
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.557180023199888e-06,
-      "loss": 0.572,
+      "loss": 0.7304,
       "step": 17145
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.556074192468503e-06,
-      "loss": 0.6209,
+      "loss": 0.7645,
       "step": 17146
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.554968409516702e-06,
-      "loss": 0.6786,
+      "loss": 0.8808,
       "step": 17147
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5538626743598255e-06,
-      "loss": 0.7465,
+      "loss": 0.8237,
       "step": 17148
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.552756987013218e-06,
-      "loss": 0.6209,
+      "loss": 0.7412,
       "step": 17149
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.551651347492212e-06,
-      "loss": 0.666,
+      "loss": 0.796,
       "step": 17150
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.550545755812156e-06,
-      "loss": 0.6791,
+      "loss": 0.7808,
       "step": 17151
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5494402119883805e-06,
-      "loss": 0.7503,
+      "loss": 0.8385,
       "step": 17152
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5483347160362265e-06,
-      "loss": 0.7288,
+      "loss": 0.8346,
       "step": 17153
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.547229267971031e-06,
-      "loss": 0.6756,
+      "loss": 0.7762,
       "step": 17154
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.54612386780813e-06,
-      "loss": 0.7096,
+      "loss": 0.7991,
       "step": 17155
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.545018515562864e-06,
-      "loss": 0.7519,
+      "loss": 0.8426,
       "step": 17156
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.543913211250559e-06,
-      "loss": 0.7178,
+      "loss": 0.8753,
       "step": 17157
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.542807954886559e-06,
-      "loss": 0.8619,
+      "loss": 0.8993,
       "step": 17158
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.541702746486189e-06,
-      "loss": 0.7214,
+      "loss": 0.8482,
       "step": 17159
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.54059758606479e-06,
-      "loss": 0.7575,
+      "loss": 0.8485,
       "step": 17160
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.539492473637688e-06,
-      "loss": 0.8263,
+      "loss": 0.9622,
       "step": 17161
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.538387409220224e-06,
-      "loss": 0.8146,
+      "loss": 0.9155,
       "step": 17162
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.537282392827715e-06,
-      "loss": 0.8048,
+      "loss": 0.92,
       "step": 17163
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.536177424475507e-06,
-      "loss": 0.6273,
+      "loss": 0.8099,
       "step": 17164
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5350725041789185e-06,
-      "loss": 0.7008,
+      "loss": 0.8269,
       "step": 17165
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.533967631953286e-06,
-      "loss": 0.6122,
+      "loss": 0.7745,
       "step": 17166
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.532862807813931e-06,
-      "loss": 0.802,
+      "loss": 0.8403,
       "step": 17167
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5317580317761894e-06,
-      "loss": 0.9546,
+      "loss": 0.9803,
       "step": 17168
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.530653303855381e-06,
-      "loss": 0.77,
+      "loss": 0.9385,
       "step": 17169
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5295486240668395e-06,
-      "loss": 0.7558,
+      "loss": 0.7983,
       "step": 17170
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.528443992425886e-06,
-      "loss": 0.785,
+      "loss": 0.9127,
       "step": 17171
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.527339408947843e-06,
-      "loss": 0.6686,
+      "loss": 0.7573,
       "step": 17172
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.526234873648043e-06,
-      "loss": 0.6625,
+      "loss": 0.7651,
       "step": 17173
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.525130386541802e-06,
-      "loss": 0.552,
+      "loss": 0.8125,
       "step": 17174
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.52402594764445e-06,
-      "loss": 0.6044,
+      "loss": 0.7195,
       "step": 17175
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5229215569713025e-06,
-      "loss": 0.5935,
+      "loss": 0.7422,
       "step": 17176
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.521817214537689e-06,
-      "loss": 0.7238,
+      "loss": 0.8508,
       "step": 17177
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.520712920358923e-06,
-      "loss": 0.7242,
+      "loss": 0.8248,
       "step": 17178
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.519608674450332e-06,
-      "loss": 0.6708,
+      "loss": 0.8214,
       "step": 17179
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.51850447682723e-06,
-      "loss": 0.7559,
+      "loss": 0.8477,
       "step": 17180
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.517400327504939e-06,
-      "loss": 0.6952,
+      "loss": 0.831,
       "step": 17181
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.516296226498777e-06,
-      "loss": 0.6661,
+      "loss": 0.7261,
       "step": 17182
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5151921738240675e-06,
-      "loss": 0.6797,
+      "loss": 0.8383,
       "step": 17183
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.5140881694961154e-06,
-      "loss": 0.6811,
+      "loss": 0.7993,
       "step": 17184
     },
     {
       "epoch": 0.62,
       "learning_rate": 6.512984213530248e-06,
-      "loss": 0.6892,
+      "loss": 0.8326,
       "step": 17185
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.511880305941775e-06,
-      "loss": 0.7679,
+      "loss": 0.8293,
       "step": 17186
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.510776446746014e-06,
-      "loss": 0.7216,
+      "loss": 0.8105,
       "step": 17187
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.509672635958279e-06,
-      "loss": 0.5162,
+      "loss": 0.7298,
       "step": 17188
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.508568873593888e-06,
-      "loss": 0.6046,
+      "loss": 0.7724,
       "step": 17189
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.507465159668149e-06,
-      "loss": 0.7298,
+      "loss": 0.8527,
       "step": 17190
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.506361494196371e-06,
-      "loss": 0.5726,
+      "loss": 0.7662,
       "step": 17191
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.5052578771938754e-06,
-      "loss": 0.7748,
+      "loss": 0.8374,
       "step": 17192
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.5041543086759655e-06,
-      "loss": 0.8602,
+      "loss": 0.8622,
       "step": 17193
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.503050788657958e-06,
-      "loss": 0.6708,
+      "loss": 0.816,
       "step": 17194
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.501947317155153e-06,
-      "loss": 0.5823,
+      "loss": 0.6636,
       "step": 17195
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.5008438941828726e-06,
-      "loss": 0.8696,
+      "loss": 0.9378,
       "step": 17196
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.499740519756413e-06,
-      "loss": 0.7027,
+      "loss": 0.8489,
       "step": 17197
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.498637193891091e-06,
-      "loss": 0.6888,
+      "loss": 0.7858,
       "step": 17198
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.497533916602208e-06,
-      "loss": 0.6216,
+      "loss": 0.8016,
       "step": 17199
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.496430687905074e-06,
-      "loss": 0.6572,
+      "loss": 0.8039,
       "step": 17200
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4953275078149925e-06,
-      "loss": 0.6475,
+      "loss": 0.7622,
       "step": 17201
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4942243763472716e-06,
-      "loss": 0.6339,
+      "loss": 0.8285,
       "step": 17202
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.493121293517208e-06,
-      "loss": 0.672,
+      "loss": 0.8139,
       "step": 17203
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4920182593401186e-06,
-      "loss": 0.6651,
+      "loss": 0.8113,
       "step": 17204
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.490915273831293e-06,
-      "loss": 0.7858,
+      "loss": 0.8362,
       "step": 17205
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4898123370060405e-06,
-      "loss": 0.7932,
+      "loss": 0.7952,
       "step": 17206
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.488709448879661e-06,
-      "loss": 0.6529,
+      "loss": 0.7651,
       "step": 17207
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.487606609467457e-06,
-      "loss": 0.6414,
+      "loss": 0.7921,
       "step": 17208
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.486503818784731e-06,
-      "loss": 0.604,
+      "loss": 0.7597,
       "step": 17209
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4854010768467745e-06,
-      "loss": 0.7753,
+      "loss": 0.9,
       "step": 17210
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.484298383668895e-06,
-      "loss": 0.5989,
+      "loss": 0.8217,
       "step": 17211
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.483195739266383e-06,
-      "loss": 0.7384,
+      "loss": 0.8812,
       "step": 17212
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.482093143654544e-06,
-      "loss": 0.6992,
+      "loss": 0.8327,
       "step": 17213
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.480990596848668e-06,
-      "loss": 0.8176,
+      "loss": 0.8278,
       "step": 17214
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.47988809886406e-06,
-      "loss": 0.7331,
+      "loss": 0.8105,
       "step": 17215
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.478785649716006e-06,
-      "loss": 0.8244,
+      "loss": 0.8568,
       "step": 17216
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.477683249419807e-06,
-      "loss": 0.6626,
+      "loss": 0.8475,
       "step": 17217
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.476580897990754e-06,
-      "loss": 0.6574,
+      "loss": 0.8707,
       "step": 17218
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.475478595444144e-06,
-      "loss": 0.7899,
+      "loss": 0.858,
       "step": 17219
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.474376341795265e-06,
-      "loss": 0.7583,
+      "loss": 0.8081,
       "step": 17220
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.473274137059415e-06,
-      "loss": 0.8211,
+      "loss": 0.9002,
       "step": 17221
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.472171981251878e-06,
-      "loss": 0.6753,
+      "loss": 0.7101,
       "step": 17222
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.471069874387957e-06,
-      "loss": 0.6794,
+      "loss": 0.7617,
       "step": 17223
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.469967816482927e-06,
-      "loss": 0.6105,
+      "loss": 0.7595,
       "step": 17224
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4688658075520895e-06,
-      "loss": 0.8114,
+      "loss": 0.829,
       "step": 17225
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.467763847610726e-06,
-      "loss": 0.6781,
+      "loss": 0.8224,
       "step": 17226
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.46666193667413e-06,
-      "loss": 0.8089,
+      "loss": 0.9539,
       "step": 17227
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.465560074757589e-06,
-      "loss": 0.7372,
+      "loss": 0.797,
       "step": 17228
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.464458261876383e-06,
-      "loss": 0.8087,
+      "loss": 0.8776,
       "step": 17229
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.463356498045803e-06,
-      "loss": 0.7009,
+      "loss": 0.8184,
       "step": 17230
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4622547832811335e-06,
-      "loss": 0.6151,
+      "loss": 0.7792,
       "step": 17231
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.461153117597661e-06,
-      "loss": 0.8295,
+      "loss": 0.8923,
       "step": 17232
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4600515010106645e-06,
-      "loss": 0.663,
+      "loss": 0.8369,
       "step": 17233
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4589499335354365e-06,
-      "loss": 0.6922,
+      "loss": 0.7978,
       "step": 17234
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.457848415187248e-06,
-      "loss": 0.7414,
+      "loss": 0.8247,
       "step": 17235
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4567469459813934e-06,
-      "loss": 0.7167,
+      "loss": 0.8172,
       "step": 17236
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.455645525933143e-06,
-      "loss": 0.7006,
+      "loss": 0.8295,
       "step": 17237
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.454544155057784e-06,
-      "loss": 0.5585,
+      "loss": 0.6831,
       "step": 17238
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.453442833370592e-06,
-      "loss": 0.735,
+      "loss": 0.8653,
       "step": 17239
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.452341560886852e-06,
-      "loss": 0.7761,
+      "loss": 0.8687,
       "step": 17240
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.451240337621837e-06,
-      "loss": 0.7994,
+      "loss": 0.815,
       "step": 17241
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.450139163590832e-06,
-      "loss": 0.6089,
+      "loss": 0.7504,
       "step": 17242
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.449038038809103e-06,
-      "loss": 0.6914,
+      "loss": 0.7782,
       "step": 17243
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.447936963291938e-06,
-      "loss": 0.8287,
+      "loss": 0.8362,
       "step": 17244
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.446835937054604e-06,
-      "loss": 0.5992,
+      "loss": 0.7343,
       "step": 17245
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.445734960112383e-06,
-      "loss": 0.6325,
+      "loss": 0.8015,
       "step": 17246
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.444634032480548e-06,
-      "loss": 0.7342,
+      "loss": 0.8392,
       "step": 17247
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.443533154174366e-06,
-      "loss": 0.7101,
+      "loss": 0.7408,
       "step": 17248
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.442432325209121e-06,
-      "loss": 0.7843,
+      "loss": 0.8432,
       "step": 17249
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4413315456000754e-06,
-      "loss": 0.7427,
+      "loss": 0.9188,
       "step": 17250
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.440230815362509e-06,
-      "loss": 0.7397,
+      "loss": 0.765,
       "step": 17251
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.439130134511686e-06,
-      "loss": 0.5631,
+      "loss": 0.7441,
       "step": 17252
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.438029503062884e-06,
-      "loss": 0.7541,
+      "loss": 0.8214,
       "step": 17253
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.436928921031366e-06,
-      "loss": 0.6678,
+      "loss": 0.8314,
       "step": 17254
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.435828388432408e-06,
-      "loss": 0.7393,
+      "loss": 0.8535,
       "step": 17255
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4347279052812695e-06,
-      "loss": 0.8172,
+      "loss": 0.8647,
       "step": 17256
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4336274715932245e-06,
-      "loss": 0.7063,
+      "loss": 0.8859,
       "step": 17257
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.432527087383538e-06,
-      "loss": 0.792,
+      "loss": 0.8229,
       "step": 17258
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.431426752667478e-06,
-      "loss": 0.6256,
+      "loss": 0.7169,
       "step": 17259
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.430326467460306e-06,
-      "loss": 0.7453,
+      "loss": 0.8764,
       "step": 17260
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.429226231777293e-06,
-      "loss": 0.8189,
+      "loss": 0.8323,
       "step": 17261
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.428126045633695e-06,
-      "loss": 0.72,
+      "loss": 0.8658,
       "step": 17262
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.427025909044786e-06,
-      "loss": 0.6576,
+      "loss": 0.7551,
       "step": 17263
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.425925822025818e-06,
-      "loss": 0.6936,
+      "loss": 0.7724,
       "step": 17264
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.424825784592061e-06,
-      "loss": 0.6204,
+      "loss": 0.8347,
       "step": 17265
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.423725796758777e-06,
-      "loss": 0.7442,
+      "loss": 0.7935,
       "step": 17266
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.422625858541216e-06,
-      "loss": 0.8505,
+      "loss": 0.9312,
       "step": 17267
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.421525969954653e-06,
-      "loss": 0.6748,
+      "loss": 0.8229,
       "step": 17268
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.420426131014335e-06,
-      "loss": 0.7097,
+      "loss": 0.8466,
       "step": 17269
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.419326341735528e-06,
-      "loss": 0.8085,
+      "loss": 0.9047,
       "step": 17270
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.418226602133485e-06,
-      "loss": 0.7065,
+      "loss": 0.7812,
       "step": 17271
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.417126912223468e-06,
-      "loss": 0.9244,
+      "loss": 0.9619,
       "step": 17272
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.416027272020731e-06,
-      "loss": 0.8033,
+      "loss": 0.7954,
       "step": 17273
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4149276815405325e-06,
-      "loss": 0.6547,
+      "loss": 0.7382,
       "step": 17274
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.413828140798121e-06,
-      "loss": 0.6042,
+      "loss": 0.7573,
       "step": 17275
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.412728649808763e-06,
-      "loss": 0.7599,
+      "loss": 0.8838,
       "step": 17276
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.4116292085877e-06,
-      "loss": 0.7344,
+      "loss": 0.7626,
       "step": 17277
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.410529817150193e-06,
-      "loss": 0.766,
+      "loss": 0.8665,
       "step": 17278
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.40943047551149e-06,
-      "loss": 0.6664,
+      "loss": 0.7868,
       "step": 17279
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.408331183686847e-06,
-      "loss": 0.7004,
+      "loss": 0.7847,
       "step": 17280
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.40723194169151e-06,
-      "loss": 0.7277,
+      "loss": 0.8332,
       "step": 17281
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.406132749540737e-06,
-      "loss": 0.6909,
+      "loss": 0.8156,
       "step": 17282
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.405033607249769e-06,
-      "loss": 0.7211,
+      "loss": 0.8019,
       "step": 17283
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.40393451483386e-06,
-      "loss": 0.7205,
+      "loss": 0.7709,
       "step": 17284
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.40283547230826e-06,
-      "loss": 0.6143,
+      "loss": 0.8104,
       "step": 17285
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.40173647968821e-06,
-      "loss": 0.7978,
+      "loss": 0.8314,
       "step": 17286
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.400637536988965e-06,
-      "loss": 0.9114,
+      "loss": 1.0098,
       "step": 17287
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.399538644225764e-06,
-      "loss": 0.8375,
+      "loss": 0.9005,
       "step": 17288
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.398439801413858e-06,
-      "loss": 0.7064,
+      "loss": 0.7437,
       "step": 17289
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.397341008568488e-06,
-      "loss": 0.6353,
+      "loss": 0.7645,
       "step": 17290
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.396242265704902e-06,
-      "loss": 0.7994,
+      "loss": 0.872,
       "step": 17291
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.395143572838341e-06,
-      "loss": 0.6539,
+      "loss": 0.7719,
       "step": 17292
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.394044929984046e-06,
-      "loss": 0.6423,
+      "loss": 0.7981,
       "step": 17293
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.392946337157263e-06,
-      "loss": 0.6409,
+      "loss": 0.7657,
       "step": 17294
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.391847794373236e-06,
-      "loss": 0.6956,
+      "loss": 0.8205,
       "step": 17295
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3907493016471955e-06,
-      "loss": 0.7955,
+      "loss": 0.8682,
       "step": 17296
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.389650858994391e-06,
-      "loss": 0.7448,
+      "loss": 0.8507,
       "step": 17297
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.388552466430056e-06,
-      "loss": 0.7362,
+      "loss": 0.803,
       "step": 17298
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.387454123969433e-06,
-      "loss": 0.7568,
+      "loss": 0.8191,
       "step": 17299
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.386355831627757e-06,
-      "loss": 0.7723,
+      "loss": 0.8011,
       "step": 17300
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3852575894202705e-06,
-      "loss": 0.736,
+      "loss": 0.9164,
       "step": 17301
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3841593973622e-06,
-      "loss": 0.7863,
+      "loss": 0.8961,
       "step": 17302
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.383061255468794e-06,
-      "loss": 0.69,
+      "loss": 0.7909,
       "step": 17303
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.38196316375528e-06,
-      "loss": 0.5957,
+      "loss": 0.6882,
       "step": 17304
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.380865122236892e-06,
-      "loss": 0.6818,
+      "loss": 0.8164,
       "step": 17305
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3797671309288685e-06,
-      "loss": 0.7345,
+      "loss": 0.83,
       "step": 17306
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.378669189846434e-06,
-      "loss": 0.6814,
+      "loss": 0.7666,
       "step": 17307
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.377571299004832e-06,
-      "loss": 0.7608,
+      "loss": 0.8188,
       "step": 17308
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3764734584192855e-06,
-      "loss": 0.7247,
+      "loss": 0.7454,
       "step": 17309
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.375375668105031e-06,
-      "loss": 0.7574,
+      "loss": 0.9107,
       "step": 17310
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.374277928077292e-06,
-      "loss": 0.4998,
+      "loss": 0.6913,
       "step": 17311
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.373180238351306e-06,
-      "loss": 0.7249,
+      "loss": 0.8633,
       "step": 17312
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.372082598942297e-06,
-      "loss": 0.7695,
+      "loss": 0.8935,
       "step": 17313
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.370985009865498e-06,
-      "loss": 0.7635,
+      "loss": 0.8528,
       "step": 17314
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.369887471136131e-06,
-      "loss": 0.7013,
+      "loss": 0.7802,
       "step": 17315
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.368789982769425e-06,
-      "loss": 0.8026,
+      "loss": 0.8597,
       "step": 17316
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.367692544780604e-06,
-      "loss": 0.6466,
+      "loss": 0.739,
       "step": 17317
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.366595157184896e-06,
-      "loss": 0.8903,
+      "loss": 0.9166,
       "step": 17318
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.365497819997526e-06,
-      "loss": 0.7178,
+      "loss": 0.8654,
       "step": 17319
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3644005332337164e-06,
-      "loss": 0.7045,
+      "loss": 0.8454,
       "step": 17320
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.363303296908692e-06,
-      "loss": 0.7944,
+      "loss": 0.8521,
       "step": 17321
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3622061110376765e-06,
-      "loss": 0.7622,
+      "loss": 0.8643,
       "step": 17322
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.361108975635888e-06,
-      "loss": 0.677,
+      "loss": 0.8772,
       "step": 17323
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.360011890718547e-06,
-      "loss": 0.6517,
+      "loss": 0.8206,
       "step": 17324
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3589148563008796e-06,
-      "loss": 0.7906,
+      "loss": 0.8076,
       "step": 17325
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.357817872398101e-06,
-      "loss": 0.7449,
+      "loss": 0.7988,
       "step": 17326
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.356720939025436e-06,
-      "loss": 0.7658,
+      "loss": 0.8552,
       "step": 17327
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.355624056198093e-06,
-      "loss": 0.6287,
+      "loss": 0.8055,
       "step": 17328
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.354527223931299e-06,
-      "loss": 0.7645,
+      "loss": 0.8376,
       "step": 17329
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.353430442240265e-06,
-      "loss": 0.6966,
+      "loss": 0.8106,
       "step": 17330
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.352333711140212e-06,
-      "loss": 0.6986,
+      "loss": 0.7768,
       "step": 17331
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3512370306463515e-06,
-      "loss": 0.6724,
+      "loss": 0.725,
       "step": 17332
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.350140400773903e-06,
-      "loss": 0.816,
+      "loss": 0.8505,
       "step": 17333
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.349043821538074e-06,
-      "loss": 0.6999,
+      "loss": 0.7955,
       "step": 17334
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.347947292954087e-06,
-      "loss": 0.7908,
+      "loss": 0.8812,
       "step": 17335
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.346850815037145e-06,
-      "loss": 0.7276,
+      "loss": 0.8246,
       "step": 17336
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.345754387802467e-06,
-      "loss": 0.7039,
+      "loss": 0.8395,
       "step": 17337
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.344658011265261e-06,
-      "loss": 0.8446,
+      "loss": 0.9403,
       "step": 17338
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.343561685440741e-06,
-      "loss": 0.6654,
+      "loss": 0.763,
       "step": 17339
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.342465410344112e-06,
-      "loss": 0.7576,
+      "loss": 0.819,
       "step": 17340
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.341369185990591e-06,
-      "loss": 0.799,
+      "loss": 0.8648,
       "step": 17341
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.340273012395379e-06,
-      "loss": 0.6981,
+      "loss": 0.83,
       "step": 17342
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.339176889573684e-06,
-      "loss": 0.7204,
+      "loss": 0.838,
       "step": 17343
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.338080817540719e-06,
-      "loss": 0.757,
+      "loss": 0.8635,
       "step": 17344
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3369847963116835e-06,
-      "loss": 0.7304,
+      "loss": 0.7808,
       "step": 17345
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.335888825901792e-06,
-      "loss": 0.7538,
+      "loss": 0.8156,
       "step": 17346
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.334792906326237e-06,
-      "loss": 0.7422,
+      "loss": 0.7868,
       "step": 17347
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.333697037600237e-06,
-      "loss": 0.6639,
+      "loss": 0.8681,
       "step": 17348
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.332601219738985e-06,
-      "loss": 0.8312,
+      "loss": 0.9354,
       "step": 17349
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.331505452757691e-06,
-      "loss": 0.7863,
+      "loss": 0.8154,
       "step": 17350
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.33040973667155e-06,
-      "loss": 0.7876,
+      "loss": 0.9088,
       "step": 17351
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.329314071495769e-06,
-      "loss": 0.7153,
+      "loss": 0.797,
       "step": 17352
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.328218457245547e-06,
-      "loss": 0.6839,
+      "loss": 0.8191,
       "step": 17353
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.327122893936087e-06,
-      "loss": 0.818,
+      "loss": 0.8482,
       "step": 17354
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.326027381582583e-06,
-      "loss": 0.7508,
+      "loss": 0.8324,
       "step": 17355
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.324931920200237e-06,
-      "loss": 0.6678,
+      "loss": 0.7957,
       "step": 17356
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.323836509804244e-06,
-      "loss": 0.7175,
+      "loss": 0.8165,
       "step": 17357
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.322741150409808e-06,
-      "loss": 0.661,
+      "loss": 0.8806,
       "step": 17358
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.3216458420321185e-06,
-      "loss": 0.7374,
+      "loss": 0.81,
       "step": 17359
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.320550584686376e-06,
-      "loss": 0.7983,
+      "loss": 0.9237,
       "step": 17360
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.319455378387773e-06,
-      "loss": 0.7673,
+      "loss": 0.808,
       "step": 17361
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.318360223151503e-06,
-      "loss": 0.7747,
+      "loss": 0.8197,
       "step": 17362
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.317265118992764e-06,
-      "loss": 0.6947,
+      "loss": 0.821,
       "step": 17363
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.316170065926742e-06,
-      "loss": 0.6628,
+      "loss": 0.7119,
       "step": 17364
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.315075063968636e-06,
-      "loss": 0.7406,
+      "loss": 0.7734,
       "step": 17365
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.313980113133636e-06,
-      "loss": 0.7664,
+      "loss": 0.8352,
       "step": 17366
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.312885213436934e-06,
-      "loss": 0.8362,
+      "loss": 0.9047,
       "step": 17367
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.311790364893715e-06,
-      "loss": 0.7232,
+      "loss": 0.825,
       "step": 17368
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.310695567519174e-06,
-      "loss": 0.7595,
+      "loss": 0.837,
       "step": 17369
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.309600821328495e-06,
-      "loss": 0.7109,
+      "loss": 0.8388,
       "step": 17370
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.308506126336871e-06,
-      "loss": 0.6149,
+      "loss": 0.6927,
       "step": 17371
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.307411482559484e-06,
-      "loss": 0.6358,
+      "loss": 0.728,
       "step": 17372
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.306316890011528e-06,
-      "loss": 0.6356,
+      "loss": 0.7749,
       "step": 17373
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.305222348708178e-06,
-      "loss": 0.7211,
+      "loss": 0.8419,
       "step": 17374
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.304127858664632e-06,
-      "loss": 0.7828,
+      "loss": 0.8536,
       "step": 17375
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.303033419896065e-06,
-      "loss": 0.774,
+      "loss": 0.8478,
       "step": 17376
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.301939032417667e-06,
-      "loss": 0.7226,
+      "loss": 0.7772,
       "step": 17377
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.300844696244615e-06,
-      "loss": 0.6596,
+      "loss": 0.8182,
       "step": 17378
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.299750411392096e-06,
-      "loss": 0.751,
+      "loss": 0.8934,
       "step": 17379
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2986561778752875e-06,
-      "loss": 0.6358,
+      "loss": 0.761,
       "step": 17380
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.297561995709378e-06,
-      "loss": 0.6791,
+      "loss": 0.8192,
       "step": 17381
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.29646786490954e-06,
-      "loss": 0.8002,
+      "loss": 0.8989,
       "step": 17382
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.295373785490955e-06,
-      "loss": 0.7709,
+      "loss": 0.8353,
       "step": 17383
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.294279757468803e-06,
-      "loss": 0.7269,
+      "loss": 0.7853,
       "step": 17384
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.293185780858261e-06,
-      "loss": 0.7565,
+      "loss": 0.8577,
       "step": 17385
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.29209185567451e-06,
-      "loss": 0.692,
+      "loss": 0.8255,
       "step": 17386
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.29099798193272e-06,
-      "loss": 0.7331,
+      "loss": 0.8023,
       "step": 17387
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.289904159648073e-06,
-      "loss": 0.7019,
+      "loss": 0.8537,
       "step": 17388
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.28881038883574e-06,
-      "loss": 0.7078,
+      "loss": 0.8186,
       "step": 17389
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.287716669510898e-06,
-      "loss": 0.6965,
+      "loss": 0.7745,
       "step": 17390
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.286623001688718e-06,
-      "loss": 0.7548,
+      "loss": 0.8324,
       "step": 17391
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.285529385384379e-06,
-      "loss": 0.686,
+      "loss": 0.9034,
       "step": 17392
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.284435820613047e-06,
-      "loss": 0.8558,
+      "loss": 0.9426,
       "step": 17393
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.283342307389901e-06,
-      "loss": 0.7566,
+      "loss": 0.869,
       "step": 17394
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.282248845730102e-06,
-      "loss": 0.7487,
+      "loss": 0.8486,
       "step": 17395
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.281155435648829e-06,
-      "loss": 0.7461,
+      "loss": 0.9022,
       "step": 17396
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.280062077161244e-06,
-      "loss": 0.708,
+      "loss": 0.8082,
       "step": 17397
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.278968770282523e-06,
-      "loss": 0.7334,
+      "loss": 0.8779,
       "step": 17398
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.277875515027829e-06,
-      "loss": 0.7947,
+      "loss": 0.8149,
       "step": 17399
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.276782311412335e-06,
-      "loss": 0.634,
+      "loss": 0.7167,
       "step": 17400
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.275689159451202e-06,
-      "loss": 0.6575,
+      "loss": 0.7907,
       "step": 17401
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.274596059159595e-06,
-      "loss": 0.752,
+      "loss": 0.8432,
       "step": 17402
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.273503010552684e-06,
-      "loss": 0.7974,
+      "loss": 0.8301,
       "step": 17403
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.27241001364563e-06,
-      "loss": 0.7554,
+      "loss": 0.8251,
       "step": 17404
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2713170684536015e-06,
-      "loss": 0.6945,
+      "loss": 0.7085,
       "step": 17405
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.270224174991755e-06,
-      "loss": 0.6853,
+      "loss": 0.8021,
       "step": 17406
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.26913133327526e-06,
-      "loss": 0.8417,
+      "loss": 0.9074,
       "step": 17407
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.26803854331927e-06,
-      "loss": 0.7383,
+      "loss": 0.8851,
       "step": 17408
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.266945805138953e-06,
-      "loss": 0.615,
+      "loss": 0.6888,
       "step": 17409
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.265853118749466e-06,
-      "loss": 0.8414,
+      "loss": 0.9637,
       "step": 17410
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2647604841659684e-06,
-      "loss": 0.6968,
+      "loss": 0.7625,
       "step": 17411
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.263667901403618e-06,
-      "loss": 0.6934,
+      "loss": 0.7862,
       "step": 17412
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.262575370477579e-06,
-      "loss": 0.6296,
+      "loss": 0.8161,
       "step": 17413
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.261482891402997e-06,
-      "loss": 0.7316,
+      "loss": 0.8502,
       "step": 17414
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.260390464195043e-06,
-      "loss": 0.7597,
+      "loss": 0.8403,
       "step": 17415
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.259298088868861e-06,
-      "loss": 0.7192,
+      "loss": 0.8244,
       "step": 17416
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.258205765439612e-06,
-      "loss": 0.6808,
+      "loss": 0.8387,
       "step": 17417
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.257113493922449e-06,
-      "loss": 0.6988,
+      "loss": 0.8832,
       "step": 17418
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2560212743325254e-06,
-      "loss": 0.7935,
+      "loss": 0.8437,
       "step": 17419
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.254929106684997e-06,
-      "loss": 0.7345,
+      "loss": 0.8146,
       "step": 17420
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.25383699099501e-06,
-      "loss": 0.6456,
+      "loss": 0.819,
       "step": 17421
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.252744927277723e-06,
-      "loss": 0.7496,
+      "loss": 0.838,
       "step": 17422
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.25165291554828e-06,
-      "loss": 0.8665,
+      "loss": 0.8923,
       "step": 17423
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.250560955821837e-06,
-      "loss": 0.6595,
+      "loss": 0.8122,
       "step": 17424
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.249469048113539e-06,
-      "loss": 0.7765,
+      "loss": 0.8958,
       "step": 17425
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.248377192438542e-06,
-      "loss": 0.6693,
+      "loss": 0.745,
       "step": 17426
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.247285388811982e-06,
-      "loss": 0.7765,
+      "loss": 0.783,
       "step": 17427
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.246193637249016e-06,
-      "loss": 0.7371,
+      "loss": 0.8271,
       "step": 17428
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2451019377647855e-06,
-      "loss": 0.8067,
+      "loss": 0.8576,
       "step": 17429
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.244010290374441e-06,
-      "loss": 0.7833,
+      "loss": 0.8667,
       "step": 17430
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.242918695093121e-06,
-      "loss": 0.7057,
+      "loss": 0.8332,
       "step": 17431
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.241827151935977e-06,
-      "loss": 0.634,
+      "loss": 0.7793,
       "step": 17432
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.240735660918144e-06,
-      "loss": 0.6917,
+      "loss": 0.8621,
       "step": 17433
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.239644222054777e-06,
-      "loss": 0.7527,
+      "loss": 0.8393,
       "step": 17434
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.238552835361007e-06,
-      "loss": 0.7691,
+      "loss": 0.863,
       "step": 17435
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.237461500851981e-06,
-      "loss": 0.8039,
+      "loss": 0.9116,
       "step": 17436
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2363702185428375e-06,
-      "loss": 0.7355,
+      "loss": 0.7688,
       "step": 17437
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.235278988448717e-06,
-      "loss": 0.766,
+      "loss": 0.8152,
       "step": 17438
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.234187810584763e-06,
-      "loss": 0.7143,
+      "loss": 0.7951,
       "step": 17439
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.233096684966105e-06,
-      "loss": 0.723,
+      "loss": 0.7655,
       "step": 17440
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.23200561160789e-06,
-      "loss": 0.8312,
+      "loss": 0.9248,
       "step": 17441
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2309145905252465e-06,
-      "loss": 0.7678,
+      "loss": 0.918,
       "step": 17442
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.229823621733319e-06,
-      "loss": 0.7061,
+      "loss": 0.7953,
       "step": 17443
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.228732705247236e-06,
-      "loss": 0.7251,
+      "loss": 0.7646,
       "step": 17444
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.227641841082142e-06,
-      "loss": 0.709,
+      "loss": 0.8472,
       "step": 17445
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.226551029253157e-06,
-      "loss": 0.6784,
+      "loss": 0.7898,
       "step": 17446
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2254602697754296e-06,
-      "loss": 0.7179,
+      "loss": 0.8285,
       "step": 17447
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2243695626640836e-06,
-      "loss": 0.6252,
+      "loss": 0.7557,
       "step": 17448
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.223278907934251e-06,
-      "loss": 0.6985,
+      "loss": 0.8727,
       "step": 17449
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.222188305601067e-06,
-      "loss": 0.7104,
+      "loss": 0.883,
       "step": 17450
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.221097755679659e-06,
-      "loss": 0.7902,
+      "loss": 0.8126,
       "step": 17451
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.220007258185159e-06,
-      "loss": 0.689,
+      "loss": 0.8069,
       "step": 17452
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.218916813132697e-06,
-      "loss": 0.7981,
+      "loss": 0.9111,
       "step": 17453
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.217826420537396e-06,
-      "loss": 0.5881,
+      "loss": 0.7094,
       "step": 17454
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.216736080414389e-06,
-      "loss": 0.6585,
+      "loss": 0.823,
       "step": 17455
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.215645792778801e-06,
-      "loss": 0.6769,
+      "loss": 0.7518,
       "step": 17456
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.214555557645759e-06,
-      "loss": 0.6004,
+      "loss": 0.7899,
       "step": 17457
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.21346537503039e-06,
-      "loss": 0.6586,
+      "loss": 0.813,
       "step": 17458
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.2123752449478125e-06,
-      "loss": 0.8073,
+      "loss": 0.9059,
       "step": 17459
     },
     {
       "epoch": 0.63,
       "learning_rate": 6.211285167413156e-06,
-      "loss": 0.7915,
+      "loss": 0.831,
       "step": 17460
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.210195142441541e-06,
-      "loss": 0.6943,
+      "loss": 0.8016,
       "step": 17461
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.209105170048093e-06,
-      "loss": 0.6897,
+      "loss": 0.7607,
       "step": 17462
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.20801525024793e-06,
-      "loss": 0.7801,
+      "loss": 0.897,
       "step": 17463
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.206925383056177e-06,
-      "loss": 0.6508,
+      "loss": 0.8295,
       "step": 17464
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.205835568487952e-06,
-      "loss": 0.6664,
+      "loss": 0.7936,
       "step": 17465
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.204745806558379e-06,
-      "loss": 0.6056,
+      "loss": 0.7356,
       "step": 17466
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.2036560972825675e-06,
-      "loss": 0.7947,
+      "loss": 0.8969,
       "step": 17467
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.202566440675645e-06,
-      "loss": 0.82,
+      "loss": 0.899,
       "step": 17468
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.201476836752723e-06,
-      "loss": 0.8114,
+      "loss": 0.9013,
       "step": 17469
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.2003872855289216e-06,
-      "loss": 0.7295,
+      "loss": 0.8064,
       "step": 17470
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.199297787019354e-06,
-      "loss": 0.7323,
+      "loss": 0.7698,
       "step": 17471
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.198208341239143e-06,
-      "loss": 0.8273,
+      "loss": 0.915,
       "step": 17472
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.197118948203389e-06,
-      "loss": 0.6418,
+      "loss": 0.719,
       "step": 17473
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.196029607927221e-06,
-      "loss": 0.6892,
+      "loss": 0.7792,
       "step": 17474
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.194940320425739e-06,
-      "loss": 0.8006,
+      "loss": 0.8775,
       "step": 17475
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.193851085714066e-06,
-      "loss": 0.7239,
+      "loss": 0.8755,
       "step": 17476
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.19276190380731e-06,
-      "loss": 0.7075,
+      "loss": 0.7726,
       "step": 17477
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.191672774720575e-06,
-      "loss": 0.8039,
+      "loss": 0.8284,
       "step": 17478
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.190583698468984e-06,
-      "loss": 0.605,
+      "loss": 0.7766,
       "step": 17479
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.189494675067633e-06,
-      "loss": 0.7062,
+      "loss": 0.8477,
       "step": 17480
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1884057045316405e-06,
-      "loss": 0.839,
+      "loss": 0.914,
       "step": 17481
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.187316786876109e-06,
-      "loss": 0.7332,
+      "loss": 0.801,
       "step": 17482
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.186227922116148e-06,
-      "loss": 0.7801,
+      "loss": 0.8371,
       "step": 17483
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.185139110266864e-06,
-      "loss": 0.692,
+      "loss": 0.8301,
       "step": 17484
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.184050351343365e-06,
-      "loss": 0.8213,
+      "loss": 0.8799,
       "step": 17485
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.182961645360747e-06,
-      "loss": 0.7487,
+      "loss": 0.8137,
       "step": 17486
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.181872992334126e-06,
-      "loss": 0.6226,
+      "loss": 0.7025,
       "step": 17487
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.180784392278597e-06,
-      "loss": 0.6573,
+      "loss": 0.8138,
       "step": 17488
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.179695845209268e-06,
-      "loss": 0.7596,
+      "loss": 0.8656,
       "step": 17489
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.178607351141236e-06,
-      "loss": 0.7581,
+      "loss": 0.8579,
       "step": 17490
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1775189100896075e-06,
-      "loss": 0.6703,
+      "loss": 0.782,
       "step": 17491
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.176430522069479e-06,
-      "loss": 0.7834,
+      "loss": 0.8943,
       "step": 17492
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.175342187095956e-06,
-      "loss": 0.7221,
+      "loss": 0.7905,
       "step": 17493
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.174253905184129e-06,
-      "loss": 0.7531,
+      "loss": 0.8367,
       "step": 17494
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.173165676349103e-06,
-      "loss": 0.6655,
+      "loss": 0.8418,
       "step": 17495
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.172077500605975e-06,
-      "loss": 0.7355,
+      "loss": 0.8323,
       "step": 17496
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.170989377969838e-06,
-      "loss": 0.6402,
+      "loss": 0.8652,
       "step": 17497
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1699013084557944e-06,
-      "loss": 0.7245,
+      "loss": 0.9306,
       "step": 17498
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.168813292078932e-06,
-      "loss": 0.6956,
+      "loss": 0.8541,
       "step": 17499
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.16772532885435e-06,
-      "loss": 0.7689,
+      "loss": 0.9211,
       "step": 17500
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.166637418797142e-06,
-      "loss": 0.8172,
+      "loss": 0.8635,
       "step": 17501
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.165549561922403e-06,
-      "loss": 0.7681,
+      "loss": 0.9027,
       "step": 17502
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.164461758245221e-06,
-      "loss": 0.7355,
+      "loss": 0.8412,
       "step": 17503
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.163374007780694e-06,
-      "loss": 0.8226,
+      "loss": 0.8733,
       "step": 17504
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.162286310543903e-06,
-      "loss": 0.7999,
+      "loss": 0.8527,
       "step": 17505
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.16119866654995e-06,
-      "loss": 0.8453,
+      "loss": 0.8918,
       "step": 17506
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.160111075813915e-06,
-      "loss": 0.6738,
+      "loss": 0.755,
       "step": 17507
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.159023538350893e-06,
-      "loss": 0.6988,
+      "loss": 0.8152,
       "step": 17508
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.157936054175968e-06,
-      "loss": 0.7284,
+      "loss": 0.8021,
       "step": 17509
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.156848623304232e-06,
-      "loss": 0.6475,
+      "loss": 0.7804,
       "step": 17510
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1557612457507655e-06,
-      "loss": 0.7417,
+      "loss": 0.7934,
       "step": 17511
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.154673921530661e-06,
-      "loss": 0.7157,
+      "loss": 0.7824,
       "step": 17512
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1535866506589945e-06,
-      "loss": 0.7035,
+      "loss": 0.8047,
       "step": 17513
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.152499433150862e-06,
-      "loss": 0.7094,
+      "loss": 0.886,
       "step": 17514
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.15141226902134e-06,
-      "loss": 0.6241,
+      "loss": 0.8066,
       "step": 17515
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.150325158285509e-06,
-      "loss": 0.8729,
+      "loss": 0.9121,
       "step": 17516
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.149238100958459e-06,
-      "loss": 0.7579,
+      "loss": 0.7631,
       "step": 17517
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.14815109705526e-06,
-      "loss": 0.7584,
+      "loss": 0.8298,
       "step": 17518
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.147064146591006e-06,
-      "loss": 0.7014,
+      "loss": 0.8313,
       "step": 17519
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1459772495807665e-06,
-      "loss": 0.7464,
+      "loss": 0.7657,
       "step": 17520
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.144890406039627e-06,
-      "loss": 0.7732,
+      "loss": 0.8308,
       "step": 17521
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.143803615982661e-06,
-      "loss": 0.744,
+      "loss": 0.8282,
       "step": 17522
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.14271687942495e-06,
-      "loss": 0.7217,
+      "loss": 0.8315,
       "step": 17523
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1416301963815685e-06,
-      "loss": 0.7579,
+      "loss": 0.8326,
       "step": 17524
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.140543566867597e-06,
-      "loss": 0.8002,
+      "loss": 0.9222,
       "step": 17525
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.139456990898105e-06,
-      "loss": 0.6433,
+      "loss": 0.7401,
       "step": 17526
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.138370468488171e-06,
-      "loss": 0.7723,
+      "loss": 0.8565,
       "step": 17527
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.137283999652864e-06,
-      "loss": 0.8621,
+      "loss": 0.8889,
       "step": 17528
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.136197584407265e-06,
-      "loss": 0.8051,
+      "loss": 0.8589,
       "step": 17529
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.135111222766441e-06,
-      "loss": 0.597,
+      "loss": 0.7864,
       "step": 17530
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.134024914745466e-06,
-      "loss": 0.6687,
+      "loss": 0.7477,
       "step": 17531
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.132938660359409e-06,
-      "loss": 0.6749,
+      "loss": 0.8205,
       "step": 17532
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.131852459623344e-06,
-      "loss": 0.7528,
+      "loss": 0.8638,
       "step": 17533
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.130766312552336e-06,
-      "loss": 0.7917,
+      "loss": 0.8949,
       "step": 17534
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.129680219161456e-06,
-      "loss": 0.7341,
+      "loss": 0.8351,
       "step": 17535
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.128594179465772e-06,
-      "loss": 0.6981,
+      "loss": 0.8417,
       "step": 17536
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.12750819348035e-06,
-      "loss": 0.6691,
+      "loss": 0.8107,
       "step": 17537
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1264222612202596e-06,
-      "loss": 0.5904,
+      "loss": 0.6579,
       "step": 17538
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.125336382700562e-06,
-      "loss": 0.5891,
+      "loss": 0.755,
       "step": 17539
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.124250557936326e-06,
-      "loss": 0.7065,
+      "loss": 0.8149,
       "step": 17540
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.123164786942612e-06,
-      "loss": 0.6498,
+      "loss": 0.7203,
       "step": 17541
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.12207906973449e-06,
-      "loss": 0.7244,
+      "loss": 0.7505,
       "step": 17542
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.120993406327016e-06,
-      "loss": 0.5931,
+      "loss": 0.7848,
       "step": 17543
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.119907796735258e-06,
-      "loss": 0.7566,
+      "loss": 0.8878,
       "step": 17544
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.118822240974269e-06,
-      "loss": 0.6515,
+      "loss": 0.7496,
       "step": 17545
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1177367390591215e-06,
-      "loss": 0.7846,
+      "loss": 0.8902,
       "step": 17546
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.116651291004863e-06,
-      "loss": 0.7268,
+      "loss": 0.9452,
       "step": 17547
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1155658968265605e-06,
-      "loss": 0.7276,
+      "loss": 0.7961,
       "step": 17548
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1144805565392685e-06,
-      "loss": 0.7194,
+      "loss": 0.7958,
       "step": 17549
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.113395270158047e-06,
-      "loss": 0.6853,
+      "loss": 0.7656,
       "step": 17550
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.112310037697951e-06,
-      "loss": 0.624,
+      "loss": 0.7496,
       "step": 17551
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1112248591740395e-06,
-      "loss": 0.8106,
+      "loss": 0.8313,
       "step": 17552
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.110139734601364e-06,
-      "loss": 0.8056,
+      "loss": 0.9053,
       "step": 17553
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.10905466399498e-06,
-      "loss": 0.7223,
+      "loss": 0.8062,
       "step": 17554
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.107969647369944e-06,
-      "loss": 0.7318,
+      "loss": 0.8695,
       "step": 17555
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.1068846847413044e-06,
-      "loss": 0.7548,
+      "loss": 0.842,
       "step": 17556
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.105799776124119e-06,
-      "loss": 0.6921,
+      "loss": 0.8339,
       "step": 17557
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.104714921533432e-06,
-      "loss": 0.6962,
+      "loss": 0.8756,
       "step": 17558
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.103630120984305e-06,
-      "loss": 0.7355,
+      "loss": 0.7836,
       "step": 17559
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.102545374491777e-06,
-      "loss": 0.5689,
+      "loss": 0.6941,
       "step": 17560
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.101460682070905e-06,
-      "loss": 0.5697,
+      "loss": 0.7065,
       "step": 17561
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.100376043736733e-06,
-      "loss": 0.7703,
+      "loss": 0.8222,
       "step": 17562
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.099291459504311e-06,
-      "loss": 0.6181,
+      "loss": 0.7614,
       "step": 17563
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.098206929388686e-06,
-      "loss": 0.7112,
+      "loss": 0.7846,
       "step": 17564
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.097122453404906e-06,
-      "loss": 0.7907,
+      "loss": 0.8751,
       "step": 17565
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.09603803156801e-06,
-      "loss": 0.7244,
+      "loss": 0.8416,
       "step": 17566
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.094953663893051e-06,
-      "loss": 0.7222,
+      "loss": 0.7957,
       "step": 17567
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0938693503950665e-06,
-      "loss": 0.6816,
+      "loss": 0.8668,
       "step": 17568
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.092785091089104e-06,
-      "loss": 0.8457,
+      "loss": 0.8409,
       "step": 17569
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.091700885990204e-06,
-      "loss": 0.8031,
+      "loss": 0.8456,
       "step": 17570
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.090616735113413e-06,
-      "loss": 0.6947,
+      "loss": 0.7663,
       "step": 17571
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.089532638473765e-06,
-      "loss": 0.8306,
+      "loss": 0.8897,
       "step": 17572
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.088448596086303e-06,
-      "loss": 0.6774,
+      "loss": 0.7909,
       "step": 17573
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.087364607966069e-06,
-      "loss": 0.6851,
+      "loss": 0.8152,
       "step": 17574
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.086280674128096e-06,
-      "loss": 0.7328,
+      "loss": 0.7686,
       "step": 17575
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.08519679458743e-06,
-      "loss": 0.6327,
+      "loss": 0.7699,
       "step": 17576
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0841129693590995e-06,
-      "loss": 0.726,
+      "loss": 0.8586,
       "step": 17577
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.08302919845815e-06,
-      "loss": 0.7124,
+      "loss": 0.8108,
       "step": 17578
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.08194548189961e-06,
-      "loss": 0.7655,
+      "loss": 0.9097,
       "step": 17579
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.08086181969852e-06,
-      "loss": 0.8103,
+      "loss": 0.9679,
       "step": 17580
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.079778211869909e-06,
-      "loss": 0.8219,
+      "loss": 0.8703,
       "step": 17581
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0786946584288145e-06,
-      "loss": 0.8217,
+      "loss": 0.9066,
       "step": 17582
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.077611159390267e-06,
-      "loss": 0.6382,
+      "loss": 0.8023,
       "step": 17583
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.076527714769304e-06,
-      "loss": 0.7052,
+      "loss": 0.7622,
       "step": 17584
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.075444324580945e-06,
-      "loss": 0.8278,
+      "loss": 0.8703,
       "step": 17585
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.074360988840234e-06,
-      "loss": 0.6976,
+      "loss": 0.8577,
       "step": 17586
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.073277707562191e-06,
-      "loss": 0.6147,
+      "loss": 0.7814,
       "step": 17587
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.072194480761851e-06,
-      "loss": 0.6901,
+      "loss": 0.7887,
       "step": 17588
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.071111308454236e-06,
-      "loss": 0.9181,
+      "loss": 0.9612,
       "step": 17589
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.07002819065438e-06,
-      "loss": 0.6,
+      "loss": 0.7208,
       "step": 17590
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.06894512737731e-06,
-      "loss": 0.8574,
+      "loss": 0.8481,
       "step": 17591
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.067862118638044e-06,
-      "loss": 0.7318,
+      "loss": 0.7759,
       "step": 17592
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0667791644516146e-06,
-      "loss": 0.7589,
+      "loss": 0.8972,
       "step": 17593
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.065696264833042e-06,
-      "loss": 0.6224,
+      "loss": 0.7728,
       "step": 17594
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.064613419797353e-06,
-      "loss": 0.6464,
+      "loss": 0.7519,
       "step": 17595
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.063530629359568e-06,
-      "loss": 0.802,
+      "loss": 0.925,
       "step": 17596
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.062447893534713e-06,
-      "loss": 0.7911,
+      "loss": 0.7746,
       "step": 17597
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.061365212337804e-06,
-      "loss": 0.6711,
+      "loss": 0.8238,
       "step": 17598
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0602825857838655e-06,
-      "loss": 0.7397,
+      "loss": 0.7941,
       "step": 17599
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.059200013887916e-06,
-      "loss": 0.7257,
+      "loss": 0.8128,
       "step": 17600
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0581174966649755e-06,
-      "loss": 0.8017,
+      "loss": 0.8574,
       "step": 17601
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.057035034130061e-06,
-      "loss": 0.8235,
+      "loss": 0.8151,
       "step": 17602
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.055952626298191e-06,
-      "loss": 0.8159,
+      "loss": 0.8988,
       "step": 17603
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.054870273184383e-06,
-      "loss": 0.7117,
+      "loss": 0.8367,
       "step": 17604
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.053787974803656e-06,
-      "loss": 0.905,
+      "loss": 0.8777,
       "step": 17605
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.052705731171018e-06,
-      "loss": 0.7745,
+      "loss": 0.7952,
       "step": 17606
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.051623542301488e-06,
-      "loss": 0.7106,
+      "loss": 0.8101,
       "step": 17607
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.050541408210078e-06,
-      "loss": 0.7346,
+      "loss": 0.8069,
       "step": 17608
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.049459328911806e-06,
-      "loss": 0.6487,
+      "loss": 0.8201,
       "step": 17609
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.048377304421681e-06,
-      "loss": 0.7469,
+      "loss": 0.8335,
       "step": 17610
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0472953347547105e-06,
-      "loss": 0.7882,
+      "loss": 0.8643,
       "step": 17611
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0462134199259115e-06,
-      "loss": 0.6439,
+      "loss": 0.8248,
       "step": 17612
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.045131559950289e-06,
-      "loss": 0.7186,
+      "loss": 0.853,
       "step": 17613
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.044049754842858e-06,
-      "loss": 0.7858,
+      "loss": 0.8464,
       "step": 17614
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.042968004618621e-06,
-      "loss": 0.6686,
+      "loss": 0.8898,
       "step": 17615
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.041886309292593e-06,
-      "loss": 0.789,
+      "loss": 0.8108,
       "step": 17616
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.040804668879769e-06,
-      "loss": 0.7344,
+      "loss": 0.8489,
       "step": 17617
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.039723083395171e-06,
-      "loss": 0.7452,
+      "loss": 0.8484,
       "step": 17618
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.038641552853792e-06,
-      "loss": 0.7951,
+      "loss": 0.8011,
       "step": 17619
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0375600772706435e-06,
-      "loss": 0.6426,
+      "loss": 0.6956,
       "step": 17620
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.036478656660722e-06,
-      "loss": 0.7376,
+      "loss": 0.8568,
       "step": 17621
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.03539729103904e-06,
-      "loss": 0.6916,
+      "loss": 0.8239,
       "step": 17622
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.034315980420594e-06,
-      "loss": 0.787,
+      "loss": 0.8409,
       "step": 17623
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0332347248203915e-06,
-      "loss": 0.71,
+      "loss": 0.7935,
       "step": 17624
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.032153524253424e-06,
-      "loss": 0.7343,
+      "loss": 0.8971,
       "step": 17625
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.031072378734698e-06,
-      "loss": 0.7088,
+      "loss": 0.8427,
       "step": 17626
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.029991288279211e-06,
-      "loss": 0.7593,
+      "loss": 0.8357,
       "step": 17627
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0289102529019625e-06,
-      "loss": 0.7687,
+      "loss": 0.8719,
       "step": 17628
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.027829272617949e-06,
-      "loss": 0.8345,
+      "loss": 0.909,
       "step": 17629
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.02674834744217e-06,
-      "loss": 0.6461,
+      "loss": 0.7976,
       "step": 17630
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.025667477389625e-06,
-      "loss": 0.7946,
+      "loss": 0.9004,
       "step": 17631
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.024586662475298e-06,
-      "loss": 0.782,
+      "loss": 0.8058,
       "step": 17632
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.023505902714194e-06,
-      "loss": 0.8028,
+      "loss": 0.8757,
       "step": 17633
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.022425198121301e-06,
-      "loss": 0.7363,
+      "loss": 0.7832,
       "step": 17634
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0213445487116175e-06,
-      "loss": 0.7039,
+      "loss": 0.7845,
       "step": 17635
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.020263954500131e-06,
-      "loss": 0.7051,
+      "loss": 0.7205,
       "step": 17636
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0191834155018395e-06,
-      "loss": 0.7367,
+      "loss": 0.8651,
       "step": 17637
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.018102931731726e-06,
-      "loss": 0.6863,
+      "loss": 0.7734,
       "step": 17638
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.017022503204788e-06,
-      "loss": 0.6334,
+      "loss": 0.7336,
       "step": 17639
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.015942129936008e-06,
-      "loss": 0.7212,
+      "loss": 0.8344,
       "step": 17640
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.014861811940381e-06,
-      "loss": 0.6035,
+      "loss": 0.8106,
       "step": 17641
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.013781549232891e-06,
-      "loss": 0.6461,
+      "loss": 0.8269,
       "step": 17642
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.012701341828528e-06,
-      "loss": 0.7523,
+      "loss": 0.8066,
       "step": 17643
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0116211897422735e-06,
-      "loss": 0.7025,
+      "loss": 0.7609,
       "step": 17644
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.010541092989121e-06,
-      "loss": 0.6863,
+      "loss": 0.8218,
       "step": 17645
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.009461051584047e-06,
-      "loss": 0.6139,
+      "loss": 0.7052,
       "step": 17646
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0083810655420394e-06,
-      "loss": 0.6764,
+      "loss": 0.7763,
       "step": 17647
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.007301134878081e-06,
-      "loss": 0.7824,
+      "loss": 0.815,
       "step": 17648
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.006221259607154e-06,
-      "loss": 0.68,
+      "loss": 0.7705,
       "step": 17649
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.005141439744247e-06,
-      "loss": 0.816,
+      "loss": 0.8618,
       "step": 17650
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.004061675304327e-06,
-      "loss": 0.8175,
+      "loss": 0.8221,
       "step": 17651
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.002981966302384e-06,
-      "loss": 0.6173,
+      "loss": 0.7323,
       "step": 17652
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0019023127533935e-06,
-      "loss": 0.8064,
+      "loss": 0.8993,
       "step": 17653
     },
     {
       "epoch": 0.64,
       "learning_rate": 6.0008227146723386e-06,
-      "loss": 0.6639,
+      "loss": 0.7873,
       "step": 17654
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.999743172074193e-06,
-      "loss": 0.7703,
+      "loss": 0.8672,
       "step": 17655
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.998663684973938e-06,
-      "loss": 0.7813,
+      "loss": 0.8199,
       "step": 17656
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.997584253386541e-06,
-      "loss": 0.7615,
+      "loss": 0.8277,
       "step": 17657
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.996504877326989e-06,
-      "loss": 0.7896,
+      "loss": 0.8881,
       "step": 17658
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.995425556810251e-06,
-      "loss": 0.5867,
+      "loss": 0.6743,
       "step": 17659
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.994346291851301e-06,
-      "loss": 0.6828,
+      "loss": 0.8226,
       "step": 17660
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9932670824651105e-06,
-      "loss": 0.6646,
+      "loss": 0.7997,
       "step": 17661
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9921879286666575e-06,
-      "loss": 0.6034,
+      "loss": 0.7865,
       "step": 17662
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.991108830470909e-06,
-      "loss": 0.6935,
+      "loss": 0.8942,
       "step": 17663
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.990029787892841e-06,
-      "loss": 0.5775,
+      "loss": 0.7356,
       "step": 17664
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.988950800947417e-06,
-      "loss": 0.8679,
+      "loss": 0.9145,
       "step": 17665
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.987871869649611e-06,
-      "loss": 0.6651,
+      "loss": 0.7908,
       "step": 17666
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9867929940143875e-06,
-      "loss": 0.738,
+      "loss": 0.8227,
       "step": 17667
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.985714174056721e-06,
-      "loss": 0.7088,
+      "loss": 0.8699,
       "step": 17668
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.984635409791576e-06,
-      "loss": 0.8231,
+      "loss": 0.9254,
       "step": 17669
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.983556701233913e-06,
-      "loss": 0.6878,
+      "loss": 0.7835,
       "step": 17670
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.982478048398706e-06,
-      "loss": 0.6339,
+      "loss": 0.8083,
       "step": 17671
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9813994513009125e-06,
-      "loss": 0.6645,
+      "loss": 0.7385,
       "step": 17672
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.980320909955501e-06,
-      "loss": 0.6878,
+      "loss": 0.8525,
       "step": 17673
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.979242424377434e-06,
-      "loss": 0.7167,
+      "loss": 0.7427,
       "step": 17674
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.978163994581674e-06,
-      "loss": 0.6789,
+      "loss": 0.7918,
       "step": 17675
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.97708562058318e-06,
-      "loss": 0.6595,
+      "loss": 0.8645,
       "step": 17676
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.976007302396923e-06,
-      "loss": 0.6735,
+      "loss": 0.7676,
       "step": 17677
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.974929040037847e-06,
-      "loss": 0.6365,
+      "loss": 0.7831,
       "step": 17678
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.973850833520923e-06,
-      "loss": 0.6174,
+      "loss": 0.8193,
       "step": 17679
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.972772682861105e-06,
-      "loss": 0.6527,
+      "loss": 0.734,
       "step": 17680
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.971694588073353e-06,
-      "loss": 0.8922,
+      "loss": 0.9298,
       "step": 17681
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9706165491726235e-06,
-      "loss": 0.6945,
+      "loss": 0.7682,
       "step": 17682
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9695385661738735e-06,
-      "loss": 0.9152,
+      "loss": 0.951,
       "step": 17683
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.968460639092053e-06,
-      "loss": 0.6773,
+      "loss": 0.8744,
       "step": 17684
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.967382767942127e-06,
-      "loss": 0.691,
+      "loss": 0.8169,
       "step": 17685
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.966304952739039e-06,
-      "loss": 0.8399,
+      "loss": 0.9015,
       "step": 17686
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.96522719349775e-06,
-      "loss": 0.601,
+      "loss": 0.7581,
       "step": 17687
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.964149490233211e-06,
-      "loss": 0.6859,
+      "loss": 0.7667,
       "step": 17688
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.963071842960364e-06,
-      "loss": 0.6316,
+      "loss": 0.7824,
       "step": 17689
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.961994251694174e-06,
-      "loss": 0.716,
+      "loss": 0.8741,
       "step": 17690
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.960916716449581e-06,
-      "loss": 0.6543,
+      "loss": 0.8185,
       "step": 17691
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.95983923724154e-06,
-      "loss": 0.7933,
+      "loss": 0.8467,
       "step": 17692
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9587618140849966e-06,
-      "loss": 0.5683,
+      "loss": 0.6924,
       "step": 17693
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.957684446994901e-06,
-      "loss": 0.6712,
+      "loss": 0.7919,
       "step": 17694
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9566071359861945e-06,
-      "loss": 0.6283,
+      "loss": 0.7926,
       "step": 17695
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.955529881073832e-06,
-      "loss": 0.6503,
+      "loss": 0.7496,
       "step": 17696
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.954452682272749e-06,
-      "loss": 0.7506,
+      "loss": 0.833,
       "step": 17697
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9533755395978985e-06,
-      "loss": 0.775,
+      "loss": 0.8033,
       "step": 17698
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.952298453064218e-06,
-      "loss": 0.5762,
+      "loss": 0.7829,
       "step": 17699
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.951221422686657e-06,
-      "loss": 0.6392,
+      "loss": 0.8425,
       "step": 17700
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.950144448480152e-06,
-      "loss": 0.6756,
+      "loss": 0.7755,
       "step": 17701
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.949067530459648e-06,
-      "loss": 0.7737,
+      "loss": 0.8778,
       "step": 17702
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9479906686400826e-06,
-      "loss": 0.7045,
+      "loss": 0.8266,
       "step": 17703
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.946913863036402e-06,
-      "loss": 0.7987,
+      "loss": 0.9137,
       "step": 17704
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.945837113663537e-06,
-      "loss": 0.6514,
+      "loss": 0.7258,
       "step": 17705
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9447604205364304e-06,
-      "loss": 0.7363,
+      "loss": 0.758,
       "step": 17706
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.943683783670022e-06,
-      "loss": 0.7156,
+      "loss": 0.8438,
       "step": 17707
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.942607203079243e-06,
-      "loss": 0.701,
+      "loss": 0.8213,
       "step": 17708
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.941530678779035e-06,
-      "loss": 0.8182,
+      "loss": 0.9114,
       "step": 17709
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.940454210784328e-06,
-      "loss": 0.74,
+      "loss": 0.7841,
       "step": 17710
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.93937779911006e-06,
-      "loss": 0.7872,
+      "loss": 0.8381,
       "step": 17711
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.938301443771163e-06,
-      "loss": 0.8391,
+      "loss": 0.9188,
       "step": 17712
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.937225144782572e-06,
-      "loss": 0.6875,
+      "loss": 0.794,
       "step": 17713
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9361489021592155e-06,
-      "loss": 0.5588,
+      "loss": 0.7167,
       "step": 17714
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.935072715916033e-06,
-      "loss": 0.8273,
+      "loss": 0.9211,
       "step": 17715
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.933996586067941e-06,
-      "loss": 0.582,
+      "loss": 0.6991,
       "step": 17716
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.932920512629884e-06,
-      "loss": 0.6369,
+      "loss": 0.7356,
       "step": 17717
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.931844495616781e-06,
-      "loss": 0.6662,
+      "loss": 0.8036,
       "step": 17718
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.930768535043565e-06,
-      "loss": 0.5339,
+      "loss": 0.7709,
       "step": 17719
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.92969263092516e-06,
-      "loss": 0.7397,
+      "loss": 0.7908,
       "step": 17720
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.928616783276498e-06,
-      "loss": 0.7661,
+      "loss": 0.8747,
       "step": 17721
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.927540992112498e-06,
-      "loss": 0.7096,
+      "loss": 0.7483,
       "step": 17722
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.926465257448093e-06,
-      "loss": 0.6395,
+      "loss": 0.8469,
       "step": 17723
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9253895792981976e-06,
-      "loss": 0.7416,
+      "loss": 0.8336,
       "step": 17724
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.924313957677747e-06,
-      "loss": 0.6728,
+      "loss": 0.815,
       "step": 17725
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.923238392601655e-06,
-      "loss": 0.769,
+      "loss": 0.8363,
       "step": 17726
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.922162884084843e-06,
-      "loss": 0.7083,
+      "loss": 0.7666,
       "step": 17727
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.921087432142241e-06,
-      "loss": 0.6755,
+      "loss": 0.8191,
       "step": 17728
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.920012036788758e-06,
-      "loss": 0.747,
+      "loss": 0.783,
       "step": 17729
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.918936698039324e-06,
-      "loss": 0.7346,
+      "loss": 0.793,
       "step": 17730
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.91786141590885e-06,
-      "loss": 0.7233,
+      "loss": 0.8549,
       "step": 17731
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.9167861904122585e-06,
-      "loss": 0.8597,
+      "loss": 0.8682,
       "step": 17732
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.915711021564463e-06,
-      "loss": 0.6775,
+      "loss": 0.8034,
       "step": 17733
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.914635909380384e-06,
-      "loss": 0.7081,
+      "loss": 0.8373,
       "step": 17734
     },
     {
       "epoch": 0.64,
       "learning_rate": 5.913560853874933e-06,
-      "loss": 0.7488,
+      "loss": 0.8459,
       "step": 17735
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.9124858550630326e-06,
-      "loss": 0.7757,
+      "loss": 0.867,
       "step": 17736
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.911410912959588e-06,
-      "loss": 0.6852,
+      "loss": 0.808,
       "step": 17737
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.9103360275795166e-06,
-      "loss": 0.8771,
+      "loss": 0.901,
       "step": 17738
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.909261198937728e-06,
-      "loss": 0.7133,
+      "loss": 0.8059,
       "step": 17739
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.908186427049139e-06,
-      "loss": 0.7347,
+      "loss": 0.8532,
       "step": 17740
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.907111711928657e-06,
-      "loss": 0.5308,
+      "loss": 0.7015,
       "step": 17741
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.906037053591196e-06,
-      "loss": 0.6334,
+      "loss": 0.8231,
       "step": 17742
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.904962452051655e-06,
-      "loss": 0.6996,
+      "loss": 0.8141,
       "step": 17743
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.903887907324958e-06,
-      "loss": 0.8625,
+      "loss": 0.914,
       "step": 17744
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.902813419426001e-06,
-      "loss": 0.7925,
+      "loss": 0.8549,
       "step": 17745
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.9017389883696945e-06,
-      "loss": 0.716,
+      "loss": 0.7761,
       "step": 17746
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.9006646141709455e-06,
-      "loss": 0.7814,
+      "loss": 0.8636,
       "step": 17747
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.899590296844658e-06,
-      "loss": 0.7136,
+      "loss": 0.911,
       "step": 17748
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8985160364057395e-06,
-      "loss": 0.6343,
+      "loss": 0.7778,
       "step": 17749
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.897441832869089e-06,
-      "loss": 0.6693,
+      "loss": 0.8113,
       "step": 17750
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.896367686249614e-06,
-      "loss": 0.7345,
+      "loss": 0.8061,
       "step": 17751
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.895293596562212e-06,
-      "loss": 0.8306,
+      "loss": 0.8255,
       "step": 17752
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.894219563821791e-06,
-      "loss": 0.7394,
+      "loss": 0.7983,
       "step": 17753
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.893145588043245e-06,
-      "loss": 0.6115,
+      "loss": 0.7814,
       "step": 17754
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.892071669241482e-06,
-      "loss": 0.6902,
+      "loss": 0.7567,
       "step": 17755
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.89099780743139e-06,
-      "loss": 0.783,
+      "loss": 0.8918,
       "step": 17756
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.889924002627881e-06,
-      "loss": 0.4879,
+      "loss": 0.6788,
       "step": 17757
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8888502548458384e-06,
-      "loss": 0.6104,
+      "loss": 0.7931,
       "step": 17758
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.887776564100169e-06,
-      "loss": 0.7391,
+      "loss": 0.7829,
       "step": 17759
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.886702930405764e-06,
-      "loss": 0.6626,
+      "loss": 0.724,
       "step": 17760
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.88562935377752e-06,
-      "loss": 0.7792,
+      "loss": 0.9216,
       "step": 17761
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.88455583423033e-06,
-      "loss": 0.6555,
+      "loss": 0.7696,
       "step": 17762
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.883482371779093e-06,
-      "loss": 0.7403,
+      "loss": 0.8168,
       "step": 17763
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.882408966438695e-06,
-      "loss": 0.5957,
+      "loss": 0.7639,
       "step": 17764
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8813356182240285e-06,
-      "loss": 0.8146,
+      "loss": 0.9206,
       "step": 17765
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.880262327149989e-06,
-      "loss": 0.6346,
+      "loss": 0.7674,
       "step": 17766
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.879189093231463e-06,
-      "loss": 0.6767,
+      "loss": 0.8513,
       "step": 17767
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.878115916483344e-06,
-      "loss": 0.6495,
+      "loss": 0.728,
       "step": 17768
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.877042796920514e-06,
-      "loss": 0.7932,
+      "loss": 0.9142,
       "step": 17769
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.875969734557872e-06,
-      "loss": 0.7783,
+      "loss": 0.8606,
       "step": 17770
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.874896729410293e-06,
-      "loss": 0.793,
+      "loss": 0.908,
       "step": 17771
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8738237814926735e-06,
-      "loss": 0.7648,
+      "loss": 0.8859,
       "step": 17772
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.87275089081989e-06,
-      "loss": 0.7667,
+      "loss": 0.8535,
       "step": 17773
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.871678057406835e-06,
-      "loss": 0.6846,
+      "loss": 0.7678,
       "step": 17774
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.870605281268389e-06,
-      "loss": 0.8756,
+      "loss": 0.8568,
       "step": 17775
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8695325624194385e-06,
-      "loss": 0.7349,
+      "loss": 0.8387,
       "step": 17776
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.86845990087486e-06,
-      "loss": 0.7734,
+      "loss": 0.8012,
       "step": 17777
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.86738729664954e-06,
-      "loss": 0.6498,
+      "loss": 0.7705,
       "step": 17778
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.866314749758356e-06,
-      "loss": 0.7009,
+      "loss": 0.8718,
       "step": 17779
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.865242260216193e-06,
-      "loss": 0.7327,
+      "loss": 0.8564,
       "step": 17780
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.864169828037923e-06,
-      "loss": 0.7091,
+      "loss": 0.8064,
       "step": 17781
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8630974532384345e-06,
-      "loss": 0.7211,
+      "loss": 0.7864,
       "step": 17782
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.862025135832596e-06,
-      "loss": 0.7494,
+      "loss": 0.8818,
       "step": 17783
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.860952875835286e-06,
-      "loss": 0.8623,
+      "loss": 0.8413,
       "step": 17784
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.859880673261384e-06,
-      "loss": 0.5936,
+      "loss": 0.792,
       "step": 17785
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.858808528125763e-06,
-      "loss": 0.6801,
+      "loss": 0.8268,
       "step": 17786
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.857736440443302e-06,
-      "loss": 0.718,
+      "loss": 0.806,
       "step": 17787
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.856664410228864e-06,
-      "loss": 0.6849,
+      "loss": 0.8113,
       "step": 17788
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.855592437497335e-06,
-      "loss": 0.7267,
+      "loss": 0.7613,
       "step": 17789
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.854520522263576e-06,
-      "loss": 0.7705,
+      "loss": 0.8815,
       "step": 17790
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.853448664542466e-06,
-      "loss": 0.8295,
+      "loss": 0.8901,
       "step": 17791
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.85237686434887e-06,
-      "loss": 0.6491,
+      "loss": 0.8254,
       "step": 17792
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.851305121697663e-06,
-      "loss": 0.6408,
+      "loss": 0.7297,
       "step": 17793
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.850233436603709e-06,
-      "loss": 0.7087,
+      "loss": 0.8546,
       "step": 17794
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.849161809081882e-06,
-      "loss": 0.5975,
+      "loss": 0.7165,
       "step": 17795
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.84809023914704e-06,
-      "loss": 0.75,
+      "loss": 0.8595,
       "step": 17796
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.84701872681406e-06,
-      "loss": 0.6359,
+      "loss": 0.7915,
       "step": 17797
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.845947272097801e-06,
-      "loss": 0.8473,
+      "loss": 0.885,
       "step": 17798
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.844875875013129e-06,
-      "loss": 0.6541,
+      "loss": 0.8845,
       "step": 17799
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.843804535574908e-06,
-      "loss": 0.7709,
+      "loss": 0.9174,
       "step": 17800
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.842733253798005e-06,
-      "loss": 0.7585,
+      "loss": 0.8258,
       "step": 17801
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8416620296972815e-06,
-      "loss": 0.6965,
+      "loss": 0.8112,
       "step": 17802
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.840590863287594e-06,
-      "loss": 0.7705,
+      "loss": 0.9044,
       "step": 17803
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.839519754583808e-06,
-      "loss": 0.586,
+      "loss": 0.741,
       "step": 17804
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.838448703600781e-06,
-      "loss": 0.6927,
+      "loss": 0.8477,
       "step": 17805
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8373777103533746e-06,
-      "loss": 0.6721,
+      "loss": 0.8103,
       "step": 17806
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8363067748564474e-06,
-      "loss": 0.6511,
+      "loss": 0.7414,
       "step": 17807
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.835235897124856e-06,
-      "loss": 0.7292,
+      "loss": 0.8763,
       "step": 17808
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.834165077173452e-06,
-      "loss": 0.7631,
+      "loss": 0.791,
       "step": 17809
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.833094315017104e-06,
-      "loss": 0.7485,
+      "loss": 0.8292,
       "step": 17810
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.832023610670659e-06,
-      "loss": 0.602,
+      "loss": 0.7543,
       "step": 17811
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8309529641489726e-06,
-      "loss": 0.6528,
+      "loss": 0.7407,
       "step": 17812
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.829882375466897e-06,
-      "loss": 0.689,
+      "loss": 0.8079,
       "step": 17813
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.828811844639291e-06,
-      "loss": 0.6739,
+      "loss": 0.8286,
       "step": 17814
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.827741371680995e-06,
-      "loss": 0.6997,
+      "loss": 0.8193,
       "step": 17815
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8266709566068726e-06,
-      "loss": 0.7056,
+      "loss": 0.8207,
       "step": 17816
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.82560059943177e-06,
-      "loss": 0.6357,
+      "loss": 0.7444,
       "step": 17817
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.824530300170537e-06,
-      "loss": 0.6669,
+      "loss": 0.8345,
       "step": 17818
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.823460058838015e-06,
-      "loss": 0.6144,
+      "loss": 0.807,
       "step": 17819
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.822389875449071e-06,
-      "loss": 0.687,
+      "loss": 0.7806,
       "step": 17820
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.821319750018533e-06,
-      "loss": 0.7848,
+      "loss": 0.784,
       "step": 17821
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.820249682561252e-06,
-      "loss": 0.7187,
+      "loss": 0.8092,
       "step": 17822
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.819179673092084e-06,
-      "loss": 0.7973,
+      "loss": 0.8865,
       "step": 17823
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8181097216258585e-06,
-      "loss": 0.8367,
+      "loss": 0.8459,
       "step": 17824
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.817039828177431e-06,
-      "loss": 0.6695,
+      "loss": 0.7503,
       "step": 17825
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.815969992761642e-06,
-      "loss": 0.7109,
+      "loss": 0.8564,
       "step": 17826
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.814900215393332e-06,
-      "loss": 0.715,
+      "loss": 0.7844,
       "step": 17827
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.813830496087341e-06,
-      "loss": 0.7369,
+      "loss": 0.8351,
       "step": 17828
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.812760834858514e-06,
-      "loss": 0.6699,
+      "loss": 0.7895,
       "step": 17829
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.811691231721691e-06,
-      "loss": 0.6534,
+      "loss": 0.7771,
       "step": 17830
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.81062168669171e-06,
-      "loss": 0.6938,
+      "loss": 0.7816,
       "step": 17831
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.809552199783405e-06,
-      "loss": 0.6421,
+      "loss": 0.8094,
       "step": 17832
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.808482771011625e-06,
-      "loss": 0.6408,
+      "loss": 0.788,
       "step": 17833
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.807413400391192e-06,
-      "loss": 0.8049,
+      "loss": 0.8874,
       "step": 17834
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.806344087936954e-06,
-      "loss": 0.6961,
+      "loss": 0.8384,
       "step": 17835
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.805274833663741e-06,
-      "loss": 0.7637,
+      "loss": 0.8213,
       "step": 17836
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8042056375863885e-06,
-      "loss": 0.6911,
+      "loss": 0.7768,
       "step": 17837
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.8031364997197255e-06,
-      "loss": 0.6949,
+      "loss": 0.888,
       "step": 17838
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.802067420078599e-06,
-      "loss": 0.8255,
+      "loss": 0.8828,
       "step": 17839
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.800998398677824e-06,
-      "loss": 0.7778,
+      "loss": 0.8425,
       "step": 17840
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.799929435532237e-06,
-      "loss": 0.6878,
+      "loss": 0.8487,
       "step": 17841
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.798860530656671e-06,
-      "loss": 0.7004,
+      "loss": 0.8,
       "step": 17842
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7977916840659564e-06,
-      "loss": 0.7997,
+      "loss": 0.8357,
       "step": 17843
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.79672289577492e-06,
-      "loss": 0.7034,
+      "loss": 0.7955,
       "step": 17844
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.79565416579839e-06,
-      "loss": 0.6486,
+      "loss": 0.7845,
       "step": 17845
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7945854941511925e-06,
-      "loss": 0.679,
+      "loss": 0.791,
       "step": 17846
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.793516880848151e-06,
-      "loss": 0.7883,
+      "loss": 0.8463,
       "step": 17847
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.792448325904099e-06,
-      "loss": 0.7101,
+      "loss": 0.8674,
       "step": 17848
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.791379829333856e-06,
-      "loss": 0.6664,
+      "loss": 0.786,
       "step": 17849
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.790311391152248e-06,
-      "loss": 0.6891,
+      "loss": 0.841,
       "step": 17850
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.78924301137409e-06,
-      "loss": 0.8289,
+      "loss": 0.9411,
       "step": 17851
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.788174690014222e-06,
-      "loss": 0.7544,
+      "loss": 0.8691,
       "step": 17852
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.787106427087446e-06,
-      "loss": 0.7331,
+      "loss": 0.8765,
       "step": 17853
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7860382226085924e-06,
-      "loss": 0.8021,
+      "loss": 0.8938,
       "step": 17854
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.784970076592482e-06,
-      "loss": 0.5777,
+      "loss": 0.703,
       "step": 17855
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.783901989053931e-06,
-      "loss": 0.5837,
+      "loss": 0.7461,
       "step": 17856
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.782833960007752e-06,
-      "loss": 0.8294,
+      "loss": 0.9078,
       "step": 17857
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.78176598946878e-06,
-      "loss": 0.78,
+      "loss": 0.8751,
       "step": 17858
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.780698077451809e-06,
-      "loss": 0.7515,
+      "loss": 0.8231,
       "step": 17859
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.779630223971671e-06,
-      "loss": 0.6231,
+      "loss": 0.7685,
       "step": 17860
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.778562429043175e-06,
-      "loss": 0.6907,
+      "loss": 0.8757,
       "step": 17861
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.777494692681137e-06,
-      "loss": 0.6443,
+      "loss": 0.7715,
       "step": 17862
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.776427014900368e-06,
-      "loss": 0.6617,
+      "loss": 0.8222,
       "step": 17863
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7753593957156785e-06,
-      "loss": 0.7047,
+      "loss": 0.8262,
       "step": 17864
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.774291835141892e-06,
-      "loss": 0.6656,
+      "loss": 0.8127,
       "step": 17865
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.773224333193801e-06,
-      "loss": 0.7277,
+      "loss": 0.8578,
       "step": 17866
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.772156889886231e-06,
-      "loss": 0.7615,
+      "loss": 0.8311,
       "step": 17867
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7710895052339844e-06,
-      "loss": 0.6127,
+      "loss": 0.7532,
       "step": 17868
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.770022179251873e-06,
-      "loss": 0.6054,
+      "loss": 0.7141,
       "step": 17869
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7689549119546965e-06,
-      "loss": 0.6435,
+      "loss": 0.7388,
       "step": 17870
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.767887703357275e-06,
-      "loss": 0.6243,
+      "loss": 0.7879,
       "step": 17871
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7668205534744e-06,
-      "loss": 0.6939,
+      "loss": 0.8079,
       "step": 17872
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.765753462320888e-06,
-      "loss": 0.6935,
+      "loss": 0.7934,
       "step": 17873
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7646864299115394e-06,
-      "loss": 0.7747,
+      "loss": 0.7736,
       "step": 17874
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.763619456261156e-06,
-      "loss": 0.6554,
+      "loss": 0.7703,
       "step": 17875
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.762552541384539e-06,
-      "loss": 0.7683,
+      "loss": 0.8425,
       "step": 17876
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.761485685296497e-06,
-      "loss": 0.7448,
+      "loss": 0.86,
       "step": 17877
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.760418888011827e-06,
-      "loss": 0.7928,
+      "loss": 0.849,
       "step": 17878
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.759352149545331e-06,
-      "loss": 0.7479,
+      "loss": 0.8068,
       "step": 17879
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7582854699118065e-06,
-      "loss": 0.64,
+      "loss": 0.7485,
       "step": 17880
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.757218849126053e-06,
-      "loss": 0.9379,
+      "loss": 0.9341,
       "step": 17881
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.756152287202867e-06,
-      "loss": 0.6363,
+      "loss": 0.771,
       "step": 17882
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.755085784157044e-06,
-      "loss": 0.6575,
+      "loss": 0.745,
       "step": 17883
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.754019340003392e-06,
-      "loss": 0.7198,
+      "loss": 0.7706,
       "step": 17884
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.752952954756688e-06,
-      "loss": 0.6887,
+      "loss": 0.8332,
       "step": 17885
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.75188662843174e-06,
-      "loss": 0.6793,
+      "loss": 0.8401,
       "step": 17886
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.750820361043339e-06,
-      "loss": 0.717,
+      "loss": 0.8526,
       "step": 17887
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.749754152606275e-06,
-      "loss": 0.6882,
+      "loss": 0.7312,
       "step": 17888
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.748688003135338e-06,
-      "loss": 0.7155,
+      "loss": 0.7905,
       "step": 17889
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.747621912645333e-06,
-      "loss": 0.7737,
+      "loss": 0.931,
       "step": 17890
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.74655588115103e-06,
-      "loss": 0.6709,
+      "loss": 0.8264,
       "step": 17891
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7454899086672355e-06,
-      "loss": 0.7478,
+      "loss": 0.8587,
       "step": 17892
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.744423995208732e-06,
-      "loss": 0.7118,
+      "loss": 0.8461,
       "step": 17893
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.743358140790307e-06,
-      "loss": 0.7638,
+      "loss": 0.8915,
       "step": 17894
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.742292345426743e-06,
-      "loss": 0.8901,
+      "loss": 0.9278,
       "step": 17895
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.741226609132838e-06,
-      "loss": 0.8269,
+      "loss": 0.7987,
       "step": 17896
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7401609319233705e-06,
-      "loss": 0.7748,
+      "loss": 0.8281,
       "step": 17897
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.739095313813126e-06,
-      "loss": 0.668,
+      "loss": 0.8287,
       "step": 17898
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.73802975481689e-06,
-      "loss": 0.7265,
+      "loss": 0.7792,
       "step": 17899
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.736964254949442e-06,
-      "loss": 0.7213,
+      "loss": 0.8828,
       "step": 17900
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.735898814225569e-06,
-      "loss": 0.7015,
+      "loss": 0.8063,
       "step": 17901
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.734833432660043e-06,
-      "loss": 0.7312,
+      "loss": 0.8004,
       "step": 17902
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.733768110267661e-06,
-      "loss": 0.6641,
+      "loss": 0.7736,
       "step": 17903
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.732702847063187e-06,
-      "loss": 0.76,
+      "loss": 0.8564,
       "step": 17904
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.73163764306141e-06,
-      "loss": 0.714,
+      "loss": 0.801,
       "step": 17905
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.730572498277102e-06,
-      "loss": 0.8548,
+      "loss": 0.8284,
       "step": 17906
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.729507412725046e-06,
-      "loss": 0.7784,
+      "loss": 0.8906,
       "step": 17907
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.72844238642001e-06,
-      "loss": 0.76,
+      "loss": 0.8657,
       "step": 17908
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7273774193767805e-06,
-      "loss": 0.7404,
+      "loss": 0.8604,
       "step": 17909
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.726312511610126e-06,
-      "loss": 0.6134,
+      "loss": 0.8555,
       "step": 17910
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.725247663134823e-06,
-      "loss": 0.7381,
+      "loss": 0.7989,
       "step": 17911
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.724182873965642e-06,
-      "loss": 0.7278,
+      "loss": 0.8632,
       "step": 17912
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.723118144117359e-06,
-      "loss": 0.7876,
+      "loss": 0.8335,
       "step": 17913
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.722053473604738e-06,
-      "loss": 0.7021,
+      "loss": 0.8553,
       "step": 17914
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.72098886244256e-06,
-      "loss": 0.8287,
+      "loss": 0.8597,
       "step": 17915
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.719924310645591e-06,
-      "loss": 0.6064,
+      "loss": 0.7574,
       "step": 17916
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.718859818228599e-06,
-      "loss": 0.7589,
+      "loss": 0.8468,
       "step": 17917
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.717795385206353e-06,
-      "loss": 0.7924,
+      "loss": 0.8379,
       "step": 17918
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.716731011593622e-06,
-      "loss": 0.6535,
+      "loss": 0.7785,
       "step": 17919
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.715666697405169e-06,
-      "loss": 0.7613,
+      "loss": 0.7627,
       "step": 17920
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.714602442655758e-06,
-      "loss": 0.5991,
+      "loss": 0.7943,
       "step": 17921
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.713538247360166e-06,
-      "loss": 0.7498,
+      "loss": 0.8252,
       "step": 17922
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.712474111533142e-06,
-      "loss": 0.6746,
+      "loss": 0.797,
       "step": 17923
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.711410035189459e-06,
-      "loss": 0.7072,
+      "loss": 0.8342,
       "step": 17924
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.710346018343878e-06,
-      "loss": 0.7192,
+      "loss": 0.7661,
       "step": 17925
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.70928206101116e-06,
-      "loss": 0.7646,
+      "loss": 0.9284,
       "step": 17926
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.708218163206061e-06,
-      "loss": 0.6995,
+      "loss": 0.7893,
       "step": 17927
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.70715432494335e-06,
-      "loss": 0.7271,
+      "loss": 0.8427,
       "step": 17928
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.706090546237782e-06,
-      "loss": 0.8232,
+      "loss": 0.9348,
       "step": 17929
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.705026827104117e-06,
-      "loss": 0.7297,
+      "loss": 0.8474,
       "step": 17930
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.703963167557105e-06,
-      "loss": 0.7421,
+      "loss": 0.9076,
       "step": 17931
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.702899567611517e-06,
-      "loss": 0.6967,
+      "loss": 0.7757,
       "step": 17932
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.701836027282093e-06,
-      "loss": 0.6306,
+      "loss": 0.8174,
       "step": 17933
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.7007725465836e-06,
-      "loss": 0.6249,
+      "loss": 0.8075,
       "step": 17934
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.699709125530788e-06,
-      "loss": 0.7384,
+      "loss": 0.7721,
       "step": 17935
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.698645764138411e-06,
-      "loss": 0.6826,
+      "loss": 0.7452,
       "step": 17936
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.697582462421221e-06,
-      "loss": 0.7261,
+      "loss": 0.8102,
       "step": 17937
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.696519220393969e-06,
-      "loss": 0.7439,
+      "loss": 0.8792,
       "step": 17938
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.695456038071407e-06,
-      "loss": 0.6944,
+      "loss": 0.8255,
       "step": 17939
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.694392915468283e-06,
-      "loss": 0.7235,
+      "loss": 0.757,
       "step": 17940
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.693329852599351e-06,
-      "loss": 0.8249,
+      "loss": 0.9226,
       "step": 17941
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.692266849479356e-06,
-      "loss": 0.5538,
+      "loss": 0.7492,
       "step": 17942
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.691203906123049e-06,
-      "loss": 0.8097,
+      "loss": 0.8639,
       "step": 17943
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.690141022545172e-06,
-      "loss": 0.7009,
+      "loss": 0.7791,
       "step": 17944
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.689078198760476e-06,
-      "loss": 0.7471,
+      "loss": 0.8792,
       "step": 17945
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.688015434783698e-06,
-      "loss": 0.5705,
+      "loss": 0.7187,
       "step": 17946
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.686952730629591e-06,
-      "loss": 0.7752,
+      "loss": 0.8,
       "step": 17947
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.685890086312897e-06,
-      "loss": 0.8391,
+      "loss": 0.8488,
       "step": 17948
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.684827501848358e-06,
-      "loss": 0.6561,
+      "loss": 0.7718,
       "step": 17949
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.683764977250708e-06,
-      "loss": 0.8365,
+      "loss": 0.8841,
       "step": 17950
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.682702512534706e-06,
-      "loss": 0.6261,
+      "loss": 0.8351,
       "step": 17951
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.681640107715072e-06,
-      "loss": 0.7921,
+      "loss": 0.8976,
       "step": 17952
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.680577762806558e-06,
-      "loss": 0.6317,
+      "loss": 0.7252,
       "step": 17953
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.679515477823899e-06,
-      "loss": 0.6879,
+      "loss": 0.7794,
       "step": 17954
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.678453252781834e-06,
-      "loss": 0.6949,
+      "loss": 0.7913,
       "step": 17955
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.677391087695098e-06,
-      "loss": 0.6861,
+      "loss": 0.7807,
       "step": 17956
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6763289825784274e-06,
-      "loss": 0.6898,
+      "loss": 0.7905,
       "step": 17957
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.675266937446558e-06,
-      "loss": 0.7384,
+      "loss": 0.8293,
       "step": 17958
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6742049523142196e-06,
-      "loss": 0.8312,
+      "loss": 0.9014,
       "step": 17959
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6731430271961545e-06,
-      "loss": 0.7841,
+      "loss": 0.7907,
       "step": 17960
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.672081162107091e-06,
-      "loss": 0.8164,
+      "loss": 0.9289,
       "step": 17961
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.67101935706176e-06,
-      "loss": 0.703,
+      "loss": 0.7841,
       "step": 17962
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.66995761207489e-06,
-      "loss": 0.7809,
+      "loss": 0.8891,
       "step": 17963
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.668895927161222e-06,
-      "loss": 0.6148,
+      "loss": 0.7657,
       "step": 17964
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.667834302335471e-06,
-      "loss": 0.673,
+      "loss": 0.8315,
       "step": 17965
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.666772737612376e-06,
-      "loss": 0.8282,
+      "loss": 0.8909,
       "step": 17966
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.665711233006661e-06,
-      "loss": 0.7568,
+      "loss": 0.8408,
       "step": 17967
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.664649788533052e-06,
-      "loss": 0.7923,
+      "loss": 0.8741,
       "step": 17968
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.663588404206274e-06,
-      "loss": 0.7139,
+      "loss": 0.7697,
       "step": 17969
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.662527080041063e-06,
-      "loss": 0.6695,
+      "loss": 0.825,
       "step": 17970
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.661465816052124e-06,
-      "loss": 0.6894,
+      "loss": 0.7813,
       "step": 17971
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6604046122541965e-06,
-      "loss": 0.6016,
+      "loss": 0.7232,
       "step": 17972
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6593434686619985e-06,
-      "loss": 0.7181,
+      "loss": 0.7872,
       "step": 17973
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.65828238529025e-06,
-      "loss": 0.7831,
+      "loss": 0.8945,
       "step": 17974
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6572213621536755e-06,
-      "loss": 0.8637,
+      "loss": 0.9275,
       "step": 17975
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6561603992669875e-06,
-      "loss": 0.7142,
+      "loss": 0.8121,
       "step": 17976
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.655099496644918e-06,
-      "loss": 0.7849,
+      "loss": 0.8782,
       "step": 17977
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.654038654302172e-06,
-      "loss": 0.6926,
+      "loss": 0.8685,
       "step": 17978
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.652977872253476e-06,
-      "loss": 0.7569,
+      "loss": 0.8725,
       "step": 17979
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6519171505135465e-06,
-      "loss": 0.7114,
+      "loss": 0.7802,
       "step": 17980
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.650856489097097e-06,
-      "loss": 0.6989,
+      "loss": 0.8437,
       "step": 17981
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.649795888018836e-06,
-      "loss": 0.7078,
+      "loss": 0.7656,
       "step": 17982
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.648735347293497e-06,
-      "loss": 0.6827,
+      "loss": 0.7893,
       "step": 17983
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.647674866935771e-06,
-      "loss": 0.7293,
+      "loss": 0.8362,
       "step": 17984
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.646614446960386e-06,
-      "loss": 0.631,
+      "loss": 0.701,
       "step": 17985
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.645554087382048e-06,
-      "loss": 0.6228,
+      "loss": 0.7988,
       "step": 17986
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.644493788215471e-06,
-      "loss": 0.7038,
+      "loss": 0.7789,
       "step": 17987
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.643433549475357e-06,
-      "loss": 0.804,
+      "loss": 0.8773,
       "step": 17988
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.64237337117643e-06,
-      "loss": 0.73,
+      "loss": 0.8527,
       "step": 17989
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.641313253333381e-06,
-      "loss": 0.6598,
+      "loss": 0.7992,
       "step": 17990
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6402531959609315e-06,
-      "loss": 0.6527,
+      "loss": 0.7875,
       "step": 17991
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.639193199073782e-06,
-      "loss": 0.7129,
+      "loss": 0.8061,
       "step": 17992
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.638133262686641e-06,
-      "loss": 0.6937,
+      "loss": 0.7624,
       "step": 17993
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6370733868142126e-06,
-      "loss": 0.7507,
+      "loss": 0.7684,
       "step": 17994
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.636013571471196e-06,
-      "loss": 0.7276,
+      "loss": 0.8509,
       "step": 17995
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6349538166723074e-06,
-      "loss": 0.743,
+      "loss": 0.8158,
       "step": 17996
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.633894122432233e-06,
-      "loss": 0.7902,
+      "loss": 0.8031,
       "step": 17997
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.632834488765688e-06,
-      "loss": 0.7086,
+      "loss": 0.8341,
       "step": 17998
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.631774915687367e-06,
-      "loss": 0.793,
+      "loss": 0.7873,
       "step": 17999
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.630715403211973e-06,
-      "loss": 0.7313,
+      "loss": 0.788,
       "step": 18000
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.629655951354198e-06,
-      "loss": 0.7103,
+      "loss": 0.8197,
       "step": 18001
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.628596560128755e-06,
-      "loss": 0.7437,
+      "loss": 0.8469,
       "step": 18002
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.627537229550324e-06,
-      "loss": 0.7343,
+      "loss": 0.8258,
       "step": 18003
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.626477959633614e-06,
-      "loss": 0.6449,
+      "loss": 0.7686,
       "step": 18004
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.625418750393318e-06,
-      "loss": 0.713,
+      "loss": 0.8673,
       "step": 18005
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6243596018441295e-06,
-      "loss": 0.6456,
+      "loss": 0.7617,
       "step": 18006
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6233005140007394e-06,
-      "loss": 0.8,
+      "loss": 0.882,
       "step": 18007
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.622241486877849e-06,
-      "loss": 0.741,
+      "loss": 0.8466,
       "step": 18008
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.621182520490147e-06,
-      "loss": 0.6038,
+      "loss": 0.7825,
       "step": 18009
     },
     {
       "epoch": 0.65,
       "learning_rate": 5.6201236148523245e-06,
-      "loss": 0.5842,
+      "loss": 0.7243,
       "step": 18010
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.619064769979072e-06,
-      "loss": 0.9569,
+      "loss": 0.9466,
       "step": 18011
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.61800598588508e-06,
-      "loss": 0.7126,
+      "loss": 0.8487,
       "step": 18012
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.61694726258504e-06,
-      "loss": 0.7961,
+      "loss": 0.8407,
       "step": 18013
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.615888600093631e-06,
-      "loss": 0.8693,
+      "loss": 0.9197,
       "step": 18014
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.614829998425558e-06,
-      "loss": 0.5834,
+      "loss": 0.7945,
       "step": 18015
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.6137714575954875e-06,
-      "loss": 0.6538,
+      "loss": 0.7537,
       "step": 18016
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.612712977618118e-06,
-      "loss": 0.6532,
+      "loss": 0.7177,
       "step": 18017
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.611654558508133e-06,
-      "loss": 0.7323,
+      "loss": 0.8032,
       "step": 18018
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.610596200280214e-06,
-      "loss": 0.766,
+      "loss": 0.7625,
       "step": 18019
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.60953790294904e-06,
-      "loss": 0.7347,
+      "loss": 0.8747,
       "step": 18020
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.608479666529304e-06,
-      "loss": 0.8193,
+      "loss": 0.8519,
       "step": 18021
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.6074214910356804e-06,
-      "loss": 0.6967,
+      "loss": 0.7665,
       "step": 18022
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.606363376482853e-06,
-      "loss": 0.7011,
+      "loss": 0.835,
       "step": 18023
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.605305322885499e-06,
-      "loss": 0.7438,
+      "loss": 0.8635,
       "step": 18024
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.604247330258299e-06,
-      "loss": 0.7076,
+      "loss": 0.7756,
       "step": 18025
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.603189398615926e-06,
-      "loss": 0.7069,
+      "loss": 0.851,
       "step": 18026
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.602131527973066e-06,
-      "loss": 0.7186,
+      "loss": 0.8055,
       "step": 18027
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.601073718344391e-06,
-      "loss": 0.7588,
+      "loss": 0.8578,
       "step": 18028
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.600015969744576e-06,
-      "loss": 0.553,
+      "loss": 0.7776,
       "step": 18029
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.598958282188295e-06,
-      "loss": 0.78,
+      "loss": 0.8323,
       "step": 18030
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.597900655690231e-06,
-      "loss": 0.671,
+      "loss": 0.8122,
       "step": 18031
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.596843090265045e-06,
-      "loss": 0.7233,
+      "loss": 0.8559,
       "step": 18032
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.595785585927409e-06,
-      "loss": 0.7822,
+      "loss": 0.8717,
       "step": 18033
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.594728142692007e-06,
-      "loss": 0.6835,
+      "loss": 0.8437,
       "step": 18034
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5936707605734965e-06,
-      "loss": 0.693,
+      "loss": 0.853,
       "step": 18035
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.592613439586554e-06,
-      "loss": 0.6813,
+      "loss": 0.802,
       "step": 18036
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.591556179745848e-06,
-      "loss": 0.7962,
+      "loss": 0.827,
       "step": 18037
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5904989810660445e-06,
-      "loss": 0.6735,
+      "loss": 0.8322,
       "step": 18038
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.589441843561808e-06,
-      "loss": 0.8434,
+      "loss": 0.911,
       "step": 18039
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.588384767247812e-06,
-      "loss": 0.6839,
+      "loss": 0.82,
       "step": 18040
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.587327752138718e-06,
-      "loss": 0.7302,
+      "loss": 0.8318,
       "step": 18041
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.586270798249191e-06,
-      "loss": 0.5893,
+      "loss": 0.6838,
       "step": 18042
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.585213905593889e-06,
-      "loss": 0.7276,
+      "loss": 0.8715,
       "step": 18043
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.58415707418749e-06,
-      "loss": 0.8037,
+      "loss": 0.8609,
       "step": 18044
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.583100304044637e-06,
-      "loss": 0.6626,
+      "loss": 0.7583,
       "step": 18045
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.582043595180005e-06,
-      "loss": 0.9011,
+      "loss": 0.9598,
       "step": 18046
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.58098694760825e-06,
-      "loss": 0.6467,
+      "loss": 0.642,
       "step": 18047
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.579930361344032e-06,
-      "loss": 0.8402,
+      "loss": 0.8926,
       "step": 18048
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.578873836402003e-06,
-      "loss": 0.6388,
+      "loss": 0.752,
       "step": 18049
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.577817372796836e-06,
-      "loss": 0.748,
+      "loss": 0.8962,
       "step": 18050
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.576760970543176e-06,
-      "loss": 0.6946,
+      "loss": 0.8122,
       "step": 18051
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.575704629655676e-06,
-      "loss": 0.7842,
+      "loss": 0.9148,
       "step": 18052
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.574648350149001e-06,
-      "loss": 0.5476,
+      "loss": 0.687,
       "step": 18053
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.573592132037803e-06,
-      "loss": 0.6817,
+      "loss": 0.7742,
       "step": 18054
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.572535975336732e-06,
-      "loss": 0.7099,
+      "loss": 0.7831,
       "step": 18055
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.571479880060444e-06,
-      "loss": 0.6774,
+      "loss": 0.8467,
       "step": 18056
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.570423846223589e-06,
-      "loss": 0.6867,
+      "loss": 0.8418,
       "step": 18057
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.569367873840814e-06,
-      "loss": 0.8014,
+      "loss": 0.8284,
       "step": 18058
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.568311962926778e-06,
-      "loss": 0.7205,
+      "loss": 0.8574,
       "step": 18059
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.567256113496125e-06,
-      "loss": 0.786,
+      "loss": 0.8167,
       "step": 18060
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.566200325563505e-06,
-      "loss": 0.7448,
+      "loss": 0.8003,
       "step": 18061
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.565144599143562e-06,
-      "loss": 0.6311,
+      "loss": 0.7755,
       "step": 18062
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.564088934250951e-06,
-      "loss": 0.6044,
+      "loss": 0.7618,
       "step": 18063
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.563033330900305e-06,
-      "loss": 0.8063,
+      "loss": 0.8506,
       "step": 18064
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.561977789106281e-06,
-      "loss": 0.6301,
+      "loss": 0.7202,
       "step": 18065
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.560922308883518e-06,
-      "loss": 0.6848,
+      "loss": 0.8654,
       "step": 18066
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.559866890246659e-06,
-      "loss": 0.7814,
+      "loss": 0.8842,
       "step": 18067
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.558811533210344e-06,
-      "loss": 0.9193,
+      "loss": 0.8868,
       "step": 18068
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.557756237789225e-06,
-      "loss": 0.732,
+      "loss": 0.7747,
       "step": 18069
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.556701003997933e-06,
-      "loss": 0.7613,
+      "loss": 0.851,
       "step": 18070
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.555645831851105e-06,
-      "loss": 0.7792,
+      "loss": 0.8446,
       "step": 18071
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5545907213633905e-06,
-      "loss": 0.8533,
+      "loss": 0.8938,
       "step": 18072
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.553535672549421e-06,
-      "loss": 0.7382,
+      "loss": 0.8417,
       "step": 18073
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5524806854238366e-06,
-      "loss": 0.6571,
+      "loss": 0.8032,
       "step": 18074
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.551425760001267e-06,
-      "loss": 0.6889,
+      "loss": 0.8573,
       "step": 18075
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.550370896296365e-06,
-      "loss": 0.7518,
+      "loss": 0.8888,
       "step": 18076
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5493160943237425e-06,
-      "loss": 0.8129,
+      "loss": 0.8931,
       "step": 18077
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5482613540980475e-06,
-      "loss": 0.7466,
+      "loss": 0.8027,
       "step": 18078
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.547206675633912e-06,
-      "loss": 0.7121,
+      "loss": 0.7912,
       "step": 18079
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.546152058945966e-06,
-      "loss": 0.7999,
+      "loss": 0.9255,
       "step": 18080
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5450975040488364e-06,
-      "loss": 0.7172,
+      "loss": 0.7573,
       "step": 18081
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5440430109571664e-06,
-      "loss": 0.6513,
+      "loss": 0.8189,
       "step": 18082
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.54298857968557e-06,
-      "loss": 0.6777,
+      "loss": 0.8297,
       "step": 18083
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.541934210248688e-06,
-      "loss": 0.6387,
+      "loss": 0.7836,
       "step": 18084
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.540879902661141e-06,
-      "loss": 0.694,
+      "loss": 0.7591,
       "step": 18085
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.539825656937561e-06,
-      "loss": 0.7349,
+      "loss": 0.8436,
       "step": 18086
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.538771473092567e-06,
-      "loss": 0.8267,
+      "loss": 0.9232,
       "step": 18087
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.537717351140797e-06,
-      "loss": 0.6737,
+      "loss": 0.8355,
       "step": 18088
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.536663291096864e-06,
-      "loss": 0.8591,
+      "loss": 0.9375,
       "step": 18089
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.535609292975392e-06,
-      "loss": 0.7839,
+      "loss": 0.8337,
       "step": 18090
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.534555356791009e-06,
-      "loss": 0.7344,
+      "loss": 0.7643,
       "step": 18091
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.533501482558337e-06,
-      "loss": 0.6549,
+      "loss": 0.7852,
       "step": 18092
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.532447670291993e-06,
-      "loss": 0.7521,
+      "loss": 0.8207,
       "step": 18093
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.531393920006596e-06,
-      "loss": 0.6963,
+      "loss": 0.7685,
       "step": 18094
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5303402317167755e-06,
-      "loss": 0.6931,
+      "loss": 0.806,
       "step": 18095
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.529286605437134e-06,
-      "loss": 0.808,
+      "loss": 0.8688,
       "step": 18096
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.528233041182302e-06,
-      "loss": 0.7485,
+      "loss": 0.7978,
       "step": 18097
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5271795389668935e-06,
-      "loss": 0.8283,
+      "loss": 1.0107,
       "step": 18098
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.526126098805522e-06,
-      "loss": 0.7314,
+      "loss": 0.8246,
       "step": 18099
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.525072720712799e-06,
-      "loss": 0.788,
+      "loss": 0.8444,
       "step": 18100
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.524019404703351e-06,
-      "loss": 0.5989,
+      "loss": 0.7457,
       "step": 18101
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.522966150791776e-06,
-      "loss": 0.7547,
+      "loss": 0.9062,
       "step": 18102
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.521912958992697e-06,
-      "loss": 0.8329,
+      "loss": 0.991,
       "step": 18103
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.520859829320721e-06,
-      "loss": 0.7417,
+      "loss": 0.749,
       "step": 18104
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.519806761790462e-06,
-      "loss": 0.7233,
+      "loss": 0.7522,
       "step": 18105
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.518753756416522e-06,
-      "loss": 0.7656,
+      "loss": 0.7891,
       "step": 18106
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.51770081321352e-06,
-      "loss": 0.709,
+      "loss": 0.7966,
       "step": 18107
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.51664793219606e-06,
-      "loss": 0.682,
+      "loss": 0.8207,
       "step": 18108
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.515595113378749e-06,
-      "loss": 0.6403,
+      "loss": 0.8026,
       "step": 18109
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.514542356776192e-06,
-      "loss": 0.7856,
+      "loss": 0.9298,
       "step": 18110
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.513489662402998e-06,
-      "loss": 0.6763,
+      "loss": 0.8481,
       "step": 18111
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.512437030273768e-06,
-      "loss": 0.6964,
+      "loss": 0.8167,
       "step": 18112
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5113844604031015e-06,
-      "loss": 0.6643,
+      "loss": 0.8329,
       "step": 18113
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5103319528056176e-06,
-      "loss": 0.7153,
+      "loss": 0.8303,
       "step": 18114
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.509279507495899e-06,
-      "loss": 0.7549,
+      "loss": 0.7815,
       "step": 18115
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.50822712448856e-06,
-      "loss": 0.7681,
+      "loss": 0.8149,
       "step": 18116
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.507174803798196e-06,
-      "loss": 0.6933,
+      "loss": 0.8566,
       "step": 18117
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.506122545439407e-06,
-      "loss": 0.6852,
+      "loss": 0.7697,
       "step": 18118
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.5050703494267885e-06,
-      "loss": 0.9088,
+      "loss": 0.9127,
       "step": 18119
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.504018215774945e-06,
-      "loss": 0.7381,
+      "loss": 0.783,
       "step": 18120
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.502966144498471e-06,
-      "loss": 0.7678,
+      "loss": 0.8569,
       "step": 18121
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.50191413561196e-06,
-      "loss": 0.6707,
+      "loss": 0.7922,
       "step": 18122
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.500862189130009e-06,
-      "loss": 0.7216,
+      "loss": 0.8133,
       "step": 18123
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.49981030506721e-06,
-      "loss": 0.718,
+      "loss": 0.7547,
       "step": 18124
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.498758483438156e-06,
-      "loss": 0.719,
+      "loss": 0.8641,
       "step": 18125
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.497706724257446e-06,
-      "loss": 0.7776,
+      "loss": 0.8077,
       "step": 18126
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.496655027539666e-06,
-      "loss": 0.6666,
+      "loss": 0.7988,
       "step": 18127
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.495603393299408e-06,
-      "loss": 0.7842,
+      "loss": 0.8104,
       "step": 18128
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.494551821551263e-06,
-      "loss": 0.6897,
+      "loss": 0.7903,
       "step": 18129
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.493500312309819e-06,
-      "loss": 0.595,
+      "loss": 0.7046,
       "step": 18130
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4924488655896635e-06,
-      "loss": 0.7181,
+      "loss": 0.8399,
       "step": 18131
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.491397481405381e-06,
-      "loss": 0.8071,
+      "loss": 0.8684,
       "step": 18132
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.490346159771569e-06,
-      "loss": 0.7068,
+      "loss": 0.8725,
       "step": 18133
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4892949007027975e-06,
-      "loss": 0.6077,
+      "loss": 0.8292,
       "step": 18134
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.488243704213665e-06,
-      "loss": 0.7442,
+      "loss": 0.8951,
       "step": 18135
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.487192570318749e-06,
-      "loss": 0.7665,
+      "loss": 0.8272,
       "step": 18136
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4861414990326325e-06,
-      "loss": 0.6591,
+      "loss": 0.8302,
       "step": 18137
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.485090490369895e-06,
-      "loss": 0.6381,
+      "loss": 0.8554,
       "step": 18138
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.484039544345126e-06,
-      "loss": 0.8122,
+      "loss": 0.8643,
       "step": 18139
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4829886609728985e-06,
-      "loss": 0.5658,
+      "loss": 0.6805,
       "step": 18140
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.481937840267797e-06,
-      "loss": 0.6491,
+      "loss": 0.7703,
       "step": 18141
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4808870822443926e-06,
-      "loss": 0.5979,
+      "loss": 0.7232,
       "step": 18142
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.479836386917277e-06,
-      "loss": 0.7024,
+      "loss": 0.8575,
       "step": 18143
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.478785754301008e-06,
-      "loss": 0.8289,
+      "loss": 0.8572,
       "step": 18144
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.477735184410178e-06,
-      "loss": 0.6116,
+      "loss": 0.7697,
       "step": 18145
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.476684677259356e-06,
-      "loss": 0.7178,
+      "loss": 0.8671,
       "step": 18146
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.475634232863117e-06,
-      "loss": 0.6797,
+      "loss": 0.8494,
       "step": 18147
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.474583851236031e-06,
-      "loss": 0.7291,
+      "loss": 0.8493,
       "step": 18148
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4735335323926754e-06,
-      "loss": 0.6935,
+      "loss": 0.8133,
       "step": 18149
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.47248327634762e-06,
-      "loss": 0.685,
+      "loss": 0.7886,
       "step": 18150
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.471433083115431e-06,
-      "loss": 0.8171,
+      "loss": 0.8834,
       "step": 18151
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.470382952710687e-06,
-      "loss": 0.705,
+      "loss": 0.8389,
       "step": 18152
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.469332885147951e-06,
-      "loss": 0.7104,
+      "loss": 0.7889,
       "step": 18153
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.468282880441795e-06,
-      "loss": 0.7738,
+      "loss": 0.8935,
       "step": 18154
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.467232938606784e-06,
-      "loss": 0.8587,
+      "loss": 0.8999,
       "step": 18155
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.466183059657484e-06,
-      "loss": 0.7824,
+      "loss": 0.8763,
       "step": 18156
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4651332436084585e-06,
-      "loss": 0.7119,
+      "loss": 0.7707,
       "step": 18157
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.464083490474279e-06,
-      "loss": 0.7544,
+      "loss": 0.8705,
       "step": 18158
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.463033800269506e-06,
-      "loss": 0.7412,
+      "loss": 0.7909,
       "step": 18159
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.461984173008701e-06,
-      "loss": 0.6856,
+      "loss": 0.7962,
       "step": 18160
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.460934608706423e-06,
-      "loss": 0.6241,
+      "loss": 0.813,
       "step": 18161
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.459885107377245e-06,
-      "loss": 0.7259,
+      "loss": 0.8477,
       "step": 18162
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.458835669035712e-06,
-      "loss": 0.6018,
+      "loss": 0.7288,
       "step": 18163
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.457786293696395e-06,
-      "loss": 0.7617,
+      "loss": 0.7607,
       "step": 18164
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.456736981373848e-06,
-      "loss": 0.6563,
+      "loss": 0.7605,
       "step": 18165
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.455687732082631e-06,
-      "loss": 0.831,
+      "loss": 0.8771,
       "step": 18166
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.454638545837296e-06,
-      "loss": 0.632,
+      "loss": 0.7061,
       "step": 18167
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.453589422652406e-06,
-      "loss": 0.6651,
+      "loss": 0.8042,
       "step": 18168
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.452540362542509e-06,
-      "loss": 0.6425,
+      "loss": 0.7027,
       "step": 18169
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.45149136552216e-06,
-      "loss": 0.6107,
+      "loss": 0.7833,
       "step": 18170
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.450442431605919e-06,
-      "loss": 0.6369,
+      "loss": 0.8269,
       "step": 18171
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.449393560808334e-06,
-      "loss": 0.7229,
+      "loss": 0.7676,
       "step": 18172
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.448344753143957e-06,
-      "loss": 0.698,
+      "loss": 0.8233,
       "step": 18173
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.447296008627334e-06,
-      "loss": 0.7599,
+      "loss": 0.7991,
       "step": 18174
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.446247327273029e-06,
-      "loss": 0.7725,
+      "loss": 0.8734,
       "step": 18175
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.445198709095572e-06,
-      "loss": 0.7495,
+      "loss": 0.8136,
       "step": 18176
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4441501541095245e-06,
-      "loss": 0.8472,
+      "loss": 0.904,
       "step": 18177
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4431016623294305e-06,
-      "loss": 0.7004,
+      "loss": 0.819,
       "step": 18178
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.442053233769835e-06,
-      "loss": 0.8187,
+      "loss": 0.8814,
       "step": 18179
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.44100486844528e-06,
-      "loss": 0.7491,
+      "loss": 0.7833,
       "step": 18180
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.439956566370322e-06,
-      "loss": 0.5856,
+      "loss": 0.7411,
       "step": 18181
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.438908327559489e-06,
-      "loss": 0.6919,
+      "loss": 0.7666,
       "step": 18182
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.437860152027334e-06,
-      "loss": 0.6793,
+      "loss": 0.7843,
       "step": 18183
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.436812039788399e-06,
-      "loss": 0.8321,
+      "loss": 0.894,
       "step": 18184
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.435763990857222e-06,
-      "loss": 0.6792,
+      "loss": 0.7925,
       "step": 18185
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4347160052483435e-06,
-      "loss": 0.7208,
+      "loss": 0.8261,
       "step": 18186
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4336680829763e-06,
-      "loss": 0.6621,
+      "loss": 0.7959,
       "step": 18187
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.43262022405564e-06,
-      "loss": 0.7528,
+      "loss": 0.8686,
       "step": 18188
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.431572428500886e-06,
-      "loss": 0.8206,
+      "loss": 0.8594,
       "step": 18189
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.430524696326587e-06,
-      "loss": 0.7661,
+      "loss": 0.9037,
       "step": 18190
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.429477027547274e-06,
-      "loss": 0.7448,
+      "loss": 0.8383,
       "step": 18191
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.428429422177484e-06,
-      "loss": 0.6678,
+      "loss": 0.8091,
       "step": 18192
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.427381880231746e-06,
-      "loss": 0.6631,
+      "loss": 0.7101,
       "step": 18193
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.426334401724603e-06,
-      "loss": 0.7606,
+      "loss": 0.8237,
       "step": 18194
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.425286986670573e-06,
-      "loss": 0.8671,
+      "loss": 0.892,
       "step": 18195
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4242396350842006e-06,
-      "loss": 0.7071,
+      "loss": 0.7753,
       "step": 18196
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.42319234698001e-06,
-      "loss": 0.8371,
+      "loss": 0.8681,
       "step": 18197
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.422145122372533e-06,
-      "loss": 0.8652,
+      "loss": 0.994,
       "step": 18198
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.421097961276291e-06,
-      "loss": 0.728,
+      "loss": 0.8549,
       "step": 18199
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.420050863705829e-06,
-      "loss": 0.7536,
+      "loss": 0.8673,
       "step": 18200
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.419003829675654e-06,
-      "loss": 0.7621,
+      "loss": 0.8328,
       "step": 18201
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.417956859200306e-06,
-      "loss": 0.7383,
+      "loss": 0.7816,
       "step": 18202
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.416909952294304e-06,
-      "loss": 0.7124,
+      "loss": 0.827,
       "step": 18203
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.415863108972175e-06,
-      "loss": 0.639,
+      "loss": 0.7555,
       "step": 18204
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.414816329248441e-06,
-      "loss": 0.6949,
+      "loss": 0.7948,
       "step": 18205
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.41376961313762e-06,
-      "loss": 0.8872,
+      "loss": 0.9578,
       "step": 18206
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.412722960654248e-06,
-      "loss": 0.7145,
+      "loss": 0.7832,
       "step": 18207
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.411676371812828e-06,
-      "loss": 0.622,
+      "loss": 0.7876,
       "step": 18208
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.410629846627893e-06,
-      "loss": 0.7085,
+      "loss": 0.7859,
       "step": 18209
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4095833851139565e-06,
-      "loss": 0.6131,
+      "loss": 0.7483,
       "step": 18210
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.408536987285538e-06,
-      "loss": 0.6728,
+      "loss": 0.7692,
       "step": 18211
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.40749065315715e-06,
-      "loss": 0.6193,
+      "loss": 0.7488,
       "step": 18212
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.406444382743321e-06,
-      "loss": 0.7661,
+      "loss": 0.802,
       "step": 18213
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.405398176058551e-06,
-      "loss": 0.7542,
+      "loss": 0.857,
       "step": 18214
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.4043520331173685e-06,
-      "loss": 0.6085,
+      "loss": 0.7552,
       "step": 18215
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.403305953934278e-06,
-      "loss": 0.6387,
+      "loss": 0.7147,
       "step": 18216
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.402259938523798e-06,
-      "loss": 0.8499,
+      "loss": 0.8555,
       "step": 18217
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.401213986900432e-06,
-      "loss": 0.6955,
+      "loss": 0.7297,
       "step": 18218
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.400168099078702e-06,
-      "loss": 0.7858,
+      "loss": 0.8665,
       "step": 18219
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3991222750731135e-06,
-      "loss": 0.6316,
+      "loss": 0.8035,
       "step": 18220
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.398076514898175e-06,
-      "loss": 0.7668,
+      "loss": 0.8934,
       "step": 18221
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.397030818568396e-06,
-      "loss": 0.6485,
+      "loss": 0.8012,
       "step": 18222
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.395985186098284e-06,
-      "loss": 0.7166,
+      "loss": 0.8416,
       "step": 18223
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3949396175023446e-06,
-      "loss": 0.6999,
+      "loss": 0.7684,
       "step": 18224
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.393894112795082e-06,
-      "loss": 0.7615,
+      "loss": 0.8403,
       "step": 18225
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3928486719910085e-06,
-      "loss": 0.7463,
+      "loss": 0.8483,
       "step": 18226
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.391803295104617e-06,
-      "loss": 0.7681,
+      "loss": 0.8236,
       "step": 18227
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.390757982150421e-06,
-      "loss": 0.7886,
+      "loss": 0.8077,
       "step": 18228
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.389712733142917e-06,
-      "loss": 0.7274,
+      "loss": 0.79,
       "step": 18229
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.388667548096608e-06,
-      "loss": 0.715,
+      "loss": 0.8054,
       "step": 18230
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.38762242702599e-06,
-      "loss": 0.6508,
+      "loss": 0.7614,
       "step": 18231
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.386577369945575e-06,
-      "loss": 0.7322,
+      "loss": 0.7989,
       "step": 18232
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.385532376869845e-06,
-      "loss": 0.7008,
+      "loss": 0.8047,
       "step": 18233
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3844874478133115e-06,
-      "loss": 0.8371,
+      "loss": 0.9077,
       "step": 18234
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.383442582790464e-06,
-      "loss": 0.7675,
+      "loss": 0.8257,
       "step": 18235
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.382397781815803e-06,
-      "loss": 0.7296,
+      "loss": 0.8426,
       "step": 18236
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.381353044903816e-06,
-      "loss": 0.7164,
+      "loss": 0.9042,
       "step": 18237
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.380308372069006e-06,
-      "loss": 0.7888,
+      "loss": 0.7835,
       "step": 18238
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.379263763325864e-06,
-      "loss": 0.5234,
+      "loss": 0.6972,
       "step": 18239
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3782192186888805e-06,
-      "loss": 0.854,
+      "loss": 0.9177,
       "step": 18240
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.377174738172544e-06,
-      "loss": 0.7665,
+      "loss": 0.8553,
       "step": 18241
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.376130321791358e-06,
-      "loss": 0.6395,
+      "loss": 0.7805,
       "step": 18242
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.375085969559798e-06,
-      "loss": 0.7098,
+      "loss": 0.743,
       "step": 18243
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3740416814923544e-06,
-      "loss": 0.7842,
+      "loss": 0.8344,
       "step": 18244
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.372997457603528e-06,
-      "loss": 0.8113,
+      "loss": 0.8609,
       "step": 18245
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.371953297907788e-06,
-      "loss": 0.7625,
+      "loss": 0.8645,
       "step": 18246
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.370909202419634e-06,
-      "loss": 0.6797,
+      "loss": 0.7547,
       "step": 18247
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.369865171153548e-06,
-      "loss": 0.622,
+      "loss": 0.7575,
       "step": 18248
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.368821204124013e-06,
-      "loss": 0.7909,
+      "loss": 0.9148,
       "step": 18249
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.367777301345508e-06,
-      "loss": 0.7184,
+      "loss": 0.7844,
       "step": 18250
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.366733462832525e-06,
-      "loss": 0.776,
+      "loss": 0.8916,
       "step": 18251
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.365689688599543e-06,
-      "loss": 0.7666,
+      "loss": 0.8736,
       "step": 18252
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.36464597866104e-06,
-      "loss": 0.629,
+      "loss": 0.7448,
       "step": 18253
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.363602333031497e-06,
-      "loss": 0.7568,
+      "loss": 0.9139,
       "step": 18254
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.362558751725394e-06,
-      "loss": 0.7237,
+      "loss": 0.8501,
       "step": 18255
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3615152347572045e-06,
-      "loss": 0.8686,
+      "loss": 0.955,
       "step": 18256
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.360471782141414e-06,
-      "loss": 0.7024,
+      "loss": 0.713,
       "step": 18257
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.359428393892495e-06,
-      "loss": 0.9137,
+      "loss": 0.9153,
       "step": 18258
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.358385070024923e-06,
-      "loss": 0.6958,
+      "loss": 0.8299,
       "step": 18259
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.357341810553169e-06,
-      "loss": 0.9577,
+      "loss": 0.8826,
       "step": 18260
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3562986154917175e-06,
-      "loss": 0.8298,
+      "loss": 0.9052,
       "step": 18261
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.355255484855031e-06,
-      "loss": 0.8543,
+      "loss": 0.9622,
       "step": 18262
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.354212418657579e-06,
-      "loss": 0.9206,
+      "loss": 0.9361,
       "step": 18263
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.353169416913843e-06,
-      "loss": 0.7842,
+      "loss": 0.8888,
       "step": 18264
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3521264796382884e-06,
-      "loss": 0.6853,
+      "loss": 0.834,
       "step": 18265
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.351083606845385e-06,
-      "loss": 0.7459,
+      "loss": 0.7994,
       "step": 18266
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3500407985496e-06,
-      "loss": 0.6406,
+      "loss": 0.7473,
       "step": 18267
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.348998054765402e-06,
-      "loss": 0.668,
+      "loss": 0.7386,
       "step": 18268
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.347955375507252e-06,
-      "loss": 0.6712,
+      "loss": 0.7636,
       "step": 18269
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.346912760789626e-06,
-      "loss": 0.5961,
+      "loss": 0.753,
       "step": 18270
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.345870210626983e-06,
-      "loss": 0.6794,
+      "loss": 0.7926,
       "step": 18271
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.344827725033788e-06,
-      "loss": 0.8352,
+      "loss": 0.8001,
       "step": 18272
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.343785304024499e-06,
-      "loss": 0.7475,
+      "loss": 0.8144,
       "step": 18273
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.34274294761359e-06,
-      "loss": 0.7333,
+      "loss": 0.8305,
       "step": 18274
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3417006558155085e-06,
-      "loss": 0.8022,
+      "loss": 0.906,
       "step": 18275
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.340658428644724e-06,
-      "loss": 0.7779,
+      "loss": 0.9095,
       "step": 18276
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.339616266115693e-06,
-      "loss": 0.7679,
+      "loss": 0.8761,
       "step": 18277
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3385741682428735e-06,
-      "loss": 0.8148,
+      "loss": 0.8775,
       "step": 18278
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3375321350407215e-06,
-      "loss": 0.6292,
+      "loss": 0.8202,
       "step": 18279
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.3364901665237024e-06,
-      "loss": 0.6579,
+      "loss": 0.8249,
       "step": 18280
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.335448262706262e-06,
-      "loss": 0.845,
+      "loss": 0.92,
       "step": 18281
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.334406423602856e-06,
-      "loss": 0.6923,
+      "loss": 0.8038,
       "step": 18282
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.333364649227944e-06,
-      "loss": 0.6145,
+      "loss": 0.7375,
       "step": 18283
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.332322939595977e-06,
-      "loss": 0.7583,
+      "loss": 0.8527,
       "step": 18284
     },
     {
       "epoch": 0.66,
       "learning_rate": 5.331281294721406e-06,
-      "loss": 0.7437,
+      "loss": 0.8395,
       "step": 18285
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.330239714618679e-06,
-      "loss": 0.7427,
+      "loss": 0.908,
       "step": 18286
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.329198199302259e-06,
-      "loss": 0.6418,
+      "loss": 0.7356,
       "step": 18287
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.32815674878658e-06,
-      "loss": 0.7141,
+      "loss": 0.8301,
       "step": 18288
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.327115363086099e-06,
-      "loss": 0.8276,
+      "loss": 0.8916,
       "step": 18289
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.3260740422152635e-06,
-      "loss": 0.7689,
+      "loss": 0.8825,
       "step": 18290
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.32503278618852e-06,
-      "loss": 0.7814,
+      "loss": 0.8618,
       "step": 18291
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.323991595020308e-06,
-      "loss": 0.6862,
+      "loss": 0.7739,
       "step": 18292
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.322950468725087e-06,
-      "loss": 0.8451,
+      "loss": 0.8873,
       "step": 18293
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.321909407317285e-06,
-      "loss": 0.7999,
+      "loss": 0.8714,
       "step": 18294
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.320868410811357e-06,
-      "loss": 0.6483,
+      "loss": 0.7386,
       "step": 18295
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.319827479221739e-06,
-      "loss": 0.6745,
+      "loss": 0.7509,
       "step": 18296
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.318786612562877e-06,
-      "loss": 0.7716,
+      "loss": 0.8063,
       "step": 18297
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.317745810849204e-06,
-      "loss": 0.6913,
+      "loss": 0.7617,
       "step": 18298
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.316705074095172e-06,
-      "loss": 0.7796,
+      "loss": 0.9484,
       "step": 18299
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.315664402315208e-06,
-      "loss": 0.8383,
+      "loss": 0.9391,
       "step": 18300
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.314623795523753e-06,
-      "loss": 0.7075,
+      "loss": 0.8134,
       "step": 18301
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.313583253735246e-06,
-      "loss": 0.7782,
+      "loss": 0.8686,
       "step": 18302
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.312542776964124e-06,
-      "loss": 0.7439,
+      "loss": 0.8456,
       "step": 18303
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.3115023652248185e-06,
-      "loss": 0.6964,
+      "loss": 0.7971,
       "step": 18304
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.310462018531763e-06,
-      "loss": 0.693,
+      "loss": 0.8377,
       "step": 18305
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.3094217368994e-06,
-      "loss": 0.7285,
+      "loss": 0.7592,
       "step": 18306
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.308381520342149e-06,
-      "loss": 0.7733,
+      "loss": 0.8199,
       "step": 18307
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.307341368874452e-06,
-      "loss": 0.6699,
+      "loss": 0.7876,
       "step": 18308
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.306301282510735e-06,
-      "loss": 0.7692,
+      "loss": 0.8733,
       "step": 18309
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.305261261265429e-06,
-      "loss": 0.6678,
+      "loss": 0.7462,
       "step": 18310
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.304221305152957e-06,
-      "loss": 0.8,
+      "loss": 0.811,
       "step": 18311
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.303181414187759e-06,
-      "loss": 0.7348,
+      "loss": 0.8694,
       "step": 18312
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.302141588384249e-06,
-      "loss": 0.591,
+      "loss": 0.73,
       "step": 18313
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.30110182775686e-06,
-      "loss": 0.8091,
+      "loss": 0.864,
       "step": 18314
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.30006213232002e-06,
-      "loss": 0.7742,
+      "loss": 0.8631,
       "step": 18315
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.299022502088147e-06,
-      "loss": 0.7585,
+      "loss": 0.8451,
       "step": 18316
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.297982937075666e-06,
-      "loss": 0.7616,
+      "loss": 0.8023,
       "step": 18317
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.296943437297001e-06,
-      "loss": 0.7104,
+      "loss": 0.7963,
       "step": 18318
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.295904002766579e-06,
-      "loss": 0.7646,
+      "loss": 0.8663,
       "step": 18319
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.294864633498805e-06,
-      "loss": 0.6605,
+      "loss": 0.8234,
       "step": 18320
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.293825329508113e-06,
-      "loss": 0.6777,
+      "loss": 0.7958,
       "step": 18321
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.292786090808919e-06,
-      "loss": 0.7181,
+      "loss": 0.8063,
       "step": 18322
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2917469174156386e-06,
-      "loss": 0.6651,
+      "loss": 0.7717,
       "step": 18323
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.290707809342686e-06,
-      "loss": 0.6695,
+      "loss": 0.8026,
       "step": 18324
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2896687666044875e-06,
-      "loss": 0.7038,
+      "loss": 0.7718,
       "step": 18325
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2886297892154445e-06,
-      "loss": 0.7324,
+      "loss": 0.7756,
       "step": 18326
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.287590877189982e-06,
-      "loss": 0.7782,
+      "loss": 0.8683,
       "step": 18327
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.286552030542511e-06,
-      "loss": 0.6368,
+      "loss": 0.7705,
       "step": 18328
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.285513249287443e-06,
-      "loss": 0.5882,
+      "loss": 0.7705,
       "step": 18329
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2844745334391855e-06,
-      "loss": 0.8484,
+      "loss": 0.8498,
       "step": 18330
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2834358830121565e-06,
-      "loss": 0.6452,
+      "loss": 0.7235,
       "step": 18331
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.282397298020764e-06,
-      "loss": 0.7674,
+      "loss": 0.8775,
       "step": 18332
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.281358778479415e-06,
-      "loss": 0.7533,
+      "loss": 0.8198,
       "step": 18333
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.280320324402518e-06,
-      "loss": 0.7496,
+      "loss": 0.8553,
       "step": 18334
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.279281935804481e-06,
-      "loss": 0.7132,
+      "loss": 0.8246,
       "step": 18335
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.278243612699704e-06,
-      "loss": 0.7058,
+      "loss": 0.856,
       "step": 18336
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.277205355102603e-06,
-      "loss": 0.7955,
+      "loss": 0.8086,
       "step": 18337
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.276167163027576e-06,
-      "loss": 0.6902,
+      "loss": 0.7404,
       "step": 18338
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.275129036489026e-06,
-      "loss": 0.8226,
+      "loss": 0.8631,
       "step": 18339
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.274090975501359e-06,
-      "loss": 0.8324,
+      "loss": 0.8422,
       "step": 18340
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2730529800789745e-06,
-      "loss": 0.7243,
+      "loss": 0.8457,
       "step": 18341
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.272015050236272e-06,
-      "loss": 0.6946,
+      "loss": 0.8252,
       "step": 18342
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2709771859876495e-06,
-      "loss": 0.6736,
+      "loss": 0.781,
       "step": 18343
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.269939387347517e-06,
-      "loss": 0.7724,
+      "loss": 0.8347,
       "step": 18344
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.268901654330256e-06,
-      "loss": 0.7356,
+      "loss": 0.8829,
       "step": 18345
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2678639869502765e-06,
-      "loss": 0.726,
+      "loss": 0.8263,
       "step": 18346
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.266826385221969e-06,
-      "loss": 0.7158,
+      "loss": 0.7901,
       "step": 18347
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.265788849159731e-06,
-      "loss": 0.7896,
+      "loss": 0.9073,
       "step": 18348
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.264751378777951e-06,
-      "loss": 0.753,
+      "loss": 0.8475,
       "step": 18349
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.263713974091032e-06,
-      "loss": 0.746,
+      "loss": 0.8589,
       "step": 18350
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2626766351133615e-06,
-      "loss": 0.7724,
+      "loss": 0.8082,
       "step": 18351
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.261639361859331e-06,
-      "loss": 0.8142,
+      "loss": 0.846,
       "step": 18352
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.260602154343327e-06,
-      "loss": 0.8824,
+      "loss": 0.9334,
       "step": 18353
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.259565012579753e-06,
-      "loss": 0.7355,
+      "loss": 0.9126,
       "step": 18354
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.25852793658298e-06,
-      "loss": 0.6325,
+      "loss": 0.8037,
       "step": 18355
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.257490926367408e-06,
-      "loss": 0.7819,
+      "loss": 0.8691,
       "step": 18356
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.256453981947421e-06,
-      "loss": 0.7956,
+      "loss": 0.9081,
       "step": 18357
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.255417103337406e-06,
-      "loss": 0.6715,
+      "loss": 0.7783,
       "step": 18358
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.254380290551746e-06,
-      "loss": 0.7135,
+      "loss": 0.7458,
       "step": 18359
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.253343543604827e-06,
-      "loss": 0.6578,
+      "loss": 0.7819,
       "step": 18360
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.25230686251103e-06,
-      "loss": 0.7854,
+      "loss": 0.8286,
       "step": 18361
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.251270247284736e-06,
-      "loss": 0.7877,
+      "loss": 0.804,
       "step": 18362
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.250233697940334e-06,
-      "loss": 0.7013,
+      "loss": 0.8499,
       "step": 18363
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.249197214492201e-06,
-      "loss": 0.5034,
+      "loss": 0.6978,
       "step": 18364
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.248160796954715e-06,
-      "loss": 0.7025,
+      "loss": 0.7569,
       "step": 18365
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.247124445342258e-06,
-      "loss": 0.712,
+      "loss": 0.8414,
       "step": 18366
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.246088159669204e-06,
-      "loss": 0.6529,
+      "loss": 0.8427,
       "step": 18367
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.245051939949928e-06,
-      "loss": 0.7222,
+      "loss": 0.8773,
       "step": 18368
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.244015786198814e-06,
-      "loss": 0.639,
+      "loss": 0.7349,
       "step": 18369
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.242979698430233e-06,
-      "loss": 0.7918,
+      "loss": 0.9106,
       "step": 18370
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.241943676658561e-06,
-      "loss": 0.7791,
+      "loss": 0.8225,
       "step": 18371
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2409077208981625e-06,
-      "loss": 0.6338,
+      "loss": 0.7668,
       "step": 18372
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.239871831163428e-06,
-      "loss": 0.834,
+      "loss": 0.8518,
       "step": 18373
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2388360074687095e-06,
-      "loss": 0.6437,
+      "loss": 0.7325,
       "step": 18374
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2378002498283885e-06,
-      "loss": 0.8342,
+      "loss": 0.9773,
       "step": 18375
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.236764558256835e-06,
-      "loss": 0.8223,
+      "loss": 0.8384,
       "step": 18376
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.235728932768414e-06,
-      "loss": 0.754,
+      "loss": 0.8818,
       "step": 18377
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2346933733774926e-06,
-      "loss": 0.6822,
+      "loss": 0.7996,
       "step": 18378
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.233657880098441e-06,
-      "loss": 0.7742,
+      "loss": 0.8324,
       "step": 18379
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.232622452945625e-06,
-      "loss": 0.7795,
+      "loss": 0.838,
       "step": 18380
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.231587091933403e-06,
-      "loss": 0.8593,
+      "loss": 0.8704,
       "step": 18381
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.230551797076149e-06,
-      "loss": 0.8004,
+      "loss": 0.8536,
       "step": 18382
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.229516568388222e-06,
-      "loss": 0.6251,
+      "loss": 0.7272,
       "step": 18383
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.228481405883984e-06,
-      "loss": 0.6766,
+      "loss": 0.8089,
       "step": 18384
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.227446309577793e-06,
-      "loss": 0.7161,
+      "loss": 0.7737,
       "step": 18385
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2264112794840205e-06,
-      "loss": 0.6719,
+      "loss": 0.8368,
       "step": 18386
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.22537631561701e-06,
-      "loss": 0.6529,
+      "loss": 0.8172,
       "step": 18387
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.224341417991134e-06,
-      "loss": 0.7029,
+      "loss": 0.7695,
       "step": 18388
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.223306586620744e-06,
-      "loss": 0.7989,
+      "loss": 0.8359,
       "step": 18389
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2222718215201995e-06,
-      "loss": 0.694,
+      "loss": 0.8526,
       "step": 18390
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.22123712270385e-06,
-      "loss": 0.7119,
+      "loss": 0.808,
       "step": 18391
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.220202490186063e-06,
-      "loss": 0.7066,
+      "loss": 0.8254,
       "step": 18392
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2191679239811764e-06,
-      "loss": 0.7082,
+      "loss": 0.8355,
       "step": 18393
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.218133424103558e-06,
-      "loss": 0.6741,
+      "loss": 0.787,
       "step": 18394
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2170989905675505e-06,
-      "loss": 0.8035,
+      "loss": 0.8656,
       "step": 18395
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.21606462338751e-06,
-      "loss": 0.7078,
+      "loss": 0.7541,
       "step": 18396
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2150303225777854e-06,
-      "loss": 0.8408,
+      "loss": 0.8953,
       "step": 18397
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2139960881527266e-06,
-      "loss": 0.8374,
+      "loss": 0.8861,
       "step": 18398
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.21296192012668e-06,
-      "loss": 0.6628,
+      "loss": 0.7875,
       "step": 18399
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.211927818513992e-06,
-      "loss": 0.7608,
+      "loss": 0.8266,
       "step": 18400
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.210893783329016e-06,
-      "loss": 0.7749,
+      "loss": 0.8408,
       "step": 18401
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.2098598145860934e-06,
-      "loss": 0.6499,
+      "loss": 0.7845,
       "step": 18402
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.208825912299571e-06,
-      "loss": 0.7789,
+      "loss": 0.8471,
       "step": 18403
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.207792076483786e-06,
-      "loss": 0.7491,
+      "loss": 0.885,
       "step": 18404
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.206758307153095e-06,
-      "loss": 0.6558,
+      "loss": 0.7683,
       "step": 18405
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.205724604321823e-06,
-      "loss": 0.6919,
+      "loss": 0.8434,
       "step": 18406
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.204690968004325e-06,
-      "loss": 0.6857,
+      "loss": 0.8076,
       "step": 18407
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.203657398214936e-06,
-      "loss": 0.727,
+      "loss": 0.9052,
       "step": 18408
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.202623894967996e-06,
-      "loss": 0.7485,
+      "loss": 0.8781,
       "step": 18409
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.201590458277837e-06,
-      "loss": 0.762,
+      "loss": 0.8063,
       "step": 18410
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.200557088158812e-06,
-      "loss": 0.6692,
+      "loss": 0.7306,
       "step": 18411
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.19952378462524e-06,
-      "loss": 0.6308,
+      "loss": 0.8235,
       "step": 18412
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1984905476914686e-06,
-      "loss": 0.5177,
+      "loss": 0.7535,
       "step": 18413
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.197457377371826e-06,
-      "loss": 0.6886,
+      "loss": 0.8386,
       "step": 18414
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.196424273680651e-06,
-      "loss": 0.678,
+      "loss": 0.7433,
       "step": 18415
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.195391236632272e-06,
-      "loss": 0.6575,
+      "loss": 0.8075,
       "step": 18416
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.19435826624102e-06,
-      "loss": 0.6747,
+      "loss": 0.8228,
       "step": 18417
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.193325362521236e-06,
-      "loss": 0.6156,
+      "loss": 0.8062,
       "step": 18418
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1922925254872355e-06,
-      "loss": 0.7848,
+      "loss": 0.8495,
       "step": 18419
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.191259755153359e-06,
-      "loss": 0.7812,
+      "loss": 0.8807,
       "step": 18420
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1902270515339305e-06,
-      "loss": 0.7093,
+      "loss": 0.8507,
       "step": 18421
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.189194414643277e-06,
-      "loss": 0.7642,
+      "loss": 0.8398,
       "step": 18422
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1881618444957224e-06,
-      "loss": 0.7653,
+      "loss": 0.8523,
       "step": 18423
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.187129341105602e-06,
-      "loss": 0.6469,
+      "loss": 0.7312,
       "step": 18424
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.186096904487225e-06,
-      "loss": 0.775,
+      "loss": 0.8996,
       "step": 18425
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.185064534654929e-06,
-      "loss": 0.7133,
+      "loss": 0.8472,
       "step": 18426
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.18403223162303e-06,
-      "loss": 0.8416,
+      "loss": 0.9064,
       "step": 18427
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.182999995405851e-06,
-      "loss": 0.7033,
+      "loss": 0.8114,
       "step": 18428
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1819678260177086e-06,
-      "loss": 0.7391,
+      "loss": 0.8282,
       "step": 18429
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1809357234729314e-06,
-      "loss": 0.6437,
+      "loss": 0.8183,
       "step": 18430
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.179903687785834e-06,
-      "loss": 0.7376,
+      "loss": 0.8575,
       "step": 18431
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1788717189707326e-06,
-      "loss": 0.7499,
+      "loss": 0.8236,
       "step": 18432
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1778398170419465e-06,
-      "loss": 0.6508,
+      "loss": 0.7382,
       "step": 18433
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.176807982013792e-06,
-      "loss": 0.6908,
+      "loss": 0.7149,
       "step": 18434
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.175776213900582e-06,
-      "loss": 0.6362,
+      "loss": 0.774,
       "step": 18435
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.17474451271663e-06,
-      "loss": 0.7165,
+      "loss": 0.7975,
       "step": 18436
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.173712878476258e-06,
-      "loss": 0.6895,
+      "loss": 0.775,
       "step": 18437
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.172681311193765e-06,
-      "loss": 0.758,
+      "loss": 0.8743,
       "step": 18438
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.171649810883474e-06,
-      "loss": 0.5826,
+      "loss": 0.7482,
       "step": 18439
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.170618377559691e-06,
-      "loss": 0.7799,
+      "loss": 0.903,
       "step": 18440
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.169587011236725e-06,
-      "loss": 0.5945,
+      "loss": 0.8376,
       "step": 18441
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.168555711928884e-06,
-      "loss": 0.7015,
+      "loss": 0.8076,
       "step": 18442
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1675244796504845e-06,
-      "loss": 0.9401,
+      "loss": 0.9106,
       "step": 18443
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.166493314415818e-06,
-      "loss": 0.7058,
+      "loss": 0.7753,
       "step": 18444
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.165462216239203e-06,
-      "loss": 0.7105,
+      "loss": 0.7836,
       "step": 18445
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.164431185134942e-06,
-      "loss": 0.8037,
+      "loss": 0.8594,
       "step": 18446
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.163400221117336e-06,
-      "loss": 0.6356,
+      "loss": 0.7762,
       "step": 18447
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.162369324200686e-06,
-      "loss": 0.7386,
+      "loss": 0.8177,
       "step": 18448
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.161338494399302e-06,
-      "loss": 0.7718,
+      "loss": 0.8336,
       "step": 18449
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.160307731727481e-06,
-      "loss": 0.745,
+      "loss": 0.7774,
       "step": 18450
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.159277036199524e-06,
-      "loss": 0.578,
+      "loss": 0.7382,
       "step": 18451
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.158246407829726e-06,
-      "loss": 0.6885,
+      "loss": 0.7717,
       "step": 18452
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.157215846632396e-06,
-      "loss": 0.8184,
+      "loss": 0.8954,
       "step": 18453
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.156185352621823e-06,
-      "loss": 0.6869,
+      "loss": 0.8094,
       "step": 18454
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.155154925812301e-06,
-      "loss": 0.8349,
+      "loss": 0.89,
       "step": 18455
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.15412456621814e-06,
-      "loss": 0.679,
+      "loss": 0.7628,
       "step": 18456
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.153094273853614e-06,
-      "loss": 0.6714,
+      "loss": 0.7738,
       "step": 18457
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.152064048733034e-06,
-      "loss": 0.7496,
+      "loss": 0.8434,
       "step": 18458
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1510338908706866e-06,
-      "loss": 0.8045,
+      "loss": 0.8419,
       "step": 18459
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.150003800280865e-06,
-      "loss": 0.8327,
+      "loss": 0.8885,
       "step": 18460
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.148973776977856e-06,
-      "loss": 0.7582,
+      "loss": 0.8513,
       "step": 18461
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.147943820975956e-06,
-      "loss": 0.7169,
+      "loss": 0.8553,
       "step": 18462
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1469139322894514e-06,
-      "loss": 0.6859,
+      "loss": 0.8121,
       "step": 18463
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.145884110932631e-06,
-      "loss": 0.7319,
+      "loss": 0.8272,
       "step": 18464
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.144854356919779e-06,
-      "loss": 0.8671,
+      "loss": 0.8724,
       "step": 18465
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.143824670265187e-06,
-      "loss": 0.8039,
+      "loss": 0.8445,
       "step": 18466
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.142795050983134e-06,
-      "loss": 0.8516,
+      "loss": 0.9528,
       "step": 18467
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.141765499087909e-06,
-      "loss": 0.7104,
+      "loss": 0.8388,
       "step": 18468
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.140736014593798e-06,
-      "loss": 0.6426,
+      "loss": 0.7678,
       "step": 18469
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.139706597515079e-06,
-      "loss": 0.6934,
+      "loss": 0.8102,
       "step": 18470
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.138677247866031e-06,
-      "loss": 0.775,
+      "loss": 0.8125,
       "step": 18471
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.137647965660948e-06,
-      "loss": 0.721,
+      "loss": 0.8116,
       "step": 18472
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.136618750914096e-06,
-      "loss": 0.6246,
+      "loss": 0.6977,
       "step": 18473
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.135589603639755e-06,
-      "loss": 0.6504,
+      "loss": 0.8042,
       "step": 18474
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.134560523852209e-06,
-      "loss": 0.709,
+      "loss": 0.7907,
       "step": 18475
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.133531511565734e-06,
-      "loss": 0.7044,
+      "loss": 0.7767,
       "step": 18476
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.132502566794604e-06,
-      "loss": 0.6782,
+      "loss": 0.7625,
       "step": 18477
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.131473689553094e-06,
-      "loss": 0.8243,
+      "loss": 0.9481,
       "step": 18478
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.13044487985548e-06,
-      "loss": 0.7436,
+      "loss": 0.8039,
       "step": 18479
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.129416137716031e-06,
-      "loss": 0.7754,
+      "loss": 0.8419,
       "step": 18480
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.128387463149027e-06,
-      "loss": 0.688,
+      "loss": 0.8116,
       "step": 18481
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.127358856168734e-06,
-      "loss": 0.6907,
+      "loss": 0.7631,
       "step": 18482
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.126330316789423e-06,
-      "loss": 0.7629,
+      "loss": 0.8185,
       "step": 18483
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.125301845025361e-06,
-      "loss": 0.7468,
+      "loss": 0.8551,
       "step": 18484
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.124273440890827e-06,
-      "loss": 0.6371,
+      "loss": 0.7067,
       "step": 18485
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.123245104400073e-06,
-      "loss": 0.7158,
+      "loss": 0.8527,
       "step": 18486
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.122216835567378e-06,
-      "loss": 0.6342,
+      "loss": 0.7348,
       "step": 18487
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.121188634407005e-06,
-      "loss": 0.6761,
+      "loss": 0.8877,
       "step": 18488
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.120160500933216e-06,
-      "loss": 0.5591,
+      "loss": 0.782,
       "step": 18489
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.119132435160275e-06,
-      "loss": 0.6139,
+      "loss": 0.7351,
       "step": 18490
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.118104437102451e-06,
-      "loss": 0.8103,
+      "loss": 0.8535,
       "step": 18491
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.117076506773999e-06,
-      "loss": 0.7882,
+      "loss": 0.8527,
       "step": 18492
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1160486441891785e-06,
-      "loss": 0.7181,
+      "loss": 0.8053,
       "step": 18493
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.115020849362258e-06,
-      "loss": 0.7606,
+      "loss": 0.9046,
       "step": 18494
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.113993122307492e-06,
-      "loss": 0.61,
+      "loss": 0.7957,
       "step": 18495
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.112965463039139e-06,
-      "loss": 0.7314,
+      "loss": 0.7907,
       "step": 18496
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.111937871571453e-06,
-      "loss": 0.7993,
+      "loss": 0.907,
       "step": 18497
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1109103479187e-06,
-      "loss": 0.7749,
+      "loss": 0.86,
       "step": 18498
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.109882892095124e-06,
-      "loss": 0.728,
+      "loss": 0.8289,
       "step": 18499
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.108855504114987e-06,
-      "loss": 0.7768,
+      "loss": 0.902,
       "step": 18500
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.10782818399254e-06,
-      "loss": 0.6891,
+      "loss": 0.7685,
       "step": 18501
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1068009317420365e-06,
-      "loss": 0.7554,
+      "loss": 0.8841,
       "step": 18502
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1057737473777225e-06,
-      "loss": 0.6094,
+      "loss": 0.7305,
       "step": 18503
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.104746630913863e-06,
-      "loss": 0.7004,
+      "loss": 0.8404,
       "step": 18504
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.10371958236469e-06,
-      "loss": 0.7003,
+      "loss": 0.7709,
       "step": 18505
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.102692601744466e-06,
-      "loss": 0.71,
+      "loss": 0.8182,
       "step": 18506
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.1016656890674325e-06,
-      "loss": 0.7256,
+      "loss": 0.7856,
       "step": 18507
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.100638844347838e-06,
-      "loss": 0.6826,
+      "loss": 0.7957,
       "step": 18508
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.099612067599924e-06,
-      "loss": 0.6619,
+      "loss": 0.8252,
       "step": 18509
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.098585358837949e-06,
-      "loss": 0.7183,
+      "loss": 0.7309,
       "step": 18510
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.097558718076142e-06,
-      "loss": 0.857,
+      "loss": 0.8596,
       "step": 18511
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.09653214532875e-06,
-      "loss": 0.6393,
+      "loss": 0.7494,
       "step": 18512
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.09550564061002e-06,
-      "loss": 0.6695,
+      "loss": 0.6927,
       "step": 18513
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.09447920393419e-06,
-      "loss": 0.8537,
+      "loss": 0.8615,
       "step": 18514
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.093452835315503e-06,
-      "loss": 0.7754,
+      "loss": 0.8815,
       "step": 18515
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0924265347681924e-06,
-      "loss": 0.7108,
+      "loss": 0.8142,
       "step": 18516
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.091400302306509e-06,
-      "loss": 0.8064,
+      "loss": 0.8598,
       "step": 18517
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.090374137944671e-06,
-      "loss": 0.5961,
+      "loss": 0.7664,
       "step": 18518
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0893480416969334e-06,
-      "loss": 0.6436,
+      "loss": 0.8153,
       "step": 18519
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0883220135775225e-06,
-      "loss": 0.7435,
+      "loss": 0.8903,
       "step": 18520
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.087296053600675e-06,
-      "loss": 0.7025,
+      "loss": 0.8473,
       "step": 18521
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.086270161780622e-06,
-      "loss": 0.7559,
+      "loss": 0.8105,
       "step": 18522
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.085244338131605e-06,
-      "loss": 0.6904,
+      "loss": 0.8082,
       "step": 18523
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0842185826678425e-06,
-      "loss": 0.6296,
+      "loss": 0.6944,
       "step": 18524
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.083192895403577e-06,
-      "loss": 0.6501,
+      "loss": 0.7631,
       "step": 18525
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.082167276353033e-06,
-      "loss": 0.6704,
+      "loss": 0.7353,
       "step": 18526
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.081141725530441e-06,
-      "loss": 0.6169,
+      "loss": 0.7088,
       "step": 18527
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.080116242950025e-06,
-      "loss": 0.7671,
+      "loss": 0.8597,
       "step": 18528
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.079090828626019e-06,
-      "loss": 0.8061,
+      "loss": 0.8956,
       "step": 18529
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.078065482572651e-06,
-      "loss": 0.6794,
+      "loss": 0.7625,
       "step": 18530
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.077040204804132e-06,
-      "loss": 0.7133,
+      "loss": 0.8313,
       "step": 18531
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0760149953347e-06,
-      "loss": 0.7479,
+      "loss": 0.8301,
       "step": 18532
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0749898541785735e-06,
-      "loss": 0.6437,
+      "loss": 0.6918,
       "step": 18533
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.073964781349976e-06,
-      "loss": 0.8024,
+      "loss": 0.8919,
       "step": 18534
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.072939776863123e-06,
-      "loss": 0.7675,
+      "loss": 0.8296,
       "step": 18535
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.071914840732248e-06,
-      "loss": 0.5856,
+      "loss": 0.7202,
       "step": 18536
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.070889972971554e-06,
-      "loss": 0.6719,
+      "loss": 0.8048,
       "step": 18537
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0698651735952746e-06,
-      "loss": 0.8984,
+      "loss": 0.8791,
       "step": 18538
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.068840442617617e-06,
-      "loss": 0.6898,
+      "loss": 0.7944,
       "step": 18539
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.067815780052805e-06,
-      "loss": 0.7665,
+      "loss": 0.853,
       "step": 18540
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.066791185915046e-06,
-      "loss": 0.7598,
+      "loss": 0.8271,
       "step": 18541
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.06576666021857e-06,
-      "loss": 0.7984,
+      "loss": 0.8414,
       "step": 18542
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0647422029775704e-06,
-      "loss": 0.9117,
+      "loss": 0.9158,
       "step": 18543
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0637178142062745e-06,
-      "loss": 0.6878,
+      "loss": 0.7374,
       "step": 18544
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.062693493918891e-06,
-      "loss": 0.6499,
+      "loss": 0.7812,
       "step": 18545
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.061669242129631e-06,
-      "loss": 0.8045,
+      "loss": 0.9325,
       "step": 18546
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.060645058852698e-06,
-      "loss": 0.5865,
+      "loss": 0.8012,
       "step": 18547
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.059620944102313e-06,
-      "loss": 0.672,
+      "loss": 0.8626,
       "step": 18548
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.058596897892681e-06,
-      "loss": 0.904,
+      "loss": 0.99,
       "step": 18549
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0575729202379965e-06,
-      "loss": 0.6544,
+      "loss": 0.7508,
       "step": 18550
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.056549011152481e-06,
-      "loss": 0.7664,
+      "loss": 0.8113,
       "step": 18551
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.055525170650333e-06,
-      "loss": 0.5869,
+      "loss": 0.7325,
       "step": 18552
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.054501398745759e-06,
-      "loss": 0.7605,
+      "loss": 0.8201,
       "step": 18553
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.053477695452958e-06,
-      "loss": 0.7861,
+      "loss": 0.7685,
       "step": 18554
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.0524540607861416e-06,
-      "loss": 0.7519,
+      "loss": 0.8563,
       "step": 18555
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.051430494759498e-06,
-      "loss": 0.6606,
+      "loss": 0.7515,
       "step": 18556
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.05040699738724e-06,
-      "loss": 0.6522,
+      "loss": 0.7786,
       "step": 18557
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.049383568683562e-06,
-      "loss": 0.7563,
+      "loss": 0.8283,
       "step": 18558
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.048360208662664e-06,
-      "loss": 0.6915,
+      "loss": 0.8135,
       "step": 18559
     },
     {
       "epoch": 0.67,
       "learning_rate": 5.047336917338738e-06,
-      "loss": 0.6378,
+      "loss": 0.7752,
       "step": 18560
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.046313694725988e-06,
-      "loss": 0.7816,
+      "loss": 0.8839,
       "step": 18561
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.045290540838609e-06,
-      "loss": 0.6282,
+      "loss": 0.7162,
       "step": 18562
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.044267455690792e-06,
-      "loss": 0.6742,
+      "loss": 0.77,
       "step": 18563
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.043244439296735e-06,
-      "loss": 0.7496,
+      "loss": 0.839,
       "step": 18564
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.042221491670627e-06,
-      "loss": 0.7512,
+      "loss": 0.8937,
       "step": 18565
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.041198612826659e-06,
-      "loss": 0.768,
+      "loss": 0.8785,
       "step": 18566
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.040175802779028e-06,
-      "loss": 0.7602,
+      "loss": 0.9015,
       "step": 18567
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.0391530615419205e-06,
-      "loss": 0.7228,
+      "loss": 0.8651,
       "step": 18568
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.038130389129525e-06,
-      "loss": 0.7287,
+      "loss": 0.8555,
       "step": 18569
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.0371077855560324e-06,
-      "loss": 0.8089,
+      "loss": 0.8898,
       "step": 18570
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.036085250835626e-06,
-      "loss": 0.7402,
+      "loss": 0.8591,
       "step": 18571
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.035062784982494e-06,
-      "loss": 0.8089,
+      "loss": 0.832,
       "step": 18572
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.034040388010819e-06,
-      "loss": 0.6993,
+      "loss": 0.8604,
       "step": 18573
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.033018059934791e-06,
-      "loss": 0.6747,
+      "loss": 0.767,
       "step": 18574
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.0319958007685895e-06,
-      "loss": 0.6896,
+      "loss": 0.8582,
       "step": 18575
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.030973610526398e-06,
-      "loss": 0.6543,
+      "loss": 0.7612,
       "step": 18576
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.029951489222398e-06,
-      "loss": 0.6406,
+      "loss": 0.82,
       "step": 18577
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.028929436870768e-06,
-      "loss": 0.6697,
+      "loss": 0.7976,
       "step": 18578
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.027907453485687e-06,
-      "loss": 0.8675,
+      "loss": 0.9101,
       "step": 18579
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.02688553908134e-06,
-      "loss": 0.6738,
+      "loss": 0.7644,
       "step": 18580
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.025863693671899e-06,
-      "loss": 0.7562,
+      "loss": 0.8317,
       "step": 18581
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.024841917271541e-06,
-      "loss": 0.6994,
+      "loss": 0.8434,
       "step": 18582
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.023820209894441e-06,
-      "loss": 0.7478,
+      "loss": 0.8361,
       "step": 18583
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.022798571554781e-06,
-      "loss": 0.695,
+      "loss": 0.7717,
       "step": 18584
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.021777002266722e-06,
-      "loss": 0.6904,
+      "loss": 0.8265,
       "step": 18585
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.020755502044447e-06,
-      "loss": 0.778,
+      "loss": 0.8415,
       "step": 18586
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.019734070902127e-06,
-      "loss": 0.7286,
+      "loss": 0.8172,
       "step": 18587
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.018712708853929e-06,
-      "loss": 0.8228,
+      "loss": 0.9432,
       "step": 18588
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.017691415914026e-06,
-      "loss": 0.6232,
+      "loss": 0.7632,
       "step": 18589
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.016670192096585e-06,
-      "loss": 0.7489,
+      "loss": 0.8676,
       "step": 18590
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.015649037415773e-06,
-      "loss": 0.6912,
+      "loss": 0.781,
       "step": 18591
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.0146279518857556e-06,
-      "loss": 0.7446,
+      "loss": 0.8341,
       "step": 18592
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.013606935520708e-06,
-      "loss": 0.7613,
+      "loss": 0.8427,
       "step": 18593
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.012585988334786e-06,
-      "loss": 0.6523,
+      "loss": 0.8255,
       "step": 18594
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.01156511034216e-06,
-      "loss": 0.6011,
+      "loss": 0.7629,
       "step": 18595
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.010544301556985e-06,
-      "loss": 0.7386,
+      "loss": 0.8062,
       "step": 18596
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.009523561993437e-06,
-      "loss": 0.6233,
+      "loss": 0.6786,
       "step": 18597
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.008502891665662e-06,
-      "loss": 0.6909,
+      "loss": 0.7799,
       "step": 18598
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.0074822905878305e-06,
-      "loss": 0.5968,
+      "loss": 0.6889,
       "step": 18599
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.006461758774099e-06,
-      "loss": 0.8162,
+      "loss": 0.9139,
       "step": 18600
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.005441296238625e-06,
-      "loss": 0.8666,
+      "loss": 0.9013,
       "step": 18601
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.004420902995564e-06,
-      "loss": 0.696,
+      "loss": 0.8289,
       "step": 18602
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.003400579059081e-06,
-      "loss": 0.7616,
+      "loss": 0.8523,
       "step": 18603
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.00238032444332e-06,
-      "loss": 0.7395,
+      "loss": 0.8398,
       "step": 18604
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.001360139162444e-06,
-      "loss": 0.6617,
+      "loss": 0.8046,
       "step": 18605
     },
     {
       "epoch": 0.68,
       "learning_rate": 5.000340023230606e-06,
-      "loss": 0.7496,
+      "loss": 0.8575,
       "step": 18606
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.999319976661954e-06,
-      "loss": 0.6754,
+      "loss": 0.7583,
       "step": 18607
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.998299999470643e-06,
-      "loss": 0.7389,
+      "loss": 0.8759,
       "step": 18608
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.997280091670823e-06,
-      "loss": 0.669,
+      "loss": 0.7587,
       "step": 18609
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.996260253276643e-06,
-      "loss": 0.6851,
+      "loss": 0.7355,
       "step": 18610
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.995240484302251e-06,
-      "loss": 0.7115,
+      "loss": 0.868,
       "step": 18611
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.994220784761798e-06,
-      "loss": 0.8039,
+      "loss": 0.8457,
       "step": 18612
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.993201154669432e-06,
-      "loss": 0.7176,
+      "loss": 0.8557,
       "step": 18613
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.992181594039294e-06,
-      "loss": 0.6553,
+      "loss": 0.7877,
       "step": 18614
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.991162102885529e-06,
-      "loss": 0.7398,
+      "loss": 0.8191,
       "step": 18615
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9901426812222886e-06,
-      "loss": 0.59,
+      "loss": 0.6976,
       "step": 18616
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.989123329063704e-06,
-      "loss": 0.6804,
+      "loss": 0.802,
       "step": 18617
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.988104046423928e-06,
-      "loss": 0.7533,
+      "loss": 0.8759,
       "step": 18618
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.987084833317097e-06,
-      "loss": 0.6672,
+      "loss": 0.7937,
       "step": 18619
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.986065689757352e-06,
-      "loss": 0.6841,
+      "loss": 0.8047,
       "step": 18620
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.985046615758827e-06,
-      "loss": 0.6694,
+      "loss": 0.8166,
       "step": 18621
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.984027611335674e-06,
-      "loss": 0.5906,
+      "loss": 0.7385,
       "step": 18622
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.983008676502011e-06,
-      "loss": 0.6711,
+      "loss": 0.804,
       "step": 18623
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.98198981127199e-06,
-      "loss": 0.8083,
+      "loss": 0.8323,
       "step": 18624
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.980971015659741e-06,
-      "loss": 0.6737,
+      "loss": 0.7848,
       "step": 18625
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.979952289679397e-06,
-      "loss": 0.5229,
+      "loss": 0.704,
       "step": 18626
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.978933633345093e-06,
-      "loss": 0.6968,
+      "loss": 0.7877,
       "step": 18627
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.977915046670958e-06,
-      "loss": 0.5784,
+      "loss": 0.6544,
       "step": 18628
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.976896529671133e-06,
-      "loss": 0.733,
+      "loss": 0.886,
       "step": 18629
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.975878082359734e-06,
-      "loss": 0.7608,
+      "loss": 0.8368,
       "step": 18630
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.974859704750903e-06,
-      "loss": 0.8335,
+      "loss": 0.8662,
       "step": 18631
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.973841396858763e-06,
-      "loss": 0.6178,
+      "loss": 0.7021,
       "step": 18632
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.972823158697443e-06,
-      "loss": 0.6222,
+      "loss": 0.7221,
       "step": 18633
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.971804990281065e-06,
-      "loss": 0.6144,
+      "loss": 0.7132,
       "step": 18634
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.970786891623766e-06,
-      "loss": 0.7112,
+      "loss": 0.8029,
       "step": 18635
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.969768862739656e-06,
-      "loss": 0.5748,
+      "loss": 0.7191,
       "step": 18636
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.968750903642869e-06,
-      "loss": 0.6863,
+      "loss": 0.8407,
       "step": 18637
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.967733014347527e-06,
-      "loss": 0.666,
+      "loss": 0.7223,
       "step": 18638
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.966715194867747e-06,
-      "loss": 0.6494,
+      "loss": 0.7486,
       "step": 18639
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.96569744521765e-06,
-      "loss": 0.6894,
+      "loss": 0.8732,
       "step": 18640
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.964679765411362e-06,
-      "loss": 0.7663,
+      "loss": 0.8822,
       "step": 18641
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.963662155462998e-06,
-      "loss": 0.7565,
+      "loss": 0.8393,
       "step": 18642
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.962644615386676e-06,
-      "loss": 0.6387,
+      "loss": 0.8433,
       "step": 18643
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.961627145196512e-06,
-      "loss": 0.7508,
+      "loss": 0.816,
       "step": 18644
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.960609744906624e-06,
-      "loss": 0.8304,
+      "loss": 0.8656,
       "step": 18645
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.959592414531126e-06,
-      "loss": 0.7165,
+      "loss": 0.9156,
       "step": 18646
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.958575154084126e-06,
-      "loss": 0.7207,
+      "loss": 0.8466,
       "step": 18647
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.957557963579753e-06,
-      "loss": 0.6963,
+      "loss": 0.8089,
       "step": 18648
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.956540843032101e-06,
-      "loss": 0.6515,
+      "loss": 0.723,
       "step": 18649
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.955523792455291e-06,
-      "loss": 0.8158,
+      "loss": 0.86,
       "step": 18650
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.954506811863433e-06,
-      "loss": 0.689,
+      "loss": 0.8214,
       "step": 18651
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.953489901270634e-06,
-      "loss": 0.7713,
+      "loss": 0.8477,
       "step": 18652
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.952473060690997e-06,
-      "loss": 0.7607,
+      "loss": 0.8217,
       "step": 18653
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.951456290138645e-06,
-      "loss": 0.8111,
+      "loss": 0.8019,
       "step": 18654
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9504395896276645e-06,
-      "loss": 0.7373,
+      "loss": 0.8524,
       "step": 18655
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.949422959172175e-06,
-      "loss": 0.7092,
+      "loss": 0.841,
       "step": 18656
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.948406398786277e-06,
-      "loss": 0.7759,
+      "loss": 0.8232,
       "step": 18657
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.947389908484071e-06,
-      "loss": 0.7844,
+      "loss": 0.8532,
       "step": 18658
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.946373488279659e-06,
-      "loss": 0.6612,
+      "loss": 0.8864,
       "step": 18659
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.945357138187148e-06,
-      "loss": 0.601,
+      "loss": 0.8321,
       "step": 18660
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.944340858220635e-06,
-      "loss": 0.6723,
+      "loss": 0.7975,
       "step": 18661
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9433246483942195e-06,
-      "loss": 0.6377,
+      "loss": 0.7831,
       "step": 18662
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9423085087219965e-06,
-      "loss": 0.6824,
+      "loss": 0.7907,
       "step": 18663
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9412924392180735e-06,
-      "loss": 0.7206,
+      "loss": 0.8019,
       "step": 18664
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.940276439896538e-06,
-      "loss": 0.8125,
+      "loss": 0.8067,
       "step": 18665
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.939260510771484e-06,
-      "loss": 0.7926,
+      "loss": 0.8256,
       "step": 18666
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9382446518570185e-06,
-      "loss": 0.5855,
+      "loss": 0.7616,
       "step": 18667
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.937228863167218e-06,
-      "loss": 0.8502,
+      "loss": 0.8666,
       "step": 18668
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.936213144716189e-06,
-      "loss": 0.6086,
+      "loss": 0.8116,
       "step": 18669
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.935197496518016e-06,
-      "loss": 0.6544,
+      "loss": 0.802,
       "step": 18670
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.934181918586793e-06,
-      "loss": 0.6999,
+      "loss": 0.7232,
       "step": 18671
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.933166410936605e-06,
-      "loss": 0.7351,
+      "loss": 0.7531,
       "step": 18672
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.932150973581547e-06,
-      "loss": 0.5869,
+      "loss": 0.7029,
       "step": 18673
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.931135606535704e-06,
-      "loss": 0.7174,
+      "loss": 0.7787,
       "step": 18674
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.930120309813163e-06,
-      "loss": 0.6509,
+      "loss": 0.7872,
       "step": 18675
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.92910508342801e-06,
-      "loss": 0.7478,
+      "loss": 0.8277,
       "step": 18676
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9280899273943284e-06,
-      "loss": 0.8407,
+      "loss": 0.9002,
       "step": 18677
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.927074841726201e-06,
-      "loss": 0.7359,
+      "loss": 0.877,
       "step": 18678
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9260598264377135e-06,
-      "loss": 0.7732,
+      "loss": 0.8988,
       "step": 18679
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.925044881542949e-06,
-      "loss": 0.7593,
+      "loss": 0.8711,
       "step": 18680
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.924030007055985e-06,
-      "loss": 0.6664,
+      "loss": 0.7668,
       "step": 18681
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.923015202990901e-06,
-      "loss": 0.6621,
+      "loss": 0.7303,
       "step": 18682
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.922000469361784e-06,
-      "loss": 0.6715,
+      "loss": 0.7711,
       "step": 18683
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.920985806182702e-06,
-      "loss": 0.7912,
+      "loss": 0.8367,
       "step": 18684
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9199712134677316e-06,
-      "loss": 0.7315,
+      "loss": 0.8554,
       "step": 18685
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9189566912309575e-06,
-      "loss": 0.5802,
+      "loss": 0.8088,
       "step": 18686
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.91794223948645e-06,
-      "loss": 0.899,
+      "loss": 0.9703,
       "step": 18687
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9169278582482835e-06,
-      "loss": 0.7793,
+      "loss": 0.8341,
       "step": 18688
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9159135475305305e-06,
-      "loss": 0.894,
+      "loss": 0.9061,
       "step": 18689
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.914899307347264e-06,
-      "loss": 0.5699,
+      "loss": 0.762,
       "step": 18690
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9138851377125515e-06,
-      "loss": 0.8233,
+      "loss": 0.9049,
       "step": 18691
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.91287103864047e-06,
-      "loss": 0.6777,
+      "loss": 0.7358,
       "step": 18692
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.911857010145085e-06,
-      "loss": 0.586,
+      "loss": 0.8013,
       "step": 18693
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9108430522404646e-06,
-      "loss": 0.6518,
+      "loss": 0.7349,
       "step": 18694
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.909829164940673e-06,
-      "loss": 0.6547,
+      "loss": 0.8162,
       "step": 18695
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.908815348259786e-06,
-      "loss": 0.799,
+      "loss": 0.8401,
       "step": 18696
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.907801602211857e-06,
-      "loss": 0.8123,
+      "loss": 0.8753,
       "step": 18697
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9067879268109585e-06,
-      "loss": 0.7548,
+      "loss": 0.8425,
       "step": 18698
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.905774322071152e-06,
-      "loss": 0.6512,
+      "loss": 0.7374,
       "step": 18699
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.904760788006498e-06,
-      "loss": 0.6359,
+      "loss": 0.7639,
       "step": 18700
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.903747324631055e-06,
-      "loss": 0.8835,
+      "loss": 0.9278,
       "step": 18701
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.902733931958896e-06,
-      "loss": 0.7482,
+      "loss": 0.8224,
       "step": 18702
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.9017206100040675e-06,
-      "loss": 0.6143,
+      "loss": 0.7737,
       "step": 18703
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.900707358780627e-06,
-      "loss": 0.6884,
+      "loss": 0.7972,
       "step": 18704
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.899694178302642e-06,
-      "loss": 0.6215,
+      "loss": 0.7586,
       "step": 18705
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.898681068584164e-06,
-      "loss": 0.6738,
+      "loss": 0.8314,
       "step": 18706
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.897668029639248e-06,
-      "loss": 0.6765,
+      "loss": 0.7206,
       "step": 18707
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.896655061481949e-06,
-      "loss": 0.7554,
+      "loss": 0.8031,
       "step": 18708
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.89564216412632e-06,
-      "loss": 0.6754,
+      "loss": 0.8296,
       "step": 18709
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.89462933758641e-06,
-      "loss": 0.7385,
+      "loss": 0.8361,
       "step": 18710
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.893616581876278e-06,
-      "loss": 0.8126,
+      "loss": 0.8708,
       "step": 18711
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.892603897009971e-06,
-      "loss": 0.7659,
+      "loss": 0.8594,
       "step": 18712
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.891591283001539e-06,
-      "loss": 0.7133,
+      "loss": 0.7628,
       "step": 18713
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.890578739865025e-06,
-      "loss": 0.6802,
+      "loss": 0.8058,
       "step": 18714
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8895662676144905e-06,
-      "loss": 0.6696,
+      "loss": 0.9258,
       "step": 18715
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.888553866263965e-06,
-      "loss": 0.6814,
+      "loss": 0.8221,
       "step": 18716
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.887541535827507e-06,
-      "loss": 0.6823,
+      "loss": 0.8273,
       "step": 18717
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8865292763191565e-06,
-      "loss": 0.7229,
+      "loss": 0.7555,
       "step": 18718
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.885517087752957e-06,
-      "loss": 0.7121,
+      "loss": 0.8328,
       "step": 18719
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.884504970142947e-06,
-      "loss": 0.6504,
+      "loss": 0.8188,
       "step": 18720
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8834929235031805e-06,
-      "loss": 0.6656,
+      "loss": 0.7635,
       "step": 18721
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.882480947847687e-06,
-      "loss": 0.7538,
+      "loss": 0.8015,
       "step": 18722
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8814690431905056e-06,
-      "loss": 0.6829,
+      "loss": 0.8994,
       "step": 18723
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8804572095456826e-06,
-      "loss": 0.6313,
+      "loss": 0.7615,
       "step": 18724
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.879445446927251e-06,
-      "loss": 0.7101,
+      "loss": 0.8457,
       "step": 18725
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.87843375534925e-06,
-      "loss": 0.5605,
+      "loss": 0.671,
       "step": 18726
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.87742213482571e-06,
-      "loss": 0.7074,
+      "loss": 0.7853,
       "step": 18727
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.876410585370678e-06,
-      "loss": 0.6617,
+      "loss": 0.8228,
       "step": 18728
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.875399106998172e-06,
-      "loss": 0.638,
+      "loss": 0.6342,
       "step": 18729
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.874387699722236e-06,
-      "loss": 0.5532,
+      "loss": 0.7102,
       "step": 18730
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8733763635568985e-06,
-      "loss": 0.7062,
+      "loss": 0.7811,
       "step": 18731
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.87236509851619e-06,
-      "loss": 0.7188,
+      "loss": 0.8603,
       "step": 18732
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.871353904614136e-06,
-      "loss": 0.7913,
+      "loss": 0.8829,
       "step": 18733
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.87034278186478e-06,
-      "loss": 0.775,
+      "loss": 0.8031,
       "step": 18734
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.86933173028213e-06,
-      "loss": 0.5788,
+      "loss": 0.7404,
       "step": 18735
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8683207498802276e-06,
-      "loss": 0.6778,
+      "loss": 0.8357,
       "step": 18736
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.867309840673093e-06,
-      "loss": 0.8883,
+      "loss": 0.8775,
       "step": 18737
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.866299002674754e-06,
-      "loss": 0.8062,
+      "loss": 0.8533,
       "step": 18738
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.865288235899227e-06,
-      "loss": 0.7702,
+      "loss": 0.8563,
       "step": 18739
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.864277540360545e-06,
-      "loss": 0.7943,
+      "loss": 0.8551,
       "step": 18740
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.863266916072728e-06,
-      "loss": 0.859,
+      "loss": 0.9166,
       "step": 18741
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.862256363049789e-06,
-      "loss": 0.6197,
+      "loss": 0.7222,
       "step": 18742
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.861245881305756e-06,
-      "loss": 0.7506,
+      "loss": 0.8563,
       "step": 18743
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.860235470854645e-06,
-      "loss": 0.7042,
+      "loss": 0.7239,
       "step": 18744
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.859225131710476e-06,
-      "loss": 0.696,
+      "loss": 0.8222,
       "step": 18745
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.858214863887259e-06,
-      "loss": 0.6812,
+      "loss": 0.7924,
       "step": 18746
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8572046673990245e-06,
-      "loss": 0.6195,
+      "loss": 0.7955,
       "step": 18747
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.85619454225977e-06,
-      "loss": 0.8806,
+      "loss": 0.9308,
       "step": 18748
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.855184488483523e-06,
-      "loss": 0.7989,
+      "loss": 0.8481,
       "step": 18749
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8541745060842905e-06,
-      "loss": 0.6723,
+      "loss": 0.7884,
       "step": 18750
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.853164595076086e-06,
-      "loss": 0.6934,
+      "loss": 0.7946,
       "step": 18751
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.852154755472917e-06,
-      "loss": 0.7079,
+      "loss": 0.8015,
       "step": 18752
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.851144987288805e-06,
-      "loss": 0.7256,
+      "loss": 0.8914,
       "step": 18753
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.850135290537742e-06,
-      "loss": 0.765,
+      "loss": 0.8664,
       "step": 18754
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.849125665233751e-06,
-      "loss": 0.6649,
+      "loss": 0.8202,
       "step": 18755
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.848116111390832e-06,
-      "loss": 0.6517,
+      "loss": 0.7139,
       "step": 18756
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.847106629022992e-06,
-      "loss": 0.7271,
+      "loss": 0.774,
       "step": 18757
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.846097218144233e-06,
-      "loss": 0.7337,
+      "loss": 0.7841,
       "step": 18758
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.845087878768566e-06,
-      "loss": 0.6367,
+      "loss": 0.7829,
       "step": 18759
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8440786109099955e-06,
-      "loss": 0.8245,
+      "loss": 0.8881,
       "step": 18760
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.843069414582511e-06,
-      "loss": 0.8565,
+      "loss": 0.9364,
       "step": 18761
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.842060289800126e-06,
-      "loss": 0.7408,
+      "loss": 0.8449,
       "step": 18762
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.841051236576834e-06,
-      "loss": 0.7121,
+      "loss": 0.8037,
       "step": 18763
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.840042254926639e-06,
-      "loss": 0.7502,
+      "loss": 0.8416,
       "step": 18764
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.839033344863533e-06,
-      "loss": 0.8518,
+      "loss": 0.8702,
       "step": 18765
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.838024506401523e-06,
-      "loss": 0.7042,
+      "loss": 0.7319,
       "step": 18766
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.837015739554592e-06,
-      "loss": 0.6167,
+      "loss": 0.7673,
       "step": 18767
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.836007044336747e-06,
-      "loss": 0.687,
+      "loss": 0.8062,
       "step": 18768
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.834998420761976e-06,
-      "loss": 0.5945,
+      "loss": 0.7435,
       "step": 18769
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.833989868844276e-06,
-      "loss": 0.666,
+      "loss": 0.8111,
       "step": 18770
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.832981388597631e-06,
-      "loss": 0.6609,
+      "loss": 0.7531,
       "step": 18771
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.831972980036042e-06,
-      "loss": 0.7543,
+      "loss": 0.817,
       "step": 18772
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.830964643173497e-06,
-      "loss": 0.7672,
+      "loss": 0.8459,
       "step": 18773
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.829956378023982e-06,
-      "loss": 0.69,
+      "loss": 0.73,
       "step": 18774
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.828948184601488e-06,
-      "loss": 0.7018,
+      "loss": 0.7884,
       "step": 18775
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.827940062920001e-06,
-      "loss": 0.767,
+      "loss": 0.8611,
       "step": 18776
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8269320129935036e-06,
-      "loss": 0.7599,
+      "loss": 0.797,
       "step": 18777
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.825924034835988e-06,
-      "loss": 0.5538,
+      "loss": 0.7134,
       "step": 18778
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.82491612846144e-06,
-      "loss": 0.6994,
+      "loss": 0.8389,
       "step": 18779
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8239082938838314e-06,
-      "loss": 0.78,
+      "loss": 0.8708,
       "step": 18780
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.822900531117153e-06,
-      "loss": 0.7533,
+      "loss": 0.8318,
       "step": 18781
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8218928401753864e-06,
-      "loss": 0.6134,
+      "loss": 0.7515,
       "step": 18782
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.820885221072509e-06,
-      "loss": 0.8915,
+      "loss": 0.8896,
       "step": 18783
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.819877673822496e-06,
-      "loss": 0.6555,
+      "loss": 0.828,
       "step": 18784
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.818870198439336e-06,
-      "loss": 0.7458,
+      "loss": 0.8058,
       "step": 18785
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.817862794937e-06,
-      "loss": 0.8135,
+      "loss": 0.8197,
       "step": 18786
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8168554633294665e-06,
-      "loss": 0.791,
+      "loss": 0.8794,
       "step": 18787
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8158482036307085e-06,
-      "loss": 0.7689,
+      "loss": 0.8179,
       "step": 18788
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.814841015854702e-06,
-      "loss": 0.7186,
+      "loss": 0.7867,
       "step": 18789
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.813833900015415e-06,
-      "loss": 0.7278,
+      "loss": 0.7839,
       "step": 18790
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.812826856126829e-06,
-      "loss": 0.6262,
+      "loss": 0.7973,
       "step": 18791
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.811819884202909e-06,
-      "loss": 0.6446,
+      "loss": 0.8116,
       "step": 18792
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.810812984257629e-06,
-      "loss": 0.7719,
+      "loss": 0.8455,
       "step": 18793
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.809806156304952e-06,
-      "loss": 0.8369,
+      "loss": 0.8824,
       "step": 18794
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.808799400358859e-06,
-      "loss": 0.6621,
+      "loss": 0.823,
       "step": 18795
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.807792716433301e-06,
-      "loss": 0.6691,
+      "loss": 0.8137,
       "step": 18796
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.806786104542255e-06,
-      "loss": 0.699,
+      "loss": 0.8064,
       "step": 18797
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.805779564699687e-06,
-      "loss": 0.8657,
+      "loss": 0.8636,
       "step": 18798
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.8047730969195525e-06,
-      "loss": 0.6152,
+      "loss": 0.7211,
       "step": 18799
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.803766701215823e-06,
-      "loss": 0.6262,
+      "loss": 0.7046,
       "step": 18800
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.802760377602458e-06,
-      "loss": 0.8022,
+      "loss": 0.8668,
       "step": 18801
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.801754126093419e-06,
-      "loss": 0.6139,
+      "loss": 0.7598,
       "step": 18802
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.800747946702662e-06,
-      "loss": 0.827,
+      "loss": 0.9416,
       "step": 18803
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.799741839444154e-06,
-      "loss": 0.6715,
+      "loss": 0.7828,
       "step": 18804
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.79873580433185e-06,
-      "loss": 0.8003,
+      "loss": 0.8385,
       "step": 18805
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.797729841379707e-06,
-      "loss": 0.6401,
+      "loss": 0.8,
       "step": 18806
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.796723950601678e-06,
-      "loss": 0.7398,
+      "loss": 0.8134,
       "step": 18807
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.7957181320117295e-06,
-      "loss": 0.6463,
+      "loss": 0.7917,
       "step": 18808
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.7947123856237995e-06,
-      "loss": 0.607,
+      "loss": 0.7365,
       "step": 18809
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.793706711451855e-06,
-      "loss": 0.8118,
+      "loss": 0.8397,
       "step": 18810
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.792701109509842e-06,
-      "loss": 0.6853,
+      "loss": 0.8168,
       "step": 18811
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.791695579811713e-06,
-      "loss": 0.7429,
+      "loss": 0.9071,
       "step": 18812
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.790690122371414e-06,
-      "loss": 0.828,
+      "loss": 0.9056,
       "step": 18813
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.789684737202908e-06,
-      "loss": 0.6303,
+      "loss": 0.7536,
       "step": 18814
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.788679424320124e-06,
-      "loss": 0.749,
+      "loss": 0.823,
       "step": 18815
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.7876741837370245e-06,
-      "loss": 0.7678,
+      "loss": 0.917,
       "step": 18816
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.786669015467549e-06,
-      "loss": 0.6502,
+      "loss": 0.7468,
       "step": 18817
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.785663919525645e-06,
-      "loss": 0.6312,
+      "loss": 0.7922,
       "step": 18818
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.784658895925256e-06,
-      "loss": 0.6372,
+      "loss": 0.6664,
       "step": 18819
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.783653944680324e-06,
-      "loss": 0.7009,
+      "loss": 0.8468,
       "step": 18820
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.782649065804794e-06,
-      "loss": 0.7253,
+      "loss": 0.8343,
       "step": 18821
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.781644259312602e-06,
-      "loss": 0.6709,
+      "loss": 0.819,
       "step": 18822
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.780639525217695e-06,
-      "loss": 0.712,
+      "loss": 0.7728,
       "step": 18823
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.779634863534009e-06,
-      "loss": 0.6422,
+      "loss": 0.7402,
       "step": 18824
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.7786302742754845e-06,
-      "loss": 0.8131,
+      "loss": 0.8465,
       "step": 18825
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.77762575745605e-06,
-      "loss": 0.8839,
+      "loss": 0.9484,
       "step": 18826
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.776621313089659e-06,
-      "loss": 0.5955,
+      "loss": 0.7175,
       "step": 18827
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.775616941190227e-06,
-      "loss": 0.8014,
+      "loss": 0.8208,
       "step": 18828
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.774612641771702e-06,
-      "loss": 0.7808,
+      "loss": 0.874,
       "step": 18829
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.773608414848012e-06,
-      "loss": 0.7572,
+      "loss": 0.8183,
       "step": 18830
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.77260426043309e-06,
-      "loss": 0.7253,
+      "loss": 0.867,
       "step": 18831
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.771600178540864e-06,
-      "loss": 0.733,
+      "loss": 0.8327,
       "step": 18832
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.770596169185275e-06,
-      "loss": 0.6915,
+      "loss": 0.7418,
       "step": 18833
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.769592232380237e-06,
-      "loss": 0.6231,
+      "loss": 0.8605,
       "step": 18834
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.768588368139689e-06,
-      "loss": 0.7055,
+      "loss": 0.7682,
       "step": 18835
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.767584576477555e-06,
-      "loss": 0.8043,
+      "loss": 0.7966,
       "step": 18836
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.766580857407762e-06,
-      "loss": 0.7482,
+      "loss": 0.8327,
       "step": 18837
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.765577210944234e-06,
-      "loss": 0.7585,
+      "loss": 0.8232,
       "step": 18838
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.764573637100892e-06,
-      "loss": 0.6945,
+      "loss": 0.7464,
       "step": 18839
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.76357013589167e-06,
-      "loss": 0.7247,
+      "loss": 0.818,
       "step": 18840
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7625667073304745e-06,
-      "loss": 0.6654,
+      "loss": 0.7631,
       "step": 18841
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.76156335143124e-06,
-      "loss": 0.7359,
+      "loss": 0.8604,
       "step": 18842
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7605600682078816e-06,
-      "loss": 0.6238,
+      "loss": 0.807,
       "step": 18843
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.759556857674317e-06,
-      "loss": 0.6843,
+      "loss": 0.8633,
       "step": 18844
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.758553719844463e-06,
-      "loss": 0.7536,
+      "loss": 0.7925,
       "step": 18845
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.757550654732247e-06,
-      "loss": 0.7557,
+      "loss": 0.8647,
       "step": 18846
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.756547662351571e-06,
-      "loss": 0.7066,
+      "loss": 0.8467,
       "step": 18847
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.755544742716358e-06,
-      "loss": 0.6843,
+      "loss": 0.7934,
       "step": 18848
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7545418958405235e-06,
-      "loss": 0.7248,
+      "loss": 0.8873,
       "step": 18849
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.753539121737976e-06,
-      "loss": 0.7033,
+      "loss": 0.8127,
       "step": 18850
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7525364204226265e-06,
-      "loss": 0.6267,
+      "loss": 0.8099,
       "step": 18851
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.751533791908396e-06,
-      "loss": 0.8018,
+      "loss": 0.7769,
       "step": 18852
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.750531236209181e-06,
-      "loss": 0.643,
+      "loss": 0.7707,
       "step": 18853
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7495287533389e-06,
-      "loss": 0.6881,
+      "loss": 0.6843,
       "step": 18854
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.748526343311458e-06,
-      "loss": 0.7499,
+      "loss": 0.8741,
       "step": 18855
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.747524006140762e-06,
-      "loss": 0.7785,
+      "loss": 0.8924,
       "step": 18856
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.746521741840719e-06,
-      "loss": 0.6978,
+      "loss": 0.8104,
       "step": 18857
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.74551955042523e-06,
-      "loss": 0.774,
+      "loss": 0.8531,
       "step": 18858
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.744517431908208e-06,
-      "loss": 0.6457,
+      "loss": 0.7488,
       "step": 18859
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.743515386303542e-06,
-      "loss": 0.5971,
+      "loss": 0.7479,
       "step": 18860
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.742513413625147e-06,
-      "loss": 0.8185,
+      "loss": 0.8911,
       "step": 18861
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.74151151388692e-06,
-      "loss": 0.6236,
+      "loss": 0.7967,
       "step": 18862
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.740509687102758e-06,
-      "loss": 0.8392,
+      "loss": 0.8397,
       "step": 18863
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.739507933286559e-06,
-      "loss": 0.6608,
+      "loss": 0.7921,
       "step": 18864
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.738506252452231e-06,
-      "loss": 0.7281,
+      "loss": 0.8322,
       "step": 18865
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.737504644613656e-06,
-      "loss": 0.8451,
+      "loss": 0.8765,
       "step": 18866
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.736503109784743e-06,
-      "loss": 0.678,
+      "loss": 0.8715,
       "step": 18867
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.735501647979379e-06,
-      "loss": 0.7306,
+      "loss": 0.8186,
       "step": 18868
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.734500259211462e-06,
-      "loss": 0.7404,
+      "loss": 0.8159,
       "step": 18869
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.733498943494879e-06,
-      "loss": 0.6697,
+      "loss": 0.857,
       "step": 18870
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.732497700843529e-06,
-      "loss": 0.7535,
+      "loss": 0.7885,
       "step": 18871
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7314965312713e-06,
-      "loss": 0.7216,
+      "loss": 0.8366,
       "step": 18872
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.730495434792082e-06,
-      "loss": 0.6919,
+      "loss": 0.806,
       "step": 18873
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.729494411419763e-06,
-      "loss": 0.7012,
+      "loss": 0.7978,
       "step": 18874
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.728493461168232e-06,
-      "loss": 0.764,
+      "loss": 0.8109,
       "step": 18875
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7274925840513745e-06,
-      "loss": 0.7681,
+      "loss": 0.8432,
       "step": 18876
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.7264917800830726e-06,
-      "loss": 0.7374,
+      "loss": 0.7821,
       "step": 18877
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.725491049277222e-06,
-      "loss": 0.5356,
+      "loss": 0.7359,
       "step": 18878
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.724490391647694e-06,
-      "loss": 0.9103,
+      "loss": 0.8522,
       "step": 18879
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.723489807208379e-06,
-      "loss": 0.7118,
+      "loss": 0.8355,
       "step": 18880
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.722489295973156e-06,
-      "loss": 0.6569,
+      "loss": 0.7869,
       "step": 18881
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.721488857955908e-06,
-      "loss": 0.6065,
+      "loss": 0.7683,
       "step": 18882
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.720488493170509e-06,
-      "loss": 0.626,
+      "loss": 0.8223,
       "step": 18883
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.719488201630844e-06,
-      "loss": 0.6751,
+      "loss": 0.8142,
       "step": 18884
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.718487983350788e-06,
-      "loss": 0.6364,
+      "loss": 0.8135,
       "step": 18885
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.717487838344219e-06,
-      "loss": 0.6323,
+      "loss": 0.7292,
       "step": 18886
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.716487766625011e-06,
-      "loss": 0.7587,
+      "loss": 0.8257,
       "step": 18887
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.715487768207039e-06,
-      "loss": 0.6479,
+      "loss": 0.7682,
       "step": 18888
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.714487843104172e-06,
-      "loss": 0.699,
+      "loss": 0.7227,
       "step": 18889
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.713487991330291e-06,
-      "loss": 0.6754,
+      "loss": 0.8035,
       "step": 18890
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.712488212899263e-06,
-      "loss": 0.7306,
+      "loss": 0.7784,
       "step": 18891
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.711488507824961e-06,
-      "loss": 0.6312,
+      "loss": 0.7771,
       "step": 18892
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.710488876121248e-06,
-      "loss": 0.715,
+      "loss": 0.8379,
       "step": 18893
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.709489317802003e-06,
-      "loss": 0.7225,
+      "loss": 0.9033,
       "step": 18894
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.708489832881085e-06,
-      "loss": 0.6536,
+      "loss": 0.8381,
       "step": 18895
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.707490421372359e-06,
-      "loss": 0.7296,
+      "loss": 0.7509,
       "step": 18896
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.706491083289702e-06,
-      "loss": 0.7121,
+      "loss": 0.82,
       "step": 18897
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.705491818646962e-06,
-      "loss": 0.8043,
+      "loss": 0.8719,
       "step": 18898
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.704492627458016e-06,
-      "loss": 0.7074,
+      "loss": 0.851,
       "step": 18899
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.703493509736722e-06,
-      "loss": 0.9095,
+      "loss": 0.9315,
       "step": 18900
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.70249446549694e-06,
-      "loss": 0.7404,
+      "loss": 0.8171,
       "step": 18901
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.701495494752527e-06,
-      "loss": 0.696,
+      "loss": 0.777,
       "step": 18902
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.700496597517351e-06,
-      "loss": 0.7438,
+      "loss": 0.8159,
       "step": 18903
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.699497773805265e-06,
-      "loss": 0.7873,
+      "loss": 0.8808,
       "step": 18904
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.698499023630128e-06,
-      "loss": 0.7886,
+      "loss": 0.8883,
       "step": 18905
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.69750034700579e-06,
-      "loss": 0.7786,
+      "loss": 0.8152,
       "step": 18906
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.696501743946119e-06,
-      "loss": 0.7705,
+      "loss": 0.8307,
       "step": 18907
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.695503214464955e-06,
-      "loss": 0.6971,
+      "loss": 0.7364,
       "step": 18908
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.694504758576161e-06,
-      "loss": 0.7592,
+      "loss": 0.815,
       "step": 18909
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.693506376293587e-06,
-      "loss": 0.771,
+      "loss": 0.8109,
       "step": 18910
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.692508067631082e-06,
-      "loss": 0.6743,
+      "loss": 0.7663,
       "step": 18911
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6915098326024945e-06,
-      "loss": 0.6779,
+      "loss": 0.8289,
       "step": 18912
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.690511671221683e-06,
-      "loss": 0.8183,
+      "loss": 0.8565,
       "step": 18913
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.689513583502485e-06,
-      "loss": 0.7518,
+      "loss": 0.8471,
       "step": 18914
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.688515569458748e-06,
-      "loss": 0.7468,
+      "loss": 0.8144,
       "step": 18915
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.687517629104325e-06,
-      "loss": 0.5326,
+      "loss": 0.7095,
       "step": 18916
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.686519762453058e-06,
-      "loss": 0.8559,
+      "loss": 0.9523,
       "step": 18917
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.685521969518789e-06,
-      "loss": 0.6202,
+      "loss": 0.8315,
       "step": 18918
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.684524250315364e-06,
-      "loss": 0.6572,
+      "loss": 0.7975,
       "step": 18919
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.683526604856623e-06,
-      "loss": 0.8389,
+      "loss": 0.977,
       "step": 18920
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.682529033156403e-06,
-      "loss": 0.6489,
+      "loss": 0.8002,
       "step": 18921
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.681531535228553e-06,
-      "loss": 0.7849,
+      "loss": 0.8489,
       "step": 18922
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6805341110869065e-06,
-      "loss": 0.7351,
+      "loss": 0.8207,
       "step": 18923
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.679536760745301e-06,
-      "loss": 0.6333,
+      "loss": 0.7747,
       "step": 18924
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.678539484217571e-06,
-      "loss": 0.7047,
+      "loss": 0.8425,
       "step": 18925
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6775422815175635e-06,
-      "loss": 0.6497,
+      "loss": 0.7786,
       "step": 18926
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.676545152659098e-06,
-      "loss": 0.7052,
+      "loss": 0.8206,
       "step": 18927
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6755480976560175e-06,
-      "loss": 0.656,
+      "loss": 0.7664,
       "step": 18928
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6745511165221545e-06,
-      "loss": 0.6523,
+      "loss": 0.7719,
       "step": 18929
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.673554209271338e-06,
-      "loss": 0.5951,
+      "loss": 0.6881,
       "step": 18930
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.672557375917396e-06,
-      "loss": 0.6958,
+      "loss": 0.8205,
       "step": 18931
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.67156061647417e-06,
-      "loss": 0.643,
+      "loss": 0.7987,
       "step": 18932
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.670563930955475e-06,
-      "loss": 0.8607,
+      "loss": 0.913,
       "step": 18933
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.669567319375142e-06,
-      "loss": 0.6712,
+      "loss": 0.7934,
       "step": 18934
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.668570781747003e-06,
-      "loss": 0.684,
+      "loss": 0.7394,
       "step": 18935
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.667574318084879e-06,
-      "loss": 0.7859,
+      "loss": 0.8474,
       "step": 18936
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.666577928402597e-06,
-      "loss": 0.7642,
+      "loss": 0.8313,
       "step": 18937
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.665581612713974e-06,
-      "loss": 0.7715,
+      "loss": 0.8559,
       "step": 18938
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.664585371032846e-06,
-      "loss": 0.6847,
+      "loss": 0.8668,
       "step": 18939
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.663589203373018e-06,
-      "loss": 0.8425,
+      "loss": 0.8552,
       "step": 18940
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.662593109748323e-06,
-      "loss": 0.8098,
+      "loss": 0.8678,
       "step": 18941
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.661597090172575e-06,
-      "loss": 0.7919,
+      "loss": 0.9053,
       "step": 18942
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.660601144659594e-06,
-      "loss": 0.7674,
+      "loss": 0.8789,
       "step": 18943
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.659605273223191e-06,
-      "loss": 0.6538,
+      "loss": 0.7372,
       "step": 18944
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.658609475877197e-06,
-      "loss": 0.7242,
+      "loss": 0.8407,
       "step": 18945
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.657613752635409e-06,
-      "loss": 0.6554,
+      "loss": 0.7492,
       "step": 18946
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.656618103511655e-06,
-      "loss": 0.6924,
+      "loss": 0.7754,
       "step": 18947
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.655622528519744e-06,
-      "loss": 0.8404,
+      "loss": 0.8692,
       "step": 18948
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.654627027673489e-06,
-      "loss": 0.5601,
+      "loss": 0.715,
       "step": 18949
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6536316009866935e-06,
-      "loss": 0.7658,
+      "loss": 0.8605,
       "step": 18950
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.652636248473179e-06,
-      "loss": 0.6768,
+      "loss": 0.7804,
       "step": 18951
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6516409701467525e-06,
-      "loss": 0.7594,
+      "loss": 0.8506,
       "step": 18952
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6506457660212124e-06,
-      "loss": 0.6057,
+      "loss": 0.7137,
       "step": 18953
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.649650636110377e-06,
-      "loss": 0.7746,
+      "loss": 0.8618,
       "step": 18954
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.648655580428047e-06,
-      "loss": 0.8296,
+      "loss": 0.8539,
       "step": 18955
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.647660598988028e-06,
-      "loss": 0.7336,
+      "loss": 0.7792,
       "step": 18956
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.646665691804121e-06,
-      "loss": 0.8745,
+      "loss": 0.913,
       "step": 18957
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.64567085889014e-06,
-      "loss": 0.624,
+      "loss": 0.8421,
       "step": 18958
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6446761002598704e-06,
-      "loss": 0.7534,
+      "loss": 0.8196,
       "step": 18959
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.643681415927127e-06,
-      "loss": 0.6281,
+      "loss": 0.8599,
       "step": 18960
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.642686805905703e-06,
-      "loss": 0.7241,
+      "loss": 0.8035,
       "step": 18961
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.641692270209399e-06,
-      "loss": 0.6806,
+      "loss": 0.7899,
       "step": 18962
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.640697808852008e-06,
-      "loss": 0.6988,
+      "loss": 0.8196,
       "step": 18963
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.63970342184734e-06,
-      "loss": 0.6039,
+      "loss": 0.6086,
       "step": 18964
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.638709109209173e-06,
-      "loss": 0.6846,
+      "loss": 0.7951,
       "step": 18965
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.637714870951314e-06,
-      "loss": 0.7907,
+      "loss": 0.8565,
       "step": 18966
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.636720707087554e-06,
-      "loss": 0.8042,
+      "loss": 0.8858,
       "step": 18967
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.635726617631685e-06,
-      "loss": 0.6432,
+      "loss": 0.8731,
       "step": 18968
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.634732602597494e-06,
-      "loss": 0.7142,
+      "loss": 0.8663,
       "step": 18969
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.633738661998779e-06,
-      "loss": 0.6636,
+      "loss": 0.8005,
       "step": 18970
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.63274479584933e-06,
-      "loss": 0.7179,
+      "loss": 0.787,
       "step": 18971
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.631751004162925e-06,
-      "loss": 0.7326,
+      "loss": 0.8496,
       "step": 18972
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.630757286953363e-06,
-      "loss": 0.8449,
+      "loss": 0.9147,
       "step": 18973
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.629763644234424e-06,
-      "loss": 0.7265,
+      "loss": 0.8154,
       "step": 18974
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.628770076019895e-06,
-      "loss": 0.6923,
+      "loss": 0.8072,
       "step": 18975
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.627776582323558e-06,
-      "loss": 0.5957,
+      "loss": 0.7616,
       "step": 18976
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.626783163159205e-06,
-      "loss": 0.6512,
+      "loss": 0.8191,
       "step": 18977
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.625789818540604e-06,
-      "loss": 0.7187,
+      "loss": 0.8307,
       "step": 18978
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.624796548481549e-06,
-      "loss": 0.6164,
+      "loss": 0.7217,
       "step": 18979
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.623803352995816e-06,
-      "loss": 0.7604,
+      "loss": 0.8092,
       "step": 18980
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.622810232097183e-06,
-      "loss": 0.7278,
+      "loss": 0.7914,
       "step": 18981
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.621817185799426e-06,
-      "loss": 0.6603,
+      "loss": 0.7921,
       "step": 18982
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.620824214116327e-06,
-      "loss": 0.6903,
+      "loss": 0.7825,
       "step": 18983
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.619831317061662e-06,
-      "loss": 0.6814,
+      "loss": 0.7532,
       "step": 18984
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6188384946492014e-06,
-      "loss": 0.7348,
+      "loss": 0.8714,
       "step": 18985
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.617845746892724e-06,
-      "loss": 0.7208,
+      "loss": 0.7795,
       "step": 18986
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.616853073805999e-06,
-      "loss": 0.6091,
+      "loss": 0.7469,
       "step": 18987
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.615860475402798e-06,
-      "loss": 0.7467,
+      "loss": 0.8708,
       "step": 18988
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.614867951696896e-06,
-      "loss": 0.7764,
+      "loss": 0.8283,
       "step": 18989
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.613875502702065e-06,
-      "loss": 0.7276,
+      "loss": 0.8927,
       "step": 18990
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6128831284320615e-06,
-      "loss": 0.6697,
+      "loss": 0.7527,
       "step": 18991
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.611890828900666e-06,
-      "loss": 0.7439,
+      "loss": 0.857,
       "step": 18992
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.61089860412164e-06,
-      "loss": 0.7298,
+      "loss": 0.8355,
       "step": 18993
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.60990645410875e-06,
-      "loss": 0.7638,
+      "loss": 0.8972,
       "step": 18994
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.608914378875756e-06,
-      "loss": 0.6463,
+      "loss": 0.7324,
       "step": 18995
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.60792237843643e-06,
-      "loss": 0.7589,
+      "loss": 0.887,
       "step": 18996
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.606930452804531e-06,
-      "loss": 0.6684,
+      "loss": 0.7404,
       "step": 18997
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.60593860199382e-06,
-      "loss": 0.6524,
+      "loss": 0.7845,
       "step": 18998
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.604946826018057e-06,
-      "loss": 0.7189,
+      "loss": 0.7943,
       "step": 18999
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.603955124891004e-06,
-      "loss": 0.7863,
+      "loss": 0.8603,
       "step": 19000
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.602963498626412e-06,
-      "loss": 0.6584,
+      "loss": 0.8015,
       "step": 19001
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.6019719472380494e-06,
-      "loss": 0.8054,
+      "loss": 0.7972,
       "step": 19002
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.600980470739667e-06,
-      "loss": 0.7405,
+      "loss": 0.7736,
       "step": 19003
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.599989069145019e-06,
-      "loss": 0.7777,
+      "loss": 0.865,
       "step": 19004
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.598997742467859e-06,
-      "loss": 0.7361,
+      "loss": 0.8061,
       "step": 19005
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.598006490721948e-06,
-      "loss": 0.6374,
+      "loss": 0.8353,
       "step": 19006
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.597015313921026e-06,
-      "loss": 0.6725,
+      "loss": 0.7617,
       "step": 19007
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.596024212078854e-06,
-      "loss": 0.845,
+      "loss": 0.8904,
       "step": 19008
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.595033185209184e-06,
-      "loss": 0.7824,
+      "loss": 0.8166,
       "step": 19009
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.594042233325752e-06,
-      "loss": 0.719,
+      "loss": 0.8059,
       "step": 19010
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.593051356442317e-06,
-      "loss": 0.6625,
+      "loss": 0.7134,
       "step": 19011
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5920605545726225e-06,
-      "loss": 0.593,
+      "loss": 0.7961,
       "step": 19012
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.591069827730416e-06,
-      "loss": 0.7864,
+      "loss": 0.8146,
       "step": 19013
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.590079175929437e-06,
-      "loss": 0.6863,
+      "loss": 0.7654,
       "step": 19014
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.589088599183437e-06,
-      "loss": 0.6287,
+      "loss": 0.7495,
       "step": 19015
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.588098097506156e-06,
-      "loss": 0.7288,
+      "loss": 0.7619,
       "step": 19016
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.587107670911336e-06,
-      "loss": 0.6268,
+      "loss": 0.762,
       "step": 19017
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5861173194127175e-06,
-      "loss": 0.7188,
+      "loss": 0.8154,
       "step": 19018
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.585127043024039e-06,
-      "loss": 0.7062,
+      "loss": 0.8386,
       "step": 19019
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5841368417590355e-06,
-      "loss": 0.6783,
+      "loss": 0.7499,
       "step": 19020
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.583146715631455e-06,
-      "loss": 0.6653,
+      "loss": 0.7457,
       "step": 19021
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.582156664655028e-06,
-      "loss": 0.7647,
+      "loss": 0.6954,
       "step": 19022
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5811666888434895e-06,
-      "loss": 0.7158,
+      "loss": 0.7268,
       "step": 19023
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5801767882105706e-06,
-      "loss": 0.838,
+      "loss": 0.9634,
       "step": 19024
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.579186962770018e-06,
-      "loss": 0.7884,
+      "loss": 0.841,
       "step": 19025
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.578197212535547e-06,
-      "loss": 0.7104,
+      "loss": 0.7881,
       "step": 19026
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5772075375209e-06,
-      "loss": 0.709,
+      "loss": 0.8044,
       "step": 19027
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.576217937739805e-06,
-      "loss": 0.7966,
+      "loss": 0.8599,
       "step": 19028
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.575228413205992e-06,
-      "loss": 0.6562,
+      "loss": 0.7661,
       "step": 19029
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.574238963933187e-06,
-      "loss": 0.7828,
+      "loss": 0.7899,
       "step": 19030
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.573249589935118e-06,
-      "loss": 0.7117,
+      "loss": 0.8141,
       "step": 19031
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.572260291225512e-06,
-      "loss": 0.7791,
+      "loss": 0.8034,
       "step": 19032
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.57127106781809e-06,
-      "loss": 0.8171,
+      "loss": 0.856,
       "step": 19033
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.570281919726583e-06,
-      "loss": 0.848,
+      "loss": 0.8703,
       "step": 19034
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5692928469647105e-06,
-      "loss": 0.5263,
+      "loss": 0.7218,
       "step": 19035
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.568303849546194e-06,
-      "loss": 0.5713,
+      "loss": 0.7254,
       "step": 19036
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.567314927484752e-06,
-      "loss": 0.7602,
+      "loss": 0.811,
       "step": 19037
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.566326080794115e-06,
-      "loss": 0.6791,
+      "loss": 0.8562,
       "step": 19038
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.565337309487986e-06,
-      "loss": 0.753,
+      "loss": 0.8182,
       "step": 19039
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.564348613580095e-06,
-      "loss": 0.9133,
+      "loss": 0.9037,
       "step": 19040
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5633599930841555e-06,
-      "loss": 0.7869,
+      "loss": 0.8908,
       "step": 19041
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5623714480138815e-06,
-      "loss": 0.6983,
+      "loss": 0.7516,
       "step": 19042
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.561382978382986e-06,
-      "loss": 0.6231,
+      "loss": 0.7801,
       "step": 19043
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.560394584205192e-06,
-      "loss": 0.7567,
+      "loss": 0.8861,
       "step": 19044
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.559406265494198e-06,
-      "loss": 0.785,
+      "loss": 0.8277,
       "step": 19045
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.558418022263726e-06,
-      "loss": 0.6498,
+      "loss": 0.7596,
       "step": 19046
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5574298545274845e-06,
-      "loss": 0.7921,
+      "loss": 0.8697,
       "step": 19047
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.556441762299182e-06,
-      "loss": 0.5842,
+      "loss": 0.7929,
       "step": 19048
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.555453745592525e-06,
-      "loss": 0.7531,
+      "loss": 0.8665,
       "step": 19049
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5544658044212195e-06,
-      "loss": 0.671,
+      "loss": 0.7714,
       "step": 19050
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.553477938798983e-06,
-      "loss": 0.7444,
+      "loss": 0.8289,
       "step": 19051
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.552490148739505e-06,
-      "loss": 0.5787,
+      "loss": 0.7925,
       "step": 19052
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.551502434256501e-06,
-      "loss": 0.6656,
+      "loss": 0.7362,
       "step": 19053
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.550514795363669e-06,
-      "loss": 0.7771,
+      "loss": 0.8295,
       "step": 19054
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.549527232074713e-06,
-      "loss": 0.8017,
+      "loss": 0.9174,
       "step": 19055
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.54853974440333e-06,
-      "loss": 0.706,
+      "loss": 0.777,
       "step": 19056
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.547552332363231e-06,
-      "loss": 0.7011,
+      "loss": 0.8325,
       "step": 19057
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.546564995968099e-06,
-      "loss": 0.7469,
+      "loss": 0.7701,
       "step": 19058
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.545577735231645e-06,
-      "loss": 0.6703,
+      "loss": 0.7543,
       "step": 19059
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.544590550167561e-06,
-      "loss": 0.752,
+      "loss": 0.8186,
       "step": 19060
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.543603440789542e-06,
-      "loss": 0.7847,
+      "loss": 0.8607,
       "step": 19061
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.542616407111281e-06,
-      "loss": 0.7361,
+      "loss": 0.8014,
       "step": 19062
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.541629449146482e-06,
-      "loss": 0.7533,
+      "loss": 0.8526,
       "step": 19063
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.540642566908821e-06,
-      "loss": 0.5048,
+      "loss": 0.692,
       "step": 19064
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.539655760412004e-06,
-      "loss": 0.73,
+      "loss": 0.8365,
       "step": 19065
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.538669029669715e-06,
-      "loss": 0.8097,
+      "loss": 0.8819,
       "step": 19066
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.537682374695644e-06,
-      "loss": 0.5815,
+      "loss": 0.7803,
       "step": 19067
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.536695795503482e-06,
-      "loss": 0.6199,
+      "loss": 0.7441,
       "step": 19068
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.535709292106909e-06,
-      "loss": 0.6327,
+      "loss": 0.707,
       "step": 19069
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.534722864519625e-06,
-      "loss": 0.6876,
+      "loss": 0.8318,
       "step": 19070
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5337365127553e-06,
-      "loss": 0.7376,
+      "loss": 0.8394,
       "step": 19071
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.532750236827628e-06,
-      "loss": 0.8029,
+      "loss": 0.8864,
       "step": 19072
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.53176403675029e-06,
-      "loss": 0.6781,
+      "loss": 0.7541,
       "step": 19073
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5307779125369675e-06,
-      "loss": 0.8348,
+      "loss": 0.8772,
       "step": 19074
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.529791864201338e-06,
-      "loss": 0.6859,
+      "loss": 0.8143,
       "step": 19075
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5288058917570925e-06,
-      "loss": 0.7003,
+      "loss": 0.7458,
       "step": 19076
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.527819995217895e-06,
-      "loss": 0.534,
+      "loss": 0.7242,
       "step": 19077
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.526834174597435e-06,
-      "loss": 0.7459,
+      "loss": 0.8507,
       "step": 19078
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.525848429909385e-06,
-      "loss": 0.7612,
+      "loss": 0.8708,
       "step": 19079
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.524862761167422e-06,
-      "loss": 0.7769,
+      "loss": 0.8815,
       "step": 19080
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.523877168385216e-06,
-      "loss": 0.6082,
+      "loss": 0.7168,
       "step": 19081
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.522891651576449e-06,
-      "loss": 0.7701,
+      "loss": 0.8612,
       "step": 19082
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.521906210754789e-06,
-      "loss": 0.5592,
+      "loss": 0.6895,
       "step": 19083
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5209208459339075e-06,
-      "loss": 0.7164,
+      "loss": 0.7676,
       "step": 19084
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.519935557127475e-06,
-      "loss": 0.664,
+      "loss": 0.732,
       "step": 19085
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.518950344349162e-06,
-      "loss": 0.847,
+      "loss": 0.8415,
       "step": 19086
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.517965207612637e-06,
-      "loss": 0.6237,
+      "loss": 0.775,
       "step": 19087
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.5169801469315625e-06,
-      "loss": 0.7877,
+      "loss": 0.8859,
       "step": 19088
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.515995162319615e-06,
-      "loss": 0.9139,
+      "loss": 0.9586,
       "step": 19089
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.515010253790448e-06,
-      "loss": 0.7169,
+      "loss": 0.9301,
       "step": 19090
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.514025421357735e-06,
-      "loss": 0.8374,
+      "loss": 0.9278,
       "step": 19091
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.513040665035137e-06,
-      "loss": 0.7203,
+      "loss": 0.846,
       "step": 19092
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.512055984836313e-06,
-      "loss": 0.7597,
+      "loss": 0.8195,
       "step": 19093
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.511071380774923e-06,
-      "loss": 0.6049,
+      "loss": 0.7607,
       "step": 19094
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.510086852864632e-06,
-      "loss": 0.6947,
+      "loss": 0.7667,
       "step": 19095
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.509102401119099e-06,
-      "loss": 0.7237,
+      "loss": 0.8189,
       "step": 19096
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.508118025551979e-06,
-      "loss": 0.6585,
+      "loss": 0.6954,
       "step": 19097
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.507133726176928e-06,
-      "loss": 0.7745,
+      "loss": 0.8441,
       "step": 19098
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.506149503007605e-06,
-      "loss": 0.7058,
+      "loss": 0.7813,
       "step": 19099
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.505165356057658e-06,
-      "loss": 0.5984,
+      "loss": 0.7236,
       "step": 19100
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.504181285340749e-06,
-      "loss": 0.5922,
+      "loss": 0.6676,
       "step": 19101
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.503197290870529e-06,
-      "loss": 0.7063,
+      "loss": 0.8104,
       "step": 19102
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.502213372660646e-06,
-      "loss": 0.7687,
+      "loss": 0.8496,
       "step": 19103
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.501229530724749e-06,
-      "loss": 0.8097,
+      "loss": 0.8784,
       "step": 19104
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.500245765076498e-06,
-      "loss": 0.7163,
+      "loss": 0.7437,
       "step": 19105
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.499262075729529e-06,
-      "loss": 0.8316,
+      "loss": 0.8696,
       "step": 19106
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.498278462697492e-06,
-      "loss": 0.7743,
+      "loss": 0.8459,
       "step": 19107
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.4972949259940425e-06,
-      "loss": 0.7541,
+      "loss": 0.8851,
       "step": 19108
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.496311465632811e-06,
-      "loss": 0.7377,
+      "loss": 0.7813,
       "step": 19109
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.495328081627453e-06,
-      "loss": 0.74,
+      "loss": 0.8145,
       "step": 19110
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.494344773991607e-06,
-      "loss": 0.6896,
+      "loss": 0.8087,
       "step": 19111
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4933615427389165e-06,
-      "loss": 0.7962,
+      "loss": 0.8181,
       "step": 19112
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4923783878830175e-06,
-      "loss": 0.7794,
+      "loss": 0.8721,
       "step": 19113
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.491395309437558e-06,
-      "loss": 0.6754,
+      "loss": 0.8152,
       "step": 19114
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.490412307416173e-06,
-      "loss": 0.7427,
+      "loss": 0.8595,
       "step": 19115
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.489429381832501e-06,
-      "loss": 0.7992,
+      "loss": 0.8472,
       "step": 19116
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.488446532700173e-06,
-      "loss": 0.7333,
+      "loss": 0.8142,
       "step": 19117
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.487463760032837e-06,
-      "loss": 0.7695,
+      "loss": 0.7754,
       "step": 19118
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.486481063844112e-06,
-      "loss": 0.7138,
+      "loss": 0.8772,
       "step": 19119
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.485498444147645e-06,
-      "loss": 0.7957,
+      "loss": 0.876,
       "step": 19120
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.484515900957063e-06,
-      "loss": 0.7209,
+      "loss": 0.8049,
       "step": 19121
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.483533434285996e-06,
-      "loss": 0.5136,
+      "loss": 0.6649,
       "step": 19122
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.482551044148072e-06,
-      "loss": 0.6587,
+      "loss": 0.7945,
       "step": 19123
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.481568730556932e-06,
-      "loss": 0.7941,
+      "loss": 0.8181,
       "step": 19124
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.480586493526194e-06,
-      "loss": 0.5976,
+      "loss": 0.7402,
       "step": 19125
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.479604333069482e-06,
-      "loss": 0.6627,
+      "loss": 0.8704,
       "step": 19126
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.47862224920043e-06,
-      "loss": 0.7062,
+      "loss": 0.8177,
       "step": 19127
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.477640241932663e-06,
-      "loss": 0.6272,
+      "loss": 0.7979,
       "step": 19128
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.476658311279801e-06,
-      "loss": 0.8222,
+      "loss": 0.8937,
       "step": 19129
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.475676457255469e-06,
-      "loss": 0.8194,
+      "loss": 0.8565,
       "step": 19130
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.474694679873287e-06,
-      "loss": 0.7245,
+      "loss": 0.7772,
       "step": 19131
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4737129791468745e-06,
-      "loss": 0.6574,
+      "loss": 0.783,
       "step": 19132
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.472731355089857e-06,
-      "loss": 0.8125,
+      "loss": 0.8792,
       "step": 19133
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.471749807715849e-06,
-      "loss": 0.6958,
+      "loss": 0.8799,
       "step": 19134
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.47076833703847e-06,
-      "loss": 0.6604,
+      "loss": 0.7908,
       "step": 19135
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.469786943071332e-06,
-      "loss": 0.7139,
+      "loss": 0.7412,
       "step": 19136
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.46880562582806e-06,
-      "loss": 0.6928,
+      "loss": 0.8076,
       "step": 19137
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.467824385322256e-06,
-      "loss": 0.6596,
+      "loss": 0.7253,
       "step": 19138
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.466843221567542e-06,
-      "loss": 0.752,
+      "loss": 0.8328,
       "step": 19139
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.465862134577527e-06,
-      "loss": 0.7741,
+      "loss": 0.8071,
       "step": 19140
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.464881124365825e-06,
-      "loss": 0.7702,
+      "loss": 0.8657,
       "step": 19141
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.46390019094604e-06,
-      "loss": 0.6575,
+      "loss": 0.7742,
       "step": 19142
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.462919334331792e-06,
-      "loss": 0.6234,
+      "loss": 0.69,
       "step": 19143
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.461938554536679e-06,
-      "loss": 0.8439,
+      "loss": 0.9044,
       "step": 19144
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.460957851574307e-06,
-      "loss": 0.7697,
+      "loss": 0.8491,
       "step": 19145
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.459977225458289e-06,
-      "loss": 0.5346,
+      "loss": 0.7152,
       "step": 19146
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.458996676202228e-06,
-      "loss": 0.6975,
+      "loss": 0.8009,
       "step": 19147
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.458016203819726e-06,
-      "loss": 0.7611,
+      "loss": 0.8923,
       "step": 19148
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.457035808324382e-06,
-      "loss": 0.7282,
+      "loss": 0.8089,
       "step": 19149
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.456055489729809e-06,
-      "loss": 0.6377,
+      "loss": 0.7876,
       "step": 19150
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.455075248049594e-06,
-      "loss": 0.6816,
+      "loss": 0.8627,
       "step": 19151
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.454095083297345e-06,
-      "loss": 0.7985,
+      "loss": 0.831,
       "step": 19152
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4531149954866594e-06,
-      "loss": 0.6764,
+      "loss": 0.8405,
       "step": 19153
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.452134984631133e-06,
-      "loss": 0.7773,
+      "loss": 0.8396,
       "step": 19154
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4511550507443565e-06,
-      "loss": 0.5515,
+      "loss": 0.7049,
       "step": 19155
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.450175193839941e-06,
-      "loss": 0.8345,
+      "loss": 0.8831,
       "step": 19156
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.449195413931462e-06,
-      "loss": 0.6879,
+      "loss": 0.7895,
       "step": 19157
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.448215711032525e-06,
-      "loss": 0.7739,
+      "loss": 0.9393,
       "step": 19158
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.447236085156718e-06,
-      "loss": 0.7068,
+      "loss": 0.8568,
       "step": 19159
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.446256536317631e-06,
-      "loss": 0.7188,
+      "loss": 0.7588,
       "step": 19160
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.445277064528852e-06,
-      "loss": 0.6991,
+      "loss": 0.7706,
       "step": 19161
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.444297669803981e-06,
-      "loss": 0.6596,
+      "loss": 0.7032,
       "step": 19162
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.443318352156593e-06,
-      "loss": 0.7286,
+      "loss": 0.7768,
       "step": 19163
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.442339111600274e-06,
-      "loss": 0.7083,
+      "loss": 0.782,
       "step": 19164
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.441359948148619e-06,
-      "loss": 0.6679,
+      "loss": 0.7897,
       "step": 19165
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.440380861815209e-06,
-      "loss": 0.6554,
+      "loss": 0.7642,
       "step": 19166
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.439401852613625e-06,
-      "loss": 0.6929,
+      "loss": 0.7778,
       "step": 19167
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.438422920557448e-06,
-      "loss": 0.8777,
+      "loss": 0.9107,
       "step": 19168
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.437444065660269e-06,
-      "loss": 0.6721,
+      "loss": 0.7564,
       "step": 19169
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.436465287935655e-06,
-      "loss": 0.7568,
+      "loss": 0.808,
       "step": 19170
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.435486587397194e-06,
-      "loss": 0.7143,
+      "loss": 0.8355,
       "step": 19171
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.434507964058462e-06,
-      "loss": 0.8325,
+      "loss": 0.8843,
       "step": 19172
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.433529417933036e-06,
-      "loss": 0.9021,
+      "loss": 0.9228,
       "step": 19173
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.432550949034487e-06,
-      "loss": 0.7002,
+      "loss": 0.7738,
       "step": 19174
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.431572557376402e-06,
-      "loss": 0.6584,
+      "loss": 0.7526,
       "step": 19175
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.430594242972342e-06,
-      "loss": 0.691,
+      "loss": 0.7764,
       "step": 19176
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.429616005835886e-06,
-      "loss": 0.7097,
+      "loss": 0.7834,
       "step": 19177
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.428637845980604e-06,
-      "loss": 0.6338,
+      "loss": 0.7739,
       "step": 19178
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.427659763420069e-06,
-      "loss": 0.7134,
+      "loss": 0.8322,
       "step": 19179
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.426681758167844e-06,
-      "loss": 0.6941,
+      "loss": 0.8466,
       "step": 19180
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4257038302375076e-06,
-      "loss": 0.7059,
+      "loss": 0.8186,
       "step": 19181
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.424725979642623e-06,
-      "loss": 0.7323,
+      "loss": 0.802,
       "step": 19182
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.42374820639675e-06,
-      "loss": 0.8716,
+      "loss": 0.9069,
       "step": 19183
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.422770510513461e-06,
-      "loss": 0.7024,
+      "loss": 0.7408,
       "step": 19184
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.421792892006318e-06,
-      "loss": 0.7685,
+      "loss": 0.8634,
       "step": 19185
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.420815350888885e-06,
-      "loss": 0.6791,
+      "loss": 0.8052,
       "step": 19186
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4198378871747195e-06,
-      "loss": 0.7088,
+      "loss": 0.8296,
       "step": 19187
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.418860500877394e-06,
-      "loss": 0.6905,
+      "loss": 0.7647,
       "step": 19188
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.417883192010451e-06,
-      "loss": 0.6803,
+      "loss": 0.7593,
       "step": 19189
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.416905960587464e-06,
-      "loss": 0.6498,
+      "loss": 0.8128,
       "step": 19190
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.415928806621985e-06,
-      "loss": 0.6752,
+      "loss": 0.815,
       "step": 19191
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.414951730127571e-06,
-      "loss": 0.7345,
+      "loss": 0.7678,
       "step": 19192
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.413974731117775e-06,
-      "loss": 0.5464,
+      "loss": 0.6931,
       "step": 19193
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.412997809606156e-06,
-      "loss": 0.6812,
+      "loss": 0.8534,
       "step": 19194
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.412020965606266e-06,
-      "loss": 0.7456,
+      "loss": 0.9122,
       "step": 19195
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4110441991316574e-06,
-      "loss": 0.7315,
+      "loss": 0.8442,
       "step": 19196
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.410067510195881e-06,
-      "loss": 0.6917,
+      "loss": 0.8531,
       "step": 19197
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4090908988124855e-06,
-      "loss": 0.7213,
+      "loss": 0.8029,
       "step": 19198
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.408114364995018e-06,
-      "loss": 0.7562,
+      "loss": 0.8312,
       "step": 19199
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.407137908757035e-06,
-      "loss": 0.6842,
+      "loss": 0.7962,
       "step": 19200
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4061615301120805e-06,
-      "loss": 0.7772,
+      "loss": 0.8033,
       "step": 19201
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.405185229073692e-06,
-      "loss": 0.7832,
+      "loss": 0.8777,
       "step": 19202
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4042090056554236e-06,
-      "loss": 0.7573,
+      "loss": 0.8985,
       "step": 19203
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.4032328598708154e-06,
-      "loss": 0.6842,
+      "loss": 0.7985,
       "step": 19204
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.40225679173341e-06,
-      "loss": 0.5614,
+      "loss": 0.7323,
       "step": 19205
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.401280801256747e-06,
-      "loss": 0.6579,
+      "loss": 0.6763,
       "step": 19206
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.400304888454376e-06,
-      "loss": 0.6673,
+      "loss": 0.8269,
       "step": 19207
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3993290533398235e-06,
-      "loss": 0.7126,
+      "loss": 0.8587,
       "step": 19208
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.398353295926636e-06,
-      "loss": 0.7514,
+      "loss": 0.8597,
       "step": 19209
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.39737761622835e-06,
-      "loss": 0.6208,
+      "loss": 0.7368,
       "step": 19210
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.396402014258501e-06,
-      "loss": 0.7463,
+      "loss": 0.8561,
       "step": 19211
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.395426490030619e-06,
-      "loss": 0.6681,
+      "loss": 0.7859,
       "step": 19212
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.394451043558245e-06,
-      "loss": 0.6831,
+      "loss": 0.7458,
       "step": 19213
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.393475674854911e-06,
-      "loss": 0.7103,
+      "loss": 0.7674,
       "step": 19214
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.392500383934149e-06,
-      "loss": 0.6592,
+      "loss": 0.8349,
       "step": 19215
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.391525170809482e-06,
-      "loss": 0.7732,
+      "loss": 0.7999,
       "step": 19216
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.390550035494454e-06,
-      "loss": 0.7442,
+      "loss": 0.7918,
       "step": 19217
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3895749780025775e-06,
-      "loss": 0.6429,
+      "loss": 0.701,
       "step": 19218
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.388599998347393e-06,
-      "loss": 0.7683,
+      "loss": 0.8354,
       "step": 19219
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.387625096542425e-06,
-      "loss": 0.7893,
+      "loss": 0.8777,
       "step": 19220
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.386650272601188e-06,
-      "loss": 0.6235,
+      "loss": 0.7121,
       "step": 19221
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3856755265372185e-06,
-      "loss": 0.5558,
+      "loss": 0.7477,
       "step": 19222
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3847008583640345e-06,
-      "loss": 0.6623,
+      "loss": 0.7375,
       "step": 19223
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.383726268095161e-06,
-      "loss": 0.7222,
+      "loss": 0.7451,
       "step": 19224
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.382751755744113e-06,
-      "loss": 0.6067,
+      "loss": 0.7538,
       "step": 19225
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3817773213244165e-06,
-      "loss": 0.777,
+      "loss": 0.9357,
       "step": 19226
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.380802964849589e-06,
-      "loss": 0.783,
+      "loss": 0.8367,
       "step": 19227
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.379828686333148e-06,
-      "loss": 0.7358,
+      "loss": 0.8242,
       "step": 19228
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.37885448578861e-06,
-      "loss": 0.7,
+      "loss": 0.8143,
       "step": 19229
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.377880363229491e-06,
-      "loss": 0.7164,
+      "loss": 0.837,
       "step": 19230
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3769063186693015e-06,
-      "loss": 0.8127,
+      "loss": 0.8723,
       "step": 19231
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.375932352121562e-06,
-      "loss": 0.6385,
+      "loss": 0.6757,
       "step": 19232
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.374958463599783e-06,
-      "loss": 0.6627,
+      "loss": 0.8261,
       "step": 19233
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3739846531174726e-06,
-      "loss": 0.7366,
+      "loss": 0.8635,
       "step": 19234
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.37301092068814e-06,
-      "loss": 0.7538,
+      "loss": 0.8081,
       "step": 19235
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.372037266325304e-06,
-      "loss": 0.7084,
+      "loss": 0.7799,
       "step": 19236
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.371063690042459e-06,
-      "loss": 0.7745,
+      "loss": 0.852,
       "step": 19237
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.370090191853123e-06,
-      "loss": 0.6459,
+      "loss": 0.7795,
       "step": 19238
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.369116771770796e-06,
-      "loss": 0.7803,
+      "loss": 0.874,
       "step": 19239
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.368143429808986e-06,
-      "loss": 0.7639,
+      "loss": 0.8172,
       "step": 19240
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.367170165981196e-06,
-      "loss": 0.8809,
+      "loss": 0.968,
       "step": 19241
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.366196980300927e-06,
-      "loss": 0.6676,
+      "loss": 0.7859,
       "step": 19242
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.365223872781682e-06,
-      "loss": 0.6851,
+      "loss": 0.7834,
       "step": 19243
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.364250843436957e-06,
-      "loss": 0.7253,
+      "loss": 0.8311,
       "step": 19244
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.363277892280259e-06,
-      "loss": 0.6311,
+      "loss": 0.7673,
       "step": 19245
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.362305019325084e-06,
-      "loss": 0.7952,
+      "loss": 0.8955,
       "step": 19246
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.361332224584926e-06,
-      "loss": 0.6538,
+      "loss": 0.6693,
       "step": 19247
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.360359508073281e-06,
-      "loss": 0.8071,
+      "loss": 0.8145,
       "step": 19248
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.359386869803654e-06,
-      "loss": 0.7011,
+      "loss": 0.8094,
       "step": 19249
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3584143097895226e-06,
-      "loss": 0.735,
+      "loss": 0.8106,
       "step": 19250
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.357441828044392e-06,
-      "loss": 0.7112,
+      "loss": 0.8605,
       "step": 19251
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3564694245817505e-06,
-      "loss": 0.7826,
+      "loss": 0.8333,
       "step": 19252
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.355497099415089e-06,
-      "loss": 0.7441,
+      "loss": 0.8354,
       "step": 19253
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3545248525578935e-06,
-      "loss": 0.723,
+      "loss": 0.7879,
       "step": 19254
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.353552684023662e-06,
-      "loss": 0.6843,
+      "loss": 0.8495,
       "step": 19255
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.352580593825869e-06,
-      "loss": 0.8689,
+      "loss": 0.8721,
       "step": 19256
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.351608581978012e-06,
-      "loss": 0.6142,
+      "loss": 0.7259,
       "step": 19257
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.350636648493571e-06,
-      "loss": 0.6626,
+      "loss": 0.7666,
       "step": 19258
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.349664793386032e-06,
-      "loss": 0.5963,
+      "loss": 0.7365,
       "step": 19259
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.348693016668875e-06,
-      "loss": 0.8856,
+      "loss": 0.8976,
       "step": 19260
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.347721318355582e-06,
-      "loss": 0.7039,
+      "loss": 0.7892,
       "step": 19261
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.346749698459643e-06,
-      "loss": 0.7444,
+      "loss": 0.7713,
       "step": 19262
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.345778156994526e-06,
-      "loss": 0.5851,
+      "loss": 0.7228,
       "step": 19263
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.344806693973715e-06,
-      "loss": 0.6783,
+      "loss": 0.8739,
       "step": 19264
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.34383530941069e-06,
-      "loss": 0.6946,
+      "loss": 0.799,
       "step": 19265
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.342864003318924e-06,
-      "loss": 0.6848,
+      "loss": 0.804,
       "step": 19266
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.34189277571189e-06,
-      "loss": 0.6223,
+      "loss": 0.743,
       "step": 19267
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.340921626603073e-06,
-      "loss": 0.802,
+      "loss": 0.8875,
       "step": 19268
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.339950556005933e-06,
-      "loss": 0.6902,
+      "loss": 0.7508,
       "step": 19269
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.33897956393395e-06,
-      "loss": 0.7086,
+      "loss": 0.7542,
       "step": 19270
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.338008650400596e-06,
-      "loss": 0.7105,
+      "loss": 0.7899,
       "step": 19271
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.337037815419337e-06,
-      "loss": 0.6417,
+      "loss": 0.7838,
       "step": 19272
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.336067059003641e-06,
-      "loss": 0.824,
+      "loss": 0.8642,
       "step": 19273
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3350963811669855e-06,
-      "loss": 0.8047,
+      "loss": 0.8527,
       "step": 19274
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.334125781922822e-06,
-      "loss": 0.7132,
+      "loss": 0.8741,
       "step": 19275
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.33315526128463e-06,
-      "loss": 0.777,
+      "loss": 0.8139,
       "step": 19276
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.332184819265867e-06,
-      "loss": 0.7677,
+      "loss": 0.9057,
       "step": 19277
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.331214455879998e-06,
-      "loss": 0.773,
+      "loss": 0.8882,
       "step": 19278
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.330244171140486e-06,
-      "loss": 0.8558,
+      "loss": 0.9006,
       "step": 19279
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.329273965060789e-06,
-      "loss": 0.8358,
+      "loss": 0.8658,
       "step": 19280
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.328303837654376e-06,
-      "loss": 0.7186,
+      "loss": 0.8294,
       "step": 19281
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3273337889346935e-06,
-      "loss": 0.7711,
+      "loss": 0.825,
       "step": 19282
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.32636381891521e-06,
-      "loss": 0.7092,
+      "loss": 0.7944,
       "step": 19283
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.325393927609378e-06,
-      "loss": 0.8398,
+      "loss": 0.8454,
       "step": 19284
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.324424115030654e-06,
-      "loss": 0.6452,
+      "loss": 0.7252,
       "step": 19285
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.32345438119249e-06,
-      "loss": 0.6539,
+      "loss": 0.7645,
       "step": 19286
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3224847261083495e-06,
-      "loss": 0.6116,
+      "loss": 0.7747,
       "step": 19287
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.321515149791671e-06,
-      "loss": 0.7888,
+      "loss": 0.8883,
       "step": 19288
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.320545652255916e-06,
-      "loss": 0.7402,
+      "loss": 0.9073,
       "step": 19289
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.319576233514533e-06,
-      "loss": 0.7593,
+      "loss": 0.8124,
       "step": 19290
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.318606893580969e-06,
-      "loss": 0.8521,
+      "loss": 0.8518,
       "step": 19291
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.317637632468671e-06,
-      "loss": 0.7964,
+      "loss": 0.8968,
       "step": 19292
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.316668450191092e-06,
-      "loss": 0.7199,
+      "loss": 0.8055,
       "step": 19293
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.315699346761674e-06,
-      "loss": 0.7574,
+      "loss": 0.8363,
       "step": 19294
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.314730322193863e-06,
-      "loss": 0.7638,
+      "loss": 0.8762,
       "step": 19295
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.313761376501102e-06,
-      "loss": 0.6497,
+      "loss": 0.7101,
       "step": 19296
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.312792509696835e-06,
-      "loss": 0.7592,
+      "loss": 0.8243,
       "step": 19297
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3118237217945024e-06,
-      "loss": 0.8228,
+      "loss": 0.8978,
       "step": 19298
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.310855012807542e-06,
-      "loss": 0.7635,
+      "loss": 0.8616,
       "step": 19299
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.309886382749403e-06,
-      "loss": 0.7195,
+      "loss": 0.854,
       "step": 19300
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3089178316335115e-06,
-      "loss": 0.6777,
+      "loss": 0.8868,
       "step": 19301
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3079493594733126e-06,
-      "loss": 0.6042,
+      "loss": 0.7475,
       "step": 19302
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.30698096628224e-06,
-      "loss": 0.7435,
+      "loss": 0.8604,
       "step": 19303
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.306012652073731e-06,
-      "loss": 0.7171,
+      "loss": 0.8321,
       "step": 19304
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.305044416861213e-06,
-      "loss": 0.7071,
+      "loss": 0.8261,
       "step": 19305
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.304076260658127e-06,
-      "loss": 0.7004,
+      "loss": 0.7184,
       "step": 19306
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.303108183477902e-06,
-      "loss": 0.5764,
+      "loss": 0.7189,
       "step": 19307
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.3021401853339676e-06,
-      "loss": 0.8084,
+      "loss": 0.9209,
       "step": 19308
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.301172266239754e-06,
-      "loss": 0.7664,
+      "loss": 0.8448,
       "step": 19309
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.300204426208689e-06,
-      "loss": 0.6438,
+      "loss": 0.7951,
       "step": 19310
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.299236665254199e-06,
-      "loss": 0.783,
+      "loss": 0.9299,
       "step": 19311
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.298268983389714e-06,
-      "loss": 0.6478,
+      "loss": 0.7629,
       "step": 19312
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2973013806286575e-06,
-      "loss": 0.8516,
+      "loss": 0.8923,
       "step": 19313
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.296333856984454e-06,
-      "loss": 0.6565,
+      "loss": 0.763,
       "step": 19314
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.295366412470521e-06,
-      "loss": 0.7385,
+      "loss": 0.8578,
       "step": 19315
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2943990471002925e-06,
-      "loss": 0.612,
+      "loss": 0.7625,
       "step": 19316
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.293431760887179e-06,
-      "loss": 0.8191,
+      "loss": 0.8803,
       "step": 19317
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2924645538446e-06,
-      "loss": 0.5586,
+      "loss": 0.7618,
       "step": 19318
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.291497425985983e-06,
-      "loss": 0.6404,
+      "loss": 0.7469,
       "step": 19319
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.290530377324734e-06,
-      "loss": 0.6926,
+      "loss": 0.8362,
       "step": 19320
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.289563407874279e-06,
-      "loss": 0.6577,
+      "loss": 0.7289,
       "step": 19321
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.288596517648029e-06,
-      "loss": 0.6282,
+      "loss": 0.7603,
       "step": 19322
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2876297066594e-06,
-      "loss": 0.7182,
+      "loss": 0.8441,
       "step": 19323
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.286662974921799e-06,
-      "loss": 0.7261,
+      "loss": 0.8139,
       "step": 19324
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.285696322448648e-06,
-      "loss": 0.7112,
+      "loss": 0.8116,
       "step": 19325
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.284729749253353e-06,
-      "loss": 0.7873,
+      "loss": 0.8877,
       "step": 19326
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.283763255349322e-06,
-      "loss": 0.6934,
+      "loss": 0.7677,
       "step": 19327
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2827968407499675e-06,
-      "loss": 0.6665,
+      "loss": 0.7379,
       "step": 19328
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.281830505468694e-06,
-      "loss": 0.4975,
+      "loss": 0.6908,
       "step": 19329
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.280864249518905e-06,
-      "loss": 0.7943,
+      "loss": 0.8402,
       "step": 19330
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.279898072914016e-06,
-      "loss": 0.7232,
+      "loss": 0.7542,
       "step": 19331
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.278931975667424e-06,
-      "loss": 0.6879,
+      "loss": 0.7465,
       "step": 19332
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.277965957792534e-06,
-      "loss": 0.6166,
+      "loss": 0.773,
       "step": 19333
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.277000019302744e-06,
-      "loss": 0.6753,
+      "loss": 0.7609,
       "step": 19334
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2760341602114665e-06,
-      "loss": 0.792,
+      "loss": 0.8536,
       "step": 19335
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.27506838053209e-06,
-      "loss": 0.772,
+      "loss": 0.8603,
       "step": 19336
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.274102680278013e-06,
-      "loss": 0.7577,
+      "loss": 0.8495,
       "step": 19337
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.273137059462642e-06,
-      "loss": 0.6629,
+      "loss": 0.7945,
       "step": 19338
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2721715180993686e-06,
-      "loss": 0.7614,
+      "loss": 0.8347,
       "step": 19339
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.271206056201588e-06,
-      "loss": 0.6986,
+      "loss": 0.8846,
       "step": 19340
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2702406737826956e-06,
-      "loss": 0.7481,
+      "loss": 0.8307,
       "step": 19341
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2692753708560845e-06,
-      "loss": 0.7147,
+      "loss": 0.8465,
       "step": 19342
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.268310147435143e-06,
-      "loss": 0.6706,
+      "loss": 0.7539,
       "step": 19343
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.26734500353327e-06,
-      "loss": 0.6834,
+      "loss": 0.8319,
       "step": 19344
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.266379939163851e-06,
-      "loss": 0.6485,
+      "loss": 0.7604,
       "step": 19345
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.265414954340276e-06,
-      "loss": 0.6694,
+      "loss": 0.7396,
       "step": 19346
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.264450049075928e-06,
-      "loss": 0.6919,
+      "loss": 0.7365,
       "step": 19347
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.263485223384205e-06,
-      "loss": 0.6319,
+      "loss": 0.7745,
       "step": 19348
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.262520477278478e-06,
-      "loss": 0.6145,
+      "loss": 0.6763,
       "step": 19349
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.261555810772143e-06,
-      "loss": 0.7214,
+      "loss": 0.7793,
       "step": 19350
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.260591223878579e-06,
-      "loss": 0.7057,
+      "loss": 0.8406,
       "step": 19351
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2596267166111705e-06,
-      "loss": 0.801,
+      "loss": 0.8113,
       "step": 19352
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.258662288983292e-06,
-      "loss": 0.8027,
+      "loss": 0.8928,
       "step": 19353
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.257697941008334e-06,
-      "loss": 0.6519,
+      "loss": 0.8549,
       "step": 19354
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2567336726996674e-06,
-      "loss": 0.7009,
+      "loss": 0.872,
       "step": 19355
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2557694840706695e-06,
-      "loss": 0.7414,
+      "loss": 0.7322,
       "step": 19356
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.254805375134723e-06,
-      "loss": 0.639,
+      "loss": 0.7643,
       "step": 19357
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.253841345905201e-06,
-      "loss": 0.6883,
+      "loss": 0.8317,
       "step": 19358
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.252877396395477e-06,
-      "loss": 0.8099,
+      "loss": 0.8706,
       "step": 19359
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.251913526618922e-06,
-      "loss": 0.7436,
+      "loss": 0.7913,
       "step": 19360
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2509497365889185e-06,
-      "loss": 0.7229,
+      "loss": 0.8312,
       "step": 19361
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.249986026318823e-06,
-      "loss": 0.7785,
+      "loss": 0.8643,
       "step": 19362
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.249022395822017e-06,
-      "loss": 0.6101,
+      "loss": 0.7819,
       "step": 19363
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.248058845111866e-06,
-      "loss": 0.7201,
+      "loss": 0.7491,
       "step": 19364
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.247095374201736e-06,
-      "loss": 0.712,
+      "loss": 0.7671,
       "step": 19365
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.246131983104993e-06,
-      "loss": 0.6723,
+      "loss": 0.8372,
       "step": 19366
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.245168671835013e-06,
-      "loss": 0.6728,
+      "loss": 0.7602,
       "step": 19367
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.244205440405144e-06,
-      "loss": 0.8453,
+      "loss": 0.9495,
       "step": 19368
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.243242288828763e-06,
-      "loss": 0.7912,
+      "loss": 0.8435,
       "step": 19369
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.242279217119226e-06,
-      "loss": 0.803,
+      "loss": 0.8284,
       "step": 19370
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2413162252898964e-06,
-      "loss": 0.6458,
+      "loss": 0.7955,
       "step": 19371
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2403533133541305e-06,
-      "loss": 0.6316,
+      "loss": 0.7845,
       "step": 19372
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.239390481325297e-06,
-      "loss": 0.6171,
+      "loss": 0.7301,
       "step": 19373
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.238427729216744e-06,
-      "loss": 0.8743,
+      "loss": 0.8401,
       "step": 19374
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2374650570418285e-06,
-      "loss": 0.7641,
+      "loss": 0.8332,
       "step": 19375
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.236502464813913e-06,
-      "loss": 0.7436,
+      "loss": 0.9047,
       "step": 19376
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.235539952546348e-06,
-      "loss": 0.7223,
+      "loss": 0.8003,
       "step": 19377
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.234577520252488e-06,
-      "loss": 0.7062,
+      "loss": 0.813,
       "step": 19378
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.23361516794568e-06,
-      "loss": 0.6849,
+      "loss": 0.8152,
       "step": 19379
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.232652895639289e-06,
-      "loss": 0.7185,
+      "loss": 0.7981,
       "step": 19380
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.231690703346648e-06,
-      "loss": 0.7236,
+      "loss": 0.8374,
       "step": 19381
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.230728591081118e-06,
-      "loss": 0.726,
+      "loss": 0.8088,
       "step": 19382
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2297665588560434e-06,
-      "loss": 0.8264,
+      "loss": 0.9059,
       "step": 19383
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.228804606684771e-06,
-      "loss": 0.6306,
+      "loss": 0.7308,
       "step": 19384
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.2278427345806426e-06,
-      "loss": 0.7581,
+      "loss": 0.856,
       "step": 19385
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.226880942557015e-06,
-      "loss": 0.6133,
+      "loss": 0.7519,
       "step": 19386
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2259192306272145e-06,
-      "loss": 0.6218,
+      "loss": 0.7251,
       "step": 19387
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2249575988045965e-06,
-      "loss": 0.6585,
+      "loss": 0.757,
       "step": 19388
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.223996047102498e-06,
-      "loss": 0.6352,
+      "loss": 0.7654,
       "step": 19389
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.223034575534259e-06,
-      "loss": 0.7767,
+      "loss": 0.8307,
       "step": 19390
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.222073184113216e-06,
-      "loss": 0.6056,
+      "loss": 0.6987,
       "step": 19391
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.221111872852713e-06,
-      "loss": 0.5799,
+      "loss": 0.7156,
       "step": 19392
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.220150641766088e-06,
-      "loss": 0.5746,
+      "loss": 0.7614,
       "step": 19393
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.219189490866665e-06,
-      "loss": 0.9836,
+      "loss": 0.9601,
       "step": 19394
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.218228420167788e-06,
-      "loss": 0.7183,
+      "loss": 0.8252,
       "step": 19395
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.217267429682789e-06,
-      "loss": 0.7742,
+      "loss": 0.8093,
       "step": 19396
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.216306519425e-06,
-      "loss": 0.6709,
+      "loss": 0.7511,
       "step": 19397
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.215345689407748e-06,
-      "loss": 0.6569,
+      "loss": 0.8171,
       "step": 19398
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.214384939644375e-06,
-      "loss": 0.7029,
+      "loss": 0.822,
       "step": 19399
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2134242701481955e-06,
-      "loss": 0.7695,
+      "loss": 0.8843,
       "step": 19400
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.212463680932547e-06,
-      "loss": 0.6817,
+      "loss": 0.8732,
       "step": 19401
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.211503172010753e-06,
-      "loss": 0.6352,
+      "loss": 0.6762,
       "step": 19402
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.210542743396141e-06,
-      "loss": 0.8196,
+      "loss": 0.8807,
       "step": 19403
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2095823951020294e-06,
-      "loss": 0.7691,
+      "loss": 0.8642,
       "step": 19404
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.208622127141752e-06,
-      "loss": 0.6322,
+      "loss": 0.7655,
       "step": 19405
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2076619395286244e-06,
-      "loss": 0.6875,
+      "loss": 0.8121,
       "step": 19406
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.206701832275969e-06,
-      "loss": 0.5723,
+      "loss": 0.7465,
       "step": 19407
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.205741805397107e-06,
-      "loss": 0.6716,
+      "loss": 0.802,
       "step": 19408
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2047818589053546e-06,
-      "loss": 0.818,
+      "loss": 0.874,
       "step": 19409
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.20382199281403e-06,
-      "loss": 0.5652,
+      "loss": 0.7816,
       "step": 19410
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.2028622071364535e-06,
-      "loss": 0.6617,
+      "loss": 0.7711,
       "step": 19411
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.201902501885944e-06,
-      "loss": 0.7298,
+      "loss": 0.8185,
       "step": 19412
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.200942877075802e-06,
-      "loss": 0.7807,
+      "loss": 0.8464,
       "step": 19413
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.199983332719355e-06,
-      "loss": 0.706,
+      "loss": 0.7886,
       "step": 19414
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.199023868829909e-06,
-      "loss": 0.7129,
+      "loss": 0.7396,
       "step": 19415
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.198064485420777e-06,
-      "loss": 0.7798,
+      "loss": 0.8734,
       "step": 19416
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.197105182505265e-06,
-      "loss": 0.6541,
+      "loss": 0.7704,
       "step": 19417
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1961459600966916e-06,
-      "loss": 0.7405,
+      "loss": 0.8528,
       "step": 19418
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.195186818208351e-06,
-      "loss": 0.7269,
+      "loss": 0.8302,
       "step": 19419
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.194227756853562e-06,
-      "loss": 0.7578,
+      "loss": 0.8823,
       "step": 19420
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.193268776045625e-06,
-      "loss": 0.756,
+      "loss": 0.9121,
       "step": 19421
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.192309875797845e-06,
-      "loss": 0.641,
+      "loss": 0.7473,
       "step": 19422
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1913510561235225e-06,
-      "loss": 0.664,
+      "loss": 0.7957,
       "step": 19423
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.190392317035966e-06,
-      "loss": 0.5895,
+      "loss": 0.7676,
       "step": 19424
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.189433658548473e-06,
-      "loss": 0.7643,
+      "loss": 0.8765,
       "step": 19425
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.188475080674345e-06,
-      "loss": 0.5676,
+      "loss": 0.7446,
       "step": 19426
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.187516583426874e-06,
-      "loss": 0.6402,
+      "loss": 0.8394,
       "step": 19427
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.186558166819372e-06,
-      "loss": 0.8337,
+      "loss": 0.8919,
       "step": 19428
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.18559983086512e-06,
-      "loss": 0.7855,
+      "loss": 0.9126,
       "step": 19429
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.184641575577425e-06,
-      "loss": 0.775,
+      "loss": 0.8185,
       "step": 19430
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.18368340096958e-06,
-      "loss": 0.7491,
+      "loss": 0.8331,
       "step": 19431
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.182725307054869e-06,
-      "loss": 0.6316,
+      "loss": 0.8534,
       "step": 19432
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1817672938465935e-06,
-      "loss": 0.6603,
+      "loss": 0.7985,
       "step": 19433
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.180809361358043e-06,
-      "loss": 0.8275,
+      "loss": 0.8326,
       "step": 19434
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.179851509602505e-06,
-      "loss": 0.6738,
+      "loss": 0.821,
       "step": 19435
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.178893738593267e-06,
-      "loss": 0.7459,
+      "loss": 0.8064,
       "step": 19436
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1779360483436215e-06,
-      "loss": 0.7213,
+      "loss": 0.7978,
       "step": 19437
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.176978438866854e-06,
-      "loss": 0.7107,
+      "loss": 0.8404,
       "step": 19438
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.176020910176249e-06,
-      "loss": 0.6954,
+      "loss": 0.748,
       "step": 19439
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.175063462285091e-06,
-      "loss": 0.789,
+      "loss": 0.8436,
       "step": 19440
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.174106095206662e-06,
-      "loss": 0.7102,
+      "loss": 0.8577,
       "step": 19441
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.173148808954242e-06,
-      "loss": 0.6916,
+      "loss": 0.7637,
       "step": 19442
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.172191603541118e-06,
-      "loss": 0.6607,
+      "loss": 0.8803,
       "step": 19443
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1712344789805684e-06,
-      "loss": 0.702,
+      "loss": 0.7795,
       "step": 19444
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.170277435285869e-06,
-      "loss": 0.6565,
+      "loss": 0.8235,
       "step": 19445
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.169320472470295e-06,
-      "loss": 0.8464,
+      "loss": 0.851,
       "step": 19446
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.168363590547135e-06,
-      "loss": 0.6535,
+      "loss": 0.8345,
       "step": 19447
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.167406789529649e-06,
-      "loss": 0.6475,
+      "loss": 0.7956,
       "step": 19448
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.166450069431122e-06,
-      "loss": 0.8204,
+      "loss": 0.8657,
       "step": 19449
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1654934302648234e-06,
-      "loss": 0.7887,
+      "loss": 0.8483,
       "step": 19450
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.164536872044025e-06,
-      "loss": 0.7258,
+      "loss": 0.8221,
       "step": 19451
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.163580394781999e-06,
-      "loss": 0.7611,
+      "loss": 0.7957,
       "step": 19452
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.162623998492012e-06,
-      "loss": 0.6352,
+      "loss": 0.7641,
       "step": 19453
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.161667683187337e-06,
-      "loss": 0.6678,
+      "loss": 0.7854,
       "step": 19454
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.160711448881236e-06,
-      "loss": 0.7217,
+      "loss": 0.8153,
       "step": 19455
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.159755295586982e-06,
-      "loss": 0.7106,
+      "loss": 0.7432,
       "step": 19456
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1587992233178366e-06,
-      "loss": 0.828,
+      "loss": 0.9096,
       "step": 19457
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.157843232087064e-06,
-      "loss": 0.7192,
+      "loss": 0.8783,
       "step": 19458
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1568873219079245e-06,
-      "loss": 0.7716,
+      "loss": 0.8888,
       "step": 19459
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.15593149279369e-06,
-      "loss": 0.6878,
+      "loss": 0.7344,
       "step": 19460
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1549757447576076e-06,
-      "loss": 0.7239,
+      "loss": 0.8681,
       "step": 19461
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.154020077812947e-06,
-      "loss": 0.6072,
+      "loss": 0.7155,
       "step": 19462
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.153064491972963e-06,
-      "loss": 0.6361,
+      "loss": 0.8075,
       "step": 19463
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.152108987250915e-06,
-      "loss": 0.5931,
+      "loss": 0.7631,
       "step": 19464
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.151153563660053e-06,
-      "loss": 0.7828,
+      "loss": 0.8633,
       "step": 19465
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.150198221213645e-06,
-      "loss": 0.6825,
+      "loss": 0.7793,
       "step": 19466
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.149242959924928e-06,
-      "loss": 0.7155,
+      "loss": 0.8987,
       "step": 19467
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1482877798071694e-06,
-      "loss": 0.6216,
+      "loss": 0.7182,
       "step": 19468
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.147332680873615e-06,
-      "loss": 0.585,
+      "loss": 0.7159,
       "step": 19469
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.146377663137514e-06,
-      "loss": 0.7128,
+      "loss": 0.8341,
       "step": 19470
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.145422726612119e-06,
-      "loss": 0.8172,
+      "loss": 0.8858,
       "step": 19471
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.144467871310674e-06,
-      "loss": 0.654,
+      "loss": 0.7031,
       "step": 19472
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1435130972464355e-06,
-      "loss": 0.8556,
+      "loss": 0.9697,
       "step": 19473
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1425584044326376e-06,
-      "loss": 0.7088,
+      "loss": 0.8299,
       "step": 19474
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1416037928825334e-06,
-      "loss": 0.7109,
+      "loss": 0.8097,
       "step": 19475
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.140649262609366e-06,
-      "loss": 0.946,
+      "loss": 1.0286,
       "step": 19476
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.139694813626376e-06,
-      "loss": 0.7334,
+      "loss": 0.7402,
       "step": 19477
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1387404459468015e-06,
-      "loss": 0.7708,
+      "loss": 0.8857,
       "step": 19478
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.137786159583895e-06,
-      "loss": 0.7965,
+      "loss": 0.8557,
       "step": 19479
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.136831954550882e-06,
-      "loss": 0.702,
+      "loss": 0.7378,
       "step": 19480
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.135877830861009e-06,
-      "loss": 0.6585,
+      "loss": 0.846,
       "step": 19481
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.134923788527512e-06,
-      "loss": 0.7737,
+      "loss": 0.9011,
       "step": 19482
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1339698275636255e-06,
-      "loss": 0.7977,
+      "loss": 0.8679,
       "step": 19483
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.133015947982581e-06,
-      "loss": 0.846,
+      "loss": 0.9187,
       "step": 19484
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.132062149797623e-06,
-      "loss": 0.704,
+      "loss": 0.8087,
       "step": 19485
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.131108433021971e-06,
-      "loss": 0.7199,
+      "loss": 0.7692,
       "step": 19486
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.130154797668865e-06,
-      "loss": 0.7869,
+      "loss": 0.887,
       "step": 19487
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.129201243751534e-06,
-      "loss": 0.6211,
+      "loss": 0.7312,
       "step": 19488
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.128247771283206e-06,
-      "loss": 0.6118,
+      "loss": 0.687,
       "step": 19489
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.127294380277109e-06,
-      "loss": 0.8064,
+      "loss": 0.8872,
       "step": 19490
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.126341070746467e-06,
-      "loss": 0.6452,
+      "loss": 0.7007,
       "step": 19491
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1253878427045155e-06,
-      "loss": 0.7763,
+      "loss": 0.8711,
       "step": 19492
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.124434696164465e-06,
-      "loss": 0.6865,
+      "loss": 0.7202,
       "step": 19493
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.123481631139552e-06,
-      "loss": 0.7423,
+      "loss": 0.8856,
       "step": 19494
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.122528647642993e-06,
-      "loss": 0.6887,
+      "loss": 0.7732,
       "step": 19495
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.12157574568801e-06,
-      "loss": 0.662,
+      "loss": 0.7768,
       "step": 19496
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.120622925287819e-06,
-      "loss": 0.7469,
+      "loss": 0.8365,
       "step": 19497
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.119670186455651e-06,
-      "loss": 0.7616,
+      "loss": 0.8545,
       "step": 19498
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.118717529204709e-06,
-      "loss": 0.6022,
+      "loss": 0.759,
       "step": 19499
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.11776495354822e-06,
-      "loss": 0.7536,
+      "loss": 0.894,
       "step": 19500
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.116812459499397e-06,
-      "loss": 0.5747,
+      "loss": 0.7115,
       "step": 19501
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.115860047071453e-06,
-      "loss": 0.7954,
+      "loss": 0.8636,
       "step": 19502
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1149077162776e-06,
-      "loss": 0.84,
+      "loss": 0.8493,
       "step": 19503
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.113955467131054e-06,
-      "loss": 0.5841,
+      "loss": 0.7061,
       "step": 19504
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.113003299645028e-06,
-      "loss": 0.6198,
+      "loss": 0.7595,
       "step": 19505
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1120512138327265e-06,
-      "loss": 0.6687,
+      "loss": 0.7982,
       "step": 19506
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.11109920970736e-06,
-      "loss": 0.8167,
+      "loss": 0.8575,
       "step": 19507
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.110147287282138e-06,
-      "loss": 0.7351,
+      "loss": 0.8192,
       "step": 19508
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.109195446570265e-06,
-      "loss": 0.707,
+      "loss": 0.8258,
       "step": 19509
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.108243687584944e-06,
-      "loss": 0.7456,
+      "loss": 0.811,
       "step": 19510
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.10729201033939e-06,
-      "loss": 0.7738,
+      "loss": 0.8846,
       "step": 19511
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.106340414846791e-06,
-      "loss": 0.7406,
+      "loss": 0.7913,
       "step": 19512
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.105388901120361e-06,
-      "loss": 0.5832,
+      "loss": 0.7348,
       "step": 19513
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.104437469173296e-06,
-      "loss": 0.6475,
+      "loss": 0.8479,
       "step": 19514
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1034861190187956e-06,
-      "loss": 0.7175,
+      "loss": 0.8377,
       "step": 19515
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1025348506700565e-06,
-      "loss": 0.7184,
+      "loss": 0.8103,
       "step": 19516
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.101583664140286e-06,
-      "loss": 0.6425,
+      "loss": 0.8096,
       "step": 19517
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.1006325594426664e-06,
-      "loss": 0.6707,
+      "loss": 0.8598,
       "step": 19518
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.099681536590404e-06,
-      "loss": 0.5872,
+      "loss": 0.7756,
       "step": 19519
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0987305955966885e-06,
-      "loss": 0.7333,
+      "loss": 0.9044,
       "step": 19520
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.097779736474713e-06,
-      "loss": 0.7547,
+      "loss": 0.9023,
       "step": 19521
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0968289592376664e-06,
-      "loss": 0.6212,
+      "loss": 0.7795,
       "step": 19522
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.095878263898745e-06,
-      "loss": 0.7345,
+      "loss": 0.8711,
       "step": 19523
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.094927650471136e-06,
-      "loss": 0.7785,
+      "loss": 0.8848,
       "step": 19524
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0939771189680285e-06,
-      "loss": 0.7096,
+      "loss": 0.7837,
       "step": 19525
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0930266694026055e-06,
-      "loss": 0.578,
+      "loss": 0.6943,
       "step": 19526
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.092076301788063e-06,
-      "loss": 0.8122,
+      "loss": 0.8995,
       "step": 19527
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.091126016137577e-06,
-      "loss": 0.6231,
+      "loss": 0.7255,
       "step": 19528
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.090175812464328e-06,
-      "loss": 0.7241,
+      "loss": 0.8607,
       "step": 19529
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0892256907815145e-06,
-      "loss": 0.7032,
+      "loss": 0.762,
       "step": 19530
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.088275651102299e-06,
-      "loss": 0.7588,
+      "loss": 0.84,
       "step": 19531
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.087325693439877e-06,
-      "loss": 0.7066,
+      "loss": 0.768,
       "step": 19532
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.08637581780742e-06,
-      "loss": 0.7683,
+      "loss": 0.8682,
       "step": 19533
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.085426024218109e-06,
-      "loss": 0.6663,
+      "loss": 0.8788,
       "step": 19534
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.084476312685116e-06,
-      "loss": 0.6048,
+      "loss": 0.776,
       "step": 19535
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.083526683221626e-06,
-      "loss": 0.5887,
+      "loss": 0.8078,
       "step": 19536
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0825771358408075e-06,
-      "loss": 0.6357,
+      "loss": 0.7639,
       "step": 19537
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.081627670555836e-06,
-      "loss": 0.6757,
+      "loss": 0.7779,
       "step": 19538
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.080678287379884e-06,
-      "loss": 0.7045,
+      "loss": 0.8596,
       "step": 19539
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.079728986326122e-06,
-      "loss": 0.6561,
+      "loss": 0.7566,
       "step": 19540
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0787797674077166e-06,
-      "loss": 0.7286,
+      "loss": 0.7462,
       "step": 19541
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.077830630637846e-06,
-      "loss": 0.6671,
+      "loss": 0.7915,
       "step": 19542
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.076881576029671e-06,
-      "loss": 0.7255,
+      "loss": 0.8261,
       "step": 19543
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.075932603596361e-06,
-      "loss": 0.7554,
+      "loss": 0.8567,
       "step": 19544
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.074983713351077e-06,
-      "loss": 0.6664,
+      "loss": 0.8029,
       "step": 19545
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0740349053069946e-06,
-      "loss": 0.7408,
+      "loss": 0.8063,
       "step": 19546
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.073086179477268e-06,
-      "loss": 0.6104,
+      "loss": 0.8072,
       "step": 19547
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.072137535875057e-06,
-      "loss": 0.7966,
+      "loss": 0.843,
       "step": 19548
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.07118897451353e-06,
-      "loss": 0.6479,
+      "loss": 0.789,
       "step": 19549
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.070240495405845e-06,
-      "loss": 0.7178,
+      "loss": 0.821,
       "step": 19550
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.06929209856516e-06,
-      "loss": 0.7106,
+      "loss": 0.7661,
       "step": 19551
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.068343784004632e-06,
-      "loss": 0.5121,
+      "loss": 0.7275,
       "step": 19552
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0673955517374174e-06,
-      "loss": 0.7147,
+      "loss": 0.8237,
       "step": 19553
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.06644740177667e-06,
-      "loss": 0.7596,
+      "loss": 0.8403,
       "step": 19554
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.065499334135549e-06,
-      "loss": 0.8153,
+      "loss": 0.8575,
       "step": 19555
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.064551348827205e-06,
-      "loss": 0.7547,
+      "loss": 0.8438,
       "step": 19556
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.06360344586479e-06,
-      "loss": 0.7472,
+      "loss": 0.7805,
       "step": 19557
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0626556252614515e-06,
-      "loss": 0.6769,
+      "loss": 0.6815,
       "step": 19558
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.061707887030348e-06,
-      "loss": 0.7412,
+      "loss": 0.8817,
       "step": 19559
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.060760231184616e-06,
-      "loss": 0.692,
+      "loss": 0.8157,
       "step": 19560
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.059812657737412e-06,
-      "loss": 0.5758,
+      "loss": 0.6566,
       "step": 19561
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.05886516670188e-06,
-      "loss": 0.679,
+      "loss": 0.7616,
       "step": 19562
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0579177580911644e-06,
-      "loss": 0.6927,
+      "loss": 0.8451,
       "step": 19563
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.056970431918406e-06,
-      "loss": 0.7471,
+      "loss": 0.8092,
       "step": 19564
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.056023188196757e-06,
-      "loss": 0.7277,
+      "loss": 0.897,
       "step": 19565
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.05507602693935e-06,
-      "loss": 0.6556,
+      "loss": 0.7682,
       "step": 19566
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.054128948159324e-06,
-      "loss": 0.8012,
+      "loss": 0.8047,
       "step": 19567
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0531819518698265e-06,
-      "loss": 0.6911,
+      "loss": 0.7615,
       "step": 19568
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.052235038083993e-06,
-      "loss": 0.7668,
+      "loss": 0.8045,
       "step": 19569
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.051288206814959e-06,
-      "loss": 0.7986,
+      "loss": 0.7981,
       "step": 19570
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.050341458075857e-06,
-      "loss": 0.6497,
+      "loss": 0.7927,
       "step": 19571
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0493947918798334e-06,
-      "loss": 0.689,
+      "loss": 0.7816,
       "step": 19572
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.048448208240006e-06,
-      "loss": 0.5621,
+      "loss": 0.7769,
       "step": 19573
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.047501707169521e-06,
-      "loss": 0.6374,
+      "loss": 0.6956,
       "step": 19574
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.046555288681504e-06,
-      "loss": 0.817,
+      "loss": 0.8602,
       "step": 19575
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.045608952789085e-06,
-      "loss": 0.5979,
+      "loss": 0.6978,
       "step": 19576
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.04466269950539e-06,
-      "loss": 0.6377,
+      "loss": 0.8138,
       "step": 19577
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.043716528843557e-06,
-      "loss": 0.7541,
+      "loss": 0.8131,
       "step": 19578
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0427704408167e-06,
-      "loss": 0.6952,
+      "loss": 0.7927,
       "step": 19579
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.041824435437954e-06,
-      "loss": 0.7224,
+      "loss": 0.7296,
       "step": 19580
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.040878512720442e-06,
-      "loss": 0.7672,
+      "loss": 0.8826,
       "step": 19581
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.039932672677283e-06,
-      "loss": 0.7698,
+      "loss": 0.8024,
       "step": 19582
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.038986915321601e-06,
-      "loss": 0.6815,
+      "loss": 0.7852,
       "step": 19583
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.038041240666524e-06,
-      "loss": 0.5673,
+      "loss": 0.7213,
       "step": 19584
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.037095648725162e-06,
-      "loss": 0.7491,
+      "loss": 0.9096,
       "step": 19585
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.036150139510635e-06,
-      "loss": 0.8481,
+      "loss": 0.8912,
       "step": 19586
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.035204713036066e-06,
-      "loss": 0.7875,
+      "loss": 0.8058,
       "step": 19587
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0342593693145685e-06,
-      "loss": 0.7389,
+      "loss": 0.8457,
       "step": 19588
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.033314108359259e-06,
-      "loss": 0.7121,
+      "loss": 0.8634,
       "step": 19589
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.032368930183246e-06,
-      "loss": 0.7122,
+      "loss": 0.8153,
       "step": 19590
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.031423834799655e-06,
-      "loss": 0.6107,
+      "loss": 0.7314,
       "step": 19591
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0304788222215815e-06,
-      "loss": 0.7296,
+      "loss": 0.7825,
       "step": 19592
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0295338924621495e-06,
-      "loss": 0.6886,
+      "loss": 0.7904,
       "step": 19593
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.028589045534463e-06,
-      "loss": 0.7831,
+      "loss": 0.8565,
       "step": 19594
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.027644281451632e-06,
-      "loss": 0.7203,
+      "loss": 0.8036,
       "step": 19595
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0266996002267585e-06,
-      "loss": 0.6758,
+      "loss": 0.7244,
       "step": 19596
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.025755001872962e-06,
-      "loss": 0.6637,
+      "loss": 0.7549,
       "step": 19597
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.02481048640333e-06,
-      "loss": 0.8187,
+      "loss": 0.8279,
       "step": 19598
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.023866053830978e-06,
-      "loss": 0.7404,
+      "loss": 0.8181,
       "step": 19599
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.022921704169006e-06,
-      "loss": 0.802,
+      "loss": 0.9538,
       "step": 19600
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0219774374305155e-06,
-      "loss": 0.5973,
+      "loss": 0.7603,
       "step": 19601
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.021033253628601e-06,
-      "loss": 0.7315,
+      "loss": 0.7481,
       "step": 19602
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.020089152776373e-06,
-      "loss": 0.7163,
+      "loss": 0.8375,
       "step": 19603
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.019145134886926e-06,
-      "loss": 0.7178,
+      "loss": 0.8172,
       "step": 19604
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.018201199973347e-06,
-      "loss": 0.7374,
+      "loss": 0.8141,
       "step": 19605
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.017257348048743e-06,
-      "loss": 0.5946,
+      "loss": 0.6883,
       "step": 19606
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.016313579126206e-06,
-      "loss": 0.6498,
+      "loss": 0.7903,
       "step": 19607
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.015369893218826e-06,
-      "loss": 0.7072,
+      "loss": 0.7565,
       "step": 19608
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.014426290339697e-06,
-      "loss": 0.713,
+      "loss": 0.8985,
       "step": 19609
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.013482770501917e-06,
-      "loss": 0.8176,
+      "loss": 0.8493,
       "step": 19610
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.012539333718561e-06,
-      "loss": 0.7454,
+      "loss": 0.8078,
       "step": 19611
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.011595980002732e-06,
-      "loss": 0.8111,
+      "loss": 0.8736,
       "step": 19612
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.010652709367512e-06,
-      "loss": 0.6878,
+      "loss": 0.8448,
       "step": 19613
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.009709521825988e-06,
-      "loss": 0.6374,
+      "loss": 0.7351,
       "step": 19614
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.008766417391241e-06,
-      "loss": 0.5617,
+      "loss": 0.6779,
       "step": 19615
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.007823396076363e-06,
-      "loss": 0.664,
+      "loss": 0.812,
       "step": 19616
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.006880457894434e-06,
-      "loss": 0.837,
+      "loss": 0.8948,
       "step": 19617
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.005937602858536e-06,
-      "loss": 0.6575,
+      "loss": 0.8042,
       "step": 19618
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.004994830981749e-06,
-      "loss": 0.76,
+      "loss": 0.8995,
       "step": 19619
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.004052142277153e-06,
-      "loss": 0.7664,
+      "loss": 0.89,
       "step": 19620
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.003109536757822e-06,
-      "loss": 0.5894,
+      "loss": 0.8217,
       "step": 19621
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.0021670144368415e-06,
-      "loss": 0.7044,
+      "loss": 0.86,
       "step": 19622
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.001224575327288e-06,
-      "loss": 0.685,
+      "loss": 0.7493,
       "step": 19623
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.000282219442225e-06,
-      "loss": 0.8381,
+      "loss": 0.918,
       "step": 19624
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.999339946794737e-06,
-      "loss": 0.7471,
+      "loss": 0.8897,
       "step": 19625
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.998397757397894e-06,
-      "loss": 0.6748,
+      "loss": 0.7395,
       "step": 19626
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.997455651264767e-06,
-      "loss": 0.7043,
+      "loss": 0.8277,
       "step": 19627
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.996513628408423e-06,
-      "loss": 0.7568,
+      "loss": 0.8441,
       "step": 19628
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.995571688841941e-06,
-      "loss": 0.7232,
+      "loss": 0.8135,
       "step": 19629
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.994629832578375e-06,
-      "loss": 0.643,
+      "loss": 0.8158,
       "step": 19630
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.993688059630804e-06,
-      "loss": 0.7212,
+      "loss": 0.7805,
       "step": 19631
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.992746370012289e-06,
-      "loss": 0.7255,
+      "loss": 0.8331,
       "step": 19632
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.991804763735895e-06,
-      "loss": 0.7124,
+      "loss": 0.756,
       "step": 19633
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.990863240814682e-06,
-      "loss": 0.7574,
+      "loss": 0.8504,
       "step": 19634
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.989921801261719e-06,
-      "loss": 0.6615,
+      "loss": 0.7859,
       "step": 19635
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.9889804450900645e-06,
-      "loss": 0.807,
+      "loss": 0.8613,
       "step": 19636
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.988039172312778e-06,
-      "loss": 0.6741,
+      "loss": 0.8002,
       "step": 19637
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.987097982942917e-06,
-      "loss": 0.7948,
+      "loss": 0.8832,
       "step": 19638
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.98615687699354e-06,
-      "loss": 0.7249,
+      "loss": 0.8156,
       "step": 19639
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.985215854477701e-06,
-      "loss": 0.6939,
+      "loss": 0.8292,
       "step": 19640
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.984274915408461e-06,
-      "loss": 0.6753,
+      "loss": 0.7125,
       "step": 19641
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.983334059798877e-06,
-      "loss": 0.697,
+      "loss": 0.8218,
       "step": 19642
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.982393287661988e-06,
-      "loss": 0.7039,
+      "loss": 0.8311,
       "step": 19643
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.981452599010857e-06,
-      "loss": 0.6562,
+      "loss": 0.8345,
       "step": 19644
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.980511993858532e-06,
-      "loss": 0.5292,
+      "loss": 0.719,
       "step": 19645
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.979571472218063e-06,
-      "loss": 0.7427,
+      "loss": 0.8203,
       "step": 19646
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.978631034102496e-06,
-      "loss": 0.6008,
+      "loss": 0.8232,
       "step": 19647
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.9776906795248816e-06,
-      "loss": 0.7308,
+      "loss": 0.813,
       "step": 19648
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.976750408498266e-06,
-      "loss": 0.6196,
+      "loss": 0.7271,
       "step": 19649
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.97581022103569e-06,
-      "loss": 0.8291,
+      "loss": 0.8658,
       "step": 19650
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.974870117150203e-06,
-      "loss": 0.7357,
+      "loss": 0.8332,
       "step": 19651
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.9739300968548435e-06,
-      "loss": 0.7736,
+      "loss": 0.8847,
       "step": 19652
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.9729901601626504e-06,
-      "loss": 0.751,
+      "loss": 0.8398,
       "step": 19653
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.972050307086671e-06,
-      "loss": 0.5703,
+      "loss": 0.746,
       "step": 19654
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.971110537639941e-06,
-      "loss": 0.8294,
+      "loss": 0.8452,
       "step": 19655
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.970170851835499e-06,
-      "loss": 0.7258,
+      "loss": 0.8203,
       "step": 19656
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.969231249686377e-06,
-      "loss": 0.5374,
+      "loss": 0.6835,
       "step": 19657
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.968291731205623e-06,
-      "loss": 0.7294,
+      "loss": 0.8057,
       "step": 19658
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.967352296406255e-06,
-      "loss": 0.7059,
+      "loss": 0.7929,
       "step": 19659
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.966412945301319e-06,
-      "loss": 0.739,
+      "loss": 0.8342,
       "step": 19660
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.965473677903846e-06,
-      "loss": 0.7352,
+      "loss": 0.7756,
       "step": 19661
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.964534494226857e-06,
-      "loss": 0.7625,
+      "loss": 0.8147,
       "step": 19662
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9635953942833915e-06,
-      "loss": 0.7511,
+      "loss": 0.769,
       "step": 19663
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.962656378086477e-06,
-      "loss": 0.71,
+      "loss": 0.8616,
       "step": 19664
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.961717445649138e-06,
-      "loss": 0.7048,
+      "loss": 0.9012,
       "step": 19665
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.960778596984399e-06,
-      "loss": 0.8082,
+      "loss": 0.8425,
       "step": 19666
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.959839832105292e-06,
-      "loss": 0.6769,
+      "loss": 0.7669,
       "step": 19667
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9589011510248375e-06,
-      "loss": 0.6514,
+      "loss": 0.8148,
       "step": 19668
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.957962553756058e-06,
-      "loss": 0.6652,
+      "loss": 0.7582,
       "step": 19669
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.957024040311973e-06,
-      "loss": 0.69,
+      "loss": 0.7638,
       "step": 19670
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.956085610705611e-06,
-      "loss": 0.6625,
+      "loss": 0.8612,
       "step": 19671
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.95514726494998e-06,
-      "loss": 0.7257,
+      "loss": 0.8667,
       "step": 19672
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.954209003058108e-06,
-      "loss": 0.7087,
+      "loss": 0.7911,
       "step": 19673
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9532708250430075e-06,
-      "loss": 0.8502,
+      "loss": 0.926,
       "step": 19674
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.952332730917696e-06,
-      "loss": 0.7739,
+      "loss": 0.8764,
       "step": 19675
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.951394720695183e-06,
-      "loss": 0.6826,
+      "loss": 0.7721,
       "step": 19676
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9504567943884945e-06,
-      "loss": 0.724,
+      "loss": 0.8355,
       "step": 19677
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.949518952010626e-06,
-      "loss": 0.7693,
+      "loss": 0.897,
       "step": 19678
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.948581193574602e-06,
-      "loss": 0.7715,
+      "loss": 0.8247,
       "step": 19679
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.947643519093428e-06,
-      "loss": 0.8036,
+      "loss": 0.7787,
       "step": 19680
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9467059285801135e-06,
-      "loss": 0.6477,
+      "loss": 0.7903,
       "step": 19681
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9457684220476655e-06,
-      "loss": 0.7355,
+      "loss": 0.8663,
       "step": 19682
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.944830999509091e-06,
-      "loss": 0.7833,
+      "loss": 0.7893,
       "step": 19683
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.943893660977395e-06,
-      "loss": 0.6995,
+      "loss": 0.7501,
       "step": 19684
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.942956406465578e-06,
-      "loss": 0.7041,
+      "loss": 0.7973,
       "step": 19685
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.94201923598665e-06,
-      "loss": 0.7368,
+      "loss": 0.8153,
       "step": 19686
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.941082149553611e-06,
-      "loss": 0.8305,
+      "loss": 0.9384,
       "step": 19687
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.94014514717946e-06,
-      "loss": 0.7598,
+      "loss": 0.8663,
       "step": 19688
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.939208228877195e-06,
-      "loss": 0.7059,
+      "loss": 0.736,
       "step": 19689
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.938271394659823e-06,
-      "loss": 0.7719,
+      "loss": 0.8652,
       "step": 19690
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9373346445403285e-06,
-      "loss": 0.6626,
+      "loss": 0.7622,
       "step": 19691
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.936397978531717e-06,
-      "loss": 0.6994,
+      "loss": 0.8224,
       "step": 19692
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.93546139664698e-06,
-      "loss": 0.6856,
+      "loss": 0.8386,
       "step": 19693
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.934524898899112e-06,
-      "loss": 0.6577,
+      "loss": 0.7519,
       "step": 19694
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.933588485301102e-06,
-      "loss": 0.6392,
+      "loss": 0.7204,
       "step": 19695
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.932652155865951e-06,
-      "loss": 0.8645,
+      "loss": 0.9007,
       "step": 19696
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.931715910606636e-06,
-      "loss": 0.7174,
+      "loss": 0.8434,
       "step": 19697
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9307797495361564e-06,
-      "loss": 0.8548,
+      "loss": 0.8817,
       "step": 19698
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9298436726674985e-06,
-      "loss": 0.787,
+      "loss": 0.8841,
       "step": 19699
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.928907680013645e-06,
-      "loss": 0.7645,
+      "loss": 0.7715,
       "step": 19700
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.927971771587584e-06,
-      "loss": 0.6551,
+      "loss": 0.7799,
       "step": 19701
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9270359474022965e-06,
-      "loss": 0.629,
+      "loss": 0.782,
       "step": 19702
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.926100207470777e-06,
-      "loss": 0.6968,
+      "loss": 0.7594,
       "step": 19703
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.925164551805991e-06,
-      "loss": 0.6251,
+      "loss": 0.7752,
       "step": 19704
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.924228980420931e-06,
-      "loss": 0.6539,
+      "loss": 0.7723,
       "step": 19705
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.923293493328574e-06,
-      "loss": 0.7575,
+      "loss": 0.8171,
       "step": 19706
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.922358090541898e-06,
-      "loss": 0.6435,
+      "loss": 0.7857,
       "step": 19707
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.921422772073878e-06,
-      "loss": 0.74,
+      "loss": 0.8314,
       "step": 19708
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.920487537937499e-06,
-      "loss": 0.7268,
+      "loss": 0.8239,
       "step": 19709
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.919552388145721e-06,
-      "loss": 0.699,
+      "loss": 0.8702,
       "step": 19710
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.918617322711532e-06,
-      "loss": 0.4983,
+      "loss": 0.6839,
       "step": 19711
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.917682341647898e-06,
-      "loss": 0.6221,
+      "loss": 0.6958,
       "step": 19712
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9167474449677934e-06,
-      "loss": 0.7728,
+      "loss": 0.8769,
       "step": 19713
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.915812632684183e-06,
-      "loss": 0.6489,
+      "loss": 0.8269,
       "step": 19714
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.914877904810043e-06,
-      "loss": 0.7044,
+      "loss": 0.7733,
       "step": 19715
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.913943261358338e-06,
-      "loss": 0.8385,
+      "loss": 0.8756,
       "step": 19716
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.913008702342036e-06,
-      "loss": 0.7409,
+      "loss": 0.8324,
       "step": 19717
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.912074227774102e-06,
-      "loss": 0.7083,
+      "loss": 0.7963,
       "step": 19718
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9111398376675e-06,
-      "loss": 0.7251,
+      "loss": 0.8647,
       "step": 19719
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.910205532035195e-06,
-      "loss": 0.6458,
+      "loss": 0.8204,
       "step": 19720
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.909271310890142e-06,
-      "loss": 0.6726,
+      "loss": 0.8286,
       "step": 19721
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.908337174245318e-06,
-      "loss": 0.61,
+      "loss": 0.7765,
       "step": 19722
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9074031221136635e-06,
-      "loss": 0.6366,
+      "loss": 0.746,
       "step": 19723
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9064691545081515e-06,
-      "loss": 0.7254,
+      "loss": 0.7845,
       "step": 19724
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.905535271441733e-06,
-      "loss": 0.6575,
+      "loss": 0.7803,
       "step": 19725
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9046014729273665e-06,
-      "loss": 0.7267,
+      "loss": 0.8113,
       "step": 19726
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.903667758978003e-06,
-      "loss": 0.7278,
+      "loss": 0.8068,
       "step": 19727
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9027341296066056e-06,
-      "loss": 0.8084,
+      "loss": 0.8465,
       "step": 19728
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.9018005848261155e-06,
-      "loss": 0.7689,
+      "loss": 0.803,
       "step": 19729
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.900867124649492e-06,
-      "loss": 0.7292,
+      "loss": 0.8153,
       "step": 19730
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.899933749089685e-06,
-      "loss": 0.7455,
+      "loss": 0.8621,
       "step": 19731
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.899000458159642e-06,
-      "loss": 0.7118,
+      "loss": 0.8388,
       "step": 19732
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.898067251872308e-06,
-      "loss": 0.8123,
+      "loss": 0.8056,
       "step": 19733
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.897134130240637e-06,
-      "loss": 0.6694,
+      "loss": 0.7453,
       "step": 19734
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8962010932775715e-06,
-      "loss": 0.7372,
+      "loss": 0.8495,
       "step": 19735
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.895268140996055e-06,
-      "loss": 0.8858,
+      "loss": 0.8953,
       "step": 19736
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.894335273409031e-06,
-      "loss": 0.7819,
+      "loss": 0.8549,
       "step": 19737
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.893402490529443e-06,
-      "loss": 0.725,
+      "loss": 0.7806,
       "step": 19738
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.892469792370232e-06,
-      "loss": 0.7247,
+      "loss": 0.8244,
       "step": 19739
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.891537178944333e-06,
-      "loss": 0.6132,
+      "loss": 0.7061,
       "step": 19740
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8906046502646955e-06,
-      "loss": 0.7577,
+      "loss": 0.8578,
       "step": 19741
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.889672206344244e-06,
-      "loss": 0.7233,
+      "loss": 0.8275,
       "step": 19742
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.888739847195925e-06,
-      "loss": 0.7677,
+      "loss": 0.8879,
       "step": 19743
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.887807572832669e-06,
-      "loss": 0.7181,
+      "loss": 0.7301,
       "step": 19744
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.886875383267411e-06,
-      "loss": 0.7593,
+      "loss": 0.8326,
       "step": 19745
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.88594327851308e-06,
-      "loss": 0.6637,
+      "loss": 0.7982,
       "step": 19746
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.885011258582616e-06,
-      "loss": 0.679,
+      "loss": 0.7498,
       "step": 19747
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.884079323488944e-06,
-      "loss": 0.6768,
+      "loss": 0.7162,
       "step": 19748
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.883147473244994e-06,
-      "loss": 0.727,
+      "loss": 0.8383,
       "step": 19749
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8822157078636934e-06,
-      "loss": 0.9783,
+      "loss": 1.0458,
       "step": 19750
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.881284027357971e-06,
-      "loss": 0.7388,
+      "loss": 0.7843,
       "step": 19751
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.880352431740747e-06,
-      "loss": 0.8134,
+      "loss": 0.9373,
       "step": 19752
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.879420921024955e-06,
-      "loss": 0.5429,
+      "loss": 0.7272,
       "step": 19753
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.878489495223513e-06,
-      "loss": 0.7654,
+      "loss": 0.8263,
       "step": 19754
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.877558154349343e-06,
-      "loss": 0.6979,
+      "loss": 0.8384,
       "step": 19755
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.876626898415369e-06,
-      "loss": 0.7799,
+      "loss": 0.8255,
       "step": 19756
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.875695727434508e-06,
-      "loss": 0.6803,
+      "loss": 0.824,
       "step": 19757
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.874764641419679e-06,
-      "loss": 0.7112,
+      "loss": 0.8103,
       "step": 19758
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8738336403837975e-06,
-      "loss": 0.6045,
+      "loss": 0.8066,
       "step": 19759
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.872902724339785e-06,
-      "loss": 0.7484,
+      "loss": 0.7954,
       "step": 19760
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8719718933005535e-06,
-      "loss": 0.6498,
+      "loss": 0.7999,
       "step": 19761
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.871041147279019e-06,
-      "loss": 0.7077,
+      "loss": 0.8295,
       "step": 19762
     },
     {
@@ -118588,13891 +118588,13891 @@
     {
       "epoch": 0.72,
       "learning_rate": 3.869179910340685e-06,
-      "loss": 0.5753,
+      "loss": 0.7694,
       "step": 19764
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.868249419449704e-06,
-      "loss": 0.8096,
+      "loss": 0.8535,
       "step": 19765
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.867319013628066e-06,
-      "loss": 0.706,
+      "loss": 0.84,
       "step": 19766
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8663886928886765e-06,
-      "loss": 0.7225,
+      "loss": 0.8297,
       "step": 19767
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.865458457244441e-06,
-      "loss": 0.7029,
+      "loss": 0.8437,
       "step": 19768
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.864528306708262e-06,
-      "loss": 0.738,
+      "loss": 0.8366,
       "step": 19769
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.863598241293055e-06,
-      "loss": 0.6653,
+      "loss": 0.7106,
       "step": 19770
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.862668261011708e-06,
-      "loss": 0.5979,
+      "loss": 0.7361,
       "step": 19771
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.861738365877136e-06,
-      "loss": 0.795,
+      "loss": 0.8595,
       "step": 19772
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.860808555902234e-06,
-      "loss": 0.7103,
+      "loss": 0.821,
       "step": 19773
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.859878831099903e-06,
-      "loss": 0.7459,
+      "loss": 0.804,
       "step": 19774
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.858949191483039e-06,
-      "loss": 0.7896,
+      "loss": 0.8693,
       "step": 19775
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.858019637064548e-06,
-      "loss": 0.7443,
+      "loss": 0.89,
       "step": 19776
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.857090167857318e-06,
-      "loss": 0.6906,
+      "loss": 0.8622,
       "step": 19777
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.856160783874241e-06,
-      "loss": 0.7173,
+      "loss": 0.7552,
       "step": 19778
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.855231485128221e-06,
-      "loss": 0.7532,
+      "loss": 0.7965,
       "step": 19779
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.854302271632146e-06,
-      "loss": 0.7293,
+      "loss": 0.7799,
       "step": 19780
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.853373143398907e-06,
-      "loss": 0.7242,
+      "loss": 0.8286,
       "step": 19781
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.852444100441391e-06,
-      "loss": 0.7092,
+      "loss": 0.8004,
       "step": 19782
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.851515142772497e-06,
-      "loss": 0.5794,
+      "loss": 0.6911,
       "step": 19783
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8505862704051015e-06,
-      "loss": 0.7726,
+      "loss": 0.8155,
       "step": 19784
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.849657483352099e-06,
-      "loss": 0.7606,
+      "loss": 0.8598,
       "step": 19785
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.848728781626373e-06,
-      "loss": 0.7338,
+      "loss": 0.7619,
       "step": 19786
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.847800165240807e-06,
-      "loss": 0.7648,
+      "loss": 0.8864,
       "step": 19787
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.846871634208283e-06,
-      "loss": 0.7471,
+      "loss": 0.9331,
       "step": 19788
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.84594318854169e-06,
-      "loss": 0.736,
+      "loss": 0.8475,
       "step": 19789
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.845014828253898e-06,
-      "loss": 0.5687,
+      "loss": 0.7154,
       "step": 19790
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.844086553357795e-06,
-      "loss": 0.6818,
+      "loss": 0.8287,
       "step": 19791
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.843158363866259e-06,
-      "loss": 0.5515,
+      "loss": 0.7053,
       "step": 19792
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.842230259792164e-06,
-      "loss": 0.7447,
+      "loss": 0.8694,
       "step": 19793
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.841302241148383e-06,
-      "loss": 0.6769,
+      "loss": 0.729,
       "step": 19794
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.840374307947804e-06,
-      "loss": 0.5281,
+      "loss": 0.7579,
       "step": 19795
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8394464602032875e-06,
-      "loss": 0.7414,
+      "loss": 0.872,
       "step": 19796
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.838518697927708e-06,
-      "loss": 0.6844,
+      "loss": 0.7815,
       "step": 19797
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.837591021133942e-06,
-      "loss": 0.6623,
+      "loss": 0.7429,
       "step": 19798
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.836663429834858e-06,
-      "loss": 0.6266,
+      "loss": 0.7506,
       "step": 19799
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.835735924043325e-06,
-      "loss": 0.6602,
+      "loss": 0.7203,
       "step": 19800
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8348085037722045e-06,
-      "loss": 0.7544,
+      "loss": 0.8405,
       "step": 19801
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.833881169034378e-06,
-      "loss": 0.8001,
+      "loss": 0.9257,
       "step": 19802
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.832953919842693e-06,
-      "loss": 0.7274,
+      "loss": 0.8085,
       "step": 19803
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.832026756210027e-06,
-      "loss": 0.6522,
+      "loss": 0.7648,
       "step": 19804
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.831099678149238e-06,
-      "loss": 0.6186,
+      "loss": 0.7326,
       "step": 19805
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.830172685673189e-06,
-      "loss": 0.7289,
+      "loss": 0.9558,
       "step": 19806
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.829245778794737e-06,
-      "loss": 0.5456,
+      "loss": 0.6909,
       "step": 19807
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.82831895752675e-06,
-      "loss": 0.6913,
+      "loss": 0.7408,
       "step": 19808
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.827392221882076e-06,
-      "loss": 0.823,
+      "loss": 0.8955,
       "step": 19809
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.826465571873578e-06,
-      "loss": 0.7344,
+      "loss": 0.8505,
       "step": 19810
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.825539007514114e-06,
-      "loss": 0.8287,
+      "loss": 0.856,
       "step": 19811
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.824612528816535e-06,
-      "loss": 0.5925,
+      "loss": 0.7567,
       "step": 19812
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.823686135793691e-06,
-      "loss": 0.6886,
+      "loss": 0.7735,
       "step": 19813
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.822759828458442e-06,
-      "loss": 0.6822,
+      "loss": 0.8185,
       "step": 19814
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.821833606823641e-06,
-      "loss": 0.7406,
+      "loss": 0.8275,
       "step": 19815
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.820907470902125e-06,
-      "loss": 0.768,
+      "loss": 0.8477,
       "step": 19816
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.819981420706756e-06,
-      "loss": 0.6849,
+      "loss": 0.7578,
       "step": 19817
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.819055456250374e-06,
-      "loss": 0.8918,
+      "loss": 0.9179,
       "step": 19818
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.81812957754583e-06,
-      "loss": 0.8101,
+      "loss": 0.8534,
       "step": 19819
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.817203784605963e-06,
-      "loss": 0.6482,
+      "loss": 0.7501,
       "step": 19820
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.816278077443627e-06,
-      "loss": 0.7123,
+      "loss": 0.8964,
       "step": 19821
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.815352456071653e-06,
-      "loss": 0.6418,
+      "loss": 0.7907,
       "step": 19822
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.814426920502893e-06,
-      "loss": 0.7274,
+      "loss": 0.8052,
       "step": 19823
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.813501470750183e-06,
-      "loss": 0.6709,
+      "loss": 0.7868,
       "step": 19824
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8125761068263625e-06,
-      "loss": 0.7134,
+      "loss": 0.8293,
       "step": 19825
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8116508287442665e-06,
-      "loss": 0.7198,
+      "loss": 0.8071,
       "step": 19826
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.810725636516741e-06,
-      "loss": 0.6488,
+      "loss": 0.7794,
       "step": 19827
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.80980053015661e-06,
-      "loss": 0.6826,
+      "loss": 0.7821,
       "step": 19828
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.808875509676717e-06,
-      "loss": 0.7666,
+      "loss": 0.8126,
       "step": 19829
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.807950575089893e-06,
-      "loss": 0.6992,
+      "loss": 0.7828,
       "step": 19830
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8070257264089683e-06,
-      "loss": 0.7674,
+      "loss": 0.8448,
       "step": 19831
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8061009636467716e-06,
-      "loss": 0.7583,
+      "loss": 0.8574,
       "step": 19832
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.80517628681614e-06,
-      "loss": 0.8083,
+      "loss": 0.8403,
       "step": 19833
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8042516959299015e-06,
-      "loss": 0.638,
+      "loss": 0.7448,
       "step": 19834
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.803327191000873e-06,
-      "loss": 0.6682,
+      "loss": 0.7958,
       "step": 19835
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.802402772041891e-06,
-      "loss": 0.6652,
+      "loss": 0.8762,
       "step": 19836
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.801478439065778e-06,
-      "loss": 0.6976,
+      "loss": 0.7888,
       "step": 19837
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.8005541920853563e-06,
-      "loss": 0.6772,
+      "loss": 0.8104,
       "step": 19838
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.799630031113446e-06,
-      "loss": 0.696,
+      "loss": 0.8264,
       "step": 19839
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.798705956162879e-06,
-      "loss": 0.6768,
+      "loss": 0.8049,
       "step": 19840
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.797781967246461e-06,
-      "loss": 0.7778,
+      "loss": 0.8346,
       "step": 19841
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7968580643770224e-06,
-      "loss": 0.686,
+      "loss": 0.8088,
       "step": 19842
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7959342475673776e-06,
-      "loss": 0.6967,
+      "loss": 0.7939,
       "step": 19843
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.795010516830342e-06,
-      "loss": 0.7886,
+      "loss": 0.8882,
       "step": 19844
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7940868721787282e-06,
-      "loss": 0.7165,
+      "loss": 0.8304,
       "step": 19845
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.793163313625358e-06,
-      "loss": 0.7973,
+      "loss": 0.8345,
       "step": 19846
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.79223984118304e-06,
-      "loss": 0.653,
+      "loss": 0.8183,
       "step": 19847
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7913164548645866e-06,
-      "loss": 0.872,
+      "loss": 0.8874,
       "step": 19848
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7903931546828086e-06,
-      "loss": 0.7339,
+      "loss": 0.8749,
       "step": 19849
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7894699406505155e-06,
-      "loss": 0.7484,
+      "loss": 0.825,
       "step": 19850
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7885468127805114e-06,
-      "loss": 0.7966,
+      "loss": 0.861,
       "step": 19851
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.787623771085611e-06,
-      "loss": 0.6851,
+      "loss": 0.7935,
       "step": 19852
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.78670081557862e-06,
-      "loss": 0.6762,
+      "loss": 0.8744,
       "step": 19853
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.785777946272332e-06,
-      "loss": 0.6711,
+      "loss": 0.8158,
       "step": 19854
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7848551631795626e-06,
-      "loss": 0.6555,
+      "loss": 0.7432,
       "step": 19855
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7839324663131093e-06,
-      "loss": 0.7093,
+      "loss": 0.868,
       "step": 19856
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7830098556857743e-06,
-      "loss": 0.7365,
+      "loss": 0.8069,
       "step": 19857
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7820873313103534e-06,
-      "loss": 0.799,
+      "loss": 0.853,
       "step": 19858
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7811648931996515e-06,
-      "loss": 0.5947,
+      "loss": 0.6694,
       "step": 19859
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7802425413664636e-06,
-      "loss": 0.6645,
+      "loss": 0.7705,
       "step": 19860
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.779320275823585e-06,
-      "loss": 0.5927,
+      "loss": 0.7134,
       "step": 19861
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.778398096583812e-06,
-      "loss": 0.7029,
+      "loss": 0.7626,
       "step": 19862
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7774760036599387e-06,
-      "loss": 0.6902,
+      "loss": 0.7875,
       "step": 19863
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7765539970647534e-06,
-      "loss": 0.5201,
+      "loss": 0.6264,
       "step": 19864
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.775632076811054e-06,
-      "loss": 0.6967,
+      "loss": 0.836,
       "step": 19865
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.77471024291163e-06,
-      "loss": 0.7832,
+      "loss": 0.8515,
       "step": 19866
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.773788495379268e-06,
-      "loss": 0.7595,
+      "loss": 0.8489,
       "step": 19867
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7728668342267538e-06,
-      "loss": 0.6297,
+      "loss": 0.7845,
       "step": 19868
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.771945259466884e-06,
-      "loss": 0.7147,
+      "loss": 0.8301,
       "step": 19869
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7710237711124287e-06,
-      "loss": 0.7344,
+      "loss": 0.7809,
       "step": 19870
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7701023691761864e-06,
-      "loss": 0.7433,
+      "loss": 0.7946,
       "step": 19871
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7691810536709374e-06,
-      "loss": 0.6914,
+      "loss": 0.8636,
       "step": 19872
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7682598246094537e-06,
-      "loss": 0.6448,
+      "loss": 0.7986,
       "step": 19873
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7673386820045264e-06,
-      "loss": 0.6722,
+      "loss": 0.8731,
       "step": 19874
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.766417625868932e-06,
-      "loss": 0.6942,
+      "loss": 0.8398,
       "step": 19875
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7654966562154495e-06,
-      "loss": 0.836,
+      "loss": 0.9224,
       "step": 19876
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.764575773056852e-06,
-      "loss": 0.7634,
+      "loss": 0.8129,
       "step": 19877
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7636549764059217e-06,
-      "loss": 0.8394,
+      "loss": 0.9487,
       "step": 19878
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7627342662754297e-06,
-      "loss": 0.7976,
+      "loss": 0.8176,
       "step": 19879
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7618136426781514e-06,
-      "loss": 0.7685,
+      "loss": 0.8457,
       "step": 19880
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7608931056268537e-06,
-      "loss": 0.7198,
+      "loss": 0.8888,
       "step": 19881
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7599726551343195e-06,
-      "loss": 0.7335,
+      "loss": 0.8409,
       "step": 19882
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7590522912133045e-06,
-      "loss": 0.6135,
+      "loss": 0.7057,
       "step": 19883
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7581320138765864e-06,
-      "loss": 0.9244,
+      "loss": 1.0019,
       "step": 19884
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7572118231369314e-06,
-      "loss": 0.7414,
+      "loss": 0.8166,
       "step": 19885
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.756291719007106e-06,
-      "loss": 0.7814,
+      "loss": 0.8055,
       "step": 19886
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.755371701499869e-06,
-      "loss": 0.601,
+      "loss": 0.7146,
       "step": 19887
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.754451770627997e-06,
-      "loss": 0.9053,
+      "loss": 0.9325,
       "step": 19888
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.753531926404239e-06,
-      "loss": 0.8813,
+      "loss": 1.0135,
       "step": 19889
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7526121688413663e-06,
-      "loss": 0.7312,
+      "loss": 0.8674,
       "step": 19890
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7516924979521364e-06,
-      "loss": 0.7521,
+      "loss": 0.8658,
       "step": 19891
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.750772913749308e-06,
-      "loss": 0.7935,
+      "loss": 0.8867,
       "step": 19892
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7498534162456377e-06,
-      "loss": 0.7628,
+      "loss": 0.8054,
       "step": 19893
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.748934005453885e-06,
-      "loss": 0.6192,
+      "loss": 0.7236,
       "step": 19894
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7480146813868024e-06,
-      "loss": 0.7927,
+      "loss": 0.8129,
       "step": 19895
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7470954440571426e-06,
-      "loss": 0.6854,
+      "loss": 0.7573,
       "step": 19896
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7461762934776657e-06,
-      "loss": 0.6353,
+      "loss": 0.7197,
       "step": 19897
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.745257229661119e-06,
-      "loss": 0.713,
+      "loss": 0.8245,
       "step": 19898
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7443382526202553e-06,
-      "loss": 0.6649,
+      "loss": 0.8208,
       "step": 19899
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7434193623678183e-06,
-      "loss": 0.6325,
+      "loss": 0.7938,
       "step": 19900
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7425005589165674e-06,
-      "loss": 0.6776,
+      "loss": 0.7726,
       "step": 19901
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.741581842279236e-06,
-      "loss": 0.6369,
+      "loss": 0.7429,
       "step": 19902
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7406632124685816e-06,
-      "loss": 0.7008,
+      "loss": 0.7868,
       "step": 19903
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7397446694973427e-06,
-      "loss": 0.6758,
+      "loss": 0.784,
       "step": 19904
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.738826213378266e-06,
-      "loss": 0.677,
+      "loss": 0.7518,
       "step": 19905
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.737907844124088e-06,
-      "loss": 0.7815,
+      "loss": 0.8346,
       "step": 19906
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7369895617475605e-06,
-      "loss": 0.547,
+      "loss": 0.7731,
       "step": 19907
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7360713662614103e-06,
-      "loss": 0.6164,
+      "loss": 0.8188,
       "step": 19908
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7351532576783857e-06,
-      "loss": 0.6325,
+      "loss": 0.7308,
       "step": 19909
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.734235236011221e-06,
-      "loss": 0.5633,
+      "loss": 0.7525,
       "step": 19910
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.733317301272653e-06,
-      "loss": 0.8372,
+      "loss": 0.8237,
       "step": 19911
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7323994534754147e-06,
-      "loss": 0.5568,
+      "loss": 0.7228,
       "step": 19912
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7314816926322396e-06,
-      "loss": 0.6213,
+      "loss": 0.634,
       "step": 19913
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.730564018755869e-06,
-      "loss": 0.6665,
+      "loss": 0.7656,
       "step": 19914
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.72964643185902e-06,
-      "loss": 0.7501,
+      "loss": 0.8394,
       "step": 19915
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.728728931954433e-06,
-      "loss": 0.7004,
+      "loss": 0.7774,
       "step": 19916
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7278115190548357e-06,
-      "loss": 0.6587,
+      "loss": 0.7123,
       "step": 19917
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7268941931729542e-06,
-      "loss": 0.6745,
+      "loss": 0.8066,
       "step": 19918
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.725976954321511e-06,
-      "loss": 0.6494,
+      "loss": 0.7109,
       "step": 19919
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7250598025132433e-06,
-      "loss": 0.6495,
+      "loss": 0.753,
       "step": 19920
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.724142737760861e-06,
-      "loss": 0.6439,
+      "loss": 0.7387,
       "step": 19921
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.723225760077097e-06,
-      "loss": 0.6241,
+      "loss": 0.771,
       "step": 19922
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.722308869474671e-06,
-      "loss": 0.6325,
+      "loss": 0.7193,
       "step": 19923
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7213920659663015e-06,
-      "loss": 0.7687,
+      "loss": 0.8475,
       "step": 19924
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7204753495647063e-06,
-      "loss": 0.7252,
+      "loss": 0.7754,
       "step": 19925
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.71955872028261e-06,
-      "loss": 0.5255,
+      "loss": 0.6798,
       "step": 19926
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7186421781327243e-06,
-      "loss": 0.8691,
+      "loss": 0.9203,
       "step": 19927
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7177257231277675e-06,
-      "loss": 0.7614,
+      "loss": 0.8369,
       "step": 19928
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7168093552804516e-06,
-      "loss": 0.7003,
+      "loss": 0.784,
       "step": 19929
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7158930746034927e-06,
-      "loss": 0.6949,
+      "loss": 0.7595,
       "step": 19930
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7149768811096e-06,
-      "loss": 0.6579,
+      "loss": 0.7189,
       "step": 19931
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7140607748114833e-06,
-      "loss": 0.8602,
+      "loss": 0.8989,
       "step": 19932
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7131447557218613e-06,
-      "loss": 0.6599,
+      "loss": 0.8337,
       "step": 19933
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.7122288238534287e-06,
-      "loss": 0.6355,
+      "loss": 0.734,
       "step": 19934
     },
     {
       "epoch": 0.72,
       "learning_rate": 3.711312979218904e-06,
-      "loss": 0.7467,
+      "loss": 0.8654,
       "step": 19935
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7103972218309894e-06,
-      "loss": 0.6887,
+      "loss": 0.7897,
       "step": 19936
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7094815517023897e-06,
-      "loss": 0.5631,
+      "loss": 0.7499,
       "step": 19937
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.708565968845804e-06,
-      "loss": 0.6721,
+      "loss": 0.8536,
       "step": 19938
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7076504732739473e-06,
-      "loss": 0.729,
+      "loss": 0.7999,
       "step": 19939
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7067350649995037e-06,
-      "loss": 0.7838,
+      "loss": 0.9136,
       "step": 19940
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7058197440351872e-06,
-      "loss": 0.5819,
+      "loss": 0.709,
       "step": 19941
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.704904510393691e-06,
-      "loss": 0.7396,
+      "loss": 0.8764,
       "step": 19942
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.703989364087713e-06,
-      "loss": 0.7772,
+      "loss": 0.8903,
       "step": 19943
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7030743051299467e-06,
-      "loss": 0.6948,
+      "loss": 0.8167,
       "step": 19944
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.702159333533093e-06,
-      "loss": 0.7202,
+      "loss": 0.874,
       "step": 19945
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7012444493098442e-06,
-      "loss": 0.6719,
+      "loss": 0.7397,
       "step": 19946
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.7003296524728917e-06,
-      "loss": 0.6324,
+      "loss": 0.7813,
       "step": 19947
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.699414943034926e-06,
-      "loss": 0.752,
+      "loss": 0.902,
       "step": 19948
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.698500321008639e-06,
-      "loss": 0.6886,
+      "loss": 0.7783,
       "step": 19949
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.69758578640672e-06,
-      "loss": 0.6226,
+      "loss": 0.8026,
       "step": 19950
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.696671339241853e-06,
-      "loss": 0.6757,
+      "loss": 0.7425,
       "step": 19951
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6957569795267344e-06,
-      "loss": 0.8006,
+      "loss": 0.9122,
       "step": 19952
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6948427072740357e-06,
-      "loss": 0.6915,
+      "loss": 0.7962,
       "step": 19953
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6939285224964526e-06,
-      "loss": 0.8169,
+      "loss": 0.8633,
       "step": 19954
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6930144252066635e-06,
-      "loss": 0.7948,
+      "loss": 0.8583,
       "step": 19955
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6921004154173523e-06,
-      "loss": 0.5501,
+      "loss": 0.7586,
       "step": 19956
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6911864931411936e-06,
-      "loss": 0.6519,
+      "loss": 0.7244,
       "step": 19957
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.690272658390874e-06,
-      "loss": 0.6998,
+      "loss": 0.8165,
       "step": 19958
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6893589111790707e-06,
-      "loss": 0.754,
+      "loss": 0.7636,
       "step": 19959
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6884452515184575e-06,
-      "loss": 0.7924,
+      "loss": 0.8847,
       "step": 19960
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.687531679421712e-06,
-      "loss": 0.6524,
+      "loss": 0.7514,
       "step": 19961
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.686618194901508e-06,
-      "loss": 0.7599,
+      "loss": 0.7884,
       "step": 19962
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6857047979705164e-06,
-      "loss": 0.6431,
+      "loss": 0.7301,
       "step": 19963
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.684791488641414e-06,
-      "loss": 0.6781,
+      "loss": 0.7568,
       "step": 19964
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.68387826692687e-06,
-      "loss": 0.6859,
+      "loss": 0.8264,
       "step": 19965
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6829651328395544e-06,
-      "loss": 0.8286,
+      "loss": 0.8829,
       "step": 19966
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.682052086392134e-06,
-      "loss": 0.9252,
+      "loss": 0.8788,
       "step": 19967
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6811391275972776e-06,
-      "loss": 0.7542,
+      "loss": 0.7325,
       "step": 19968
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6802262564676506e-06,
-      "loss": 0.6881,
+      "loss": 0.868,
       "step": 19969
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.679313473015914e-06,
-      "loss": 0.6559,
+      "loss": 0.7853,
       "step": 19970
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6784007772547413e-06,
-      "loss": 0.601,
+      "loss": 0.7374,
       "step": 19971
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6774881691967824e-06,
-      "loss": 0.5677,
+      "loss": 0.6723,
       "step": 19972
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6765756488547078e-06,
-      "loss": 0.6964,
+      "loss": 0.7788,
       "step": 19973
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6756632162411755e-06,
-      "loss": 0.6138,
+      "loss": 0.7645,
       "step": 19974
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.674750871368842e-06,
-      "loss": 0.8722,
+      "loss": 0.9149,
       "step": 19975
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.673838614250362e-06,
-      "loss": 0.8275,
+      "loss": 0.8309,
       "step": 19976
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6729264448983994e-06,
-      "loss": 0.7709,
+      "loss": 0.8312,
       "step": 19977
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6720143633256057e-06,
-      "loss": 0.7932,
+      "loss": 0.8361,
       "step": 19978
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6711023695446336e-06,
-      "loss": 0.7741,
+      "loss": 0.812,
       "step": 19979
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.670190463568134e-06,
-      "loss": 0.6625,
+      "loss": 0.7951,
       "step": 19980
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.669278645408768e-06,
-      "loss": 0.6756,
+      "loss": 0.8454,
       "step": 19981
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.668366915079171e-06,
-      "loss": 0.6954,
+      "loss": 0.7915,
       "step": 19982
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.667455272592003e-06,
-      "loss": 0.8224,
+      "loss": 0.8212,
       "step": 19983
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6665437179599094e-06,
-      "loss": 0.8146,
+      "loss": 0.8431,
       "step": 19984
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6656322511955358e-06,
-      "loss": 0.5328,
+      "loss": 0.785,
       "step": 19985
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6647208723115257e-06,
-      "loss": 0.6842,
+      "loss": 0.7771,
       "step": 19986
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6638095813205265e-06,
-      "loss": 0.7299,
+      "loss": 0.8141,
       "step": 19987
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.662898378235179e-06,
-      "loss": 0.7308,
+      "loss": 0.8069,
       "step": 19988
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.661987263068122e-06,
-      "loss": 0.6755,
+      "loss": 0.7999,
       "step": 19989
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6610762358320018e-06,
-      "loss": 0.743,
+      "loss": 0.8541,
       "step": 19990
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6601652965394562e-06,
-      "loss": 0.7761,
+      "loss": 0.8937,
       "step": 19991
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.659254445203122e-06,
-      "loss": 0.7342,
+      "loss": 0.8387,
       "step": 19992
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6583436818356354e-06,
-      "loss": 0.6273,
+      "loss": 0.8409,
       "step": 19993
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6574330064496335e-06,
-      "loss": 0.7138,
+      "loss": 0.8006,
       "step": 19994
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6565224190577454e-06,
-      "loss": 0.5605,
+      "loss": 0.6785,
       "step": 19995
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6556119196726127e-06,
-      "loss": 0.6539,
+      "loss": 0.7221,
       "step": 19996
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6547015083068617e-06,
-      "loss": 0.738,
+      "loss": 0.8962,
       "step": 19997
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6537911849731255e-06,
-      "loss": 0.7244,
+      "loss": 0.8177,
       "step": 19998
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.652880949684029e-06,
-      "loss": 0.6349,
+      "loss": 0.6946,
       "step": 19999
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6519708024522107e-06,
-      "loss": 0.5961,
+      "loss": 0.8038,
       "step": 20000
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6510607432902846e-06,
-      "loss": 0.6833,
+      "loss": 0.8431,
       "step": 20001
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6501507722108853e-06,
-      "loss": 0.6959,
+      "loss": 0.7882,
       "step": 20002
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6492408892266352e-06,
-      "loss": 0.8111,
+      "loss": 0.8522,
       "step": 20003
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6483310943501573e-06,
-      "loss": 0.8759,
+      "loss": 0.9558,
       "step": 20004
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.64742138759407e-06,
-      "loss": 0.6406,
+      "loss": 0.7445,
       "step": 20005
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.646511768971004e-06,
-      "loss": 0.8747,
+      "loss": 0.8672,
       "step": 20006
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.645602238493571e-06,
-      "loss": 0.6913,
+      "loss": 0.784,
       "step": 20007
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.644692796174386e-06,
-      "loss": 0.6873,
+      "loss": 0.8057,
       "step": 20008
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6437834420260764e-06,
-      "loss": 0.8146,
+      "loss": 0.8854,
       "step": 20009
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6428741760612517e-06,
-      "loss": 0.5895,
+      "loss": 0.7349,
       "step": 20010
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.64196499829253e-06,
-      "loss": 0.7231,
+      "loss": 0.7879,
       "step": 20011
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6410559087325182e-06,
-      "loss": 0.6628,
+      "loss": 0.7804,
       "step": 20012
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.640146907393841e-06,
-      "loss": 0.7126,
+      "loss": 0.8146,
       "step": 20013
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6392379942890955e-06,
-      "loss": 0.5568,
+      "loss": 0.6898,
       "step": 20014
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6383291694309007e-06,
-      "loss": 0.6921,
+      "loss": 0.8141,
       "step": 20015
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.637420432831863e-06,
-      "loss": 0.811,
+      "loss": 0.8361,
       "step": 20016
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6365117845045896e-06,
-      "loss": 0.7967,
+      "loss": 0.8896,
       "step": 20017
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6356032244616836e-06,
-      "loss": 0.6394,
+      "loss": 0.7396,
       "step": 20018
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6346947527157582e-06,
-      "loss": 0.6642,
+      "loss": 0.8032,
       "step": 20019
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6337863692794064e-06,
-      "loss": 0.7974,
+      "loss": 0.888,
       "step": 20020
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.632878074165239e-06,
-      "loss": 0.7003,
+      "loss": 0.7643,
       "step": 20021
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.631969867385855e-06,
-      "loss": 0.7034,
+      "loss": 0.7454,
       "step": 20022
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.631061748953855e-06,
-      "loss": 0.5725,
+      "loss": 0.777,
       "step": 20023
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6301537188818316e-06,
-      "loss": 0.7275,
+      "loss": 0.8538,
       "step": 20024
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.629245777182392e-06,
-      "loss": 0.789,
+      "loss": 0.8519,
       "step": 20025
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6283379238681326e-06,
-      "loss": 0.7625,
+      "loss": 0.848,
       "step": 20026
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6274301589516357e-06,
-      "loss": 0.798,
+      "loss": 0.863,
       "step": 20027
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.626522482445508e-06,
-      "loss": 0.7786,
+      "loss": 0.8294,
       "step": 20028
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.625614894362339e-06,
-      "loss": 0.6424,
+      "loss": 0.7977,
       "step": 20029
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.624707394714718e-06,
-      "loss": 0.6991,
+      "loss": 0.8456,
       "step": 20030
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6237999835152336e-06,
-      "loss": 0.8607,
+      "loss": 0.8738,
       "step": 20031
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6228926607764846e-06,
-      "loss": 0.7507,
+      "loss": 0.803,
       "step": 20032
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6219854265110456e-06,
-      "loss": 0.7603,
+      "loss": 0.8551,
       "step": 20033
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6210782807315124e-06,
-      "loss": 0.6391,
+      "loss": 0.8046,
       "step": 20034
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.620171223450468e-06,
-      "loss": 0.7034,
+      "loss": 0.824,
       "step": 20035
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6192642546804956e-06,
-      "loss": 0.6781,
+      "loss": 0.7702,
       "step": 20036
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.618357374434176e-06,
-      "loss": 0.6612,
+      "loss": 0.7368,
       "step": 20037
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6174505827241003e-06,
-      "loss": 0.5899,
+      "loss": 0.6987,
       "step": 20038
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.616543879562834e-06,
-      "loss": 0.779,
+      "loss": 0.8274,
       "step": 20039
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.615637264962969e-06,
-      "loss": 0.6434,
+      "loss": 0.7856,
       "step": 20040
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.614730738937078e-06,
-      "loss": 0.751,
+      "loss": 0.8601,
       "step": 20041
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6138243014977394e-06,
-      "loss": 0.6623,
+      "loss": 0.6767,
       "step": 20042
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.612917952657523e-06,
-      "loss": 0.6984,
+      "loss": 0.8551,
       "step": 20043
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.612011692429013e-06,
-      "loss": 0.7567,
+      "loss": 0.8098,
       "step": 20044
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.611105520824779e-06,
-      "loss": 0.6857,
+      "loss": 0.8104,
       "step": 20045
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6101994378573855e-06,
-      "loss": 0.6879,
+      "loss": 0.8243,
       "step": 20046
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.609293443539412e-06,
-      "loss": 0.7508,
+      "loss": 0.8875,
       "step": 20047
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6083875378834253e-06,
-      "loss": 0.7291,
+      "loss": 0.7389,
       "step": 20048
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6074817209019918e-06,
-      "loss": 0.6504,
+      "loss": 0.7841,
       "step": 20049
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.606575992607677e-06,
-      "loss": 0.8325,
+      "loss": 0.8624,
       "step": 20050
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6056703530130566e-06,
-      "loss": 0.6837,
+      "loss": 0.7934,
       "step": 20051
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6047648021306805e-06,
-      "loss": 0.8795,
+      "loss": 0.8992,
       "step": 20052
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6038593399731213e-06,
-      "loss": 0.6405,
+      "loss": 0.8155,
       "step": 20053
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.60295396655294e-06,
-      "loss": 0.5367,
+      "loss": 0.7461,
       "step": 20054
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.602048681882696e-06,
-      "loss": 0.6699,
+      "loss": 0.8278,
       "step": 20055
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.601143485974946e-06,
-      "loss": 0.6751,
+      "loss": 0.8341,
       "step": 20056
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6002383788422548e-06,
-      "loss": 0.7944,
+      "loss": 0.9192,
       "step": 20057
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5993333604971748e-06,
-      "loss": 0.7754,
+      "loss": 0.8032,
       "step": 20058
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5984284309522642e-06,
-      "loss": 0.8064,
+      "loss": 0.8336,
       "step": 20059
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5975235902200766e-06,
-      "loss": 0.6857,
+      "loss": 0.8796,
       "step": 20060
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.596618838313165e-06,
-      "loss": 0.658,
+      "loss": 0.7197,
       "step": 20061
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5957141752440785e-06,
-      "loss": 0.796,
+      "loss": 0.8996,
       "step": 20062
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.594809601025374e-06,
-      "loss": 0.5673,
+      "loss": 0.6414,
       "step": 20063
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5939051156696025e-06,
-      "loss": 0.5685,
+      "loss": 0.7071,
       "step": 20064
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.593000719189301e-06,
-      "loss": 0.7114,
+      "loss": 0.7981,
       "step": 20065
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.592096411597028e-06,
-      "loss": 0.7546,
+      "loss": 0.849,
       "step": 20066
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5911921929053263e-06,
-      "loss": 0.7876,
+      "loss": 0.8613,
       "step": 20067
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5902880631267387e-06,
-      "loss": 0.8031,
+      "loss": 0.8331,
       "step": 20068
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5893840222738074e-06,
-      "loss": 0.7026,
+      "loss": 0.7195,
       "step": 20069
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5884800703590818e-06,
-      "loss": 0.5678,
+      "loss": 0.6977,
       "step": 20070
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.587576207395097e-06,
-      "loss": 0.6743,
+      "loss": 0.7766,
       "step": 20071
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.586672433394395e-06,
-      "loss": 0.7058,
+      "loss": 0.7626,
       "step": 20072
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5857687483695124e-06,
-      "loss": 0.8024,
+      "loss": 0.8801,
       "step": 20073
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.584865152332989e-06,
-      "loss": 0.7071,
+      "loss": 0.8542,
       "step": 20074
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.583961645297356e-06,
-      "loss": 0.756,
+      "loss": 0.8452,
       "step": 20075
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.583058227275156e-06,
-      "loss": 0.8392,
+      "loss": 0.8325,
       "step": 20076
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5821548982789176e-06,
-      "loss": 0.7981,
+      "loss": 0.7497,
       "step": 20077
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5812516583211755e-06,
-      "loss": 0.768,
+      "loss": 0.8354,
       "step": 20078
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.580348507414455e-06,
-      "loss": 0.6548,
+      "loss": 0.8055,
       "step": 20079
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.579445445571298e-06,
-      "loss": 0.6399,
+      "loss": 0.7702,
       "step": 20080
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5785424728042197e-06,
-      "loss": 0.7927,
+      "loss": 0.8645,
       "step": 20081
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5776395891257554e-06,
-      "loss": 0.702,
+      "loss": 0.8424,
       "step": 20082
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.576736794548434e-06,
-      "loss": 0.7989,
+      "loss": 0.8647,
       "step": 20083
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.57583408908477e-06,
-      "loss": 0.7806,
+      "loss": 0.817,
       "step": 20084
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.574931472747296e-06,
-      "loss": 0.6313,
+      "loss": 0.7581,
       "step": 20085
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.574028945548532e-06,
-      "loss": 0.7867,
+      "loss": 0.9014,
       "step": 20086
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.573126507500999e-06,
-      "loss": 0.6239,
+      "loss": 0.8209,
       "step": 20087
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5722241586172147e-06,
-      "loss": 0.7452,
+      "loss": 0.7317,
       "step": 20088
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5713218989097033e-06,
-      "loss": 0.6744,
+      "loss": 0.7513,
       "step": 20089
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.570419728390978e-06,
-      "loss": 0.6485,
+      "loss": 0.7685,
       "step": 20090
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.569517647073558e-06,
-      "loss": 0.6004,
+      "loss": 0.7565,
       "step": 20091
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5686156549699534e-06,
-      "loss": 0.6822,
+      "loss": 0.8166,
       "step": 20092
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5677137520926883e-06,
-      "loss": 0.7336,
+      "loss": 0.8733,
       "step": 20093
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5668119384542623e-06,
-      "loss": 0.7059,
+      "loss": 0.8636,
       "step": 20094
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5659102140671964e-06,
-      "loss": 0.7204,
+      "loss": 0.7593,
       "step": 20095
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5650085789439968e-06,
-      "loss": 0.7127,
+      "loss": 0.7627,
       "step": 20096
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.564107033097174e-06,
-      "loss": 0.6872,
+      "loss": 0.8385,
       "step": 20097
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5632055765392304e-06,
-      "loss": 0.6137,
+      "loss": 0.7151,
       "step": 20098
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5623042092826842e-06,
-      "loss": 0.7051,
+      "loss": 0.7351,
       "step": 20099
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5614029313400266e-06,
-      "loss": 0.7022,
+      "loss": 0.8341,
       "step": 20100
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.560501742723772e-06,
-      "loss": 0.6823,
+      "loss": 0.7142,
       "step": 20101
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5596006434464182e-06,
-      "loss": 0.7425,
+      "loss": 0.843,
       "step": 20102
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.558699633520468e-06,
-      "loss": 0.6227,
+      "loss": 0.6972,
       "step": 20103
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.557798712958421e-06,
-      "loss": 0.7807,
+      "loss": 0.9212,
       "step": 20104
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5568978817727783e-06,
-      "loss": 0.6836,
+      "loss": 0.7717,
       "step": 20105
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.555997139976034e-06,
-      "loss": 0.6751,
+      "loss": 0.8823,
       "step": 20106
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5550964875806836e-06,
-      "loss": 0.6186,
+      "loss": 0.6947,
       "step": 20107
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.55419592459923e-06,
-      "loss": 0.6784,
+      "loss": 0.7644,
       "step": 20108
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.553295451044161e-06,
-      "loss": 0.7704,
+      "loss": 0.8194,
       "step": 20109
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5523950669279718e-06,
-      "loss": 0.5777,
+      "loss": 0.7382,
       "step": 20110
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.551494772263149e-06,
-      "loss": 0.7565,
+      "loss": 0.7768,
       "step": 20111
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5505945670621945e-06,
-      "loss": 0.7444,
+      "loss": 0.7987,
       "step": 20112
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5496944513375833e-06,
-      "loss": 0.8384,
+      "loss": 0.8591,
       "step": 20113
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5487944251018115e-06,
-      "loss": 0.6058,
+      "loss": 0.6938,
       "step": 20114
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.547894488367365e-06,
-      "loss": 0.6809,
+      "loss": 0.8118,
       "step": 20115
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.546994641146728e-06,
-      "loss": 0.6552,
+      "loss": 0.8115,
       "step": 20116
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.546094883452382e-06,
-      "loss": 0.8422,
+      "loss": 0.8624,
       "step": 20117
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5451952152968195e-06,
-      "loss": 0.7808,
+      "loss": 0.803,
       "step": 20118
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5442956366925074e-06,
-      "loss": 0.6988,
+      "loss": 0.7316,
       "step": 20119
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5433961476519383e-06,
-      "loss": 0.687,
+      "loss": 0.7743,
       "step": 20120
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.542496748187586e-06,
-      "loss": 0.6749,
+      "loss": 0.7989,
       "step": 20121
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.54159743831193e-06,
-      "loss": 0.722,
+      "loss": 0.8066,
       "step": 20122
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.540698218037446e-06,
-      "loss": 0.6979,
+      "loss": 0.7239,
       "step": 20123
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5397990873766063e-06,
-      "loss": 0.7633,
+      "loss": 0.7801,
       "step": 20124
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5389000463418956e-06,
-      "loss": 0.5332,
+      "loss": 0.6748,
       "step": 20125
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.538001094945772e-06,
-      "loss": 0.6834,
+      "loss": 0.7558,
       "step": 20126
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5371022332007188e-06,
-      "loss": 0.7446,
+      "loss": 0.8189,
       "step": 20127
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.536203461119202e-06,
-      "loss": 0.6573,
+      "loss": 0.8387,
       "step": 20128
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5353047787136918e-06,
-      "loss": 0.7139,
+      "loss": 0.7833,
       "step": 20129
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5344061859966527e-06,
-      "loss": 0.5263,
+      "loss": 0.7698,
       "step": 20130
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5335076829805603e-06,
-      "loss": 0.7142,
+      "loss": 0.7534,
       "step": 20131
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.532609269677867e-06,
-      "loss": 0.7175,
+      "loss": 0.8296,
       "step": 20132
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5317109461010477e-06,
-      "loss": 0.6967,
+      "loss": 0.8055,
       "step": 20133
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5308127122625625e-06,
-      "loss": 0.5949,
+      "loss": 0.8144,
       "step": 20134
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5299145681748714e-06,
-      "loss": 0.6492,
+      "loss": 0.7678,
       "step": 20135
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.529016513850433e-06,
-      "loss": 0.6522,
+      "loss": 0.7846,
       "step": 20136
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5281185493017157e-06,
-      "loss": 0.7074,
+      "loss": 0.8026,
       "step": 20137
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.527220674541164e-06,
-      "loss": 0.7406,
+      "loss": 0.8632,
       "step": 20138
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5263228895812462e-06,
-      "loss": 0.6283,
+      "loss": 0.8538,
       "step": 20139
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.525425194434412e-06,
-      "loss": 0.6503,
+      "loss": 0.7779,
       "step": 20140
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.524527589113118e-06,
-      "loss": 0.6613,
+      "loss": 0.8101,
       "step": 20141
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5236300736298157e-06,
-      "loss": 0.6022,
+      "loss": 0.7141,
       "step": 20142
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.522732647996955e-06,
-      "loss": 0.7022,
+      "loss": 0.8053,
       "step": 20143
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5218353122269943e-06,
-      "loss": 0.6963,
+      "loss": 0.7927,
       "step": 20144
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.520938066332371e-06,
-      "loss": 0.6898,
+      "loss": 0.7855,
       "step": 20145
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.520040910325543e-06,
-      "loss": 0.9103,
+      "loss": 0.903,
       "step": 20146
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.519143844218954e-06,
-      "loss": 0.6269,
+      "loss": 0.7619,
       "step": 20147
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5182468680250483e-06,
-      "loss": 0.6867,
+      "loss": 0.8508,
       "step": 20148
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.517349981756267e-06,
-      "loss": 0.7165,
+      "loss": 0.8116,
       "step": 20149
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.516453185425065e-06,
-      "loss": 0.7776,
+      "loss": 0.8398,
       "step": 20150
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.515556479043869e-06,
-      "loss": 0.5816,
+      "loss": 0.7505,
       "step": 20151
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5146598626251292e-06,
-      "loss": 0.7796,
+      "loss": 0.8356,
       "step": 20152
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.513763336181283e-06,
-      "loss": 0.7744,
+      "loss": 0.809,
       "step": 20153
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.512866899724768e-06,
-      "loss": 0.6509,
+      "loss": 0.7815,
       "step": 20154
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.511970553268017e-06,
-      "loss": 0.6735,
+      "loss": 0.8187,
       "step": 20155
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5110742968234724e-06,
-      "loss": 0.8358,
+      "loss": 0.8842,
       "step": 20156
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5101781304035654e-06,
-      "loss": 0.6498,
+      "loss": 0.7454,
       "step": 20157
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5092820540207294e-06,
-      "loss": 0.749,
+      "loss": 0.8652,
       "step": 20158
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.508386067687396e-06,
-      "loss": 0.7043,
+      "loss": 0.7808,
       "step": 20159
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5074901714159947e-06,
-      "loss": 0.724,
+      "loss": 0.795,
       "step": 20160
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.506594365218956e-06,
-      "loss": 0.6728,
+      "loss": 0.8188,
       "step": 20161
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5056986491087042e-06,
-      "loss": 0.7923,
+      "loss": 0.9229,
       "step": 20162
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.504803023097676e-06,
-      "loss": 0.7561,
+      "loss": 0.7993,
       "step": 20163
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.503907487198284e-06,
-      "loss": 0.7123,
+      "loss": 0.8086,
       "step": 20164
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.503012041422962e-06,
-      "loss": 0.7298,
+      "loss": 0.7979,
       "step": 20165
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5021166857841303e-06,
-      "loss": 0.577,
+      "loss": 0.6706,
       "step": 20166
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5012214202942095e-06,
-      "loss": 0.8007,
+      "loss": 0.8976,
       "step": 20167
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.5003262449656185e-06,
-      "loss": 0.7162,
+      "loss": 0.86,
       "step": 20168
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.499431159810781e-06,
-      "loss": 0.6799,
+      "loss": 0.7576,
       "step": 20169
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4985361648421146e-06,
-      "loss": 0.7034,
+      "loss": 0.7589,
       "step": 20170
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4976412600720333e-06,
-      "loss": 0.8242,
+      "loss": 0.9184,
       "step": 20171
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4967464455129553e-06,
-      "loss": 0.6022,
+      "loss": 0.7892,
       "step": 20172
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4958517211772924e-06,
-      "loss": 0.7484,
+      "loss": 0.7848,
       "step": 20173
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.494957087077455e-06,
-      "loss": 0.6591,
+      "loss": 0.7991,
       "step": 20174
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4940625432258625e-06,
-      "loss": 0.7883,
+      "loss": 0.9279,
       "step": 20175
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.493168089634922e-06,
-      "loss": 0.5763,
+      "loss": 0.8358,
       "step": 20176
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4922737263170412e-06,
-      "loss": 0.7177,
+      "loss": 0.8161,
       "step": 20177
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4913794532846302e-06,
-      "loss": 0.6693,
+      "loss": 0.7438,
       "step": 20178
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4904852705500937e-06,
-      "loss": 0.7411,
+      "loss": 0.7584,
       "step": 20179
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4895911781258386e-06,
-      "loss": 0.6387,
+      "loss": 0.7458,
       "step": 20180
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.488697176024266e-06,
-      "loss": 0.7452,
+      "loss": 0.8506,
       "step": 20181
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4878032642577884e-06,
-      "loss": 0.6373,
+      "loss": 0.7752,
       "step": 20182
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.486909442838794e-06,
-      "loss": 0.6902,
+      "loss": 0.8053,
       "step": 20183
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4860157117796935e-06,
-      "loss": 0.7445,
+      "loss": 0.8285,
       "step": 20184
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4851220710928835e-06,
-      "loss": 0.6852,
+      "loss": 0.7821,
       "step": 20185
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.48422852079076e-06,
-      "loss": 0.7417,
+      "loss": 0.8931,
       "step": 20186
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4833350608857185e-06,
-      "loss": 0.693,
+      "loss": 0.7423,
       "step": 20187
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4824416913901604e-06,
-      "loss": 0.8512,
+      "loss": 0.8617,
       "step": 20188
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.481548412316477e-06,
-      "loss": 0.7966,
+      "loss": 0.8752,
       "step": 20189
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.480655223677061e-06,
-      "loss": 0.7659,
+      "loss": 0.926,
       "step": 20190
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4797621254842994e-06,
-      "loss": 0.755,
+      "loss": 0.8579,
       "step": 20191
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4788691177505952e-06,
-      "loss": 0.7282,
+      "loss": 0.8524,
       "step": 20192
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4779762004883234e-06,
-      "loss": 0.6561,
+      "loss": 0.7764,
       "step": 20193
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.477083373709881e-06,
-      "loss": 0.6499,
+      "loss": 0.7553,
       "step": 20194
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.476190637427651e-06,
-      "loss": 0.644,
+      "loss": 0.7701,
       "step": 20195
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.475297991654021e-06,
-      "loss": 0.7163,
+      "loss": 0.7858,
       "step": 20196
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4744054364013737e-06,
-      "loss": 0.6686,
+      "loss": 0.7955,
       "step": 20197
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4735129716820923e-06,
-      "loss": 0.7299,
+      "loss": 0.7603,
       "step": 20198
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.472620597508559e-06,
-      "loss": 0.6101,
+      "loss": 0.765,
       "step": 20199
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.47172831389315e-06,
-      "loss": 0.63,
+      "loss": 0.7762,
       "step": 20200
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4708361208482523e-06,
-      "loss": 0.5753,
+      "loss": 0.7601,
       "step": 20201
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.469944018386239e-06,
-      "loss": 0.6265,
+      "loss": 0.7607,
       "step": 20202
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4690520065194876e-06,
-      "loss": 0.7176,
+      "loss": 0.8071,
       "step": 20203
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4681600852603737e-06,
-      "loss": 0.8044,
+      "loss": 0.8562,
       "step": 20204
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.46726825462127e-06,
-      "loss": 0.6761,
+      "loss": 0.7837,
       "step": 20205
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4663765146145488e-06,
-      "loss": 0.6095,
+      "loss": 0.7078,
       "step": 20206
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4654848652525862e-06,
-      "loss": 0.6072,
+      "loss": 0.7477,
       "step": 20207
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.464593306547749e-06,
-      "loss": 0.791,
+      "loss": 0.8601,
       "step": 20208
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4637018385124076e-06,
-      "loss": 0.7577,
+      "loss": 0.8463,
       "step": 20209
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.4628104611589265e-06,
-      "loss": 0.7669,
+      "loss": 0.8099,
       "step": 20210
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.461919174499682e-06,
-      "loss": 0.7023,
+      "loss": 0.7388,
       "step": 20211
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4610279785470258e-06,
-      "loss": 0.7423,
+      "loss": 0.7704,
       "step": 20212
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4601368733133323e-06,
-      "loss": 0.8046,
+      "loss": 0.8162,
       "step": 20213
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.45924585881096e-06,
-      "loss": 0.7256,
+      "loss": 0.8074,
       "step": 20214
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4583549350522716e-06,
-      "loss": 0.626,
+      "loss": 0.7822,
       "step": 20215
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.457464102049628e-06,
-      "loss": 0.7349,
+      "loss": 0.8592,
       "step": 20216
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.456573359815386e-06,
-      "loss": 0.6519,
+      "loss": 0.7364,
       "step": 20217
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.455682708361906e-06,
-      "loss": 0.6983,
+      "loss": 0.775,
       "step": 20218
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.454792147701539e-06,
-      "loss": 0.7176,
+      "loss": 0.7766,
       "step": 20219
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4539016778466485e-06,
-      "loss": 0.7143,
+      "loss": 0.7778,
       "step": 20220
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4530112988095843e-06,
-      "loss": 0.6655,
+      "loss": 0.7437,
       "step": 20221
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4521210106026992e-06,
-      "loss": 0.6877,
+      "loss": 0.8434,
       "step": 20222
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.451230813238341e-06,
-      "loss": 0.6716,
+      "loss": 0.7913,
       "step": 20223
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.450340706728871e-06,
-      "loss": 0.8005,
+      "loss": 0.8425,
       "step": 20224
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4494506910866245e-06,
-      "loss": 0.7813,
+      "loss": 0.8267,
       "step": 20225
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4485607663239584e-06,
-      "loss": 0.6416,
+      "loss": 0.7291,
       "step": 20226
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4476709324532165e-06,
-      "loss": 0.8484,
+      "loss": 0.8786,
       "step": 20227
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.446781189486744e-06,
-      "loss": 0.7389,
+      "loss": 0.8745,
       "step": 20228
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.445891537436882e-06,
-      "loss": 0.6826,
+      "loss": 0.8031,
       "step": 20229
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4450019763159827e-06,
-      "loss": 0.688,
+      "loss": 0.887,
       "step": 20230
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4441125061363723e-06,
-      "loss": 0.8128,
+      "loss": 0.8797,
       "step": 20231
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4432231269104043e-06,
-      "loss": 0.6574,
+      "loss": 0.8016,
       "step": 20232
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4423338386504114e-06,
-      "loss": 0.7361,
+      "loss": 0.8307,
       "step": 20233
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.441444641368733e-06,
-      "loss": 0.8453,
+      "loss": 0.9157,
       "step": 20234
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4405555350777008e-06,
-      "loss": 0.628,
+      "loss": 0.7366,
       "step": 20235
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4396665197896584e-06,
-      "loss": 0.8166,
+      "loss": 0.8032,
       "step": 20236
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4387775955169376e-06,
-      "loss": 0.6885,
+      "loss": 0.8321,
       "step": 20237
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4378887622718627e-06,
-      "loss": 0.6518,
+      "loss": 0.7078,
       "step": 20238
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.437000020066773e-06,
-      "loss": 0.7327,
+      "loss": 0.8517,
       "step": 20239
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.436111368913997e-06,
-      "loss": 0.608,
+      "loss": 0.7468,
       "step": 20240
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4352228088258634e-06,
-      "loss": 0.7461,
+      "loss": 0.8437,
       "step": 20241
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4343343398146945e-06,
-      "loss": 0.7596,
+      "loss": 0.7963,
       "step": 20242
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4334459618928284e-06,
-      "loss": 0.6892,
+      "loss": 0.8635,
       "step": 20243
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4325576750725766e-06,
-      "loss": 0.6994,
+      "loss": 0.7098,
       "step": 20244
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.431669479366272e-06,
-      "loss": 0.7086,
+      "loss": 0.8497,
       "step": 20245
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.430781374786234e-06,
-      "loss": 0.9435,
+      "loss": 0.9216,
       "step": 20246
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.429893361344784e-06,
-      "loss": 0.7185,
+      "loss": 0.7912,
       "step": 20247
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.429005439054238e-06,
-      "loss": 0.7051,
+      "loss": 0.8493,
       "step": 20248
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4281176079269262e-06,
-      "loss": 0.6696,
+      "loss": 0.8157,
       "step": 20249
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4272298679751514e-06,
-      "loss": 0.7192,
+      "loss": 0.8126,
       "step": 20250
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4263422192112395e-06,
-      "loss": 0.7435,
+      "loss": 0.7963,
       "step": 20251
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4254546616475027e-06,
-      "loss": 0.7167,
+      "loss": 0.7711,
       "step": 20252
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4245671952962543e-06,
-      "loss": 0.5973,
+      "loss": 0.7305,
       "step": 20253
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.423679820169804e-06,
-      "loss": 0.7203,
+      "loss": 0.8514,
       "step": 20254
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4227925362804692e-06,
-      "loss": 0.7999,
+      "loss": 0.8629,
       "step": 20255
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.421905343640558e-06,
-      "loss": 0.5549,
+      "loss": 0.7307,
       "step": 20256
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4210182422623717e-06,
-      "loss": 0.6706,
+      "loss": 0.8249,
       "step": 20257
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.420131232158226e-06,
-      "loss": 0.7117,
+      "loss": 0.8141,
       "step": 20258
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.419244313340424e-06,
-      "loss": 0.713,
+      "loss": 0.8154,
       "step": 20259
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4183574858212696e-06,
-      "loss": 0.7839,
+      "loss": 0.8354,
       "step": 20260
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4174707496130634e-06,
-      "loss": 0.5415,
+      "loss": 0.6472,
       "step": 20261
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.416584104728119e-06,
-      "loss": 0.7195,
+      "loss": 0.8141,
       "step": 20262
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4156975511787217e-06,
-      "loss": 0.5079,
+      "loss": 0.6339,
       "step": 20263
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4148110889771837e-06,
-      "loss": 0.8122,
+      "loss": 0.8686,
       "step": 20264
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.413924718135797e-06,
-      "loss": 0.7797,
+      "loss": 0.8915,
       "step": 20265
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4130384386668615e-06,
-      "loss": 0.7392,
+      "loss": 0.7787,
       "step": 20266
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.412152250582669e-06,
-      "loss": 0.8064,
+      "loss": 0.8895,
       "step": 20267
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4112661538955194e-06,
-      "loss": 0.7267,
+      "loss": 0.8382,
       "step": 20268
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4103801486177034e-06,
-      "loss": 0.6981,
+      "loss": 0.8158,
       "step": 20269
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4094942347615146e-06,
-      "loss": 0.6618,
+      "loss": 0.7945,
       "step": 20270
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4086084123392417e-06,
-      "loss": 0.7451,
+      "loss": 0.8746,
       "step": 20271
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.407722681363176e-06,
-      "loss": 0.836,
+      "loss": 0.854,
       "step": 20272
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4068370418456008e-06,
-      "loss": 0.7096,
+      "loss": 0.7601,
       "step": 20273
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4059514937988103e-06,
-      "loss": 0.7983,
+      "loss": 0.8347,
       "step": 20274
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.405066037235092e-06,
-      "loss": 0.6996,
+      "loss": 0.8288,
       "step": 20275
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.404180672166717e-06,
-      "loss": 0.6315,
+      "loss": 0.7355,
       "step": 20276
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4032953986059815e-06,
-      "loss": 0.7015,
+      "loss": 0.8189,
       "step": 20277
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4024102165651616e-06,
-      "loss": 0.6759,
+      "loss": 0.6733,
       "step": 20278
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4015251260565397e-06,
-      "loss": 0.771,
+      "loss": 0.7954,
       "step": 20279
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.4006401270923915e-06,
-      "loss": 0.751,
+      "loss": 0.8999,
       "step": 20280
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3997552196850048e-06,
-      "loss": 0.7011,
+      "loss": 0.8604,
       "step": 20281
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3988704038466434e-06,
-      "loss": 0.7834,
+      "loss": 0.7951,
       "step": 20282
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3979856795895927e-06,
-      "loss": 0.718,
+      "loss": 0.7773,
       "step": 20283
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3971010469261245e-06,
-      "loss": 0.709,
+      "loss": 0.774,
       "step": 20284
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3962165058685093e-06,
-      "loss": 0.6127,
+      "loss": 0.7132,
       "step": 20285
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.395332056429018e-06,
-      "loss": 0.661,
+      "loss": 0.8787,
       "step": 20286
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.394447698619927e-06,
-      "loss": 0.6985,
+      "loss": 0.7904,
       "step": 20287
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3935634324535026e-06,
-      "loss": 0.757,
+      "loss": 0.8272,
       "step": 20288
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.392679257942012e-06,
-      "loss": 0.8432,
+      "loss": 0.9102,
       "step": 20289
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3917951750977186e-06,
-      "loss": 0.6732,
+      "loss": 0.8126,
       "step": 20290
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.390911183932898e-06,
-      "loss": 0.7659,
+      "loss": 0.8311,
       "step": 20291
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3900272844598004e-06,
-      "loss": 0.6426,
+      "loss": 0.7765,
       "step": 20292
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3891434766906995e-06,
-      "loss": 0.6017,
+      "loss": 0.7406,
       "step": 20293
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3882597606378577e-06,
-      "loss": 0.6356,
+      "loss": 0.7537,
       "step": 20294
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3873761363135227e-06,
-      "loss": 0.6755,
+      "loss": 0.761,
       "step": 20295
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3864926037299663e-06,
-      "loss": 0.8537,
+      "loss": 0.9152,
       "step": 20296
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3856091628994412e-06,
-      "loss": 0.6795,
+      "loss": 0.7839,
       "step": 20297
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.384725813834204e-06,
-      "loss": 0.77,
+      "loss": 0.8376,
       "step": 20298
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3838425565465073e-06,
-      "loss": 0.7532,
+      "loss": 0.854,
       "step": 20299
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3829593910486115e-06,
-      "loss": 0.7929,
+      "loss": 0.8376,
       "step": 20300
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.382076317352766e-06,
-      "loss": 0.6596,
+      "loss": 0.8276,
       "step": 20301
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3811933354712214e-06,
-      "loss": 0.7239,
+      "loss": 0.8234,
       "step": 20302
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3803104454162283e-06,
-      "loss": 0.7098,
+      "loss": 0.7841,
       "step": 20303
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3794276472000364e-06,
-      "loss": 0.5594,
+      "loss": 0.6848,
       "step": 20304
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.378544940834888e-06,
-      "loss": 0.6566,
+      "loss": 0.7148,
       "step": 20305
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3776623263330387e-06,
-      "loss": 0.7414,
+      "loss": 0.7831,
       "step": 20306
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.376779803706728e-06,
-      "loss": 0.7348,
+      "loss": 0.8003,
       "step": 20307
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.375897372968201e-06,
-      "loss": 0.6987,
+      "loss": 0.813,
       "step": 20308
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.375015034129695e-06,
-      "loss": 0.6767,
+      "loss": 0.7765,
       "step": 20309
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3741327872034633e-06,
-      "loss": 0.6847,
+      "loss": 0.8691,
       "step": 20310
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3732506322017324e-06,
-      "loss": 0.6981,
+      "loss": 0.7896,
       "step": 20311
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.372368569136749e-06,
-      "loss": 0.8027,
+      "loss": 0.8402,
       "step": 20312
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.371486598020749e-06,
-      "loss": 0.7574,
+      "loss": 0.8604,
       "step": 20313
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3706047188659675e-06,
-      "loss": 0.8463,
+      "loss": 0.8683,
       "step": 20314
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3697229316846393e-06,
-      "loss": 0.685,
+      "loss": 0.8643,
       "step": 20315
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.368841236488999e-06,
-      "loss": 0.6317,
+      "loss": 0.7609,
       "step": 20316
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3679596332912766e-06,
-      "loss": 0.7099,
+      "loss": 0.7524,
       "step": 20317
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.367078122103703e-06,
-      "loss": 0.6552,
+      "loss": 0.8148,
       "step": 20318
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3661967029385122e-06,
-      "loss": 0.7309,
+      "loss": 0.8726,
       "step": 20319
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3653153758079303e-06,
-      "loss": 0.7323,
+      "loss": 0.7823,
       "step": 20320
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3644341407241845e-06,
-      "loss": 0.7622,
+      "loss": 0.858,
       "step": 20321
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3635529976994963e-06,
-      "loss": 0.6669,
+      "loss": 0.7543,
       "step": 20322
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.362671946746101e-06,
-      "loss": 0.7124,
+      "loss": 0.786,
       "step": 20323
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3617909878762087e-06,
-      "loss": 0.6199,
+      "loss": 0.7059,
       "step": 20324
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3609101211020523e-06,
-      "loss": 0.6114,
+      "loss": 0.764,
       "step": 20325
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.360029346435848e-06,
-      "loss": 0.8305,
+      "loss": 0.8946,
       "step": 20326
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.359148663889815e-06,
-      "loss": 0.7532,
+      "loss": 0.8395,
       "step": 20327
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3582680734761695e-06,
-      "loss": 0.7688,
+      "loss": 0.8801,
       "step": 20328
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.357387575207137e-06,
-      "loss": 0.7371,
+      "loss": 0.8163,
       "step": 20329
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3565071690949215e-06,
-      "loss": 0.6313,
+      "loss": 0.8285,
       "step": 20330
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3556268551517458e-06,
-      "loss": 0.824,
+      "loss": 0.891,
       "step": 20331
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3547466333898206e-06,
-      "loss": 0.7675,
+      "loss": 0.8555,
       "step": 20332
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3538665038213568e-06,
-      "loss": 0.7846,
+      "loss": 0.8358,
       "step": 20333
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3529864664585664e-06,
-      "loss": 0.7588,
+      "loss": 0.8625,
       "step": 20334
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.352106521313655e-06,
-      "loss": 0.6577,
+      "loss": 0.8095,
       "step": 20335
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3512266683988394e-06,
-      "loss": 0.6609,
+      "loss": 0.7273,
       "step": 20336
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.350346907726314e-06,
-      "loss": 0.7722,
+      "loss": 0.9152,
       "step": 20337
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.349467239308294e-06,
-      "loss": 0.7702,
+      "loss": 0.8533,
       "step": 20338
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3485876631569804e-06,
-      "loss": 0.6999,
+      "loss": 0.7978,
       "step": 20339
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3477081792845746e-06,
-      "loss": 0.6549,
+      "loss": 0.7653,
       "step": 20340
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.346828787703277e-06,
-      "loss": 0.6592,
+      "loss": 0.779,
       "step": 20341
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3459494884252973e-06,
-      "loss": 0.6888,
+      "loss": 0.7739,
       "step": 20342
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.345070281462821e-06,
-      "loss": 0.7053,
+      "loss": 0.7801,
       "step": 20343
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3441911668280548e-06,
-      "loss": 0.5987,
+      "loss": 0.7416,
       "step": 20344
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.343312144533194e-06,
-      "loss": 0.6709,
+      "loss": 0.7776,
       "step": 20345
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3424332145904315e-06,
-      "loss": 0.6972,
+      "loss": 0.7725,
       "step": 20346
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3415543770119596e-06,
-      "loss": 0.653,
+      "loss": 0.6904,
       "step": 20347
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3406756318099798e-06,
-      "loss": 0.7166,
+      "loss": 0.8896,
       "step": 20348
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.339796978996671e-06,
-      "loss": 0.6981,
+      "loss": 0.8121,
       "step": 20349
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.338918418584234e-06,
-      "loss": 0.7383,
+      "loss": 0.7826,
       "step": 20350
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3380399505848516e-06,
-      "loss": 0.6911,
+      "loss": 0.7535,
       "step": 20351
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.337161575010712e-06,
-      "loss": 0.7407,
+      "loss": 0.7569,
       "step": 20352
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.336283291874004e-06,
-      "loss": 0.7025,
+      "loss": 0.7973,
       "step": 20353
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.335405101186907e-06,
-      "loss": 0.7584,
+      "loss": 0.8943,
       "step": 20354
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.334527002961615e-06,
-      "loss": 0.7339,
+      "loss": 0.8089,
       "step": 20355
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.333648997210297e-06,
-      "loss": 0.6903,
+      "loss": 0.8375,
       "step": 20356
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3327710839451434e-06,
-      "loss": 0.6965,
+      "loss": 0.8017,
       "step": 20357
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.331893263178333e-06,
-      "loss": 0.7535,
+      "loss": 0.8537,
       "step": 20358
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3310155349220417e-06,
-      "loss": 0.5771,
+      "loss": 0.7883,
       "step": 20359
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.330137899188445e-06,
-      "loss": 0.6563,
+      "loss": 0.8254,
       "step": 20360
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3292603559897274e-06,
-      "loss": 0.7348,
+      "loss": 0.7925,
       "step": 20361
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3283829053380524e-06,
-      "loss": 0.7596,
+      "loss": 0.8608,
       "step": 20362
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.327505547245601e-06,
-      "loss": 0.7239,
+      "loss": 0.8507,
       "step": 20363
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.326628281724543e-06,
-      "loss": 0.7042,
+      "loss": 0.7151,
       "step": 20364
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.325751108787051e-06,
-      "loss": 0.8347,
+      "loss": 0.8984,
       "step": 20365
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3248740284452873e-06,
-      "loss": 0.8205,
+      "loss": 0.8917,
       "step": 20366
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3239970407114307e-06,
-      "loss": 0.6734,
+      "loss": 0.8051,
       "step": 20367
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3231201455976424e-06,
-      "loss": 0.758,
+      "loss": 0.7532,
       "step": 20368
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3222433431160896e-06,
-      "loss": 0.6929,
+      "loss": 0.8024,
       "step": 20369
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3213666332789352e-06,
-      "loss": 0.8301,
+      "loss": 0.8965,
       "step": 20370
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3204900160983435e-06,
-      "loss": 0.684,
+      "loss": 0.857,
       "step": 20371
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3196134915864753e-06,
-      "loss": 0.6175,
+      "loss": 0.737,
       "step": 20372
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.318737059755488e-06,
-      "loss": 0.7783,
+      "loss": 0.7997,
       "step": 20373
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3178607206175517e-06,
-      "loss": 0.8323,
+      "loss": 0.9257,
       "step": 20374
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.316984474184811e-06,
-      "loss": 0.6575,
+      "loss": 0.8365,
       "step": 20375
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.316108320469431e-06,
-      "loss": 0.627,
+      "loss": 0.7586,
       "step": 20376
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3152322594835663e-06,
-      "loss": 0.7557,
+      "loss": 0.8203,
       "step": 20377
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.314356291239368e-06,
-      "loss": 0.873,
+      "loss": 0.9597,
       "step": 20378
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.313480415748989e-06,
-      "loss": 0.6431,
+      "loss": 0.8203,
       "step": 20379
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3126046330245843e-06,
-      "loss": 0.6795,
+      "loss": 0.7304,
       "step": 20380
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.311728943078302e-06,
-      "loss": 0.6347,
+      "loss": 0.7504,
       "step": 20381
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3108533459222914e-06,
-      "loss": 0.6677,
+      "loss": 0.8089,
       "step": 20382
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3099778415687e-06,
-      "loss": 0.7005,
+      "loss": 0.772,
       "step": 20383
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.309102430029675e-06,
-      "loss": 0.7144,
+      "loss": 0.8381,
       "step": 20384
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.308227111317356e-06,
-      "loss": 0.7662,
+      "loss": 0.8462,
       "step": 20385
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.307351885443896e-06,
-      "loss": 0.67,
+      "loss": 0.7808,
       "step": 20386
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3064767524214315e-06,
-      "loss": 0.737,
+      "loss": 0.8536,
       "step": 20387
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.3056017122621063e-06,
-      "loss": 0.6427,
+      "loss": 0.7948,
       "step": 20388
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.304726764978059e-06,
-      "loss": 0.6923,
+      "loss": 0.7523,
       "step": 20389
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.303851910581428e-06,
-      "loss": 0.7618,
+      "loss": 0.8295,
       "step": 20390
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.302977149084352e-06,
-      "loss": 0.6113,
+      "loss": 0.7569,
       "step": 20391
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.302102480498962e-06,
-      "loss": 0.6665,
+      "loss": 0.7092,
       "step": 20392
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.301227904837405e-06,
-      "loss": 0.7392,
+      "loss": 0.8615,
       "step": 20393
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.300353422111798e-06,
-      "loss": 0.7077,
+      "loss": 0.7936,
       "step": 20394
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2994790323342863e-06,
-      "loss": 0.7007,
+      "loss": 0.8237,
       "step": 20395
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.298604735516996e-06,
-      "loss": 0.8469,
+      "loss": 0.8682,
       "step": 20396
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2977305316720576e-06,
-      "loss": 0.6125,
+      "loss": 0.7265,
       "step": 20397
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.296856420811595e-06,
-      "loss": 0.7939,
+      "loss": 0.8984,
       "step": 20398
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2959824029477415e-06,
-      "loss": 0.6559,
+      "loss": 0.7305,
       "step": 20399
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2951084780926214e-06,
-      "loss": 0.7586,
+      "loss": 0.7738,
       "step": 20400
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2942346462583576e-06,
-      "loss": 0.7456,
+      "loss": 0.8531,
       "step": 20401
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2933609074570707e-06,
-      "loss": 0.7755,
+      "loss": 0.8599,
       "step": 20402
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2924872617008927e-06,
-      "loss": 0.6933,
+      "loss": 0.7599,
       "step": 20403
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.291613709001931e-06,
-      "loss": 0.7122,
+      "loss": 0.8238,
       "step": 20404
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.290740249372313e-06,
-      "loss": 0.6996,
+      "loss": 0.8008,
       "step": 20405
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2898668828241554e-06,
-      "loss": 0.6773,
+      "loss": 0.7323,
       "step": 20406
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.288993609369575e-06,
-      "loss": 0.8076,
+      "loss": 0.9072,
       "step": 20407
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.288120429020686e-06,
-      "loss": 0.7086,
+      "loss": 0.8376,
       "step": 20408
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2872473417896034e-06,
-      "loss": 0.724,
+      "loss": 0.8476,
       "step": 20409
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2863743476884402e-06,
-      "loss": 0.6855,
+      "loss": 0.8143,
       "step": 20410
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2855014467293035e-06,
-      "loss": 0.7358,
+      "loss": 0.8031,
       "step": 20411
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.284628638924311e-06,
-      "loss": 0.7289,
+      "loss": 0.8135,
       "step": 20412
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2837559242855687e-06,
-      "loss": 0.6234,
+      "loss": 0.8065,
       "step": 20413
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.282883302825184e-06,
-      "loss": 0.7432,
+      "loss": 0.7964,
       "step": 20414
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2820107745552623e-06,
-      "loss": 0.6919,
+      "loss": 0.7603,
       "step": 20415
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2811383394879093e-06,
-      "loss": 0.6875,
+      "loss": 0.744,
       "step": 20416
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.280265997635227e-06,
-      "loss": 0.7172,
+      "loss": 0.8094,
       "step": 20417
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.279393749009322e-06,
-      "loss": 0.7277,
+      "loss": 0.8582,
       "step": 20418
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2785215936222947e-06,
-      "loss": 0.7174,
+      "loss": 0.8209,
       "step": 20419
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.277649531486242e-06,
-      "loss": 0.596,
+      "loss": 0.6881,
       "step": 20420
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2767775626132627e-06,
-      "loss": 0.7755,
+      "loss": 0.9059,
       "step": 20421
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2759056870154616e-06,
-      "loss": 0.6989,
+      "loss": 0.7389,
       "step": 20422
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.275033904704923e-06,
-      "loss": 0.6875,
+      "loss": 0.812,
       "step": 20423
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.274162215693749e-06,
-      "loss": 0.61,
+      "loss": 0.7141,
       "step": 20424
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.273290619994033e-06,
-      "loss": 0.7083,
+      "loss": 0.8016,
       "step": 20425
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2724191176178656e-06,
-      "loss": 0.724,
+      "loss": 0.8401,
       "step": 20426
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2715477085773374e-06,
-      "loss": 0.7722,
+      "loss": 0.7774,
       "step": 20427
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2706763928845385e-06,
-      "loss": 0.7043,
+      "loss": 0.7886,
       "step": 20428
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2698051705515564e-06,
-      "loss": 0.6943,
+      "loss": 0.8425,
       "step": 20429
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.268934041590476e-06,
-      "loss": 0.7715,
+      "loss": 0.9189,
       "step": 20430
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.268063006013389e-06,
-      "loss": 0.7254,
+      "loss": 0.8377,
       "step": 20431
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2671920638323762e-06,
-      "loss": 0.7587,
+      "loss": 0.7986,
       "step": 20432
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2663212150595214e-06,
-      "loss": 0.5802,
+      "loss": 0.6792,
       "step": 20433
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2654504597069014e-06,
-      "loss": 0.5806,
+      "loss": 0.7554,
       "step": 20434
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.264579797786609e-06,
-      "loss": 0.7303,
+      "loss": 0.8069,
       "step": 20435
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2637092293107087e-06,
-      "loss": 0.7197,
+      "loss": 0.8555,
       "step": 20436
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.262838754291289e-06,
-      "loss": 0.6946,
+      "loss": 0.8681,
       "step": 20437
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2619683727404215e-06,
-      "loss": 0.7901,
+      "loss": 0.8323,
       "step": 20438
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2610980846701833e-06,
-      "loss": 0.7428,
+      "loss": 0.8196,
       "step": 20439
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2602278900926454e-06,
-      "loss": 0.719,
+      "loss": 0.7838,
       "step": 20440
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.259357789019889e-06,
-      "loss": 0.6572,
+      "loss": 0.8058,
       "step": 20441
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2584877814639727e-06,
-      "loss": 0.6597,
+      "loss": 0.7805,
       "step": 20442
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2576178674369776e-06,
-      "loss": 0.7345,
+      "loss": 0.8514,
       "step": 20443
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.256748046950967e-06,
-      "loss": 0.7026,
+      "loss": 0.7739,
       "step": 20444
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.255878320018011e-06,
-      "loss": 0.8625,
+      "loss": 0.8569,
       "step": 20445
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.255008686650175e-06,
-      "loss": 0.76,
+      "loss": 0.8561,
       "step": 20446
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.254139146859523e-06,
-      "loss": 0.6076,
+      "loss": 0.7619,
       "step": 20447
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2532697006581192e-06,
-      "loss": 0.6612,
+      "loss": 0.7857,
       "step": 20448
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2524003480580226e-06,
-      "loss": 0.7274,
+      "loss": 0.7837,
       "step": 20449
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2515310890713004e-06,
-      "loss": 0.6264,
+      "loss": 0.7229,
       "step": 20450
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2506619237100102e-06,
-      "loss": 0.745,
+      "loss": 0.8577,
       "step": 20451
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.24979285198621e-06,
-      "loss": 0.6526,
+      "loss": 0.6626,
       "step": 20452
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2489238739119534e-06,
-      "loss": 0.7294,
+      "loss": 0.8687,
       "step": 20453
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.248054989499305e-06,
-      "loss": 0.5861,
+      "loss": 0.7452,
       "step": 20454
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2471861987603083e-06,
-      "loss": 0.7184,
+      "loss": 0.7747,
       "step": 20455
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2463175017070248e-06,
-      "loss": 0.6464,
+      "loss": 0.8223,
       "step": 20456
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2454488983515043e-06,
-      "loss": 0.7968,
+      "loss": 0.8691,
       "step": 20457
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.244580388705796e-06,
-      "loss": 0.82,
+      "loss": 0.8557,
       "step": 20458
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2437119727819476e-06,
-      "loss": 0.735,
+      "loss": 0.9058,
       "step": 20459
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2428436505920157e-06,
-      "loss": 0.7281,
+      "loss": 0.8092,
       "step": 20460
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2419754221480347e-06,
-      "loss": 0.7097,
+      "loss": 0.8495,
       "step": 20461
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2411072874620608e-06,
-      "loss": 0.8826,
+      "loss": 0.9239,
       "step": 20462
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.240239246546132e-06,
-      "loss": 0.7886,
+      "loss": 0.8238,
       "step": 20463
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2393712994122937e-06,
-      "loss": 0.7009,
+      "loss": 0.7925,
       "step": 20464
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.238503446072586e-06,
-      "loss": 0.6189,
+      "loss": 0.7189,
       "step": 20465
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.237635686539047e-06,
-      "loss": 0.6526,
+      "loss": 0.8253,
       "step": 20466
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.236768020823724e-06,
-      "loss": 0.5841,
+      "loss": 0.8103,
       "step": 20467
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2359004489386427e-06,
-      "loss": 0.6605,
+      "loss": 0.7998,
       "step": 20468
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2350329708958495e-06,
-      "loss": 0.7063,
+      "loss": 0.7615,
       "step": 20469
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.234165586707375e-06,
-      "loss": 0.69,
+      "loss": 0.8256,
       "step": 20470
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2332982963852545e-06,
-      "loss": 0.7564,
+      "loss": 0.8082,
       "step": 20471
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2324310999415144e-06,
-      "loss": 0.6837,
+      "loss": 0.7133,
       "step": 20472
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2315639973881984e-06,
-      "loss": 0.8608,
+      "loss": 0.8932,
       "step": 20473
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.230696988737323e-06,
-      "loss": 0.656,
+      "loss": 0.7474,
       "step": 20474
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2298300740009246e-06,
-      "loss": 0.7457,
+      "loss": 0.7833,
       "step": 20475
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2289632531910287e-06,
-      "loss": 0.6575,
+      "loss": 0.8169,
       "step": 20476
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2280965263196604e-06,
-      "loss": 0.7951,
+      "loss": 0.842,
       "step": 20477
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.227229893398842e-06,
-      "loss": 0.736,
+      "loss": 0.9466,
       "step": 20478
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2263633544406037e-06,
-      "loss": 0.7942,
+      "loss": 0.8696,
       "step": 20479
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.225496909456962e-06,
-      "loss": 0.7586,
+      "loss": 0.871,
       "step": 20480
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.224630558459939e-06,
-      "loss": 0.7597,
+      "loss": 0.851,
       "step": 20481
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2237643014615537e-06,
-      "loss": 0.5861,
+      "loss": 0.7152,
       "step": 20482
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2228981384738255e-06,
-      "loss": 0.7052,
+      "loss": 0.7926,
       "step": 20483
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.222032069508767e-06,
-      "loss": 0.7379,
+      "loss": 0.8412,
       "step": 20484
     },
     {
       "epoch": 0.74,
       "learning_rate": 3.2211660945783996e-06,
-      "loss": 0.6444,
+      "loss": 0.7702,
       "step": 20485
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.220300213694737e-06,
-      "loss": 0.7003,
+      "loss": 0.8141,
       "step": 20486
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2194344268697853e-06,
-      "loss": 0.7857,
+      "loss": 0.8491,
       "step": 20487
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2185687341155627e-06,
-      "loss": 0.721,
+      "loss": 0.8371,
       "step": 20488
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2177031354440767e-06,
-      "loss": 0.7469,
+      "loss": 0.803,
       "step": 20489
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2168376308673378e-06,
-      "loss": 0.623,
+      "loss": 0.7102,
       "step": 20490
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2159722203973497e-06,
-      "loss": 0.6857,
+      "loss": 0.8593,
       "step": 20491
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.215106904046127e-06,
-      "loss": 0.7519,
+      "loss": 0.8653,
       "step": 20492
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.214241681825664e-06,
-      "loss": 0.7557,
+      "loss": 0.8215,
       "step": 20493
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2133765537479733e-06,
-      "loss": 0.6404,
+      "loss": 0.7847,
       "step": 20494
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2125115198250534e-06,
-      "loss": 0.763,
+      "loss": 0.803,
       "step": 20495
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.211646580068907e-06,
-      "loss": 0.7326,
+      "loss": 0.7796,
       "step": 20496
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2107817344915284e-06,
-      "loss": 0.8185,
+      "loss": 0.8781,
       "step": 20497
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.209916983104924e-06,
-      "loss": 0.7313,
+      "loss": 0.8244,
       "step": 20498
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2090523259210883e-06,
-      "loss": 0.7338,
+      "loss": 0.8485,
       "step": 20499
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2081877629520154e-06,
-      "loss": 0.8701,
+      "loss": 0.8723,
       "step": 20500
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.207323294209699e-06,
-      "loss": 0.6226,
+      "loss": 0.7363,
       "step": 20501
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2064589197061403e-06,
-      "loss": 0.6819,
+      "loss": 0.7696,
       "step": 20502
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.205594639453319e-06,
-      "loss": 0.6916,
+      "loss": 0.757,
       "step": 20503
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.204730453463235e-06,
-      "loss": 0.7951,
+      "loss": 0.8685,
       "step": 20504
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.203866361747877e-06,
-      "loss": 0.6964,
+      "loss": 0.7528,
       "step": 20505
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2030023643192266e-06,
-      "loss": 0.8067,
+      "loss": 0.9025,
       "step": 20506
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2021384611892757e-06,
-      "loss": 0.6774,
+      "loss": 0.8861,
       "step": 20507
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.20127465237001e-06,
-      "loss": 0.6684,
+      "loss": 0.7419,
       "step": 20508
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.2004109378734126e-06,
-      "loss": 0.5589,
+      "loss": 0.6884,
       "step": 20509
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1995473177114624e-06,
-      "loss": 0.6127,
+      "loss": 0.8083,
       "step": 20510
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.198683791896148e-06,
-      "loss": 0.6521,
+      "loss": 0.7423,
       "step": 20511
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.197820360439445e-06,
-      "loss": 0.5802,
+      "loss": 0.7135,
       "step": 20512
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1969570233533352e-06,
-      "loss": 0.7232,
+      "loss": 0.8275,
       "step": 20513
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1960937806497938e-06,
-      "loss": 0.7628,
+      "loss": 0.8357,
       "step": 20514
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1952306323407966e-06,
-      "loss": 0.6269,
+      "loss": 0.7167,
       "step": 20515
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1943675784383177e-06,
-      "loss": 0.6705,
+      "loss": 0.7547,
       "step": 20516
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1935046189543364e-06,
-      "loss": 0.6821,
+      "loss": 0.8188,
       "step": 20517
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1926417539008203e-06,
-      "loss": 0.8019,
+      "loss": 0.8755,
       "step": 20518
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1917789832897418e-06,
-      "loss": 0.6888,
+      "loss": 0.7447,
       "step": 20519
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1909163071330674e-06,
-      "loss": 0.6994,
+      "loss": 0.7749,
       "step": 20520
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1900537254427745e-06,
-      "loss": 0.6511,
+      "loss": 0.7878,
       "step": 20521
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.189191238230818e-06,
-      "loss": 0.7087,
+      "loss": 0.7799,
       "step": 20522
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1883288455091722e-06,
-      "loss": 0.694,
+      "loss": 0.7384,
       "step": 20523
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1874665472898003e-06,
-      "loss": 0.6177,
+      "loss": 0.7551,
       "step": 20524
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1866043435846627e-06,
-      "loss": 0.744,
+      "loss": 0.8784,
       "step": 20525
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.185742234405724e-06,
-      "loss": 0.6932,
+      "loss": 0.8297,
       "step": 20526
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.184880219764944e-06,
-      "loss": 0.5086,
+      "loss": 0.6567,
       "step": 20527
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.184018299674281e-06,
-      "loss": 0.8991,
+      "loss": 0.9305,
       "step": 20528
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.183156474145691e-06,
-      "loss": 0.6534,
+      "loss": 0.714,
       "step": 20529
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1822947431911356e-06,
-      "loss": 0.7996,
+      "loss": 0.8777,
       "step": 20530
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.181433106822568e-06,
-      "loss": 0.8486,
+      "loss": 0.8978,
       "step": 20531
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1805715650519407e-06,
-      "loss": 0.6289,
+      "loss": 0.8583,
       "step": 20532
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.179710117891205e-06,
-      "loss": 0.6884,
+      "loss": 0.8192,
       "step": 20533
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1788487653523214e-06,
-      "loss": 0.6747,
+      "loss": 0.7891,
       "step": 20534
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1779875074472266e-06,
-      "loss": 0.7739,
+      "loss": 0.8303,
       "step": 20535
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1771263441878786e-06,
-      "loss": 0.7542,
+      "loss": 0.8016,
       "step": 20536
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1762652755862234e-06,
-      "loss": 0.7903,
+      "loss": 0.8883,
       "step": 20537
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1754043016542046e-06,
-      "loss": 0.807,
+      "loss": 0.8478,
       "step": 20538
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1745434224037654e-06,
-      "loss": 0.6795,
+      "loss": 0.7957,
       "step": 20539
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1736826378468587e-06,
-      "loss": 0.6964,
+      "loss": 0.7788,
       "step": 20540
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1728219479954137e-06,
-      "loss": 0.5691,
+      "loss": 0.6821,
       "step": 20541
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.171961352861381e-06,
-      "loss": 0.8394,
+      "loss": 0.8739,
       "step": 20542
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1711008524566966e-06,
-      "loss": 0.624,
+      "loss": 0.8551,
       "step": 20543
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1702404467933003e-06,
-      "loss": 0.6565,
+      "loss": 0.7983,
       "step": 20544
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1693801358831256e-06,
-      "loss": 0.6843,
+      "loss": 0.8017,
       "step": 20545
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1685199197381077e-06,
-      "loss": 0.606,
+      "loss": 0.7899,
       "step": 20546
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1676597983701896e-06,
-      "loss": 0.7836,
+      "loss": 0.859,
       "step": 20547
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1667997717912923e-06,
-      "loss": 0.7409,
+      "loss": 0.8276,
       "step": 20548
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1659398400133557e-06,
-      "loss": 0.828,
+      "loss": 0.8411,
       "step": 20549
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1650800030483075e-06,
-      "loss": 0.6926,
+      "loss": 0.7564,
       "step": 20550
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1642202609080773e-06,
-      "loss": 0.6722,
+      "loss": 0.7888,
       "step": 20551
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1633606136045882e-06,
-      "loss": 0.7102,
+      "loss": 0.8193,
       "step": 20552
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1625010611497776e-06,
-      "loss": 0.8425,
+      "loss": 0.9348,
       "step": 20553
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1616416035555577e-06,
-      "loss": 0.6783,
+      "loss": 0.7645,
       "step": 20554
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1607822408338606e-06,
-      "loss": 0.6962,
+      "loss": 0.8145,
       "step": 20555
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.159922972996606e-06,
-      "loss": 0.7013,
+      "loss": 0.8095,
       "step": 20556
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.159063800055716e-06,
-      "loss": 0.6387,
+      "loss": 0.7838,
       "step": 20557
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.158204722023107e-06,
-      "loss": 0.7375,
+      "loss": 0.8525,
       "step": 20558
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1573457389107065e-06,
-      "loss": 0.7016,
+      "loss": 0.9219,
       "step": 20559
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1564868507304182e-06,
-      "loss": 0.7287,
+      "loss": 0.8206,
       "step": 20560
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1556280574941687e-06,
-      "loss": 0.6577,
+      "loss": 0.7492,
       "step": 20561
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1547693592138697e-06,
-      "loss": 0.6381,
+      "loss": 0.7304,
       "step": 20562
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1539107559014325e-06,
-      "loss": 0.6122,
+      "loss": 0.7412,
       "step": 20563
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.153052247568771e-06,
-      "loss": 0.7104,
+      "loss": 0.8336,
       "step": 20564
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1521938342277925e-06,
-      "loss": 0.7294,
+      "loss": 0.7263,
       "step": 20565
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1513355158904137e-06,
-      "loss": 0.4136,
+      "loss": 0.6588,
       "step": 20566
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1504772925685323e-06,
-      "loss": 0.6419,
+      "loss": 0.7769,
       "step": 20567
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1496191642740625e-06,
-      "loss": 0.6742,
+      "loss": 0.7343,
       "step": 20568
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1487611310189083e-06,
-      "loss": 0.7317,
+      "loss": 0.8306,
       "step": 20569
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.147903192814973e-06,
-      "loss": 0.6473,
+      "loss": 0.7535,
       "step": 20570
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.147045349674155e-06,
-      "loss": 0.6849,
+      "loss": 0.7974,
       "step": 20571
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.146187601608367e-06,
-      "loss": 0.5295,
+      "loss": 0.6555,
       "step": 20572
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.145329948629494e-06,
-      "loss": 0.7328,
+      "loss": 0.8566,
       "step": 20573
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.144472390749447e-06,
-      "loss": 0.7602,
+      "loss": 0.8031,
       "step": 20574
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1436149279801174e-06,
-      "loss": 0.553,
+      "loss": 0.7621,
       "step": 20575
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1427575603334036e-06,
-      "loss": 0.7734,
+      "loss": 0.8447,
       "step": 20576
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.141900287821197e-06,
-      "loss": 0.7222,
+      "loss": 0.7672,
       "step": 20577
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1410431104553962e-06,
-      "loss": 0.7898,
+      "loss": 0.8578,
       "step": 20578
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1401860282478903e-06,
-      "loss": 0.6043,
+      "loss": 0.749,
       "step": 20579
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1393290412105702e-06,
-      "loss": 0.6783,
+      "loss": 0.8254,
       "step": 20580
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1384721493553263e-06,
-      "loss": 0.7852,
+      "loss": 0.7754,
       "step": 20581
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1376153526940457e-06,
-      "loss": 0.5911,
+      "loss": 0.7421,
       "step": 20582
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.136758651238616e-06,
-      "loss": 0.7239,
+      "loss": 0.8237,
       "step": 20583
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.135902045000919e-06,
-      "loss": 0.6638,
+      "loss": 0.8554,
       "step": 20584
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1350455339928488e-06,
-      "loss": 0.6308,
+      "loss": 0.7204,
       "step": 20585
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1341891182262752e-06,
-      "loss": 0.7753,
+      "loss": 0.8012,
       "step": 20586
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.133332797713089e-06,
-      "loss": 0.6459,
+      "loss": 0.8206,
       "step": 20587
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1324765724651683e-06,
-      "loss": 0.6604,
+      "loss": 0.83,
       "step": 20588
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1316204424943907e-06,
-      "loss": 0.6215,
+      "loss": 0.765,
       "step": 20589
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.130764407812632e-06,
-      "loss": 0.7407,
+      "loss": 0.8711,
       "step": 20590
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.129908468431774e-06,
-      "loss": 0.7932,
+      "loss": 0.8387,
       "step": 20591
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1290526243636888e-06,
-      "loss": 0.709,
+      "loss": 0.8652,
       "step": 20592
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1281968756202498e-06,
-      "loss": 0.5227,
+      "loss": 0.6304,
       "step": 20593
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1273412222133294e-06,
-      "loss": 0.7614,
+      "loss": 0.8613,
       "step": 20594
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.126485664154798e-06,
-      "loss": 0.7064,
+      "loss": 0.7425,
       "step": 20595
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.125630201456522e-06,
-      "loss": 0.7163,
+      "loss": 0.7834,
       "step": 20596
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1247748341303784e-06,
-      "loss": 0.789,
+      "loss": 0.794,
       "step": 20597
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.123919562188228e-06,
-      "loss": 0.6292,
+      "loss": 0.7822,
       "step": 20598
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1230643856419384e-06,
-      "loss": 0.8385,
+      "loss": 0.9167,
       "step": 20599
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.122209304503373e-06,
-      "loss": 0.6976,
+      "loss": 0.7936,
       "step": 20600
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1213543187843953e-06,
-      "loss": 0.6577,
+      "loss": 0.7945,
       "step": 20601
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.120499428496867e-06,
-      "loss": 0.7887,
+      "loss": 0.8454,
       "step": 20602
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1196446336526454e-06,
-      "loss": 0.7033,
+      "loss": 0.8463,
       "step": 20603
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1187899342635986e-06,
-      "loss": 0.6441,
+      "loss": 0.8447,
       "step": 20604
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1179353303415727e-06,
-      "loss": 0.7181,
+      "loss": 0.8572,
       "step": 20605
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1170808218984317e-06,
-      "loss": 0.8251,
+      "loss": 0.8973,
       "step": 20606
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.116226408946029e-06,
-      "loss": 0.702,
+      "loss": 0.8299,
       "step": 20607
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1153720914962194e-06,
-      "loss": 0.783,
+      "loss": 0.8511,
       "step": 20608
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1145178695608492e-06,
-      "loss": 0.7802,
+      "loss": 0.7801,
       "step": 20609
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.113663743151778e-06,
-      "loss": 0.7352,
+      "loss": 0.8303,
       "step": 20610
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.112809712280852e-06,
-      "loss": 0.6743,
+      "loss": 0.7658,
       "step": 20611
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1119557769599197e-06,
-      "loss": 0.6891,
+      "loss": 0.7388,
       "step": 20612
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.111101937200829e-06,
-      "loss": 0.7263,
+      "loss": 0.8067,
       "step": 20613
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1102481930154237e-06,
-      "loss": 0.5792,
+      "loss": 0.8013,
       "step": 20614
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.109394544415546e-06,
-      "loss": 0.6795,
+      "loss": 0.7951,
       "step": 20615
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.108540991413046e-06,
-      "loss": 0.7009,
+      "loss": 0.8238,
       "step": 20616
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.107687534019762e-06,
-      "loss": 0.679,
+      "loss": 0.7439,
       "step": 20617
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1068341722475347e-06,
-      "loss": 0.7558,
+      "loss": 0.8591,
       "step": 20618
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1059809061082038e-06,
-      "loss": 0.7111,
+      "loss": 0.8379,
       "step": 20619
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1051277356136055e-06,
-      "loss": 0.8024,
+      "loss": 0.8206,
       "step": 20620
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.104274660775578e-06,
-      "loss": 0.8461,
+      "loss": 0.8804,
       "step": 20621
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1034216816059525e-06,
-      "loss": 0.6084,
+      "loss": 0.727,
       "step": 20622
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1025687981165698e-06,
-      "loss": 0.7421,
+      "loss": 0.8157,
       "step": 20623
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.101716010319258e-06,
-      "loss": 0.7308,
+      "loss": 0.8333,
       "step": 20624
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1008633182258494e-06,
-      "loss": 0.6634,
+      "loss": 0.8178,
       "step": 20625
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.1000107218481733e-06,
-      "loss": 0.7621,
+      "loss": 0.8013,
       "step": 20626
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.09915822119806e-06,
-      "loss": 0.6914,
+      "loss": 0.739,
       "step": 20627
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0983058162873313e-06,
-      "loss": 0.7619,
+      "loss": 0.8756,
       "step": 20628
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.09745350712782e-06,
-      "loss": 0.6689,
+      "loss": 0.7899,
       "step": 20629
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.096601293731348e-06,
-      "loss": 0.7238,
+      "loss": 0.8544,
       "step": 20630
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.095749176109739e-06,
-      "loss": 0.6616,
+      "loss": 0.7786,
       "step": 20631
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.094897154274811e-06,
-      "loss": 0.8453,
+      "loss": 0.8959,
       "step": 20632
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0940452282383937e-06,
-      "loss": 0.7993,
+      "loss": 0.8424,
       "step": 20633
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.093193398012294e-06,
-      "loss": 0.7074,
+      "loss": 0.8541,
       "step": 20634
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0923416636083404e-06,
-      "loss": 0.5767,
+      "loss": 0.78,
       "step": 20635
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.091490025038345e-06,
-      "loss": 0.7674,
+      "loss": 0.8419,
       "step": 20636
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0906384823141244e-06,
-      "loss": 0.7999,
+      "loss": 0.86,
       "step": 20637
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0897870354474914e-06,
-      "loss": 0.7175,
+      "loss": 0.8498,
       "step": 20638
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.088935684450258e-06,
-      "loss": 0.6605,
+      "loss": 0.861,
       "step": 20639
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0880844293342382e-06,
-      "loss": 0.6917,
+      "loss": 0.7467,
       "step": 20640
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0872332701112362e-06,
-      "loss": 0.7242,
+      "loss": 0.8165,
       "step": 20641
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.086382206793067e-06,
-      "loss": 0.679,
+      "loss": 0.8298,
       "step": 20642
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.085531239391537e-06,
-      "loss": 0.7659,
+      "loss": 0.8816,
       "step": 20643
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.08468036791845e-06,
-      "loss": 0.7279,
+      "loss": 0.8394,
       "step": 20644
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0838295923856077e-06,
-      "loss": 0.7641,
+      "loss": 0.8116,
       "step": 20645
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0829789128048237e-06,
-      "loss": 0.8519,
+      "loss": 0.8926,
       "step": 20646
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.082128329187888e-06,
-      "loss": 0.6203,
+      "loss": 0.7515,
       "step": 20647
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.081277841546608e-06,
-      "loss": 0.7317,
+      "loss": 0.8848,
       "step": 20648
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0804274498927834e-06,
-      "loss": 0.6379,
+      "loss": 0.7827,
       "step": 20649
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.079577154238209e-06,
-      "loss": 0.6285,
+      "loss": 0.7551,
       "step": 20650
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.07872695459468e-06,
-      "loss": 0.7278,
+      "loss": 0.8195,
       "step": 20651
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.077876850974001e-06,
-      "loss": 0.6602,
+      "loss": 0.8257,
       "step": 20652
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0770268433879525e-06,
-      "loss": 0.6432,
+      "loss": 0.8493,
       "step": 20653
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.076176931848338e-06,
-      "loss": 0.7532,
+      "loss": 0.8279,
       "step": 20654
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.075327116366944e-06,
-      "loss": 0.75,
+      "loss": 0.8825,
       "step": 20655
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.074477396955563e-06,
-      "loss": 0.7408,
+      "loss": 0.8043,
       "step": 20656
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.073627773625981e-06,
-      "loss": 0.639,
+      "loss": 0.7467,
       "step": 20657
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0727782463899868e-06,
-      "loss": 0.7453,
+      "loss": 0.8693,
       "step": 20658
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0719288152593664e-06,
-      "loss": 0.811,
+      "loss": 0.8399,
       "step": 20659
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0710794802459e-06,
-      "loss": 0.6261,
+      "loss": 0.8303,
       "step": 20660
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0702302413613783e-06,
-      "loss": 0.5878,
+      "loss": 0.7135,
       "step": 20661
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.06938109861758e-06,
-      "loss": 0.7308,
+      "loss": 0.8514,
       "step": 20662
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0685320520262853e-06,
-      "loss": 0.7956,
+      "loss": 0.8451,
       "step": 20663
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0676831015992714e-06,
-      "loss": 0.5993,
+      "loss": 0.7097,
       "step": 20664
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0668342473483248e-06,
-      "loss": 0.7039,
+      "loss": 0.7725,
       "step": 20665
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0659854892852105e-06,
-      "loss": 0.8575,
+      "loss": 0.8492,
       "step": 20666
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.065136827421712e-06,
-      "loss": 0.7878,
+      "loss": 0.8662,
       "step": 20667
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0642882617696003e-06,
-      "loss": 0.7589,
+      "loss": 0.7967,
       "step": 20668
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.063439792340649e-06,
-      "loss": 0.7486,
+      "loss": 0.8255,
       "step": 20669
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0625914191466265e-06,
-      "loss": 0.6979,
+      "loss": 0.7068,
       "step": 20670
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0617431421993105e-06,
-      "loss": 0.6619,
+      "loss": 0.7559,
       "step": 20671
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0608949615104577e-06,
-      "loss": 0.718,
+      "loss": 0.8026,
       "step": 20672
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0600468770918455e-06,
-      "loss": 0.751,
+      "loss": 0.8672,
       "step": 20673
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0591988889552372e-06,
-      "loss": 0.6356,
+      "loss": 0.7743,
       "step": 20674
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.058350997112396e-06,
-      "loss": 0.654,
+      "loss": 0.7865,
       "step": 20675
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0575032015750863e-06,
-      "loss": 0.759,
+      "loss": 0.8411,
       "step": 20676
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.056655502355065e-06,
-      "loss": 0.6411,
+      "loss": 0.7512,
       "step": 20677
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.055807899464105e-06,
-      "loss": 0.8591,
+      "loss": 0.8897,
       "step": 20678
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0549603929139505e-06,
-      "loss": 0.6587,
+      "loss": 0.7968,
       "step": 20679
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0541129827163705e-06,
-      "loss": 0.6277,
+      "loss": 0.7627,
       "step": 20680
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0532656688831174e-06,
-      "loss": 0.7273,
+      "loss": 0.8409,
       "step": 20681
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.052418451425948e-06,
-      "loss": 0.8208,
+      "loss": 0.7864,
       "step": 20682
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.051571330356612e-06,
-      "loss": 0.6974,
+      "loss": 0.7999,
       "step": 20683
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0507243056868707e-06,
-      "loss": 0.6859,
+      "loss": 0.7681,
       "step": 20684
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.049877377428464e-06,
-      "loss": 0.7335,
+      "loss": 0.8199,
       "step": 20685
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0490305455931514e-06,
-      "loss": 0.6609,
+      "loss": 0.7438,
       "step": 20686
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0481838101926774e-06,
-      "loss": 0.5668,
+      "loss": 0.6984,
       "step": 20687
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.04733717123879e-06,
-      "loss": 0.7183,
+      "loss": 0.843,
       "step": 20688
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0464906287432315e-06,
-      "loss": 0.7855,
+      "loss": 0.8106,
       "step": 20689
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.045644182717753e-06,
-      "loss": 0.6505,
+      "loss": 0.8284,
       "step": 20690
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0447978331740946e-06,
-      "loss": 0.5769,
+      "loss": 0.8042,
       "step": 20691
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0439515801239973e-06,
-      "loss": 0.8065,
+      "loss": 0.794,
       "step": 20692
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0431054235792033e-06,
-      "loss": 0.689,
+      "loss": 0.8153,
       "step": 20693
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.042259363551451e-06,
-      "loss": 0.6688,
+      "loss": 0.8006,
       "step": 20694
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0414134000524785e-06,
-      "loss": 0.8075,
+      "loss": 0.883,
       "step": 20695
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0405675330940186e-06,
-      "loss": 0.6075,
+      "loss": 0.7507,
       "step": 20696
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.039721762687816e-06,
-      "loss": 0.8362,
+      "loss": 0.9034,
       "step": 20697
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0388760888455925e-06,
-      "loss": 0.6471,
+      "loss": 0.78,
       "step": 20698
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.03803051157909e-06,
-      "loss": 0.5789,
+      "loss": 0.7055,
       "step": 20699
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.037185030900036e-06,
-      "loss": 0.7188,
+      "loss": 0.7717,
       "step": 20700
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.036339646820162e-06,
-      "loss": 0.7272,
+      "loss": 0.8546,
       "step": 20701
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.035494359351191e-06,
-      "loss": 0.7093,
+      "loss": 0.7719,
       "step": 20702
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0346491685048606e-06,
-      "loss": 0.6347,
+      "loss": 0.7504,
       "step": 20703
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.033804074292885e-06,
-      "loss": 0.6379,
+      "loss": 0.7552,
       "step": 20704
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0329590767269967e-06,
-      "loss": 0.7017,
+      "loss": 0.8552,
       "step": 20705
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0321141758189166e-06,
-      "loss": 0.7871,
+      "loss": 0.7866,
       "step": 20706
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0312693715803654e-06,
-      "loss": 0.7342,
+      "loss": 0.729,
       "step": 20707
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0304246640230617e-06,
-      "loss": 0.7131,
+      "loss": 0.8049,
       "step": 20708
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0295800531587303e-06,
-      "loss": 0.6402,
+      "loss": 0.7579,
       "step": 20709
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.028735538999087e-06,
-      "loss": 0.8324,
+      "loss": 0.9107,
       "step": 20710
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.027891121555846e-06,
-      "loss": 0.6836,
+      "loss": 0.7881,
       "step": 20711
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0270468008407204e-06,
-      "loss": 0.7098,
+      "loss": 0.8243,
       "step": 20712
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0262025768654334e-06,
-      "loss": 0.743,
+      "loss": 0.8054,
       "step": 20713
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0253584496416853e-06,
-      "loss": 0.7325,
+      "loss": 0.8235,
       "step": 20714
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.024514419181196e-06,
-      "loss": 0.6304,
+      "loss": 0.7564,
       "step": 20715
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0236704854956756e-06,
-      "loss": 0.7614,
+      "loss": 0.8594,
       "step": 20716
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0228266485968226e-06,
-      "loss": 0.7997,
+      "loss": 0.8751,
       "step": 20717
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0219829084963536e-06,
-      "loss": 0.7821,
+      "loss": 0.8196,
       "step": 20718
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.021139265205971e-06,
-      "loss": 0.8416,
+      "loss": 0.8501,
       "step": 20719
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0202957187373795e-06,
-      "loss": 0.7524,
+      "loss": 0.8527,
       "step": 20720
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.019452269102279e-06,
-      "loss": 0.7021,
+      "loss": 0.7917,
       "step": 20721
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0186089163123776e-06,
-      "loss": 0.6451,
+      "loss": 0.7792,
       "step": 20722
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0177656603793713e-06,
-      "loss": 0.6413,
+      "loss": 0.8699,
       "step": 20723
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0169225013149604e-06,
-      "loss": 0.6206,
+      "loss": 0.7253,
       "step": 20724
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0160794391308423e-06,
-      "loss": 0.6022,
+      "loss": 0.7482,
       "step": 20725
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.015236473838713e-06,
-      "loss": 0.7067,
+      "loss": 0.7833,
       "step": 20726
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0143936054502652e-06,
-      "loss": 0.6297,
+      "loss": 0.7166,
       "step": 20727
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.013550833977197e-06,
-      "loss": 0.6788,
+      "loss": 0.7491,
       "step": 20728
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.012708159431199e-06,
-      "loss": 0.7184,
+      "loss": 0.8516,
       "step": 20729
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0118655818239605e-06,
-      "loss": 0.6338,
+      "loss": 0.7065,
       "step": 20730
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.01102310116717e-06,
-      "loss": 0.7487,
+      "loss": 0.8097,
       "step": 20731
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.010180717472524e-06,
-      "loss": 0.7031,
+      "loss": 0.9026,
       "step": 20732
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0093384307516973e-06,
-      "loss": 0.7657,
+      "loss": 0.8358,
       "step": 20733
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0084962410163845e-06,
-      "loss": 0.6458,
+      "loss": 0.7619,
       "step": 20734
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.007654148278266e-06,
-      "loss": 0.6858,
+      "loss": 0.8409,
       "step": 20735
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0068121525490266e-06,
-      "loss": 0.5998,
+      "loss": 0.7597,
       "step": 20736
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0059702538403457e-06,
-      "loss": 0.7894,
+      "loss": 0.8695,
       "step": 20737
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0051284521639044e-06,
-      "loss": 0.7264,
+      "loss": 0.8144,
       "step": 20738
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.004286747531382e-06,
-      "loss": 0.912,
+      "loss": 0.988,
       "step": 20739
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0034451399544506e-06,
-      "loss": 0.7803,
+      "loss": 0.8186,
       "step": 20740
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.002603629444796e-06,
-      "loss": 0.7595,
+      "loss": 0.7536,
       "step": 20741
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.001762216014087e-06,
-      "loss": 0.8326,
+      "loss": 0.859,
       "step": 20742
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.000920899673998e-06,
-      "loss": 0.6219,
+      "loss": 0.7722,
       "step": 20743
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.0000796804361985e-06,
-      "loss": 0.7172,
+      "loss": 0.8427,
       "step": 20744
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9992385583123683e-06,
-      "loss": 0.7281,
+      "loss": 0.778,
       "step": 20745
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9983975333141635e-06,
-      "loss": 0.7445,
+      "loss": 0.8637,
       "step": 20746
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9975566054532625e-06,
-      "loss": 0.6955,
+      "loss": 0.7937,
       "step": 20747
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9967157747413278e-06,
-      "loss": 0.6618,
+      "loss": 0.8067,
       "step": 20748
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9958750411900263e-06,
-      "loss": 0.7598,
+      "loss": 0.7824,
       "step": 20749
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.995034404811017e-06,
-      "loss": 0.7042,
+      "loss": 0.8056,
       "step": 20750
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.994193865615973e-06,
-      "loss": 0.6878,
+      "loss": 0.7941,
       "step": 20751
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.993353423616544e-06,
-      "loss": 0.8315,
+      "loss": 0.8512,
       "step": 20752
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.992513078824397e-06,
-      "loss": 0.8654,
+      "loss": 0.8839,
       "step": 20753
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9916728312511888e-06,
-      "loss": 0.7601,
+      "loss": 0.7913,
       "step": 20754
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9908326809085773e-06,
-      "loss": 0.768,
+      "loss": 0.8629,
       "step": 20755
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.989992627808217e-06,
-      "loss": 0.7041,
+      "loss": 0.7707,
       "step": 20756
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9891526719617636e-06,
-      "loss": 0.6524,
+      "loss": 0.6924,
       "step": 20757
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.9883128133808693e-06,
-      "loss": 0.6629,
+      "loss": 0.7549,
       "step": 20758
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.987473052077183e-06,
-      "loss": 0.5903,
+      "loss": 0.7216,
       "step": 20759
     },
     {
       "epoch": 0.75,
       "learning_rate": 2.986633388062363e-06,
-      "loss": 0.6955,
+      "loss": 0.7811,
       "step": 20760
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9857938213480532e-06,
-      "loss": 0.6866,
+      "loss": 0.7452,
       "step": 20761
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9849543519459023e-06,
-      "loss": 0.699,
+      "loss": 0.8675,
       "step": 20762
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9841149798675538e-06,
-      "loss": 0.7112,
+      "loss": 0.7569,
       "step": 20763
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.983275705124662e-06,
-      "loss": 0.788,
+      "loss": 0.8461,
       "step": 20764
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.982436527728858e-06,
-      "loss": 0.747,
+      "loss": 0.7808,
       "step": 20765
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.981597447691793e-06,
-      "loss": 0.6824,
+      "loss": 0.7829,
       "step": 20766
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9807584650251064e-06,
-      "loss": 0.8436,
+      "loss": 0.9382,
       "step": 20767
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9799195797404377e-06,
-      "loss": 0.771,
+      "loss": 0.798,
       "step": 20768
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.97908079184942e-06,
-      "loss": 0.6301,
+      "loss": 0.8016,
       "step": 20769
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.978242101363702e-06,
-      "loss": 0.6763,
+      "loss": 0.7696,
       "step": 20770
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.977403508294906e-06,
-      "loss": 0.7665,
+      "loss": 0.9075,
       "step": 20771
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.976565012654676e-06,
-      "loss": 0.6796,
+      "loss": 0.7628,
       "step": 20772
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9757266144546415e-06,
-      "loss": 0.6312,
+      "loss": 0.7987,
       "step": 20773
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9748883137064335e-06,
-      "loss": 0.6352,
+      "loss": 0.806,
       "step": 20774
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9740501104216836e-06,
-      "loss": 0.693,
+      "loss": 0.7898,
       "step": 20775
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.973212004612016e-06,
-      "loss": 0.6614,
+      "loss": 0.7952,
       "step": 20776
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9723739962890684e-06,
-      "loss": 0.6816,
+      "loss": 0.7351,
       "step": 20777
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9715360854644546e-06,
-      "loss": 0.7855,
+      "loss": 0.8168,
       "step": 20778
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.970698272149809e-06,
-      "loss": 0.5137,
+      "loss": 0.7022,
       "step": 20779
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9698605563567506e-06,
-      "loss": 0.5956,
+      "loss": 0.6492,
       "step": 20780
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.969022938096904e-06,
-      "loss": 0.6105,
+      "loss": 0.8082,
       "step": 20781
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9681854173818845e-06,
-      "loss": 0.7648,
+      "loss": 0.8398,
       "step": 20782
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9673479942233217e-06,
-      "loss": 0.6742,
+      "loss": 0.7131,
       "step": 20783
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9665106686328217e-06,
-      "loss": 0.8253,
+      "loss": 0.8268,
       "step": 20784
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.965673440622009e-06,
-      "loss": 0.8081,
+      "loss": 0.8522,
       "step": 20785
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.964836310202498e-06,
-      "loss": 0.6459,
+      "loss": 0.7302,
       "step": 20786
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9639992773859016e-06,
-      "loss": 0.7667,
+      "loss": 0.8367,
       "step": 20787
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9631623421838284e-06,
-      "loss": 0.66,
+      "loss": 0.7648,
       "step": 20788
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.962325504607897e-06,
-      "loss": 0.6266,
+      "loss": 0.7523,
       "step": 20789
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9614887646697153e-06,
-      "loss": 0.7128,
+      "loss": 0.8002,
       "step": 20790
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.960652122380889e-06,
-      "loss": 0.8099,
+      "loss": 0.865,
       "step": 20791
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.959815577753028e-06,
-      "loss": 0.6763,
+      "loss": 0.8196,
       "step": 20792
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9589791307977376e-06,
-      "loss": 0.6719,
+      "loss": 0.792,
       "step": 20793
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9581427815266205e-06,
-      "loss": 0.6921,
+      "loss": 0.7561,
       "step": 20794
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9573065299512793e-06,
-      "loss": 0.6618,
+      "loss": 0.7691,
       "step": 20795
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9564703760833236e-06,
-      "loss": 0.758,
+      "loss": 0.838,
       "step": 20796
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.955634319934342e-06,
-      "loss": 0.5519,
+      "loss": 0.729,
       "step": 20797
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9547983615159413e-06,
-      "loss": 0.6721,
+      "loss": 0.7219,
       "step": 20798
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9539625008397187e-06,
-      "loss": 0.7059,
+      "loss": 0.8546,
       "step": 20799
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.953126737917269e-06,
-      "loss": 0.8385,
+      "loss": 0.9277,
       "step": 20800
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.952291072760185e-06,
-      "loss": 0.6439,
+      "loss": 0.8056,
       "step": 20801
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9514555053800686e-06,
-      "loss": 0.6968,
+      "loss": 0.8209,
       "step": 20802
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9506200357885005e-06,
-      "loss": 0.6982,
+      "loss": 0.8312,
       "step": 20803
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9497846639970807e-06,
-      "loss": 0.7463,
+      "loss": 0.8171,
       "step": 20804
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.948949390017396e-06,
-      "loss": 0.8019,
+      "loss": 0.9179,
       "step": 20805
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.948114213861034e-06,
-      "loss": 0.8579,
+      "loss": 0.8889,
       "step": 20806
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.947279135539579e-06,
-      "loss": 0.7903,
+      "loss": 0.8508,
       "step": 20807
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9464441550646227e-06,
-      "loss": 0.6029,
+      "loss": 0.7203,
       "step": 20808
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.945609272447746e-06,
-      "loss": 0.6515,
+      "loss": 0.7711,
       "step": 20809
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9447744877005323e-06,
-      "loss": 0.6982,
+      "loss": 0.7689,
       "step": 20810
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.943939800834562e-06,
-      "loss": 0.767,
+      "loss": 0.8657,
       "step": 20811
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9431052118614158e-06,
-      "loss": 0.7287,
+      "loss": 0.8047,
       "step": 20812
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9422707207926715e-06,
-      "loss": 0.7386,
+      "loss": 0.8179,
       "step": 20813
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.941436327639906e-06,
-      "loss": 0.7215,
+      "loss": 0.7656,
       "step": 20814
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.940602032414701e-06,
-      "loss": 0.7314,
+      "loss": 0.7845,
       "step": 20815
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.939767835128622e-06,
-      "loss": 0.5951,
+      "loss": 0.7177,
       "step": 20816
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.93893373579325e-06,
-      "loss": 0.7939,
+      "loss": 0.899,
       "step": 20817
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9380997344201533e-06,
-      "loss": 0.5952,
+      "loss": 0.72,
       "step": 20818
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9372658310209036e-06,
-      "loss": 0.7428,
+      "loss": 0.7894,
       "step": 20819
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9364320256070676e-06,
-      "loss": 0.6575,
+      "loss": 0.766,
       "step": 20820
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.935598318190217e-06,
-      "loss": 0.7117,
+      "loss": 0.8331,
       "step": 20821
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9347647087819174e-06,
-      "loss": 0.6337,
+      "loss": 0.7684,
       "step": 20822
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9339311973937335e-06,
-      "loss": 0.8318,
+      "loss": 0.9195,
       "step": 20823
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9330977840372287e-06,
-      "loss": 0.7036,
+      "loss": 0.8285,
       "step": 20824
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9322644687239656e-06,
-      "loss": 0.7775,
+      "loss": 0.8817,
       "step": 20825
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.931431251465501e-06,
-      "loss": 0.7952,
+      "loss": 0.8825,
       "step": 20826
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9305981322734035e-06,
-      "loss": 0.783,
+      "loss": 0.876,
       "step": 20827
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9297651111592253e-06,
-      "loss": 0.7255,
+      "loss": 0.8741,
       "step": 20828
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9289321881345257e-06,
-      "loss": 0.7946,
+      "loss": 0.7889,
       "step": 20829
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.928099363210859e-06,
-      "loss": 0.5983,
+      "loss": 0.8094,
       "step": 20830
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9272666363997805e-06,
-      "loss": 0.7569,
+      "loss": 0.8325,
       "step": 20831
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9264340077128416e-06,
-      "loss": 0.7211,
+      "loss": 0.8157,
       "step": 20832
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9256014771615924e-06,
-      "loss": 0.8441,
+      "loss": 0.9158,
       "step": 20833
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9247690447575883e-06,
-      "loss": 0.5541,
+      "loss": 0.7653,
       "step": 20834
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.923936710512376e-06,
-      "loss": 0.6596,
+      "loss": 0.7582,
       "step": 20835
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9231044744375014e-06,
-      "loss": 0.6494,
+      "loss": 0.7521,
       "step": 20836
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9222723365445117e-06,
-      "loss": 0.7315,
+      "loss": 0.784,
       "step": 20837
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9214402968449507e-06,
-      "loss": 0.7025,
+      "loss": 0.8239,
       "step": 20838
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9206083553503607e-06,
-      "loss": 0.6851,
+      "loss": 0.808,
       "step": 20839
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.919776512072288e-06,
-      "loss": 0.7341,
+      "loss": 0.8154,
       "step": 20840
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9189447670222693e-06,
-      "loss": 0.6578,
+      "loss": 0.7802,
       "step": 20841
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.918113120211846e-06,
-      "loss": 0.8259,
+      "loss": 0.9026,
       "step": 20842
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9172815716525527e-06,
-      "loss": 0.7187,
+      "loss": 0.8114,
       "step": 20843
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.916450121355935e-06,
-      "loss": 0.7495,
+      "loss": 0.7551,
       "step": 20844
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.915618769333515e-06,
-      "loss": 0.7177,
+      "loss": 0.8335,
       "step": 20845
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9147875155968365e-06,
-      "loss": 0.6445,
+      "loss": 0.7549,
       "step": 20846
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.913956360157427e-06,
-      "loss": 0.7055,
+      "loss": 0.8527,
       "step": 20847
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9131253030268214e-06,
-      "loss": 0.7041,
+      "loss": 0.8077,
       "step": 20848
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9122943442165464e-06,
-      "loss": 0.6891,
+      "loss": 0.8319,
       "step": 20849
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9114634837381305e-06,
-      "loss": 0.6845,
+      "loss": 0.7487,
       "step": 20850
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.910632721603103e-06,
-      "loss": 0.646,
+      "loss": 0.745,
       "step": 20851
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9098020578229834e-06,
-      "loss": 0.7548,
+      "loss": 0.8416,
       "step": 20852
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.908971492409305e-06,
-      "loss": 0.7055,
+      "loss": 0.8146,
       "step": 20853
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9081410253735854e-06,
-      "loss": 0.724,
+      "loss": 0.8146,
       "step": 20854
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.907310656727347e-06,
-      "loss": 0.7178,
+      "loss": 0.7975,
       "step": 20855
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9064803864821077e-06,
-      "loss": 0.5847,
+      "loss": 0.733,
       "step": 20856
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9056502146493935e-06,
-      "loss": 0.691,
+      "loss": 0.8571,
       "step": 20857
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9048201412407118e-06,
-      "loss": 0.6717,
+      "loss": 0.727,
       "step": 20858
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.903990166267587e-06,
-      "loss": 0.7119,
+      "loss": 0.7932,
       "step": 20859
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.903160289741531e-06,
-      "loss": 0.691,
+      "loss": 0.8743,
       "step": 20860
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9023305116740565e-06,
-      "loss": 0.7886,
+      "loss": 0.7953,
       "step": 20861
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9015008320766723e-06,
-      "loss": 0.637,
+      "loss": 0.7579,
       "step": 20862
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.9006712509608993e-06,
-      "loss": 0.6771,
+      "loss": 0.8007,
       "step": 20863
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.899841768338234e-06,
-      "loss": 0.7006,
+      "loss": 0.8276,
       "step": 20864
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8990123842201936e-06,
-      "loss": 0.6216,
+      "loss": 0.7479,
       "step": 20865
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.89818309861828e-06,
-      "loss": 0.6032,
+      "loss": 0.7336,
       "step": 20866
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8973539115440008e-06,
-      "loss": 0.7271,
+      "loss": 0.8618,
       "step": 20867
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.896524823008859e-06,
-      "loss": 0.7626,
+      "loss": 0.8484,
       "step": 20868
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.895695833024357e-06,
-      "loss": 0.7189,
+      "loss": 0.8197,
       "step": 20869
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8948669416019946e-06,
-      "loss": 0.7092,
+      "loss": 0.7739,
       "step": 20870
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.89403814875327e-06,
-      "loss": 0.8295,
+      "loss": 0.8754,
       "step": 20871
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.893209454489687e-06,
-      "loss": 0.7059,
+      "loss": 0.8287,
       "step": 20872
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.89238085882274e-06,
-      "loss": 0.5498,
+      "loss": 0.6781,
       "step": 20873
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.891552361763924e-06,
-      "loss": 0.7129,
+      "loss": 0.7891,
       "step": 20874
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.89072396332473e-06,
-      "loss": 0.6547,
+      "loss": 0.7695,
       "step": 20875
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.889895663516661e-06,
-      "loss": 0.7118,
+      "loss": 0.8985,
       "step": 20876
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.889067462351195e-06,
-      "loss": 0.598,
+      "loss": 0.7806,
       "step": 20877
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8882393598398318e-06,
-      "loss": 0.6614,
+      "loss": 0.7715,
       "step": 20878
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8874113559940574e-06,
-      "loss": 0.7502,
+      "loss": 0.7344,
       "step": 20879
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.88658345082536e-06,
-      "loss": 0.7364,
+      "loss": 0.8499,
       "step": 20880
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8857556443452206e-06,
-      "loss": 0.6551,
+      "loss": 0.8004,
       "step": 20881
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.884927936565134e-06,
-      "loss": 0.6864,
+      "loss": 0.7843,
       "step": 20882
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8841003274965706e-06,
-      "loss": 0.6763,
+      "loss": 0.7232,
       "step": 20883
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8832728171510225e-06,
-      "loss": 0.6381,
+      "loss": 0.7878,
       "step": 20884
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8824454055399674e-06,
-      "loss": 0.7152,
+      "loss": 0.7912,
       "step": 20885
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8816180926748837e-06,
-      "loss": 0.6882,
+      "loss": 0.8351,
       "step": 20886
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.880790878567249e-06,
-      "loss": 0.7892,
+      "loss": 0.8072,
       "step": 20887
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.879963763228537e-06,
-      "loss": 0.7521,
+      "loss": 0.7907,
       "step": 20888
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8791367466702315e-06,
-      "loss": 0.712,
+      "loss": 0.8223,
       "step": 20889
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.878309828903795e-06,
-      "loss": 0.6866,
+      "loss": 0.6945,
       "step": 20890
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.877483009940708e-06,
-      "loss": 0.6673,
+      "loss": 0.794,
       "step": 20891
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8766562897924388e-06,
-      "loss": 0.7584,
+      "loss": 0.8069,
       "step": 20892
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8758296684704566e-06,
-      "loss": 0.661,
+      "loss": 0.7549,
       "step": 20893
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.875003145986227e-06,
-      "loss": 0.7219,
+      "loss": 0.8153,
       "step": 20894
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8741767223512264e-06,
-      "loss": 0.7269,
+      "loss": 0.7997,
       "step": 20895
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8733503975769064e-06,
-      "loss": 0.7282,
+      "loss": 0.8363,
       "step": 20896
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.872524171674741e-06,
-      "loss": 0.6543,
+      "loss": 0.759,
       "step": 20897
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8716980446561914e-06,
-      "loss": 0.7028,
+      "loss": 0.7795,
       "step": 20898
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8708720165327177e-06,
-      "loss": 0.6652,
+      "loss": 0.7816,
       "step": 20899
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.870046087315775e-06,
-      "loss": 0.7607,
+      "loss": 0.7858,
       "step": 20900
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8692202570168315e-06,
-      "loss": 0.6279,
+      "loss": 0.8138,
       "step": 20901
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.86839452564734e-06,
-      "loss": 0.6752,
+      "loss": 0.821,
       "step": 20902
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.867568893218755e-06,
-      "loss": 0.6869,
+      "loss": 0.794,
       "step": 20903
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8667433597425322e-06,
-      "loss": 0.8269,
+      "loss": 0.9049,
       "step": 20904
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.865917925230125e-06,
-      "loss": 0.6812,
+      "loss": 0.7825,
       "step": 20905
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.865092589692984e-06,
-      "loss": 0.7314,
+      "loss": 0.8904,
       "step": 20906
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.864267353142558e-06,
-      "loss": 0.6821,
+      "loss": 0.7823,
       "step": 20907
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.863442215590303e-06,
-      "loss": 0.723,
+      "loss": 0.7676,
       "step": 20908
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8626171770476573e-06,
-      "loss": 0.7301,
+      "loss": 0.8138,
       "step": 20909
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.861792237526073e-06,
-      "loss": 0.6902,
+      "loss": 0.8256,
       "step": 20910
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8609673970369954e-06,
-      "loss": 0.6579,
+      "loss": 0.756,
       "step": 20911
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8601426555918653e-06,
-      "loss": 0.8873,
+      "loss": 0.9224,
       "step": 20912
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8593180132021227e-06,
-      "loss": 0.6712,
+      "loss": 0.7628,
       "step": 20913
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8584934698792166e-06,
-      "loss": 0.6124,
+      "loss": 0.7514,
       "step": 20914
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8576690256345763e-06,
-      "loss": 0.5991,
+      "loss": 0.6718,
       "step": 20915
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.856844680479647e-06,
-      "loss": 0.7507,
+      "loss": 0.8095,
       "step": 20916
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8560204344258637e-06,
-      "loss": 0.6818,
+      "loss": 0.7506,
       "step": 20917
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.855196287484661e-06,
-      "loss": 0.7007,
+      "loss": 0.7676,
       "step": 20918
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.854372239667468e-06,
-      "loss": 0.6864,
+      "loss": 0.7999,
       "step": 20919
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.853548290985726e-06,
-      "loss": 0.6574,
+      "loss": 0.7799,
       "step": 20920
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8527244414508615e-06,
-      "loss": 0.7622,
+      "loss": 0.8563,
       "step": 20921
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8519006910743054e-06,
-      "loss": 0.527,
+      "loss": 0.6756,
       "step": 20922
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.851077039867485e-06,
-      "loss": 0.736,
+      "loss": 0.818,
       "step": 20923
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8502534878418275e-06,
-      "loss": 0.6028,
+      "loss": 0.7559,
       "step": 20924
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8494300350087587e-06,
-      "loss": 0.6675,
+      "loss": 0.7496,
       "step": 20925
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.848606681379701e-06,
-      "loss": 0.647,
+      "loss": 0.7499,
       "step": 20926
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.847783426966084e-06,
-      "loss": 0.7596,
+      "loss": 0.826,
       "step": 20927
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8469602717793177e-06,
-      "loss": 0.7246,
+      "loss": 0.8307,
       "step": 20928
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.846137215830833e-06,
-      "loss": 0.78,
+      "loss": 0.8191,
       "step": 20929
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.845314259132044e-06,
-      "loss": 0.6892,
+      "loss": 0.7428,
       "step": 20930
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.844491401694368e-06,
-      "loss": 0.8585,
+      "loss": 0.8501,
       "step": 20931
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8436686435292193e-06,
-      "loss": 0.7088,
+      "loss": 0.8355,
       "step": 20932
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.842845984648017e-06,
-      "loss": 0.7167,
+      "loss": 0.7555,
       "step": 20933
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8420234250621724e-06,
-      "loss": 0.8152,
+      "loss": 0.8313,
       "step": 20934
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8412009647830973e-06,
-      "loss": 0.6909,
+      "loss": 0.7831,
       "step": 20935
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8403786038222013e-06,
-      "loss": 0.6662,
+      "loss": 0.7942,
       "step": 20936
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8395563421908934e-06,
-      "loss": 0.6796,
+      "loss": 0.7889,
       "step": 20937
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.83873417990058e-06,
-      "loss": 0.6427,
+      "loss": 0.6685,
       "step": 20938
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.837912116962671e-06,
-      "loss": 0.7116,
+      "loss": 0.7221,
       "step": 20939
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.837090153388571e-06,
-      "loss": 0.8141,
+      "loss": 0.8806,
       "step": 20940
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.836268289189682e-06,
-      "loss": 0.7072,
+      "loss": 0.827,
       "step": 20941
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8354465243774034e-06,
-      "loss": 0.6981,
+      "loss": 0.8871,
       "step": 20942
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.834624858963145e-06,
-      "loss": 0.6947,
+      "loss": 0.7876,
       "step": 20943
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.833803292958297e-06,
-      "loss": 0.7476,
+      "loss": 0.8408,
       "step": 20944
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8329818263742583e-06,
-      "loss": 0.7671,
+      "loss": 0.8433,
       "step": 20945
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.832160459222434e-06,
-      "loss": 0.8332,
+      "loss": 0.8641,
       "step": 20946
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8313391915142073e-06,
-      "loss": 0.6305,
+      "loss": 0.767,
       "step": 20947
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.830518023260982e-06,
-      "loss": 0.6353,
+      "loss": 0.7622,
       "step": 20948
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.829696954474146e-06,
-      "loss": 0.7163,
+      "loss": 0.7547,
       "step": 20949
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8288759851650915e-06,
-      "loss": 0.6982,
+      "loss": 0.7876,
       "step": 20950
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.828055115345206e-06,
-      "loss": 0.6943,
+      "loss": 0.8256,
       "step": 20951
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8272343450258834e-06,
-      "loss": 0.6642,
+      "loss": 0.8313,
       "step": 20952
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8264136742185066e-06,
-      "loss": 0.6525,
+      "loss": 0.7502,
       "step": 20953
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.825593102934462e-06,
-      "loss": 0.7048,
+      "loss": 0.8076,
       "step": 20954
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.824772631185131e-06,
-      "loss": 0.7323,
+      "loss": 0.8958,
       "step": 20955
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.823952258981906e-06,
-      "loss": 0.6058,
+      "loss": 0.7021,
       "step": 20956
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.823131986336155e-06,
-      "loss": 0.7544,
+      "loss": 0.8013,
       "step": 20957
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8223118132592688e-06,
-      "loss": 0.7119,
+      "loss": 0.75,
       "step": 20958
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8214917397626217e-06,
-      "loss": 0.7122,
+      "loss": 0.7179,
       "step": 20959
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.820671765857592e-06,
-      "loss": 0.7162,
+      "loss": 0.8255,
       "step": 20960
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8198518915555528e-06,
-      "loss": 0.5757,
+      "loss": 0.7005,
       "step": 20961
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.819032116867887e-06,
-      "loss": 0.6897,
+      "loss": 0.7921,
       "step": 20962
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8182124418059566e-06,
-      "loss": 0.6671,
+      "loss": 0.7153,
       "step": 20963
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.817392866381141e-06,
-      "loss": 0.8122,
+      "loss": 0.817,
       "step": 20964
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.81657339060481e-06,
-      "loss": 0.6141,
+      "loss": 0.7524,
       "step": 20965
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.81575401448833e-06,
-      "loss": 0.6624,
+      "loss": 0.8174,
       "step": 20966
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8149347380430703e-06,
-      "loss": 0.7092,
+      "loss": 0.8275,
       "step": 20967
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.814115561280397e-06,
-      "loss": 0.607,
+      "loss": 0.7897,
       "step": 20968
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.813296484211675e-06,
-      "loss": 0.7533,
+      "loss": 0.7641,
       "step": 20969
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8124775068482646e-06,
-      "loss": 0.7725,
+      "loss": 0.8899,
       "step": 20970
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8116586292015336e-06,
-      "loss": 0.8558,
+      "loss": 0.8985,
       "step": 20971
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8108398512828405e-06,
-      "loss": 0.734,
+      "loss": 0.8847,
       "step": 20972
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.810021173103544e-06,
-      "loss": 0.8197,
+      "loss": 0.8867,
       "step": 20973
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.809202594675e-06,
-      "loss": 0.6844,
+      "loss": 0.7759,
       "step": 20974
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8083841160085733e-06,
-      "loss": 0.7644,
+      "loss": 0.7757,
       "step": 20975
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8075657371156074e-06,
-      "loss": 0.8696,
+      "loss": 0.9686,
       "step": 20976
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.806747458007465e-06,
-      "loss": 0.6304,
+      "loss": 0.7248,
       "step": 20977
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.805929278695496e-06,
-      "loss": 0.771,
+      "loss": 0.8277,
       "step": 20978
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8051111991910518e-06,
-      "loss": 0.7298,
+      "loss": 0.8034,
       "step": 20979
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.804293219505477e-06,
-      "loss": 0.63,
+      "loss": 0.8192,
       "step": 20980
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8034753396501325e-06,
-      "loss": 0.5621,
+      "loss": 0.7654,
       "step": 20981
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8026575596363493e-06,
-      "loss": 0.718,
+      "loss": 0.8291,
       "step": 20982
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.801839879475484e-06,
-      "loss": 0.6303,
+      "loss": 0.7581,
       "step": 20983
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.801022299178878e-06,
-      "loss": 0.7536,
+      "loss": 0.8149,
       "step": 20984
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.8002048187578745e-06,
-      "loss": 0.6744,
+      "loss": 0.7708,
       "step": 20985
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.799387438223813e-06,
-      "loss": 0.7517,
+      "loss": 0.8026,
       "step": 20986
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.798570157588031e-06,
-      "loss": 0.695,
+      "loss": 0.7987,
       "step": 20987
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.797752976861877e-06,
-      "loss": 0.8367,
+      "loss": 0.8382,
       "step": 20988
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7969358960566763e-06,
-      "loss": 0.8046,
+      "loss": 0.8629,
       "step": 20989
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.796118915183772e-06,
-      "loss": 0.7498,
+      "loss": 0.7885,
       "step": 20990
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.795302034254498e-06,
-      "loss": 0.813,
+      "loss": 0.9029,
       "step": 20991
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7944852532801857e-06,
-      "loss": 0.6839,
+      "loss": 0.7758,
       "step": 20992
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7936685722721648e-06,
-      "loss": 0.7323,
+      "loss": 0.7984,
       "step": 20993
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.792851991241773e-06,
-      "loss": 0.8071,
+      "loss": 0.862,
       "step": 20994
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7920355102003283e-06,
-      "loss": 0.6349,
+      "loss": 0.7594,
       "step": 20995
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.791219129159167e-06,
-      "loss": 0.5181,
+      "loss": 0.7403,
       "step": 20996
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.790402848129613e-06,
-      "loss": 0.8524,
+      "loss": 0.8681,
       "step": 20997
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.789586667122991e-06,
-      "loss": 0.8336,
+      "loss": 0.8418,
       "step": 20998
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.78877058615062e-06,
-      "loss": 0.7754,
+      "loss": 0.8458,
       "step": 20999
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7879546052238282e-06,
-      "loss": 0.7165,
+      "loss": 0.8117,
       "step": 21000
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.787138724353934e-06,
-      "loss": 0.901,
+      "loss": 0.9564,
       "step": 21001
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7863229435522564e-06,
-      "loss": 0.7361,
+      "loss": 0.7998,
       "step": 21002
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.785507262830114e-06,
-      "loss": 0.868,
+      "loss": 0.8812,
       "step": 21003
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7846916821988214e-06,
-      "loss": 0.7612,
+      "loss": 0.8107,
       "step": 21004
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.783876201669694e-06,
-      "loss": 0.6786,
+      "loss": 0.7749,
       "step": 21005
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.783060821254043e-06,
-      "loss": 0.6603,
+      "loss": 0.7596,
       "step": 21006
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.782245540963191e-06,
-      "loss": 0.6857,
+      "loss": 0.7836,
       "step": 21007
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7814303608084347e-06,
-      "loss": 0.825,
+      "loss": 0.9075,
       "step": 21008
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7806152808010933e-06,
-      "loss": 0.5901,
+      "loss": 0.7849,
       "step": 21009
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.779800300952472e-06,
-      "loss": 0.6535,
+      "loss": 0.8055,
       "step": 21010
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7789854212738775e-06,
-      "loss": 0.685,
+      "loss": 0.824,
       "step": 21011
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.778170641776612e-06,
-      "loss": 0.7258,
+      "loss": 0.7976,
       "step": 21012
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.777355962471988e-06,
-      "loss": 0.63,
+      "loss": 0.7475,
       "step": 21013
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7765413833712964e-06,
-      "loss": 0.7032,
+      "loss": 0.7595,
       "step": 21014
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7757269044858472e-06,
-      "loss": 0.7511,
+      "loss": 0.8539,
       "step": 21015
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7749125258269383e-06,
-      "loss": 0.7922,
+      "loss": 0.8961,
       "step": 21016
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7740982474058664e-06,
-      "loss": 0.6305,
+      "loss": 0.7802,
       "step": 21017
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.773284069233926e-06,
-      "loss": 0.6046,
+      "loss": 0.7678,
       "step": 21018
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7724699913224187e-06,
-      "loss": 0.7714,
+      "loss": 0.864,
       "step": 21019
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7716560136826354e-06,
-      "loss": 0.6606,
+      "loss": 0.7916,
       "step": 21020
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.77084213632587e-06,
-      "loss": 0.6364,
+      "loss": 0.7016,
       "step": 21021
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7700283592634135e-06,
-      "loss": 0.6956,
+      "loss": 0.7649,
       "step": 21022
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7692146825065547e-06,
-      "loss": 0.7484,
+      "loss": 0.8042,
       "step": 21023
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7684011060665827e-06,
-      "loss": 0.7205,
+      "loss": 0.8592,
       "step": 21024
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.767587629954783e-06,
-      "loss": 0.629,
+      "loss": 0.7952,
       "step": 21025
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7667742541824495e-06,
-      "loss": 0.6767,
+      "loss": 0.8614,
       "step": 21026
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7659609787608545e-06,
-      "loss": 0.5592,
+      "loss": 0.6992,
       "step": 21027
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.765147803701289e-06,
-      "loss": 0.6945,
+      "loss": 0.7537,
       "step": 21028
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7643347290150336e-06,
-      "loss": 0.5652,
+      "loss": 0.7837,
       "step": 21029
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7635217547133684e-06,
-      "loss": 0.7475,
+      "loss": 0.8499,
       "step": 21030
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7627088808075674e-06,
-      "loss": 0.7811,
+      "loss": 0.8535,
       "step": 21031
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.761896107308917e-06,
-      "loss": 0.7363,
+      "loss": 0.738,
       "step": 21032
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7610834342286874e-06,
-      "loss": 0.8598,
+      "loss": 0.9299,
       "step": 21033
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.7602708615781548e-06,
-      "loss": 0.7673,
+      "loss": 0.8269,
       "step": 21034
     },
     {
       "epoch": 0.76,
       "learning_rate": 2.759458389368591e-06,
-      "loss": 0.6542,
+      "loss": 0.8412,
       "step": 21035
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7586460176112707e-06,
-      "loss": 0.6673,
+      "loss": 0.7328,
       "step": 21036
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7578337463174585e-06,
-      "loss": 0.7416,
+      "loss": 0.7634,
       "step": 21037
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.757021575498432e-06,
-      "loss": 0.6659,
+      "loss": 0.8295,
       "step": 21038
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.756209505165455e-06,
-      "loss": 0.7538,
+      "loss": 0.8807,
       "step": 21039
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7553975353297922e-06,
-      "loss": 0.6865,
+      "loss": 0.7769,
       "step": 21040
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7545856660027102e-06,
-      "loss": 0.7169,
+      "loss": 0.8664,
       "step": 21041
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7537738971954718e-06,
-      "loss": 0.7133,
+      "loss": 0.7688,
       "step": 21042
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7529622289193403e-06,
-      "loss": 0.7072,
+      "loss": 0.7622,
       "step": 21043
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.752150661185572e-06,
-      "loss": 0.819,
+      "loss": 0.8334,
       "step": 21044
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7513391940054336e-06,
-      "loss": 0.7101,
+      "loss": 0.8324,
       "step": 21045
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7505278273901794e-06,
-      "loss": 0.6551,
+      "loss": 0.73,
       "step": 21046
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.749716561351066e-06,
-      "loss": 0.5976,
+      "loss": 0.7672,
       "step": 21047
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7489053958993472e-06,
-      "loss": 0.8287,
+      "loss": 0.8111,
       "step": 21048
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.748094331046279e-06,
-      "loss": 0.6394,
+      "loss": 0.7224,
       "step": 21049
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.747283366803111e-06,
-      "loss": 0.5973,
+      "loss": 0.7745,
       "step": 21050
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7464725031810967e-06,
-      "loss": 0.7701,
+      "loss": 0.8592,
       "step": 21051
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7456617401914866e-06,
-      "loss": 0.6572,
+      "loss": 0.8156,
       "step": 21052
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7448510778455274e-06,
-      "loss": 0.7476,
+      "loss": 0.8145,
       "step": 21053
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7440405161544615e-06,
-      "loss": 0.7491,
+      "loss": 0.8238,
       "step": 21054
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7432300551295453e-06,
-      "loss": 0.755,
+      "loss": 0.8707,
       "step": 21055
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7424196947820103e-06,
-      "loss": 0.7148,
+      "loss": 0.7996,
       "step": 21056
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7416094351231084e-06,
-      "loss": 0.6922,
+      "loss": 0.767,
       "step": 21057
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7407992761640758e-06,
-      "loss": 0.6539,
+      "loss": 0.7491,
       "step": 21058
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.739989217916155e-06,
-      "loss": 0.7226,
+      "loss": 0.8361,
       "step": 21059
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.739179260390582e-06,
-      "loss": 0.6801,
+      "loss": 0.7879,
       "step": 21060
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7383694035985965e-06,
-      "loss": 0.698,
+      "loss": 0.856,
       "step": 21061
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.737559647551432e-06,
-      "loss": 0.6495,
+      "loss": 0.7309,
       "step": 21062
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7367499922603204e-06,
-      "loss": 0.6528,
+      "loss": 0.7505,
       "step": 21063
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7359404377364994e-06,
-      "loss": 0.725,
+      "loss": 0.7245,
       "step": 21064
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7351309839911998e-06,
-      "loss": 0.5906,
+      "loss": 0.8102,
       "step": 21065
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7343216310356504e-06,
-      "loss": 0.6021,
+      "loss": 0.7795,
       "step": 21066
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.733512378881078e-06,
-      "loss": 0.8102,
+      "loss": 0.8104,
       "step": 21067
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7327032275387134e-06,
-      "loss": 0.7563,
+      "loss": 0.8408,
       "step": 21068
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.731894177019777e-06,
-      "loss": 0.7162,
+      "loss": 0.779,
       "step": 21069
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7310852273354995e-06,
-      "loss": 0.5372,
+      "loss": 0.675,
       "step": 21070
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7302763784971008e-06,
-      "loss": 0.6918,
+      "loss": 0.8541,
       "step": 21071
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.729467630515804e-06,
-      "loss": 0.7217,
+      "loss": 0.872,
       "step": 21072
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7286589834028245e-06,
-      "loss": 0.7381,
+      "loss": 0.8026,
       "step": 21073
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7278504371693916e-06,
-      "loss": 0.5533,
+      "loss": 0.7165,
       "step": 21074
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.72704199182671e-06,
-      "loss": 0.6719,
+      "loss": 0.791,
       "step": 21075
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.726233647386004e-06,
-      "loss": 0.7222,
+      "loss": 0.8505,
       "step": 21076
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.725425403858487e-06,
-      "loss": 0.5777,
+      "loss": 0.7392,
       "step": 21077
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.724617261255371e-06,
-      "loss": 0.7518,
+      "loss": 0.7275,
       "step": 21078
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7238092195878673e-06,
-      "loss": 0.8425,
+      "loss": 0.8607,
       "step": 21079
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.723001278867188e-06,
-      "loss": 0.8457,
+      "loss": 0.8443,
       "step": 21080
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.722193439104541e-06,
-      "loss": 0.7638,
+      "loss": 0.8194,
       "step": 21081
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.721385700311131e-06,
-      "loss": 0.8068,
+      "loss": 0.8028,
       "step": 21082
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7205780624981702e-06,
-      "loss": 0.6099,
+      "loss": 0.7828,
       "step": 21083
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7197705256768614e-06,
-      "loss": 0.7901,
+      "loss": 0.8482,
       "step": 21084
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7189630898584065e-06,
-      "loss": 0.7315,
+      "loss": 0.8737,
       "step": 21085
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.718155755054006e-06,
-      "loss": 0.7246,
+      "loss": 0.7971,
       "step": 21086
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7173485212748685e-06,
-      "loss": 0.6649,
+      "loss": 0.7484,
       "step": 21087
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.716541388532181e-06,
-      "loss": 0.888,
+      "loss": 0.9268,
       "step": 21088
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.715734356837151e-06,
-      "loss": 0.6242,
+      "loss": 0.7018,
       "step": 21089
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7149274262009728e-06,
-      "loss": 0.7477,
+      "loss": 0.8259,
       "step": 21090
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7141205966348395e-06,
-      "loss": 0.7174,
+      "loss": 0.8385,
       "step": 21091
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.713313868149943e-06,
-      "loss": 0.7858,
+      "loss": 0.8241,
       "step": 21092
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7125072407574836e-06,
-      "loss": 0.7134,
+      "loss": 0.7025,
       "step": 21093
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.71170071446864e-06,
-      "loss": 0.8007,
+      "loss": 0.9507,
       "step": 21094
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7108942892946134e-06,
-      "loss": 0.6712,
+      "loss": 0.7763,
       "step": 21095
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.710087965246585e-06,
-      "loss": 0.7175,
+      "loss": 0.7936,
       "step": 21096
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.709281742335743e-06,
-      "loss": 0.7237,
+      "loss": 0.8635,
       "step": 21097
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7084756205732732e-06,
-      "loss": 0.7333,
+      "loss": 0.8335,
       "step": 21098
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.707669599970356e-06,
-      "loss": 0.6877,
+      "loss": 0.782,
       "step": 21099
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.706863680538182e-06,
-      "loss": 0.8009,
+      "loss": 0.8063,
       "step": 21100
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.706057862287922e-06,
-      "loss": 0.7173,
+      "loss": 0.7486,
       "step": 21101
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7052521452307624e-06,
-      "loss": 0.7553,
+      "loss": 0.7579,
       "step": 21102
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.70444652937788e-06,
-      "loss": 0.7745,
+      "loss": 0.879,
       "step": 21103
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.70364101474045e-06,
-      "loss": 0.7176,
+      "loss": 0.7536,
       "step": 21104
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7028356013296462e-06,
-      "loss": 0.7082,
+      "loss": 0.8314,
       "step": 21105
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7020302891566518e-06,
-      "loss": 0.7628,
+      "loss": 0.8367,
       "step": 21106
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.7012250782326255e-06,
-      "loss": 0.7107,
+      "loss": 0.8263,
       "step": 21107
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.700419968568749e-06,
-      "loss": 0.7361,
+      "loss": 0.8146,
       "step": 21108
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.69961496017619e-06,
-      "loss": 0.8901,
+      "loss": 0.8976,
       "step": 21109
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.698810053066114e-06,
-      "loss": 0.7612,
+      "loss": 0.737,
       "step": 21110
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6980052472496874e-06,
-      "loss": 0.7566,
+      "loss": 0.8043,
       "step": 21111
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.697200542738083e-06,
-      "loss": 0.7012,
+      "loss": 0.758,
       "step": 21112
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.696395939542454e-06,
-      "loss": 0.7701,
+      "loss": 0.8466,
       "step": 21113
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.695591437673972e-06,
-      "loss": 0.7254,
+      "loss": 0.7927,
       "step": 21114
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6947870371437956e-06,
-      "loss": 0.676,
+      "loss": 0.7373,
       "step": 21115
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6939827379630836e-06,
-      "loss": 0.6384,
+      "loss": 0.7296,
       "step": 21116
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6931785401429957e-06,
-      "loss": 0.7934,
+      "loss": 0.8669,
       "step": 21117
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.692374443694685e-06,
-      "loss": 0.717,
+      "loss": 0.8142,
       "step": 21118
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.691570448629318e-06,
-      "loss": 0.7932,
+      "loss": 0.8969,
       "step": 21119
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6907665549580342e-06,
-      "loss": 0.7117,
+      "loss": 0.8162,
       "step": 21120
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.689962762691998e-06,
-      "loss": 0.7087,
+      "loss": 0.8016,
       "step": 21121
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6891590718423568e-06,
-      "loss": 0.6691,
+      "loss": 0.7247,
       "step": 21122
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6883554824202605e-06,
-      "loss": 0.562,
+      "loss": 0.8408,
       "step": 21123
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6875519944368543e-06,
-      "loss": 0.6277,
+      "loss": 0.7311,
       "step": 21124
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6867486079032966e-06,
-      "loss": 0.6297,
+      "loss": 0.7518,
       "step": 21125
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6859453228307185e-06,
-      "loss": 0.6689,
+      "loss": 0.7842,
       "step": 21126
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6851421392302756e-06,
-      "loss": 0.6628,
+      "loss": 0.7025,
       "step": 21127
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.684339057113107e-06,
-      "loss": 0.6288,
+      "loss": 0.7533,
       "step": 21128
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.683536076490354e-06,
-      "loss": 0.6267,
+      "loss": 0.7628,
       "step": 21129
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6827331973731542e-06,
-      "loss": 0.7356,
+      "loss": 0.7919,
       "step": 21130
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6819304197726525e-06,
-      "loss": 0.583,
+      "loss": 0.7294,
       "step": 21131
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6811277436999838e-06,
-      "loss": 0.7396,
+      "loss": 0.801,
       "step": 21132
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6803251691662833e-06,
-      "loss": 0.8058,
+      "loss": 0.8499,
       "step": 21133
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.679522696182686e-06,
-      "loss": 0.8089,
+      "loss": 0.8642,
       "step": 21134
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.678720324760324e-06,
-      "loss": 0.7646,
+      "loss": 0.8464,
       "step": 21135
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.677918054910331e-06,
-      "loss": 0.7972,
+      "loss": 0.9218,
       "step": 21136
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.677115886643833e-06,
-      "loss": 0.6372,
+      "loss": 0.798,
       "step": 21137
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6763138199719686e-06,
-      "loss": 0.8098,
+      "loss": 0.8635,
       "step": 21138
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6755118549058522e-06,
-      "loss": 0.5608,
+      "loss": 0.7189,
       "step": 21139
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6747099914566197e-06,
-      "loss": 0.7372,
+      "loss": 0.7968,
       "step": 21140
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6739082296353937e-06,
-      "loss": 0.7949,
+      "loss": 0.8665,
       "step": 21141
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6731065694532955e-06,
-      "loss": 0.7315,
+      "loss": 0.8053,
       "step": 21142
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6723050109214465e-06,
-      "loss": 0.7367,
+      "loss": 0.828,
       "step": 21143
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6715035540509704e-06,
-      "loss": 0.6036,
+      "loss": 0.7202,
       "step": 21144
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6707021988529857e-06,
-      "loss": 0.706,
+      "loss": 0.8038,
       "step": 21145
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6699009453386095e-06,
-      "loss": 0.6722,
+      "loss": 0.7267,
       "step": 21146
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6690997935189567e-06,
-      "loss": 0.7183,
+      "loss": 0.8208,
       "step": 21147
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.668298743405143e-06,
-      "loss": 0.7602,
+      "loss": 0.8175,
       "step": 21148
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6674977950082793e-06,
-      "loss": 0.7763,
+      "loss": 0.8412,
       "step": 21149
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6666969483394834e-06,
-      "loss": 0.7172,
+      "loss": 0.821,
       "step": 21150
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6658962034098624e-06,
-      "loss": 0.8092,
+      "loss": 0.836,
       "step": 21151
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.665095560230526e-06,
-      "loss": 0.632,
+      "loss": 0.7492,
       "step": 21152
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.664295018812578e-06,
-      "loss": 0.7224,
+      "loss": 0.7592,
       "step": 21153
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6634945791671365e-06,
-      "loss": 0.6705,
+      "loss": 0.8367,
       "step": 21154
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6626942413052937e-06,
-      "loss": 0.7703,
+      "loss": 0.8369,
       "step": 21155
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.661894005238156e-06,
-      "loss": 0.8131,
+      "loss": 0.8978,
       "step": 21156
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.661093870976834e-06,
-      "loss": 0.6128,
+      "loss": 0.7203,
       "step": 21157
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.660293838532415e-06,
-      "loss": 0.6807,
+      "loss": 0.8472,
       "step": 21158
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6594939079160087e-06,
-      "loss": 0.6371,
+      "loss": 0.7611,
       "step": 21159
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6586940791387096e-06,
-      "loss": 0.8202,
+      "loss": 0.8083,
       "step": 21160
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.657894352211614e-06,
-      "loss": 0.7285,
+      "loss": 0.8257,
       "step": 21161
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6570947271458146e-06,
-      "loss": 0.7071,
+      "loss": 0.7312,
       "step": 21162
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6562952039524106e-06,
-      "loss": 0.8662,
+      "loss": 0.8553,
       "step": 21163
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.655495782642491e-06,
-      "loss": 0.7853,
+      "loss": 0.8297,
       "step": 21164
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6546964632271464e-06,
-      "loss": 0.6836,
+      "loss": 0.7897,
       "step": 21165
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.653897245717465e-06,
-      "loss": 0.5566,
+      "loss": 0.6589,
       "step": 21166
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6530981301245416e-06,
-      "loss": 0.6465,
+      "loss": 0.7777,
       "step": 21167
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6522991164594525e-06,
-      "loss": 0.7481,
+      "loss": 0.791,
       "step": 21168
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6515002047332907e-06,
-      "loss": 0.5991,
+      "loss": 0.7896,
       "step": 21169
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.650701394957137e-06,
-      "loss": 0.7128,
+      "loss": 0.7333,
       "step": 21170
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6499026871420753e-06,
-      "loss": 0.7514,
+      "loss": 0.8399,
       "step": 21171
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.649104081299181e-06,
-      "loss": 0.8603,
+      "loss": 0.8835,
       "step": 21172
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.648305577439545e-06,
-      "loss": 0.7347,
+      "loss": 0.8351,
       "step": 21173
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6475071755742353e-06,
-      "loss": 0.7732,
+      "loss": 0.8491,
       "step": 21174
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.646708875714329e-06,
-      "loss": 0.6355,
+      "loss": 0.7422,
       "step": 21175
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6459106778709066e-06,
-      "loss": 0.6587,
+      "loss": 0.8224,
       "step": 21176
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6451125820550393e-06,
-      "loss": 0.7096,
+      "loss": 0.7855,
       "step": 21177
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6443145882778e-06,
-      "loss": 0.8126,
+      "loss": 0.8784,
       "step": 21178
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.643516696550259e-06,
-      "loss": 0.719,
+      "loss": 0.7776,
       "step": 21179
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.642718906883487e-06,
-      "loss": 0.6526,
+      "loss": 0.8508,
       "step": 21180
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6419212192885478e-06,
-      "loss": 0.6802,
+      "loss": 0.7928,
       "step": 21181
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6411236337765154e-06,
-      "loss": 0.8204,
+      "loss": 0.9049,
       "step": 21182
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.640326150358451e-06,
-      "loss": 0.7889,
+      "loss": 0.9493,
       "step": 21183
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.639528769045421e-06,
-      "loss": 0.7069,
+      "loss": 0.8439,
       "step": 21184
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.638731489848482e-06,
-      "loss": 0.7697,
+      "loss": 0.8636,
       "step": 21185
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.637934312778705e-06,
-      "loss": 0.6446,
+      "loss": 0.7566,
       "step": 21186
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6371372378471384e-06,
-      "loss": 0.8886,
+      "loss": 0.9801,
       "step": 21187
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6363402650648496e-06,
-      "loss": 0.6875,
+      "loss": 0.7515,
       "step": 21188
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.635543394442891e-06,
-      "loss": 0.783,
+      "loss": 0.8511,
       "step": 21189
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6347466259923194e-06,
-      "loss": 0.666,
+      "loss": 0.7961,
       "step": 21190
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6339499597241856e-06,
-      "loss": 0.7122,
+      "loss": 0.7856,
       "step": 21191
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6331533956495503e-06,
-      "loss": 0.742,
+      "loss": 0.789,
       "step": 21192
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.632356933779454e-06,
-      "loss": 0.7104,
+      "loss": 0.8348,
       "step": 21193
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6315605741249563e-06,
-      "loss": 0.7044,
+      "loss": 0.8252,
       "step": 21194
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6307643166970997e-06,
-      "loss": 0.709,
+      "loss": 0.8596,
       "step": 21195
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.629968161506933e-06,
-      "loss": 0.7198,
+      "loss": 0.8955,
       "step": 21196
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.629172108565501e-06,
-      "loss": 0.7394,
+      "loss": 0.8367,
       "step": 21197
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6283761578838453e-06,
-      "loss": 0.7331,
+      "loss": 0.8142,
       "step": 21198
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6275803094730167e-06,
-      "loss": 0.6924,
+      "loss": 0.7954,
       "step": 21199
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6267845633440457e-06,
-      "loss": 0.6473,
+      "loss": 0.7148,
       "step": 21200
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6259889195079814e-06,
-      "loss": 0.7396,
+      "loss": 0.8726,
       "step": 21201
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.625193377975859e-06,
-      "loss": 0.8143,
+      "loss": 0.935,
       "step": 21202
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.624397938758714e-06,
-      "loss": 0.7745,
+      "loss": 0.8525,
       "step": 21203
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6236026018675796e-06,
-      "loss": 0.714,
+      "loss": 0.8434,
       "step": 21204
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.622807367313499e-06,
-      "loss": 0.8035,
+      "loss": 0.8256,
       "step": 21205
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.622012235107495e-06,
-      "loss": 0.8372,
+      "loss": 0.8808,
       "step": 21206
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6212172052606054e-06,
-      "loss": 0.686,
+      "loss": 0.705,
       "step": 21207
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.620422277783857e-06,
-      "loss": 0.7968,
+      "loss": 0.8257,
       "step": 21208
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6196274526882803e-06,
-      "loss": 0.7992,
+      "loss": 0.7848,
       "step": 21209
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.618832729984898e-06,
-      "loss": 0.6832,
+      "loss": 0.8222,
       "step": 21210
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.618038109684743e-06,
-      "loss": 0.6457,
+      "loss": 0.7338,
       "step": 21211
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6172435917988346e-06,
-      "loss": 0.7257,
+      "loss": 0.7747,
       "step": 21212
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.616449176338197e-06,
-      "loss": 0.6784,
+      "loss": 0.7417,
       "step": 21213
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.615654863313851e-06,
-      "loss": 0.7381,
+      "loss": 0.7521,
       "step": 21214
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.614860652736817e-06,
-      "loss": 0.6543,
+      "loss": 0.8097,
       "step": 21215
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6140665446181124e-06,
-      "loss": 0.6735,
+      "loss": 0.7688,
       "step": 21216
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.613272538968753e-06,
-      "loss": 0.6563,
+      "loss": 0.7868,
       "step": 21217
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6124786357997622e-06,
-      "loss": 0.7079,
+      "loss": 0.8637,
       "step": 21218
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6116848351221435e-06,
-      "loss": 0.6764,
+      "loss": 0.7598,
       "step": 21219
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.610891136946918e-06,
-      "loss": 0.6296,
+      "loss": 0.7167,
       "step": 21220
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6100975412850936e-06,
-      "loss": 0.7673,
+      "loss": 0.8401,
       "step": 21221
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6093040481476815e-06,
-      "loss": 0.6315,
+      "loss": 0.7133,
       "step": 21222
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6085106575456863e-06,
-      "loss": 0.7615,
+      "loss": 0.7466,
       "step": 21223
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.607717369490125e-06,
-      "loss": 0.9288,
+      "loss": 0.9358,
       "step": 21224
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6069241839919913e-06,
-      "loss": 0.8244,
+      "loss": 0.8274,
       "step": 21225
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.606131101062297e-06,
-      "loss": 0.7318,
+      "loss": 0.8374,
       "step": 21226
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6053381207120443e-06,
-      "loss": 0.7146,
+      "loss": 0.7598,
       "step": 21227
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.604545242952232e-06,
-      "loss": 0.6218,
+      "loss": 0.7948,
       "step": 21228
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6037524677938596e-06,
-      "loss": 0.6702,
+      "loss": 0.7467,
       "step": 21229
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6029597952479315e-06,
-      "loss": 0.6398,
+      "loss": 0.81,
       "step": 21230
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.6021672253254405e-06,
-      "loss": 0.6356,
+      "loss": 0.7477,
       "step": 21231
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.601374758037383e-06,
-      "loss": 0.6683,
+      "loss": 0.7645,
       "step": 21232
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.600582393394754e-06,
-      "loss": 0.5353,
+      "loss": 0.7439,
       "step": 21233
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.599790131408545e-06,
-      "loss": 0.7889,
+      "loss": 0.802,
       "step": 21234
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5989979720897485e-06,
-      "loss": 0.6158,
+      "loss": 0.7633,
       "step": 21235
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.598205915449352e-06,
-      "loss": 0.8208,
+      "loss": 0.8312,
       "step": 21236
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5974139614983527e-06,
-      "loss": 0.6391,
+      "loss": 0.6896,
       "step": 21237
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.596622110247725e-06,
-      "loss": 0.8138,
+      "loss": 0.904,
       "step": 21238
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5958303617084657e-06,
-      "loss": 0.7118,
+      "loss": 0.8466,
       "step": 21239
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5950387158915548e-06,
-      "loss": 0.7263,
+      "loss": 0.7856,
       "step": 21240
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5942471728079742e-06,
-      "loss": 0.583,
+      "loss": 0.6671,
       "step": 21241
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.593455732468705e-06,
-      "loss": 0.7257,
+      "loss": 0.8665,
       "step": 21242
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5926643948847306e-06,
-      "loss": 0.6809,
+      "loss": 0.7401,
       "step": 21243
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.59187316006703e-06,
-      "loss": 0.8333,
+      "loss": 0.8687,
       "step": 21244
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.591082028026577e-06,
-      "loss": 0.6875,
+      "loss": 0.7863,
       "step": 21245
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5902909987743496e-06,
-      "loss": 0.8697,
+      "loss": 0.7972,
       "step": 21246
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5895000723213213e-06,
-      "loss": 0.6121,
+      "loss": 0.6756,
       "step": 21247
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5887092486784628e-06,
-      "loss": 0.7588,
+      "loss": 0.783,
       "step": 21248
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5879185278567507e-06,
-      "loss": 0.6882,
+      "loss": 0.8332,
       "step": 21249
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.587127909867153e-06,
-      "loss": 0.905,
+      "loss": 0.9377,
       "step": 21250
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5863373947206383e-06,
-      "loss": 0.7136,
+      "loss": 0.8204,
       "step": 21251
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.585546982428173e-06,
-      "loss": 0.7225,
+      "loss": 0.884,
       "step": 21252
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5847566730007247e-06,
-      "loss": 0.7857,
+      "loss": 0.8506,
       "step": 21253
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.583966466449256e-06,
-      "loss": 0.7432,
+      "loss": 0.8469,
       "step": 21254
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5831763627847274e-06,
-      "loss": 0.718,
+      "loss": 0.8321,
       "step": 21255
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.58238636201811e-06,
-      "loss": 0.6522,
+      "loss": 0.7704,
       "step": 21256
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.581596464160351e-06,
-      "loss": 0.8669,
+      "loss": 0.8706,
       "step": 21257
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5808066692224197e-06,
-      "loss": 0.7113,
+      "loss": 0.7734,
       "step": 21258
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5800169772152683e-06,
-      "loss": 0.6993,
+      "loss": 0.8217,
       "step": 21259
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.579227388149854e-06,
-      "loss": 0.7635,
+      "loss": 0.868,
       "step": 21260
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5784379020371286e-06,
-      "loss": 0.6971,
+      "loss": 0.7725,
       "step": 21261
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5776485188880507e-06,
-      "loss": 0.7435,
+      "loss": 0.8267,
       "step": 21262
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.576859238713568e-06,
-      "loss": 0.7946,
+      "loss": 0.9057,
       "step": 21263
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5760700615246314e-06,
-      "loss": 0.6669,
+      "loss": 0.8741,
       "step": 21264
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.575280987332187e-06,
-      "loss": 0.7595,
+      "loss": 0.7989,
       "step": 21265
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5744920161471897e-06,
-      "loss": 0.6896,
+      "loss": 0.7488,
       "step": 21266
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.573703147980575e-06,
-      "loss": 0.6475,
+      "loss": 0.7763,
       "step": 21267
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5729143828432957e-06,
-      "loss": 0.6608,
+      "loss": 0.7391,
       "step": 21268
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.572125720746291e-06,
-      "loss": 0.5746,
+      "loss": 0.7291,
       "step": 21269
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5713371617005033e-06,
-      "loss": 0.7182,
+      "loss": 0.728,
       "step": 21270
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5705487057168734e-06,
-      "loss": 0.6028,
+      "loss": 0.794,
       "step": 21271
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5697603528063386e-06,
-      "loss": 0.7785,
+      "loss": 0.8921,
       "step": 21272
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5689721029798364e-06,
-      "loss": 0.6274,
+      "loss": 0.7564,
       "step": 21273
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5681839562483e-06,
-      "loss": 0.7709,
+      "loss": 0.8925,
       "step": 21274
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5673959126226698e-06,
-      "loss": 0.717,
+      "loss": 0.8387,
       "step": 21275
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.566607972113877e-06,
-      "loss": 0.7214,
+      "loss": 0.8376,
       "step": 21276
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5658201347328517e-06,
-      "loss": 0.7555,
+      "loss": 0.7953,
       "step": 21277
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5650324004905238e-06,
-      "loss": 0.818,
+      "loss": 0.8526,
       "step": 21278
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.564244769397822e-06,
-      "loss": 0.7745,
+      "loss": 0.8373,
       "step": 21279
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.563457241465671e-06,
-      "loss": 0.7717,
+      "loss": 0.8889,
       "step": 21280
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5626698167050036e-06,
-      "loss": 0.5506,
+      "loss": 0.7102,
       "step": 21281
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5618824951267395e-06,
-      "loss": 0.6054,
+      "loss": 0.7094,
       "step": 21282
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.561095276741804e-06,
-      "loss": 0.673,
+      "loss": 0.7964,
       "step": 21283
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5603081615611125e-06,
-      "loss": 0.777,
+      "loss": 0.829,
       "step": 21284
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5595211495955965e-06,
-      "loss": 0.6838,
+      "loss": 0.8121,
       "step": 21285
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.558734240856161e-06,
-      "loss": 0.6417,
+      "loss": 0.7112,
       "step": 21286
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.557947435353735e-06,
-      "loss": 0.7846,
+      "loss": 0.8343,
       "step": 21287
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.557160733099228e-06,
-      "loss": 0.6192,
+      "loss": 0.7705,
       "step": 21288
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5563741341035555e-06,
-      "loss": 0.6473,
+      "loss": 0.7172,
       "step": 21289
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5555876383776314e-06,
-      "loss": 0.7352,
+      "loss": 0.8268,
       "step": 21290
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5548012459323667e-06,
-      "loss": 0.8166,
+      "loss": 0.8742,
       "step": 21291
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.554014956778671e-06,
-      "loss": 0.7521,
+      "loss": 0.7527,
       "step": 21292
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5532287709274506e-06,
-      "loss": 0.6524,
+      "loss": 0.7685,
       "step": 21293
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.552442688389618e-06,
-      "loss": 0.7972,
+      "loss": 0.9165,
       "step": 21294
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.551656709176077e-06,
-      "loss": 0.7127,
+      "loss": 0.7921,
       "step": 21295
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5508708332977304e-06,
-      "loss": 0.7144,
+      "loss": 0.7699,
       "step": 21296
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.55008506076548e-06,
-      "loss": 0.7477,
+      "loss": 0.8714,
       "step": 21297
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5492993915902353e-06,
-      "loss": 0.7494,
+      "loss": 0.8467,
       "step": 21298
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5485138257828845e-06,
-      "loss": 0.6694,
+      "loss": 0.7924,
       "step": 21299
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.547728363354335e-06,
-      "loss": 0.7742,
+      "loss": 0.8253,
       "step": 21300
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.54694300431548e-06,
-      "loss": 0.5791,
+      "loss": 0.7301,
       "step": 21301
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.546157748677217e-06,
-      "loss": 0.8368,
+      "loss": 0.8705,
       "step": 21302
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5453725964504372e-06,
-      "loss": 0.6236,
+      "loss": 0.7456,
       "step": 21303
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5445875476460413e-06,
-      "loss": 0.7964,
+      "loss": 0.8619,
       "step": 21304
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5438026022749096e-06,
-      "loss": 0.6571,
+      "loss": 0.743,
       "step": 21305
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.54301776034794e-06,
-      "loss": 0.6368,
+      "loss": 0.7717,
       "step": 21306
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5422330218760195e-06,
-      "loss": 0.7869,
+      "loss": 0.8158,
       "step": 21307
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5414483868700344e-06,
-      "loss": 0.7749,
+      "loss": 0.7459,
       "step": 21308
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.5406638553408714e-06,
-      "loss": 0.695,
+      "loss": 0.7256,
       "step": 21309
     },
     {
       "epoch": 0.77,
       "learning_rate": 2.539879427299411e-06,
-      "loss": 0.638,
+      "loss": 0.7784,
       "step": 21310
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5390951027565438e-06,
-      "loss": 0.8183,
+      "loss": 0.8424,
       "step": 21311
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5383108817231416e-06,
-      "loss": 0.6453,
+      "loss": 0.7872,
       "step": 21312
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.537526764210092e-06,
-      "loss": 0.6941,
+      "loss": 0.7517,
       "step": 21313
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5367427502282704e-06,
-      "loss": 0.8151,
+      "loss": 0.8823,
       "step": 21314
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5359588397885536e-06,
-      "loss": 0.684,
+      "loss": 0.7822,
       "step": 21315
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5351750329018156e-06,
-      "loss": 0.7109,
+      "loss": 0.81,
       "step": 21316
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.534391329578938e-06,
-      "loss": 0.7394,
+      "loss": 0.8287,
       "step": 21317
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5336077298307836e-06,
-      "loss": 0.6344,
+      "loss": 0.7776,
       "step": 21318
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.532824233668231e-06,
-      "loss": 0.8122,
+      "loss": 0.8855,
       "step": 21319
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.532040841102147e-06,
-      "loss": 0.7844,
+      "loss": 0.8604,
       "step": 21320
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5312575521434014e-06,
-      "loss": 0.6608,
+      "loss": 0.7463,
       "step": 21321
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5304743668028574e-06,
-      "loss": 0.6643,
+      "loss": 0.8082,
       "step": 21322
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.529691285091389e-06,
-      "loss": 0.7772,
+      "loss": 0.8328,
       "step": 21323
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5289083070198496e-06,
-      "loss": 0.7887,
+      "loss": 0.843,
       "step": 21324
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.528125432599111e-06,
-      "loss": 0.7352,
+      "loss": 0.8121,
       "step": 21325
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5273426618400297e-06,
-      "loss": 0.7019,
+      "loss": 0.8323,
       "step": 21326
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5265599947534667e-06,
-      "loss": 0.6835,
+      "loss": 0.7815,
       "step": 21327
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5257774313502816e-06,
-      "loss": 0.7245,
+      "loss": 0.8228,
       "step": 21328
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.524994971641326e-06,
-      "loss": 0.5778,
+      "loss": 0.7454,
       "step": 21329
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5242126156374656e-06,
-      "loss": 0.7831,
+      "loss": 0.7993,
       "step": 21330
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5234303633495425e-06,
-      "loss": 0.7276,
+      "loss": 0.8309,
       "step": 21331
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.522648214788419e-06,
-      "loss": 0.6678,
+      "loss": 0.8134,
       "step": 21332
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5218661699649416e-06,
-      "loss": 0.6818,
+      "loss": 0.8076,
       "step": 21333
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5210842288899617e-06,
-      "loss": 0.663,
+      "loss": 0.7595,
       "step": 21334
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.520302391574323e-06,
-      "loss": 0.7599,
+      "loss": 0.8171,
       "step": 21335
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5195206580288833e-06,
-      "loss": 0.7076,
+      "loss": 0.8329,
       "step": 21336
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.518739028264474e-06,
-      "loss": 0.7651,
+      "loss": 0.8273,
       "step": 21337
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.51795750229195e-06,
-      "loss": 0.6315,
+      "loss": 0.7861,
       "step": 21338
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.51717608012215e-06,
-      "loss": 0.683,
+      "loss": 0.8393,
       "step": 21339
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5163947617659144e-06,
-      "loss": 0.7427,
+      "loss": 0.8029,
       "step": 21340
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5156135472340816e-06,
-      "loss": 0.7382,
+      "loss": 0.8775,
       "step": 21341
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5148324365374945e-06,
-      "loss": 0.6564,
+      "loss": 0.762,
       "step": 21342
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.514051429686988e-06,
-      "loss": 0.7815,
+      "loss": 0.7934,
       "step": 21343
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5132705266933953e-06,
-      "loss": 0.679,
+      "loss": 0.7714,
       "step": 21344
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.512489727567553e-06,
-      "loss": 0.6574,
+      "loss": 0.8777,
       "step": 21345
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.511709032320291e-06,
-      "loss": 0.7582,
+      "loss": 0.8054,
       "step": 21346
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5109284409624422e-06,
-      "loss": 0.7125,
+      "loss": 0.7508,
       "step": 21347
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5101479535048334e-06,
-      "loss": 0.6874,
+      "loss": 0.7937,
       "step": 21348
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.509367569958301e-06,
-      "loss": 0.6307,
+      "loss": 0.7472,
       "step": 21349
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5085872903336593e-06,
-      "loss": 0.7824,
+      "loss": 0.8585,
       "step": 21350
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5078071146417428e-06,
-      "loss": 0.7139,
+      "loss": 0.7926,
       "step": 21351
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5070270428933728e-06,
-      "loss": 0.8337,
+      "loss": 0.8677,
       "step": 21352
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.506247075099372e-06,
-      "loss": 0.655,
+      "loss": 0.7846,
       "step": 21353
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.505467211270557e-06,
-      "loss": 0.6647,
+      "loss": 0.7704,
       "step": 21354
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5046874514177544e-06,
-      "loss": 0.6267,
+      "loss": 0.7538,
       "step": 21355
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5039077955517787e-06,
-      "loss": 0.6416,
+      "loss": 0.8215,
       "step": 21356
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.503128243683447e-06,
-      "loss": 0.4909,
+      "loss": 0.6557,
       "step": 21357
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5023487958235738e-06,
-      "loss": 0.7119,
+      "loss": 0.827,
       "step": 21358
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.501569451982974e-06,
-      "loss": 0.7783,
+      "loss": 0.8792,
       "step": 21359
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.500790212172456e-06,
-      "loss": 0.7493,
+      "loss": 0.747,
       "step": 21360
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.5000110764028363e-06,
-      "loss": 0.6472,
+      "loss": 0.7991,
       "step": 21361
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4992320446849226e-06,
-      "loss": 0.6739,
+      "loss": 0.8516,
       "step": 21362
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.498453117029521e-06,
-      "loss": 0.6292,
+      "loss": 0.7445,
       "step": 21363
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4976742934474376e-06,
-      "loss": 0.7513,
+      "loss": 0.8495,
       "step": 21364
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.496895573949484e-06,
-      "loss": 0.676,
+      "loss": 0.7573,
       "step": 21365
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4961169585464553e-06,
-      "loss": 0.7242,
+      "loss": 0.7824,
       "step": 21366
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.495338447249155e-06,
-      "loss": 0.7203,
+      "loss": 0.7336,
       "step": 21367
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.494560040068392e-06,
-      "loss": 0.7122,
+      "loss": 0.7767,
       "step": 21368
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.493781737014953e-06,
-      "loss": 0.7215,
+      "loss": 0.7622,
       "step": 21369
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4930035380996467e-06,
-      "loss": 0.7193,
+      "loss": 0.8272,
       "step": 21370
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.492225443333264e-06,
-      "loss": 0.6417,
+      "loss": 0.7977,
       "step": 21371
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.491447452726602e-06,
-      "loss": 0.7932,
+      "loss": 0.8439,
       "step": 21372
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.49066956629045e-06,
-      "loss": 0.8138,
+      "loss": 0.8887,
       "step": 21373
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4898917840356062e-06,
-      "loss": 0.7943,
+      "loss": 0.8395,
       "step": 21374
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4891141059728587e-06,
-      "loss": 0.7966,
+      "loss": 0.8283,
       "step": 21375
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.488336532112996e-06,
-      "loss": 0.7363,
+      "loss": 0.8228,
       "step": 21376
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4875590624668046e-06,
-      "loss": 0.6107,
+      "loss": 0.7479,
       "step": 21377
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4867816970450775e-06,
-      "loss": 0.5846,
+      "loss": 0.6987,
       "step": 21378
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.486004435858588e-06,
-      "loss": 0.7278,
+      "loss": 0.8093,
       "step": 21379
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.48522727891813e-06,
-      "loss": 0.7413,
+      "loss": 0.8528,
       "step": 21380
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4844502262344805e-06,
-      "loss": 0.6091,
+      "loss": 0.716,
       "step": 21381
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4836732778184214e-06,
-      "loss": 0.6439,
+      "loss": 0.8073,
       "step": 21382
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4828964336807283e-06,
-      "loss": 0.674,
+      "loss": 0.7911,
       "step": 21383
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.482119693832188e-06,
-      "loss": 0.6734,
+      "loss": 0.7518,
       "step": 21384
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4813430582835676e-06,
-      "loss": 0.6708,
+      "loss": 0.7753,
       "step": 21385
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.48056652704564e-06,
-      "loss": 0.7882,
+      "loss": 0.8304,
       "step": 21386
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.479790100129188e-06,
-      "loss": 0.8237,
+      "loss": 0.8699,
       "step": 21387
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4790137775449787e-06,
-      "loss": 0.7373,
+      "loss": 0.753,
       "step": 21388
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.478237559303781e-06,
-      "loss": 0.7274,
+      "loss": 0.778,
       "step": 21389
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.477461445416366e-06,
-      "loss": 0.6722,
+      "loss": 0.7318,
       "step": 21390
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4766854358935e-06,
-      "loss": 0.6689,
+      "loss": 0.8202,
       "step": 21391
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4759095307459467e-06,
-      "loss": 0.7498,
+      "loss": 0.8289,
       "step": 21392
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.475133729984477e-06,
-      "loss": 0.6912,
+      "loss": 0.7782,
       "step": 21393
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.474358033619849e-06,
-      "loss": 0.6305,
+      "loss": 0.7214,
       "step": 21394
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4735824416628262e-06,
-      "loss": 0.7693,
+      "loss": 0.8051,
       "step": 21395
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.472806954124166e-06,
-      "loss": 0.6566,
+      "loss": 0.7559,
       "step": 21396
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.472031571014636e-06,
-      "loss": 0.6798,
+      "loss": 0.7819,
       "step": 21397
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4712562923449802e-06,
-      "loss": 0.5966,
+      "loss": 0.7535,
       "step": 21398
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4704811181259646e-06,
-      "loss": 0.6859,
+      "loss": 0.7917,
       "step": 21399
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.469706048368341e-06,
-      "loss": 0.753,
+      "loss": 0.7757,
       "step": 21400
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.468931083082862e-06,
-      "loss": 0.7623,
+      "loss": 0.8691,
       "step": 21401
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4681562222802745e-06,
-      "loss": 0.6899,
+      "loss": 0.8405,
       "step": 21402
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4673814659713404e-06,
-      "loss": 0.7628,
+      "loss": 0.8302,
       "step": 21403
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4666068141667988e-06,
-      "loss": 0.6439,
+      "loss": 0.7394,
       "step": 21404
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.465832266877395e-06,
-      "loss": 0.6539,
+      "loss": 0.7138,
       "step": 21405
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.465057824113882e-06,
-      "loss": 0.8472,
+      "loss": 0.8627,
       "step": 21406
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4642834858870014e-06,
-      "loss": 0.7912,
+      "loss": 0.8549,
       "step": 21407
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4635092522074954e-06,
-      "loss": 0.7157,
+      "loss": 0.8285,
       "step": 21408
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4627351230861028e-06,
-      "loss": 0.751,
+      "loss": 0.8003,
       "step": 21409
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.461961098533572e-06,
-      "loss": 0.6673,
+      "loss": 0.8036,
       "step": 21410
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.46118717856063e-06,
-      "loss": 0.7856,
+      "loss": 0.8811,
       "step": 21411
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4604133631780227e-06,
-      "loss": 0.6739,
+      "loss": 0.7638,
       "step": 21412
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4596396523964825e-06,
-      "loss": 0.6906,
+      "loss": 0.7762,
       "step": 21413
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4588660462267445e-06,
-      "loss": 0.6446,
+      "loss": 0.817,
       "step": 21414
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.458092544679538e-06,
-      "loss": 0.6542,
+      "loss": 0.8006,
       "step": 21415
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.457319147765601e-06,
-      "loss": 0.6331,
+      "loss": 0.7656,
       "step": 21416
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4565458554956546e-06,
-      "loss": 0.7612,
+      "loss": 0.8306,
       "step": 21417
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.455772667880435e-06,
-      "loss": 0.7527,
+      "loss": 0.8018,
       "step": 21418
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.454999584930665e-06,
-      "loss": 0.6712,
+      "loss": 0.7991,
       "step": 21419
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.454226606657071e-06,
-      "loss": 0.778,
+      "loss": 0.7757,
       "step": 21420
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4534537330703733e-06,
-      "loss": 0.7461,
+      "loss": 0.8302,
       "step": 21421
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4526809641813042e-06,
-      "loss": 0.574,
+      "loss": 0.6466,
       "step": 21422
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.451908300000576e-06,
-      "loss": 0.6639,
+      "loss": 0.7889,
       "step": 21423
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.451135740538908e-06,
-      "loss": 0.6438,
+      "loss": 0.7108,
       "step": 21424
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4503632858070225e-06,
-      "loss": 0.7061,
+      "loss": 0.8036,
       "step": 21425
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4495909358156365e-06,
-      "loss": 0.6968,
+      "loss": 0.7585,
       "step": 21426
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4488186905754627e-06,
-      "loss": 0.7468,
+      "loss": 0.8487,
       "step": 21427
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4480465500972137e-06,
-      "loss": 0.7824,
+      "loss": 0.86,
       "step": 21428
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.44727451439161e-06,
-      "loss": 0.8327,
+      "loss": 0.9595,
       "step": 21429
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4465025834693502e-06,
-      "loss": 0.7628,
+      "loss": 0.7849,
       "step": 21430
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4457307573411525e-06,
-      "loss": 0.8275,
+      "loss": 0.9082,
       "step": 21431
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4449590360177224e-06,
-      "loss": 0.7021,
+      "loss": 0.8365,
       "step": 21432
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.444187419509767e-06,
-      "loss": 0.7671,
+      "loss": 0.8959,
       "step": 21433
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4434159078279882e-06,
-      "loss": 0.6679,
+      "loss": 0.7405,
       "step": 21434
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.442644500983098e-06,
-      "loss": 0.7307,
+      "loss": 0.7925,
       "step": 21435
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4418731989857857e-06,
-      "loss": 0.7722,
+      "loss": 0.7719,
       "step": 21436
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4411020018467633e-06,
-      "loss": 0.6005,
+      "loss": 0.7685,
       "step": 21437
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4403309095767256e-06,
-      "loss": 0.507,
+      "loss": 0.6936,
       "step": 21438
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.439559922186372e-06,
-      "loss": 0.6314,
+      "loss": 0.7688,
       "step": 21439
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.438789039686393e-06,
-      "loss": 0.7149,
+      "loss": 0.8109,
       "step": 21440
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4380182620874915e-06,
-      "loss": 0.699,
+      "loss": 0.7702,
       "step": 21441
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4372475894003576e-06,
-      "loss": 0.7911,
+      "loss": 0.8252,
       "step": 21442
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4364770216356837e-06,
-      "loss": 0.7815,
+      "loss": 0.8351,
       "step": 21443
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4357065588041605e-06,
-      "loss": 0.8066,
+      "loss": 0.8394,
       "step": 21444
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4349362009164755e-06,
-      "loss": 0.8053,
+      "loss": 0.8556,
       "step": 21445
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.434165947983317e-06,
-      "loss": 0.722,
+      "loss": 0.8126,
       "step": 21446
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.433395800015369e-06,
-      "loss": 0.6699,
+      "loss": 0.7215,
       "step": 21447
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4326257570233247e-06,
-      "loss": 0.6811,
+      "loss": 0.7604,
       "step": 21448
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4318558190178553e-06,
-      "loss": 0.6905,
+      "loss": 0.7788,
       "step": 21449
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.431085986009651e-06,
-      "loss": 0.6772,
+      "loss": 0.7995,
       "step": 21450
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4303162580093908e-06,
-      "loss": 0.6647,
+      "loss": 0.7844,
       "step": 21451
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.429546635027752e-06,
-      "loss": 0.7211,
+      "loss": 0.81,
       "step": 21452
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4287771170754095e-06,
-      "loss": 0.6303,
+      "loss": 0.7883,
       "step": 21453
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4280077041630444e-06,
-      "loss": 0.7139,
+      "loss": 0.7778,
       "step": 21454
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4272383963013313e-06,
-      "loss": 0.7658,
+      "loss": 0.8218,
       "step": 21455
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.42646919350094e-06,
-      "loss": 0.5729,
+      "loss": 0.7049,
       "step": 21456
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.425700095772543e-06,
-      "loss": 0.6378,
+      "loss": 0.7124,
       "step": 21457
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4249311031268107e-06,
-      "loss": 0.635,
+      "loss": 0.7584,
       "step": 21458
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.424162215574408e-06,
-      "loss": 0.7891,
+      "loss": 0.8164,
       "step": 21459
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.42339343312601e-06,
-      "loss": 0.7611,
+      "loss": 0.8054,
       "step": 21460
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4226247557922778e-06,
-      "loss": 0.7487,
+      "loss": 0.8039,
       "step": 21461
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.421856183583877e-06,
-      "loss": 0.6647,
+      "loss": 0.7463,
       "step": 21462
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.421087716511469e-06,
-      "loss": 0.6679,
+      "loss": 0.7226,
       "step": 21463
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.420319354585716e-06,
-      "loss": 0.7005,
+      "loss": 0.7519,
       "step": 21464
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4195510978172776e-06,
-      "loss": 0.8165,
+      "loss": 0.8914,
       "step": 21465
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.41878294621681e-06,
-      "loss": 0.6248,
+      "loss": 0.7616,
       "step": 21466
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.418014899794978e-06,
-      "loss": 0.7978,
+      "loss": 0.817,
       "step": 21467
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.417246958562426e-06,
-      "loss": 0.5905,
+      "loss": 0.6815,
       "step": 21468
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.416479122529817e-06,
-      "loss": 0.7542,
+      "loss": 0.7883,
       "step": 21469
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4157113917078e-06,
-      "loss": 0.7512,
+      "loss": 0.8152,
       "step": 21470
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4149437661070286e-06,
-      "loss": 0.8883,
+      "loss": 0.9082,
       "step": 21471
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.414176245738146e-06,
-      "loss": 0.7714,
+      "loss": 0.868,
       "step": 21472
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4134088306118086e-06,
-      "loss": 0.8372,
+      "loss": 0.8633,
       "step": 21473
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.41264152073866e-06,
-      "loss": 0.8216,
+      "loss": 0.8594,
       "step": 21474
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4118743161293446e-06,
-      "loss": 0.6835,
+      "loss": 0.7783,
       "step": 21475
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4111072167945038e-06,
-      "loss": 0.8137,
+      "loss": 0.8406,
       "step": 21476
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4103402227447893e-06,
-      "loss": 0.5466,
+      "loss": 0.7462,
       "step": 21477
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.409573333990831e-06,
-      "loss": 0.7047,
+      "loss": 0.843,
       "step": 21478
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4088065505432746e-06,
-      "loss": 0.8121,
+      "loss": 0.9102,
       "step": 21479
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4080398724127586e-06,
-      "loss": 0.6773,
+      "loss": 0.7184,
       "step": 21480
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4072732996099167e-06,
-      "loss": 0.7192,
+      "loss": 0.8765,
       "step": 21481
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4065068321453843e-06,
-      "loss": 0.7578,
+      "loss": 0.8247,
       "step": 21482
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.405740470029797e-06,
-      "loss": 0.6887,
+      "loss": 0.8061,
       "step": 21483
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4049742132737863e-06,
-      "loss": 0.6939,
+      "loss": 0.7697,
       "step": 21484
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4042080618879782e-06,
-      "loss": 0.6534,
+      "loss": 0.7863,
       "step": 21485
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4034420158830096e-06,
-      "loss": 0.7871,
+      "loss": 0.8678,
       "step": 21486
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4026760752695056e-06,
-      "loss": 0.7158,
+      "loss": 0.7734,
       "step": 21487
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4019102400580918e-06,
-      "loss": 0.6966,
+      "loss": 0.7952,
       "step": 21488
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.401144510259392e-06,
-      "loss": 0.7503,
+      "loss": 0.8749,
       "step": 21489
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.4003788858840316e-06,
-      "loss": 0.7935,
+      "loss": 0.8154,
       "step": 21490
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.399613366942629e-06,
-      "loss": 0.6818,
+      "loss": 0.8385,
       "step": 21491
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.39884795344581e-06,
-      "loss": 0.7475,
+      "loss": 0.876,
       "step": 21492
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3980826454041906e-06,
-      "loss": 0.7076,
+      "loss": 0.8107,
       "step": 21493
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3973174428283896e-06,
-      "loss": 0.7229,
+      "loss": 0.8171,
       "step": 21494
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3965523457290187e-06,
-      "loss": 0.6739,
+      "loss": 0.7732,
       "step": 21495
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.395787354116702e-06,
-      "loss": 0.7682,
+      "loss": 0.7496,
       "step": 21496
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.395022468002042e-06,
-      "loss": 0.7832,
+      "loss": 0.9089,
       "step": 21497
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3942576873956567e-06,
-      "loss": 0.5807,
+      "loss": 0.7027,
       "step": 21498
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3934930123081558e-06,
-      "loss": 0.737,
+      "loss": 0.8006,
       "step": 21499
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3927284427501464e-06,
-      "loss": 0.7016,
+      "loss": 0.8451,
       "step": 21500
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.391963978732237e-06,
-      "loss": 0.6955,
+      "loss": 0.7411,
       "step": 21501
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3911996202650323e-06,
-      "loss": 0.7208,
+      "loss": 0.7948,
       "step": 21502
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3904353673591387e-06,
-      "loss": 0.7805,
+      "loss": 0.8551,
       "step": 21503
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.389671220025154e-06,
-      "loss": 0.6444,
+      "loss": 0.7054,
       "step": 21504
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.388907178273686e-06,
-      "loss": 0.6287,
+      "loss": 0.7714,
       "step": 21505
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.388143242115333e-06,
-      "loss": 0.6897,
+      "loss": 0.7887,
       "step": 21506
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3873794115606928e-06,
-      "loss": 0.7735,
+      "loss": 0.888,
       "step": 21507
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3866156866203594e-06,
-      "loss": 0.6681,
+      "loss": 0.7834,
       "step": 21508
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3858520673049367e-06,
-      "loss": 0.6922,
+      "loss": 0.7427,
       "step": 21509
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.385088553625008e-06,
-      "loss": 0.7009,
+      "loss": 0.7366,
       "step": 21510
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.384325145591174e-06,
-      "loss": 0.6947,
+      "loss": 0.9064,
       "step": 21511
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.383561843214024e-06,
-      "loss": 0.7055,
+      "loss": 0.7701,
       "step": 21512
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3827986465041462e-06,
-      "loss": 0.6387,
+      "loss": 0.796,
       "step": 21513
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3820355554721274e-06,
-      "loss": 0.7394,
+      "loss": 0.8495,
       "step": 21514
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3812725701285633e-06,
-      "loss": 0.5556,
+      "loss": 0.6582,
       "step": 21515
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3805096904840265e-06,
-      "loss": 0.6773,
+      "loss": 0.7959,
       "step": 21516
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3797469165491092e-06,
-      "loss": 0.6645,
+      "loss": 0.7474,
       "step": 21517
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.378984248334393e-06,
-      "loss": 0.6237,
+      "loss": 0.782,
       "step": 21518
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.378221685850456e-06,
-      "loss": 0.6969,
+      "loss": 0.7805,
       "step": 21519
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.37745922910788e-06,
-      "loss": 0.608,
+      "loss": 0.7479,
       "step": 21520
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3766968781172395e-06,
-      "loss": 0.824,
+      "loss": 0.9301,
       "step": 21521
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3759346328891197e-06,
-      "loss": 0.7356,
+      "loss": 0.7604,
       "step": 21522
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3751724934340834e-06,
-      "loss": 0.7437,
+      "loss": 0.8495,
       "step": 21523
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.374410459762714e-06,
-      "loss": 0.8058,
+      "loss": 0.9281,
       "step": 21524
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3736485318855797e-06,
-      "loss": 0.6209,
+      "loss": 0.7275,
       "step": 21525
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3728867098132513e-06,
-      "loss": 0.6768,
+      "loss": 0.7728,
       "step": 21526
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.372124993556296e-06,
-      "loss": 0.6993,
+      "loss": 0.7648,
       "step": 21527
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3713633831252903e-06,
-      "loss": 0.8065,
+      "loss": 0.8784,
       "step": 21528
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3706018785307873e-06,
-      "loss": 0.7556,
+      "loss": 0.8707,
       "step": 21529
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.369840479783362e-06,
-      "loss": 0.7449,
+      "loss": 0.7866,
       "step": 21530
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3690791868935736e-06,
-      "loss": 0.7706,
+      "loss": 0.824,
       "step": 21531
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3683179998719853e-06,
-      "loss": 0.6482,
+      "loss": 0.6917,
       "step": 21532
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3675569187291537e-06,
-      "loss": 0.7415,
+      "loss": 0.7744,
       "step": 21533
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.366795943475646e-06,
-      "loss": 0.7283,
+      "loss": 0.8738,
       "step": 21534
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3660350741220094e-06,
-      "loss": 0.7483,
+      "loss": 0.8621,
       "step": 21535
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3652743106788078e-06,
-      "loss": 0.7533,
+      "loss": 0.776,
       "step": 21536
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3645136531565925e-06,
-      "loss": 0.6593,
+      "loss": 0.7904,
       "step": 21537
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3637531015659177e-06,
-      "loss": 0.7116,
+      "loss": 0.7944,
       "step": 21538
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.362992655917333e-06,
-      "loss": 0.8108,
+      "loss": 0.8507,
       "step": 21539
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3622323162213877e-06,
-      "loss": 0.7047,
+      "loss": 0.7805,
       "step": 21540
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3614720824886384e-06,
-      "loss": 0.726,
+      "loss": 0.7467,
       "step": 21541
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.36071195472962e-06,
-      "loss": 0.6437,
+      "loss": 0.8008,
       "step": 21542
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.359951932954888e-06,
-      "loss": 0.6693,
+      "loss": 0.7722,
       "step": 21543
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3591920171749826e-06,
-      "loss": 0.7356,
+      "loss": 0.7593,
       "step": 21544
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3584322074004473e-06,
-      "loss": 0.7688,
+      "loss": 0.8529,
       "step": 21545
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3576725036418204e-06,
-      "loss": 0.7904,
+      "loss": 0.8817,
       "step": 21546
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.35691290590965e-06,
-      "loss": 0.7076,
+      "loss": 0.8069,
       "step": 21547
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3561534142144626e-06,
-      "loss": 0.6962,
+      "loss": 0.7892,
       "step": 21548
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3553940285668042e-06,
-      "loss": 0.6893,
+      "loss": 0.7924,
       "step": 21549
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.354634748977208e-06,
-      "loss": 0.5568,
+      "loss": 0.7109,
       "step": 21550
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.353875575456207e-06,
-      "loss": 0.7318,
+      "loss": 0.7802,
       "step": 21551
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3531165080143315e-06,
-      "loss": 0.737,
+      "loss": 0.7797,
       "step": 21552
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3523575466621185e-06,
-      "loss": 0.7214,
+      "loss": 0.8239,
       "step": 21553
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.351598691410093e-06,
-      "loss": 0.6428,
+      "loss": 0.7274,
       "step": 21554
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3508399422687833e-06,
-      "loss": 0.7858,
+      "loss": 0.8188,
       "step": 21555
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3500812992487174e-06,
-      "loss": 0.6893,
+      "loss": 0.8307,
       "step": 21556
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3493227623604196e-06,
-      "loss": 0.6745,
+      "loss": 0.7538,
       "step": 21557
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3485643316144146e-06,
-      "loss": 0.6616,
+      "loss": 0.7366,
       "step": 21558
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3478060070212195e-06,
-      "loss": 0.7281,
+      "loss": 0.8067,
       "step": 21559
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3470477885913644e-06,
-      "loss": 0.6985,
+      "loss": 0.7963,
       "step": 21560
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3462896763353583e-06,
-      "loss": 0.7806,
+      "loss": 0.8457,
       "step": 21561
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3455316702637264e-06,
-      "loss": 0.6965,
+      "loss": 0.8381,
       "step": 21562
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.344773770386982e-06,
-      "loss": 0.6783,
+      "loss": 0.7909,
       "step": 21563
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.34401597671564e-06,
-      "loss": 0.7159,
+      "loss": 0.7683,
       "step": 21564
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3432582892602118e-06,
-      "loss": 0.7089,
+      "loss": 0.8275,
       "step": 21565
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3425007080312166e-06,
-      "loss": 0.6322,
+      "loss": 0.7742,
       "step": 21566
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3417432330391545e-06,
-      "loss": 0.6616,
+      "loss": 0.77,
       "step": 21567
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3409858642945414e-06,
-      "loss": 0.8373,
+      "loss": 0.8768,
       "step": 21568
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.340228601807882e-06,
-      "loss": 0.7692,
+      "loss": 0.802,
       "step": 21569
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.339471445589685e-06,
-      "loss": 0.8459,
+      "loss": 0.8494,
       "step": 21570
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.338714395650449e-06,
-      "loss": 0.6684,
+      "loss": 0.7718,
       "step": 21571
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3379574520006833e-06,
-      "loss": 0.7595,
+      "loss": 0.8385,
       "step": 21572
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3372006146508865e-06,
-      "loss": 0.7166,
+      "loss": 0.7988,
       "step": 21573
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3364438836115597e-06,
-      "loss": 0.7258,
+      "loss": 0.8795,
       "step": 21574
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.335687258893199e-06,
-      "loss": 0.7585,
+      "loss": 0.7872,
       "step": 21575
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.334930740506307e-06,
-      "loss": 0.5548,
+      "loss": 0.7556,
       "step": 21576
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.334174328461374e-06,
-      "loss": 0.6425,
+      "loss": 0.8048,
       "step": 21577
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.333418022768893e-06,
-      "loss": 0.5376,
+      "loss": 0.6816,
       "step": 21578
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.332661823439365e-06,
-      "loss": 0.741,
+      "loss": 0.7702,
       "step": 21579
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3319057304832703e-06,
-      "loss": 0.8217,
+      "loss": 0.9139,
       "step": 21580
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.3311497439111065e-06,
-      "loss": 0.6357,
+      "loss": 0.8158,
       "step": 21581
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.330393863733359e-06,
-      "loss": 0.6272,
+      "loss": 0.7446,
       "step": 21582
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.329638089960515e-06,
-      "loss": 0.6892,
+      "loss": 0.7857,
       "step": 21583
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.328882422603058e-06,
-      "loss": 0.7923,
+      "loss": 0.9231,
       "step": 21584
     },
     {
       "epoch": 0.78,
       "learning_rate": 2.328126861671475e-06,
-      "loss": 0.6074,
+      "loss": 0.7734,
       "step": 21585
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3273714071762477e-06,
-      "loss": 0.796,
+      "loss": 0.8617,
       "step": 21586
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.326616059127855e-06,
-      "loss": 0.8125,
+      "loss": 0.8992,
       "step": 21587
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.325860817536778e-06,
-      "loss": 0.776,
+      "loss": 0.7921,
       "step": 21588
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3251056824134944e-06,
-      "loss": 0.7503,
+      "loss": 0.8419,
       "step": 21589
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3243506537684767e-06,
-      "loss": 0.8206,
+      "loss": 0.9442,
       "step": 21590
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.323595731612207e-06,
-      "loss": 0.7779,
+      "loss": 0.8125,
       "step": 21591
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3228409159551545e-06,
-      "loss": 0.7324,
+      "loss": 0.8232,
       "step": 21592
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.322086206807792e-06,
-      "loss": 0.5619,
+      "loss": 0.6898,
       "step": 21593
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3213316041805866e-06,
-      "loss": 0.8109,
+      "loss": 0.8967,
       "step": 21594
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3205771080840166e-06,
-      "loss": 0.7747,
+      "loss": 0.7434,
       "step": 21595
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3198227185285403e-06,
-      "loss": 0.7012,
+      "loss": 0.8542,
       "step": 21596
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.319068435524625e-06,
-      "loss": 0.6253,
+      "loss": 0.7386,
       "step": 21597
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.31831425908274e-06,
-      "loss": 0.6301,
+      "loss": 0.8007,
       "step": 21598
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.317560189213346e-06,
-      "loss": 0.7109,
+      "loss": 0.8024,
       "step": 21599
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.316806225926903e-06,
-      "loss": 0.6528,
+      "loss": 0.6975,
       "step": 21600
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3160523692338743e-06,
-      "loss": 0.5742,
+      "loss": 0.7223,
       "step": 21601
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.315298619144716e-06,
-      "loss": 0.6974,
+      "loss": 0.8025,
       "step": 21602
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3145449756698834e-06,
-      "loss": 0.7305,
+      "loss": 0.8245,
       "step": 21603
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.313791438819838e-06,
-      "loss": 0.6634,
+      "loss": 0.7738,
       "step": 21604
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.313038008605031e-06,
-      "loss": 0.7039,
+      "loss": 0.8237,
       "step": 21605
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3122846850359147e-06,
-      "loss": 0.7222,
+      "loss": 0.8635,
       "step": 21606
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3115314681229397e-06,
-      "loss": 0.63,
+      "loss": 0.757,
       "step": 21607
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.310778357876562e-06,
-      "loss": 0.7721,
+      "loss": 0.7994,
       "step": 21608
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.310025354307218e-06,
-      "loss": 0.7921,
+      "loss": 0.8578,
       "step": 21609
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3092724574253655e-06,
-      "loss": 0.6758,
+      "loss": 0.7819,
       "step": 21610
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3085196672414466e-06,
-      "loss": 0.6427,
+      "loss": 0.8104,
       "step": 21611
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3077669837659032e-06,
-      "loss": 0.7393,
+      "loss": 0.8759,
       "step": 21612
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.307014407009177e-06,
-      "loss": 0.6943,
+      "loss": 0.7978,
       "step": 21613
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3062619369817174e-06,
-      "loss": 0.7531,
+      "loss": 0.8521,
       "step": 21614
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.305509573693954e-06,
-      "loss": 0.7086,
+      "loss": 0.7873,
       "step": 21615
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.304757317156324e-06,
-      "loss": 0.7894,
+      "loss": 0.7557,
       "step": 21616
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.304005167379273e-06,
-      "loss": 0.7197,
+      "loss": 0.8501,
       "step": 21617
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3032531243732292e-06,
-      "loss": 0.7317,
+      "loss": 0.8491,
       "step": 21618
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3025011881486293e-06,
-      "loss": 0.831,
+      "loss": 0.8417,
       "step": 21619
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3017493587159013e-06,
-      "loss": 0.7963,
+      "loss": 0.9265,
       "step": 21620
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3009976360854834e-06,
-      "loss": 0.6526,
+      "loss": 0.7906,
       "step": 21621
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.3002460202677946e-06,
-      "loss": 0.6794,
+      "loss": 0.8459,
       "step": 21622
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2994945112732713e-06,
-      "loss": 0.8695,
+      "loss": 0.867,
       "step": 21623
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2987431091123345e-06,
-      "loss": 0.7173,
+      "loss": 0.8,
       "step": 21624
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2979918137954115e-06,
-      "loss": 0.7481,
+      "loss": 0.8689,
       "step": 21625
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2972406253329217e-06,
-      "loss": 0.7449,
+      "loss": 0.8301,
       "step": 21626
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.296489543735295e-06,
-      "loss": 0.7281,
+      "loss": 0.8063,
       "step": 21627
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2957385690129396e-06,
-      "loss": 0.6141,
+      "loss": 0.6868,
       "step": 21628
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2949877011762843e-06,
-      "loss": 0.7833,
+      "loss": 0.8505,
       "step": 21629
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.294236940235742e-06,
-      "loss": 0.7244,
+      "loss": 0.7694,
       "step": 21630
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2934862862017304e-06,
-      "loss": 0.7533,
+      "loss": 0.787,
       "step": 21631
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2927357390846582e-06,
-      "loss": 0.7181,
+      "loss": 0.7932,
       "step": 21632
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.291985298894949e-06,
-      "loss": 0.6654,
+      "loss": 0.7718,
       "step": 21633
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2912349656430046e-06,
-      "loss": 0.7555,
+      "loss": 0.7754,
       "step": 21634
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.290484739339234e-06,
-      "loss": 0.7471,
+      "loss": 0.7993,
       "step": 21635
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.289734619994053e-06,
-      "loss": 0.7075,
+      "loss": 0.8357,
       "step": 21636
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2889846076178646e-06,
-      "loss": 0.668,
+      "loss": 0.7396,
       "step": 21637
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2882347022210738e-06,
-      "loss": 0.7677,
+      "loss": 0.799,
       "step": 21638
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.287484903814082e-06,
-      "loss": 0.6857,
+      "loss": 0.8015,
       "step": 21639
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2867352124073008e-06,
-      "loss": 0.7126,
+      "loss": 0.7888,
       "step": 21640
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.285985628011119e-06,
-      "loss": 0.7401,
+      "loss": 0.8188,
       "step": 21641
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2852361506359445e-06,
-      "loss": 0.7834,
+      "loss": 0.8249,
       "step": 21642
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2844867802921732e-06,
-      "loss": 0.6441,
+      "loss": 0.7673,
       "step": 21643
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2837375169901997e-06,
-      "loss": 0.8122,
+      "loss": 0.8455,
       "step": 21644
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.282988360740419e-06,
-      "loss": 0.6981,
+      "loss": 0.802,
       "step": 21645
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2822393115532306e-06,
-      "loss": 0.5158,
+      "loss": 0.725,
       "step": 21646
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2814903694390157e-06,
-      "loss": 0.6322,
+      "loss": 0.6949,
       "step": 21647
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.280741534408174e-06,
-      "loss": 0.6299,
+      "loss": 0.7647,
       "step": 21648
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2799928064710906e-06,
-      "loss": 0.7473,
+      "loss": 0.7971,
       "step": 21649
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2792441856381553e-06,
-      "loss": 0.7133,
+      "loss": 0.7858,
       "step": 21650
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2784956719197494e-06,
-      "loss": 0.6142,
+      "loss": 0.6863,
       "step": 21651
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2777472653262623e-06,
-      "loss": 0.6974,
+      "loss": 0.8767,
       "step": 21652
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.276998965868079e-06,
-      "loss": 0.5678,
+      "loss": 0.6773,
       "step": 21653
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2762507735555716e-06,
-      "loss": 0.6756,
+      "loss": 0.7726,
       "step": 21654
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.275502688399129e-06,
-      "loss": 0.7086,
+      "loss": 0.7914,
       "step": 21655
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2747547104091284e-06,
-      "loss": 0.7504,
+      "loss": 0.8216,
       "step": 21656
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2740068395959437e-06,
-      "loss": 0.7222,
+      "loss": 0.7467,
       "step": 21657
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2732590759699503e-06,
-      "loss": 0.8139,
+      "loss": 0.8582,
       "step": 21658
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.27251141954153e-06,
-      "loss": 0.7135,
+      "loss": 0.8509,
       "step": 21659
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.271763870321044e-06,
-      "loss": 0.7267,
+      "loss": 0.7687,
       "step": 21660
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2710164283188717e-06,
-      "loss": 0.8512,
+      "loss": 0.8648,
       "step": 21661
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2702690935453818e-06,
-      "loss": 0.7421,
+      "loss": 0.8026,
       "step": 21662
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2695218660109407e-06,
-      "loss": 0.7783,
+      "loss": 0.8418,
       "step": 21663
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2687747457259123e-06,
-      "loss": 0.7496,
+      "loss": 0.8628,
       "step": 21664
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.268027732700668e-06,
-      "loss": 0.8218,
+      "loss": 0.8582,
       "step": 21665
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.267280826945569e-06,
-      "loss": 0.7601,
+      "loss": 0.8733,
       "step": 21666
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.266534028470977e-06,
-      "loss": 0.7599,
+      "loss": 0.7729,
       "step": 21667
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.265787337287253e-06,
-      "loss": 0.6547,
+      "loss": 0.8102,
       "step": 21668
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.265040753404757e-06,
-      "loss": 0.593,
+      "loss": 0.7088,
       "step": 21669
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2642942768338426e-06,
-      "loss": 0.6714,
+      "loss": 0.8073,
       "step": 21670
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2635479075848723e-06,
-      "loss": 0.8004,
+      "loss": 0.8386,
       "step": 21671
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2628016456681977e-06,
-      "loss": 0.7468,
+      "loss": 0.7868,
       "step": 21672
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.262055491094174e-06,
-      "loss": 0.7855,
+      "loss": 0.8648,
       "step": 21673
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2613094438731507e-06,
-      "loss": 0.6541,
+      "loss": 0.7826,
       "step": 21674
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2605635040154793e-06,
-      "loss": 0.7143,
+      "loss": 0.8586,
       "step": 21675
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.259817671531508e-06,
-      "loss": 0.6808,
+      "loss": 0.778,
       "step": 21676
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2590719464315825e-06,
-      "loss": 0.6945,
+      "loss": 0.8755,
       "step": 21677
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.258326328726056e-06,
-      "loss": 0.6589,
+      "loss": 0.7697,
       "step": 21678
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2575808184252625e-06,
-      "loss": 0.7719,
+      "loss": 0.8007,
       "step": 21679
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2568354155395523e-06,
-      "loss": 0.7533,
+      "loss": 0.884,
       "step": 21680
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.256090120079265e-06,
-      "loss": 0.7739,
+      "loss": 0.788,
       "step": 21681
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2553449320547405e-06,
-      "loss": 0.6829,
+      "loss": 0.8068,
       "step": 21682
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2545998514763133e-06,
-      "loss": 0.6852,
+      "loss": 0.87,
       "step": 21683
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2538548783543267e-06,
-      "loss": 0.7849,
+      "loss": 0.8688,
       "step": 21684
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2531100126991147e-06,
-      "loss": 0.7268,
+      "loss": 0.756,
       "step": 21685
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2523652545210085e-06,
-      "loss": 0.7015,
+      "loss": 0.7902,
       "step": 21686
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.251620603830339e-06,
-      "loss": 0.7468,
+      "loss": 0.8702,
       "step": 21687
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.250876060637446e-06,
-      "loss": 0.6515,
+      "loss": 0.7517,
       "step": 21688
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2501316249526484e-06,
-      "loss": 0.7164,
+      "loss": 0.7747,
       "step": 21689
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2493872967862804e-06,
-      "loss": 0.7786,
+      "loss": 0.8866,
       "step": 21690
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2486430761486678e-06,
-      "loss": 0.5987,
+      "loss": 0.7653,
       "step": 21691
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2478989630501345e-06,
-      "loss": 0.6784,
+      "loss": 0.7195,
       "step": 21692
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.247154957501004e-06,
-      "loss": 0.7942,
+      "loss": 0.809,
       "step": 21693
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2464110595115984e-06,
-      "loss": 0.7253,
+      "loss": 0.8946,
       "step": 21694
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2456672690922386e-06,
-      "loss": 0.7487,
+      "loss": 0.8434,
       "step": 21695
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2449235862532414e-06,
-      "loss": 0.7052,
+      "loss": 0.8212,
       "step": 21696
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2441800110049294e-06,
-      "loss": 0.7632,
+      "loss": 0.8138,
       "step": 21697
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2434365433576144e-06,
-      "loss": 0.6953,
+      "loss": 0.8161,
       "step": 21698
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.242693183321614e-06,
-      "loss": 0.6448,
+      "loss": 0.805,
       "step": 21699
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.241949930907238e-06,
-      "loss": 0.7469,
+      "loss": 0.802,
       "step": 21700
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2412067861248e-06,
-      "loss": 0.7598,
+      "loss": 0.842,
       "step": 21701
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.240463748984606e-06,
-      "loss": 0.6342,
+      "loss": 0.715,
       "step": 21702
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.239720819496972e-06,
-      "loss": 0.8106,
+      "loss": 0.7971,
       "step": 21703
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2389779976722005e-06,
-      "loss": 0.626,
+      "loss": 0.8037,
       "step": 21704
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.238235283520599e-06,
-      "loss": 0.6914,
+      "loss": 0.823,
       "step": 21705
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2374926770524663e-06,
-      "loss": 0.741,
+      "loss": 0.8136,
       "step": 21706
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2367501782781155e-06,
-      "loss": 0.6743,
+      "loss": 0.7936,
       "step": 21707
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2360077872078355e-06,
-      "loss": 0.7898,
+      "loss": 0.863,
       "step": 21708
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2352655038519342e-06,
-      "loss": 0.7222,
+      "loss": 0.8258,
       "step": 21709
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.234523328220708e-06,
-      "loss": 0.6443,
+      "loss": 0.7476,
       "step": 21710
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2337812603244525e-06,
-      "loss": 0.5586,
+      "loss": 0.6536,
       "step": 21711
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.233039300173463e-06,
-      "loss": 0.8107,
+      "loss": 0.8852,
       "step": 21712
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2322974477780336e-06,
-      "loss": 0.6963,
+      "loss": 0.7671,
       "step": 21713
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.231555703148457e-06,
-      "loss": 0.6829,
+      "loss": 0.7806,
       "step": 21714
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2308140662950185e-06,
-      "loss": 0.6066,
+      "loss": 0.7487,
       "step": 21715
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2300725372280165e-06,
-      "loss": 0.717,
+      "loss": 0.7875,
       "step": 21716
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2293311159577324e-06,
-      "loss": 0.7199,
+      "loss": 0.7744,
       "step": 21717
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.228589802494454e-06,
-      "loss": 0.6726,
+      "loss": 0.7463,
       "step": 21718
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.227848596848464e-06,
-      "loss": 0.6246,
+      "loss": 0.7566,
       "step": 21719
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2271074990300523e-06,
-      "loss": 0.7168,
+      "loss": 0.8622,
       "step": 21720
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2263665090494912e-06,
-      "loss": 0.693,
+      "loss": 0.7786,
       "step": 21721
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2256256269170672e-06,
-      "loss": 0.7681,
+      "loss": 0.8377,
       "step": 21722
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.224884852643058e-06,
-      "loss": 0.7365,
+      "loss": 0.7935,
       "step": 21723
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.22414418623774e-06,
-      "loss": 0.8236,
+      "loss": 0.8296,
       "step": 21724
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2234036277113856e-06,
-      "loss": 0.6262,
+      "loss": 0.7343,
       "step": 21725
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.222663177074278e-06,
-      "loss": 0.7571,
+      "loss": 0.8098,
       "step": 21726
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.221922834336678e-06,
-      "loss": 0.658,
+      "loss": 0.7616,
       "step": 21727
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2211825995088674e-06,
-      "loss": 0.7019,
+      "loss": 0.7025,
       "step": 21728
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.22044247260111e-06,
-      "loss": 0.6796,
+      "loss": 0.8025,
       "step": 21729
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2197024536236767e-06,
-      "loss": 0.7841,
+      "loss": 0.7787,
       "step": 21730
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.218962542586832e-06,
-      "loss": 0.6474,
+      "loss": 0.7903,
       "step": 21731
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2182227395008427e-06,
-      "loss": 0.7113,
+      "loss": 0.8538,
       "step": 21732
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.217483044375972e-06,
-      "loss": 0.71,
+      "loss": 0.7028,
       "step": 21733
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.21674345722248e-06,
-      "loss": 0.6736,
+      "loss": 0.8306,
       "step": 21734
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2160039780506317e-06,
-      "loss": 0.8329,
+      "loss": 0.897,
       "step": 21735
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.215264606870684e-06,
-      "loss": 0.7179,
+      "loss": 0.8496,
       "step": 21736
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.214525343692895e-06,
-      "loss": 0.7489,
+      "loss": 0.8508,
       "step": 21737
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.213786188527518e-06,
-      "loss": 0.581,
+      "loss": 0.7174,
       "step": 21738
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.213047141384815e-06,
-      "loss": 0.7979,
+      "loss": 0.851,
       "step": 21739
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.212308202275031e-06,
-      "loss": 0.7372,
+      "loss": 0.8153,
       "step": 21740
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.211569371208423e-06,
-      "loss": 0.6017,
+      "loss": 0.7227,
       "step": 21741
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2108306481952393e-06,
-      "loss": 0.9049,
+      "loss": 0.9222,
       "step": 21742
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.21009203324573e-06,
-      "loss": 0.7604,
+      "loss": 0.8715,
       "step": 21743
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.209353526370137e-06,
-      "loss": 0.695,
+      "loss": 0.797,
       "step": 21744
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2086151275787173e-06,
-      "loss": 0.7798,
+      "loss": 0.8134,
       "step": 21745
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2078768368817028e-06,
-      "loss": 0.793,
+      "loss": 0.8414,
       "step": 21746
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2071386542893437e-06,
-      "loss": 0.7375,
+      "loss": 0.8228,
       "step": 21747
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2064005798118792e-06,
-      "loss": 0.6959,
+      "loss": 0.799,
       "step": 21748
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.205662613459548e-06,
-      "loss": 0.8894,
+      "loss": 0.8823,
       "step": 21749
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.204924755242591e-06,
-      "loss": 0.818,
+      "loss": 0.8648,
       "step": 21750
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2041870051712398e-06,
-      "loss": 0.7311,
+      "loss": 0.7783,
       "step": 21751
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2034493632557387e-06,
-      "loss": 0.7398,
+      "loss": 0.805,
       "step": 21752
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.20271182950631e-06,
-      "loss": 0.6912,
+      "loss": 0.753,
       "step": 21753
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2019744039331947e-06,
-      "loss": 0.6588,
+      "loss": 0.7853,
       "step": 21754
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2012370865466215e-06,
-      "loss": 0.7331,
+      "loss": 0.8665,
       "step": 21755
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.2004998773568175e-06,
-      "loss": 0.7119,
+      "loss": 0.7561,
       "step": 21756
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1997627763740093e-06,
-      "loss": 0.8052,
+      "loss": 0.876,
       "step": 21757
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1990257836084306e-06,
-      "loss": 0.7698,
+      "loss": 0.9076,
       "step": 21758
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.198288899070297e-06,
-      "loss": 0.6396,
+      "loss": 0.6721,
       "step": 21759
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.197552122769837e-06,
-      "loss": 0.6435,
+      "loss": 0.7102,
       "step": 21760
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1968154547172716e-06,
-      "loss": 0.7344,
+      "loss": 0.7857,
       "step": 21761
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1960788949228194e-06,
-      "loss": 0.7294,
+      "loss": 0.7922,
       "step": 21762
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1953424433966984e-06,
-      "loss": 0.7485,
+      "loss": 0.7771,
       "step": 21763
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1946061001491314e-06,
-      "loss": 0.7471,
+      "loss": 0.8036,
       "step": 21764
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.193869865190329e-06,
-      "loss": 0.6966,
+      "loss": 0.8322,
       "step": 21765
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1931337385305073e-06,
-      "loss": 0.7324,
+      "loss": 0.8208,
       "step": 21766
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1923977201798786e-06,
-      "loss": 0.7622,
+      "loss": 0.7942,
       "step": 21767
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.191661810148653e-06,
-      "loss": 0.6949,
+      "loss": 0.8085,
       "step": 21768
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.190926008447042e-06,
-      "loss": 0.7057,
+      "loss": 0.7443,
       "step": 21769
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.19019031508525e-06,
-      "loss": 0.7582,
+      "loss": 0.8022,
       "step": 21770
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1894547300734927e-06,
-      "loss": 0.7299,
+      "loss": 0.8016,
       "step": 21771
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1887192534219636e-06,
-      "loss": 0.6251,
+      "loss": 0.7304,
       "step": 21772
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.187983885140875e-06,
-      "loss": 0.6223,
+      "loss": 0.787,
       "step": 21773
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1872486252404257e-06,
-      "loss": 0.6634,
+      "loss": 0.7927,
       "step": 21774
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.186513473730817e-06,
-      "loss": 0.7424,
+      "loss": 0.856,
       "step": 21775
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.185778430622245e-06,
-      "loss": 0.7912,
+      "loss": 0.783,
       "step": 21776
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1850434959249165e-06,
-      "loss": 0.802,
+      "loss": 0.8415,
       "step": 21777
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1843086696490157e-06,
-      "loss": 0.7851,
+      "loss": 0.8722,
       "step": 21778
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.183573951804746e-06,
-      "loss": 0.7292,
+      "loss": 0.7526,
       "step": 21779
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1828393424022977e-06,
-      "loss": 0.6728,
+      "loss": 0.8165,
       "step": 21780
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.182104841451862e-06,
-      "loss": 0.5207,
+      "loss": 0.7003,
       "step": 21781
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1813704489636265e-06,
-      "loss": 0.6634,
+      "loss": 0.8141,
       "step": 21782
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.180636164947787e-06,
-      "loss": 0.7725,
+      "loss": 0.823,
       "step": 21783
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.179901989414525e-06,
-      "loss": 0.7753,
+      "loss": 0.8263,
       "step": 21784
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1791679223740283e-06,
-      "loss": 0.6737,
+      "loss": 0.7408,
       "step": 21785
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.178433963836477e-06,
-      "loss": 0.7868,
+      "loss": 0.8688,
       "step": 21786
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1777001138120634e-06,
-      "loss": 0.8507,
+      "loss": 0.9094,
       "step": 21787
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.176966372310959e-06,
-      "loss": 0.5945,
+      "loss": 0.7403,
       "step": 21788
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1762327393433447e-06,
-      "loss": 0.6428,
+      "loss": 0.7347,
       "step": 21789
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1754992149194052e-06,
-      "loss": 0.6691,
+      "loss": 0.7858,
       "step": 21790
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1747657990493075e-06,
-      "loss": 0.7455,
+      "loss": 0.8218,
       "step": 21791
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1740324917432355e-06,
-      "loss": 0.7338,
+      "loss": 0.8302,
       "step": 21792
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.173299293011358e-06,
-      "loss": 0.7258,
+      "loss": 0.8604,
       "step": 21793
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1725662028638484e-06,
-      "loss": 0.8312,
+      "loss": 0.9183,
       "step": 21794
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.171833221310874e-06,
-      "loss": 0.8242,
+      "loss": 0.8008,
       "step": 21795
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1711003483626104e-06,
-      "loss": 0.6687,
+      "loss": 0.7769,
       "step": 21796
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1703675840292215e-06,
-      "loss": 0.7658,
+      "loss": 0.8547,
       "step": 21797
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.169634928320873e-06,
-      "loss": 0.7554,
+      "loss": 0.831,
       "step": 21798
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1689023812477306e-06,
-      "loss": 0.6257,
+      "loss": 0.7399,
       "step": 21799
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1681699428199566e-06,
-      "loss": 0.7559,
+      "loss": 0.7863,
       "step": 21800
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1674376130477093e-06,
-      "loss": 0.7185,
+      "loss": 0.765,
       "step": 21801
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1667053919411562e-06,
-      "loss": 0.6922,
+      "loss": 0.7928,
       "step": 21802
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1659732795104503e-06,
-      "loss": 0.777,
+      "loss": 0.7898,
       "step": 21803
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1652412757657516e-06,
-      "loss": 0.7366,
+      "loss": 0.8345,
       "step": 21804
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.164509380717211e-06,
-      "loss": 0.6784,
+      "loss": 0.7843,
       "step": 21805
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1637775943749906e-06,
-      "loss": 0.8226,
+      "loss": 0.8889,
       "step": 21806
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1630459167492358e-06,
-      "loss": 0.6793,
+      "loss": 0.7767,
       "step": 21807
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1623143478500963e-06,
-      "loss": 0.7025,
+      "loss": 0.8241,
       "step": 21808
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1615828876877286e-06,
-      "loss": 0.732,
+      "loss": 0.8407,
       "step": 21809
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.160851536272276e-06,
-      "loss": 0.6493,
+      "loss": 0.8153,
       "step": 21810
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1601202936138864e-06,
-      "loss": 0.8019,
+      "loss": 0.8572,
       "step": 21811
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1593891597227056e-06,
-      "loss": 0.7152,
+      "loss": 0.7926,
       "step": 21812
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.158658134608874e-06,
-      "loss": 0.6013,
+      "loss": 0.6908,
       "step": 21813
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1579272182825337e-06,
-      "loss": 0.7965,
+      "loss": 0.873,
       "step": 21814
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1571964107538303e-06,
-      "loss": 0.7444,
+      "loss": 0.8535,
       "step": 21815
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.156465712032898e-06,
-      "loss": 0.8142,
+      "loss": 0.8117,
       "step": 21816
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.155735122129875e-06,
-      "loss": 0.8053,
+      "loss": 0.9113,
       "step": 21817
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1550046410548965e-06,
-      "loss": 0.656,
+      "loss": 0.7296,
       "step": 21818
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.154274268818102e-06,
-      "loss": 0.6262,
+      "loss": 0.7736,
       "step": 21819
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.153544005429615e-06,
-      "loss": 0.6876,
+      "loss": 0.7761,
       "step": 21820
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1528138508995757e-06,
-      "loss": 0.6321,
+      "loss": 0.7279,
       "step": 21821
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.15208380523811e-06,
-      "loss": 0.7005,
+      "loss": 0.7639,
       "step": 21822
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.151353868455347e-06,
-      "loss": 0.7386,
+      "loss": 0.8137,
       "step": 21823
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.150624040561411e-06,
-      "loss": 0.6676,
+      "loss": 0.7445,
       "step": 21824
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1498943215664346e-06,
-      "loss": 0.7137,
+      "loss": 0.8321,
       "step": 21825
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1491647114805346e-06,
-      "loss": 0.6638,
+      "loss": 0.721,
       "step": 21826
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1484352103138317e-06,
-      "loss": 0.7146,
+      "loss": 0.8574,
       "step": 21827
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.147705818076453e-06,
-      "loss": 0.7983,
+      "loss": 0.8573,
       "step": 21828
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1469765347785153e-06,
-      "loss": 0.4862,
+      "loss": 0.6705,
       "step": 21829
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.146247360430137e-06,
-      "loss": 0.6869,
+      "loss": 0.8156,
       "step": 21830
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.14551829504143e-06,
-      "loss": 0.7062,
+      "loss": 0.8059,
       "step": 21831
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1447893386225182e-06,
-      "loss": 0.7188,
+      "loss": 0.7782,
       "step": 21832
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.144060491183504e-06,
-      "loss": 0.6732,
+      "loss": 0.7491,
       "step": 21833
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.143331752734508e-06,
-      "loss": 0.7433,
+      "loss": 0.8351,
       "step": 21834
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.142603123285635e-06,
-      "loss": 0.6864,
+      "loss": 0.7755,
       "step": 21835
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1418746028469963e-06,
-      "loss": 0.6403,
+      "loss": 0.7732,
       "step": 21836
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1411461914286957e-06,
-      "loss": 0.6857,
+      "loss": 0.7707,
       "step": 21837
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1404178890408468e-06,
-      "loss": 0.7652,
+      "loss": 0.8403,
       "step": 21838
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.139689695693543e-06,
-      "loss": 0.6661,
+      "loss": 0.7143,
       "step": 21839
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1389616113968946e-06,
-      "loss": 0.7709,
+      "loss": 0.8058,
       "step": 21840
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1382336361610012e-06,
-      "loss": 0.9151,
+      "loss": 0.9299,
       "step": 21841
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.137505769995961e-06,
-      "loss": 0.687,
+      "loss": 0.7762,
       "step": 21842
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.136778012911871e-06,
-      "loss": 0.5415,
+      "loss": 0.6875,
       "step": 21843
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1360503649188335e-06,
-      "loss": 0.7448,
+      "loss": 0.8412,
       "step": 21844
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1353228260269387e-06,
-      "loss": 0.6334,
+      "loss": 0.7413,
       "step": 21845
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1345953962462775e-06,
-      "loss": 0.6187,
+      "loss": 0.7741,
       "step": 21846
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.133868075586948e-06,
-      "loss": 0.757,
+      "loss": 0.8623,
       "step": 21847
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.133140864059039e-06,
-      "loss": 0.7131,
+      "loss": 0.7426,
       "step": 21848
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1324137616726394e-06,
-      "loss": 0.5898,
+      "loss": 0.712,
       "step": 21849
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1316867684378327e-06,
-      "loss": 0.7631,
+      "loss": 0.8044,
       "step": 21850
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.130959884364714e-06,
-      "loss": 0.7493,
+      "loss": 0.8137,
       "step": 21851
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1302331094633568e-06,
-      "loss": 0.6937,
+      "loss": 0.7724,
       "step": 21852
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.129506443743852e-06,
-      "loss": 0.666,
+      "loss": 0.7964,
       "step": 21853
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.128779887216279e-06,
-      "loss": 0.7264,
+      "loss": 0.7814,
       "step": 21854
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.128053439890717e-06,
-      "loss": 0.7843,
+      "loss": 0.8963,
       "step": 21855
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1273271017772424e-06,
-      "loss": 0.7701,
+      "loss": 0.782,
       "step": 21856
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1266008728859387e-06,
-      "loss": 0.581,
+      "loss": 0.6503,
       "step": 21857
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.125874753226872e-06,
-      "loss": 0.7922,
+      "loss": 0.8365,
       "step": 21858
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1251487428101245e-06,
-      "loss": 0.5432,
+      "loss": 0.7279,
       "step": 21859
     },
     {
       "epoch": 0.79,
       "learning_rate": 2.1244228416457645e-06,
-      "loss": 0.6197,
+      "loss": 0.7566,
       "step": 21860
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1236970497438646e-06,
-      "loss": 0.5982,
+      "loss": 0.7536,
       "step": 21861
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1229713671144903e-06,
-      "loss": 0.62,
+      "loss": 0.7058,
       "step": 21862
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1222457937677144e-06,
-      "loss": 0.6047,
+      "loss": 0.7002,
       "step": 21863
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1215203297136035e-06,
-      "loss": 0.7143,
+      "loss": 0.8161,
       "step": 21864
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.120794974962215e-06,
-      "loss": 0.6882,
+      "loss": 0.7387,
       "step": 21865
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1200697295236193e-06,
-      "loss": 0.6223,
+      "loss": 0.6956,
       "step": 21866
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.119344593407876e-06,
-      "loss": 0.6895,
+      "loss": 0.7896,
       "step": 21867
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1186195666250455e-06,
-      "loss": 0.7114,
+      "loss": 0.7749,
       "step": 21868
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.117894649185184e-06,
-      "loss": 0.6877,
+      "loss": 0.8145,
       "step": 21869
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1171698410983554e-06,
-      "loss": 0.6298,
+      "loss": 0.7195,
       "step": 21870
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.116445142374607e-06,
-      "loss": 0.8363,
+      "loss": 0.8389,
       "step": 21871
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.115720553023999e-06,
-      "loss": 0.6341,
+      "loss": 0.7477,
       "step": 21872
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.114996073056581e-06,
-      "loss": 0.7825,
+      "loss": 0.8644,
       "step": 21873
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1142717024824067e-06,
-      "loss": 0.6898,
+      "loss": 0.775,
       "step": 21874
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1135474413115207e-06,
-      "loss": 0.751,
+      "loss": 0.8673,
       "step": 21875
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1128232895539803e-06,
-      "loss": 0.7228,
+      "loss": 0.8294,
       "step": 21876
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.11209924721982e-06,
-      "loss": 0.5997,
+      "loss": 0.7442,
       "step": 21877
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.111375314319094e-06,
-      "loss": 0.7384,
+      "loss": 0.8653,
       "step": 21878
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1106514908618437e-06,
-      "loss": 0.6641,
+      "loss": 0.7579,
       "step": 21879
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1099277768581107e-06,
-      "loss": 0.6402,
+      "loss": 0.6899,
       "step": 21880
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1092041723179312e-06,
-      "loss": 0.6618,
+      "loss": 0.7533,
       "step": 21881
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1084806772513523e-06,
-      "loss": 0.8423,
+      "loss": 0.8428,
       "step": 21882
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1077572916684095e-06,
-      "loss": 0.6509,
+      "loss": 0.7404,
       "step": 21883
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1070340155791314e-06,
-      "loss": 0.6759,
+      "loss": 0.7305,
       "step": 21884
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1063108489935604e-06,
-      "loss": 0.8954,
+      "loss": 0.9641,
       "step": 21885
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.105587791921727e-06,
-      "loss": 0.6307,
+      "loss": 0.803,
       "step": 21886
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1048648443736617e-06,
-      "loss": 0.6052,
+      "loss": 0.6887,
       "step": 21887
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.104142006359392e-06,
-      "loss": 0.7514,
+      "loss": 0.877,
       "step": 21888
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.103419277888955e-06,
-      "loss": 0.7012,
+      "loss": 0.7351,
       "step": 21889
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1026966589723665e-06,
-      "loss": 0.7805,
+      "loss": 0.7806,
       "step": 21890
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.1019741496196597e-06,
-      "loss": 0.5549,
+      "loss": 0.7195,
       "step": 21891
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.101251749840856e-06,
-      "loss": 0.786,
+      "loss": 0.8535,
       "step": 21892
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.100529459645977e-06,
-      "loss": 0.6702,
+      "loss": 0.7999,
       "step": 21893
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0998072790450406e-06,
-      "loss": 0.6976,
+      "loss": 0.842,
       "step": 21894
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.099085208048073e-06,
-      "loss": 0.7513,
+      "loss": 0.863,
       "step": 21895
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.098363246665088e-06,
-      "loss": 0.8558,
+      "loss": 0.8051,
       "step": 21896
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0976413949061025e-06,
-      "loss": 0.7608,
+      "loss": 0.815,
       "step": 21897
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.09691965278113e-06,
-      "loss": 0.7414,
+      "loss": 0.7891,
       "step": 21898
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.096198020300184e-06,
-      "loss": 0.6532,
+      "loss": 0.7112,
       "step": 21899
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0954764974732744e-06,
-      "loss": 0.6733,
+      "loss": 0.8101,
       "step": 21900
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.094755084310416e-06,
-      "loss": 0.7939,
+      "loss": 0.8345,
       "step": 21901
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.094033780821614e-06,
-      "loss": 0.8046,
+      "loss": 0.8359,
       "step": 21902
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.093312587016877e-06,
-      "loss": 0.7314,
+      "loss": 0.8152,
       "step": 21903
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0925915029062083e-06,
-      "loss": 0.7561,
+      "loss": 0.8384,
       "step": 21904
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0918705284996157e-06,
-      "loss": 0.6325,
+      "loss": 0.7569,
       "step": 21905
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.091149663807097e-06,
-      "loss": 0.7076,
+      "loss": 0.802,
       "step": 21906
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0904289088386543e-06,
-      "loss": 0.5777,
+      "loss": 0.7099,
       "step": 21907
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0897082636042908e-06,
-      "loss": 0.6161,
+      "loss": 0.6868,
       "step": 21908
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0889877281140015e-06,
-      "loss": 0.7422,
+      "loss": 0.8738,
       "step": 21909
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.088267302377783e-06,
-      "loss": 0.748,
+      "loss": 0.7988,
       "step": 21910
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.087546986405631e-06,
-      "loss": 0.8091,
+      "loss": 0.831,
       "step": 21911
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.086826780207538e-06,
-      "loss": 0.6951,
+      "loss": 0.8128,
       "step": 21912
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.086106683793494e-06,
-      "loss": 0.6266,
+      "loss": 0.7525,
       "step": 21913
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.085386697173494e-06,
-      "loss": 0.7237,
+      "loss": 0.7962,
       "step": 21914
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0846668203575225e-06,
-      "loss": 0.6797,
+      "loss": 0.8327,
       "step": 21915
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0839470533555704e-06,
-      "loss": 0.7337,
+      "loss": 0.787,
       "step": 21916
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0832273961776182e-06,
-      "loss": 0.6674,
+      "loss": 0.7792,
       "step": 21917
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.082507848833659e-06,
-      "loss": 0.633,
+      "loss": 0.738,
       "step": 21918
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0817884113336638e-06,
-      "loss": 0.6714,
+      "loss": 0.8476,
       "step": 21919
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0810690836876235e-06,
-      "loss": 0.6361,
+      "loss": 0.6674,
       "step": 21920
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0803498659055144e-06,
-      "loss": 0.7718,
+      "loss": 0.8993,
       "step": 21921
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.079630757997313e-06,
-      "loss": 0.7469,
+      "loss": 0.7629,
       "step": 21922
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.078911759972998e-06,
-      "loss": 0.7424,
+      "loss": 0.7733,
       "step": 21923
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.078192871842545e-06,
-      "loss": 0.6678,
+      "loss": 0.787,
       "step": 21924
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.077474093615924e-06,
-      "loss": 0.8869,
+      "loss": 0.8706,
       "step": 21925
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0767554253031088e-06,
-      "loss": 0.7157,
+      "loss": 0.7885,
       "step": 21926
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0760368669140716e-06,
-      "loss": 0.6066,
+      "loss": 0.6873,
       "step": 21927
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0753184184587814e-06,
-      "loss": 0.6196,
+      "loss": 0.7885,
       "step": 21928
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0746000799472043e-06,
-      "loss": 0.8856,
+      "loss": 0.9171,
       "step": 21929
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.073881851389303e-06,
-      "loss": 0.674,
+      "loss": 0.8015,
       "step": 21930
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0731637327950514e-06,
-      "loss": 0.6654,
+      "loss": 0.7305,
       "step": 21931
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0724457241744012e-06,
-      "loss": 0.6783,
+      "loss": 0.7048,
       "step": 21932
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.071727825537321e-06,
-      "loss": 0.6612,
+      "loss": 0.743,
       "step": 21933
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.071010036893768e-06,
-      "loss": 0.7898,
+      "loss": 0.826,
       "step": 21934
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0702923582537015e-06,
-      "loss": 0.6958,
+      "loss": 0.7609,
       "step": 21935
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0695747896270746e-06,
-      "loss": 0.7003,
+      "loss": 0.8013,
       "step": 21936
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0688573310238513e-06,
-      "loss": 0.5769,
+      "loss": 0.7428,
       "step": 21937
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.068139982453975e-06,
-      "loss": 0.8142,
+      "loss": 0.9164,
       "step": 21938
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0674227439274053e-06,
-      "loss": 0.6884,
+      "loss": 0.7828,
       "step": 21939
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0667056154540886e-06,
-      "loss": 0.6186,
+      "loss": 0.7097,
       "step": 21940
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0659885970439773e-06,
-      "loss": 0.7402,
+      "loss": 0.9126,
       "step": 21941
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.065271688707017e-06,
-      "loss": 0.8967,
+      "loss": 0.9315,
       "step": 21942
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.064554890453153e-06,
-      "loss": 0.6117,
+      "loss": 0.7508,
       "step": 21943
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.063838202292332e-06,
-      "loss": 0.6679,
+      "loss": 0.7388,
       "step": 21944
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0631216242344933e-06,
-      "loss": 0.7586,
+      "loss": 0.7998,
       "step": 21945
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.062405156289584e-06,
-      "loss": 0.5514,
+      "loss": 0.6841,
       "step": 21946
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0616887984675405e-06,
-      "loss": 0.5895,
+      "loss": 0.8148,
       "step": 21947
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0609725507783017e-06,
-      "loss": 0.7845,
+      "loss": 0.8489,
       "step": 21948
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.060256413231804e-06,
-      "loss": 0.7186,
+      "loss": 0.7819,
       "step": 21949
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.059540385837987e-06,
-      "loss": 0.7648,
+      "loss": 0.8675,
       "step": 21950
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0588244686067758e-06,
-      "loss": 0.7131,
+      "loss": 0.7789,
       "step": 21951
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0581086615481114e-06,
-      "loss": 0.8857,
+      "loss": 0.8459,
       "step": 21952
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.05739296467192e-06,
-      "loss": 0.7334,
+      "loss": 0.8569,
       "step": 21953
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0566773779881344e-06,
-      "loss": 0.7061,
+      "loss": 0.7765,
       "step": 21954
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0559619015066757e-06,
-      "loss": 0.6968,
+      "loss": 0.7421,
       "step": 21955
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0552465352374807e-06,
-      "loss": 0.6668,
+      "loss": 0.7911,
       "step": 21956
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0545312791904613e-06,
-      "loss": 0.6892,
+      "loss": 0.8264,
       "step": 21957
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0538161333755512e-06,
-      "loss": 0.7691,
+      "loss": 0.8292,
       "step": 21958
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.053101097802668e-06,
-      "loss": 0.7028,
+      "loss": 0.7835,
       "step": 21959
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0523861724817316e-06,
-      "loss": 0.7107,
+      "loss": 0.8593,
       "step": 21960
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.051671357422661e-06,
-      "loss": 0.8008,
+      "loss": 0.8327,
       "step": 21961
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.05095665263537e-06,
-      "loss": 0.7714,
+      "loss": 0.8022,
       "step": 21962
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0502420581297834e-06,
-      "loss": 0.6415,
+      "loss": 0.8447,
       "step": 21963
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0495275739158027e-06,
-      "loss": 0.6347,
+      "loss": 0.7986,
       "step": 21964
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0488132000033502e-06,
-      "loss": 0.7042,
+      "loss": 0.7934,
       "step": 21965
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0480989364023316e-06,
-      "loss": 0.797,
+      "loss": 0.8593,
       "step": 21966
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.04738478312266e-06,
-      "loss": 0.7965,
+      "loss": 0.7875,
       "step": 21967
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.046670740174238e-06,
-      "loss": 0.6571,
+      "loss": 0.6803,
       "step": 21968
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.04595680756698e-06,
-      "loss": 0.6492,
+      "loss": 0.7807,
       "step": 21969
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0452429853107814e-06,
-      "loss": 0.6824,
+      "loss": 0.7147,
       "step": 21970
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.044529273415552e-06,
-      "loss": 0.6893,
+      "loss": 0.8323,
       "step": 21971
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0438156718911918e-06,
-      "loss": 0.6572,
+      "loss": 0.6921,
       "step": 21972
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0431021807476004e-06,
-      "loss": 0.618,
+      "loss": 0.7797,
       "step": 21973
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0423887999946744e-06,
-      "loss": 0.6293,
+      "loss": 0.7434,
       "step": 21974
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0416755296423173e-06,
-      "loss": 0.6943,
+      "loss": 0.8182,
       "step": 21975
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.04096236970042e-06,
-      "loss": 0.6533,
+      "loss": 0.7463,
       "step": 21976
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0402493201788774e-06,
-      "loss": 0.6897,
+      "loss": 0.8368,
       "step": 21977
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0395363810875814e-06,
-      "loss": 0.6295,
+      "loss": 0.7282,
       "step": 21978
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.038823552436424e-06,
-      "loss": 0.666,
+      "loss": 0.7972,
       "step": 21979
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0381108342352953e-06,
-      "loss": 0.6772,
+      "loss": 0.7007,
       "step": 21980
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0373982264940793e-06,
-      "loss": 0.7695,
+      "loss": 0.8806,
       "step": 21981
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0366857292226707e-06,
-      "loss": 0.7143,
+      "loss": 0.735,
       "step": 21982
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0359733424309434e-06,
-      "loss": 0.9089,
+      "loss": 0.9226,
       "step": 21983
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.03526106612879e-06,
-      "loss": 0.7458,
+      "loss": 0.8267,
       "step": 21984
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0345489003260875e-06,
-      "loss": 0.6341,
+      "loss": 0.8066,
       "step": 21985
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0338368450327184e-06,
-      "loss": 0.7081,
+      "loss": 0.7772,
       "step": 21986
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0331249002585574e-06,
-      "loss": 0.745,
+      "loss": 0.8307,
       "step": 21987
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.032413066013491e-06,
-      "loss": 0.6368,
+      "loss": 0.7516,
       "step": 21988
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0317013423073827e-06,
-      "loss": 0.6971,
+      "loss": 0.805,
       "step": 21989
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.030989729150116e-06,
-      "loss": 0.6605,
+      "loss": 0.7787,
       "step": 21990
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0302782265515598e-06,
-      "loss": 0.6833,
+      "loss": 0.7267,
       "step": 21991
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0295668345215857e-06,
-      "loss": 0.7642,
+      "loss": 0.8046,
       "step": 21992
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0288555530700604e-06,
-      "loss": 0.7155,
+      "loss": 0.7773,
       "step": 21993
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.028144382206858e-06,
-      "loss": 0.7919,
+      "loss": 0.8592,
       "step": 21994
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0274333219418417e-06,
-      "loss": 0.7505,
+      "loss": 0.8466,
       "step": 21995
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.026722372284876e-06,
-      "loss": 0.8012,
+      "loss": 0.8141,
       "step": 21996
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.026011533245822e-06,
-      "loss": 0.7892,
+      "loss": 0.8531,
       "step": 21997
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.02530080483455e-06,
-      "loss": 0.7894,
+      "loss": 0.8306,
       "step": 21998
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.024590187060912e-06,
-      "loss": 0.5877,
+      "loss": 0.6795,
       "step": 21999
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0238796799347672e-06,
-      "loss": 0.7201,
+      "loss": 0.7793,
       "step": 22000
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.02316928346598e-06,
-      "loss": 0.6154,
+      "loss": 0.6828,
       "step": 22001
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.022458997664396e-06,
-      "loss": 0.7821,
+      "loss": 0.8565,
       "step": 22002
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0217488225398784e-06,
-      "loss": 0.6877,
+      "loss": 0.7835,
       "step": 22003
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.021038758102275e-06,
-      "loss": 0.6292,
+      "loss": 0.7693,
       "step": 22004
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.020328804361439e-06,
-      "loss": 0.8732,
+      "loss": 0.9036,
       "step": 22005
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0196189613272166e-06,
-      "loss": 0.7312,
+      "loss": 0.7593,
       "step": 22006
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0189092290094603e-06,
-      "loss": 0.749,
+      "loss": 0.8014,
       "step": 22007
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0181996074180154e-06,
-      "loss": 0.7756,
+      "loss": 0.7582,
       "step": 22008
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0174900965627253e-06,
-      "loss": 0.6228,
+      "loss": 0.7609,
       "step": 22009
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.016780696453435e-06,
-      "loss": 0.7141,
+      "loss": 0.8336,
       "step": 22010
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0160714070999865e-06,
-      "loss": 0.69,
+      "loss": 0.7415,
       "step": 22011
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0153622285122165e-06,
-      "loss": 0.802,
+      "loss": 0.8377,
       "step": 22012
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.014653160699969e-06,
-      "loss": 0.5784,
+      "loss": 0.7163,
       "step": 22013
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0139442036730793e-06,
-      "loss": 0.6883,
+      "loss": 0.7294,
       "step": 22014
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0132353574413832e-06,
-      "loss": 0.8062,
+      "loss": 0.8675,
       "step": 22015
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.012526622014712e-06,
-      "loss": 0.8113,
+      "loss": 0.9126,
       "step": 22016
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0118179974029072e-06,
-      "loss": 0.7452,
+      "loss": 0.8098,
       "step": 22017
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0111094836157907e-06,
-      "loss": 0.6388,
+      "loss": 0.7774,
       "step": 22018
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0104010806631924e-06,
-      "loss": 0.6766,
+      "loss": 0.8015,
       "step": 22019
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0096927885549464e-06,
-      "loss": 0.7905,
+      "loss": 0.8813,
       "step": 22020
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0089846073008758e-06,
-      "loss": 0.5895,
+      "loss": 0.643,
       "step": 22021
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.008276536910806e-06,
-      "loss": 0.729,
+      "loss": 0.792,
       "step": 22022
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.007568577394561e-06,
-      "loss": 0.5481,
+      "loss": 0.7159,
       "step": 22023
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0068607287619625e-06,
-      "loss": 0.7279,
+      "loss": 0.796,
       "step": 22024
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0061529910228273e-06,
-      "loss": 0.7069,
+      "loss": 0.9061,
       "step": 22025
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0054453641869797e-06,
-      "loss": 0.7719,
+      "loss": 0.8635,
       "step": 22026
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0047378482642354e-06,
-      "loss": 0.8286,
+      "loss": 0.8588,
       "step": 22027
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.004030443264411e-06,
-      "loss": 0.6734,
+      "loss": 0.7454,
       "step": 22028
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.003323149197315e-06,
-      "loss": 0.6916,
+      "loss": 0.8051,
       "step": 22029
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0026159660727705e-06,
-      "loss": 0.7254,
+      "loss": 0.7559,
       "step": 22030
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.001908893900577e-06,
-      "loss": 0.6797,
+      "loss": 0.8142,
       "step": 22031
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.001201932690553e-06,
-      "loss": 0.7483,
+      "loss": 0.8591,
       "step": 22032
     },
     {
       "epoch": 0.8,
       "learning_rate": 2.0004950824525027e-06,
-      "loss": 0.7347,
+      "loss": 0.7527,
       "step": 22033
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.999788343196234e-06,
-      "loss": 0.7037,
+      "loss": 0.7925,
       "step": 22034
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9990817149315478e-06,
-      "loss": 0.7281,
+      "loss": 0.8087,
       "step": 22035
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9983751976682564e-06,
-      "loss": 0.7526,
+      "loss": 0.7885,
       "step": 22036
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.997668791416154e-06,
-      "loss": 0.5663,
+      "loss": 0.6197,
       "step": 22037
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9969624961850407e-06,
-      "loss": 0.7422,
+      "loss": 0.7932,
       "step": 22038
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.996256311984721e-06,
-      "loss": 0.6196,
+      "loss": 0.7761,
       "step": 22039
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.995550238824988e-06,
-      "loss": 0.7283,
+      "loss": 0.8484,
       "step": 22040
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9948442767156395e-06,
-      "loss": 0.758,
+      "loss": 0.8259,
       "step": 22041
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9941384256664684e-06,
-      "loss": 0.8207,
+      "loss": 0.8752,
       "step": 22042
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9934326856872687e-06,
-      "loss": 0.6445,
+      "loss": 0.7825,
       "step": 22043
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9927270567878278e-06,
-      "loss": 0.7396,
+      "loss": 0.8273,
       "step": 22044
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9920215389779395e-06,
-      "loss": 0.6593,
+      "loss": 0.7664,
       "step": 22045
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9913161322673913e-06,
-      "loss": 0.7923,
+      "loss": 0.823,
       "step": 22046
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9906108366659695e-06,
-      "loss": 0.6742,
+      "loss": 0.8586,
       "step": 22047
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9899056521834546e-06,
-      "loss": 0.7922,
+      "loss": 0.926,
       "step": 22048
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9892005788296397e-06,
-      "loss": 0.5853,
+      "loss": 0.7285,
       "step": 22049
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9884956166142965e-06,
-      "loss": 0.7581,
+      "loss": 0.7486,
       "step": 22050
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9877907655472118e-06,
-      "loss": 0.622,
+      "loss": 0.7284,
       "step": 22051
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.987086025638162e-06,
-      "loss": 0.6445,
+      "loss": 0.7984,
       "step": 22052
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9863813968969257e-06,
-      "loss": 0.7494,
+      "loss": 0.813,
       "step": 22053
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.985676879333275e-06,
-      "loss": 0.6535,
+      "loss": 0.7449,
       "step": 22054
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9849724729569907e-06,
-      "loss": 0.6212,
+      "loss": 0.7681,
       "step": 22055
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9842681777778395e-06,
-      "loss": 0.6597,
+      "loss": 0.7687,
       "step": 22056
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.983563993805593e-06,
-      "loss": 0.7496,
+      "loss": 0.8429,
       "step": 22057
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.982859921050023e-06,
-      "loss": 0.7209,
+      "loss": 0.8231,
       "step": 22058
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9821559595208984e-06,
-      "loss": 0.6585,
+      "loss": 0.7575,
       "step": 22059
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9814521092279825e-06,
-      "loss": 0.8164,
+      "loss": 0.8416,
       "step": 22060
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9807483701810404e-06,
-      "loss": 0.9128,
+      "loss": 0.9057,
       "step": 22061
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.980044742389842e-06,
-      "loss": 0.7152,
+      "loss": 0.8374,
       "step": 22062
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9793412258641386e-06,
-      "loss": 0.7928,
+      "loss": 0.8643,
       "step": 22063
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.978637820613698e-06,
-      "loss": 0.7163,
+      "loss": 0.8013,
       "step": 22064
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9779345266482773e-06,
-      "loss": 0.8242,
+      "loss": 0.8369,
       "step": 22065
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9772313439776325e-06,
-      "loss": 0.7122,
+      "loss": 0.7752,
       "step": 22066
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9765282726115177e-06,
-      "loss": 0.6893,
+      "loss": 0.8014,
       "step": 22067
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.975825312559695e-06,
-      "loss": 0.6807,
+      "loss": 0.7753,
       "step": 22068
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9751224638319056e-06,
-      "loss": 0.5938,
+      "loss": 0.7078,
       "step": 22069
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9744197264379096e-06,
-      "loss": 0.7482,
+      "loss": 0.8865,
       "step": 22070
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9737171003874524e-06,
-      "loss": 0.6951,
+      "loss": 0.8497,
       "step": 22071
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9730145856902826e-06,
-      "loss": 0.6922,
+      "loss": 0.8726,
       "step": 22072
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.972312182356143e-06,
-      "loss": 0.7568,
+      "loss": 0.7632,
       "step": 22073
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.971609890394787e-06,
-      "loss": 0.8276,
+      "loss": 0.9201,
       "step": 22074
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9709077098159536e-06,
-      "loss": 0.6643,
+      "loss": 0.772,
       "step": 22075
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.97020564062938e-06,
-      "loss": 0.8188,
+      "loss": 0.8676,
       "step": 22076
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.969503682844812e-06,
-      "loss": 0.6253,
+      "loss": 0.7544,
       "step": 22077
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9688018364719875e-06,
-      "loss": 0.6563,
+      "loss": 0.7702,
       "step": 22078
     },
     {
@@ -132484,9505 +132484,9505 @@
     {
       "epoch": 0.8,
       "learning_rate": 1.9673984780005097e-06,
-      "loss": 0.7242,
+      "loss": 0.8671,
       "step": 22080
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.966696965921331e-06,
-      "loss": 0.6837,
+      "loss": 0.764,
       "step": 22081
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9659955652928297e-06,
-      "loss": 0.8003,
+      "loss": 0.7716,
       "step": 22082
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.965294276124744e-06,
-      "loss": 0.6981,
+      "loss": 0.7553,
       "step": 22083
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9645930984268004e-06,
-      "loss": 0.6371,
+      "loss": 0.7109,
       "step": 22084
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.963892032208726e-06,
-      "loss": 0.6958,
+      "loss": 0.7915,
       "step": 22085
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9631910774802463e-06,
-      "loss": 0.5916,
+      "loss": 0.7194,
       "step": 22086
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.962490234251092e-06,
-      "loss": 0.6867,
+      "loss": 0.8209,
       "step": 22087
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9617895025309773e-06,
-      "loss": 0.708,
+      "loss": 0.8522,
       "step": 22088
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9610888823296316e-06,
-      "loss": 0.8152,
+      "loss": 0.8749,
       "step": 22089
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9603883736567707e-06,
-      "loss": 0.668,
+      "loss": 0.7698,
       "step": 22090
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9596879765221154e-06,
-      "loss": 0.6974,
+      "loss": 0.8408,
       "step": 22091
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9589876909353787e-06,
-      "loss": 0.6928,
+      "loss": 0.8238,
       "step": 22092
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.958287516906282e-06,
-      "loss": 0.7155,
+      "loss": 0.7394,
       "step": 22093
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9575874544445383e-06,
-      "loss": 0.6504,
+      "loss": 0.7381,
       "step": 22094
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9568875035598536e-06,
-      "loss": 0.7906,
+      "loss": 0.8153,
       "step": 22095
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9561876642619447e-06,
-      "loss": 0.6757,
+      "loss": 0.7823,
       "step": 22096
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9554879365605207e-06,
-      "loss": 0.67,
+      "loss": 0.7522,
       "step": 22097
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9547883204652874e-06,
-      "loss": 0.7824,
+      "loss": 0.8316,
       "step": 22098
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9540888159859483e-06,
-      "loss": 0.6617,
+      "loss": 0.8114,
       "step": 22099
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.953389423132217e-06,
-      "loss": 0.7563,
+      "loss": 0.852,
       "step": 22100
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9526901419137843e-06,
-      "loss": 0.6628,
+      "loss": 0.8009,
       "step": 22101
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.951990972340362e-06,
-      "loss": 0.7237,
+      "loss": 0.8029,
       "step": 22102
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.951291914421647e-06,
-      "loss": 0.7088,
+      "loss": 0.7265,
       "step": 22103
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.950592968167335e-06,
-      "loss": 0.7399,
+      "loss": 0.8158,
       "step": 22104
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9498941335871247e-06,
-      "loss": 0.5527,
+      "loss": 0.7509,
       "step": 22105
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.949195410690713e-06,
-      "loss": 0.768,
+      "loss": 0.8772,
       "step": 22106
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9484967994877923e-06,
-      "loss": 0.5879,
+      "loss": 0.7345,
       "step": 22107
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.947798299988055e-06,
-      "loss": 0.7594,
+      "loss": 0.8263,
       "step": 22108
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9470999122011924e-06,
-      "loss": 0.687,
+      "loss": 0.7697,
       "step": 22109
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.946401636136892e-06,
-      "loss": 0.7148,
+      "loss": 0.7754,
       "step": 22110
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.945703471804841e-06,
-      "loss": 0.8112,
+      "loss": 0.7682,
       "step": 22111
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9450054192147274e-06,
-      "loss": 0.7094,
+      "loss": 0.8129,
       "step": 22112
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9443074783762393e-06,
-      "loss": 0.6195,
+      "loss": 0.7243,
       "step": 22113
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9436096492990497e-06,
-      "loss": 0.7792,
+      "loss": 0.841,
       "step": 22114
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.942911931992848e-06,
-      "loss": 0.8669,
+      "loss": 0.9184,
       "step": 22115
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.942214326467312e-06,
-      "loss": 0.7849,
+      "loss": 0.8183,
       "step": 22116
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.94151683273212e-06,
-      "loss": 0.7133,
+      "loss": 0.8269,
       "step": 22117
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.940819450796946e-06,
-      "loss": 0.6439,
+      "loss": 0.741,
       "step": 22118
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9401221806714687e-06,
-      "loss": 0.689,
+      "loss": 0.7902,
       "step": 22119
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.939425022365362e-06,
-      "loss": 0.7451,
+      "loss": 0.8197,
       "step": 22120
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.938727975888297e-06,
-      "loss": 0.5532,
+      "loss": 0.7068,
       "step": 22121
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9380310412499435e-06,
-      "loss": 0.5926,
+      "loss": 0.6786,
       "step": 22122
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.937334218459971e-06,
-      "loss": 0.819,
+      "loss": 0.8228,
       "step": 22123
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.936637507528043e-06,
-      "loss": 0.6473,
+      "loss": 0.6846,
       "step": 22124
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9359409084638324e-06,
-      "loss": 0.7788,
+      "loss": 0.7388,
       "step": 22125
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9352444212770016e-06,
-      "loss": 0.774,
+      "loss": 0.8226,
       "step": 22126
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9345480459772104e-06,
-      "loss": 0.8025,
+      "loss": 0.8814,
       "step": 22127
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.93385178257412e-06,
-      "loss": 0.7094,
+      "loss": 0.7836,
       "step": 22128
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.933155631077397e-06,
-      "loss": 0.7723,
+      "loss": 0.8671,
       "step": 22129
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.932459591496688e-06,
-      "loss": 0.8072,
+      "loss": 0.8685,
       "step": 22130
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9317636638416592e-06,
-      "loss": 0.7069,
+      "loss": 0.8041,
       "step": 22131
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9310678481219646e-06,
-      "loss": 0.6286,
+      "loss": 0.6982,
       "step": 22132
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9303721443472513e-06,
-      "loss": 0.6767,
+      "loss": 0.7299,
       "step": 22133
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.9296765525271764e-06,
-      "loss": 0.5765,
+      "loss": 0.6742,
       "step": 22134
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.928981072671389e-06,
-      "loss": 0.7059,
+      "loss": 0.7859,
       "step": 22135
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9282857047895385e-06,
-      "loss": 0.6951,
+      "loss": 0.7755,
       "step": 22136
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.927590448891269e-06,
-      "loss": 0.7565,
+      "loss": 0.8285,
       "step": 22137
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9268953049862306e-06,
-      "loss": 0.5943,
+      "loss": 0.6838,
       "step": 22138
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9262002730840667e-06,
-      "loss": 0.6955,
+      "loss": 0.7645,
       "step": 22139
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.925505353194418e-06,
-      "loss": 0.7551,
+      "loss": 0.8055,
       "step": 22140
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.924810545326924e-06,
-      "loss": 0.6871,
+      "loss": 0.7702,
       "step": 22141
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9241158494912314e-06,
-      "loss": 0.6392,
+      "loss": 0.7131,
       "step": 22142
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.923421265696969e-06,
-      "loss": 0.628,
+      "loss": 0.7052,
       "step": 22143
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9227267939537785e-06,
-      "loss": 0.7252,
+      "loss": 0.7758,
       "step": 22144
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.922032434271295e-06,
-      "loss": 0.7634,
+      "loss": 0.7944,
       "step": 22145
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9213381866591496e-06,
-      "loss": 0.6633,
+      "loss": 0.7647,
       "step": 22146
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.920644051126972e-06,
-      "loss": 0.7511,
+      "loss": 0.797,
       "step": 22147
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.919950027684401e-06,
-      "loss": 0.809,
+      "loss": 0.8546,
       "step": 22148
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9192561163410528e-06,
-      "loss": 0.6645,
+      "loss": 0.7251,
       "step": 22149
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.918562317106565e-06,
-      "loss": 0.7912,
+      "loss": 0.8802,
       "step": 22150
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.917868629990558e-06,
-      "loss": 0.8313,
+      "loss": 0.9017,
       "step": 22151
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.917175055002657e-06,
-      "loss": 0.9201,
+      "loss": 0.8906,
       "step": 22152
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9164815921524836e-06,
-      "loss": 0.8551,
+      "loss": 0.9199,
       "step": 22153
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9157882414496597e-06,
-      "loss": 0.7204,
+      "loss": 0.8291,
       "step": 22154
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.915095002903804e-06,
-      "loss": 0.7614,
+      "loss": 0.8088,
       "step": 22155
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9144018765245322e-06,
-      "loss": 0.6964,
+      "loss": 0.8286,
       "step": 22156
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.913708862321464e-06,
-      "loss": 0.6117,
+      "loss": 0.7316,
       "step": 22157
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.913015960304212e-06,
-      "loss": 0.6987,
+      "loss": 0.7712,
       "step": 22158
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9123231704823906e-06,
-      "loss": 0.8091,
+      "loss": 0.8237,
       "step": 22159
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9116304928656072e-06,
-      "loss": 0.7241,
+      "loss": 0.7712,
       "step": 22160
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9109379274634797e-06,
-      "loss": 0.693,
+      "loss": 0.8343,
       "step": 22161
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9102454742856067e-06,
-      "loss": 0.8246,
+      "loss": 0.9111,
       "step": 22162
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.909553133341603e-06,
-      "loss": 0.7478,
+      "loss": 0.8329,
       "step": 22163
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.90886090464107e-06,
-      "loss": 0.8725,
+      "loss": 0.9039,
       "step": 22164
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9081687881936127e-06,
-      "loss": 0.8124,
+      "loss": 0.8661,
       "step": 22165
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9074767840088295e-06,
-      "loss": 0.6114,
+      "loss": 0.7348,
       "step": 22166
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9067848920963294e-06,
-      "loss": 0.8027,
+      "loss": 0.8727,
       "step": 22167
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9060931124657024e-06,
-      "loss": 0.6686,
+      "loss": 0.7568,
       "step": 22168
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9054014451265502e-06,
-      "loss": 0.6789,
+      "loss": 0.7281,
       "step": 22169
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9047098900884697e-06,
-      "loss": 0.8221,
+      "loss": 0.8509,
       "step": 22170
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9040184473610522e-06,
-      "loss": 0.6705,
+      "loss": 0.8003,
       "step": 22171
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9033271169538935e-06,
-      "loss": 0.7262,
+      "loss": 0.8355,
       "step": 22172
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9026358988765802e-06,
-      "loss": 0.7662,
+      "loss": 0.7742,
       "step": 22173
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9019447931387102e-06,
-      "loss": 0.669,
+      "loss": 0.7632,
       "step": 22174
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9012537997498615e-06,
-      "loss": 0.7519,
+      "loss": 0.7927,
       "step": 22175
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.9005629187196284e-06,
-      "loss": 0.7007,
+      "loss": 0.7791,
       "step": 22176
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8998721500575934e-06,
-      "loss": 0.7248,
+      "loss": 0.848,
       "step": 22177
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8991814937733399e-06,
-      "loss": 0.7228,
+      "loss": 0.7828,
       "step": 22178
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.898490949876447e-06,
-      "loss": 0.6808,
+      "loss": 0.8083,
       "step": 22179
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8978005183765024e-06,
-      "loss": 0.6778,
+      "loss": 0.8405,
       "step": 22180
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8971101992830742e-06,
-      "loss": 0.8572,
+      "loss": 0.8423,
       "step": 22181
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8964199926057491e-06,
-      "loss": 0.7753,
+      "loss": 0.8785,
       "step": 22182
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8957298983540995e-06,
-      "loss": 0.764,
+      "loss": 0.8348,
       "step": 22183
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8950399165376976e-06,
-      "loss": 0.6841,
+      "loss": 0.689,
       "step": 22184
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8943500471661158e-06,
-      "loss": 0.7457,
+      "loss": 0.7871,
       "step": 22185
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.893660290248932e-06,
-      "loss": 0.6611,
+      "loss": 0.701,
       "step": 22186
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8929706457957053e-06,
-      "loss": 0.5454,
+      "loss": 0.7065,
       "step": 22187
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8922811138160102e-06,
-      "loss": 0.6645,
+      "loss": 0.6995,
       "step": 22188
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8915916943194123e-06,
-      "loss": 0.709,
+      "loss": 0.7446,
       "step": 22189
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.890902387315474e-06,
-      "loss": 0.6151,
+      "loss": 0.7783,
       "step": 22190
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8902131928137612e-06,
-      "loss": 0.6249,
+      "loss": 0.7388,
       "step": 22191
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8895241108238305e-06,
-      "loss": 0.568,
+      "loss": 0.7317,
       "step": 22192
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8888351413552519e-06,
-      "loss": 0.651,
+      "loss": 0.7518,
       "step": 22193
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8881462844175713e-06,
-      "loss": 0.7425,
+      "loss": 0.8539,
       "step": 22194
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8874575400203543e-06,
-      "loss": 0.7704,
+      "loss": 0.8535,
       "step": 22195
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8867689081731545e-06,
-      "loss": 0.6797,
+      "loss": 0.761,
       "step": 22196
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8860803888855239e-06,
-      "loss": 0.5889,
+      "loss": 0.7527,
       "step": 22197
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.885391982167013e-06,
-      "loss": 0.7345,
+      "loss": 0.7797,
       "step": 22198
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8847036880271819e-06,
-      "loss": 0.6571,
+      "loss": 0.7877,
       "step": 22199
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8840155064755673e-06,
-      "loss": 0.8447,
+      "loss": 0.8231,
       "step": 22200
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8833274375217248e-06,
-      "loss": 0.8326,
+      "loss": 0.8209,
       "step": 22201
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8826394811751969e-06,
-      "loss": 0.7176,
+      "loss": 0.749,
       "step": 22202
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8819516374455304e-06,
-      "loss": 0.7954,
+      "loss": 0.8692,
       "step": 22203
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8812639063422644e-06,
-      "loss": 0.6684,
+      "loss": 0.7519,
       "step": 22204
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8805762878749455e-06,
-      "loss": 0.7257,
+      "loss": 0.7857,
       "step": 22205
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8798887820531098e-06,
-      "loss": 0.8127,
+      "loss": 0.8855,
       "step": 22206
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.879201388886297e-06,
-      "loss": 0.6734,
+      "loss": 0.8124,
       "step": 22207
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.878514108384043e-06,
-      "loss": 0.7844,
+      "loss": 0.8495,
       "step": 22208
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8778269405558825e-06,
-      "loss": 0.7323,
+      "loss": 0.8339,
       "step": 22209
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8771398854113498e-06,
-      "loss": 0.8212,
+      "loss": 0.8285,
       "step": 22210
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.876452942959973e-06,
-      "loss": 0.7163,
+      "loss": 0.8333,
       "step": 22211
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8757661132112925e-06,
-      "loss": 0.7005,
+      "loss": 0.7885,
       "step": 22212
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8750793961748236e-06,
-      "loss": 0.7242,
+      "loss": 0.839,
       "step": 22213
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8743927918601035e-06,
-      "loss": 0.6456,
+      "loss": 0.7012,
       "step": 22214
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8737063002766553e-06,
-      "loss": 0.7414,
+      "loss": 0.8233,
       "step": 22215
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8730199214340017e-06,
-      "loss": 0.7618,
+      "loss": 0.7443,
       "step": 22216
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8723336553416637e-06,
-      "loss": 0.7488,
+      "loss": 0.842,
       "step": 22217
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8716475020091674e-06,
-      "loss": 0.6448,
+      "loss": 0.746,
       "step": 22218
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.870961461446028e-06,
-      "loss": 0.7996,
+      "loss": 0.907,
       "step": 22219
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.870275533661766e-06,
-      "loss": 0.5242,
+      "loss": 0.679,
       "step": 22220
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8695897186658963e-06,
-      "loss": 0.78,
+      "loss": 0.8546,
       "step": 22221
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.868904016467934e-06,
-      "loss": 0.8343,
+      "loss": 0.7973,
       "step": 22222
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8682184270773885e-06,
-      "loss": 0.7445,
+      "loss": 0.768,
       "step": 22223
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8675329505037788e-06,
-      "loss": 0.8563,
+      "loss": 0.9104,
       "step": 22224
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8668475867566093e-06,
-      "loss": 0.7339,
+      "loss": 0.793,
       "step": 22225
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8661623358453906e-06,
-      "loss": 0.6682,
+      "loss": 0.7446,
       "step": 22226
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.865477197779627e-06,
-      "loss": 0.6588,
+      "loss": 0.7542,
       "step": 22227
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8647921725688312e-06,
-      "loss": 0.7296,
+      "loss": 0.8189,
       "step": 22228
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8641072602224986e-06,
-      "loss": 0.7623,
+      "loss": 0.7713,
       "step": 22229
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8634224607501317e-06,
-      "loss": 0.7675,
+      "loss": 0.8939,
       "step": 22230
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8627377741612395e-06,
-      "loss": 0.7008,
+      "loss": 0.7737,
       "step": 22231
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8620532004653103e-06,
-      "loss": 0.7478,
+      "loss": 0.8158,
       "step": 22232
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8613687396718482e-06,
-      "loss": 0.579,
+      "loss": 0.7091,
       "step": 22233
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8606843917903493e-06,
-      "loss": 0.5576,
+      "loss": 0.7654,
       "step": 22234
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.860000156830305e-06,
-      "loss": 0.7594,
+      "loss": 0.8478,
       "step": 22235
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8593160348012075e-06,
-      "loss": 0.7069,
+      "loss": 0.839,
       "step": 22236
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8586320257125524e-06,
-      "loss": 0.6503,
+      "loss": 0.7296,
       "step": 22237
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8579481295738267e-06,
-      "loss": 0.6942,
+      "loss": 0.8028,
       "step": 22238
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8572643463945183e-06,
-      "loss": 0.7363,
+      "loss": 0.7786,
       "step": 22239
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.856580676184112e-06,
-      "loss": 0.594,
+      "loss": 0.7205,
       "step": 22240
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8558971189520991e-06,
-      "loss": 0.7182,
+      "loss": 0.76,
       "step": 22241
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.855213674707954e-06,
-      "loss": 0.7604,
+      "loss": 0.8571,
       "step": 22242
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.854530343461166e-06,
-      "loss": 0.6057,
+      "loss": 0.7014,
       "step": 22243
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8538471252212131e-06,
-      "loss": 0.584,
+      "loss": 0.6808,
       "step": 22244
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.853164019997573e-06,
-      "loss": 0.7405,
+      "loss": 0.8284,
       "step": 22245
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8524810277997208e-06,
-      "loss": 0.6805,
+      "loss": 0.765,
       "step": 22246
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8517981486371395e-06,
-      "loss": 0.7294,
+      "loss": 0.7853,
       "step": 22247
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8511153825192952e-06,
-      "loss": 0.6263,
+      "loss": 0.7755,
       "step": 22248
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8504327294556613e-06,
-      "loss": 0.6977,
+      "loss": 0.8269,
       "step": 22249
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8497501894557123e-06,
-      "loss": 0.6547,
+      "loss": 0.7258,
       "step": 22250
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8490677625289165e-06,
-      "loss": 0.7583,
+      "loss": 0.846,
       "step": 22251
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8483854486847407e-06,
-      "loss": 0.6028,
+      "loss": 0.7221,
       "step": 22252
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8477032479326508e-06,
-      "loss": 0.8021,
+      "loss": 0.8254,
       "step": 22253
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.847021160282112e-06,
-      "loss": 0.7287,
+      "loss": 0.7965,
       "step": 22254
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.846339185742585e-06,
-      "loss": 0.7083,
+      "loss": 0.8289,
       "step": 22255
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8456573243235354e-06,
-      "loss": 0.6933,
+      "loss": 0.794,
       "step": 22256
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.84497557603442e-06,
-      "loss": 0.7316,
+      "loss": 0.8598,
       "step": 22257
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8442939408846993e-06,
-      "loss": 0.7057,
+      "loss": 0.722,
       "step": 22258
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8436124188838257e-06,
-      "loss": 0.5788,
+      "loss": 0.7467,
       "step": 22259
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8429310100412624e-06,
-      "loss": 0.7561,
+      "loss": 0.7354,
       "step": 22260
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.842249714366453e-06,
-      "loss": 0.7206,
+      "loss": 0.7597,
       "step": 22261
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.841568531868857e-06,
-      "loss": 0.7804,
+      "loss": 0.8667,
       "step": 22262
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.840887462557922e-06,
-      "loss": 0.6715,
+      "loss": 0.7586,
       "step": 22263
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8402065064430974e-06,
-      "loss": 0.657,
+      "loss": 0.7783,
       "step": 22264
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8395256635338287e-06,
-      "loss": 0.579,
+      "loss": 0.702,
       "step": 22265
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8388449338395675e-06,
-      "loss": 0.6697,
+      "loss": 0.7274,
       "step": 22266
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8381643173697505e-06,
-      "loss": 0.7651,
+      "loss": 0.7904,
       "step": 22267
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8374838141338214e-06,
-      "loss": 0.778,
+      "loss": 0.8786,
       "step": 22268
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8368034241412259e-06,
-      "loss": 0.7766,
+      "loss": 0.8228,
       "step": 22269
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8361231474014007e-06,
-      "loss": 0.6281,
+      "loss": 0.7327,
       "step": 22270
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.835442983923784e-06,
-      "loss": 0.7801,
+      "loss": 0.8586,
       "step": 22271
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8347629337178097e-06,
-      "loss": 0.751,
+      "loss": 0.7781,
       "step": 22272
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8340829967929197e-06,
-      "loss": 0.7783,
+      "loss": 0.8213,
       "step": 22273
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8334031731585367e-06,
-      "loss": 0.6506,
+      "loss": 0.7656,
       "step": 22274
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8327234628241008e-06,
-      "loss": 0.6879,
+      "loss": 0.7791,
       "step": 22275
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8320438657990381e-06,
-      "loss": 0.9266,
+      "loss": 0.9149,
       "step": 22276
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8313643820927795e-06,
-      "loss": 0.7028,
+      "loss": 0.8086,
       "step": 22277
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8306850117147468e-06,
-      "loss": 0.7599,
+      "loss": 0.7667,
       "step": 22278
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8300057546743733e-06,
-      "loss": 0.7641,
+      "loss": 0.8604,
       "step": 22279
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8293266109810738e-06,
-      "loss": 0.701,
+      "loss": 0.8038,
       "step": 22280
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8286475806442772e-06,
-      "loss": 0.6982,
+      "loss": 0.7158,
       "step": 22281
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8279686636734029e-06,
-      "loss": 0.7504,
+      "loss": 0.8036,
       "step": 22282
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8272898600778677e-06,
-      "loss": 0.7926,
+      "loss": 0.8664,
       "step": 22283
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8266111698670875e-06,
-      "loss": 0.7455,
+      "loss": 0.8417,
       "step": 22284
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8259325930504835e-06,
-      "loss": 0.6471,
+      "loss": 0.7924,
       "step": 22285
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8252541296374705e-06,
-      "loss": 0.6529,
+      "loss": 0.8029,
       "step": 22286
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.824575779637452e-06,
-      "loss": 0.6967,
+      "loss": 0.8013,
       "step": 22287
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8238975430598483e-06,
-      "loss": 0.7353,
+      "loss": 0.7747,
       "step": 22288
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8232194199140652e-06,
-      "loss": 0.8012,
+      "loss": 0.9214,
       "step": 22289
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.822541410209512e-06,
-      "loss": 0.729,
+      "loss": 0.6882,
       "step": 22290
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8218635139555918e-06,
-      "loss": 0.7089,
+      "loss": 0.7493,
       "step": 22291
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8211857311617164e-06,
-      "loss": 0.6572,
+      "loss": 0.8142,
       "step": 22292
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8205080618372805e-06,
-      "loss": 0.7236,
+      "loss": 0.7766,
       "step": 22293
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8198305059916932e-06,
-      "loss": 0.6789,
+      "loss": 0.783,
       "step": 22294
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8191530636343501e-06,
-      "loss": 0.7191,
+      "loss": 0.8307,
       "step": 22295
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8184757347746518e-06,
-      "loss": 0.7643,
+      "loss": 0.8069,
       "step": 22296
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8177985194219916e-06,
-      "loss": 0.5533,
+      "loss": 0.708,
       "step": 22297
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8171214175857743e-06,
-      "loss": 0.7503,
+      "loss": 0.7376,
       "step": 22298
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8164444292753814e-06,
-      "loss": 0.6699,
+      "loss": 0.8366,
       "step": 22299
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8157675545002129e-06,
-      "loss": 0.6026,
+      "loss": 0.7482,
       "step": 22300
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8150907932696582e-06,
-      "loss": 0.737,
+      "loss": 0.8103,
       "step": 22301
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8144141455931063e-06,
-      "loss": 0.5514,
+      "loss": 0.6888,
       "step": 22302
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.813737611479941e-06,
-      "loss": 0.8202,
+      "loss": 0.9009,
       "step": 22303
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8130611909395547e-06,
-      "loss": 0.8382,
+      "loss": 0.931,
       "step": 22304
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8123848839813318e-06,
-      "loss": 0.7948,
+      "loss": 0.872,
       "step": 22305
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8117086906146454e-06,
-      "loss": 0.5757,
+      "loss": 0.7378,
       "step": 22306
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8110326108488863e-06,
-      "loss": 0.7024,
+      "loss": 0.7701,
       "step": 22307
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8103566446934317e-06,
-      "loss": 0.6636,
+      "loss": 0.745,
       "step": 22308
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8096807921576587e-06,
-      "loss": 0.7447,
+      "loss": 0.8199,
       "step": 22309
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8090050532509417e-06,
-      "loss": 0.8478,
+      "loss": 0.8934,
       "step": 22310
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8083294279826635e-06,
-      "loss": 0.7105,
+      "loss": 0.7505,
       "step": 22311
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8076539163621864e-06,
-      "loss": 0.6303,
+      "loss": 0.7864,
       "step": 22312
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8069785183988896e-06,
-      "loss": 0.7461,
+      "loss": 0.7914,
       "step": 22313
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8063032341021425e-06,
-      "loss": 0.6757,
+      "loss": 0.7915,
       "step": 22314
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8056280634813129e-06,
-      "loss": 0.6608,
+      "loss": 0.7075,
       "step": 22315
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8049530065457644e-06,
-      "loss": 0.704,
+      "loss": 0.8174,
       "step": 22316
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8042780633048685e-06,
-      "loss": 0.5965,
+      "loss": 0.6603,
       "step": 22317
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8036032337679865e-06,
-      "loss": 0.6048,
+      "loss": 0.7515,
       "step": 22318
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.80292851794448e-06,
-      "loss": 0.7412,
+      "loss": 0.7521,
       "step": 22319
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8022539158437101e-06,
-      "loss": 0.6567,
+      "loss": 0.7826,
       "step": 22320
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8015794274750365e-06,
-      "loss": 0.7222,
+      "loss": 0.7632,
       "step": 22321
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8009050528478144e-06,
-      "loss": 0.7297,
+      "loss": 0.7777,
       "step": 22322
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.8002307919714036e-06,
-      "loss": 0.6826,
+      "loss": 0.8332,
       "step": 22323
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7995566448551583e-06,
-      "loss": 0.7133,
+      "loss": 0.8448,
       "step": 22324
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7988826115084258e-06,
-      "loss": 0.7402,
+      "loss": 0.779,
       "step": 22325
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7982086919405628e-06,
-      "loss": 0.7145,
+      "loss": 0.8131,
       "step": 22326
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7975348861609177e-06,
-      "loss": 0.8162,
+      "loss": 0.8168,
       "step": 22327
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7968611941788382e-06,
-      "loss": 0.5333,
+      "loss": 0.6248,
       "step": 22328
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7961876160036685e-06,
-      "loss": 0.7407,
+      "loss": 0.8391,
       "step": 22329
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7955141516447593e-06,
-      "loss": 0.8719,
+      "loss": 0.851,
       "step": 22330
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7948408011114493e-06,
-      "loss": 0.7433,
+      "loss": 0.7931,
       "step": 22331
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7941675644130819e-06,
-      "loss": 0.69,
+      "loss": 0.8286,
       "step": 22332
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7934944415589972e-06,
-      "loss": 0.7238,
+      "loss": 0.7653,
       "step": 22333
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7928214325585336e-06,
-      "loss": 0.6679,
+      "loss": 0.7646,
       "step": 22334
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7921485374210257e-06,
-      "loss": 0.6944,
+      "loss": 0.8557,
       "step": 22335
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7914757561558139e-06,
-      "loss": 0.6527,
+      "loss": 0.6924,
       "step": 22336
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7908030887722305e-06,
-      "loss": 0.7293,
+      "loss": 0.7861,
       "step": 22337
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.790130535279605e-06,
-      "loss": 0.7912,
+      "loss": 0.7139,
       "step": 22338
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7894580956872687e-06,
-      "loss": 0.5903,
+      "loss": 0.6845,
       "step": 22339
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7887857700045574e-06,
-      "loss": 0.8046,
+      "loss": 0.8277,
       "step": 22340
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7881135582407883e-06,
-      "loss": 0.7131,
+      "loss": 0.7682,
       "step": 22341
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7874414604052937e-06,
-      "loss": 0.6678,
+      "loss": 0.7888,
       "step": 22342
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7867694765074007e-06,
-      "loss": 0.6859,
+      "loss": 0.8492,
       "step": 22343
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7860976065564229e-06,
-      "loss": 0.6507,
+      "loss": 0.7582,
       "step": 22344
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7854258505616884e-06,
-      "loss": 0.7412,
+      "loss": 0.7954,
       "step": 22345
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7847542085325154e-06,
-      "loss": 0.6307,
+      "loss": 0.7061,
       "step": 22346
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.784082680478223e-06,
-      "loss": 0.706,
+      "loss": 0.7857,
       "step": 22347
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7834112664081226e-06,
-      "loss": 0.6534,
+      "loss": 0.788,
       "step": 22348
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.782739966331537e-06,
-      "loss": 0.7558,
+      "loss": 0.8261,
       "step": 22349
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.782068780257774e-06,
-      "loss": 0.7398,
+      "loss": 0.8591,
       "step": 22350
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7813977081961486e-06,
-      "loss": 0.7521,
+      "loss": 0.7951,
       "step": 22351
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7807267501559688e-06,
-      "loss": 0.8068,
+      "loss": 0.8308,
       "step": 22352
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7800559061465427e-06,
-      "loss": 0.7222,
+      "loss": 0.83,
       "step": 22353
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7793851761771773e-06,
-      "loss": 0.6774,
+      "loss": 0.7243,
       "step": 22354
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.778714560257181e-06,
-      "loss": 0.709,
+      "loss": 0.8045,
       "step": 22355
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7780440583958546e-06,
-      "loss": 0.665,
+      "loss": 0.7961,
       "step": 22356
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7773736706025025e-06,
-      "loss": 0.6925,
+      "loss": 0.7666,
       "step": 22357
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7767033968864222e-06,
-      "loss": 0.8187,
+      "loss": 0.9214,
       "step": 22358
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7760332372569188e-06,
-      "loss": 0.8151,
+      "loss": 0.7829,
       "step": 22359
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7753631917232827e-06,
-      "loss": 0.7658,
+      "loss": 0.8,
       "step": 22360
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.774693260294814e-06,
-      "loss": 0.8637,
+      "loss": 0.9182,
       "step": 22361
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7740234429808068e-06,
-      "loss": 0.8173,
+      "loss": 0.8907,
       "step": 22362
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.773353739790553e-06,
-      "loss": 0.6376,
+      "loss": 0.7688,
       "step": 22363
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7726841507333436e-06,
-      "loss": 0.6337,
+      "loss": 0.7662,
       "step": 22364
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7720146758184698e-06,
-      "loss": 0.7055,
+      "loss": 0.7913,
       "step": 22365
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7713453150552173e-06,
-      "loss": 0.8586,
+      "loss": 0.8436,
       "step": 22366
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7706760684528713e-06,
-      "loss": 0.6629,
+      "loss": 0.7855,
       "step": 22367
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7700069360207218e-06,
-      "loss": 0.8246,
+      "loss": 0.8671,
       "step": 22368
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7693379177680492e-06,
-      "loss": 0.6547,
+      "loss": 0.8054,
       "step": 22369
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.768669013704134e-06,
-      "loss": 0.7356,
+      "loss": 0.8287,
       "step": 22370
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7680002238382565e-06,
-      "loss": 0.7507,
+      "loss": 0.8921,
       "step": 22371
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7673315481797005e-06,
-      "loss": 0.6577,
+      "loss": 0.796,
       "step": 22372
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7666629867377339e-06,
-      "loss": 0.776,
+      "loss": 0.8129,
       "step": 22373
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7659945395216394e-06,
-      "loss": 0.634,
+      "loss": 0.7105,
       "step": 22374
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7653262065406885e-06,
-      "loss": 0.7669,
+      "loss": 0.8063,
       "step": 22375
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7646579878041525e-06,
-      "loss": 0.8324,
+      "loss": 0.8598,
       "step": 22376
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7639898833213e-06,
-      "loss": 0.7421,
+      "loss": 0.8649,
       "step": 22377
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7633218931014072e-06,
-      "loss": 0.6578,
+      "loss": 0.7572,
       "step": 22378
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.762654017153732e-06,
-      "loss": 0.7578,
+      "loss": 0.7954,
       "step": 22379
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7619862554875478e-06,
-      "loss": 0.8476,
+      "loss": 0.8998,
       "step": 22380
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.761318608112116e-06,
-      "loss": 0.7783,
+      "loss": 0.8748,
       "step": 22381
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.760651075036699e-06,
-      "loss": 0.7928,
+      "loss": 0.8411,
       "step": 22382
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7599836562705585e-06,
-      "loss": 0.7106,
+      "loss": 0.848,
       "step": 22383
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7593163518229518e-06,
-      "loss": 0.7602,
+      "loss": 0.8204,
       "step": 22384
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7586491617031431e-06,
-      "loss": 0.7043,
+      "loss": 0.736,
       "step": 22385
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7579820859203789e-06,
-      "loss": 0.5279,
+      "loss": 0.6815,
       "step": 22386
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7573151244839225e-06,
-      "loss": 0.8023,
+      "loss": 0.8551,
       "step": 22387
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7566482774030235e-06,
-      "loss": 0.8443,
+      "loss": 0.8653,
       "step": 22388
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7559815446869344e-06,
-      "loss": 0.6606,
+      "loss": 0.7372,
       "step": 22389
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7553149263449022e-06,
-      "loss": 0.844,
+      "loss": 0.8609,
       "step": 22390
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7546484223861826e-06,
-      "loss": 0.7674,
+      "loss": 0.8444,
       "step": 22391
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7539820328200119e-06,
-      "loss": 0.6785,
+      "loss": 0.7982,
       "step": 22392
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7533157576556436e-06,
-      "loss": 0.595,
+      "loss": 0.6632,
       "step": 22393
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7526495969023193e-06,
-      "loss": 0.6679,
+      "loss": 0.7223,
       "step": 22394
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7519835505692796e-06,
-      "loss": 0.6918,
+      "loss": 0.7893,
       "step": 22395
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7513176186657633e-06,
-      "loss": 0.7155,
+      "loss": 0.7911,
       "step": 22396
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7506518012010176e-06,
-      "loss": 0.7055,
+      "loss": 0.8063,
       "step": 22397
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7499860981842675e-06,
-      "loss": 0.7249,
+      "loss": 0.8443,
       "step": 22398
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7493205096247589e-06,
-      "loss": 0.8338,
+      "loss": 0.8996,
       "step": 22399
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.748655035531721e-06,
-      "loss": 0.6625,
+      "loss": 0.7035,
       "step": 22400
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7479896759143877e-06,
-      "loss": 0.7378,
+      "loss": 0.8744,
       "step": 22401
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7473244307819902e-06,
-      "loss": 0.7897,
+      "loss": 0.8054,
       "step": 22402
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7466593001437537e-06,
-      "loss": 0.7882,
+      "loss": 0.8133,
       "step": 22403
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.745994284008915e-06,
-      "loss": 0.8478,
+      "loss": 0.8367,
       "step": 22404
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7453293823866901e-06,
-      "loss": 0.7602,
+      "loss": 0.7577,
       "step": 22405
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7446645952863094e-06,
-      "loss": 0.6547,
+      "loss": 0.6535,
       "step": 22406
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7439999227169957e-06,
-      "loss": 0.6888,
+      "loss": 0.7,
       "step": 22407
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7433353646879691e-06,
-      "loss": 0.7405,
+      "loss": 0.7952,
       "step": 22408
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.742670921208447e-06,
-      "loss": 0.8354,
+      "loss": 0.8321,
       "step": 22409
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.7420065922876562e-06,
-      "loss": 0.7778,
+      "loss": 0.879,
       "step": 22410
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7413423779348015e-06,
-      "loss": 0.7665,
+      "loss": 0.8023,
       "step": 22411
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.740678278159106e-06,
-      "loss": 0.7078,
+      "loss": 0.731,
       "step": 22412
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7400142929697816e-06,
-      "loss": 0.7441,
+      "loss": 0.7943,
       "step": 22413
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7393504223760394e-06,
-      "loss": 0.758,
+      "loss": 0.8586,
       "step": 22414
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7386866663870872e-06,
-      "loss": 0.878,
+      "loss": 0.8146,
       "step": 22415
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7380230250121377e-06,
-      "loss": 0.678,
+      "loss": 0.749,
       "step": 22416
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7373594982603982e-06,
-      "loss": 0.7663,
+      "loss": 0.8193,
       "step": 22417
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7366960861410708e-06,
-      "loss": 0.5952,
+      "loss": 0.8002,
       "step": 22418
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.736032788663362e-06,
-      "loss": 0.7511,
+      "loss": 0.7975,
       "step": 22419
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7353696058364733e-06,
-      "loss": 0.658,
+      "loss": 0.8335,
       "step": 22420
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.734706537669606e-06,
-      "loss": 0.7746,
+      "loss": 0.7816,
       "step": 22421
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7340435841719548e-06,
-      "loss": 0.723,
+      "loss": 0.8385,
       "step": 22422
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7333807453527273e-06,
-      "loss": 0.6889,
+      "loss": 0.7786,
       "step": 22423
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.732718021221107e-06,
-      "loss": 0.7393,
+      "loss": 0.8054,
       "step": 22424
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7320554117862976e-06,
-      "loss": 0.7033,
+      "loss": 0.7787,
       "step": 22425
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7313929170574884e-06,
-      "loss": 0.7436,
+      "loss": 0.7963,
       "step": 22426
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.730730537043871e-06,
-      "loss": 0.6974,
+      "loss": 0.7167,
       "step": 22427
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7300682717546324e-06,
-      "loss": 0.7905,
+      "loss": 0.8612,
       "step": 22428
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7294061211989655e-06,
-      "loss": 0.7006,
+      "loss": 0.8372,
       "step": 22429
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7287440853860537e-06,
-      "loss": 0.5957,
+      "loss": 0.8187,
       "step": 22430
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7280821643250823e-06,
-      "loss": 0.7668,
+      "loss": 0.8206,
       "step": 22431
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7274203580252358e-06,
-      "loss": 0.6256,
+      "loss": 0.6623,
       "step": 22432
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7267586664956938e-06,
-      "loss": 0.8112,
+      "loss": 0.8271,
       "step": 22433
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7260970897456342e-06,
-      "loss": 0.7348,
+      "loss": 0.7965,
       "step": 22434
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.725435627784241e-06,
-      "loss": 0.6198,
+      "loss": 0.7682,
       "step": 22435
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.724774280620689e-06,
-      "loss": 0.743,
+      "loss": 0.837,
       "step": 22436
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7241130482641521e-06,
-      "loss": 0.6662,
+      "loss": 0.7807,
       "step": 22437
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.723451930723803e-06,
-      "loss": 0.6659,
+      "loss": 0.7597,
       "step": 22438
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7227909280088195e-06,
-      "loss": 0.7603,
+      "loss": 0.8583,
       "step": 22439
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7221300401283658e-06,
-      "loss": 0.7013,
+      "loss": 0.7844,
       "step": 22440
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.721469267091611e-06,
-      "loss": 0.7824,
+      "loss": 0.7678,
       "step": 22441
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7208086089077292e-06,
-      "loss": 0.6796,
+      "loss": 0.8199,
       "step": 22442
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.720148065585877e-06,
-      "loss": 0.7776,
+      "loss": 0.8705,
       "step": 22443
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7194876371352243e-06,
-      "loss": 0.7737,
+      "loss": 0.7935,
       "step": 22444
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7188273235649323e-06,
-      "loss": 0.663,
+      "loss": 0.8048,
       "step": 22445
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7181671248841625e-06,
-      "loss": 0.7717,
+      "loss": 0.8702,
       "step": 22446
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7175070411020701e-06,
-      "loss": 0.6816,
+      "loss": 0.8238,
       "step": 22447
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7168470722278196e-06,
-      "loss": 0.7235,
+      "loss": 0.8493,
       "step": 22448
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.716187218270563e-06,
-      "loss": 0.6564,
+      "loss": 0.8,
       "step": 22449
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.715527479239455e-06,
-      "loss": 0.7023,
+      "loss": 0.8326,
       "step": 22450
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.714867855143647e-06,
-      "loss": 0.6377,
+      "loss": 0.7192,
       "step": 22451
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7142083459922986e-06,
-      "loss": 0.65,
+      "loss": 0.827,
       "step": 22452
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.713548951794547e-06,
-      "loss": 0.7971,
+      "loss": 0.8029,
       "step": 22453
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7128896725595501e-06,
-      "loss": 0.7614,
+      "loss": 0.8277,
       "step": 22454
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7122305082964507e-06,
-      "loss": 0.7279,
+      "loss": 0.779,
       "step": 22455
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.711571459014394e-06,
-      "loss": 0.7209,
+      "loss": 0.7742,
       "step": 22456
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7109125247225211e-06,
-      "loss": 0.6126,
+      "loss": 0.7228,
       "step": 22457
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7102537054299817e-06,
-      "loss": 0.6792,
+      "loss": 0.8143,
       "step": 22458
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7095950011459083e-06,
-      "loss": 0.648,
+      "loss": 0.8232,
       "step": 22459
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7089364118794382e-06,
-      "loss": 0.7822,
+      "loss": 0.8547,
       "step": 22460
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7082779376397152e-06,
-      "loss": 0.7073,
+      "loss": 0.8343,
       "step": 22461
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7076195784358707e-06,
-      "loss": 0.787,
+      "loss": 0.9087,
       "step": 22462
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7069613342770398e-06,
-      "loss": 0.6786,
+      "loss": 0.7971,
       "step": 22463
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7063032051723538e-06,
-      "loss": 0.7836,
+      "loss": 0.7708,
       "step": 22464
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7056451911309434e-06,
-      "loss": 0.6346,
+      "loss": 0.7673,
       "step": 22465
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.704987292161936e-06,
-      "loss": 0.7288,
+      "loss": 0.8408,
       "step": 22466
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7043295082744626e-06,
-      "loss": 0.6779,
+      "loss": 0.7586,
       "step": 22467
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7036718394776486e-06,
-      "loss": 0.7447,
+      "loss": 0.8132,
       "step": 22468
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7030142857806153e-06,
-      "loss": 0.7991,
+      "loss": 0.8357,
       "step": 22469
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7023568471924855e-06,
-      "loss": 0.6233,
+      "loss": 0.7647,
       "step": 22470
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7016995237223855e-06,
-      "loss": 0.7162,
+      "loss": 0.8538,
       "step": 22471
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7010423153794265e-06,
-      "loss": 0.6621,
+      "loss": 0.7433,
       "step": 22472
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.7003852221727335e-06,
-      "loss": 0.8245,
+      "loss": 0.8388,
       "step": 22473
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6997282441114193e-06,
-      "loss": 0.7043,
+      "loss": 0.7933,
       "step": 22474
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6990713812046e-06,
-      "loss": 0.6685,
+      "loss": 0.7747,
       "step": 22475
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.698414633461385e-06,
-      "loss": 0.6768,
+      "loss": 0.8032,
       "step": 22476
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6977580008908933e-06,
-      "loss": 0.6631,
+      "loss": 0.7515,
       "step": 22477
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.697101483502227e-06,
-      "loss": 0.8342,
+      "loss": 0.9501,
       "step": 22478
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6964450813044953e-06,
-      "loss": 0.6612,
+      "loss": 0.7575,
       "step": 22479
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6957887943068097e-06,
-      "loss": 0.6327,
+      "loss": 0.8007,
       "step": 22480
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6951326225182728e-06,
-      "loss": 0.6804,
+      "loss": 0.7367,
       "step": 22481
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6944765659479866e-06,
-      "loss": 0.7317,
+      "loss": 0.8048,
       "step": 22482
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6938206246050526e-06,
-      "loss": 0.7168,
+      "loss": 0.7737,
       "step": 22483
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6931647984985767e-06,
-      "loss": 0.6212,
+      "loss": 0.7507,
       "step": 22484
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6925090876376482e-06,
-      "loss": 0.7923,
+      "loss": 0.8668,
       "step": 22485
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6918534920313723e-06,
-      "loss": 0.7498,
+      "loss": 0.8403,
       "step": 22486
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6911980116888416e-06,
-      "loss": 0.7284,
+      "loss": 0.8383,
       "step": 22487
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6905426466191498e-06,
-      "loss": 0.7023,
+      "loss": 0.7342,
       "step": 22488
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6898873968313867e-06,
-      "loss": 0.8324,
+      "loss": 0.826,
       "step": 22489
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6892322623346502e-06,
-      "loss": 0.8293,
+      "loss": 0.8272,
       "step": 22490
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.68857724313802e-06,
-      "loss": 0.6963,
+      "loss": 0.8423,
       "step": 22491
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6879223392505905e-06,
-      "loss": 0.7141,
+      "loss": 0.7924,
       "step": 22492
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6872675506814452e-06,
-      "loss": 0.7905,
+      "loss": 0.8802,
       "step": 22493
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6866128774396684e-06,
-      "loss": 0.6434,
+      "loss": 0.7391,
       "step": 22494
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.685958319534341e-06,
-      "loss": 0.6371,
+      "loss": 0.7575,
       "step": 22495
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6853038769745466e-06,
-      "loss": 0.6645,
+      "loss": 0.8616,
       "step": 22496
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6846495497693683e-06,
-      "loss": 0.597,
+      "loss": 0.7226,
       "step": 22497
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.683995337927873e-06,
-      "loss": 0.7705,
+      "loss": 0.8308,
       "step": 22498
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6833412414591465e-06,
-      "loss": 0.6596,
+      "loss": 0.7897,
       "step": 22499
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.682687260372261e-06,
-      "loss": 0.7332,
+      "loss": 0.7838,
       "step": 22500
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.682033394676289e-06,
-      "loss": 0.6489,
+      "loss": 0.7375,
       "step": 22501
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6813796443802987e-06,
-      "loss": 0.6936,
+      "loss": 0.8085,
       "step": 22502
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6807260094933697e-06,
-      "loss": 0.7979,
+      "loss": 0.8433,
       "step": 22503
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.680072490024557e-06,
-      "loss": 0.5634,
+      "loss": 0.7461,
       "step": 22504
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6794190859829373e-06,
-      "loss": 0.6818,
+      "loss": 0.7654,
       "step": 22505
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6787657973775728e-06,
-      "loss": 0.7028,
+      "loss": 0.7593,
       "step": 22506
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.678112624217527e-06,
-      "loss": 0.7449,
+      "loss": 0.8102,
       "step": 22507
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6774595665118577e-06,
-      "loss": 0.577,
+      "loss": 0.8008,
       "step": 22508
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6768066242696346e-06,
-      "loss": 0.7988,
+      "loss": 0.8621,
       "step": 22509
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6761537974999054e-06,
-      "loss": 0.647,
+      "loss": 0.7686,
       "step": 22510
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.675501086211734e-06,
-      "loss": 0.7386,
+      "loss": 0.7961,
       "step": 22511
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.674848490414175e-06,
-      "loss": 0.6347,
+      "loss": 0.7459,
       "step": 22512
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6741960101162812e-06,
-      "loss": 0.7402,
+      "loss": 0.7795,
       "step": 22513
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6735436453271014e-06,
-      "loss": 0.6618,
+      "loss": 0.771,
       "step": 22514
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6728913960556937e-06,
-      "loss": 0.6737,
+      "loss": 0.8285,
       "step": 22515
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.672239262311105e-06,
-      "loss": 0.7587,
+      "loss": 0.823,
       "step": 22516
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6715872441023761e-06,
-      "loss": 0.6349,
+      "loss": 0.7321,
       "step": 22517
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.670935341438561e-06,
-      "loss": 0.722,
+      "loss": 0.8268,
       "step": 22518
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6702835543287e-06,
-      "loss": 0.7801,
+      "loss": 0.8112,
       "step": 22519
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6696318827818359e-06,
-      "loss": 0.923,
+      "loss": 0.9403,
       "step": 22520
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6689803268070081e-06,
-      "loss": 0.8179,
+      "loss": 0.8992,
       "step": 22521
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6683288864132629e-06,
-      "loss": 0.6778,
+      "loss": 0.7856,
       "step": 22522
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6676775616096297e-06,
-      "loss": 0.7011,
+      "loss": 0.7226,
       "step": 22523
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.667026352405149e-06,
-      "loss": 0.7717,
+      "loss": 0.804,
       "step": 22524
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6663752588088556e-06,
-      "loss": 0.7349,
+      "loss": 0.8395,
       "step": 22525
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6657242808297814e-06,
-      "loss": 0.7457,
+      "loss": 0.8517,
       "step": 22526
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6650734184769557e-06,
-      "loss": 0.6567,
+      "loss": 0.8163,
       "step": 22527
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6644226717594125e-06,
-      "loss": 0.7266,
+      "loss": 0.7966,
       "step": 22528
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6637720406861791e-06,
-      "loss": 0.7656,
+      "loss": 0.7789,
       "step": 22529
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6631215252662803e-06,
-      "loss": 0.8286,
+      "loss": 0.8586,
       "step": 22530
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6624711255087423e-06,
-      "loss": 0.5102,
+      "loss": 0.7042,
       "step": 22531
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6618208414225877e-06,
-      "loss": 0.6601,
+      "loss": 0.7378,
       "step": 22532
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6611706730168353e-06,
-      "loss": 0.6988,
+      "loss": 0.7329,
       "step": 22533
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.660520620300512e-06,
-      "loss": 0.6077,
+      "loss": 0.6834,
       "step": 22534
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.659870683282636e-06,
-      "loss": 0.6151,
+      "loss": 0.7012,
       "step": 22535
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.659220861972215e-06,
-      "loss": 0.6804,
+      "loss": 0.8175,
       "step": 22536
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6585711563782736e-06,
-      "loss": 0.846,
+      "loss": 0.8954,
       "step": 22537
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6579215665098225e-06,
-      "loss": 0.8115,
+      "loss": 0.8468,
       "step": 22538
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6572720923758734e-06,
-      "loss": 0.6304,
+      "loss": 0.7875,
       "step": 22539
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6566227339854357e-06,
-      "loss": 0.7726,
+      "loss": 0.7942,
       "step": 22540
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6559734913475234e-06,
-      "loss": 0.6909,
+      "loss": 0.7508,
       "step": 22541
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.655324364471137e-06,
-      "loss": 0.8476,
+      "loss": 0.8853,
       "step": 22542
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.654675353365287e-06,
-      "loss": 0.768,
+      "loss": 0.8638,
       "step": 22543
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6540264580389754e-06,
-      "loss": 0.7632,
+      "loss": 0.8494,
       "step": 22544
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6533776785012068e-06,
-      "loss": 0.5896,
+      "loss": 0.7247,
       "step": 22545
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6527290147609764e-06,
-      "loss": 0.7117,
+      "loss": 0.7473,
       "step": 22546
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6520804668272917e-06,
-      "loss": 0.672,
+      "loss": 0.7494,
       "step": 22547
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.651432034709146e-06,
-      "loss": 0.7006,
+      "loss": 0.765,
       "step": 22548
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.650783718415535e-06,
-      "loss": 0.7554,
+      "loss": 0.8132,
       "step": 22549
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6501355179554524e-06,
-      "loss": 0.6902,
+      "loss": 0.7999,
       "step": 22550
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6494874333378974e-06,
-      "loss": 0.7034,
+      "loss": 0.7585,
       "step": 22551
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6488394645718508e-06,
-      "loss": 0.6265,
+      "loss": 0.7698,
       "step": 22552
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.648191611666311e-06,
-      "loss": 0.6465,
+      "loss": 0.7533,
       "step": 22553
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6475438746302652e-06,
-      "loss": 0.604,
+      "loss": 0.7085,
       "step": 22554
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6468962534726917e-06,
-      "loss": 0.6321,
+      "loss": 0.6972,
       "step": 22555
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6462487482025847e-06,
-      "loss": 0.705,
+      "loss": 0.8212,
       "step": 22556
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6456013588289232e-06,
-      "loss": 0.6911,
+      "loss": 0.759,
       "step": 22557
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6449540853606905e-06,
-      "loss": 0.6792,
+      "loss": 0.7473,
       "step": 22558
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6443069278068623e-06,
-      "loss": 0.6496,
+      "loss": 0.7609,
       "step": 22559
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.643659886176423e-06,
-      "loss": 0.6845,
+      "loss": 0.7557,
       "step": 22560
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6430129604783474e-06,
-      "loss": 0.7995,
+      "loss": 0.8344,
       "step": 22561
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6423661507216094e-06,
-      "loss": 0.7379,
+      "loss": 0.7866,
       "step": 22562
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6417194569151828e-06,
-      "loss": 0.7879,
+      "loss": 0.7901,
       "step": 22563
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6410728790680396e-06,
-      "loss": 0.6569,
+      "loss": 0.6998,
       "step": 22564
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6404264171891492e-06,
-      "loss": 0.7965,
+      "loss": 0.8193,
       "step": 22565
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6397800712874834e-06,
-      "loss": 0.8748,
+      "loss": 0.8718,
       "step": 22566
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.639133841372007e-06,
-      "loss": 0.8419,
+      "loss": 0.8061,
       "step": 22567
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6384877274516864e-06,
-      "loss": 0.6944,
+      "loss": 0.7477,
       "step": 22568
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.637841729535483e-06,
-      "loss": 0.707,
+      "loss": 0.7868,
       "step": 22569
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6371958476323656e-06,
-      "loss": 0.7785,
+      "loss": 0.8322,
       "step": 22570
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6365500817512869e-06,
-      "loss": 0.7242,
+      "loss": 0.807,
       "step": 22571
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6359044319012108e-06,
-      "loss": 0.6766,
+      "loss": 0.7372,
       "step": 22572
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6352588980910944e-06,
-      "loss": 0.6932,
+      "loss": 0.7539,
       "step": 22573
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6346134803298918e-06,
-      "loss": 0.6443,
+      "loss": 0.8353,
       "step": 22574
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6339681786265593e-06,
-      "loss": 0.8175,
+      "loss": 0.7969,
       "step": 22575
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6333229929900473e-06,
-      "loss": 0.7449,
+      "loss": 0.8821,
       "step": 22576
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6326779234293088e-06,
-      "loss": 0.7722,
+      "loss": 0.8025,
       "step": 22577
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.63203296995329e-06,
-      "loss": 0.5953,
+      "loss": 0.7475,
       "step": 22578
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6313881325709435e-06,
-      "loss": 0.7534,
+      "loss": 0.8159,
       "step": 22579
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6307434112912135e-06,
-      "loss": 0.6348,
+      "loss": 0.733,
       "step": 22580
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6300988061230438e-06,
-      "loss": 0.6412,
+      "loss": 0.812,
       "step": 22581
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6294543170753751e-06,
-      "loss": 0.6735,
+      "loss": 0.7532,
       "step": 22582
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.628809944157157e-06,
-      "loss": 0.6641,
+      "loss": 0.7983,
       "step": 22583
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6281656873773187e-06,
-      "loss": 0.7038,
+      "loss": 0.8229,
       "step": 22584
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6275215467448057e-06,
-      "loss": 0.7909,
+      "loss": 0.8362,
       "step": 22585
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6268775222685518e-06,
-      "loss": 0.6884,
+      "loss": 0.7023,
       "step": 22586
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.626233613957493e-06,
-      "loss": 0.7356,
+      "loss": 0.7714,
       "step": 22587
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6255898218205591e-06,
-      "loss": 0.7235,
+      "loss": 0.8311,
       "step": 22588
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6249461458666904e-06,
-      "loss": 0.656,
+      "loss": 0.7251,
       "step": 22589
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6243025861048056e-06,
-      "loss": 0.6993,
+      "loss": 0.7639,
       "step": 22590
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.623659142543842e-06,
-      "loss": 0.7166,
+      "loss": 0.7843,
       "step": 22591
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6230158151927223e-06,
-      "loss": 0.7537,
+      "loss": 0.8759,
       "step": 22592
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6223726040603738e-06,
-      "loss": 0.6039,
+      "loss": 0.7696,
       "step": 22593
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6217295091557184e-06,
-      "loss": 0.7719,
+      "loss": 0.8138,
       "step": 22594
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6210865304876766e-06,
-      "loss": 0.8906,
+      "loss": 0.9511,
       "step": 22595
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6204436680651768e-06,
-      "loss": 0.6954,
+      "loss": 0.7834,
       "step": 22596
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6198009218971256e-06,
-      "loss": 0.6815,
+      "loss": 0.7666,
       "step": 22597
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6191582919924508e-06,
-      "loss": 0.7721,
+      "loss": 0.7992,
       "step": 22598
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6185157783600624e-06,
-      "loss": 0.6905,
+      "loss": 0.8313,
       "step": 22599
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6178733810088764e-06,
-      "loss": 0.6876,
+      "loss": 0.8155,
       "step": 22600
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6172310999478026e-06,
-      "loss": 0.7373,
+      "loss": 0.8035,
       "step": 22601
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6165889351857578e-06,
-      "loss": 0.6506,
+      "loss": 0.7357,
       "step": 22602
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.615946886731642e-06,
-      "loss": 0.7355,
+      "loss": 0.814,
       "step": 22603
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.615304954594371e-06,
-      "loss": 0.6353,
+      "loss": 0.7394,
       "step": 22604
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.614663138782847e-06,
-      "loss": 0.7423,
+      "loss": 0.8031,
       "step": 22605
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.614021439305975e-06,
-      "loss": 0.7732,
+      "loss": 0.8695,
       "step": 22606
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6133798561726544e-06,
-      "loss": 0.6892,
+      "loss": 0.7247,
       "step": 22607
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6127383893917947e-06,
-      "loss": 0.6049,
+      "loss": 0.7732,
       "step": 22608
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6120970389722846e-06,
-      "loss": 0.758,
+      "loss": 0.7877,
       "step": 22609
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.61145580492303e-06,
-      "loss": 0.7618,
+      "loss": 0.793,
       "step": 22610
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.610814687252924e-06,
-      "loss": 0.6237,
+      "loss": 0.7732,
       "step": 22611
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6101736859708618e-06,
-      "loss": 0.769,
+      "loss": 0.8291,
       "step": 22612
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.609532801085736e-06,
-      "loss": 0.5765,
+      "loss": 0.719,
       "step": 22613
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.608892032606435e-06,
-      "loss": 0.6698,
+      "loss": 0.8204,
       "step": 22614
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6082513805418564e-06,
-      "loss": 0.7783,
+      "loss": 0.8488,
       "step": 22615
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.60761084490088e-06,
-      "loss": 0.629,
+      "loss": 0.7106,
       "step": 22616
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.606970425692398e-06,
-      "loss": 0.5432,
+      "loss": 0.6788,
       "step": 22617
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6063301229252926e-06,
-      "loss": 0.652,
+      "loss": 0.7543,
       "step": 22618
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6056899366084478e-06,
-      "loss": 0.6,
+      "loss": 0.6586,
       "step": 22619
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.605049866750743e-06,
-      "loss": 0.7289,
+      "loss": 0.7507,
       "step": 22620
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6044099133610646e-06,
-      "loss": 0.6564,
+      "loss": 0.7678,
       "step": 22621
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6037700764482821e-06,
-      "loss": 0.7909,
+      "loss": 0.8379,
       "step": 22622
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6031303560212808e-06,
-      "loss": 0.6731,
+      "loss": 0.7809,
       "step": 22623
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.602490752088931e-06,
-      "loss": 0.8217,
+      "loss": 0.8096,
       "step": 22624
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6018512646601081e-06,
-      "loss": 0.7648,
+      "loss": 0.744,
       "step": 22625
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6012118937436804e-06,
-      "loss": 0.8554,
+      "loss": 0.867,
       "step": 22626
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.6005726393485243e-06,
-      "loss": 0.6368,
+      "loss": 0.7611,
       "step": 22627
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.599933501483506e-06,
-      "loss": 0.9311,
+      "loss": 0.981,
       "step": 22628
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5992944801574918e-06,
-      "loss": 0.7306,
+      "loss": 0.8104,
       "step": 22629
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5986555753793464e-06,
-      "loss": 0.6247,
+      "loss": 0.6878,
       "step": 22630
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5980167871579367e-06,
-      "loss": 0.7721,
+      "loss": 0.8068,
       "step": 22631
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.597378115502123e-06,
-      "loss": 0.7114,
+      "loss": 0.8111,
       "step": 22632
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5967395604207626e-06,
-      "loss": 0.6597,
+      "loss": 0.8204,
       "step": 22633
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.596101121922724e-06,
-      "loss": 0.7048,
+      "loss": 0.7927,
       "step": 22634
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5954628000168538e-06,
-      "loss": 0.6888,
+      "loss": 0.711,
       "step": 22635
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5948245947120155e-06,
-      "loss": 0.6505,
+      "loss": 0.7502,
       "step": 22636
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5941865060170602e-06,
-      "loss": 0.665,
+      "loss": 0.8083,
       "step": 22637
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5935485339408408e-06,
-      "loss": 0.7525,
+      "loss": 0.7535,
       "step": 22638
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5929106784922056e-06,
-      "loss": 0.8441,
+      "loss": 0.8624,
       "step": 22639
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5922729396800097e-06,
-      "loss": 0.6567,
+      "loss": 0.8271,
       "step": 22640
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5916353175130972e-06,
-      "loss": 0.7568,
+      "loss": 0.8147,
       "step": 22641
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5909978120003145e-06,
-      "loss": 0.7614,
+      "loss": 0.839,
       "step": 22642
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5903604231505065e-06,
-      "loss": 0.8766,
+      "loss": 1.0275,
       "step": 22643
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5897231509725163e-06,
-      "loss": 0.8002,
+      "loss": 0.894,
       "step": 22644
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5890859954751826e-06,
-      "loss": 0.7893,
+      "loss": 0.8456,
       "step": 22645
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.588448956667349e-06,
-      "loss": 0.6732,
+      "loss": 0.7591,
       "step": 22646
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5878120345578518e-06,
-      "loss": 0.6953,
+      "loss": 0.7391,
       "step": 22647
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5871752291555276e-06,
-      "loss": 0.6332,
+      "loss": 0.7554,
       "step": 22648
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.586538540469208e-06,
-      "loss": 0.6943,
+      "loss": 0.7969,
       "step": 22649
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5859019685077347e-06,
-      "loss": 0.7039,
+      "loss": 0.7796,
       "step": 22650
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.58526551327993e-06,
-      "loss": 0.6621,
+      "loss": 0.812,
       "step": 22651
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5846291747946264e-06,
-      "loss": 0.7519,
+      "loss": 0.7844,
       "step": 22652
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5839929530606568e-06,
-      "loss": 0.7918,
+      "loss": 0.7216,
       "step": 22653
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5833568480868388e-06,
-      "loss": 0.5461,
+      "loss": 0.6451,
       "step": 22654
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5827208598820066e-06,
-      "loss": 0.6896,
+      "loss": 0.761,
       "step": 22655
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5820849884549793e-06,
-      "loss": 0.7961,
+      "loss": 0.8513,
       "step": 22656
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5814492338145793e-06,
-      "loss": 0.6493,
+      "loss": 0.7264,
       "step": 22657
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5808135959696247e-06,
-      "loss": 0.7449,
+      "loss": 0.8183,
       "step": 22658
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5801780749289385e-06,
-      "loss": 0.7972,
+      "loss": 0.9102,
       "step": 22659
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5795426707013362e-06,
-      "loss": 0.6185,
+      "loss": 0.7763,
       "step": 22660
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5789073832956315e-06,
-      "loss": 0.7604,
+      "loss": 0.7906,
       "step": 22661
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5782722127206385e-06,
-      "loss": 0.6251,
+      "loss": 0.7867,
       "step": 22662
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5776371589851703e-06,
-      "loss": 0.6939,
+      "loss": 0.8279,
       "step": 22663
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.577002222098033e-06,
-      "loss": 0.6428,
+      "loss": 0.7532,
       "step": 22664
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.576367402068043e-06,
-      "loss": 0.6568,
+      "loss": 0.7288,
       "step": 22665
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5757326989040034e-06,
-      "loss": 0.7419,
+      "loss": 0.7905,
       "step": 22666
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5750981126147203e-06,
-      "loss": 0.6162,
+      "loss": 0.7639,
       "step": 22667
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5744636432089944e-06,
-      "loss": 0.6605,
+      "loss": 0.78,
       "step": 22668
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5738292906956364e-06,
-      "loss": 0.6647,
+      "loss": 0.7747,
       "step": 22669
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5731950550834397e-06,
-      "loss": 0.7403,
+      "loss": 0.7913,
       "step": 22670
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5725609363812033e-06,
-      "loss": 0.7485,
+      "loss": 0.8424,
       "step": 22671
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5719269345977283e-06,
-      "loss": 0.7506,
+      "loss": 0.864,
       "step": 22672
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.57129304974181e-06,
-      "loss": 0.6899,
+      "loss": 0.8012,
       "step": 22673
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5706592818222422e-06,
-      "loss": 0.735,
+      "loss": 0.7715,
       "step": 22674
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5700256308478167e-06,
-      "loss": 0.7402,
+      "loss": 0.7914,
       "step": 22675
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5693920968273247e-06,
-      "loss": 0.7633,
+      "loss": 0.8032,
       "step": 22676
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5687586797695543e-06,
-      "loss": 0.7245,
+      "loss": 0.8077,
       "step": 22677
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5681253796832963e-06,
-      "loss": 0.733,
+      "loss": 0.7547,
       "step": 22678
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.567492196577336e-06,
-      "loss": 0.5487,
+      "loss": 0.7112,
       "step": 22679
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5668591304604574e-06,
-      "loss": 0.7006,
+      "loss": 0.791,
       "step": 22680
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5662261813414404e-06,
-      "loss": 0.7163,
+      "loss": 0.7947,
       "step": 22681
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5655933492290743e-06,
-      "loss": 0.6491,
+      "loss": 0.7519,
       "step": 22682
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5649606341321289e-06,
-      "loss": 0.63,
+      "loss": 0.7239,
       "step": 22683
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5643280360593893e-06,
-      "loss": 0.7557,
+      "loss": 0.7811,
       "step": 22684
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.5636955550196298e-06,
-      "loss": 0.7825,
+      "loss": 0.818,
       "step": 22685
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5630631910216242e-06,
-      "loss": 0.8151,
+      "loss": 0.8494,
       "step": 22686
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5624309440741447e-06,
-      "loss": 0.7489,
+      "loss": 0.7746,
       "step": 22687
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5617988141859686e-06,
-      "loss": 0.7175,
+      "loss": 0.8074,
       "step": 22688
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.56116680136586e-06,
-      "loss": 0.794,
+      "loss": 0.8089,
       "step": 22689
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.560534905622586e-06,
-      "loss": 0.7747,
+      "loss": 0.8576,
       "step": 22690
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.559903126964919e-06,
-      "loss": 0.5867,
+      "loss": 0.762,
       "step": 22691
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5592714654016207e-06,
-      "loss": 0.6633,
+      "loss": 0.7748,
       "step": 22692
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5586399209414548e-06,
-      "loss": 0.8072,
+      "loss": 0.904,
       "step": 22693
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5580084935931816e-06,
-      "loss": 0.7011,
+      "loss": 0.7748,
       "step": 22694
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5573771833655672e-06,
-      "loss": 0.7948,
+      "loss": 0.9092,
       "step": 22695
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5567459902673622e-06,
-      "loss": 0.6749,
+      "loss": 0.7813,
       "step": 22696
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5561149143073284e-06,
-      "loss": 0.7083,
+      "loss": 0.78,
       "step": 22697
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5554839554942203e-06,
-      "loss": 0.6262,
+      "loss": 0.8273,
       "step": 22698
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5548531138367906e-06,
-      "loss": 0.8488,
+      "loss": 0.8462,
       "step": 22699
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5542223893437902e-06,
-      "loss": 0.7247,
+      "loss": 0.8398,
       "step": 22700
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5535917820239754e-06,
-      "loss": 0.731,
+      "loss": 0.8227,
       "step": 22701
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.552961291886086e-06,
-      "loss": 0.7473,
+      "loss": 0.8211,
       "step": 22702
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5523309189388757e-06,
-      "loss": 0.7145,
+      "loss": 0.8422,
       "step": 22703
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5517006631910892e-06,
-      "loss": 0.8533,
+      "loss": 0.884,
       "step": 22704
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.551070524651469e-06,
-      "loss": 0.8338,
+      "loss": 0.9181,
       "step": 22705
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5504405033287562e-06,
-      "loss": 0.6362,
+      "loss": 0.7485,
       "step": 22706
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.549810599231697e-06,
-      "loss": 0.7564,
+      "loss": 0.8137,
       "step": 22707
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5491808123690245e-06,
-      "loss": 0.7976,
+      "loss": 0.8673,
       "step": 22708
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5485511427494748e-06,
-      "loss": 0.7159,
+      "loss": 0.7485,
       "step": 22709
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5479215903817902e-06,
-      "loss": 0.7363,
+      "loss": 0.8372,
       "step": 22710
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.547292155274701e-06,
-      "loss": 0.6901,
+      "loss": 0.7593,
       "step": 22711
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5466628374369408e-06,
-      "loss": 0.7186,
+      "loss": 0.7841,
       "step": 22712
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5460336368772377e-06,
-      "loss": 0.7415,
+      "loss": 0.8342,
       "step": 22713
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5454045536043283e-06,
-      "loss": 0.7911,
+      "loss": 0.8261,
       "step": 22714
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5447755876269298e-06,
-      "loss": 0.7349,
+      "loss": 0.8036,
       "step": 22715
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5441467389537768e-06,
-      "loss": 0.7932,
+      "loss": 0.8269,
       "step": 22716
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5435180075935896e-06,
-      "loss": 0.7204,
+      "loss": 0.7895,
       "step": 22717
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5428893935550925e-06,
-      "loss": 0.6242,
+      "loss": 0.7391,
       "step": 22718
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5422608968470032e-06,
-      "loss": 0.7826,
+      "loss": 0.8592,
       "step": 22719
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.541632517478049e-06,
-      "loss": 0.7823,
+      "loss": 0.808,
       "step": 22720
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.541004255456937e-06,
-      "loss": 0.7774,
+      "loss": 0.7648,
       "step": 22721
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.540376110792393e-06,
-      "loss": 0.6155,
+      "loss": 0.7114,
       "step": 22722
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5397480834931279e-06,
-      "loss": 0.6948,
+      "loss": 0.8017,
       "step": 22723
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.539120173567854e-06,
-      "loss": 0.5357,
+      "loss": 0.7755,
       "step": 22724
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5384923810252805e-06,
-      "loss": 0.669,
+      "loss": 0.7259,
       "step": 22725
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5378647058741224e-06,
-      "loss": 0.6652,
+      "loss": 0.7845,
       "step": 22726
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5372371481230887e-06,
-      "loss": 0.6949,
+      "loss": 0.8086,
       "step": 22727
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5366097077808773e-06,
-      "loss": 0.799,
+      "loss": 0.8286,
       "step": 22728
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5359823848561994e-06,
-      "loss": 0.6794,
+      "loss": 0.744,
       "step": 22729
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5353551793577582e-06,
-      "loss": 0.7233,
+      "loss": 0.8436,
       "step": 22730
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5347280912942542e-06,
-      "loss": 0.7146,
+      "loss": 0.8377,
       "step": 22731
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.534101120674385e-06,
-      "loss": 0.6535,
+      "loss": 0.7194,
       "step": 22732
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5334742675068547e-06,
-      "loss": 0.7268,
+      "loss": 0.813,
       "step": 22733
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5328475318003533e-06,
-      "loss": 0.5747,
+      "loss": 0.7219,
       "step": 22734
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5322209135635802e-06,
-      "loss": 0.7988,
+      "loss": 0.7697,
       "step": 22735
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5315944128052284e-06,
-      "loss": 0.6828,
+      "loss": 0.8262,
       "step": 22736
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.530968029533989e-06,
-      "loss": 0.6795,
+      "loss": 0.7935,
       "step": 22737
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5303417637585506e-06,
-      "loss": 0.6649,
+      "loss": 0.818,
       "step": 22738
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5297156154876046e-06,
-      "loss": 0.7745,
+      "loss": 0.763,
       "step": 22739
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.529089584729837e-06,
-      "loss": 0.8888,
+      "loss": 0.8905,
       "step": 22740
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5284636714939327e-06,
-      "loss": 0.7139,
+      "loss": 0.7644,
       "step": 22741
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5278378757885748e-06,
-      "loss": 0.688,
+      "loss": 0.7825,
       "step": 22742
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5272121976224462e-06,
-      "loss": 0.7635,
+      "loss": 0.7927,
       "step": 22743
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5265866370042259e-06,
-      "loss": 0.7565,
+      "loss": 0.8385,
       "step": 22744
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5259611939425957e-06,
-      "loss": 0.7097,
+      "loss": 0.8098,
       "step": 22745
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5253358684462328e-06,
-      "loss": 0.612,
+      "loss": 0.8172,
       "step": 22746
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.524710660523806e-06,
-      "loss": 0.7074,
+      "loss": 0.7572,
       "step": 22747
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5240855701839962e-06,
-      "loss": 0.6658,
+      "loss": 0.7253,
       "step": 22748
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5234605974354732e-06,
-      "loss": 0.7338,
+      "loss": 0.7788,
       "step": 22749
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5228357422869077e-06,
-      "loss": 0.6896,
+      "loss": 0.799,
       "step": 22750
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5222110047469673e-06,
-      "loss": 0.6697,
+      "loss": 0.7566,
       "step": 22751
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5215863848243252e-06,
-      "loss": 0.7813,
+      "loss": 0.8239,
       "step": 22752
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5209618825276373e-06,
-      "loss": 0.6209,
+      "loss": 0.7174,
       "step": 22753
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.520337497865576e-06,
-      "loss": 0.7016,
+      "loss": 0.8423,
       "step": 22754
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.519713230846801e-06,
-      "loss": 0.7465,
+      "loss": 0.7731,
       "step": 22755
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.519089081479973e-06,
-      "loss": 0.7643,
+      "loss": 0.7962,
       "step": 22756
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5184650497737486e-06,
-      "loss": 0.7649,
+      "loss": 0.8206,
       "step": 22757
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5178411357367907e-06,
-      "loss": 0.6969,
+      "loss": 0.7523,
       "step": 22758
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5172173393777512e-06,
-      "loss": 0.6719,
+      "loss": 0.7744,
       "step": 22759
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5165936607052867e-06,
-      "loss": 0.724,
+      "loss": 0.7575,
       "step": 22760
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5159700997280458e-06,
-      "loss": 0.6947,
+      "loss": 0.7756,
       "step": 22761
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.515346656454688e-06,
-      "loss": 0.7709,
+      "loss": 0.8585,
       "step": 22762
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5147233308938514e-06,
-      "loss": 0.7188,
+      "loss": 0.8173,
       "step": 22763
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.514100123054193e-06,
-      "loss": 0.6958,
+      "loss": 0.8641,
       "step": 22764
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5134770329443571e-06,
-      "loss": 0.6874,
+      "loss": 0.7673,
       "step": 22765
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5128540605729824e-06,
-      "loss": 0.7903,
+      "loss": 0.8047,
       "step": 22766
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5122312059487177e-06,
-      "loss": 0.6982,
+      "loss": 0.7718,
       "step": 22767
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5116084690802023e-06,
-      "loss": 0.569,
+      "loss": 0.667,
       "step": 22768
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5109858499760765e-06,
-      "loss": 0.8068,
+      "loss": 0.8218,
       "step": 22769
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5103633486449754e-06,
-      "loss": 0.6874,
+      "loss": 0.8388,
       "step": 22770
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5097409650955398e-06,
-      "loss": 0.8033,
+      "loss": 0.8747,
       "step": 22771
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5091186993364026e-06,
-      "loss": 0.7386,
+      "loss": 0.7773,
       "step": 22772
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5084965513761963e-06,
-      "loss": 0.7726,
+      "loss": 0.8386,
       "step": 22773
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5078745212235523e-06,
-      "loss": 0.6541,
+      "loss": 0.8172,
       "step": 22774
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5072526088870997e-06,
-      "loss": 0.7647,
+      "loss": 0.8676,
       "step": 22775
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.506630814375466e-06,
-      "loss": 0.5138,
+      "loss": 0.71,
       "step": 22776
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5060091376972818e-06,
-      "loss": 0.7767,
+      "loss": 0.8564,
       "step": 22777
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5053875788611694e-06,
-      "loss": 0.611,
+      "loss": 0.6997,
       "step": 22778
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5047661378757517e-06,
-      "loss": 0.6458,
+      "loss": 0.7658,
       "step": 22779
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5041448147496484e-06,
-      "loss": 0.7403,
+      "loss": 0.832,
       "step": 22780
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5035236094914873e-06,
-      "loss": 0.8139,
+      "loss": 0.7581,
       "step": 22781
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5029025221098747e-06,
-      "loss": 0.6397,
+      "loss": 0.8195,
       "step": 22782
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5022815526134382e-06,
-      "loss": 0.7486,
+      "loss": 0.7621,
       "step": 22783
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5016607010107876e-06,
-      "loss": 0.668,
+      "loss": 0.7659,
       "step": 22784
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5010399673105368e-06,
-      "loss": 0.8456,
+      "loss": 0.8997,
       "step": 22785
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.5004193515212984e-06,
-      "loss": 0.7613,
+      "loss": 0.8453,
       "step": 22786
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4997988536516817e-06,
-      "loss": 0.7245,
+      "loss": 0.8135,
       "step": 22787
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4991784737102955e-06,
-      "loss": 0.632,
+      "loss": 0.7749,
       "step": 22788
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4985582117057452e-06,
-      "loss": 0.6922,
+      "loss": 0.756,
       "step": 22789
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4979380676466404e-06,
-      "loss": 0.6529,
+      "loss": 0.7589,
       "step": 22790
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4973180415415812e-06,
-      "loss": 0.6507,
+      "loss": 0.7638,
       "step": 22791
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4966981333991714e-06,
-      "loss": 0.7846,
+      "loss": 0.8963,
       "step": 22792
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4960783432280067e-06,
-      "loss": 0.6806,
+      "loss": 0.7825,
       "step": 22793
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4954586710366958e-06,
-      "loss": 0.7332,
+      "loss": 0.7298,
       "step": 22794
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.494839116833824e-06,
-      "loss": 0.6294,
+      "loss": 0.6618,
       "step": 22795
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.494219680627994e-06,
-      "loss": 0.7174,
+      "loss": 0.7584,
       "step": 22796
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4936003624277985e-06,
-      "loss": 0.6778,
+      "loss": 0.8148,
       "step": 22797
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.492981162241829e-06,
-      "loss": 0.5672,
+      "loss": 0.7216,
       "step": 22798
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.492362080078673e-06,
-      "loss": 0.811,
+      "loss": 0.803,
       "step": 22799
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4917431159469276e-06,
-      "loss": 0.5953,
+      "loss": 0.8167,
       "step": 22800
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4911242698551709e-06,
-      "loss": 0.686,
+      "loss": 0.7498,
       "step": 22801
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.490505541811994e-06,
-      "loss": 0.6241,
+      "loss": 0.7023,
       "step": 22802
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.489886931825979e-06,
-      "loss": 0.7357,
+      "loss": 0.7809,
       "step": 22803
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4892684399057089e-06,
-      "loss": 0.7098,
+      "loss": 0.7958,
       "step": 22804
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4886500660597636e-06,
-      "loss": 0.7218,
+      "loss": 0.766,
       "step": 22805
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4880318102967207e-06,
-      "loss": 0.7258,
+      "loss": 0.6943,
       "step": 22806
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4874136726251643e-06,
-      "loss": 0.6997,
+      "loss": 0.808,
       "step": 22807
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.486795653053661e-06,
-      "loss": 0.7865,
+      "loss": 0.7982,
       "step": 22808
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4861777515907904e-06,
-      "loss": 0.7167,
+      "loss": 0.7868,
       "step": 22809
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4855599682451248e-06,
-      "loss": 0.5943,
+      "loss": 0.7388,
       "step": 22810
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4849423030252352e-06,
-      "loss": 0.553,
+      "loss": 0.7398,
       "step": 22811
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4843247559396857e-06,
-      "loss": 0.5172,
+      "loss": 0.6737,
       "step": 22812
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4837073269970537e-06,
-      "loss": 0.5751,
+      "loss": 0.7406,
       "step": 22813
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4830900162058947e-06,
-      "loss": 0.614,
+      "loss": 0.7635,
       "step": 22814
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4824728235747809e-06,
-      "loss": 0.785,
+      "loss": 0.9027,
       "step": 22815
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4818557491122709e-06,
-      "loss": 0.7788,
+      "loss": 0.8356,
       "step": 22816
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4812387928269267e-06,
-      "loss": 0.7075,
+      "loss": 0.8271,
       "step": 22817
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.480621954727306e-06,
-      "loss": 0.7401,
+      "loss": 0.8491,
       "step": 22818
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4800052348219728e-06,
-      "loss": 0.6554,
+      "loss": 0.7232,
       "step": 22819
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.479388633119474e-06,
-      "loss": 0.6198,
+      "loss": 0.7924,
       "step": 22820
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4787721496283701e-06,
-      "loss": 0.7185,
+      "loss": 0.8572,
       "step": 22821
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4781557843572137e-06,
-      "loss": 0.7107,
+      "loss": 0.7836,
       "step": 22822
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.477539537314554e-06,
-      "loss": 0.7122,
+      "loss": 0.7906,
       "step": 22823
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4769234085089423e-06,
-      "loss": 0.6841,
+      "loss": 0.7828,
       "step": 22824
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.476307397948923e-06,
-      "loss": 0.6604,
+      "loss": 0.7332,
       "step": 22825
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.47569150564305e-06,
-      "loss": 0.644,
+      "loss": 0.7795,
       "step": 22826
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4750757315998575e-06,
-      "loss": 0.7508,
+      "loss": 0.8052,
       "step": 22827
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4744600758278971e-06,
-      "loss": 0.7245,
+      "loss": 0.8092,
       "step": 22828
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4738445383357069e-06,
-      "loss": 0.8477,
+      "loss": 0.8564,
       "step": 22829
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.473229119131826e-06,
-      "loss": 0.6246,
+      "loss": 0.6722,
       "step": 22830
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4726138182247907e-06,
-      "loss": 0.6407,
+      "loss": 0.7798,
       "step": 22831
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4719986356231453e-06,
-      "loss": 0.7699,
+      "loss": 0.9025,
       "step": 22832
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.471383571335414e-06,
-      "loss": 0.806,
+      "loss": 0.8165,
       "step": 22833
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4707686253701381e-06,
-      "loss": 0.4759,
+      "loss": 0.6276,
       "step": 22834
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.470153797735846e-06,
-      "loss": 0.5987,
+      "loss": 0.7083,
       "step": 22835
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4695390884410676e-06,
-      "loss": 0.6399,
+      "loss": 0.7717,
       "step": 22836
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4689244974943285e-06,
-      "loss": 0.7861,
+      "loss": 0.8323,
       "step": 22837
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4683100249041604e-06,
-      "loss": 0.8804,
+      "loss": 0.9099,
       "step": 22838
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4676956706790868e-06,
-      "loss": 0.6808,
+      "loss": 0.7374,
       "step": 22839
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4670814348276286e-06,
-      "loss": 0.7762,
+      "loss": 0.8333,
       "step": 22840
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.46646731735831e-06,
-      "loss": 0.6791,
+      "loss": 0.7488,
       "step": 22841
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4658533182796486e-06,
-      "loss": 0.711,
+      "loss": 0.8084,
       "step": 22842
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4652394376001656e-06,
-      "loss": 0.6864,
+      "loss": 0.8688,
       "step": 22843
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4646256753283726e-06,
-      "loss": 0.7506,
+      "loss": 0.8769,
       "step": 22844
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4640120314727935e-06,
-      "loss": 0.7301,
+      "loss": 0.791,
       "step": 22845
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4633985060419308e-06,
-      "loss": 0.7519,
+      "loss": 0.8136,
       "step": 22846
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.462785099044306e-06,
-      "loss": 0.6014,
+      "loss": 0.7349,
       "step": 22847
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4621718104884242e-06,
-      "loss": 0.6003,
+      "loss": 0.7203,
       "step": 22848
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4615586403827954e-06,
-      "loss": 0.7021,
+      "loss": 0.8663,
       "step": 22849
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4609455887359226e-06,
-      "loss": 0.7565,
+      "loss": 0.8366,
       "step": 22850
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4603326555563202e-06,
-      "loss": 0.8374,
+      "loss": 0.8679,
       "step": 22851
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4597198408524814e-06,
-      "loss": 0.6713,
+      "loss": 0.7575,
       "step": 22852
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.459107144632914e-06,
-      "loss": 0.6362,
+      "loss": 0.7955,
       "step": 22853
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4584945669061178e-06,
-      "loss": 0.8028,
+      "loss": 0.8572,
       "step": 22854
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4578821076805904e-06,
-      "loss": 0.6539,
+      "loss": 0.7947,
       "step": 22855
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4572697669648273e-06,
-      "loss": 0.6959,
+      "loss": 0.799,
       "step": 22856
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.456657544767327e-06,
-      "loss": 0.8569,
+      "loss": 0.8809,
       "step": 22857
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.456045441096583e-06,
-      "loss": 0.7124,
+      "loss": 0.7518,
       "step": 22858
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.455433455961085e-06,
-      "loss": 0.7262,
+      "loss": 0.8044,
       "step": 22859
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.454821589369324e-06,
-      "loss": 0.6867,
+      "loss": 0.7946,
       "step": 22860
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4542098413297923e-06,
-      "loss": 0.9252,
+      "loss": 0.9465,
       "step": 22861
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4535982118509728e-06,
-      "loss": 0.7629,
+      "loss": 0.7664,
       "step": 22862
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.45298670094135e-06,
-      "loss": 0.6486,
+      "loss": 0.8112,
       "step": 22863
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4523753086094138e-06,
-      "loss": 0.6583,
+      "loss": 0.7809,
       "step": 22864
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4517640348636397e-06,
-      "loss": 0.7428,
+      "loss": 0.7747,
       "step": 22865
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4511528797125119e-06,
-      "loss": 0.8637,
+      "loss": 0.8688,
       "step": 22866
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.45054184316451e-06,
-      "loss": 0.7671,
+      "loss": 0.8238,
       "step": 22867
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.449930925228109e-06,
-      "loss": 0.7296,
+      "loss": 0.8692,
       "step": 22868
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.449320125911784e-06,
-      "loss": 0.667,
+      "loss": 0.7973,
       "step": 22869
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4487094452240125e-06,
-      "loss": 0.804,
+      "loss": 0.8883,
       "step": 22870
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4480988831732634e-06,
-      "loss": 0.7862,
+      "loss": 0.8191,
       "step": 22871
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.44748843976801e-06,
-      "loss": 0.7367,
+      "loss": 0.7394,
       "step": 22872
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.44687811501672e-06,
-      "loss": 0.7463,
+      "loss": 0.7688,
       "step": 22873
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.446267908927861e-06,
-      "loss": 0.7671,
+      "loss": 0.7521,
       "step": 22874
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.445657821509896e-06,
-      "loss": 0.8292,
+      "loss": 0.8167,
       "step": 22875
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.445047852771294e-06,
-      "loss": 0.6144,
+      "loss": 0.807,
       "step": 22876
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4444380027205151e-06,
-      "loss": 0.7205,
+      "loss": 0.7499,
       "step": 22877
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.44382827136602e-06,
-      "loss": 0.6508,
+      "loss": 0.7771,
       "step": 22878
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4432186587162667e-06,
-      "loss": 0.6944,
+      "loss": 0.7915,
       "step": 22879
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.442609164779718e-06,
-      "loss": 0.7639,
+      "loss": 0.8524,
       "step": 22880
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.441999789564823e-06,
-      "loss": 0.6467,
+      "loss": 0.7757,
       "step": 22881
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4413905330800371e-06,
-      "loss": 0.8494,
+      "loss": 0.8752,
       "step": 22882
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4407813953338167e-06,
-      "loss": 0.7619,
+      "loss": 0.7943,
       "step": 22883
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4401723763346097e-06,
-      "loss": 0.6979,
+      "loss": 0.7377,
       "step": 22884
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4395634760908672e-06,
-      "loss": 0.9177,
+      "loss": 0.9443,
       "step": 22885
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4389546946110355e-06,
-      "loss": 0.7199,
+      "loss": 0.7859,
       "step": 22886
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4383460319035602e-06,
-      "loss": 0.7589,
+      "loss": 0.7791,
       "step": 22887
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4377374879768845e-06,
-      "loss": 0.6912,
+      "loss": 0.8168,
       "step": 22888
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4371290628394551e-06,
-      "loss": 0.8642,
+      "loss": 0.853,
       "step": 22889
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.436520756499712e-06,
-      "loss": 0.8125,
+      "loss": 0.8444,
       "step": 22890
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4359125689660914e-06,
-      "loss": 0.7552,
+      "loss": 0.8588,
       "step": 22891
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4353045002470324e-06,
-      "loss": 0.7778,
+      "loss": 0.8618,
       "step": 22892
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4346965503509757e-06,
-      "loss": 0.7646,
+      "loss": 0.8103,
       "step": 22893
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4340887192863473e-06,
-      "loss": 0.7537,
+      "loss": 0.8241,
       "step": 22894
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4334810070615856e-06,
-      "loss": 0.7818,
+      "loss": 0.8694,
       "step": 22895
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4328734136851218e-06,
-      "loss": 0.6579,
+      "loss": 0.7869,
       "step": 22896
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.432265939165385e-06,
-      "loss": 0.682,
+      "loss": 0.7612,
       "step": 22897
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4316585835107988e-06,
-      "loss": 0.6112,
+      "loss": 0.7179,
       "step": 22898
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.431051346729797e-06,
-      "loss": 0.7619,
+      "loss": 0.802,
       "step": 22899
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4304442288307996e-06,
-      "loss": 0.6449,
+      "loss": 0.7733,
       "step": 22900
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4298372298222262e-06,
-      "loss": 0.7,
+      "loss": 0.7522,
       "step": 22901
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4292303497125037e-06,
-      "loss": 0.6694,
+      "loss": 0.8208,
       "step": 22902
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4286235885100508e-06,
-      "loss": 0.731,
+      "loss": 0.7935,
       "step": 22903
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.428016946223284e-06,
-      "loss": 0.7244,
+      "loss": 0.7816,
       "step": 22904
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4274104228606179e-06,
-      "loss": 0.7209,
+      "loss": 0.7693,
       "step": 22905
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4268040184304732e-06,
-      "loss": 0.5977,
+      "loss": 0.7501,
       "step": 22906
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4261977329412546e-06,
-      "loss": 0.7794,
+      "loss": 0.8398,
       "step": 22907
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4255915664013808e-06,
-      "loss": 0.5982,
+      "loss": 0.7077,
       "step": 22908
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4249855188192584e-06,
-      "loss": 0.6091,
+      "loss": 0.7504,
       "step": 22909
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4243795902032953e-06,
-      "loss": 0.7157,
+      "loss": 0.8097,
       "step": 22910
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4237737805618956e-06,
-      "loss": 0.7183,
+      "loss": 0.7722,
       "step": 22911
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4231680899034718e-06,
-      "loss": 0.687,
+      "loss": 0.8095,
       "step": 22912
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4225625182364168e-06,
-      "loss": 0.614,
+      "loss": 0.7506,
       "step": 22913
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4219570655691395e-06,
-      "loss": 0.6763,
+      "loss": 0.746,
       "step": 22914
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4213517319100367e-06,
-      "loss": 0.8279,
+      "loss": 0.8539,
       "step": 22915
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4207465172675083e-06,
-      "loss": 0.7874,
+      "loss": 0.8365,
       "step": 22916
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4201414216499466e-06,
-      "loss": 0.7446,
+      "loss": 0.8179,
       "step": 22917
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4195364450657534e-06,
-      "loss": 0.6764,
+      "loss": 0.7805,
       "step": 22918
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4189315875233166e-06,
-      "loss": 0.6602,
+      "loss": 0.781,
       "step": 22919
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4183268490310265e-06,
-      "loss": 0.6116,
+      "loss": 0.7534,
       "step": 22920
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4177222295972781e-06,
-      "loss": 0.6901,
+      "loss": 0.8057,
       "step": 22921
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4171177292304571e-06,
-      "loss": 0.6747,
+      "loss": 0.7625,
       "step": 22922
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4165133479389493e-06,
-      "loss": 0.5842,
+      "loss": 0.7914,
       "step": 22923
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4159090857311386e-06,
-      "loss": 0.7513,
+      "loss": 0.8341,
       "step": 22924
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4153049426154153e-06,
-      "loss": 0.7163,
+      "loss": 0.8071,
       "step": 22925
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4147009186001503e-06,
-      "loss": 0.7029,
+      "loss": 0.7155,
       "step": 22926
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4140970136937327e-06,
-      "loss": 0.7228,
+      "loss": 0.8212,
       "step": 22927
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4134932279045355e-06,
-      "loss": 0.7775,
+      "loss": 0.8057,
       "step": 22928
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4128895612409387e-06,
-      "loss": 0.683,
+      "loss": 0.7502,
       "step": 22929
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4122860137113126e-06,
-      "loss": 0.8928,
+      "loss": 0.925,
       "step": 22930
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.411682585324038e-06,
-      "loss": 0.7099,
+      "loss": 0.8018,
       "step": 22931
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.411079276087478e-06,
-      "loss": 0.7594,
+      "loss": 0.8479,
       "step": 22932
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4104760860100098e-06,
-      "loss": 0.767,
+      "loss": 0.775,
       "step": 22933
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4098730150999985e-06,
-      "loss": 0.7223,
+      "loss": 0.7561,
       "step": 22934
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4092700633658108e-06,
-      "loss": 0.7598,
+      "loss": 0.8274,
       "step": 22935
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4086672308158112e-06,
-      "loss": 0.829,
+      "loss": 0.886,
       "step": 22936
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4080645174583652e-06,
-      "loss": 0.7175,
+      "loss": 0.8055,
       "step": 22937
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4074619233018371e-06,
-      "loss": 0.6912,
+      "loss": 0.7747,
       "step": 22938
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4068594483545783e-06,
-      "loss": 0.7341,
+      "loss": 0.7761,
       "step": 22939
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4062570926249553e-06,
-      "loss": 0.7979,
+      "loss": 0.9028,
       "step": 22940
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4056548561213223e-06,
-      "loss": 0.7094,
+      "loss": 0.7767,
       "step": 22941
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.405052738852034e-06,
-      "loss": 0.6788,
+      "loss": 0.8241,
       "step": 22942
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4044507408254426e-06,
-      "loss": 0.7266,
+      "loss": 0.8324,
       "step": 22943
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4038488620499057e-06,
-      "loss": 0.6055,
+      "loss": 0.8022,
       "step": 22944
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4032471025337645e-06,
-      "loss": 0.5818,
+      "loss": 0.7408,
       "step": 22945
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4026454622853759e-06,
-      "loss": 0.5757,
+      "loss": 0.6666,
       "step": 22946
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4020439413130826e-06,
-      "loss": 0.7401,
+      "loss": 0.7396,
       "step": 22947
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4014425396252307e-06,
-      "loss": 0.7538,
+      "loss": 0.786,
       "step": 22948
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4008412572301622e-06,
-      "loss": 0.7049,
+      "loss": 0.7948,
       "step": 22949
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4002400941362227e-06,
-      "loss": 0.8532,
+      "loss": 0.8921,
       "step": 22950
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3996390503517509e-06,
-      "loss": 0.5721,
+      "loss": 0.6396,
       "step": 22951
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3990381258850838e-06,
-      "loss": 0.7766,
+      "loss": 0.8522,
       "step": 22952
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.39843732074456e-06,
-      "loss": 0.7308,
+      "loss": 0.7551,
       "step": 22953
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3978366349385154e-06,
-      "loss": 0.7585,
+      "loss": 0.7598,
       "step": 22954
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3972360684752784e-06,
-      "loss": 0.6648,
+      "loss": 0.7248,
       "step": 22955
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3966356213631893e-06,
-      "loss": 0.5933,
+      "loss": 0.7507,
       "step": 22956
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.396035293610576e-06,
-      "loss": 0.821,
+      "loss": 0.8628,
       "step": 22957
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3954350852257602e-06,
-      "loss": 0.7304,
+      "loss": 0.785,
       "step": 22958
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.394834996217077e-06,
-      "loss": 0.7551,
+      "loss": 0.803,
       "step": 22959
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3942350265928495e-06,
-      "loss": 0.7249,
+      "loss": 0.7715,
       "step": 22960
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3936351763614008e-06,
-      "loss": 0.6578,
+      "loss": 0.7476,
       "step": 22961
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3930354455310513e-06,
-      "loss": 0.7506,
+      "loss": 0.8447,
       "step": 22962
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3924358341101274e-06,
-      "loss": 0.7814,
+      "loss": 0.8579,
       "step": 22963
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3918363421069403e-06,
-      "loss": 0.5425,
+      "loss": 0.6257,
       "step": 22964
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3912369695298111e-06,
-      "loss": 0.8507,
+      "loss": 0.9145,
       "step": 22965
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3906377163870567e-06,
-      "loss": 0.7135,
+      "loss": 0.8777,
       "step": 22966
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3900385826869867e-06,
-      "loss": 0.8755,
+      "loss": 0.899,
       "step": 22967
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3894395684379147e-06,
-      "loss": 0.7944,
+      "loss": 0.8167,
       "step": 22968
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.388840673648154e-06,
-      "loss": 0.8443,
+      "loss": 0.8864,
       "step": 22969
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3882418983260103e-06,
-      "loss": 0.6988,
+      "loss": 0.796,
       "step": 22970
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3876432424797925e-06,
-      "loss": 0.81,
+      "loss": 0.8442,
       "step": 22971
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.387044706117806e-06,
-      "loss": 0.7183,
+      "loss": 0.806,
       "step": 22972
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3864462892483533e-06,
-      "loss": 0.7602,
+      "loss": 0.8665,
       "step": 22973
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3858479918797353e-06,
-      "loss": 0.5793,
+      "loss": 0.7424,
       "step": 22974
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3852498140202563e-06,
-      "loss": 0.6225,
+      "loss": 0.7436,
       "step": 22975
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3846517556782157e-06,
-      "loss": 0.613,
+      "loss": 0.6751,
       "step": 22976
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3840538168619045e-06,
-      "loss": 0.7452,
+      "loss": 0.8143,
       "step": 22977
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3834559975796235e-06,
-      "loss": 0.892,
+      "loss": 0.9285,
       "step": 22978
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3828582978396665e-06,
-      "loss": 0.8049,
+      "loss": 0.8681,
       "step": 22979
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3822607176503233e-06,
-      "loss": 0.7432,
+      "loss": 0.8136,
       "step": 22980
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.381663257019884e-06,
-      "loss": 0.7162,
+      "loss": 0.7942,
       "step": 22981
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3810659159566408e-06,
-      "loss": 0.8136,
+      "loss": 0.8655,
       "step": 22982
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3804686944688805e-06,
-      "loss": 0.6858,
+      "loss": 0.7345,
       "step": 22983
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3798715925648865e-06,
-      "loss": 0.7433,
+      "loss": 0.8808,
       "step": 22984
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3792746102529442e-06,
-      "loss": 0.6678,
+      "loss": 0.7325,
       "step": 22985
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3786777475413359e-06,
-      "loss": 0.6416,
+      "loss": 0.7638,
       "step": 22986
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3780810044383397e-06,
-      "loss": 0.6199,
+      "loss": 0.7574,
       "step": 22987
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3774843809522376e-06,
-      "loss": 0.7326,
+      "loss": 0.7395,
       "step": 22988
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3768878770913075e-06,
-      "loss": 0.7256,
+      "loss": 0.8043,
       "step": 22989
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3762914928638227e-06,
-      "loss": 0.5864,
+      "loss": 0.7328,
       "step": 22990
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3756952282780556e-06,
-      "loss": 0.7231,
+      "loss": 0.7919,
       "step": 22991
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.375099083342285e-06,
-      "loss": 0.6928,
+      "loss": 0.7897,
       "step": 22992
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3745030580647745e-06,
-      "loss": 0.7515,
+      "loss": 0.8815,
       "step": 22993
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3739071524537973e-06,
-      "loss": 0.7088,
+      "loss": 0.7892,
       "step": 22994
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3733113665176222e-06,
-      "loss": 0.7109,
+      "loss": 0.798,
       "step": 22995
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3727157002645074e-06,
-      "loss": 0.6805,
+      "loss": 0.7102,
       "step": 22996
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.372120153702724e-06,
-      "loss": 0.7657,
+      "loss": 0.795,
       "step": 22997
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3715247268405318e-06,
-      "loss": 0.6101,
+      "loss": 0.725,
       "step": 22998
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3709294196861922e-06,
-      "loss": 0.6648,
+      "loss": 0.7699,
       "step": 22999
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3703342322479618e-06,
-      "loss": 0.691,
+      "loss": 0.853,
       "step": 23000
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3697391645341019e-06,
-      "loss": 0.7982,
+      "loss": 0.8412,
       "step": 23001
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3691442165528667e-06,
-      "loss": 0.6856,
+      "loss": 0.7619,
       "step": 23002
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.368549388312509e-06,
-      "loss": 0.8326,
+      "loss": 0.9181,
       "step": 23003
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3679546798212807e-06,
-      "loss": 0.7777,
+      "loss": 0.8621,
       "step": 23004
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3673600910874386e-06,
-      "loss": 0.6413,
+      "loss": 0.81,
       "step": 23005
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3667656221192226e-06,
-      "loss": 0.7486,
+      "loss": 0.7935,
       "step": 23006
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3661712729248866e-06,
-      "loss": 0.7055,
+      "loss": 0.7818,
       "step": 23007
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3655770435126747e-06,
-      "loss": 0.7121,
+      "loss": 0.781,
       "step": 23008
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3649829338908306e-06,
-      "loss": 0.6158,
+      "loss": 0.792,
       "step": 23009
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.364388944067594e-06,
-      "loss": 0.7239,
+      "loss": 0.8,
       "step": 23010
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.363795074051213e-06,
-      "loss": 0.6249,
+      "loss": 0.7189,
       "step": 23011
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3632013238499175e-06,
-      "loss": 0.5628,
+      "loss": 0.8633,
       "step": 23012
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3626076934719534e-06,
-      "loss": 0.6891,
+      "loss": 0.7997,
       "step": 23013
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3620141829255506e-06,
-      "loss": 0.7062,
+      "loss": 0.8066,
       "step": 23014
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.361420792218947e-06,
-      "loss": 0.6514,
+      "loss": 0.6743,
       "step": 23015
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3608275213603729e-06,
-      "loss": 0.6671,
+      "loss": 0.7255,
       "step": 23016
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.360234370358059e-06,
-      "loss": 0.6612,
+      "loss": 0.7919,
       "step": 23017
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3596413392202358e-06,
-      "loss": 0.7285,
+      "loss": 0.8047,
       "step": 23018
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3590484279551275e-06,
-      "loss": 0.6141,
+      "loss": 0.7467,
       "step": 23019
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3584556365709644e-06,
-      "loss": 0.8495,
+      "loss": 0.8223,
       "step": 23020
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3578629650759678e-06,
-      "loss": 0.7974,
+      "loss": 0.8606,
       "step": 23021
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3572704134783621e-06,
-      "loss": 0.8347,
+      "loss": 0.8539,
       "step": 23022
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.356677981786364e-06,
-      "loss": 0.7995,
+      "loss": 0.8518,
       "step": 23023
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3560856700082003e-06,
-      "loss": 0.6366,
+      "loss": 0.7396,
       "step": 23024
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3554934781520801e-06,
-      "loss": 0.6873,
+      "loss": 0.8119,
       "step": 23025
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3549014062262245e-06,
-      "loss": 0.6174,
+      "loss": 0.7714,
       "step": 23026
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3543094542388469e-06,
-      "loss": 0.7282,
+      "loss": 0.8757,
       "step": 23027
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3537176221981575e-06,
-      "loss": 0.6399,
+      "loss": 0.7535,
       "step": 23028
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3531259101123674e-06,
-      "loss": 0.6121,
+      "loss": 0.6416,
       "step": 23029
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3525343179896911e-06,
-      "loss": 0.6941,
+      "loss": 0.7495,
       "step": 23030
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3519428458383276e-06,
-      "loss": 0.7985,
+      "loss": 0.8306,
       "step": 23031
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3513514936664896e-06,
-      "loss": 0.8477,
+      "loss": 0.8631,
       "step": 23032
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.350760261482379e-06,
-      "loss": 0.6056,
+      "loss": 0.7461,
       "step": 23033
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3501691492941983e-06,
-      "loss": 0.7393,
+      "loss": 0.8053,
       "step": 23034
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3495781571101473e-06,
-      "loss": 0.6282,
+      "loss": 0.7532,
       "step": 23035
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3489872849384233e-06,
-      "loss": 0.6884,
+      "loss": 0.7996,
       "step": 23036
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3483965327872318e-06,
-      "loss": 0.6511,
+      "loss": 0.7312,
       "step": 23037
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3478059006647593e-06,
-      "loss": 0.7645,
+      "loss": 0.8117,
       "step": 23038
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3472153885792061e-06,
-      "loss": 0.6583,
+      "loss": 0.7192,
       "step": 23039
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3466249965387623e-06,
-      "loss": 0.6397,
+      "loss": 0.7967,
       "step": 23040
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3460347245516193e-06,
-      "loss": 0.595,
+      "loss": 0.7429,
       "step": 23041
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3454445726259645e-06,
-      "loss": 0.6653,
+      "loss": 0.7291,
       "step": 23042
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.344854540769991e-06,
-      "loss": 0.7557,
+      "loss": 0.7696,
       "step": 23043
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.344264628991876e-06,
-      "loss": 0.7051,
+      "loss": 0.7836,
       "step": 23044
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.343674837299811e-06,
-      "loss": 0.6253,
+      "loss": 0.7261,
       "step": 23045
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.343085165701975e-06,
-      "loss": 0.7781,
+      "loss": 0.853,
       "step": 23046
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3424956142065503e-06,
-      "loss": 0.6624,
+      "loss": 0.7194,
       "step": 23047
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3419061828217139e-06,
-      "loss": 0.7373,
+      "loss": 0.819,
       "step": 23048
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.341316871555647e-06,
-      "loss": 0.8486,
+      "loss": 0.8767,
       "step": 23049
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3407276804165238e-06,
-      "loss": 0.7028,
+      "loss": 0.7914,
       "step": 23050
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3401386094125168e-06,
-      "loss": 0.5281,
+      "loss": 0.7038,
       "step": 23051
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3395496585518009e-06,
-      "loss": 0.7325,
+      "loss": 0.7727,
       "step": 23052
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.338960827842546e-06,
-      "loss": 0.7847,
+      "loss": 0.8503,
       "step": 23053
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3383721172929209e-06,
-      "loss": 0.7695,
+      "loss": 0.8054,
       "step": 23054
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3377835269110905e-06,
-      "loss": 0.7561,
+      "loss": 0.8317,
       "step": 23055
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3371950567052294e-06,
-      "loss": 0.5452,
+      "loss": 0.6784,
       "step": 23056
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3366067066834908e-06,
-      "loss": 0.7563,
+      "loss": 0.8303,
       "step": 23057
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3360184768540442e-06,
-      "loss": 0.7706,
+      "loss": 0.8434,
       "step": 23058
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3354303672250491e-06,
-      "loss": 0.7663,
+      "loss": 0.778,
       "step": 23059
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3348423778046648e-06,
-      "loss": 0.7359,
+      "loss": 0.7774,
       "step": 23060
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.334254508601045e-06,
-      "loss": 0.6535,
+      "loss": 0.8195,
       "step": 23061
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3336667596223541e-06,
-      "loss": 0.7865,
+      "loss": 0.804,
       "step": 23062
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3330791308767366e-06,
-      "loss": 0.7067,
+      "loss": 0.8079,
       "step": 23063
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3324916223723504e-06,
-      "loss": 0.7978,
+      "loss": 0.8168,
       "step": 23064
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3319042341173471e-06,
-      "loss": 0.7741,
+      "loss": 0.8267,
       "step": 23065
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3313169661198734e-06,
-      "loss": 0.6804,
+      "loss": 0.7911,
       "step": 23066
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3307298183880745e-06,
-      "loss": 0.6475,
+      "loss": 0.8046,
       "step": 23067
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3301427909301023e-06,
-      "loss": 0.7494,
+      "loss": 0.8018,
       "step": 23068
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3295558837540979e-06,
-      "loss": 0.6028,
+      "loss": 0.7402,
       "step": 23069
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3289690968682035e-06,
-      "loss": 0.6867,
+      "loss": 0.7614,
       "step": 23070
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3283824302805602e-06,
-      "loss": 0.7586,
+      "loss": 0.7863,
       "step": 23071
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3277958839993077e-06,
-      "loss": 0.5237,
+      "loss": 0.7043,
       "step": 23072
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3272094580325834e-06,
-      "loss": 0.5774,
+      "loss": 0.7032,
       "step": 23073
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3266231523885199e-06,
-      "loss": 0.7081,
+      "loss": 0.8156,
       "step": 23074
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3260369670752593e-06,
-      "loss": 0.7255,
+      "loss": 0.8553,
       "step": 23075
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3254509021009242e-06,
-      "loss": 0.6873,
+      "loss": 0.7986,
       "step": 23076
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3248649574736528e-06,
-      "loss": 0.7349,
+      "loss": 0.8389,
       "step": 23077
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3242791332015714e-06,
-      "loss": 0.7749,
+      "loss": 0.8654,
       "step": 23078
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3236934292928071e-06,
-      "loss": 0.7664,
+      "loss": 0.7732,
       "step": 23079
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3231078457554857e-06,
-      "loss": 0.7322,
+      "loss": 0.7986,
       "step": 23080
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3225223825977329e-06,
-      "loss": 0.7599,
+      "loss": 0.8364,
       "step": 23081
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.321937039827671e-06,
-      "loss": 0.5712,
+      "loss": 0.7198,
       "step": 23082
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3213518174534201e-06,
-      "loss": 0.6707,
+      "loss": 0.7982,
       "step": 23083
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3207667154830995e-06,
-      "loss": 0.6508,
+      "loss": 0.7573,
       "step": 23084
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3201817339248257e-06,
-      "loss": 0.8078,
+      "loss": 0.8851,
       "step": 23085
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3195968727867137e-06,
-      "loss": 0.6635,
+      "loss": 0.8018,
       "step": 23086
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3190121320768811e-06,
-      "loss": 0.7105,
+      "loss": 0.778,
       "step": 23087
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3184275118034394e-06,
-      "loss": 0.616,
+      "loss": 0.7719,
       "step": 23088
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3178430119744978e-06,
-      "loss": 0.6408,
+      "loss": 0.7017,
       "step": 23089
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.317258632598165e-06,
-      "loss": 0.7113,
+      "loss": 0.7393,
       "step": 23090
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3166743736825516e-06,
-      "loss": 0.748,
+      "loss": 0.7302,
       "step": 23091
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3160902352357596e-06,
-      "loss": 0.6779,
+      "loss": 0.7842,
       "step": 23092
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3155062172658927e-06,
-      "loss": 0.6227,
+      "loss": 0.7741,
       "step": 23093
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3149223197810578e-06,
-      "loss": 0.7698,
+      "loss": 0.8151,
       "step": 23094
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.314338542789354e-06,
-      "loss": 0.7021,
+      "loss": 0.798,
       "step": 23095
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.313754886298878e-06,
-      "loss": 0.6416,
+      "loss": 0.8139,
       "step": 23096
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3131713503177301e-06,
-      "loss": 0.7024,
+      "loss": 0.8589,
       "step": 23097
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3125879348540038e-06,
-      "loss": 0.7311,
+      "loss": 0.8314,
       "step": 23098
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3120046399157926e-06,
-      "loss": 0.6817,
+      "loss": 0.7981,
       "step": 23099
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3114214655111923e-06,
-      "loss": 0.7249,
+      "loss": 0.7978,
       "step": 23100
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.310838411648292e-06,
-      "loss": 0.8686,
+      "loss": 0.8926,
       "step": 23101
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3102554783351794e-06,
-      "loss": 0.8213,
+      "loss": 0.8842,
       "step": 23102
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3096726655799418e-06,
-      "loss": 0.5439,
+      "loss": 0.6906,
       "step": 23103
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3090899733906703e-06,
-      "loss": 0.8014,
+      "loss": 0.8863,
       "step": 23104
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3085074017754395e-06,
-      "loss": 0.5344,
+      "loss": 0.6365,
       "step": 23105
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3079249507423398e-06,
-      "loss": 0.7032,
+      "loss": 0.8112,
       "step": 23106
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3073426202994488e-06,
-      "loss": 0.6493,
+      "loss": 0.7473,
       "step": 23107
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3067604104548447e-06,
-      "loss": 0.6001,
+      "loss": 0.8478,
       "step": 23108
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3061783212166047e-06,
-      "loss": 0.6962,
+      "loss": 0.7699,
       "step": 23109
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3055963525928095e-06,
-      "loss": 0.7075,
+      "loss": 0.7836,
       "step": 23110
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3050145045915275e-06,
-      "loss": 0.6473,
+      "loss": 0.7211,
       "step": 23111
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.30443277722083e-06,
-      "loss": 0.7941,
+      "loss": 0.846,
       "step": 23112
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3038511704887935e-06,
-      "loss": 0.6542,
+      "loss": 0.7496,
       "step": 23113
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.303269684403482e-06,
-      "loss": 0.8321,
+      "loss": 0.8595,
       "step": 23114
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3026883189729656e-06,
-      "loss": 0.768,
+      "loss": 0.8796,
       "step": 23115
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3021070742053077e-06,
-      "loss": 0.8415,
+      "loss": 0.8721,
       "step": 23116
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3015259501085763e-06,
-      "loss": 0.767,
+      "loss": 0.84,
       "step": 23117
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3009449466908264e-06,
-      "loss": 0.7546,
+      "loss": 0.8574,
       "step": 23118
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3003640639601267e-06,
-      "loss": 0.6212,
+      "loss": 0.7389,
       "step": 23119
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2997833019245309e-06,
-      "loss": 0.8105,
+      "loss": 0.8763,
       "step": 23120
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2992026605920982e-06,
-      "loss": 0.7636,
+      "loss": 0.8834,
       "step": 23121
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.298622139970882e-06,
-      "loss": 0.8078,
+      "loss": 0.8478,
       "step": 23122
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2980417400689427e-06,
-      "loss": 0.6969,
+      "loss": 0.7568,
       "step": 23123
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2974614608943226e-06,
-      "loss": 0.7315,
+      "loss": 0.8377,
       "step": 23124
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2968813024550796e-06,
-      "loss": 0.6938,
+      "loss": 0.8176,
       "step": 23125
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.296301264759261e-06,
-      "loss": 0.7455,
+      "loss": 0.8146,
       "step": 23126
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2957213478149132e-06,
-      "loss": 0.7916,
+      "loss": 0.8948,
       "step": 23127
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2951415516300792e-06,
-      "loss": 0.6154,
+      "loss": 0.712,
       "step": 23128
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.294561876212811e-06,
-      "loss": 0.7451,
+      "loss": 0.7347,
       "step": 23129
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2939823215711422e-06,
-      "loss": 0.8441,
+      "loss": 0.9726,
       "step": 23130
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2934028877131145e-06,
-      "loss": 0.6609,
+      "loss": 0.7877,
       "step": 23131
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.292823574646771e-06,
-      "loss": 0.7089,
+      "loss": 0.8328,
       "step": 23132
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2922443823801466e-06,
-      "loss": 0.7087,
+      "loss": 0.8134,
       "step": 23133
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2916653109212773e-06,
-      "loss": 0.6849,
+      "loss": 0.7867,
       "step": 23134
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.291086360278193e-06,
-      "loss": 0.6916,
+      "loss": 0.845,
       "step": 23135
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2905075304589343e-06,
-      "loss": 0.6521,
+      "loss": 0.7864,
       "step": 23136
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2899288214715212e-06,
-      "loss": 0.7138,
+      "loss": 0.8279,
       "step": 23137
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2893502333239904e-06,
-      "loss": 0.8775,
+      "loss": 0.9301,
       "step": 23138
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2887717660243671e-06,
-      "loss": 0.7807,
+      "loss": 0.789,
       "step": 23139
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2881934195806745e-06,
-      "loss": 0.6371,
+      "loss": 0.7311,
       "step": 23140
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2876151940009362e-06,
-      "loss": 0.662,
+      "loss": 0.7455,
       "step": 23141
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2870370892931804e-06,
-      "loss": 0.6202,
+      "loss": 0.7076,
       "step": 23142
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2864591054654184e-06,
-      "loss": 0.8297,
+      "loss": 0.8436,
       "step": 23143
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2858812425256762e-06,
-      "loss": 0.6949,
+      "loss": 0.8312,
       "step": 23144
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.285303500481968e-06,
-      "loss": 0.7046,
+      "loss": 0.7955,
       "step": 23145
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.284725879342308e-06,
-      "loss": 0.6229,
+      "loss": 0.7633,
       "step": 23146
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2841483791147102e-06,
-      "loss": 0.6728,
+      "loss": 0.754,
       "step": 23147
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2835709998071888e-06,
-      "loss": 0.7954,
+      "loss": 0.8775,
       "step": 23148
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.282993741427755e-06,
-      "loss": 0.7041,
+      "loss": 0.7423,
       "step": 23149
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.282416603984411e-06,
-      "loss": 0.6816,
+      "loss": 0.7129,
       "step": 23150
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2818395874851708e-06,
-      "loss": 0.6983,
+      "loss": 0.7807,
       "step": 23151
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2812626919380356e-06,
-      "loss": 0.7143,
+      "loss": 0.7717,
       "step": 23152
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.280685917351011e-06,
-      "loss": 0.7089,
+      "loss": 0.8481,
       "step": 23153
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2801092637320966e-06,
-      "loss": 0.7408,
+      "loss": 0.9068,
       "step": 23154
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.279532731089298e-06,
-      "loss": 0.7872,
+      "loss": 0.8057,
       "step": 23155
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2789563194306053e-06,
-      "loss": 0.7302,
+      "loss": 0.772,
       "step": 23156
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2783800287640225e-06,
-      "loss": 0.7005,
+      "loss": 0.8268,
       "step": 23157
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2778038590975428e-06,
-      "loss": 0.6508,
+      "loss": 0.7693,
       "step": 23158
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.277227810439159e-06,
-      "loss": 0.7358,
+      "loss": 0.734,
       "step": 23159
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.276651882796862e-06,
-      "loss": 0.7873,
+      "loss": 0.7816,
       "step": 23160
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2760760761786462e-06,
-      "loss": 0.6915,
+      "loss": 0.7725,
       "step": 23161
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2755003905924945e-06,
-      "loss": 0.7241,
+      "loss": 0.7891,
       "step": 23162
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2749248260463976e-06,
-      "loss": 0.721,
+      "loss": 0.7996,
       "step": 23163
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.27434938254834e-06,
-      "loss": 0.7067,
+      "loss": 0.8159,
       "step": 23164
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2737740601063042e-06,
-      "loss": 0.6104,
+      "loss": 0.7163,
       "step": 23165
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2731988587282707e-06,
-      "loss": 0.7315,
+      "loss": 0.7699,
       "step": 23166
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2726237784222228e-06,
-      "loss": 0.6951,
+      "loss": 0.8058,
       "step": 23167
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.27204881919614e-06,
-      "loss": 0.7686,
+      "loss": 0.8479,
       "step": 23168
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2714739810579923e-06,
-      "loss": 0.5765,
+      "loss": 0.673,
       "step": 23169
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2708992640157602e-06,
-      "loss": 0.7665,
+      "loss": 0.823,
       "step": 23170
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2703246680774174e-06,
-      "loss": 0.8393,
+      "loss": 0.8692,
       "step": 23171
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2697501932509326e-06,
-      "loss": 0.7062,
+      "loss": 0.8666,
       "step": 23172
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2691758395442754e-06,
-      "loss": 0.7053,
+      "loss": 0.7952,
       "step": 23173
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2686016069654216e-06,
-      "loss": 0.7911,
+      "loss": 0.8031,
       "step": 23174
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.268027495522327e-06,
-      "loss": 0.7904,
+      "loss": 0.8141,
       "step": 23175
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2674535052229642e-06,
-      "loss": 0.7329,
+      "loss": 0.7745,
       "step": 23176
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2668796360752933e-06,
-      "loss": 0.8727,
+      "loss": 0.8627,
       "step": 23177
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.266305888087278e-06,
-      "loss": 0.7452,
+      "loss": 0.8952,
       "step": 23178
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2657322612668743e-06,
-      "loss": 0.7647,
+      "loss": 0.7688,
       "step": 23179
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2651587556220447e-06,
-      "loss": 0.7489,
+      "loss": 0.8314,
       "step": 23180
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2645853711607447e-06,
-      "loss": 0.7698,
+      "loss": 0.787,
       "step": 23181
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2640121078909294e-06,
-      "loss": 0.6426,
+      "loss": 0.8018,
       "step": 23182
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2634389658205503e-06,
-      "loss": 0.7285,
+      "loss": 0.7384,
       "step": 23183
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.262865944957561e-06,
-      "loss": 0.7843,
+      "loss": 0.9238,
       "step": 23184
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.262293045309908e-06,
-      "loss": 0.6057,
+      "loss": 0.7286,
       "step": 23185
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2617202668855433e-06,
-      "loss": 0.8353,
+      "loss": 0.8258,
       "step": 23186
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2611476096924146e-06,
-      "loss": 0.6665,
+      "loss": 0.7592,
       "step": 23187
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2605750737384591e-06,
-      "loss": 0.8212,
+      "loss": 0.8681,
       "step": 23188
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2600026590316282e-06,
-      "loss": 0.7738,
+      "loss": 0.7604,
       "step": 23189
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.259430365579859e-06,
-      "loss": 0.6795,
+      "loss": 0.7453,
       "step": 23190
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2588581933910927e-06,
-      "loss": 0.7715,
+      "loss": 0.8272,
       "step": 23191
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.258286142473264e-06,
-      "loss": 0.6977,
+      "loss": 0.7662,
       "step": 23192
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2577142128343155e-06,
-      "loss": 0.6368,
+      "loss": 0.7139,
       "step": 23193
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2571424044821778e-06,
-      "loss": 0.8304,
+      "loss": 0.9172,
       "step": 23194
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2565707174247855e-06,
-      "loss": 0.7808,
+      "loss": 0.8533,
       "step": 23195
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2559991516700676e-06,
-      "loss": 0.6246,
+      "loss": 0.77,
       "step": 23196
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.255427707225957e-06,
-      "loss": 0.7196,
+      "loss": 0.834,
       "step": 23197
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.254856384100377e-06,
-      "loss": 0.7302,
+      "loss": 0.8082,
       "step": 23198
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2542851823012592e-06,
-      "loss": 0.7151,
+      "loss": 0.7892,
       "step": 23199
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2537141018365261e-06,
-      "loss": 0.8492,
+      "loss": 0.8872,
       "step": 23200
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2531431427141006e-06,
-      "loss": 0.7224,
+      "loss": 0.8217,
       "step": 23201
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2525723049419015e-06,
-      "loss": 0.7929,
+      "loss": 0.8757,
       "step": 23202
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2520015885278547e-06,
-      "loss": 0.7624,
+      "loss": 0.8471,
       "step": 23203
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2514309934798707e-06,
-      "loss": 0.6772,
+      "loss": 0.7612,
       "step": 23204
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.250860519805871e-06,
-      "loss": 0.7073,
+      "loss": 0.7828,
       "step": 23205
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.250290167513769e-06,
-      "loss": 0.6809,
+      "loss": 0.7982,
       "step": 23206
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2497199366114743e-06,
-      "loss": 0.7046,
+      "loss": 0.7794,
       "step": 23207
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2491498271069013e-06,
-      "loss": 0.7677,
+      "loss": 0.7699,
       "step": 23208
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2485798390079596e-06,
-      "loss": 0.8697,
+      "loss": 0.8682,
       "step": 23209
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2480099723225547e-06,
-      "loss": 0.687,
+      "loss": 0.7526,
       "step": 23210
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.247440227058593e-06,
-      "loss": 0.6974,
+      "loss": 0.7993,
       "step": 23211
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2468706032239808e-06,
-      "loss": 0.723,
+      "loss": 0.8337,
       "step": 23212
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2463011008266214e-06,
-      "loss": 0.6357,
+      "loss": 0.7621,
       "step": 23213
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2457317198744135e-06,
-      "loss": 0.9207,
+      "loss": 0.9493,
       "step": 23214
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2451624603752554e-06,
-      "loss": 0.7715,
+      "loss": 0.8067,
       "step": 23215
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.244593322337051e-06,
-      "loss": 0.7982,
+      "loss": 0.8236,
       "step": 23216
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.244024305767687e-06,
-      "loss": 0.5967,
+      "loss": 0.7032,
       "step": 23217
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2434554106750652e-06,
-      "loss": 0.7142,
+      "loss": 0.7172,
       "step": 23218
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2428866370670767e-06,
-      "loss": 0.7537,
+      "loss": 0.8468,
       "step": 23219
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.24231798495161e-06,
-      "loss": 0.5897,
+      "loss": 0.6239,
       "step": 23220
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2417494543365539e-06,
-      "loss": 0.6879,
+      "loss": 0.7836,
       "step": 23221
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2411810452298024e-06,
-      "loss": 0.6528,
+      "loss": 0.7677,
       "step": 23222
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2406127576392323e-06,
-      "loss": 0.724,
+      "loss": 0.7545,
       "step": 23223
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2400445915727343e-06,
-      "loss": 0.6063,
+      "loss": 0.7104,
       "step": 23224
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2394765470381886e-06,
-      "loss": 0.6364,
+      "loss": 0.7134,
       "step": 23225
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2389086240434766e-06,
-      "loss": 0.792,
+      "loss": 0.8191,
       "step": 23226
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2383408225964778e-06,
-      "loss": 0.6109,
+      "loss": 0.7857,
       "step": 23227
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2377731427050677e-06,
-      "loss": 0.718,
+      "loss": 0.7216,
       "step": 23228
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2372055843771235e-06,
-      "loss": 0.7217,
+      "loss": 0.764,
       "step": 23229
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.236638147620518e-06,
-      "loss": 0.6479,
+      "loss": 0.8231,
       "step": 23230
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.236070832443126e-06,
-      "loss": 0.8271,
+      "loss": 0.7611,
       "step": 23231
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2355036388528164e-06,
-      "loss": 0.7306,
+      "loss": 0.7969,
       "step": 23232
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2349365668574588e-06,
-      "loss": 0.6681,
+      "loss": 0.6751,
       "step": 23233
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.234369616464919e-06,
-      "loss": 0.8306,
+      "loss": 0.8927,
       "step": 23234
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.233802787683067e-06,
-      "loss": 0.7107,
+      "loss": 0.7821,
       "step": 23235
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2332360805197597e-06,
-      "loss": 0.7122,
+      "loss": 0.8266,
       "step": 23236
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2326694949828665e-06,
-      "loss": 0.7595,
+      "loss": 0.7749,
       "step": 23237
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.232103031080244e-06,
-      "loss": 0.5991,
+      "loss": 0.6777,
       "step": 23238
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2315366888197522e-06,
-      "loss": 0.7248,
+      "loss": 0.7335,
       "step": 23239
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2309704682092461e-06,
-      "loss": 0.772,
+      "loss": 0.7443,
       "step": 23240
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.230404369256587e-06,
-      "loss": 0.7989,
+      "loss": 0.8531,
       "step": 23241
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2298383919696199e-06,
-      "loss": 0.6724,
+      "loss": 0.7848,
       "step": 23242
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2292725363562052e-06,
-      "loss": 0.5623,
+      "loss": 0.6512,
       "step": 23243
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.22870680242419e-06,
-      "loss": 0.7448,
+      "loss": 0.8506,
       "step": 23244
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2281411901814222e-06,
-      "loss": 0.6704,
+      "loss": 0.7115,
       "step": 23245
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2275756996357503e-06,
-      "loss": 0.7173,
+      "loss": 0.7754,
       "step": 23246
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2270103307950165e-06,
-      "loss": 0.7342,
+      "loss": 0.845,
       "step": 23247
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2264450836670727e-06,
-      "loss": 0.7023,
+      "loss": 0.7648,
       "step": 23248
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2258799582597502e-06,
-      "loss": 0.6548,
+      "loss": 0.768,
       "step": 23249
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2253149545808962e-06,
-      "loss": 0.8647,
+      "loss": 0.8896,
       "step": 23250
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2247500726383476e-06,
-      "loss": 0.8264,
+      "loss": 0.9129,
       "step": 23251
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2241853124399417e-06,
-      "loss": 0.7435,
+      "loss": 0.793,
       "step": 23252
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2236206739935107e-06,
-      "loss": 0.7088,
+      "loss": 0.8032,
       "step": 23253
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2230561573068956e-06,
-      "loss": 0.8202,
+      "loss": 0.8631,
       "step": 23254
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2224917623879196e-06,
-      "loss": 0.7652,
+      "loss": 0.7463,
       "step": 23255
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2219274892444177e-06,
-      "loss": 0.7291,
+      "loss": 0.7722,
       "step": 23256
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2213633378842183e-06,
-      "loss": 0.6651,
+      "loss": 0.7309,
       "step": 23257
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2207993083151471e-06,
-      "loss": 0.7866,
+      "loss": 0.8184,
       "step": 23258
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2202354005450268e-06,
-      "loss": 0.7163,
+      "loss": 0.7665,
       "step": 23259
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2196716145816866e-06,
-      "loss": 0.7681,
+      "loss": 0.7924,
       "step": 23260
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2191079504329439e-06,
-      "loss": 0.6956,
+      "loss": 0.8973,
       "step": 23261
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2185444081066211e-06,
-      "loss": 0.7128,
+      "loss": 0.7287,
       "step": 23262
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2179809876105352e-06,
-      "loss": 0.6737,
+      "loss": 0.8431,
       "step": 23263
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2174176889525025e-06,
-      "loss": 0.7985,
+      "loss": 0.8952,
       "step": 23264
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2168545121403386e-06,
-      "loss": 0.6123,
+      "loss": 0.6993,
       "step": 23265
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2162914571818552e-06,
-      "loss": 0.7432,
+      "loss": 0.8058,
       "step": 23266
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2157285240848682e-06,
-      "loss": 0.675,
+      "loss": 0.7778,
       "step": 23267
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2151657128571814e-06,
-      "loss": 0.7117,
+      "loss": 0.802,
       "step": 23268
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2146030235066085e-06,
-      "loss": 0.7016,
+      "loss": 0.8451,
       "step": 23269
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2140404560409536e-06,
-      "loss": 0.7544,
+      "loss": 0.8368,
       "step": 23270
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.213478010468022e-06,
-      "loss": 0.7205,
+      "loss": 0.8446,
       "step": 23271
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2129156867956149e-06,
-      "loss": 0.6786,
+      "loss": 0.7271,
       "step": 23272
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2123534850315388e-06,
-      "loss": 0.6907,
+      "loss": 0.7904,
       "step": 23273
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2117914051835855e-06,
-      "loss": 0.6994,
+      "loss": 0.7567,
       "step": 23274
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2112294472595598e-06,
-      "loss": 0.8376,
+      "loss": 0.9264,
       "step": 23275
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2106676112672567e-06,
-      "loss": 0.7358,
+      "loss": 0.8512,
       "step": 23276
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2101058972144698e-06,
-      "loss": 0.873,
+      "loss": 0.9099,
       "step": 23277
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2095443051089906e-06,
-      "loss": 0.6766,
+      "loss": 0.7217,
       "step": 23278
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.208982834958613e-06,
-      "loss": 0.7209,
+      "loss": 0.8358,
       "step": 23279
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2084214867711263e-06,
-      "loss": 0.7793,
+      "loss": 0.8273,
       "step": 23280
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2078602605543189e-06,
-      "loss": 0.6425,
+      "loss": 0.7494,
       "step": 23281
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2072991563159742e-06,
-      "loss": 0.6488,
+      "loss": 0.7833,
       "step": 23282
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2067381740638794e-06,
-      "loss": 0.6886,
+      "loss": 0.8134,
       "step": 23283
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2061773138058176e-06,
-      "loss": 0.689,
+      "loss": 0.7806,
       "step": 23284
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2056165755495652e-06,
-      "loss": 0.6172,
+      "loss": 0.7578,
       "step": 23285
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.20505595930291e-06,
-      "loss": 0.7714,
+      "loss": 0.8097,
       "step": 23286
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2044954650736197e-06,
-      "loss": 0.7322,
+      "loss": 0.7761,
       "step": 23287
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2039350928694781e-06,
-      "loss": 0.6938,
+      "loss": 0.7796,
       "step": 23288
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2033748426982584e-06,
-      "loss": 0.7318,
+      "loss": 0.794,
       "step": 23289
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2028147145677306e-06,
-      "loss": 0.8304,
+      "loss": 0.823,
       "step": 23290
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2022547084856662e-06,
-      "loss": 0.6569,
+      "loss": 0.7656,
       "step": 23291
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.201694824459836e-06,
-      "loss": 0.7666,
+      "loss": 0.8215,
       "step": 23292
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2011350624980079e-06,
-      "loss": 0.7754,
+      "loss": 0.7988,
       "step": 23293
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2005754226079469e-06,
-      "loss": 0.6635,
+      "loss": 0.8211,
       "step": 23294
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.200015904797417e-06,
-      "loss": 0.7437,
+      "loss": 0.8954,
       "step": 23295
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1994565090741806e-06,
-      "loss": 0.6877,
+      "loss": 0.7704,
       "step": 23296
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.198897235445997e-06,
-      "loss": 0.7551,
+      "loss": 0.7839,
       "step": 23297
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1983380839206283e-06,
-      "loss": 0.6637,
+      "loss": 0.7161,
       "step": 23298
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.197779054505832e-06,
-      "loss": 0.6493,
+      "loss": 0.7926,
       "step": 23299
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1972201472093625e-06,
-      "loss": 0.7225,
+      "loss": 0.7918,
       "step": 23300
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1966613620389734e-06,
-      "loss": 0.6773,
+      "loss": 0.8236,
       "step": 23301
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1961026990024172e-06,
-      "loss": 0.6713,
+      "loss": 0.6801,
       "step": 23302
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1955441581074456e-06,
-      "loss": 0.7267,
+      "loss": 0.8022,
       "step": 23303
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1949857393618047e-06,
-      "loss": 0.7011,
+      "loss": 0.8287,
       "step": 23304
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1944274427732483e-06,
-      "loss": 0.7323,
+      "loss": 0.8126,
       "step": 23305
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1938692683495133e-06,
-      "loss": 0.8544,
+      "loss": 0.877,
       "step": 23306
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1933112160983506e-06,
-      "loss": 0.5754,
+      "loss": 0.6851,
       "step": 23307
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1927532860274993e-06,
-      "loss": 0.7582,
+      "loss": 0.7974,
       "step": 23308
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1921954781446998e-06,
-      "loss": 0.7295,
+      "loss": 0.7945,
       "step": 23309
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1916377924576905e-06,
-      "loss": 0.7081,
+      "loss": 0.7361,
       "step": 23310
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1910802289742107e-06,
-      "loss": 0.6709,
+      "loss": 0.8006,
       "step": 23311
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1905227877019943e-06,
-      "loss": 0.8618,
+      "loss": 0.9332,
       "step": 23312
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1899654686487761e-06,
-      "loss": 0.8347,
+      "loss": 0.8811,
       "step": 23313
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1894082718222844e-06,
-      "loss": 0.7636,
+      "loss": 0.8422,
       "step": 23314
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1888511972302562e-06,
-      "loss": 0.7609,
+      "loss": 0.7921,
       "step": 23315
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1882942448804124e-06,
-      "loss": 0.6203,
+      "loss": 0.7621,
       "step": 23316
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1877374147804854e-06,
-      "loss": 0.7222,
+      "loss": 0.8135,
       "step": 23317
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1871807069382e-06,
-      "loss": 0.6098,
+      "loss": 0.7444,
       "step": 23318
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1866241213612783e-06,
-      "loss": 0.6721,
+      "loss": 0.747,
       "step": 23319
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1860676580574415e-06,
-      "loss": 0.7524,
+      "loss": 0.9007,
       "step": 23320
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1855113170344113e-06,
-      "loss": 0.7063,
+      "loss": 0.8175,
       "step": 23321
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1849550982999048e-06,
-      "loss": 0.8158,
+      "loss": 0.8325,
       "step": 23322
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.184399001861638e-06,
-      "loss": 0.6987,
+      "loss": 0.7643,
       "step": 23323
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1838430277273293e-06,
-      "loss": 0.7806,
+      "loss": 0.8094,
       "step": 23324
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1832871759046893e-06,
-      "loss": 0.7052,
+      "loss": 0.7906,
       "step": 23325
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1827314464014316e-06,
-      "loss": 0.6332,
+      "loss": 0.7617,
       "step": 23326
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1821758392252646e-06,
-      "loss": 0.7864,
+      "loss": 0.8248,
       "step": 23327
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1816203543838967e-06,
-      "loss": 0.5274,
+      "loss": 0.7582,
       "step": 23328
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1810649918850325e-06,
-      "loss": 0.8142,
+      "loss": 0.9232,
       "step": 23329
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1805097517363817e-06,
-      "loss": 0.701,
+      "loss": 0.756,
       "step": 23330
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1799546339456447e-06,
-      "loss": 0.8236,
+      "loss": 0.8836,
       "step": 23331
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1793996385205242e-06,
-      "loss": 0.6823,
+      "loss": 0.7623,
       "step": 23332
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1788447654687151e-06,
-      "loss": 0.7197,
+      "loss": 0.8435,
       "step": 23333
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1782900147979247e-06,
-      "loss": 0.7284,
+      "loss": 0.8097,
       "step": 23334
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1777353865158393e-06,
-      "loss": 0.7163,
+      "loss": 0.7999,
       "step": 23335
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.17718088063016e-06,
-      "loss": 0.769,
+      "loss": 0.8689,
       "step": 23336
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1766264971485787e-06,
-      "loss": 0.657,
+      "loss": 0.6994,
       "step": 23337
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1760722360787856e-06,
-      "loss": 0.7337,
+      "loss": 0.79,
       "step": 23338
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1755180974284686e-06,
-      "loss": 0.8399,
+      "loss": 0.9016,
       "step": 23339
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.174964081205322e-06,
-      "loss": 0.8112,
+      "loss": 0.8811,
       "step": 23340
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1744101874170256e-06,
-      "loss": 0.6795,
+      "loss": 0.7402,
       "step": 23341
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.173856416071264e-06,
-      "loss": 0.7458,
+      "loss": 0.8172,
       "step": 23342
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1733027671757236e-06,
-      "loss": 0.8193,
+      "loss": 0.8557,
       "step": 23343
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1727492407380847e-06,
-      "loss": 0.7887,
+      "loss": 0.8722,
       "step": 23344
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1721958367660247e-06,
-      "loss": 0.6226,
+      "loss": 0.788,
       "step": 23345
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1716425552672205e-06,
-      "loss": 0.6833,
+      "loss": 0.7825,
       "step": 23346
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1710893962493552e-06,
-      "loss": 0.76,
+      "loss": 0.7779,
       "step": 23347
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1705363597200935e-06,
-      "loss": 0.6789,
+      "loss": 0.8172,
       "step": 23348
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.169983445687114e-06,
-      "loss": 0.7576,
+      "loss": 0.8293,
       "step": 23349
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1694306541580868e-06,
-      "loss": 0.7178,
+      "loss": 0.736,
       "step": 23350
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1688779851406795e-06,
-      "loss": 0.6289,
+      "loss": 0.79,
       "step": 23351
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.168325438642559e-06,
-      "loss": 0.775,
+      "loss": 0.8312,
       "step": 23352
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1677730146713962e-06,
-      "loss": 0.6732,
+      "loss": 0.7718,
       "step": 23353
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1672207132348478e-06,
-      "loss": 0.6872,
+      "loss": 0.8072,
       "step": 23354
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1666685343405816e-06,
-      "loss": 0.7333,
+      "loss": 0.8085,
       "step": 23355
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1661164779962576e-06,
-      "loss": 0.7609,
+      "loss": 0.798,
       "step": 23356
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.165564544209532e-06,
-      "loss": 0.6605,
+      "loss": 0.781,
       "step": 23357
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1650127329880634e-06,
-      "loss": 0.6205,
+      "loss": 0.7756,
       "step": 23358
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1644610443395098e-06,
-      "loss": 0.7483,
+      "loss": 0.8418,
       "step": 23359
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1639094782715243e-06,
-      "loss": 0.753,
+      "loss": 0.8273,
       "step": 23360
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1633580347917551e-06,
-      "loss": 0.6776,
+      "loss": 0.7889,
       "step": 23361
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.162806713907857e-06,
-      "loss": 0.7015,
+      "loss": 0.7778,
       "step": 23362
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1622555156274773e-06,
-      "loss": 0.6713,
+      "loss": 0.8009,
       "step": 23363
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1617044399582633e-06,
-      "loss": 0.6125,
+      "loss": 0.8058,
       "step": 23364
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1611534869078588e-06,
-      "loss": 0.7509,
+      "loss": 0.7856,
       "step": 23365
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1606026564839123e-06,
-      "loss": 0.7491,
+      "loss": 0.8016,
       "step": 23366
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1600519486940587e-06,
-      "loss": 0.7576,
+      "loss": 0.8919,
       "step": 23367
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1595013635459452e-06,
-      "loss": 0.7448,
+      "loss": 0.8451,
       "step": 23368
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1589509010472055e-06,
-      "loss": 0.6655,
+      "loss": 0.8044,
       "step": 23369
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1584005612054794e-06,
-      "loss": 0.7942,
+      "loss": 0.7853,
       "step": 23370
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1578503440283983e-06,
-      "loss": 0.5331,
+      "loss": 0.7543,
       "step": 23371
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.157300249523603e-06,
-      "loss": 0.7089,
+      "loss": 0.7972,
       "step": 23372
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.156750277698715e-06,
-      "loss": 0.7431,
+      "loss": 0.8286,
       "step": 23373
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1562004285613737e-06,
-      "loss": 0.6149,
+      "loss": 0.6703,
       "step": 23374
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.155650702119203e-06,
-      "loss": 0.6383,
+      "loss": 0.7567,
       "step": 23375
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1551010983798305e-06,
-      "loss": 0.4952,
+      "loss": 0.6679,
       "step": 23376
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1545516173508786e-06,
-      "loss": 0.7224,
+      "loss": 0.7596,
       "step": 23377
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1540022590399759e-06,
-      "loss": 0.8122,
+      "loss": 0.902,
       "step": 23378
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1534530234547415e-06,
-      "loss": 0.8019,
+      "loss": 0.834,
       "step": 23379
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1529039106027916e-06,
-      "loss": 0.6427,
+      "loss": 0.7321,
       "step": 23380
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1523549204917506e-06,
-      "loss": 0.7674,
+      "loss": 0.7902,
       "step": 23381
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1518060531292307e-06,
-      "loss": 0.8248,
+      "loss": 0.8488,
       "step": 23382
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1512573085228474e-06,
-      "loss": 0.6272,
+      "loss": 0.7539,
       "step": 23383
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.150708686680213e-06,
-      "loss": 0.6649,
+      "loss": 0.7701,
       "step": 23384
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1501601876089442e-06,
-      "loss": 0.5929,
+      "loss": 0.7845,
       "step": 23385
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.149611811316642e-06,
-      "loss": 0.6438,
+      "loss": 0.7787,
       "step": 23386
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1490635578109222e-06,
-      "loss": 0.6661,
+      "loss": 0.8531,
       "step": 23387
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1485154270993858e-06,
-      "loss": 0.6218,
+      "loss": 0.7441,
       "step": 23388
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1479674191896407e-06,
-      "loss": 0.6801,
+      "loss": 0.7743,
       "step": 23389
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1474195340892857e-06,
-      "loss": 0.7414,
+      "loss": 0.758,
       "step": 23390
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1468717718059264e-06,
-      "loss": 0.6787,
+      "loss": 0.7779,
       "step": 23391
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1463241323471596e-06,
-      "loss": 0.727,
+      "loss": 0.7074,
       "step": 23392
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1457766157205852e-06,
-      "loss": 0.885,
+      "loss": 0.9022,
       "step": 23393
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1452292219337967e-06,
-      "loss": 0.8558,
+      "loss": 0.7708,
       "step": 23394
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1446819509943906e-06,
-      "loss": 0.7145,
+      "loss": 0.7732,
       "step": 23395
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1441348029099552e-06,
-      "loss": 0.7239,
+      "loss": 0.7557,
       "step": 23396
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1435877776880865e-06,
-      "loss": 0.6075,
+      "loss": 0.687,
       "step": 23397
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1430408753363753e-06,
-      "loss": 0.7094,
+      "loss": 0.8078,
       "step": 23398
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1424940958623998e-06,
-      "loss": 0.7959,
+      "loss": 0.8254,
       "step": 23399
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.141947439273755e-06,
-      "loss": 0.7339,
+      "loss": 0.8022,
       "step": 23400
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1414009055780206e-06,
-      "loss": 0.7229,
+      "loss": 0.7674,
       "step": 23401
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1408544947827805e-06,
-      "loss": 0.606,
+      "loss": 0.6995,
       "step": 23402
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1403082068956118e-06,
-      "loss": 0.7422,
+      "loss": 0.8483,
       "step": 23403
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1397620419240995e-06,
-      "loss": 0.7562,
+      "loss": 0.7822,
       "step": 23404
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1392159998758178e-06,
-      "loss": 0.6985,
+      "loss": 0.7416,
       "step": 23405
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1386700807583417e-06,
-      "loss": 0.6905,
+      "loss": 0.7564,
       "step": 23406
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1381242845792462e-06,
-      "loss": 0.6717,
+      "loss": 0.7218,
       "step": 23407
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1375786113461029e-06,
-      "loss": 0.7655,
+      "loss": 0.907,
       "step": 23408
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1370330610664792e-06,
-      "loss": 0.729,
+      "loss": 0.8524,
       "step": 23409
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1364876337479502e-06,
-      "loss": 0.7043,
+      "loss": 0.7768,
       "step": 23410
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1359423293980787e-06,
-      "loss": 0.6409,
+      "loss": 0.7187,
       "step": 23411
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1353971480244318e-06,
-      "loss": 0.7634,
+      "loss": 0.765,
       "step": 23412
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1348520896345694e-06,
-      "loss": 0.7254,
+      "loss": 0.8806,
       "step": 23413
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1343071542360607e-06,
-      "loss": 0.7316,
+      "loss": 0.7822,
       "step": 23414
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1337623418364575e-06,
-      "loss": 0.7402,
+      "loss": 0.8243,
       "step": 23415
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1332176524433236e-06,
-      "loss": 0.757,
+      "loss": 0.8582,
       "step": 23416
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1326730860642178e-06,
-      "loss": 0.8122,
+      "loss": 0.8762,
       "step": 23417
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1321286427066869e-06,
-      "loss": 0.7078,
+      "loss": 0.8161,
       "step": 23418
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1315843223782908e-06,
-      "loss": 0.5705,
+      "loss": 0.7034,
       "step": 23419
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.13104012508658e-06,
-      "loss": 0.6949,
+      "loss": 0.7708,
       "step": 23420
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1304960508391037e-06,
-      "loss": 0.6515,
+      "loss": 0.7799,
       "step": 23421
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.12995209964341e-06,
-      "loss": 0.7286,
+      "loss": 0.8177,
       "step": 23422
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1294082715070464e-06,
-      "loss": 0.7603,
+      "loss": 0.8673,
       "step": 23423
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1288645664375574e-06,
-      "loss": 0.7806,
+      "loss": 0.8626,
       "step": 23424
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.128320984442487e-06,
-      "loss": 0.6236,
+      "loss": 0.7851,
       "step": 23425
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1277775255293733e-06,
-      "loss": 0.8365,
+      "loss": 0.8874,
       "step": 23426
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1272341897057627e-06,
-      "loss": 0.7737,
+      "loss": 0.8548,
       "step": 23427
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1266909769791845e-06,
-      "loss": 0.7933,
+      "loss": 0.8525,
       "step": 23428
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1261478873571807e-06,
-      "loss": 0.6698,
+      "loss": 0.7407,
       "step": 23429
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1256049208472853e-06,
-      "loss": 0.5846,
+      "loss": 0.7666,
       "step": 23430
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1250620774570308e-06,
-      "loss": 0.6609,
+      "loss": 0.7531,
       "step": 23431
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1245193571939462e-06,
-      "loss": 0.6717,
+      "loss": 0.8205,
       "step": 23432
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.123976760065566e-06,
-      "loss": 0.7997,
+      "loss": 0.8753,
       "step": 23433
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1234342860794112e-06,
-      "loss": 0.8769,
+      "loss": 0.8657,
       "step": 23434
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1228919352430135e-06,
-      "loss": 0.6608,
+      "loss": 0.7902,
       "step": 23435
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1223497075638957e-06,
-      "loss": 0.6356,
+      "loss": 0.7267,
       "step": 23436
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1218076030495795e-06,
-      "loss": 0.6646,
+      "loss": 0.7378,
       "step": 23437
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1212656217075868e-06,
-      "loss": 0.9429,
+      "loss": 0.9179,
       "step": 23438
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1207237635454348e-06,
-      "loss": 0.707,
+      "loss": 0.8242,
       "step": 23439
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1201820285706443e-06,
-      "loss": 0.6715,
+      "loss": 0.7019,
       "step": 23440
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.119640416790726e-06,
-      "loss": 0.8134,
+      "loss": 0.8815,
       "step": 23441
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1190989282132002e-06,
-      "loss": 0.7898,
+      "loss": 0.8105,
       "step": 23442
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1185575628455769e-06,
-      "loss": 0.6505,
+      "loss": 0.6574,
       "step": 23443
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1180163206953654e-06,
-      "loss": 0.6291,
+      "loss": 0.6981,
       "step": 23444
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1174752017700741e-06,
-      "loss": 0.777,
+      "loss": 0.8147,
       "step": 23445
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1169342060772147e-06,
-      "loss": 0.664,
+      "loss": 0.722,
       "step": 23446
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.116393333624287e-06,
-      "loss": 0.7408,
+      "loss": 0.7825,
       "step": 23447
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1158525844187984e-06,
-      "loss": 0.7918,
+      "loss": 0.8433,
       "step": 23448
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1153119584682514e-06,
-      "loss": 0.6236,
+      "loss": 0.8765,
       "step": 23449
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1147714557801449e-06,
-      "loss": 0.7168,
+      "loss": 0.8178,
       "step": 23450
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1142310763619756e-06,
-      "loss": 0.8289,
+      "loss": 0.8826,
       "step": 23451
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.113690820221247e-06,
-      "loss": 0.6217,
+      "loss": 0.7236,
       "step": 23452
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1131506873654464e-06,
-      "loss": 0.5535,
+      "loss": 0.7161,
       "step": 23453
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1126106778020729e-06,
-      "loss": 0.6642,
+      "loss": 0.7632,
       "step": 23454
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1120707915386176e-06,
-      "loss": 0.6492,
+      "loss": 0.7871,
       "step": 23455
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1115310285825687e-06,
-      "loss": 0.7327,
+      "loss": 0.809,
       "step": 23456
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1109913889414158e-06,
-      "loss": 0.7896,
+      "loss": 0.8258,
       "step": 23457
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1104518726226443e-06,
-      "loss": 0.9321,
+      "loss": 0.9293,
       "step": 23458
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1099124796337435e-06,
-      "loss": 0.7855,
+      "loss": 0.8496,
       "step": 23459
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1093732099821909e-06,
-      "loss": 0.6894,
+      "loss": 0.8002,
       "step": 23460
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1088340636754714e-06,
-      "loss": 0.755,
+      "loss": 0.8663,
       "step": 23461
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1082950407210646e-06,
-      "loss": 0.7619,
+      "loss": 0.8068,
       "step": 23462
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1077561411264491e-06,
-      "loss": 0.5568,
+      "loss": 0.7302,
       "step": 23463
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.107217364899099e-06,
-      "loss": 0.6914,
+      "loss": 0.7412,
       "step": 23464
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1066787120464938e-06,
-      "loss": 0.6319,
+      "loss": 0.6973,
       "step": 23465
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1061401825760999e-06,
-      "loss": 0.8001,
+      "loss": 0.8318,
       "step": 23466
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1056017764953941e-06,
-      "loss": 0.7303,
+      "loss": 0.8162,
       "step": 23467
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1050634938118443e-06,
-      "loss": 0.9631,
+      "loss": 0.9462,
       "step": 23468
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1045253345329188e-06,
-      "loss": 0.6994,
+      "loss": 0.8088,
       "step": 23469
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1039872986660805e-06,
-      "loss": 0.6666,
+      "loss": 0.7545,
       "step": 23470
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1034493862188012e-06,
-      "loss": 0.7211,
+      "loss": 0.8073,
       "step": 23471
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1029115971985338e-06,
-      "loss": 0.6842,
+      "loss": 0.8356,
       "step": 23472
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1023739316127479e-06,
-      "loss": 0.8055,
+      "loss": 0.8171,
       "step": 23473
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1018363894689e-06,
-      "loss": 0.6907,
+      "loss": 0.833,
       "step": 23474
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1012989707744458e-06,
-      "loss": 0.6669,
+      "loss": 0.7687,
       "step": 23475
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.1007616755368445e-06,
-      "loss": 0.7651,
+      "loss": 0.7851,
       "step": 23476
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.100224503763545e-06,
-      "loss": 0.9151,
+      "loss": 0.9113,
       "step": 23477
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0996874554620086e-06,
-      "loss": 0.7157,
+      "loss": 0.8021,
       "step": 23478
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0991505306396766e-06,
-      "loss": 0.7677,
+      "loss": 0.8016,
       "step": 23479
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0986137293040034e-06,
-      "loss": 0.736,
+      "loss": 0.7991,
       "step": 23480
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0980770514624361e-06,
-      "loss": 0.6869,
+      "loss": 0.77,
       "step": 23481
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0975404971224201e-06,
-      "loss": 0.7495,
+      "loss": 0.823,
       "step": 23482
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.097004066291396e-06,
-      "loss": 0.6667,
+      "loss": 0.818,
       "step": 23483
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0964677589768125e-06,
-      "loss": 0.7946,
+      "loss": 0.8651,
       "step": 23484
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0959315751861033e-06,
-      "loss": 0.6972,
+      "loss": 0.7554,
       "step": 23485
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0953955149267114e-06,
-      "loss": 0.6838,
+      "loss": 0.764,
       "step": 23486
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0948595782060734e-06,
-      "loss": 0.6884,
+      "loss": 0.7704,
       "step": 23487
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0943237650316229e-06,
-      "loss": 0.8127,
+      "loss": 0.8779,
       "step": 23488
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0937880754107943e-06,
-      "loss": 0.6349,
+      "loss": 0.7803,
       "step": 23489
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0932525093510204e-06,
-      "loss": 0.7235,
+      "loss": 0.849,
       "step": 23490
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.092717066859732e-06,
-      "loss": 0.7157,
+      "loss": 0.7818,
       "step": 23491
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0921817479443565e-06,
-      "loss": 0.6588,
+      "loss": 0.7793,
       "step": 23492
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0916465526123199e-06,
-      "loss": 0.5839,
+      "loss": 0.8035,
       "step": 23493
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0911114808710488e-06,
-      "loss": 0.6777,
+      "loss": 0.75,
       "step": 23494
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0905765327279649e-06,
-      "loss": 0.7576,
+      "loss": 0.8536,
       "step": 23495
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0900417081904901e-06,
-      "loss": 0.699,
+      "loss": 0.8092,
       "step": 23496
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0895070072660473e-06,
-      "loss": 0.716,
+      "loss": 0.741,
       "step": 23497
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0889724299620498e-06,
-      "loss": 0.771,
+      "loss": 0.8434,
       "step": 23498
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0884379762859187e-06,
-      "loss": 0.6818,
+      "loss": 0.7686,
       "step": 23499
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0879036462450664e-06,
-      "loss": 0.7142,
+      "loss": 0.8597,
       "step": 23500
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0873694398469071e-06,
-      "loss": 0.813,
+      "loss": 0.8581,
       "step": 23501
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0868353570988487e-06,
-      "loss": 0.7192,
+      "loss": 0.866,
       "step": 23502
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0863013980083059e-06,
-      "loss": 0.7331,
+      "loss": 0.8003,
       "step": 23503
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0857675625826836e-06,
-      "loss": 0.7759,
+      "loss": 0.7948,
       "step": 23504
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0852338508293892e-06,
-      "loss": 0.7554,
+      "loss": 0.8577,
       "step": 23505
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0847002627558268e-06,
-      "loss": 0.7368,
+      "loss": 0.8248,
       "step": 23506
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0841667983693993e-06,
-      "loss": 0.7979,
+      "loss": 0.8649,
       "step": 23507
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.0836334576775055e-06,
-      "loss": 0.7432,
+      "loss": 0.7712,
       "step": 23508
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.083100240687549e-06,
-      "loss": 0.7383,
+      "loss": 0.8312,
       "step": 23509
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0825671474069243e-06,
-      "loss": 0.6387,
+      "loss": 0.7407,
       "step": 23510
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0820341778430298e-06,
-      "loss": 0.7546,
+      "loss": 0.824,
       "step": 23511
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0815013320032563e-06,
-      "loss": 0.68,
+      "loss": 0.7143,
       "step": 23512
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.080968609895e-06,
-      "loss": 0.7869,
+      "loss": 0.8313,
       "step": 23513
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0804360115256485e-06,
-      "loss": 0.7557,
+      "loss": 0.8272,
       "step": 23514
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.07990353690259e-06,
-      "loss": 0.7594,
+      "loss": 0.7919,
       "step": 23515
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.079371186033219e-06,
-      "loss": 0.76,
+      "loss": 0.7693,
       "step": 23516
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0788389589249115e-06,
-      "loss": 0.7492,
+      "loss": 0.792,
       "step": 23517
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0783068555850585e-06,
-      "loss": 0.6539,
+      "loss": 0.8074,
       "step": 23518
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0777748760210382e-06,
-      "loss": 0.659,
+      "loss": 0.7452,
       "step": 23519
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.077243020240234e-06,
-      "loss": 0.7376,
+      "loss": 0.6959,
       "step": 23520
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0767112882500207e-06,
-      "loss": 0.8657,
+      "loss": 0.8514,
       "step": 23521
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0761796800577794e-06,
-      "loss": 0.657,
+      "loss": 0.8329,
       "step": 23522
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0756481956708841e-06,
-      "loss": 0.7923,
+      "loss": 0.8868,
       "step": 23523
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.075116835096708e-06,
-      "loss": 0.6786,
+      "loss": 0.8081,
       "step": 23524
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0745855983426211e-06,
-      "loss": 0.7889,
+      "loss": 0.7872,
       "step": 23525
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0740544854159995e-06,
-      "loss": 0.7154,
+      "loss": 0.7767,
       "step": 23526
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0735234963242036e-06,
-      "loss": 0.8565,
+      "loss": 0.8767,
       "step": 23527
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0729926310746064e-06,
-      "loss": 0.6804,
+      "loss": 0.7771,
       "step": 23528
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.072461889674571e-06,
-      "loss": 0.8483,
+      "loss": 0.8686,
       "step": 23529
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0719312721314601e-06,
-      "loss": 0.7691,
+      "loss": 0.7672,
       "step": 23530
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0714007784526348e-06,
-      "loss": 0.7604,
+      "loss": 0.91,
       "step": 23531
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.070870408645457e-06,
-      "loss": 0.6781,
+      "loss": 0.803,
       "step": 23532
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0703401627172827e-06,
-      "loss": 0.8172,
+      "loss": 0.8098,
       "step": 23533
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0698100406754663e-06,
-      "loss": 0.7399,
+      "loss": 0.7928,
       "step": 23534
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0692800425273685e-06,
-      "loss": 0.6913,
+      "loss": 0.7177,
       "step": 23535
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0687501682803392e-06,
-      "loss": 0.7886,
+      "loss": 0.84,
       "step": 23536
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0682204179417287e-06,
-      "loss": 0.6975,
+      "loss": 0.7701,
       "step": 23537
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0676907915188872e-06,
-      "loss": 0.6641,
+      "loss": 0.7581,
       "step": 23538
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0671612890191641e-06,
-      "loss": 0.7678,
+      "loss": 0.8522,
       "step": 23539
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0666319104499011e-06,
-      "loss": 0.6687,
+      "loss": 0.7665,
       "step": 23540
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0661026558184473e-06,
-      "loss": 0.6729,
+      "loss": 0.8207,
       "step": 23541
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0655735251321441e-06,
-      "loss": 0.7356,
+      "loss": 0.7745,
       "step": 23542
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0650445183983315e-06,
-      "loss": 0.7261,
+      "loss": 0.8097,
       "step": 23543
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.064515635624348e-06,
-      "loss": 0.5919,
+      "loss": 0.7702,
       "step": 23544
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0639868768175354e-06,
-      "loss": 0.6265,
+      "loss": 0.7448,
       "step": 23545
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0634582419852224e-06,
-      "loss": 0.6866,
+      "loss": 0.8339,
       "step": 23546
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0629297311347498e-06,
-      "loss": 0.6155,
+      "loss": 0.743,
       "step": 23547
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.062401344273446e-06,
-      "loss": 0.7727,
+      "loss": 0.8297,
       "step": 23548
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.061873081408643e-06,
-      "loss": 0.7307,
+      "loss": 0.7664,
       "step": 23549
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0613449425476696e-06,
-      "loss": 0.7706,
+      "loss": 0.8562,
       "step": 23550
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.060816927697853e-06,
-      "loss": 0.6503,
+      "loss": 0.7838,
       "step": 23551
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0602890368665186e-06,
-      "loss": 0.6938,
+      "loss": 0.7371,
       "step": 23552
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0597612700609882e-06,
-      "loss": 0.7017,
+      "loss": 0.7537,
       "step": 23553
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0592336272885872e-06,
-      "loss": 0.7672,
+      "loss": 0.7919,
       "step": 23554
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0587061085566352e-06,
-      "loss": 0.7912,
+      "loss": 0.7559,
       "step": 23555
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0581787138724487e-06,
-      "loss": 0.7676,
+      "loss": 0.8297,
       "step": 23556
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0576514432433439e-06,
-      "loss": 0.8004,
+      "loss": 0.8126,
       "step": 23557
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0571242966766426e-06,
-      "loss": 0.717,
+      "loss": 0.776,
       "step": 23558
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0565972741796494e-06,
-      "loss": 0.8558,
+      "loss": 0.9165,
       "step": 23559
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0560703757596813e-06,
-      "loss": 0.7683,
+      "loss": 0.7985,
       "step": 23560
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0555436014240471e-06,
-      "loss": 0.761,
+      "loss": 0.9078,
       "step": 23561
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0550169511800545e-06,
-      "loss": 0.6418,
+      "loss": 0.6739,
       "step": 23562
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0544904250350097e-06,
-      "loss": 0.6851,
+      "loss": 0.7791,
       "step": 23563
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0539640229962212e-06,
-      "loss": 0.7699,
+      "loss": 0.8255,
       "step": 23564
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0534377450709853e-06,
-      "loss": 0.7792,
+      "loss": 0.7945,
       "step": 23565
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0529115912666088e-06,
-      "loss": 0.5945,
+      "loss": 0.7255,
       "step": 23566
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.052385561590391e-06,
-      "loss": 0.755,
+      "loss": 0.6839,
       "step": 23567
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0518596560496275e-06,
-      "loss": 0.8119,
+      "loss": 0.8664,
       "step": 23568
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0513338746516134e-06,
-      "loss": 0.6073,
+      "loss": 0.7679,
       "step": 23569
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0508082174036483e-06,
-      "loss": 0.7231,
+      "loss": 0.82,
       "step": 23570
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0502826843130231e-06,
-      "loss": 0.8833,
+      "loss": 0.9552,
       "step": 23571
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0497572753870233e-06,
-      "loss": 0.8193,
+      "loss": 0.8835,
       "step": 23572
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0492319906329452e-06,
-      "loss": 0.6744,
+      "loss": 0.7637,
       "step": 23573
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0487068300580739e-06,
-      "loss": 0.7246,
+      "loss": 0.8684,
       "step": 23574
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0481817936696948e-06,
-      "loss": 0.7935,
+      "loss": 0.8238,
       "step": 23575
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0476568814750899e-06,
-      "loss": 0.7529,
+      "loss": 0.8304,
       "step": 23576
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0471320934815488e-06,
-      "loss": 0.7963,
+      "loss": 0.8638,
       "step": 23577
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0466074296963435e-06,
-      "loss": 0.8282,
+      "loss": 0.9404,
       "step": 23578
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0460828901267583e-06,
-      "loss": 0.7256,
+      "loss": 0.874,
       "step": 23579
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0455584747800696e-06,
-      "loss": 0.7008,
+      "loss": 0.8344,
       "step": 23580
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0450341836635524e-06,
-      "loss": 0.5824,
+      "loss": 0.7462,
       "step": 23581
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.044510016784478e-06,
-      "loss": 0.6464,
+      "loss": 0.7337,
       "step": 23582
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0439859741501245e-06,
-      "loss": 0.7524,
+      "loss": 0.8767,
       "step": 23583
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0434620557677555e-06,
-      "loss": 0.7055,
+      "loss": 0.7827,
       "step": 23584
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0429382616446448e-06,
-      "loss": 0.5936,
+      "loss": 0.746,
       "step": 23585
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0424145917880569e-06,
-      "loss": 0.6412,
+      "loss": 0.7965,
       "step": 23586
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0418910462052577e-06,
-      "loss": 0.7759,
+      "loss": 0.8043,
       "step": 23587
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0413676249035086e-06,
-      "loss": 0.7647,
+      "loss": 0.8409,
       "step": 23588
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0408443278900747e-06,
-      "loss": 0.778,
+      "loss": 0.8803,
       "step": 23589
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.040321155172217e-06,
-      "loss": 0.7283,
+      "loss": 0.7994,
       "step": 23590
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0397981067571861e-06,
-      "loss": 0.8541,
+      "loss": 0.8722,
       "step": 23591
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0392751826522462e-06,
-      "loss": 0.7923,
+      "loss": 0.8324,
       "step": 23592
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0387523828646495e-06,
-      "loss": 0.6607,
+      "loss": 0.7528,
       "step": 23593
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.038229707401649e-06,
-      "loss": 0.7267,
+      "loss": 0.7448,
       "step": 23594
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0377071562704944e-06,
-      "loss": 0.6886,
+      "loss": 0.8165,
       "step": 23595
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.037184729478441e-06,
-      "loss": 0.9036,
+      "loss": 0.8957,
       "step": 23596
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0366624270327286e-06,
-      "loss": 0.6559,
+      "loss": 0.7838,
       "step": 23597
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0361402489406102e-06,
-      "loss": 0.6375,
+      "loss": 0.8102,
       "step": 23598
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.035618195209328e-06,
-      "loss": 0.7586,
+      "loss": 0.846,
       "step": 23599
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0350962658461238e-06,
-      "loss": 0.6797,
+      "loss": 0.7008,
       "step": 23600
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0345744608582376e-06,
-      "loss": 0.6356,
+      "loss": 0.7229,
       "step": 23601
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0340527802529132e-06,
-      "loss": 0.5266,
+      "loss": 0.5658,
       "step": 23602
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0335312240373841e-06,
-      "loss": 0.7142,
+      "loss": 0.7185,
       "step": 23603
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0330097922188886e-06,
-      "loss": 0.7293,
+      "loss": 0.8479,
       "step": 23604
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0324884848046601e-06,
-      "loss": 0.7154,
+      "loss": 0.7913,
       "step": 23605
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0319673018019294e-06,
-      "loss": 0.8309,
+      "loss": 0.8562,
       "step": 23606
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0314462432179273e-06,
-      "loss": 0.7766,
+      "loss": 0.8366,
       "step": 23607
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0309253090598858e-06,
-      "loss": 0.6386,
+      "loss": 0.7332,
       "step": 23608
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0304044993350314e-06,
-      "loss": 0.6803,
+      "loss": 0.7341,
       "step": 23609
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.029883814050585e-06,
-      "loss": 0.5709,
+      "loss": 0.7127,
       "step": 23610
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0293632532137755e-06,
-      "loss": 0.6918,
+      "loss": 0.8229,
       "step": 23611
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0288428168318244e-06,
-      "loss": 0.6775,
+      "loss": 0.8475,
       "step": 23612
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0283225049119494e-06,
-      "loss": 0.7722,
+      "loss": 0.7548,
       "step": 23613
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0278023174613682e-06,
-      "loss": 0.722,
+      "loss": 0.813,
       "step": 23614
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0272822544873051e-06,
-      "loss": 0.5695,
+      "loss": 0.7123,
       "step": 23615
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0267623159969652e-06,
-      "loss": 0.6585,
+      "loss": 0.7084,
       "step": 23616
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0262425019975697e-06,
-      "loss": 0.7231,
+      "loss": 0.7819,
       "step": 23617
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.025722812496327e-06,
-      "loss": 0.8378,
+      "loss": 0.7829,
       "step": 23618
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.025203247500447e-06,
-      "loss": 0.7285,
+      "loss": 0.7339,
       "step": 23619
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0246838070171373e-06,
-      "loss": 0.8116,
+      "loss": 0.8877,
       "step": 23620
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0241644910536063e-06,
-      "loss": 0.7155,
+      "loss": 0.8102,
       "step": 23621
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0236452996170588e-06,
-      "loss": 0.6685,
+      "loss": 0.7478,
       "step": 23622
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0231262327146973e-06,
-      "loss": 0.7178,
+      "loss": 0.7297,
       "step": 23623
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0226072903537198e-06,
-      "loss": 0.7008,
+      "loss": 0.7878,
       "step": 23624
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0220884725413337e-06,
-      "loss": 0.8919,
+      "loss": 0.9225,
       "step": 23625
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0215697792847278e-06,
-      "loss": 0.6929,
+      "loss": 0.7714,
       "step": 23626
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0210512105911042e-06,
-      "loss": 0.7469,
+      "loss": 0.8042,
       "step": 23627
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0205327664676578e-06,
-      "loss": 0.7135,
+      "loss": 0.8123,
       "step": 23628
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0200144469215745e-06,
-      "loss": 0.7127,
+      "loss": 0.7336,
       "step": 23629
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0194962519600525e-06,
-      "loss": 0.5644,
+      "loss": 0.6614,
       "step": 23630
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0189781815902789e-06,
-      "loss": 0.8161,
+      "loss": 0.8787,
       "step": 23631
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0184602358194395e-06,
-      "loss": 0.551,
+      "loss": 0.7101,
       "step": 23632
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0179424146547202e-06,
-      "loss": 0.5423,
+      "loss": 0.6796,
       "step": 23633
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.017424718103307e-06,
-      "loss": 0.5568,
+      "loss": 0.6778,
       "step": 23634
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0169071461723812e-06,
-      "loss": 0.6469,
+      "loss": 0.7015,
       "step": 23635
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0163896988691247e-06,
-      "loss": 0.7344,
+      "loss": 0.7571,
       "step": 23636
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0158723762007138e-06,
-      "loss": 0.7754,
+      "loss": 0.7687,
       "step": 23637
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0153551781743275e-06,
-      "loss": 0.7116,
+      "loss": 0.7919,
       "step": 23638
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0148381047971378e-06,
-      "loss": 0.7957,
+      "loss": 0.8552,
       "step": 23639
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0143211560763245e-06,
-      "loss": 0.7421,
+      "loss": 0.8335,
       "step": 23640
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0138043320190548e-06,
-      "loss": 0.7956,
+      "loss": 0.8462,
       "step": 23641
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0132876326325014e-06,
-      "loss": 0.66,
+      "loss": 0.7482,
       "step": 23642
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0127710579238281e-06,
-      "loss": 0.6925,
+      "loss": 0.7738,
       "step": 23643
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0122546079002105e-06,
-      "loss": 0.7718,
+      "loss": 0.762,
       "step": 23644
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0117382825688037e-06,
-      "loss": 0.8567,
+      "loss": 0.8474,
       "step": 23645
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0112220819367779e-06,
-      "loss": 0.6409,
+      "loss": 0.7538,
       "step": 23646
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0107060060112917e-06,
-      "loss": 0.6878,
+      "loss": 0.7214,
       "step": 23647
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0101900547995057e-06,
-      "loss": 0.824,
+      "loss": 0.8914,
       "step": 23648
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.009674228308578e-06,
-      "loss": 0.6667,
+      "loss": 0.8671,
       "step": 23649
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0091585265456649e-06,
-      "loss": 0.7794,
+      "loss": 0.8306,
       "step": 23650
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0086429495179217e-06,
-      "loss": 0.6509,
+      "loss": 0.799,
       "step": 23651
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0081274972324972e-06,
-      "loss": 0.7165,
+      "loss": 0.7094,
       "step": 23652
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.007612169696548e-06,
-      "loss": 0.7824,
+      "loss": 0.786,
       "step": 23653
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0070969669172226e-06,
-      "loss": 0.6469,
+      "loss": 0.775,
       "step": 23654
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0065818889016665e-06,
-      "loss": 0.6878,
+      "loss": 0.7877,
       "step": 23655
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.006066935657024e-06,
-      "loss": 0.8072,
+      "loss": 0.8187,
       "step": 23656
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0055521071904473e-06,
-      "loss": 0.8241,
+      "loss": 0.8314,
       "step": 23657
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0050374035090693e-06,
-      "loss": 0.6784,
+      "loss": 0.7243,
       "step": 23658
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0045228246200367e-06,
-      "loss": 0.7038,
+      "loss": 0.7603,
       "step": 23659
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0040083705304882e-06,
-      "loss": 0.969,
+      "loss": 0.9597,
       "step": 23660
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.003494041247559e-06,
-      "loss": 0.5419,
+      "loss": 0.716,
       "step": 23661
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0029798367783838e-06,
-      "loss": 0.743,
+      "loss": 0.7697,
       "step": 23662
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0024657571301023e-06,
-      "loss": 0.7245,
+      "loss": 0.779,
       "step": 23663
     },
     {
@@ -141994,23014 +141994,23014 @@
     {
       "epoch": 0.86,
       "learning_rate": 1.001437972324728e-06,
-      "loss": 0.7046,
+      "loss": 0.8034,
       "step": 23665
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.0009242671818998e-06,
-      "loss": 0.8583,
+      "loss": 0.8017,
       "step": 23666
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.000410686888479e-06,
-      "loss": 0.6748,
+      "loss": 0.7666,
       "step": 23667
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.99897231451591e-07,
-      "loss": 0.7602,
+      "loss": 0.8535,
       "step": 23668
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.993839008783568e-07,
-      "loss": 0.7188,
+      "loss": 0.8301,
       "step": 23669
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.988706951759052e-07,
-      "loss": 0.5814,
+      "loss": 0.7235,
       "step": 23670
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.983576143513484e-07,
-      "loss": 0.7226,
+      "loss": 0.811,
       "step": 23671
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.978446584118096e-07,
-      "loss": 0.6095,
+      "loss": 0.6289,
       "step": 23672
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.973318273644028e-07,
-      "loss": 0.6483,
+      "loss": 0.7755,
       "step": 23673
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.968191212162447e-07,
-      "loss": 0.6827,
+      "loss": 0.7986,
       "step": 23674
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.963065399744464e-07,
-      "loss": 0.6161,
+      "loss": 0.7571,
       "step": 23675
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.957940836461234e-07,
-      "loss": 0.6978,
+      "loss": 0.8144,
       "step": 23676
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.952817522383772e-07,
-      "loss": 0.6951,
+      "loss": 0.7903,
       "step": 23677
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.94769545758324e-07,
-      "loss": 0.6996,
+      "loss": 0.725,
       "step": 23678
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.942574642130665e-07,
-      "loss": 0.6596,
+      "loss": 0.8003,
       "step": 23679
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.93745507609709e-07,
-      "loss": 0.7118,
+      "loss": 0.8022,
       "step": 23680
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.932336759553519e-07,
-      "loss": 0.6682,
+      "loss": 0.7386,
       "step": 23681
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.927219692571033e-07,
-      "loss": 0.6918,
+      "loss": 0.7686,
       "step": 23682
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.922103875220534e-07,
-      "loss": 0.8083,
+      "loss": 0.8889,
       "step": 23683
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.916989307573056e-07,
-      "loss": 0.7329,
+      "loss": 0.7973,
       "step": 23684
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.911875989699548e-07,
-      "loss": 0.8293,
+      "loss": 0.8299,
       "step": 23685
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.906763921670936e-07,
-      "loss": 0.8061,
+      "loss": 0.8178,
       "step": 23686
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.90165310355815e-07,
-      "loss": 0.69,
+      "loss": 0.7757,
       "step": 23687
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.896543535432079e-07,
-      "loss": 0.6921,
+      "loss": 0.7303,
       "step": 23688
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.891435217363666e-07,
-      "loss": 0.7607,
+      "loss": 0.8047,
       "step": 23689
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.886328149423697e-07,
-      "loss": 0.6561,
+      "loss": 0.8014,
       "step": 23690
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.881222331683095e-07,
-      "loss": 0.794,
+      "loss": 0.812,
       "step": 23691
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.876117764212668e-07,
-      "loss": 0.6436,
+      "loss": 0.7387,
       "step": 23692
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.871014447083239e-07,
-      "loss": 0.6284,
+      "loss": 0.6878,
       "step": 23693
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.865912380365594e-07,
-      "loss": 0.7733,
+      "loss": 0.8374,
       "step": 23694
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.860811564130556e-07,
-      "loss": 0.6783,
+      "loss": 0.7599,
       "step": 23695
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.855711998448846e-07,
-      "loss": 0.6392,
+      "loss": 0.7566,
       "step": 23696
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.850613683391242e-07,
-      "loss": 0.7368,
+      "loss": 0.8453,
       "step": 23697
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.845516619028473e-07,
-      "loss": 0.5714,
+      "loss": 0.7306,
       "step": 23698
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.840420805431239e-07,
-      "loss": 0.7115,
+      "loss": 0.7801,
       "step": 23699
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.83532624267024e-07,
-      "loss": 0.648,
+      "loss": 0.7168,
       "step": 23700
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.830232930816174e-07,
-      "loss": 0.7351,
+      "loss": 0.802,
       "step": 23701
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.825140869939686e-07,
-      "loss": 0.7735,
+      "loss": 0.7885,
       "step": 23702
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.82005006011144e-07,
-      "loss": 0.7817,
+      "loss": 0.7513,
       "step": 23703
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.814960501402048e-07,
-      "loss": 0.6409,
+      "loss": 0.8063,
       "step": 23704
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.80987219388212e-07,
-      "loss": 0.7703,
+      "loss": 0.8112,
       "step": 23705
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.804785137622253e-07,
-      "loss": 0.7991,
+      "loss": 0.8181,
       "step": 23706
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.799699332693002e-07,
-      "loss": 0.6602,
+      "loss": 0.7405,
       "step": 23707
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.79461477916498e-07,
-      "loss": 0.8493,
+      "loss": 0.8632,
       "step": 23708
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.78953147710866e-07,
-      "loss": 0.6689,
+      "loss": 0.7378,
       "step": 23709
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.784449426594622e-07,
-      "loss": 0.7177,
+      "loss": 0.739,
       "step": 23710
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.779368627693343e-07,
-      "loss": 0.707,
+      "loss": 0.8028,
       "step": 23711
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.774289080475319e-07,
-      "loss": 0.6395,
+      "loss": 0.7839,
       "step": 23712
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.769210785010995e-07,
-      "loss": 0.7166,
+      "loss": 0.7682,
       "step": 23713
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.764133741370886e-07,
-      "loss": 0.6965,
+      "loss": 0.7874,
       "step": 23714
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.759057949625383e-07,
-      "loss": 0.7994,
+      "loss": 0.9011,
       "step": 23715
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.753983409844913e-07,
-      "loss": 0.6398,
+      "loss": 0.761,
       "step": 23716
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.748910122099886e-07,
-      "loss": 0.7608,
+      "loss": 0.826,
       "step": 23717
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.743838086460677e-07,
-      "loss": 0.8333,
+      "loss": 0.8692,
       "step": 23718
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.738767302997643e-07,
-      "loss": 0.5434,
+      "loss": 0.6741,
       "step": 23719
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.73369777178117e-07,
-      "loss": 0.8,
+      "loss": 0.8149,
       "step": 23720
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.728629492881557e-07,
-      "loss": 0.7337,
+      "loss": 0.8409,
       "step": 23721
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.72356246636914e-07,
-      "loss": 0.7029,
+      "loss": 0.8377,
       "step": 23722
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.718496692314217e-07,
-      "loss": 0.7157,
+      "loss": 0.8329,
       "step": 23723
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.713432170787051e-07,
-      "loss": 0.5285,
+      "loss": 0.6847,
       "step": 23724
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.708368901857912e-07,
-      "loss": 0.6876,
+      "loss": 0.8187,
       "step": 23725
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.70330688559703e-07,
-      "loss": 0.7052,
+      "loss": 0.8028,
       "step": 23726
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.698246122074684e-07,
-      "loss": 0.7273,
+      "loss": 0.8371,
       "step": 23727
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.693186611361017e-07,
-      "loss": 0.7012,
+      "loss": 0.7671,
       "step": 23728
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.688128353526283e-07,
-      "loss": 0.9886,
+      "loss": 0.9363,
       "step": 23729
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.683071348640627e-07,
-      "loss": 0.683,
+      "loss": 0.7288,
       "step": 23730
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.678015596774215e-07,
-      "loss": 0.7267,
+      "loss": 0.8326,
       "step": 23731
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.672961097997158e-07,
-      "loss": 0.7729,
+      "loss": 0.8005,
       "step": 23732
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.667907852379643e-07,
-      "loss": 0.6652,
+      "loss": 0.7587,
       "step": 23733
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.662855859991726e-07,
-      "loss": 0.8032,
+      "loss": 0.8607,
       "step": 23734
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.65780512090353e-07,
-      "loss": 0.9363,
+      "loss": 0.9503,
       "step": 23735
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.652755635185073e-07,
-      "loss": 0.8152,
+      "loss": 0.9061,
       "step": 23736
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.647707402906492e-07,
-      "loss": 0.6693,
+      "loss": 0.7434,
       "step": 23737
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.64266042413774e-07,
-      "loss": 0.6624,
+      "loss": 0.803,
       "step": 23738
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.637614698948893e-07,
-      "loss": 0.6968,
+      "loss": 0.8496,
       "step": 23739
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.632570227409932e-07,
-      "loss": 0.561,
+      "loss": 0.7546,
       "step": 23740
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.627527009590843e-07,
-      "loss": 0.7839,
+      "loss": 0.7797,
       "step": 23741
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.622485045561602e-07,
-      "loss": 0.7541,
+      "loss": 0.8039,
       "step": 23742
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.617444335392146e-07,
-      "loss": 0.8008,
+      "loss": 0.7981,
       "step": 23743
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.612404879152416e-07,
-      "loss": 0.822,
+      "loss": 0.9463,
       "step": 23744
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.60736667691231e-07,
-      "loss": 0.754,
+      "loss": 0.7825,
       "step": 23745
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.602329728741756e-07,
-      "loss": 0.7192,
+      "loss": 0.8036,
       "step": 23746
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.597294034710614e-07,
-      "loss": 0.7405,
+      "loss": 0.7755,
       "step": 23747
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.592259594888764e-07,
-      "loss": 0.7206,
+      "loss": 0.7557,
       "step": 23748
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.587226409346028e-07,
-      "loss": 0.833,
+      "loss": 0.8453,
       "step": 23749
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.58219447815225e-07,
-      "loss": 0.7953,
+      "loss": 0.8866,
       "step": 23750
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.577163801377232e-07,
-      "loss": 0.6238,
+      "loss": 0.7178,
       "step": 23751
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.57213437909078e-07,
-      "loss": 0.7375,
+      "loss": 0.7561,
       "step": 23752
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.567106211362664e-07,
-      "loss": 0.7153,
+      "loss": 0.7592,
       "step": 23753
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.562079298262638e-07,
-      "loss": 0.7199,
+      "loss": 0.768,
       "step": 23754
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.557053639860436e-07,
-      "loss": 0.7628,
+      "loss": 0.7654,
       "step": 23755
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.552029236225824e-07,
-      "loss": 0.7218,
+      "loss": 0.8284,
       "step": 23756
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.547006087428445e-07,
-      "loss": 0.6817,
+      "loss": 0.8297,
       "step": 23757
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.541984193538033e-07,
-      "loss": 0.7629,
+      "loss": 0.8417,
       "step": 23758
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.536963554624257e-07,
-      "loss": 0.7303,
+      "loss": 0.7946,
       "step": 23759
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.531944170756757e-07,
-      "loss": 0.6808,
+      "loss": 0.7712,
       "step": 23760
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.526926042005169e-07,
-      "loss": 0.6917,
+      "loss": 0.7931,
       "step": 23761
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.521909168439125e-07,
-      "loss": 0.701,
+      "loss": 0.8032,
       "step": 23762
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.516893550128215e-07,
-      "loss": 0.7673,
+      "loss": 0.7947,
       "step": 23763
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.511879187142015e-07,
-      "loss": 0.661,
+      "loss": 0.7406,
       "step": 23764
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.506866079550126e-07,
-      "loss": 0.7081,
+      "loss": 0.8019,
       "step": 23765
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.501854227422069e-07,
-      "loss": 0.5551,
+      "loss": 0.7849,
       "step": 23766
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.496843630827379e-07,
-      "loss": 0.7464,
+      "loss": 0.8161,
       "step": 23767
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.491834289835566e-07,
-      "loss": 0.7166,
+      "loss": 0.8034,
       "step": 23768
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.486826204516175e-07,
-      "loss": 0.5999,
+      "loss": 0.778,
       "step": 23769
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.481819374938617e-07,
-      "loss": 0.8476,
+      "loss": 0.8438,
       "step": 23770
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.476813801172402e-07,
-      "loss": 0.7282,
+      "loss": 0.7896,
       "step": 23771
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.471809483286953e-07,
-      "loss": 0.7244,
+      "loss": 0.7978,
       "step": 23772
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.466806421351715e-07,
-      "loss": 0.6634,
+      "loss": 0.8376,
       "step": 23773
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.461804615436066e-07,
-      "loss": 0.662,
+      "loss": 0.8339,
       "step": 23774
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.456804065609448e-07,
-      "loss": 0.6673,
+      "loss": 0.8185,
       "step": 23775
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.451804771941187e-07,
-      "loss": 0.674,
+      "loss": 0.7659,
       "step": 23776
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.44680673450068e-07,
-      "loss": 0.625,
+      "loss": 0.7071,
       "step": 23777
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.441809953357239e-07,
-      "loss": 0.7123,
+      "loss": 0.7809,
       "step": 23778
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.43681442858021e-07,
-      "loss": 0.7744,
+      "loss": 0.8731,
       "step": 23779
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.431820160238891e-07,
-      "loss": 0.7921,
+      "loss": 0.8525,
       "step": 23780
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.426827148402562e-07,
-      "loss": 0.6792,
+      "loss": 0.8222,
       "step": 23781
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.421835393140499e-07,
-      "loss": 0.5198,
+      "loss": 0.6861,
       "step": 23782
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.416844894521937e-07,
-      "loss": 0.6273,
+      "loss": 0.7419,
       "step": 23783
     },
     {
       "epoch": 0.86,
       "learning_rate": 9.411855652616153e-07,
-      "loss": 0.8278,
+      "loss": 0.8344,
       "step": 23784
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.406867667492347e-07,
-      "loss": 0.6594,
+      "loss": 0.83,
       "step": 23785
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.401880939219721e-07,
-      "loss": 0.6831,
+      "loss": 0.7993,
       "step": 23786
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.396895467867428e-07,
-      "loss": 0.8683,
+      "loss": 0.8948,
       "step": 23787
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.391911253504693e-07,
-      "loss": 0.6423,
+      "loss": 0.7417,
       "step": 23788
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.386928296200603e-07,
-      "loss": 0.616,
+      "loss": 0.6635,
       "step": 23789
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.381946596024349e-07,
-      "loss": 0.7988,
+      "loss": 0.831,
       "step": 23790
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.376966153044997e-07,
-      "loss": 0.6655,
+      "loss": 0.7329,
       "step": 23791
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.37198696733167e-07,
-      "loss": 0.6411,
+      "loss": 0.8168,
       "step": 23792
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.367009038953423e-07,
-      "loss": 0.7256,
+      "loss": 0.837,
       "step": 23793
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.362032367979379e-07,
-      "loss": 0.9766,
+      "loss": 0.9332,
       "step": 23794
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.357056954478482e-07,
-      "loss": 0.8211,
+      "loss": 0.9315,
       "step": 23795
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.352082798519846e-07,
-      "loss": 0.7265,
+      "loss": 0.8146,
       "step": 23796
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.347109900172446e-07,
-      "loss": 0.7701,
+      "loss": 0.8301,
       "step": 23797
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.342138259505273e-07,
-      "loss": 0.7501,
+      "loss": 0.7621,
       "step": 23798
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.337167876587305e-07,
-      "loss": 0.8029,
+      "loss": 0.8236,
       "step": 23799
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.332198751487486e-07,
-      "loss": 0.7005,
+      "loss": 0.7878,
       "step": 23800
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.327230884274807e-07,
-      "loss": 0.8215,
+      "loss": 0.872,
       "step": 23801
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.322264275018111e-07,
-      "loss": 0.5508,
+      "loss": 0.6531,
       "step": 23802
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.317298923786355e-07,
-      "loss": 0.7821,
+      "loss": 0.811,
       "step": 23803
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.312334830648417e-07,
-      "loss": 0.8862,
+      "loss": 0.927,
       "step": 23804
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.307371995673176e-07,
-      "loss": 0.7324,
+      "loss": 0.8581,
       "step": 23805
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.302410418929442e-07,
-      "loss": 0.7681,
+      "loss": 0.8054,
       "step": 23806
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.297450100486127e-07,
-      "loss": 0.7201,
+      "loss": 0.7556,
       "step": 23807
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.292491040411955e-07,
-      "loss": 0.6024,
+      "loss": 0.6869,
       "step": 23808
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.287533238775792e-07,
-      "loss": 0.7281,
+      "loss": 0.841,
       "step": 23809
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.282576695646406e-07,
-      "loss": 0.7359,
+      "loss": 0.8679,
       "step": 23810
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.277621411092552e-07,
-      "loss": 0.6604,
+      "loss": 0.7935,
       "step": 23811
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.272667385182976e-07,
-      "loss": 0.8215,
+      "loss": 0.9149,
       "step": 23812
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.267714617986423e-07,
-      "loss": 0.7858,
+      "loss": 0.7796,
       "step": 23813
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.262763109571593e-07,
-      "loss": 0.7481,
+      "loss": 0.7635,
       "step": 23814
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.257812860007198e-07,
-      "loss": 0.6587,
+      "loss": 0.7688,
       "step": 23815
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.252863869361894e-07,
-      "loss": 0.786,
+      "loss": 0.87,
       "step": 23816
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.247916137704349e-07,
-      "loss": 0.7693,
+      "loss": 0.8543,
       "step": 23817
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.242969665103197e-07,
-      "loss": 0.7761,
+      "loss": 0.8362,
       "step": 23818
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.238024451627092e-07,
-      "loss": 0.4992,
+      "loss": 0.6681,
       "step": 23819
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.233080497344637e-07,
-      "loss": 0.7772,
+      "loss": 0.7854,
       "step": 23820
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.228137802324366e-07,
-      "loss": 0.7569,
+      "loss": 0.746,
       "step": 23821
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.223196366634924e-07,
-      "loss": 0.7994,
+      "loss": 0.8802,
       "step": 23822
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.218256190344832e-07,
-      "loss": 0.7207,
+      "loss": 0.8114,
       "step": 23823
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.213317273522637e-07,
-      "loss": 0.719,
+      "loss": 0.7801,
       "step": 23824
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.208379616236828e-07,
-      "loss": 0.7242,
+      "loss": 0.782,
       "step": 23825
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.203443218555985e-07,
-      "loss": 0.6974,
+      "loss": 0.8604,
       "step": 23826
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.198508080548497e-07,
-      "loss": 0.6014,
+      "loss": 0.7043,
       "step": 23827
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.193574202282895e-07,
-      "loss": 0.7297,
+      "loss": 0.7649,
       "step": 23828
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.188641583827629e-07,
-      "loss": 0.7356,
+      "loss": 0.8019,
       "step": 23829
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.183710225251097e-07,
-      "loss": 0.5472,
+      "loss": 0.7144,
       "step": 23830
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.178780126621734e-07,
-      "loss": 0.722,
+      "loss": 0.804,
       "step": 23831
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.17385128800794e-07,
-      "loss": 0.5529,
+      "loss": 0.7807,
       "step": 23832
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.168923709478106e-07,
-      "loss": 0.8234,
+      "loss": 0.8638,
       "step": 23833
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.163997391100577e-07,
-      "loss": 0.7879,
+      "loss": 0.8066,
       "step": 23834
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.159072332943686e-07,
-      "loss": 0.6588,
+      "loss": 0.7102,
       "step": 23835
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.154148535075824e-07,
-      "loss": 0.6259,
+      "loss": 0.6651,
       "step": 23836
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.149225997565214e-07,
-      "loss": 0.7434,
+      "loss": 0.7938,
       "step": 23837
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.144304720480212e-07,
-      "loss": 0.6486,
+      "loss": 0.766,
       "step": 23838
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.139384703889098e-07,
-      "loss": 0.7794,
+      "loss": 0.8288,
       "step": 23839
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.134465947860061e-07,
-      "loss": 0.8744,
+      "loss": 0.8673,
       "step": 23840
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.129548452461412e-07,
-      "loss": 0.7507,
+      "loss": 0.7637,
       "step": 23841
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.124632217761354e-07,
-      "loss": 0.8197,
+      "loss": 0.7824,
       "step": 23842
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.119717243828075e-07,
-      "loss": 0.6814,
+      "loss": 0.7815,
       "step": 23843
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.114803530729777e-07,
-      "loss": 0.742,
+      "loss": 0.7366,
       "step": 23844
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.109891078534639e-07,
-      "loss": 0.5943,
+      "loss": 0.7816,
       "step": 23845
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.104979887310805e-07,
-      "loss": 0.7633,
+      "loss": 0.7737,
       "step": 23846
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.100069957126412e-07,
-      "loss": 0.7458,
+      "loss": 0.8321,
       "step": 23847
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.095161288049581e-07,
-      "loss": 0.7454,
+      "loss": 0.7278,
       "step": 23848
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.090253880148403e-07,
-      "loss": 0.589,
+      "loss": 0.7442,
       "step": 23849
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.085347733490956e-07,
-      "loss": 0.7551,
+      "loss": 0.7101,
       "step": 23850
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.080442848145344e-07,
-      "loss": 0.6951,
+      "loss": 0.7506,
       "step": 23851
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.075539224179574e-07,
-      "loss": 0.7357,
+      "loss": 0.8379,
       "step": 23852
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.070636861661708e-07,
-      "loss": 0.8032,
+      "loss": 0.8528,
       "step": 23853
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.065735760659711e-07,
-      "loss": 0.7968,
+      "loss": 0.8243,
       "step": 23854
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.060835921241651e-07,
-      "loss": 0.6334,
+      "loss": 0.7406,
       "step": 23855
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.055937343475441e-07,
-      "loss": 0.8091,
+      "loss": 0.7547,
       "step": 23856
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.051040027429081e-07,
-      "loss": 0.6394,
+      "loss": 0.6787,
       "step": 23857
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.046143973170496e-07,
-      "loss": 0.7258,
+      "loss": 0.8817,
       "step": 23858
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.04124918076762e-07,
-      "loss": 0.7067,
+      "loss": 0.7759,
       "step": 23859
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.036355650288353e-07,
-      "loss": 0.6327,
+      "loss": 0.7405,
       "step": 23860
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.031463381800598e-07,
-      "loss": 0.5497,
+      "loss": 0.6679,
       "step": 23861
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.026572375372222e-07,
-      "loss": 0.7524,
+      "loss": 0.8437,
       "step": 23862
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.02168263107105e-07,
-      "loss": 0.7731,
+      "loss": 0.8559,
       "step": 23863
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.01679414896498e-07,
-      "loss": 0.7534,
+      "loss": 0.8053,
       "step": 23864
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.011906929121794e-07,
-      "loss": 0.6262,
+      "loss": 0.7401,
       "step": 23865
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.007020971609315e-07,
-      "loss": 0.6767,
+      "loss": 0.7269,
       "step": 23866
     },
     {
       "epoch": 0.87,
       "learning_rate": 9.002136276495288e-07,
-      "loss": 0.665,
+      "loss": 0.746,
       "step": 23867
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.997252843847559e-07,
-      "loss": 0.7212,
+      "loss": 0.8124,
       "step": 23868
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.992370673733786e-07,
-      "loss": 0.7002,
+      "loss": 0.7753,
       "step": 23869
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.987489766221769e-07,
-      "loss": 0.6591,
+      "loss": 0.7463,
       "step": 23870
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.982610121379198e-07,
-      "loss": 0.7716,
+      "loss": 0.8147,
       "step": 23871
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.977731739273776e-07,
-      "loss": 0.8578,
+      "loss": 0.8291,
       "step": 23872
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.97285461997317e-07,
-      "loss": 0.7296,
+      "loss": 0.7537,
       "step": 23873
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.967978763545093e-07,
-      "loss": 0.6881,
+      "loss": 0.7661,
       "step": 23874
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.963104170057102e-07,
-      "loss": 0.7078,
+      "loss": 0.7912,
       "step": 23875
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.958230839576909e-07,
-      "loss": 0.6568,
+      "loss": 0.7868,
       "step": 23876
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.953358772172083e-07,
-      "loss": 0.5989,
+      "loss": 0.7124,
       "step": 23877
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.948487967910235e-07,
-      "loss": 0.7118,
+      "loss": 0.848,
       "step": 23878
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.943618426858925e-07,
-      "loss": 0.6114,
+      "loss": 0.7759,
       "step": 23879
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.938750149085695e-07,
-      "loss": 0.7191,
+      "loss": 0.7324,
       "step": 23880
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.93388313465815e-07,
-      "loss": 0.6648,
+      "loss": 0.7494,
       "step": 23881
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.929017383643735e-07,
-      "loss": 0.782,
+      "loss": 0.7375,
       "step": 23882
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.924152896110005e-07,
-      "loss": 0.6127,
+      "loss": 0.7188,
       "step": 23883
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.919289672124432e-07,
-      "loss": 0.7059,
+      "loss": 0.7847,
       "step": 23884
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.914427711754492e-07,
-      "loss": 0.879,
+      "loss": 0.9499,
       "step": 23885
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.909567015067622e-07,
-      "loss": 0.6841,
+      "loss": 0.7261,
       "step": 23886
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.9047075821313e-07,
-      "loss": 0.7174,
+      "loss": 0.8036,
       "step": 23887
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.899849413012874e-07,
-      "loss": 0.7217,
+      "loss": 0.7701,
       "step": 23888
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.894992507779798e-07,
-      "loss": 0.765,
+      "loss": 0.7538,
       "step": 23889
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.890136866499443e-07,
-      "loss": 0.735,
+      "loss": 0.8363,
       "step": 23890
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.885282489239177e-07,
-      "loss": 0.6009,
+      "loss": 0.745,
       "step": 23891
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.880429376066324e-07,
-      "loss": 0.6681,
+      "loss": 0.7805,
       "step": 23892
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.875577527048262e-07,
-      "loss": 0.7024,
+      "loss": 0.7169,
       "step": 23893
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.870726942252239e-07,
-      "loss": 0.6963,
+      "loss": 0.7602,
       "step": 23894
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.8658776217456e-07,
-      "loss": 0.8095,
+      "loss": 0.899,
       "step": 23895
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.861029565595602e-07,
-      "loss": 0.9078,
+      "loss": 0.8714,
       "step": 23896
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.856182773869526e-07,
-      "loss": 0.7535,
+      "loss": 0.802,
       "step": 23897
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.851337246634584e-07,
-      "loss": 0.7467,
+      "loss": 0.8118,
       "step": 23898
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.846492983957999e-07,
-      "loss": 0.7883,
+      "loss": 0.8358,
       "step": 23899
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.84164998590703e-07,
-      "loss": 0.7313,
+      "loss": 0.6961,
       "step": 23900
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.836808252548801e-07,
-      "loss": 0.7311,
+      "loss": 0.7979,
       "step": 23901
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.831967783950524e-07,
-      "loss": 0.6943,
+      "loss": 0.7448,
       "step": 23902
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.827128580179345e-07,
-      "loss": 0.8108,
+      "loss": 0.8555,
       "step": 23903
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.8222906413024e-07,
-      "loss": 0.7031,
+      "loss": 0.8671,
       "step": 23904
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.817453967386791e-07,
-      "loss": 0.7969,
+      "loss": 0.878,
       "step": 23905
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.812618558499674e-07,
-      "loss": 0.6177,
+      "loss": 0.7443,
       "step": 23906
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.807784414708054e-07,
-      "loss": 0.6825,
+      "loss": 0.8268,
       "step": 23907
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.802951536079063e-07,
-      "loss": 0.6399,
+      "loss": 0.7099,
       "step": 23908
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.798119922679737e-07,
-      "loss": 0.6048,
+      "loss": 0.7105,
       "step": 23909
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.793289574577091e-07,
-      "loss": 0.8048,
+      "loss": 0.8944,
       "step": 23910
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.788460491838124e-07,
-      "loss": 0.7261,
+      "loss": 0.8258,
       "step": 23911
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.783632674529874e-07,
-      "loss": 0.8151,
+      "loss": 0.8445,
       "step": 23912
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.778806122719308e-07,
-      "loss": 0.8092,
+      "loss": 0.7998,
       "step": 23913
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.773980836473373e-07,
-      "loss": 0.686,
+      "loss": 0.8333,
       "step": 23914
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.769156815859026e-07,
-      "loss": 0.7316,
+      "loss": 0.7914,
       "step": 23915
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.764334060943192e-07,
-      "loss": 0.6772,
+      "loss": 0.7688,
       "step": 23916
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.759512571792772e-07,
-      "loss": 0.7158,
+      "loss": 0.763,
       "step": 23917
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.754692348474647e-07,
-      "loss": 0.5344,
+      "loss": 0.714,
       "step": 23918
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.749873391055741e-07,
-      "loss": 0.6729,
+      "loss": 0.7925,
       "step": 23919
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.745055699602834e-07,
-      "loss": 0.6032,
+      "loss": 0.6535,
       "step": 23920
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.740239274182826e-07,
-      "loss": 0.8276,
+      "loss": 0.8515,
       "step": 23921
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.735424114862523e-07,
-      "loss": 0.6487,
+      "loss": 0.729,
       "step": 23922
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.730610221708713e-07,
-      "loss": 0.7977,
+      "loss": 0.9062,
       "step": 23923
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.725797594788176e-07,
-      "loss": 0.7676,
+      "loss": 0.8011,
       "step": 23924
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.720986234167716e-07,
-      "loss": 0.6702,
+      "loss": 0.7902,
       "step": 23925
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.716176139914067e-07,
-      "loss": 0.6025,
+      "loss": 0.6815,
       "step": 23926
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.711367312093955e-07,
-      "loss": 0.716,
+      "loss": 0.7144,
       "step": 23927
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.706559750774102e-07,
-      "loss": 0.6471,
+      "loss": 0.683,
       "step": 23928
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.7017534560212e-07,
-      "loss": 0.6919,
+      "loss": 0.8066,
       "step": 23929
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.696948427901931e-07,
-      "loss": 0.8625,
+      "loss": 0.9198,
       "step": 23930
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.69214466648296e-07,
-      "loss": 0.7749,
+      "loss": 0.8584,
       "step": 23931
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.687342171830937e-07,
-      "loss": 0.5974,
+      "loss": 0.702,
       "step": 23932
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.682540944012496e-07,
-      "loss": 0.7171,
+      "loss": 0.7739,
       "step": 23933
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.67774098309423e-07,
-      "loss": 0.7719,
+      "loss": 0.8574,
       "step": 23934
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.672942289142727e-07,
-      "loss": 0.7586,
+      "loss": 0.844,
       "step": 23935
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.668144862224593e-07,
-      "loss": 0.7225,
+      "loss": 0.6982,
       "step": 23936
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.663348702406337e-07,
-      "loss": 0.7789,
+      "loss": 0.7913,
       "step": 23937
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.658553809754567e-07,
-      "loss": 0.7901,
+      "loss": 0.7488,
       "step": 23938
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.653760184335724e-07,
-      "loss": 0.6927,
+      "loss": 0.8031,
       "step": 23939
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.648967826216382e-07,
-      "loss": 0.7593,
+      "loss": 0.8396,
       "step": 23940
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.644176735462995e-07,
-      "loss": 0.7517,
+      "loss": 0.8633,
       "step": 23941
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.639386912142045e-07,
-      "loss": 0.7459,
+      "loss": 0.7631,
       "step": 23942
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.634598356319956e-07,
-      "loss": 0.6238,
+      "loss": 0.7063,
       "step": 23943
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.62981106806321e-07,
-      "loss": 0.7399,
+      "loss": 0.7504,
       "step": 23944
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.625025047438185e-07,
-      "loss": 0.7224,
+      "loss": 0.8502,
       "step": 23945
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.620240294511295e-07,
-      "loss": 0.7086,
+      "loss": 0.829,
       "step": 23946
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.615456809348932e-07,
-      "loss": 0.6806,
+      "loss": 0.7822,
       "step": 23947
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.610674592017443e-07,
-      "loss": 0.7003,
+      "loss": 0.8255,
       "step": 23948
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.605893642583152e-07,
-      "loss": 0.7439,
+      "loss": 0.8024,
       "step": 23949
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.60111396111244e-07,
-      "loss": 0.6843,
+      "loss": 0.7926,
       "step": 23950
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.596335547671597e-07,
-      "loss": 0.7016,
+      "loss": 0.7549,
       "step": 23951
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.591558402326905e-07,
-      "loss": 0.7357,
+      "loss": 0.7831,
       "step": 23952
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.586782525144655e-07,
-      "loss": 0.7489,
+      "loss": 0.7827,
       "step": 23953
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.582007916191104e-07,
-      "loss": 0.7144,
+      "loss": 0.8145,
       "step": 23954
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.577234575532479e-07,
-      "loss": 0.7114,
+      "loss": 0.7936,
       "step": 23955
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.572462503235002e-07,
-      "loss": 0.801,
+      "loss": 0.8129,
       "step": 23956
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.5676916993649e-07,
-      "loss": 0.6907,
+      "loss": 0.7613,
       "step": 23957
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.562922163988352e-07,
-      "loss": 0.7302,
+      "loss": 0.825,
       "step": 23958
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.558153897171529e-07,
-      "loss": 0.7242,
+      "loss": 0.8423,
       "step": 23959
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.553386898980576e-07,
-      "loss": 0.7204,
+      "loss": 0.8233,
       "step": 23960
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.548621169481641e-07,
-      "loss": 0.8221,
+      "loss": 0.9028,
       "step": 23961
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.543856708740816e-07,
-      "loss": 0.6033,
+      "loss": 0.6805,
       "step": 23962
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.539093516824238e-07,
-      "loss": 0.7633,
+      "loss": 0.8557,
       "step": 23963
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.534331593797962e-07,
-      "loss": 0.7384,
+      "loss": 0.6925,
       "step": 23964
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.529570939728071e-07,
-      "loss": 0.7001,
+      "loss": 0.8117,
       "step": 23965
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.52481155468059e-07,
-      "loss": 0.7495,
+      "loss": 0.8761,
       "step": 23966
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.520053438721588e-07,
-      "loss": 0.7999,
+      "loss": 0.8141,
       "step": 23967
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.515296591917021e-07,
-      "loss": 0.6825,
+      "loss": 0.7844,
       "step": 23968
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.51054101433293e-07,
-      "loss": 0.7265,
+      "loss": 0.7488,
       "step": 23969
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.505786706035279e-07,
-      "loss": 0.7905,
+      "loss": 0.8947,
       "step": 23970
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.50103366709002e-07,
-      "loss": 0.7793,
+      "loss": 0.8328,
       "step": 23971
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.496281897563096e-07,
-      "loss": 0.6467,
+      "loss": 0.7729,
       "step": 23972
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.491531397520447e-07,
-      "loss": 0.7144,
+      "loss": 0.7556,
       "step": 23973
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.486782167027952e-07,
-      "loss": 0.7504,
+      "loss": 0.8073,
       "step": 23974
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.482034206151502e-07,
-      "loss": 0.6906,
+      "loss": 0.7331,
       "step": 23975
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.477287514957e-07,
-      "loss": 0.7087,
+      "loss": 0.8165,
       "step": 23976
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.472542093510283e-07,
-      "loss": 0.8135,
+      "loss": 0.8831,
       "step": 23977
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.467797941877176e-07,
-      "loss": 0.8155,
+      "loss": 0.8549,
       "step": 23978
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.463055060123493e-07,
-      "loss": 0.6591,
+      "loss": 0.8639,
       "step": 23979
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.458313448315081e-07,
-      "loss": 0.6537,
+      "loss": 0.7727,
       "step": 23980
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.453573106517654e-07,
-      "loss": 0.7535,
+      "loss": 0.8223,
       "step": 23981
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.448834034797027e-07,
-      "loss": 0.738,
+      "loss": 0.7568,
       "step": 23982
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.444096233218935e-07,
-      "loss": 0.7841,
+      "loss": 0.8676,
       "step": 23983
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.439359701849114e-07,
-      "loss": 0.7359,
+      "loss": 0.8069,
       "step": 23984
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.434624440753247e-07,
-      "loss": 0.6825,
+      "loss": 0.817,
       "step": 23985
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.42989044999708e-07,
-      "loss": 0.6957,
+      "loss": 0.834,
       "step": 23986
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.425157729646238e-07,
-      "loss": 0.623,
+      "loss": 0.6866,
       "step": 23987
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.420426279766414e-07,
-      "loss": 0.678,
+      "loss": 0.7066,
       "step": 23988
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.415696100423243e-07,
-      "loss": 0.723,
+      "loss": 0.7885,
       "step": 23989
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.410967191682351e-07,
-      "loss": 0.8178,
+      "loss": 0.8545,
       "step": 23990
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.406239553609341e-07,
-      "loss": 0.8423,
+      "loss": 0.877,
       "step": 23991
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.401513186269794e-07,
-      "loss": 0.7659,
+      "loss": 0.7892,
       "step": 23992
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.396788089729302e-07,
-      "loss": 0.7318,
+      "loss": 0.8125,
       "step": 23993
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.39206426405339e-07,
-      "loss": 0.6953,
+      "loss": 0.8266,
       "step": 23994
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.387341709307617e-07,
-      "loss": 0.7003,
+      "loss": 0.7853,
       "step": 23995
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.382620425557508e-07,
-      "loss": 0.8529,
+      "loss": 0.8818,
       "step": 23996
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.377900412868545e-07,
-      "loss": 0.6844,
+      "loss": 0.7254,
       "step": 23997
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.373181671306208e-07,
-      "loss": 0.7472,
+      "loss": 0.7356,
       "step": 23998
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.368464200935999e-07,
-      "loss": 0.6822,
+      "loss": 0.7485,
       "step": 23999
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.363748001823313e-07,
-      "loss": 0.7382,
+      "loss": 0.7831,
       "step": 24000
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.359033074033629e-07,
-      "loss": 0.6115,
+      "loss": 0.6999,
       "step": 24001
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.35431941763234e-07,
-      "loss": 0.6399,
+      "loss": 0.642,
       "step": 24002
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.349607032684836e-07,
-      "loss": 0.5095,
+      "loss": 0.6985,
       "step": 24003
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.344895919256479e-07,
-      "loss": 0.8078,
+      "loss": 0.867,
       "step": 24004
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.340186077412693e-07,
-      "loss": 0.6245,
+      "loss": 0.768,
       "step": 24005
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.335477507218737e-07,
-      "loss": 0.644,
+      "loss": 0.7183,
       "step": 24006
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.330770208739991e-07,
-      "loss": 0.8477,
+      "loss": 0.897,
       "step": 24007
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.326064182041748e-07,
-      "loss": 0.6071,
+      "loss": 0.7759,
       "step": 24008
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.321359427189302e-07,
-      "loss": 0.8005,
+      "loss": 0.8566,
       "step": 24009
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.316655944247909e-07,
-      "loss": 0.785,
+      "loss": 0.8402,
       "step": 24010
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.311953733282818e-07,
-      "loss": 0.8007,
+      "loss": 0.8092,
       "step": 24011
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.307252794359321e-07,
-      "loss": 0.7659,
+      "loss": 0.851,
       "step": 24012
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.302553127542567e-07,
-      "loss": 0.5977,
+      "loss": 0.7275,
       "step": 24013
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.297854732897792e-07,
-      "loss": 0.666,
+      "loss": 0.7307,
       "step": 24014
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.293157610490176e-07,
-      "loss": 0.6881,
+      "loss": 0.6835,
       "step": 24015
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.288461760384892e-07,
-      "loss": 0.6886,
+      "loss": 0.7864,
       "step": 24016
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.283767182647051e-07,
-      "loss": 0.808,
+      "loss": 0.906,
       "step": 24017
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.279073877341859e-07,
-      "loss": 0.7382,
+      "loss": 0.8368,
       "step": 24018
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.27438184453433e-07,
-      "loss": 0.6342,
+      "loss": 0.7602,
       "step": 24019
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.269691084289644e-07,
-      "loss": 0.7632,
+      "loss": 0.7493,
       "step": 24020
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.26500159667284e-07,
-      "loss": 0.6526,
+      "loss": 0.747,
       "step": 24021
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.260313381748985e-07,
-      "loss": 0.6546,
+      "loss": 0.8112,
       "step": 24022
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.255626439583098e-07,
-      "loss": 0.7475,
+      "loss": 0.7754,
       "step": 24023
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.250940770240246e-07,
-      "loss": 0.7774,
+      "loss": 0.8553,
       "step": 24024
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.246256373785421e-07,
-      "loss": 0.6688,
+      "loss": 0.7796,
       "step": 24025
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.241573250283608e-07,
-      "loss": 0.7962,
+      "loss": 0.8127,
       "step": 24026
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.236891399799773e-07,
-      "loss": 0.622,
+      "loss": 0.7617,
       "step": 24027
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.232210822398878e-07,
-      "loss": 0.6733,
+      "loss": 0.7735,
       "step": 24028
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.227531518145848e-07,
-      "loss": 0.7377,
+      "loss": 0.814,
       "step": 24029
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.222853487105598e-07,
-      "loss": 0.6942,
+      "loss": 0.8226,
       "step": 24030
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.218176729343075e-07,
-      "loss": 0.6563,
+      "loss": 0.8124,
       "step": 24031
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.213501244923094e-07,
-      "loss": 0.8145,
+      "loss": 0.8295,
       "step": 24032
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.208827033910583e-07,
-      "loss": 0.723,
+      "loss": 0.8126,
       "step": 24033
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.204154096370354e-07,
-      "loss": 0.7643,
+      "loss": 0.8491,
       "step": 24034
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.199482432367245e-07,
-      "loss": 0.724,
+      "loss": 0.8265,
       "step": 24035
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.19481204196605e-07,
-      "loss": 0.8168,
+      "loss": 0.8743,
       "step": 24036
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.190142925231614e-07,
-      "loss": 0.7757,
+      "loss": 0.8013,
       "step": 24037
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.185475082228655e-07,
-      "loss": 0.7305,
+      "loss": 0.7766,
       "step": 24038
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.180808513021987e-07,
-      "loss": 0.7333,
+      "loss": 0.7778,
       "step": 24039
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.176143217676313e-07,
-      "loss": 0.7606,
+      "loss": 0.8037,
       "step": 24040
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.171479196256382e-07,
-      "loss": 0.7405,
+      "loss": 0.8572,
       "step": 24041
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.166816448826864e-07,
-      "loss": 0.7681,
+      "loss": 0.8546,
       "step": 24042
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.162154975452486e-07,
-      "loss": 0.767,
+      "loss": 0.7666,
       "step": 24043
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.157494776197916e-07,
-      "loss": 0.7055,
+      "loss": 0.7412,
       "step": 24044
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.152835851127805e-07,
-      "loss": 0.7316,
+      "loss": 0.7541,
       "step": 24045
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.148178200306756e-07,
-      "loss": 0.8288,
+      "loss": 0.8746,
       "step": 24046
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.14352182379945e-07,
-      "loss": 0.6268,
+      "loss": 0.745,
       "step": 24047
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.138866721670413e-07,
-      "loss": 0.6718,
+      "loss": 0.8428,
       "step": 24048
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.134212893984284e-07,
-      "loss": 0.5377,
+      "loss": 0.6832,
       "step": 24049
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.129560340805632e-07,
-      "loss": 0.733,
+      "loss": 0.7505,
       "step": 24050
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.124909062198949e-07,
-      "loss": 0.5688,
+      "loss": 0.7382,
       "step": 24051
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.120259058228807e-07,
-      "loss": 0.7198,
+      "loss": 0.7968,
       "step": 24052
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.115610328959722e-07,
-      "loss": 0.7545,
+      "loss": 0.7537,
       "step": 24053
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.110962874456174e-07,
-      "loss": 0.6565,
+      "loss": 0.7685,
       "step": 24054
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.106316694782623e-07,
-      "loss": 0.8085,
+      "loss": 0.8408,
       "step": 24055
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.101671790003562e-07,
-      "loss": 0.6503,
+      "loss": 0.7362,
       "step": 24056
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.097028160183429e-07,
-      "loss": 0.774,
+      "loss": 0.8985,
       "step": 24057
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.092385805386627e-07,
-      "loss": 0.6399,
+      "loss": 0.804,
       "step": 24058
     },
     {
       "epoch": 0.87,
       "learning_rate": 8.087744725677571e-07,
-      "loss": 0.6792,
+      "loss": 0.7595,
       "step": 24059
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.083104921120654e-07,
-      "loss": 0.7376,
+      "loss": 0.8093,
       "step": 24060
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.078466391780226e-07,
-      "loss": 0.626,
+      "loss": 0.6978,
       "step": 24061
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.073829137720679e-07,
-      "loss": 0.6857,
+      "loss": 0.8759,
       "step": 24062
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.069193159006328e-07,
-      "loss": 0.6908,
+      "loss": 0.8055,
       "step": 24063
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.064558455701477e-07,
-      "loss": 0.6743,
+      "loss": 0.7831,
       "step": 24064
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.05992502787043e-07,
-      "loss": 0.6465,
+      "loss": 0.7825,
       "step": 24065
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.055292875577514e-07,
-      "loss": 0.7039,
+      "loss": 0.8329,
       "step": 24066
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.050661998886911e-07,
-      "loss": 0.8921,
+      "loss": 0.8509,
       "step": 24067
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.046032397862946e-07,
-      "loss": 0.7631,
+      "loss": 0.8273,
       "step": 24068
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.041404072569803e-07,
-      "loss": 0.6143,
+      "loss": 0.7687,
       "step": 24069
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.036777023071707e-07,
-      "loss": 0.7027,
+      "loss": 0.7644,
       "step": 24070
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.032151249432863e-07,
-      "loss": 0.7503,
+      "loss": 0.8058,
       "step": 24071
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.027526751717418e-07,
-      "loss": 0.7437,
+      "loss": 0.7969,
       "step": 24072
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.022903529989567e-07,
-      "loss": 0.7849,
+      "loss": 0.7909,
       "step": 24073
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.018281584313403e-07,
-      "loss": 0.6965,
+      "loss": 0.7187,
       "step": 24074
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.013660914753085e-07,
-      "loss": 0.7469,
+      "loss": 0.7975,
       "step": 24075
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.009041521372729e-07,
-      "loss": 0.7732,
+      "loss": 0.7662,
       "step": 24076
     },
     {
       "epoch": 0.88,
       "learning_rate": 8.004423404236395e-07,
-      "loss": 0.7844,
+      "loss": 0.8467,
       "step": 24077
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.999806563408141e-07,
-      "loss": 0.734,
+      "loss": 0.7943,
       "step": 24078
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.995190998952074e-07,
-      "loss": 0.7057,
+      "loss": 0.7727,
       "step": 24079
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.990576710932163e-07,
-      "loss": 0.7357,
+      "loss": 0.8248,
       "step": 24080
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.98596369941248e-07,
-      "loss": 0.5366,
+      "loss": 0.7032,
       "step": 24081
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.981351964456985e-07,
-      "loss": 0.7445,
+      "loss": 0.7977,
       "step": 24082
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.976741506129681e-07,
-      "loss": 0.7438,
+      "loss": 0.7995,
       "step": 24083
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.972132324494508e-07,
-      "loss": 0.7156,
+      "loss": 0.8639,
       "step": 24084
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.967524419615469e-07,
-      "loss": 0.7248,
+      "loss": 0.7078,
       "step": 24085
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.962917791556413e-07,
-      "loss": 0.762,
+      "loss": 0.7964,
       "step": 24086
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.958312440381299e-07,
-      "loss": 0.6377,
+      "loss": 0.7514,
       "step": 24087
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.953708366154023e-07,
-      "loss": 0.638,
+      "loss": 0.7138,
       "step": 24088
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.949105568938442e-07,
-      "loss": 0.7165,
+      "loss": 0.8333,
       "step": 24089
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.944504048798407e-07,
-      "loss": 0.7731,
+      "loss": 0.8086,
       "step": 24090
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.939903805797777e-07,
-      "loss": 0.8367,
+      "loss": 0.8389,
       "step": 24091
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.935304840000369e-07,
-      "loss": 0.6553,
+      "loss": 0.7566,
       "step": 24092
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.930707151469963e-07,
-      "loss": 0.722,
+      "loss": 0.7807,
       "step": 24093
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.926110740270388e-07,
-      "loss": 0.5578,
+      "loss": 0.7126,
       "step": 24094
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.92151560646538e-07,
-      "loss": 0.7702,
+      "loss": 0.8432,
       "step": 24095
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.916921750118711e-07,
-      "loss": 0.7732,
+      "loss": 0.8285,
       "step": 24096
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.912329171294087e-07,
-      "loss": 0.6277,
+      "loss": 0.7347,
       "step": 24097
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.907737870055266e-07,
-      "loss": 0.5571,
+      "loss": 0.6662,
       "step": 24098
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.903147846465887e-07,
-      "loss": 0.6269,
+      "loss": 0.6964,
       "step": 24099
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.898559100589686e-07,
-      "loss": 0.6757,
+      "loss": 0.789,
       "step": 24100
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.893971632490304e-07,
-      "loss": 0.6531,
+      "loss": 0.793,
       "step": 24101
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.889385442231378e-07,
-      "loss": 0.8375,
+      "loss": 0.8783,
       "step": 24102
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.884800529876524e-07,
-      "loss": 0.7226,
+      "loss": 0.7068,
       "step": 24103
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.880216895489412e-07,
-      "loss": 0.7066,
+      "loss": 0.7807,
       "step": 24104
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.875634539133548e-07,
-      "loss": 0.7903,
+      "loss": 0.7225,
       "step": 24105
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.871053460872558e-07,
-      "loss": 0.7722,
+      "loss": 0.8182,
       "step": 24106
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.866473660770002e-07,
-      "loss": 0.6556,
+      "loss": 0.7624,
       "step": 24107
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.861895138889398e-07,
-      "loss": 0.762,
+      "loss": 0.8877,
       "step": 24108
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.85731789529427e-07,
-      "loss": 0.8633,
+      "loss": 0.8252,
       "step": 24109
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.852741930048102e-07,
-      "loss": 0.7127,
+      "loss": 0.8017,
       "step": 24110
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.848167243214433e-07,
-      "loss": 0.7499,
+      "loss": 0.8257,
       "step": 24111
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.843593834856677e-07,
-      "loss": 0.7504,
+      "loss": 0.8244,
       "step": 24112
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.839021705038308e-07,
-      "loss": 0.8139,
+      "loss": 0.9125,
       "step": 24113
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.834450853822762e-07,
-      "loss": 0.7766,
+      "loss": 0.8531,
       "step": 24114
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.829881281273432e-07,
-      "loss": 0.86,
+      "loss": 0.8621,
       "step": 24115
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.825312987453715e-07,
-      "loss": 0.7947,
+      "loss": 0.8098,
       "step": 24116
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.820745972427047e-07,
-      "loss": 0.7648,
+      "loss": 0.8205,
       "step": 24117
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.816180236256699e-07,
-      "loss": 0.6428,
+      "loss": 0.7448,
       "step": 24118
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.811615779006077e-07,
-      "loss": 0.7596,
+      "loss": 0.7185,
       "step": 24119
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.807052600738485e-07,
-      "loss": 0.8916,
+      "loss": 0.8811,
       "step": 24120
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.80249070151724e-07,
-      "loss": 0.6666,
+      "loss": 0.6662,
       "step": 24121
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.797930081405602e-07,
-      "loss": 0.5758,
+      "loss": 0.6177,
       "step": 24122
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.793370740466888e-07,
-      "loss": 0.7009,
+      "loss": 0.7287,
       "step": 24123
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.788812678764324e-07,
-      "loss": 0.7052,
+      "loss": 0.7324,
       "step": 24124
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.784255896361159e-07,
-      "loss": 0.7323,
+      "loss": 0.8302,
       "step": 24125
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.77970039332061e-07,
-      "loss": 0.6577,
+      "loss": 0.7722,
       "step": 24126
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.775146169705871e-07,
-      "loss": 0.6502,
+      "loss": 0.8665,
       "step": 24127
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.770593225580126e-07,
-      "loss": 0.8438,
+      "loss": 0.8657,
       "step": 24128
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.766041561006532e-07,
-      "loss": 0.6566,
+      "loss": 0.7235,
       "step": 24129
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.761491176048274e-07,
-      "loss": 0.8124,
+      "loss": 0.8846,
       "step": 24130
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.756942070768437e-07,
-      "loss": 0.7158,
+      "loss": 0.8199,
       "step": 24131
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.752394245230155e-07,
-      "loss": 0.7087,
+      "loss": 0.7977,
       "step": 24132
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.747847699496525e-07,
-      "loss": 0.8086,
+      "loss": 0.8686,
       "step": 24133
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.743302433630628e-07,
-      "loss": 0.7036,
+      "loss": 0.831,
       "step": 24134
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.738758447695494e-07,
-      "loss": 0.5787,
+      "loss": 0.6222,
       "step": 24135
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.734215741754214e-07,
-      "loss": 0.5953,
+      "loss": 0.7554,
       "step": 24136
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.729674315869751e-07,
-      "loss": 0.7614,
+      "loss": 0.8335,
       "step": 24137
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.725134170105153e-07,
-      "loss": 0.7141,
+      "loss": 0.7923,
       "step": 24138
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.720595304523404e-07,
-      "loss": 0.7269,
+      "loss": 0.7047,
       "step": 24139
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.716057719187464e-07,
-      "loss": 0.7213,
+      "loss": 0.7839,
       "step": 24140
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.711521414160283e-07,
-      "loss": 0.8068,
+      "loss": 0.8456,
       "step": 24141
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.706986389504811e-07,
-      "loss": 0.6761,
+      "loss": 0.7944,
       "step": 24142
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.702452645283953e-07,
-      "loss": 0.5925,
+      "loss": 0.7301,
       "step": 24143
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.697920181560615e-07,
-      "loss": 0.7399,
+      "loss": 0.7752,
       "step": 24144
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.693388998397677e-07,
-      "loss": 0.6352,
+      "loss": 0.7678,
       "step": 24145
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.688859095858003e-07,
-      "loss": 0.6034,
+      "loss": 0.6584,
       "step": 24146
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.684330474004431e-07,
-      "loss": 0.7788,
+      "loss": 0.8157,
       "step": 24147
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.679803132899776e-07,
-      "loss": 0.6839,
+      "loss": 0.6854,
       "step": 24148
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.675277072606902e-07,
-      "loss": 0.708,
+      "loss": 0.7603,
       "step": 24149
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.670752293188544e-07,
-      "loss": 0.6047,
+      "loss": 0.6932,
       "step": 24150
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.666228794707508e-07,
-      "loss": 0.5001,
+      "loss": 0.6908,
       "step": 24151
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.661706577226535e-07,
-      "loss": 0.7541,
+      "loss": 0.7732,
       "step": 24152
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.657185640808384e-07,
-      "loss": 0.7154,
+      "loss": 0.7178,
       "step": 24153
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.65266598551575e-07,
-      "loss": 0.723,
+      "loss": 0.7873,
       "step": 24154
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.648147611411361e-07,
-      "loss": 0.7864,
+      "loss": 0.84,
       "step": 24155
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.643630518557898e-07,
-      "loss": 0.7708,
+      "loss": 0.8675,
       "step": 24156
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.639114707018025e-07,
-      "loss": 0.6859,
+      "loss": 0.8139,
       "step": 24157
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.63460017685439e-07,
-      "loss": 0.7677,
+      "loss": 0.8945,
       "step": 24158
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.630086928129632e-07,
-      "loss": 0.7098,
+      "loss": 0.7448,
       "step": 24159
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.625574960906335e-07,
-      "loss": 0.7027,
+      "loss": 0.7167,
       "step": 24160
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.621064275247148e-07,
-      "loss": 0.7007,
+      "loss": 0.7657,
       "step": 24161
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.616554871214621e-07,
-      "loss": 0.7054,
+      "loss": 0.7372,
       "step": 24162
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.612046748871327e-07,
-      "loss": 0.7139,
+      "loss": 0.7897,
       "step": 24163
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.607539908279804e-07,
-      "loss": 0.6933,
+      "loss": 0.7827,
       "step": 24164
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.60303434950257e-07,
-      "loss": 0.6786,
+      "loss": 0.856,
       "step": 24165
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.59853007260215e-07,
-      "loss": 0.6894,
+      "loss": 0.7614,
       "step": 24166
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.594027077640997e-07,
-      "loss": 0.7579,
+      "loss": 0.7969,
       "step": 24167
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.589525364681627e-07,
-      "loss": 0.7619,
+      "loss": 0.7814,
       "step": 24168
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.585024933786489e-07,
-      "loss": 0.6295,
+      "loss": 0.7191,
       "step": 24169
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.580525785018012e-07,
-      "loss": 0.6352,
+      "loss": 0.701,
       "step": 24170
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.5760279184386e-07,
-      "loss": 0.7295,
+      "loss": 0.7993,
       "step": 24171
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.571531334110682e-07,
-      "loss": 0.7957,
+      "loss": 0.866,
       "step": 24172
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.567036032096598e-07,
-      "loss": 0.6622,
+      "loss": 0.7027,
       "step": 24173
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.562542012458773e-07,
-      "loss": 0.7916,
+      "loss": 0.8248,
       "step": 24174
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.558049275259516e-07,
-      "loss": 0.6871,
+      "loss": 0.8321,
       "step": 24175
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.553557820561175e-07,
-      "loss": 0.8385,
+      "loss": 0.7961,
       "step": 24176
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.549067648426034e-07,
-      "loss": 0.766,
+      "loss": 0.8262,
       "step": 24177
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.544578758916443e-07,
-      "loss": 0.8105,
+      "loss": 0.8285,
       "step": 24178
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.540091152094609e-07,
-      "loss": 0.755,
+      "loss": 0.8133,
       "step": 24179
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.535604828022847e-07,
-      "loss": 0.883,
+      "loss": 0.8793,
       "step": 24180
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.531119786763375e-07,
-      "loss": 0.7809,
+      "loss": 0.7915,
       "step": 24181
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.526636028378431e-07,
-      "loss": 0.7867,
+      "loss": 0.8195,
       "step": 24182
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.5221535529302e-07,
-      "loss": 0.7445,
+      "loss": 0.7994,
       "step": 24183
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.517672360480876e-07,
-      "loss": 0.7017,
+      "loss": 0.7959,
       "step": 24184
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.513192451092633e-07,
-      "loss": 0.7857,
+      "loss": 0.8567,
       "step": 24185
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.508713824827618e-07,
-      "loss": 0.8221,
+      "loss": 0.9132,
       "step": 24186
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.504236481747973e-07,
-      "loss": 0.6779,
+      "loss": 0.7256,
       "step": 24187
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.499760421915825e-07,
-      "loss": 0.7404,
+      "loss": 0.789,
       "step": 24188
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.495285645393247e-07,
-      "loss": 0.7456,
+      "loss": 0.8595,
       "step": 24189
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.490812152242332e-07,
-      "loss": 0.7667,
+      "loss": 0.8716,
       "step": 24190
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.486339942525167e-07,
-      "loss": 0.6416,
+      "loss": 0.6663,
       "step": 24191
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.481869016303744e-07,
-      "loss": 0.6024,
+      "loss": 0.7346,
       "step": 24192
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.477399373640148e-07,
-      "loss": 0.6648,
+      "loss": 0.8147,
       "step": 24193
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.472931014596352e-07,
-      "loss": 0.6154,
+      "loss": 0.7065,
       "step": 24194
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.468463939234361e-07,
-      "loss": 0.7253,
+      "loss": 0.8064,
       "step": 24195
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.463998147616136e-07,
-      "loss": 0.6643,
+      "loss": 0.7626,
       "step": 24196
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.459533639803673e-07,
-      "loss": 0.7044,
+      "loss": 0.7978,
       "step": 24197
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.455070415858856e-07,
-      "loss": 0.8609,
+      "loss": 0.863,
       "step": 24198
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.450608475843646e-07,
-      "loss": 0.6628,
+      "loss": 0.7432,
       "step": 24199
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.446147819819938e-07,
-      "loss": 0.7248,
+      "loss": 0.7953,
       "step": 24200
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.441688447849604e-07,
-      "loss": 0.7917,
+      "loss": 0.7926,
       "step": 24201
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.437230359994529e-07,
-      "loss": 0.7667,
+      "loss": 0.7968,
       "step": 24202
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.432773556316541e-07,
-      "loss": 0.7023,
+      "loss": 0.8112,
       "step": 24203
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.428318036877491e-07,
-      "loss": 0.7224,
+      "loss": 0.8062,
       "step": 24204
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.423863801739173e-07,
-      "loss": 0.6861,
+      "loss": 0.7374,
       "step": 24205
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.419410850963415e-07,
-      "loss": 0.6261,
+      "loss": 0.8071,
       "step": 24206
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.41495918461197e-07,
-      "loss": 0.7565,
+      "loss": 0.8126,
       "step": 24207
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.410508802746608e-07,
-      "loss": 0.6701,
+      "loss": 0.7102,
       "step": 24208
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.406059705429047e-07,
-      "loss": 0.7667,
+      "loss": 0.8776,
       "step": 24209
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.401611892721073e-07,
-      "loss": 0.7139,
+      "loss": 0.7841,
       "step": 24210
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.397165364684322e-07,
-      "loss": 0.781,
+      "loss": 0.7793,
       "step": 24211
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.392720121380526e-07,
-      "loss": 0.7239,
+      "loss": 0.7648,
       "step": 24212
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.388276162871344e-07,
-      "loss": 0.7681,
+      "loss": 0.7742,
       "step": 24213
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.383833489218439e-07,
-      "loss": 0.7753,
+      "loss": 0.7952,
       "step": 24214
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.379392100483418e-07,
-      "loss": 0.7719,
+      "loss": 0.8016,
       "step": 24215
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.374951996727963e-07,
-      "loss": 0.6685,
+      "loss": 0.7262,
       "step": 24216
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.370513178013583e-07,
-      "loss": 0.6619,
+      "loss": 0.7979,
       "step": 24217
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.366075644401938e-07,
-      "loss": 0.7854,
+      "loss": 0.774,
       "step": 24218
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.361639395954556e-07,
-      "loss": 0.6869,
+      "loss": 0.7789,
       "step": 24219
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.357204432732989e-07,
-      "loss": 0.8,
+      "loss": 0.8445,
       "step": 24220
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.352770754798777e-07,
-      "loss": 0.8068,
+      "loss": 0.8966,
       "step": 24221
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.348338362213403e-07,
-      "loss": 0.6262,
+      "loss": 0.7786,
       "step": 24222
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.343907255038408e-07,
-      "loss": 0.7962,
+      "loss": 0.8724,
       "step": 24223
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.339477433335196e-07,
-      "loss": 0.6587,
+      "loss": 0.7241,
       "step": 24224
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.335048897165297e-07,
-      "loss": 0.6599,
+      "loss": 0.7066,
       "step": 24225
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.33062164659012e-07,
-      "loss": 0.6757,
+      "loss": 0.7902,
       "step": 24226
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.326195681671078e-07,
-      "loss": 0.6346,
+      "loss": 0.6716,
       "step": 24227
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.321771002469569e-07,
-      "loss": 0.7422,
+      "loss": 0.8238,
       "step": 24228
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.317347609047032e-07,
-      "loss": 0.7765,
+      "loss": 0.821,
       "step": 24229
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.312925501464763e-07,
-      "loss": 0.7398,
+      "loss": 0.798,
       "step": 24230
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.308504679784157e-07,
-      "loss": 0.7274,
+      "loss": 0.7411,
       "step": 24231
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.304085144066542e-07,
-      "loss": 0.7939,
+      "loss": 0.8361,
       "step": 24232
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.299666894373236e-07,
-      "loss": 0.8411,
+      "loss": 0.8583,
       "step": 24233
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.295249930765503e-07,
-      "loss": 0.7084,
+      "loss": 0.7254,
       "step": 24234
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.290834253304657e-07,
-      "loss": 0.6149,
+      "loss": 0.6017,
       "step": 24235
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.286419862051953e-07,
-      "loss": 0.784,
+      "loss": 0.8208,
       "step": 24236
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.282006757068627e-07,
-      "loss": 0.9137,
+      "loss": 0.8721,
       "step": 24237
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.27759493841591e-07,
-      "loss": 0.7635,
+      "loss": 0.7457,
       "step": 24238
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.273184406155009e-07,
-      "loss": 0.8445,
+      "loss": 0.8603,
       "step": 24239
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.268775160347119e-07,
-      "loss": 0.8408,
+      "loss": 0.9002,
       "step": 24240
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.264367201053379e-07,
-      "loss": 0.6353,
+      "loss": 0.7709,
       "step": 24241
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.259960528335008e-07,
-      "loss": 0.7128,
+      "loss": 0.7273,
       "step": 24242
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.255555142253068e-07,
-      "loss": 0.6417,
+      "loss": 0.7797,
       "step": 24243
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.251151042868731e-07,
-      "loss": 0.6389,
+      "loss": 0.7835,
       "step": 24244
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.246748230243073e-07,
-      "loss": 0.7736,
+      "loss": 0.7594,
       "step": 24245
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.242346704437198e-07,
-      "loss": 0.6359,
+      "loss": 0.7444,
       "step": 24246
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.237946465512124e-07,
-      "loss": 0.6878,
+      "loss": 0.7432,
       "step": 24247
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.233547513528971e-07,
-      "loss": 0.7659,
+      "loss": 0.7893,
       "step": 24248
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.2291498485487e-07,
-      "loss": 0.7666,
+      "loss": 0.7741,
       "step": 24249
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.224753470632362e-07,
-      "loss": 0.6955,
+      "loss": 0.7141,
       "step": 24250
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.220358379840931e-07,
-      "loss": 0.8825,
+      "loss": 0.8578,
       "step": 24251
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.215964576235401e-07,
-      "loss": 0.6905,
+      "loss": 0.7583,
       "step": 24252
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.211572059876703e-07,
-      "loss": 0.7663,
+      "loss": 0.8075,
       "step": 24253
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.2071808308258e-07,
-      "loss": 0.698,
+      "loss": 0.6881,
       "step": 24254
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.202790889143619e-07,
-      "loss": 0.6474,
+      "loss": 0.7239,
       "step": 24255
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.198402234891044e-07,
-      "loss": 0.6518,
+      "loss": 0.7368,
       "step": 24256
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.194014868128973e-07,
-      "loss": 0.7238,
+      "loss": 0.7675,
       "step": 24257
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.189628788918268e-07,
-      "loss": 0.6984,
+      "loss": 0.7591,
       "step": 24258
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.185243997319791e-07,
-      "loss": 0.6166,
+      "loss": 0.8286,
       "step": 24259
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.180860493394337e-07,
-      "loss": 0.7124,
+      "loss": 0.8257,
       "step": 24260
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.176478277202792e-07,
-      "loss": 0.7662,
+      "loss": 0.804,
       "step": 24261
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.172097348805874e-07,
-      "loss": 0.7513,
+      "loss": 0.7821,
       "step": 24262
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.167717708264422e-07,
-      "loss": 0.5904,
+      "loss": 0.7405,
       "step": 24263
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.163339355639176e-07,
-      "loss": 0.5872,
+      "loss": 0.6889,
       "step": 24264
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.158962290990878e-07,
-      "loss": 0.7092,
+      "loss": 0.7239,
       "step": 24265
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.154586514380235e-07,
-      "loss": 0.8057,
+      "loss": 0.8165,
       "step": 24266
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.150212025867987e-07,
-      "loss": 0.8337,
+      "loss": 0.8012,
       "step": 24267
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.145838825514817e-07,
-      "loss": 0.844,
+      "loss": 0.87,
       "step": 24268
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.141466913381378e-07,
-      "loss": 0.6772,
+      "loss": 0.7552,
       "step": 24269
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.137096289528345e-07,
-      "loss": 0.704,
+      "loss": 0.7825,
       "step": 24270
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.132726954016356e-07,
-      "loss": 0.7187,
+      "loss": 0.8063,
       "step": 24271
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.128358906905986e-07,
-      "loss": 0.6449,
+      "loss": 0.736,
       "step": 24272
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.123992148257896e-07,
-      "loss": 0.63,
+      "loss": 0.7896,
       "step": 24273
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.119626678132629e-07,
-      "loss": 0.8164,
+      "loss": 0.9045,
       "step": 24274
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.11526249659078e-07,
-      "loss": 0.6418,
+      "loss": 0.7518,
       "step": 24275
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.110899603692845e-07,
-      "loss": 0.5248,
+      "loss": 0.7381,
       "step": 24276
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.106537999499429e-07,
-      "loss": 0.7619,
+      "loss": 0.8335,
       "step": 24277
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.102177684070977e-07,
-      "loss": 0.8472,
+      "loss": 0.8679,
       "step": 24278
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.097818657467992e-07,
-      "loss": 0.7643,
+      "loss": 0.7567,
       "step": 24279
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.093460919750994e-07,
-      "loss": 0.6315,
+      "loss": 0.7549,
       "step": 24280
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.089104470980368e-07,
-      "loss": 0.6999,
+      "loss": 0.7408,
       "step": 24281
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.084749311216621e-07,
-      "loss": 0.7711,
+      "loss": 0.7628,
       "step": 24282
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.080395440520138e-07,
-      "loss": 0.7564,
+      "loss": 0.7541,
       "step": 24283
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.076042858951349e-07,
-      "loss": 0.6689,
+      "loss": 0.7494,
       "step": 24284
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.071691566570593e-07,
-      "loss": 0.844,
+      "loss": 0.8715,
       "step": 24285
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.067341563438291e-07,
-      "loss": 0.891,
+      "loss": 0.8922,
       "step": 24286
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.062992849614769e-07,
-      "loss": 0.6805,
+      "loss": 0.8306,
       "step": 24287
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.05864542516036e-07,
-      "loss": 0.6638,
+      "loss": 0.7645,
       "step": 24288
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.054299290135358e-07,
-      "loss": 0.7796,
+      "loss": 0.8588,
       "step": 24289
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.049954444600115e-07,
-      "loss": 0.7934,
+      "loss": 0.7908,
       "step": 24290
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.045610888614851e-07,
-      "loss": 0.7401,
+      "loss": 0.7432,
       "step": 24291
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.041268622239861e-07,
-      "loss": 0.7597,
+      "loss": 0.7914,
       "step": 24292
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.036927645535385e-07,
-      "loss": 0.697,
+      "loss": 0.7196,
       "step": 24293
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.032587958561632e-07,
-      "loss": 0.7005,
+      "loss": 0.8478,
       "step": 24294
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.028249561378797e-07,
-      "loss": 0.7564,
+      "loss": 0.8323,
       "step": 24295
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.023912454047132e-07,
-      "loss": 0.654,
+      "loss": 0.767,
       "step": 24296
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.019576636626734e-07,
-      "loss": 0.6398,
+      "loss": 0.7287,
       "step": 24297
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.0152421091778e-07,
-      "loss": 0.6741,
+      "loss": 0.7983,
       "step": 24298
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.010908871760458e-07,
-      "loss": 0.75,
+      "loss": 0.835,
       "step": 24299
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.006576924434815e-07,
-      "loss": 0.5698,
+      "loss": 0.6826,
       "step": 24300
     },
     {
       "epoch": 0.88,
       "learning_rate": 7.002246267260982e-07,
-      "loss": 0.7835,
+      "loss": 0.8189,
       "step": 24301
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.997916900299029e-07,
-      "loss": 0.6416,
+      "loss": 0.6898,
       "step": 24302
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.993588823609032e-07,
-      "loss": 0.8497,
+      "loss": 0.8455,
       "step": 24303
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.98926203725101e-07,
-      "loss": 0.828,
+      "loss": 0.8783,
       "step": 24304
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.984936541285037e-07,
-      "loss": 0.6909,
+      "loss": 0.8243,
       "step": 24305
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.980612335771087e-07,
-      "loss": 0.7537,
+      "loss": 0.816,
       "step": 24306
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.976289420769167e-07,
-      "loss": 0.7771,
+      "loss": 0.7892,
       "step": 24307
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.97196779633924e-07,
-      "loss": 0.6996,
+      "loss": 0.7469,
       "step": 24308
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.967647462541293e-07,
-      "loss": 0.8277,
+      "loss": 0.8771,
       "step": 24309
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.963328419435211e-07,
-      "loss": 0.8517,
+      "loss": 0.9012,
       "step": 24310
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.959010667080968e-07,
-      "loss": 0.7422,
+      "loss": 0.7962,
       "step": 24311
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.954694205538437e-07,
-      "loss": 0.7372,
+      "loss": 0.7689,
       "step": 24312
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.950379034867516e-07,
-      "loss": 0.6031,
+      "loss": 0.6914,
       "step": 24313
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.946065155128046e-07,
-      "loss": 0.7556,
+      "loss": 0.796,
       "step": 24314
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.941752566379934e-07,
-      "loss": 0.7243,
+      "loss": 0.8326,
       "step": 24315
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.937441268682932e-07,
-      "loss": 0.8643,
+      "loss": 0.8391,
       "step": 24316
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.933131262096914e-07,
-      "loss": 0.6194,
+      "loss": 0.6905,
       "step": 24317
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.928822546681657e-07,
-      "loss": 0.7082,
+      "loss": 0.8218,
       "step": 24318
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.924515122496934e-07,
-      "loss": 0.7597,
+      "loss": 0.8432,
       "step": 24319
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.920208989602505e-07,
-      "loss": 0.6295,
+      "loss": 0.6918,
       "step": 24320
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.915904148058095e-07,
-      "loss": 0.7035,
+      "loss": 0.7253,
       "step": 24321
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.911600597923485e-07,
-      "loss": 0.6573,
+      "loss": 0.7968,
       "step": 24322
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.907298339258306e-07,
-      "loss": 0.738,
+      "loss": 0.8152,
       "step": 24323
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.9029973721223e-07,
-      "loss": 0.7824,
+      "loss": 0.8497,
       "step": 24324
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.898697696575108e-07,
-      "loss": 0.6752,
+      "loss": 0.7668,
       "step": 24325
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.894399312676403e-07,
-      "loss": 0.7152,
+      "loss": 0.87,
       "step": 24326
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.890102220485784e-07,
-      "loss": 0.7228,
+      "loss": 0.7225,
       "step": 24327
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.885806420062935e-07,
-      "loss": 0.6759,
+      "loss": 0.7544,
       "step": 24328
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.881511911467365e-07,
-      "loss": 0.7395,
+      "loss": 0.7937,
       "step": 24329
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.877218694758725e-07,
-      "loss": 0.7484,
+      "loss": 0.8354,
       "step": 24330
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.872926769996546e-07,
-      "loss": 0.7282,
+      "loss": 0.7793,
       "step": 24331
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.868636137240381e-07,
-      "loss": 0.7794,
+      "loss": 0.7823,
       "step": 24332
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.864346796549736e-07,
-      "loss": 0.7589,
+      "loss": 0.7996,
       "step": 24333
     },
     {
       "epoch": 0.88,
       "learning_rate": 6.860058747984155e-07,
-      "loss": 0.7781,
+      "loss": 0.8484,
       "step": 24334
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.855771991603099e-07,
-      "loss": 0.7378,
+      "loss": 0.7978,
       "step": 24335
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.851486527466067e-07,
-      "loss": 0.8851,
+      "loss": 0.8774,
       "step": 24336
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.847202355632498e-07,
-      "loss": 0.7318,
+      "loss": 0.7477,
       "step": 24337
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.842919476161824e-07,
-      "loss": 0.758,
+      "loss": 0.8367,
       "step": 24338
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.838637889113475e-07,
-      "loss": 0.8065,
+      "loss": 0.8557,
       "step": 24339
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.834357594546825e-07,
-      "loss": 0.5991,
+      "loss": 0.7606,
       "step": 24340
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.830078592521306e-07,
-      "loss": 0.8912,
+      "loss": 0.838,
       "step": 24341
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.825800883096234e-07,
-      "loss": 0.6734,
+      "loss": 0.8163,
       "step": 24342
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.821524466330987e-07,
-      "loss": 0.73,
+      "loss": 0.7428,
       "step": 24343
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.817249342284893e-07,
-      "loss": 0.7067,
+      "loss": 0.7883,
       "step": 24344
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.81297551101725e-07,
-      "loss": 0.6486,
+      "loss": 0.7209,
       "step": 24345
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.808702972587344e-07,
-      "loss": 0.6816,
+      "loss": 0.7854,
       "step": 24346
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.804431727054495e-07,
-      "loss": 0.7885,
+      "loss": 0.8857,
       "step": 24347
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.800161774477887e-07,
-      "loss": 0.9064,
+      "loss": 0.8767,
       "step": 24348
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.795893114916829e-07,
-      "loss": 0.6773,
+      "loss": 0.7329,
       "step": 24349
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.791625748430508e-07,
-      "loss": 0.7755,
+      "loss": 0.7813,
       "step": 24350
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.787359675078131e-07,
-      "loss": 0.638,
+      "loss": 0.7689,
       "step": 24351
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.783094894918885e-07,
-      "loss": 0.6695,
+      "loss": 0.7737,
       "step": 24352
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.778831408011943e-07,
-      "loss": 0.8581,
+      "loss": 0.8559,
       "step": 24353
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.77456921441646e-07,
-      "loss": 0.7419,
+      "loss": 0.7486,
       "step": 24354
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.770308314191554e-07,
-      "loss": 0.8729,
+      "loss": 0.9193,
       "step": 24355
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.766048707396345e-07,
-      "loss": 0.6767,
+      "loss": 0.8089,
       "step": 24356
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.761790394089929e-07,
-      "loss": 0.6887,
+      "loss": 0.8143,
       "step": 24357
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.757533374331382e-07,
-      "loss": 0.8113,
+      "loss": 0.8508,
       "step": 24358
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.753277648179746e-07,
-      "loss": 0.7939,
+      "loss": 0.85,
       "step": 24359
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.749023215694129e-07,
-      "loss": 0.7631,
+      "loss": 0.8166,
       "step": 24360
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.74477007693346e-07,
-      "loss": 0.8156,
+      "loss": 0.8307,
       "step": 24361
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.740518231956827e-07,
-      "loss": 0.7636,
+      "loss": 0.8705,
       "step": 24362
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.736267680823172e-07,
-      "loss": 0.6766,
+      "loss": 0.7419,
       "step": 24363
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.73201842359149e-07,
-      "loss": 0.7662,
+      "loss": 0.807,
       "step": 24364
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.727770460320693e-07,
-      "loss": 0.7436,
+      "loss": 0.834,
       "step": 24365
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.723523791069775e-07,
-      "loss": 0.7443,
+      "loss": 0.8256,
       "step": 24366
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.719278415897613e-07,
-      "loss": 0.7976,
+      "loss": 0.8297,
       "step": 24367
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.715034334863124e-07,
-      "loss": 0.6569,
+      "loss": 0.7242,
       "step": 24368
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.710791548025164e-07,
-      "loss": 0.6953,
+      "loss": 0.7446,
       "step": 24369
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.706550055442629e-07,
-      "loss": 0.6947,
+      "loss": 0.7617,
       "step": 24370
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.702309857174316e-07,
-      "loss": 0.7929,
+      "loss": 0.874,
       "step": 24371
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.698070953279101e-07,
-      "loss": 0.7437,
+      "loss": 0.8063,
       "step": 24372
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.693833343815781e-07,
-      "loss": 0.7358,
+      "loss": 0.7253,
       "step": 24373
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.689597028843131e-07,
-      "loss": 0.6798,
+      "loss": 0.7816,
       "step": 24374
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.685362008419938e-07,
-      "loss": 0.7497,
+      "loss": 0.8213,
       "step": 24375
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.681128282604942e-07,
-      "loss": 0.8146,
+      "loss": 0.8884,
       "step": 24376
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.6768958514569e-07,
-      "loss": 0.6949,
+      "loss": 0.7475,
       "step": 24377
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.672664715034494e-07,
-      "loss": 0.6884,
+      "loss": 0.757,
       "step": 24378
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.668434873396478e-07,
-      "loss": 0.6573,
+      "loss": 0.7664,
       "step": 24379
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.664206326601497e-07,
-      "loss": 0.7782,
+      "loss": 0.8294,
       "step": 24380
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.659979074708234e-07,
-      "loss": 0.8249,
+      "loss": 0.8233,
       "step": 24381
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.655753117775332e-07,
-      "loss": 0.8436,
+      "loss": 0.8362,
       "step": 24382
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.651528455861411e-07,
-      "loss": 0.7963,
+      "loss": 0.845,
       "step": 24383
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.647305089025069e-07,
-      "loss": 0.6716,
+      "loss": 0.7423,
       "step": 24384
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.643083017324936e-07,
-      "loss": 0.7822,
+      "loss": 0.8326,
       "step": 24385
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.638862240819576e-07,
-      "loss": 0.7564,
+      "loss": 0.8067,
       "step": 24386
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.634642759567544e-07,
-      "loss": 0.6619,
+      "loss": 0.7621,
       "step": 24387
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.630424573627348e-07,
-      "loss": 0.7137,
+      "loss": 0.7363,
       "step": 24388
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.626207683057573e-07,
-      "loss": 0.811,
+      "loss": 0.8707,
       "step": 24389
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.621992087916662e-07,
-      "loss": 0.7466,
+      "loss": 0.824,
       "step": 24390
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.617777788263124e-07,
-      "loss": 0.717,
+      "loss": 0.7841,
       "step": 24391
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.613564784155446e-07,
-      "loss": 0.7774,
+      "loss": 0.8568,
       "step": 24392
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.609353075652047e-07,
-      "loss": 0.7178,
+      "loss": 0.8111,
       "step": 24393
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.605142662811371e-07,
-      "loss": 0.654,
+      "loss": 0.7136,
       "step": 24394
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.600933545691834e-07,
-      "loss": 0.7405,
+      "loss": 0.8316,
       "step": 24395
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.596725724351827e-07,
-      "loss": 0.7648,
+      "loss": 0.8278,
       "step": 24396
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.592519198849712e-07,
-      "loss": 0.6183,
+      "loss": 0.6411,
       "step": 24397
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.588313969243887e-07,
-      "loss": 0.7078,
+      "loss": 0.8079,
       "step": 24398
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.584110035592661e-07,
-      "loss": 0.7706,
+      "loss": 0.8933,
       "step": 24399
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.579907397954377e-07,
-      "loss": 0.7404,
+      "loss": 0.7853,
       "step": 24400
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.575706056387332e-07,
-      "loss": 0.8023,
+      "loss": 0.8756,
       "step": 24401
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.571506010949802e-07,
-      "loss": 0.7045,
+      "loss": 0.7547,
       "step": 24402
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.567307261700063e-07,
-      "loss": 0.7167,
+      "loss": 0.8366,
       "step": 24403
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.563109808696389e-07,
-      "loss": 0.8399,
+      "loss": 0.806,
       "step": 24404
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.55891365199699e-07,
-      "loss": 0.725,
+      "loss": 0.851,
       "step": 24405
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.554718791660097e-07,
-      "loss": 0.8171,
+      "loss": 0.9344,
       "step": 24406
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.550525227743876e-07,
-      "loss": 0.6861,
+      "loss": 0.7592,
       "step": 24407
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.546332960306567e-07,
-      "loss": 0.9042,
+      "loss": 0.8864,
       "step": 24408
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.542141989406259e-07,
-      "loss": 0.7358,
+      "loss": 0.7216,
       "step": 24409
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.537952315101148e-07,
-      "loss": 0.7803,
+      "loss": 0.7876,
       "step": 24410
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.533763937449344e-07,
-      "loss": 0.6117,
+      "loss": 0.7556,
       "step": 24411
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.529576856508957e-07,
-      "loss": 0.7173,
+      "loss": 0.8108,
       "step": 24412
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.525391072338072e-07,
-      "loss": 0.7038,
+      "loss": 0.833,
       "step": 24413
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.521206584994766e-07,
-      "loss": 0.6807,
+      "loss": 0.7488,
       "step": 24414
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.517023394537081e-07,
-      "loss": 0.6507,
+      "loss": 0.7006,
       "step": 24415
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.512841501023049e-07,
-      "loss": 0.6526,
+      "loss": 0.7396,
       "step": 24416
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.508660904510722e-07,
-      "loss": 0.6742,
+      "loss": 0.8339,
       "step": 24417
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.504481605058066e-07,
-      "loss": 0.7645,
+      "loss": 0.7617,
       "step": 24418
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.500303602723091e-07,
-      "loss": 0.6808,
+      "loss": 0.788,
       "step": 24419
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.496126897563715e-07,
-      "loss": 0.8463,
+      "loss": 0.8716,
       "step": 24420
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.491951489637949e-07,
-      "loss": 0.6955,
+      "loss": 0.7866,
       "step": 24421
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.487777379003657e-07,
-      "loss": 0.6832,
+      "loss": 0.7575,
       "step": 24422
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.483604565718793e-07,
-      "loss": 0.6923,
+      "loss": 0.8984,
       "step": 24423
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.479433049841233e-07,
-      "loss": 0.694,
+      "loss": 0.8164,
       "step": 24424
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.475262831428852e-07,
-      "loss": 0.793,
+      "loss": 0.8262,
       "step": 24425
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.471093910539484e-07,
-      "loss": 0.6872,
+      "loss": 0.6974,
       "step": 24426
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.466926287231023e-07,
-      "loss": 0.8317,
+      "loss": 0.9066,
       "step": 24427
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.462759961561227e-07,
-      "loss": 0.727,
+      "loss": 0.7654,
       "step": 24428
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.458594933587936e-07,
-      "loss": 0.7237,
+      "loss": 0.787,
       "step": 24429
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.454431203368916e-07,
-      "loss": 0.6356,
+      "loss": 0.7191,
       "step": 24430
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.450268770961943e-07,
-      "loss": 0.7226,
+      "loss": 0.7792,
       "step": 24431
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.446107636424769e-07,
-      "loss": 0.8091,
+      "loss": 0.8575,
       "step": 24432
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.441947799815096e-07,
-      "loss": 0.7051,
+      "loss": 0.7737,
       "step": 24433
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.437789261190686e-07,
-      "loss": 0.7342,
+      "loss": 0.7659,
       "step": 24434
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.433632020609159e-07,
-      "loss": 0.6499,
+      "loss": 0.7665,
       "step": 24435
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.429476078128271e-07,
-      "loss": 0.7306,
+      "loss": 0.7851,
       "step": 24436
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.425321433805631e-07,
-      "loss": 0.7984,
+      "loss": 0.891,
       "step": 24437
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.421168087698893e-07,
-      "loss": 0.7085,
+      "loss": 0.8088,
       "step": 24438
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.417016039865664e-07,
-      "loss": 0.8055,
+      "loss": 0.8283,
       "step": 24439
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.41286529036359e-07,
-      "loss": 0.554,
+      "loss": 0.7092,
       "step": 24440
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.408715839250191e-07,
-      "loss": 0.7779,
+      "loss": 0.8262,
       "step": 24441
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.404567686583096e-07,
-      "loss": 0.7398,
+      "loss": 0.7905,
       "step": 24442
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.400420832419818e-07,
-      "loss": 0.6744,
+      "loss": 0.7648,
       "step": 24443
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.396275276817909e-07,
-      "loss": 0.691,
+      "loss": 0.7876,
       "step": 24444
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.392131019834858e-07,
-      "loss": 0.6483,
+      "loss": 0.7428,
       "step": 24445
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.387988061528216e-07,
-      "loss": 0.602,
+      "loss": 0.6952,
       "step": 24446
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.383846401955373e-07,
-      "loss": 0.7438,
+      "loss": 0.7613,
       "step": 24447
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.379706041173861e-07,
-      "loss": 0.8169,
+      "loss": 0.8333,
       "step": 24448
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.375566979241099e-07,
-      "loss": 0.7055,
+      "loss": 0.7599,
       "step": 24449
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.371429216214509e-07,
-      "loss": 0.5743,
+      "loss": 0.7033,
       "step": 24450
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.367292752151499e-07,
-      "loss": 0.6435,
+      "loss": 0.7503,
       "step": 24451
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.363157587109436e-07,
-      "loss": 0.7019,
+      "loss": 0.8234,
       "step": 24452
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.359023721145741e-07,
-      "loss": 0.7335,
+      "loss": 0.7562,
       "step": 24453
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.3548911543177e-07,
-      "loss": 0.71,
+      "loss": 0.7603,
       "step": 24454
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.3507598866827e-07,
-      "loss": 0.7456,
+      "loss": 0.7832,
       "step": 24455
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.346629918298031e-07,
-      "loss": 0.821,
+      "loss": 0.8544,
       "step": 24456
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.342501249221e-07,
-      "loss": 0.6555,
+      "loss": 0.7346,
       "step": 24457
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.338373879508852e-07,
-      "loss": 0.5904,
+      "loss": 0.7075,
       "step": 24458
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.334247809218929e-07,
-      "loss": 0.7399,
+      "loss": 0.7836,
       "step": 24459
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.330123038408376e-07,
-      "loss": 0.6371,
+      "loss": 0.7815,
       "step": 24460
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.325999567134489e-07,
-      "loss": 0.6482,
+      "loss": 0.6961,
       "step": 24461
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.321877395454446e-07,
-      "loss": 0.6544,
+      "loss": 0.7793,
       "step": 24462
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.317756523425445e-07,
-      "loss": 0.8464,
+      "loss": 0.8918,
       "step": 24463
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.31363695110464e-07,
-      "loss": 0.7492,
+      "loss": 0.745,
       "step": 24464
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.30951867854922e-07,
-      "loss": 0.5895,
+      "loss": 0.7399,
       "step": 24465
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.305401705816294e-07,
-      "loss": 0.8663,
+      "loss": 0.8088,
       "step": 24466
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.301286032962983e-07,
-      "loss": 0.6828,
+      "loss": 0.78,
       "step": 24467
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.297171660046387e-07,
-      "loss": 0.7316,
+      "loss": 0.7681,
       "step": 24468
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.293058587123591e-07,
-      "loss": 0.6953,
+      "loss": 0.7355,
       "step": 24469
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.288946814251651e-07,
-      "loss": 0.6698,
+      "loss": 0.7538,
       "step": 24470
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.284836341487599e-07,
-      "loss": 0.7343,
+      "loss": 0.6841,
       "step": 24471
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.280727168888512e-07,
-      "loss": 0.7704,
+      "loss": 0.7973,
       "step": 24472
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.276619296511333e-07,
-      "loss": 0.73,
+      "loss": 0.7535,
       "step": 24473
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.272512724413105e-07,
-      "loss": 0.7511,
+      "loss": 0.837,
       "step": 24474
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.268407452650771e-07,
-      "loss": 0.7506,
+      "loss": 0.7845,
       "step": 24475
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.264303481281297e-07,
-      "loss": 0.766,
+      "loss": 0.8278,
       "step": 24476
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.260200810361617e-07,
-      "loss": 0.6732,
+      "loss": 0.7857,
       "step": 24477
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.256099439948648e-07,
-      "loss": 0.767,
+      "loss": 0.8336,
       "step": 24478
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.251999370099305e-07,
-      "loss": 0.7579,
+      "loss": 0.8249,
       "step": 24479
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.247900600870461e-07,
-      "loss": 0.8331,
+      "loss": 0.8102,
       "step": 24480
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.243803132318982e-07,
-      "loss": 0.5843,
+      "loss": 0.7411,
       "step": 24481
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.239706964501701e-07,
-      "loss": 0.6208,
+      "loss": 0.7506,
       "step": 24482
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.235612097475452e-07,
-      "loss": 0.6978,
+      "loss": 0.77,
       "step": 24483
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.231518531297054e-07,
-      "loss": 0.8652,
+      "loss": 0.8664,
       "step": 24484
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.227426266023307e-07,
-      "loss": 0.6681,
+      "loss": 0.7817,
       "step": 24485
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.223335301710975e-07,
-      "loss": 0.6727,
+      "loss": 0.7466,
       "step": 24486
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.219245638416794e-07,
-      "loss": 0.651,
+      "loss": 0.7977,
       "step": 24487
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.215157276197559e-07,
-      "loss": 0.7576,
+      "loss": 0.7653,
       "step": 24488
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.211070215109927e-07,
-      "loss": 0.5883,
+      "loss": 0.6413,
       "step": 24489
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.206984455210619e-07,
-      "loss": 0.8076,
+      "loss": 0.8453,
       "step": 24490
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.202899996556355e-07,
-      "loss": 0.7361,
+      "loss": 0.8324,
       "step": 24491
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.198816839203747e-07,
-      "loss": 0.7803,
+      "loss": 0.8323,
       "step": 24492
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.194734983209483e-07,
-      "loss": 0.6463,
+      "loss": 0.768,
       "step": 24493
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.190654428630171e-07,
-      "loss": 0.6763,
+      "loss": 0.7727,
       "step": 24494
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.186575175522436e-07,
-      "loss": 0.6932,
+      "loss": 0.7246,
       "step": 24495
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.182497223942852e-07,
-      "loss": 0.7606,
+      "loss": 0.7468,
       "step": 24496
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.178420573948029e-07,
-      "loss": 0.6628,
+      "loss": 0.7988,
       "step": 24497
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.174345225594502e-07,
-      "loss": 0.7993,
+      "loss": 0.8119,
       "step": 24498
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.170271178938814e-07,
-      "loss": 0.7938,
+      "loss": 0.8183,
       "step": 24499
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.166198434037462e-07,
-      "loss": 0.6488,
+      "loss": 0.7378,
       "step": 24500
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.162126990947015e-07,
-      "loss": 0.8605,
+      "loss": 0.8585,
       "step": 24501
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.158056849723882e-07,
-      "loss": 0.7815,
+      "loss": 0.8583,
       "step": 24502
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.153988010424583e-07,
-      "loss": 0.6981,
+      "loss": 0.752,
       "step": 24503
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.149920473105553e-07,
-      "loss": 0.7643,
+      "loss": 0.9042,
       "step": 24504
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.145854237823212e-07,
-      "loss": 0.6606,
+      "loss": 0.827,
       "step": 24505
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.141789304633971e-07,
-      "loss": 0.7026,
+      "loss": 0.7748,
       "step": 24506
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.137725673594275e-07,
-      "loss": 0.8,
+      "loss": 0.9043,
       "step": 24507
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.133663344760444e-07,
-      "loss": 0.6633,
+      "loss": 0.7662,
       "step": 24508
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.129602318188832e-07,
-      "loss": 0.719,
+      "loss": 0.7765,
       "step": 24509
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.125542593935819e-07,
-      "loss": 0.7585,
+      "loss": 0.834,
       "step": 24510
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.121484172057712e-07,
-      "loss": 0.7131,
+      "loss": 0.8018,
       "step": 24511
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.117427052610825e-07,
-      "loss": 0.7157,
+      "loss": 0.7694,
       "step": 24512
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.113371235651433e-07,
-      "loss": 0.7846,
+      "loss": 0.8475,
       "step": 24513
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.109316721235803e-07,
-      "loss": 0.7316,
+      "loss": 0.7974,
       "step": 24514
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.105263509420178e-07,
-      "loss": 0.7457,
+      "loss": 0.7984,
       "step": 24515
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.101211600260815e-07,
-      "loss": 0.6822,
+      "loss": 0.7601,
       "step": 24516
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.097160993813911e-07,
-      "loss": 0.7064,
+      "loss": 0.8154,
       "step": 24517
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.093111690135678e-07,
-      "loss": 0.6963,
+      "loss": 0.7989,
       "step": 24518
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.08906368928226e-07,
-      "loss": 0.7166,
+      "loss": 0.8152,
       "step": 24519
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.085016991309878e-07,
-      "loss": 0.7384,
+      "loss": 0.7804,
       "step": 24520
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.08097159627461e-07,
-      "loss": 0.7468,
+      "loss": 0.8017,
       "step": 24521
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.076927504232611e-07,
-      "loss": 0.7286,
+      "loss": 0.7859,
       "step": 24522
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.07288471523999e-07,
-      "loss": 0.7741,
+      "loss": 0.9163,
       "step": 24523
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.068843229352827e-07,
-      "loss": 0.7353,
+      "loss": 0.7114,
       "step": 24524
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.064803046627177e-07,
-      "loss": 0.814,
+      "loss": 0.8191,
       "step": 24525
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.060764167119149e-07,
-      "loss": 0.7646,
+      "loss": 0.873,
       "step": 24526
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.056726590884698e-07,
-      "loss": 0.592,
+      "loss": 0.7733,
       "step": 24527
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.052690317979893e-07,
-      "loss": 0.7618,
+      "loss": 0.7773,
       "step": 24528
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.048655348460708e-07,
-      "loss": 0.715,
+      "loss": 0.7685,
       "step": 24529
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.044621682383133e-07,
-      "loss": 0.756,
+      "loss": 0.8293,
       "step": 24530
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.040589319803136e-07,
-      "loss": 0.6302,
+      "loss": 0.7841,
       "step": 24531
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.036558260776637e-07,
-      "loss": 0.7056,
+      "loss": 0.7488,
       "step": 24532
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.032528505359592e-07,
-      "loss": 0.7578,
+      "loss": 0.8907,
       "step": 24533
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.028500053607877e-07,
-      "loss": 0.6276,
+      "loss": 0.6897,
       "step": 24534
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.024472905577406e-07,
-      "loss": 0.7714,
+      "loss": 0.7564,
       "step": 24535
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.020447061324031e-07,
-      "loss": 0.6742,
+      "loss": 0.7769,
       "step": 24536
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.01642252090362e-07,
-      "loss": 0.7301,
+      "loss": 0.8213,
       "step": 24537
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.012399284371983e-07,
-      "loss": 0.7022,
+      "loss": 0.7611,
       "step": 24538
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.008377351784989e-07,
-      "loss": 0.7076,
+      "loss": 0.7117,
       "step": 24539
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.004356723198368e-07,
-      "loss": 0.6618,
+      "loss": 0.7147,
       "step": 24540
     },
     {
       "epoch": 0.89,
       "learning_rate": 6.000337398667955e-07,
-      "loss": 0.805,
+      "loss": 0.8002,
       "step": 24541
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.996319378249493e-07,
-      "loss": 0.8451,
+      "loss": 0.8291,
       "step": 24542
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.992302661998728e-07,
-      "loss": 0.7855,
+      "loss": 0.8445,
       "step": 24543
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.988287249971359e-07,
-      "loss": 0.8631,
+      "loss": 0.8862,
       "step": 24544
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.98427314222314e-07,
-      "loss": 0.6996,
+      "loss": 0.7903,
       "step": 24545
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.98026033880974e-07,
-      "loss": 0.6845,
+      "loss": 0.8201,
       "step": 24546
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.976248839786835e-07,
-      "loss": 0.7106,
+      "loss": 0.7238,
       "step": 24547
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.972238645210071e-07,
-      "loss": 0.6697,
+      "loss": 0.7804,
       "step": 24548
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.968229755135091e-07,
-      "loss": 0.7,
+      "loss": 0.767,
       "step": 24549
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.964222169617506e-07,
-      "loss": 0.722,
+      "loss": 0.8331,
       "step": 24550
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.960215888712905e-07,
-      "loss": 0.7997,
+      "loss": 0.8831,
       "step": 24551
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.956210912476912e-07,
-      "loss": 0.6934,
+      "loss": 0.6974,
       "step": 24552
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.952207240965036e-07,
-      "loss": 0.7329,
+      "loss": 0.8036,
       "step": 24553
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.948204874232854e-07,
-      "loss": 0.6501,
+      "loss": 0.6692,
       "step": 24554
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.944203812335891e-07,
-      "loss": 0.7384,
+      "loss": 0.7881,
       "step": 24555
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.940204055329657e-07,
-      "loss": 0.8043,
+      "loss": 0.8058,
       "step": 24556
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.936205603269629e-07,
-      "loss": 0.8407,
+      "loss": 0.8304,
       "step": 24557
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.932208456211308e-07,
-      "loss": 0.7943,
+      "loss": 0.8017,
       "step": 24558
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.928212614210105e-07,
-      "loss": 0.7036,
+      "loss": 0.7572,
       "step": 24559
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.924218077321497e-07,
-      "loss": 0.7501,
+      "loss": 0.8502,
       "step": 24560
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.920224845600886e-07,
-      "loss": 0.7578,
+      "loss": 0.7944,
       "step": 24561
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.916232919103671e-07,
-      "loss": 0.716,
+      "loss": 0.7192,
       "step": 24562
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.912242297885218e-07,
-      "loss": 0.756,
+      "loss": 0.777,
       "step": 24563
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.90825298200094e-07,
-      "loss": 0.727,
+      "loss": 0.867,
       "step": 24564
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.904264971506135e-07,
-      "loss": 0.7505,
+      "loss": 0.7771,
       "step": 24565
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.900278266456161e-07,
-      "loss": 0.7672,
+      "loss": 0.8054,
       "step": 24566
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.896292866906317e-07,
-      "loss": 0.673,
+      "loss": 0.7534,
       "step": 24567
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.892308772911881e-07,
-      "loss": 0.8077,
+      "loss": 0.7455,
       "step": 24568
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.888325984528154e-07,
-      "loss": 0.7231,
+      "loss": 0.7657,
       "step": 24569
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.884344501810357e-07,
-      "loss": 0.697,
+      "loss": 0.7855,
       "step": 24570
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.880364324813781e-07,
-      "loss": 0.6528,
+      "loss": 0.7842,
       "step": 24571
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.87638545359357e-07,
-      "loss": 0.7032,
+      "loss": 0.6947,
       "step": 24572
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.872407888204989e-07,
-      "loss": 0.7269,
+      "loss": 0.7996,
       "step": 24573
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.868431628703197e-07,
-      "loss": 0.8641,
+      "loss": 0.8894,
       "step": 24574
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.864456675143371e-07,
-      "loss": 0.7055,
+      "loss": 0.8398,
       "step": 24575
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.860483027580622e-07,
-      "loss": 0.7344,
+      "loss": 0.772,
       "step": 24576
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.856510686070127e-07,
-      "loss": 0.7468,
+      "loss": 0.8085,
       "step": 24577
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.852539650666967e-07,
-      "loss": 0.6524,
+      "loss": 0.7127,
       "step": 24578
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.848569921426239e-07,
-      "loss": 0.694,
+      "loss": 0.7654,
       "step": 24579
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.844601498403025e-07,
-      "loss": 0.7299,
+      "loss": 0.7885,
       "step": 24580
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.840634381652377e-07,
-      "loss": 0.8342,
+      "loss": 0.7544,
       "step": 24581
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.836668571229321e-07,
-      "loss": 0.7287,
+      "loss": 0.7927,
       "step": 24582
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.832704067188899e-07,
-      "loss": 0.9117,
+      "loss": 0.8936,
       "step": 24583
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.828740869586102e-07,
-      "loss": 0.5888,
+      "loss": 0.7447,
       "step": 24584
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.824778978475909e-07,
-      "loss": 0.9077,
+      "loss": 0.8643,
       "step": 24585
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.820818393913296e-07,
-      "loss": 0.8529,
+      "loss": 0.8835,
       "step": 24586
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.8168591159532e-07,
-      "loss": 0.761,
+      "loss": 0.8052,
       "step": 24587
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.812901144650562e-07,
-      "loss": 0.7738,
+      "loss": 0.857,
       "step": 24588
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.808944480060275e-07,
-      "loss": 0.7429,
+      "loss": 0.8147,
       "step": 24589
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.80498912223727e-07,
-      "loss": 0.6896,
+      "loss": 0.7134,
       "step": 24590
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.801035071236361e-07,
-      "loss": 0.7399,
+      "loss": 0.8287,
       "step": 24591
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.797082327112458e-07,
-      "loss": 0.5612,
+      "loss": 0.663,
       "step": 24592
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.793130889920373e-07,
-      "loss": 0.7796,
+      "loss": 0.81,
       "step": 24593
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.789180759714941e-07,
-      "loss": 0.7807,
+      "loss": 0.7655,
       "step": 24594
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.78523193655095e-07,
-      "loss": 0.6991,
+      "loss": 0.6927,
       "step": 24595
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.781284420483202e-07,
-      "loss": 0.7389,
+      "loss": 0.7594,
       "step": 24596
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.777338211566453e-07,
-      "loss": 0.7529,
+      "loss": 0.7992,
       "step": 24597
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.773393309855457e-07,
-      "loss": 0.7301,
+      "loss": 0.7703,
       "step": 24598
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.769449715404917e-07,
-      "loss": 0.6587,
+      "loss": 0.7914,
       "step": 24599
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.765507428269601e-07,
-      "loss": 0.81,
+      "loss": 0.8073,
       "step": 24600
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.76156644850413e-07,
-      "loss": 0.7148,
+      "loss": 0.7693,
       "step": 24601
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.757626776163239e-07,
-      "loss": 0.7055,
+      "loss": 0.7626,
       "step": 24602
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.753688411301573e-07,
-      "loss": 0.7994,
+      "loss": 0.779,
       "step": 24603
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.749751353973743e-07,
-      "loss": 0.8654,
+      "loss": 0.953,
       "step": 24604
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.745815604234406e-07,
-      "loss": 0.8471,
+      "loss": 0.8181,
       "step": 24605
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.741881162138141e-07,
-      "loss": 0.7008,
+      "loss": 0.8379,
       "step": 24606
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.737948027739537e-07,
-      "loss": 0.8008,
+      "loss": 0.8314,
       "step": 24607
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.734016201093151e-07,
-      "loss": 0.8135,
+      "loss": 0.7941,
       "step": 24608
     },
     {
       "epoch": 0.89,
       "learning_rate": 5.730085682253561e-07,
-      "loss": 0.7442,
+      "loss": 0.7531,
       "step": 24609
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.72615647127529e-07,
-      "loss": 0.7803,
+      "loss": 0.8128,
       "step": 24610
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.722228568212828e-07,
-      "loss": 0.6754,
+      "loss": 0.7999,
       "step": 24611
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.718301973120688e-07,
-      "loss": 0.7854,
+      "loss": 0.9073,
       "step": 24612
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.714376686053336e-07,
-      "loss": 0.7142,
+      "loss": 0.753,
       "step": 24613
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.710452707065217e-07,
-      "loss": 0.8471,
+      "loss": 0.8937,
       "step": 24614
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.706530036210811e-07,
-      "loss": 0.7388,
+      "loss": 0.8051,
       "step": 24615
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.702608673544508e-07,
-      "loss": 0.8441,
+      "loss": 0.845,
       "step": 24616
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.698688619120718e-07,
-      "loss": 0.6058,
+      "loss": 0.7378,
       "step": 24617
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.694769872993822e-07,
-      "loss": 0.8434,
+      "loss": 0.8687,
       "step": 24618
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.690852435218208e-07,
-      "loss": 0.7758,
+      "loss": 0.773,
       "step": 24619
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.68693630584819e-07,
-      "loss": 0.6378,
+      "loss": 0.7652,
       "step": 24620
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.683021484938134e-07,
-      "loss": 0.8085,
+      "loss": 0.8742,
       "step": 24621
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.67910797254233e-07,
-      "loss": 0.8652,
+      "loss": 0.8456,
       "step": 24622
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.67519576871508e-07,
-      "loss": 0.75,
+      "loss": 0.8029,
       "step": 24623
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.671284873510663e-07,
-      "loss": 0.876,
+      "loss": 0.8796,
       "step": 24624
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.667375286983334e-07,
-      "loss": 0.7441,
+      "loss": 0.8346,
       "step": 24625
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.663467009187329e-07,
-      "loss": 0.9351,
+      "loss": 0.8999,
       "step": 24626
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.659560040176859e-07,
-      "loss": 0.7008,
+      "loss": 0.8708,
       "step": 24627
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.655654380006159e-07,
-      "loss": 0.7026,
+      "loss": 0.7847,
       "step": 24628
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.651750028729397e-07,
-      "loss": 0.6471,
+      "loss": 0.697,
       "step": 24629
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.647846986400741e-07,
-      "loss": 0.7186,
+      "loss": 0.7359,
       "step": 24630
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.643945253074323e-07,
-      "loss": 0.784,
+      "loss": 0.8257,
       "step": 24631
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.640044828804314e-07,
-      "loss": 0.6465,
+      "loss": 0.6667,
       "step": 24632
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.636145713644781e-07,
-      "loss": 0.7824,
+      "loss": 0.8044,
       "step": 24633
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.632247907649857e-07,
-      "loss": 0.8029,
+      "loss": 0.8094,
       "step": 24634
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.6283514108736e-07,
-      "loss": 0.6586,
+      "loss": 0.7028,
       "step": 24635
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.624456223370078e-07,
-      "loss": 0.6909,
+      "loss": 0.8585,
       "step": 24636
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.620562345193303e-07,
-      "loss": 0.7433,
+      "loss": 0.8275,
       "step": 24637
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.616669776397354e-07,
-      "loss": 0.8117,
+      "loss": 0.8572,
       "step": 24638
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.612778517036166e-07,
-      "loss": 0.78,
+      "loss": 0.8378,
       "step": 24639
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.608888567163773e-07,
-      "loss": 0.7123,
+      "loss": 0.7709,
       "step": 24640
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.604999926834121e-07,
-      "loss": 0.7307,
+      "loss": 0.8113,
       "step": 24641
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.601112596101177e-07,
-      "loss": 0.7615,
+      "loss": 0.7814,
       "step": 24642
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.597226575018855e-07,
-      "loss": 0.8154,
+      "loss": 0.7207,
       "step": 24643
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.593341863641067e-07,
-      "loss": 0.7314,
+      "loss": 0.7867,
       "step": 24644
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.589458462021736e-07,
-      "loss": 0.6668,
+      "loss": 0.758,
       "step": 24645
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.585576370214696e-07,
-      "loss": 0.7319,
+      "loss": 0.7631,
       "step": 24646
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.581695588273827e-07,
-      "loss": 0.7275,
+      "loss": 0.7731,
       "step": 24647
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.577816116252988e-07,
-      "loss": 0.7215,
+      "loss": 0.7061,
       "step": 24648
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.573937954205977e-07,
-      "loss": 0.6384,
+      "loss": 0.686,
       "step": 24649
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.570061102186585e-07,
-      "loss": 0.8346,
+      "loss": 0.8648,
       "step": 24650
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.566185560248649e-07,
-      "loss": 0.7999,
+      "loss": 0.8366,
       "step": 24651
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.562311328445869e-07,
-      "loss": 0.7216,
+      "loss": 0.8517,
       "step": 24652
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.558438406832056e-07,
-      "loss": 0.7685,
+      "loss": 0.7633,
       "step": 24653
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.554566795460903e-07,
-      "loss": 0.7221,
+      "loss": 0.6774,
       "step": 24654
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.550696494386143e-07,
-      "loss": 0.709,
+      "loss": 0.7705,
       "step": 24655
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.546827503661445e-07,
-      "loss": 0.7037,
+      "loss": 0.7807,
       "step": 24656
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.542959823340532e-07,
-      "loss": 0.8399,
+      "loss": 0.8873,
       "step": 24657
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.539093453477007e-07,
-      "loss": 0.6629,
+      "loss": 0.7527,
       "step": 24658
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.535228394124558e-07,
-      "loss": 0.737,
+      "loss": 0.7809,
       "step": 24659
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.531364645336778e-07,
-      "loss": 0.7126,
+      "loss": 0.8119,
       "step": 24660
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.52750220716729e-07,
-      "loss": 0.774,
+      "loss": 0.7902,
       "step": 24661
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.523641079669662e-07,
-      "loss": 0.6513,
+      "loss": 0.7523,
       "step": 24662
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.51978126289745e-07,
-      "loss": 0.7624,
+      "loss": 0.8359,
       "step": 24663
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.515922756904268e-07,
-      "loss": 0.7341,
+      "loss": 0.8167,
       "step": 24664
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.512065561743563e-07,
-      "loss": 0.7331,
+      "loss": 0.8375,
       "step": 24665
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.508209677468901e-07,
-      "loss": 0.7172,
+      "loss": 0.8203,
       "step": 24666
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.504355104133762e-07,
-      "loss": 0.6627,
+      "loss": 0.7143,
       "step": 24667
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.500501841791628e-07,
-      "loss": 0.7193,
+      "loss": 0.8465,
       "step": 24668
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.496649890495942e-07,
-      "loss": 0.7634,
+      "loss": 0.7893,
       "step": 24669
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.492799250300185e-07,
-      "loss": 0.7871,
+      "loss": 0.8143,
       "step": 24670
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.488949921257714e-07,
-      "loss": 0.6212,
+      "loss": 0.66,
       "step": 24671
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.485101903421986e-07,
-      "loss": 0.617,
+      "loss": 0.6995,
       "step": 24672
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.48125519684638e-07,
-      "loss": 0.7224,
+      "loss": 0.8391,
       "step": 24673
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.477409801584243e-07,
-      "loss": 0.8104,
+      "loss": 0.8913,
       "step": 24674
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.473565717688922e-07,
-      "loss": 0.6679,
+      "loss": 0.783,
       "step": 24675
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.469722945213774e-07,
-      "loss": 0.7702,
+      "loss": 0.8194,
       "step": 24676
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.46588148421211e-07,
-      "loss": 0.6436,
+      "loss": 0.7382,
       "step": 24677
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.4620413347372e-07,
-      "loss": 0.7215,
+      "loss": 0.7682,
       "step": 24678
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.458202496842346e-07,
-      "loss": 0.7088,
+      "loss": 0.7916,
       "step": 24679
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.454364970580783e-07,
-      "loss": 0.6483,
+      "loss": 0.7896,
       "step": 24680
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.450528756005779e-07,
-      "loss": 0.6945,
+      "loss": 0.7128,
       "step": 24681
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.446693853170515e-07,
-      "loss": 0.6994,
+      "loss": 0.7239,
       "step": 24682
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.442860262128247e-07,
-      "loss": 0.5634,
+      "loss": 0.7292,
       "step": 24683
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.43902798293211e-07,
-      "loss": 0.7591,
+      "loss": 0.8869,
       "step": 24684
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.435197015635308e-07,
-      "loss": 0.8331,
+      "loss": 0.8633,
       "step": 24685
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.431367360290984e-07,
-      "loss": 0.7109,
+      "loss": 0.7532,
       "step": 24686
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.427539016952255e-07,
-      "loss": 0.7438,
+      "loss": 0.7818,
       "step": 24687
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.423711985672231e-07,
-      "loss": 0.7279,
+      "loss": 0.7748,
       "step": 24688
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.419886266504026e-07,
-      "loss": 0.6689,
+      "loss": 0.8081,
       "step": 24689
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.416061859500709e-07,
-      "loss": 0.7695,
+      "loss": 0.8689,
       "step": 24690
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.412238764715338e-07,
-      "loss": 0.7831,
+      "loss": 0.7726,
       "step": 24691
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.408416982200959e-07,
-      "loss": 0.7354,
+      "loss": 0.8554,
       "step": 24692
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.404596512010574e-07,
-      "loss": 0.6958,
+      "loss": 0.7276,
       "step": 24693
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.400777354197195e-07,
-      "loss": 0.7583,
+      "loss": 0.7246,
       "step": 24694
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.396959508813815e-07,
-      "loss": 0.7312,
+      "loss": 0.7737,
       "step": 24695
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.393142975913401e-07,
-      "loss": 0.8103,
+      "loss": 0.8057,
       "step": 24696
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.389327755548901e-07,
-      "loss": 0.78,
+      "loss": 0.7717,
       "step": 24697
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.385513847773238e-07,
-      "loss": 0.7171,
+      "loss": 0.817,
       "step": 24698
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.381701252639349e-07,
-      "loss": 0.6624,
+      "loss": 0.7109,
       "step": 24699
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.37788997020009e-07,
-      "loss": 0.6486,
+      "loss": 0.7192,
       "step": 24700
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.374080000508353e-07,
-      "loss": 0.8322,
+      "loss": 0.7884,
       "step": 24701
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.370271343617018e-07,
-      "loss": 0.6936,
+      "loss": 0.7971,
       "step": 24702
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.366463999578885e-07,
-      "loss": 0.6801,
+      "loss": 0.8102,
       "step": 24703
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.362657968446805e-07,
-      "loss": 0.7319,
+      "loss": 0.833,
       "step": 24704
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.358853250273577e-07,
-      "loss": 0.8517,
+      "loss": 0.8438,
       "step": 24705
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.355049845111981e-07,
-      "loss": 0.8625,
+      "loss": 0.8882,
       "step": 24706
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.351247753014777e-07,
-      "loss": 0.5751,
+      "loss": 0.6946,
       "step": 24707
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.347446974034731e-07,
-      "loss": 0.72,
+      "loss": 0.7706,
       "step": 24708
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.343647508224559e-07,
-      "loss": 0.779,
+      "loss": 0.7962,
       "step": 24709
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.339849355636984e-07,
-      "loss": 0.6458,
+      "loss": 0.7521,
       "step": 24710
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.336052516324675e-07,
-      "loss": 0.7161,
+      "loss": 0.7179,
       "step": 24711
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.332256990340357e-07,
-      "loss": 0.7328,
+      "loss": 0.7855,
       "step": 24712
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.328462777736631e-07,
-      "loss": 0.8698,
+      "loss": 0.8739,
       "step": 24713
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.324669878566169e-07,
-      "loss": 0.7854,
+      "loss": 0.7737,
       "step": 24714
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.320878292881593e-07,
-      "loss": 0.7263,
+      "loss": 0.8141,
       "step": 24715
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.317088020735494e-07,
-      "loss": 0.6176,
+      "loss": 0.7688,
       "step": 24716
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.313299062180455e-07,
-      "loss": 0.6293,
+      "loss": 0.7442,
       "step": 24717
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.309511417269076e-07,
-      "loss": 0.7263,
+      "loss": 0.7391,
       "step": 24718
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.305725086053859e-07,
-      "loss": 0.7487,
+      "loss": 0.7358,
       "step": 24719
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.301940068587342e-07,
-      "loss": 0.6741,
+      "loss": 0.793,
       "step": 24720
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.298156364922058e-07,
-      "loss": 0.599,
+      "loss": 0.7168,
       "step": 24721
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.29437397511049e-07,
-      "loss": 0.7234,
+      "loss": 0.8199,
       "step": 24722
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.290592899205115e-07,
-      "loss": 0.7279,
+      "loss": 0.753,
       "step": 24723
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.286813137258384e-07,
-      "loss": 0.8065,
+      "loss": 0.8352,
       "step": 24724
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.283034689322752e-07,
-      "loss": 0.8012,
+      "loss": 0.8304,
       "step": 24725
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.279257555450601e-07,
-      "loss": 0.6958,
+      "loss": 0.7657,
       "step": 24726
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.275481735694377e-07,
-      "loss": 0.6257,
+      "loss": 0.7826,
       "step": 24727
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.271707230106459e-07,
-      "loss": 0.5902,
+      "loss": 0.654,
       "step": 24728
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.267934038739186e-07,
-      "loss": 0.6128,
+      "loss": 0.7223,
       "step": 24729
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.264162161644915e-07,
-      "loss": 0.7971,
+      "loss": 0.9057,
       "step": 24730
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.260391598876014e-07,
-      "loss": 0.6775,
+      "loss": 0.7586,
       "step": 24731
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.256622350484719e-07,
-      "loss": 0.8368,
+      "loss": 0.8692,
       "step": 24732
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.252854416523389e-07,
-      "loss": 0.5529,
+      "loss": 0.6063,
       "step": 24733
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.249087797044272e-07,
-      "loss": 0.7792,
+      "loss": 0.8089,
       "step": 24734
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.245322492099636e-07,
-      "loss": 0.8949,
+      "loss": 0.862,
       "step": 24735
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.241558501741695e-07,
-      "loss": 0.7482,
+      "loss": 0.7923,
       "step": 24736
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.237795826022707e-07,
-      "loss": 0.6282,
+      "loss": 0.779,
       "step": 24737
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.234034464994842e-07,
-      "loss": 0.6718,
+      "loss": 0.7655,
       "step": 24738
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.230274418710279e-07,
-      "loss": 0.7107,
+      "loss": 0.7462,
       "step": 24739
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.226515687221212e-07,
-      "loss": 0.6703,
+      "loss": 0.7664,
       "step": 24740
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.222758270579764e-07,
-      "loss": 0.7413,
+      "loss": 0.8432,
       "step": 24741
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.219002168838083e-07,
-      "loss": 0.7365,
+      "loss": 0.8329,
       "step": 24742
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.215247382048261e-07,
-      "loss": 0.6867,
+      "loss": 0.7647,
       "step": 24743
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.211493910262421e-07,
-      "loss": 0.8428,
+      "loss": 0.8152,
       "step": 24744
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.207741753532591e-07,
-      "loss": 0.6227,
+      "loss": 0.6583,
       "step": 24745
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.203990911910861e-07,
-      "loss": 0.8239,
+      "loss": 0.8647,
       "step": 24746
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.200241385449256e-07,
-      "loss": 0.6886,
+      "loss": 0.8623,
       "step": 24747
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.196493174199813e-07,
-      "loss": 0.8432,
+      "loss": 0.8781,
       "step": 24748
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.192746278214489e-07,
-      "loss": 0.7621,
+      "loss": 0.7991,
       "step": 24749
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.189000697545332e-07,
-      "loss": 0.6164,
+      "loss": 0.7363,
       "step": 24750
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.185256432244246e-07,
-      "loss": 0.7594,
+      "loss": 0.8664,
       "step": 24751
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.18151348236321e-07,
-      "loss": 0.6625,
+      "loss": 0.7789,
       "step": 24752
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.177771847954149e-07,
-      "loss": 0.6867,
+      "loss": 0.7658,
       "step": 24753
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.174031529068968e-07,
-      "loss": 0.6711,
+      "loss": 0.7803,
       "step": 24754
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.170292525759546e-07,
-      "loss": 0.6769,
+      "loss": 0.7494,
       "step": 24755
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.166554838077786e-07,
-      "loss": 0.6713,
+      "loss": 0.7232,
       "step": 24756
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.162818466075525e-07,
-      "loss": 0.8519,
+      "loss": 0.8634,
       "step": 24757
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.159083409804588e-07,
-      "loss": 0.713,
+      "loss": 0.7861,
       "step": 24758
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.155349669316812e-07,
-      "loss": 0.7706,
+      "loss": 0.7805,
       "step": 24759
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.151617244663998e-07,
-      "loss": 0.7171,
+      "loss": 0.7795,
       "step": 24760
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.147886135897929e-07,
-      "loss": 0.6417,
+      "loss": 0.7772,
       "step": 24761
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.144156343070339e-07,
-      "loss": 0.753,
+      "loss": 0.8419,
       "step": 24762
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.140427866233033e-07,
-      "loss": 0.7449,
+      "loss": 0.8159,
       "step": 24763
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.136700705437658e-07,
-      "loss": 0.7186,
+      "loss": 0.8516,
       "step": 24764
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.132974860735995e-07,
-      "loss": 0.5639,
+      "loss": 0.7031,
       "step": 24765
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.129250332179702e-07,
-      "loss": 0.6269,
+      "loss": 0.7311,
       "step": 24766
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.125527119820461e-07,
-      "loss": 0.7653,
+      "loss": 0.7965,
       "step": 24767
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.121805223709908e-07,
-      "loss": 0.7524,
+      "loss": 0.7781,
       "step": 24768
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.118084643899723e-07,
-      "loss": 0.7602,
+      "loss": 0.8095,
       "step": 24769
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.114365380441455e-07,
-      "loss": 0.8292,
+      "loss": 0.9044,
       "step": 24770
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.110647433386761e-07,
-      "loss": 0.7874,
+      "loss": 0.8723,
       "step": 24771
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.106930802787213e-07,
-      "loss": 0.8602,
+      "loss": 0.9152,
       "step": 24772
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.103215488694357e-07,
-      "loss": 0.9197,
+      "loss": 0.8712,
       "step": 24773
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.099501491159731e-07,
-      "loss": 0.7466,
+      "loss": 0.726,
       "step": 24774
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.095788810234881e-07,
-      "loss": 0.7141,
+      "loss": 0.7215,
       "step": 24775
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.092077445971321e-07,
-      "loss": 0.7954,
+      "loss": 0.8964,
       "step": 24776
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.088367398420524e-07,
-      "loss": 0.7298,
+      "loss": 0.7688,
       "step": 24777
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.084658667633968e-07,
-      "loss": 0.6816,
+      "loss": 0.7737,
       "step": 24778
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.080951253663113e-07,
-      "loss": 0.7766,
+      "loss": 0.8367,
       "step": 24779
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.077245156559385e-07,
-      "loss": 0.6448,
+      "loss": 0.7555,
       "step": 24780
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.073540376374187e-07,
-      "loss": 0.6943,
+      "loss": 0.7542,
       "step": 24781
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.069836913158954e-07,
-      "loss": 0.6711,
+      "loss": 0.7392,
       "step": 24782
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.066134766965025e-07,
-      "loss": 0.6144,
+      "loss": 0.722,
       "step": 24783
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.062433937843792e-07,
-      "loss": 0.8034,
+      "loss": 0.8643,
       "step": 24784
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.05873442584659e-07,
-      "loss": 0.6508,
+      "loss": 0.7993,
       "step": 24785
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.055036231024746e-07,
-      "loss": 0.8957,
+      "loss": 0.8797,
       "step": 24786
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.051339353429552e-07,
-      "loss": 0.7156,
+      "loss": 0.8127,
       "step": 24787
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.047643793112323e-07,
-      "loss": 0.6997,
+      "loss": 0.8404,
       "step": 24788
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.043949550124316e-07,
-      "loss": 0.7619,
+      "loss": 0.7199,
       "step": 24789
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.040256624516792e-07,
-      "loss": 0.7754,
+      "loss": 0.7927,
       "step": 24790
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.036565016340966e-07,
-      "loss": 0.689,
+      "loss": 0.82,
       "step": 24791
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.032874725648074e-07,
-      "loss": 0.8016,
+      "loss": 0.8488,
       "step": 24792
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.029185752489285e-07,
-      "loss": 0.7646,
+      "loss": 0.834,
       "step": 24793
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.025498096915804e-07,
-      "loss": 0.7156,
+      "loss": 0.7625,
       "step": 24794
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.021811758978801e-07,
-      "loss": 0.7603,
+      "loss": 0.8087,
       "step": 24795
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.018126738729389e-07,
-      "loss": 0.7604,
+      "loss": 0.7663,
       "step": 24796
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.014443036218719e-07,
-      "loss": 0.6402,
+      "loss": 0.7132,
       "step": 24797
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.01076065149787e-07,
-      "loss": 0.7162,
+      "loss": 0.856,
       "step": 24798
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.007079584617958e-07,
-      "loss": 0.7441,
+      "loss": 0.7937,
       "step": 24799
     },
     {
       "epoch": 0.9,
       "learning_rate": 5.00339983563003e-07,
-      "loss": 0.631,
+      "loss": 0.6744,
       "step": 24800
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.999721404585156e-07,
-      "loss": 0.7592,
+      "loss": 0.7724,
       "step": 24801
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.996044291534352e-07,
-      "loss": 0.7697,
+      "loss": 0.764,
       "step": 24802
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.992368496528644e-07,
-      "loss": 0.6329,
+      "loss": 0.7906,
       "step": 24803
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.988694019619022e-07,
-      "loss": 0.7724,
+      "loss": 0.7808,
       "step": 24804
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.985020860856471e-07,
-      "loss": 0.7061,
+      "loss": 0.7733,
       "step": 24805
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.981349020291936e-07,
-      "loss": 0.796,
+      "loss": 0.8379,
       "step": 24806
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.977678497976379e-07,
-      "loss": 0.667,
+      "loss": 0.7433,
       "step": 24807
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.974009293960713e-07,
-      "loss": 0.6835,
+      "loss": 0.7915,
       "step": 24808
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.970341408295842e-07,
-      "loss": 0.7877,
+      "loss": 0.8495,
       "step": 24809
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.966674841032648e-07,
-      "loss": 0.706,
+      "loss": 0.7801,
       "step": 24810
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.963009592222034e-07,
-      "loss": 0.832,
+      "loss": 0.8228,
       "step": 24811
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.959345661914783e-07,
-      "loss": 0.8135,
+      "loss": 0.8519,
       "step": 24812
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.955683050161786e-07,
-      "loss": 0.7437,
+      "loss": 0.754,
       "step": 24813
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.952021757013848e-07,
-      "loss": 0.6358,
+      "loss": 0.689,
       "step": 24814
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.948361782521737e-07,
-      "loss": 0.6827,
+      "loss": 0.7523,
       "step": 24815
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.94470312673625e-07,
-      "loss": 0.7522,
+      "loss": 0.81,
       "step": 24816
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.941045789708144e-07,
-      "loss": 0.7157,
+      "loss": 0.7627,
       "step": 24817
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.937389771488155e-07,
-      "loss": 0.8668,
+      "loss": 0.8731,
       "step": 24818
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.933735072126988e-07,
-      "loss": 0.87,
+      "loss": 0.8233,
       "step": 24819
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.930081691675393e-07,
-      "loss": 0.6513,
+      "loss": 0.7841,
       "step": 24820
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.926429630184015e-07,
-      "loss": 0.7865,
+      "loss": 0.8251,
       "step": 24821
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.922778887703539e-07,
-      "loss": 0.758,
+      "loss": 0.787,
       "step": 24822
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.919129464284611e-07,
-      "loss": 0.8454,
+      "loss": 0.8204,
       "step": 24823
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.915481359977847e-07,
-      "loss": 0.8067,
+      "loss": 0.8965,
       "step": 24824
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.911834574833874e-07,
-      "loss": 0.7879,
+      "loss": 0.8502,
       "step": 24825
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.908189108903294e-07,
-      "loss": 0.6521,
+      "loss": 0.7222,
       "step": 24826
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.904544962236668e-07,
-      "loss": 0.694,
+      "loss": 0.7482,
       "step": 24827
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.900902134884556e-07,
-      "loss": 0.8514,
+      "loss": 0.8241,
       "step": 24828
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.897260626897493e-07,
-      "loss": 0.7377,
+      "loss": 0.7122,
       "step": 24829
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.89362043832603e-07,
-      "loss": 0.7722,
+      "loss": 0.9002,
       "step": 24830
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.889981569220614e-07,
-      "loss": 0.7928,
+      "loss": 0.8054,
       "step": 24831
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.886344019631773e-07,
-      "loss": 0.6292,
+      "loss": 0.7956,
       "step": 24832
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.882707789609964e-07,
-      "loss": 0.7051,
+      "loss": 0.7579,
       "step": 24833
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.879072879205638e-07,
-      "loss": 0.7179,
+      "loss": 0.7907,
       "step": 24834
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.875439288469208e-07,
-      "loss": 0.7539,
+      "loss": 0.829,
       "step": 24835
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.871807017451091e-07,
-      "loss": 0.7556,
+      "loss": 0.8548,
       "step": 24836
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.868176066201691e-07,
-      "loss": 0.6843,
+      "loss": 0.8005,
       "step": 24837
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.864546434771356e-07,
-      "loss": 0.6641,
+      "loss": 0.7011,
       "step": 24838
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.860918123210467e-07,
-      "loss": 0.7487,
+      "loss": 0.7735,
       "step": 24839
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.857291131569364e-07,
-      "loss": 0.7285,
+      "loss": 0.7317,
       "step": 24840
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.853665459898349e-07,
-      "loss": 0.7069,
+      "loss": 0.7621,
       "step": 24841
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.850041108247727e-07,
-      "loss": 0.8425,
+      "loss": 0.8321,
       "step": 24842
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.846418076667802e-07,
-      "loss": 0.6342,
+      "loss": 0.8106,
       "step": 24843
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.842796365208791e-07,
-      "loss": 0.8037,
+      "loss": 0.8452,
       "step": 24844
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.839175973920995e-07,
-      "loss": 0.5592,
+      "loss": 0.717,
       "step": 24845
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.83555690285461e-07,
-      "loss": 0.6376,
+      "loss": 0.7734,
       "step": 24846
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.831939152059862e-07,
-      "loss": 0.665,
+      "loss": 0.6965,
       "step": 24847
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.828322721586909e-07,
-      "loss": 0.6597,
+      "loss": 0.7303,
       "step": 24848
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.824707611485969e-07,
-      "loss": 0.7144,
+      "loss": 0.8685,
       "step": 24849
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.821093821807155e-07,
-      "loss": 0.5867,
+      "loss": 0.6939,
       "step": 24850
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.817481352600639e-07,
-      "loss": 0.7676,
+      "loss": 0.796,
       "step": 24851
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.813870203916527e-07,
-      "loss": 0.7025,
+      "loss": 0.8282,
       "step": 24852
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.810260375804898e-07,
-      "loss": 0.5771,
+      "loss": 0.7038,
       "step": 24853
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.80665186831586e-07,
-      "loss": 0.6644,
+      "loss": 0.7262,
       "step": 24854
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.803044681499458e-07,
-      "loss": 0.7278,
+      "loss": 0.8843,
       "step": 24855
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.799438815405766e-07,
-      "loss": 0.7463,
+      "loss": 0.7441,
       "step": 24856
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.795834270084754e-07,
-      "loss": 0.7419,
+      "loss": 0.7432,
       "step": 24857
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.792231045586493e-07,
-      "loss": 0.6939,
+      "loss": 0.796,
       "step": 24858
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.788629141960932e-07,
-      "loss": 0.7796,
+      "loss": 0.7316,
       "step": 24859
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.785028559258054e-07,
-      "loss": 0.5939,
+      "loss": 0.7216,
       "step": 24860
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.781429297527806e-07,
-      "loss": 0.7978,
+      "loss": 0.8803,
       "step": 24861
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.777831356820151e-07,
-      "loss": 0.7599,
+      "loss": 0.8894,
       "step": 24862
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.774234737184957e-07,
-      "loss": 0.7676,
+      "loss": 0.8069,
       "step": 24863
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.770639438672164e-07,
-      "loss": 0.7073,
+      "loss": 0.7992,
       "step": 24864
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.767045461331632e-07,
-      "loss": 0.749,
+      "loss": 0.8673,
       "step": 24865
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7634528052132313e-07,
-      "loss": 0.793,
+      "loss": 0.7647,
       "step": 24866
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7598614703667887e-07,
-      "loss": 0.6425,
+      "loss": 0.7357,
       "step": 24867
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7562714568421654e-07,
-      "loss": 0.7192,
+      "loss": 0.7574,
       "step": 24868
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.752682764689109e-07,
-      "loss": 0.665,
+      "loss": 0.8398,
       "step": 24869
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.749095393957459e-07,
-      "loss": 0.5476,
+      "loss": 0.6883,
       "step": 24870
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7455093446969636e-07,
-      "loss": 0.7777,
+      "loss": 0.8045,
       "step": 24871
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7419246169573827e-07,
-      "loss": 0.7237,
+      "loss": 0.7977,
       "step": 24872
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.738341210788444e-07,
-      "loss": 0.7072,
+      "loss": 0.8074,
       "step": 24873
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7347591262398405e-07,
-      "loss": 0.7474,
+      "loss": 0.7835,
       "step": 24874
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.731178363361333e-07,
-      "loss": 0.8499,
+      "loss": 0.8561,
       "step": 24875
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.727598922202514e-07,
-      "loss": 0.7771,
+      "loss": 0.8221,
       "step": 24876
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.724020802813112e-07,
-      "loss": 0.5592,
+      "loss": 0.7363,
       "step": 24877
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.72044400524273e-07,
-      "loss": 0.7195,
+      "loss": 0.7967,
       "step": 24878
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.716868529541008e-07,
-      "loss": 0.6644,
+      "loss": 0.7915,
       "step": 24879
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7132943757575375e-07,
-      "loss": 0.6919,
+      "loss": 0.7798,
       "step": 24880
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7097215439419364e-07,
-      "loss": 0.7824,
+      "loss": 0.8465,
       "step": 24881
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.7061500341437306e-07,
-      "loss": 0.7599,
+      "loss": 0.7893,
       "step": 24882
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.702579846412492e-07,
-      "loss": 0.8103,
+      "loss": 0.8286,
       "step": 24883
     },
     {
       "epoch": 0.9,
       "learning_rate": 4.699010980797747e-07,
-      "loss": 0.7155,
+      "loss": 0.7777,
       "step": 24884
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6954434373490234e-07,
-      "loss": 0.811,
+      "loss": 0.8754,
       "step": 24885
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.691877216115781e-07,
-      "loss": 0.816,
+      "loss": 0.8219,
       "step": 24886
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.688312317147525e-07,
-      "loss": 0.7392,
+      "loss": 0.8467,
       "step": 24887
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6847487404937166e-07,
-      "loss": 0.7823,
+      "loss": 0.8361,
       "step": 24888
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6811864862037816e-07,
-      "loss": 0.7228,
+      "loss": 0.7985,
       "step": 24889
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.677625554327148e-07,
-      "loss": 0.7654,
+      "loss": 0.8076,
       "step": 24890
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.674065944913209e-07,
-      "loss": 0.7038,
+      "loss": 0.8332,
       "step": 24891
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6705076580113473e-07,
-      "loss": 0.7566,
+      "loss": 0.7765,
       "step": 24892
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.666950693670935e-07,
-      "loss": 0.6781,
+      "loss": 0.7895,
       "step": 24893
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.663395051941344e-07,
-      "loss": 0.6375,
+      "loss": 0.7297,
       "step": 24894
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6598407328718673e-07,
-      "loss": 0.7542,
+      "loss": 0.8481,
       "step": 24895
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6562877365118333e-07,
-      "loss": 0.7167,
+      "loss": 0.785,
       "step": 24896
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.652736062910534e-07,
-      "loss": 0.7881,
+      "loss": 0.8817,
       "step": 24897
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6491857121172326e-07,
-      "loss": 0.6197,
+      "loss": 0.725,
       "step": 24898
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6456366841811875e-07,
-      "loss": 0.7921,
+      "loss": 0.7761,
       "step": 24899
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.642088979151671e-07,
-      "loss": 0.8318,
+      "loss": 0.811,
       "step": 24900
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.638542597077855e-07,
-      "loss": 0.7979,
+      "loss": 0.8284,
       "step": 24901
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6349975380089675e-07,
-      "loss": 0.6272,
+      "loss": 0.6657,
       "step": 24902
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.631453801994179e-07,
-      "loss": 0.6247,
+      "loss": 0.7054,
       "step": 24903
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.627911389082662e-07,
-      "loss": 0.7686,
+      "loss": 0.7846,
       "step": 24904
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.624370299323544e-07,
-      "loss": 0.7194,
+      "loss": 0.8081,
       "step": 24905
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6208305327659853e-07,
-      "loss": 0.6173,
+      "loss": 0.6941,
       "step": 24906
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.61729208945908e-07,
-      "loss": 0.7247,
+      "loss": 0.8315,
       "step": 24907
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6137549694519114e-07,
-      "loss": 0.7166,
+      "loss": 0.7163,
       "step": 24908
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6102191727935507e-07,
-      "loss": 0.6561,
+      "loss": 0.7112,
       "step": 24909
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6066846995330817e-07,
-      "loss": 0.7048,
+      "loss": 0.749,
       "step": 24910
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.603151549719498e-07,
-      "loss": 0.5633,
+      "loss": 0.7384,
       "step": 24911
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.599619723401827e-07,
-      "loss": 0.7277,
+      "loss": 0.7839,
       "step": 24912
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5960892206290964e-07,
-      "loss": 0.7384,
+      "loss": 0.8524,
       "step": 24913
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.592560041450256e-07,
-      "loss": 0.7869,
+      "loss": 0.8093,
       "step": 24914
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.589032185914277e-07,
-      "loss": 0.7417,
+      "loss": 0.8241,
       "step": 24915
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5855056540701217e-07,
-      "loss": 0.621,
+      "loss": 0.8331,
       "step": 24916
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.581980445966694e-07,
-      "loss": 0.6536,
+      "loss": 0.7348,
       "step": 24917
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5784565616528885e-07,
-      "loss": 0.7867,
+      "loss": 0.781,
       "step": 24918
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.574934001177622e-07,
-      "loss": 0.7978,
+      "loss": 0.9575,
       "step": 24919
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5714127645897666e-07,
-      "loss": 0.7587,
+      "loss": 0.7792,
       "step": 24920
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5678928519381607e-07,
-      "loss": 0.6246,
+      "loss": 0.7248,
       "step": 24921
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5643742632716426e-07,
-      "loss": 0.794,
+      "loss": 0.885,
       "step": 24922
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5608569986390184e-07,
-      "loss": 0.7176,
+      "loss": 0.7287,
       "step": 24923
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.557341058089093e-07,
-      "loss": 0.8304,
+      "loss": 0.7967,
       "step": 24924
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.55382644167065e-07,
-      "loss": 0.6828,
+      "loss": 0.7397,
       "step": 24925
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5503131494324395e-07,
-      "loss": 0.7187,
+      "loss": 0.7113,
       "step": 24926
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.546801181423222e-07,
-      "loss": 0.7682,
+      "loss": 0.7844,
       "step": 24927
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5432905376916804e-07,
-      "loss": 0.7883,
+      "loss": 0.7665,
       "step": 24928
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5397812182865763e-07,
-      "loss": 0.5829,
+      "loss": 0.7146,
       "step": 24929
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.53627322325656e-07,
-      "loss": 0.827,
+      "loss": 0.8885,
       "step": 24930
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.532766552650292e-07,
-      "loss": 0.6947,
+      "loss": 0.7921,
       "step": 24931
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5292612065164555e-07,
-      "loss": 0.7412,
+      "loss": 0.845,
       "step": 24932
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.525757184903656e-07,
-      "loss": 0.7604,
+      "loss": 0.789,
       "step": 24933
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.522254487860522e-07,
-      "loss": 0.827,
+      "loss": 0.896,
       "step": 24934
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.518753115435637e-07,
-      "loss": 0.6647,
+      "loss": 0.7481,
       "step": 24935
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5152530676775827e-07,
-      "loss": 0.7505,
+      "loss": 0.8078,
       "step": 24936
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.511754344634911e-07,
-      "loss": 0.8405,
+      "loss": 0.8576,
       "step": 24937
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.508256946356171e-07,
-      "loss": 0.5994,
+      "loss": 0.6977,
       "step": 24938
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.5047608728898905e-07,
-      "loss": 0.7875,
+      "loss": 0.8493,
       "step": 24939
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.501266124284553e-07,
-      "loss": 0.8218,
+      "loss": 0.8518,
       "step": 24940
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.497772700588654e-07,
-      "loss": 0.8507,
+      "loss": 0.8923,
       "step": 24941
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4942806018506757e-07,
-      "loss": 0.7671,
+      "loss": 0.847,
       "step": 24942
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.490789828119013e-07,
-      "loss": 0.8307,
+      "loss": 0.8198,
       "step": 24943
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.487300379442161e-07,
-      "loss": 0.7643,
+      "loss": 0.8384,
       "step": 24944
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4838122558684914e-07,
-      "loss": 0.8103,
+      "loss": 0.8873,
       "step": 24945
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4803254574464105e-07,
-      "loss": 0.7465,
+      "loss": 0.8096,
       "step": 24946
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.476839984224279e-07,
-      "loss": 0.6631,
+      "loss": 0.7298,
       "step": 24947
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4733558362504816e-07,
-      "loss": 0.7584,
+      "loss": 0.8036,
       "step": 24948
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4698730135733335e-07,
-      "loss": 0.8358,
+      "loss": 0.8641,
       "step": 24949
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.466391516241142e-07,
-      "loss": 0.7051,
+      "loss": 0.7759,
       "step": 24950
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4629113443022343e-07,
-      "loss": 0.7169,
+      "loss": 0.7282,
       "step": 24951
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.459432497804883e-07,
-      "loss": 0.6875,
+      "loss": 0.7675,
       "step": 24952
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.45595497679735e-07,
-      "loss": 0.6876,
+      "loss": 0.7578,
       "step": 24953
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4524787813278737e-07,
-      "loss": 0.8524,
+      "loss": 0.8103,
       "step": 24954
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.449003911444716e-07,
-      "loss": 0.6485,
+      "loss": 0.7077,
       "step": 24955
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.445530367196027e-07,
-      "loss": 0.7994,
+      "loss": 0.7849,
       "step": 24956
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4420581486300464e-07,
-      "loss": 0.8043,
+      "loss": 0.8303,
       "step": 24957
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.438587255794935e-07,
-      "loss": 0.7676,
+      "loss": 0.8959,
       "step": 24958
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4351176887388324e-07,
-      "loss": 0.7649,
+      "loss": 0.728,
       "step": 24959
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.431649447509878e-07,
-      "loss": 0.6493,
+      "loss": 0.8461,
       "step": 24960
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.428182532156222e-07,
-      "loss": 0.6074,
+      "loss": 0.7263,
       "step": 24961
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.424716942725904e-07,
-      "loss": 0.6622,
+      "loss": 0.7224,
       "step": 24962
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4212526792670406e-07,
-      "loss": 0.6771,
+      "loss": 0.7469,
       "step": 24963
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.417789741827694e-07,
-      "loss": 0.645,
+      "loss": 0.7844,
       "step": 24964
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4143281304559027e-07,
-      "loss": 0.8949,
+      "loss": 0.8938,
       "step": 24965
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4108678451996845e-07,
-      "loss": 0.6785,
+      "loss": 0.8182,
       "step": 24966
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4074088861070674e-07,
-      "loss": 0.7217,
+      "loss": 0.814,
       "step": 24967
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.4039512532260134e-07,
-      "loss": 0.6736,
+      "loss": 0.8226,
       "step": 24968
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.400494946604483e-07,
-      "loss": 0.6423,
+      "loss": 0.7604,
       "step": 24969
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3970399662904726e-07,
-      "loss": 0.7092,
+      "loss": 0.7748,
       "step": 24970
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.393586312331888e-07,
-      "loss": 0.7305,
+      "loss": 0.8271,
       "step": 24971
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3901339847766343e-07,
-      "loss": 0.8102,
+      "loss": 0.8006,
       "step": 24972
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3866829836726186e-07,
-      "loss": 0.7112,
+      "loss": 0.85,
       "step": 24973
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.383233309067736e-07,
-      "loss": 0.7321,
+      "loss": 0.8066,
       "step": 24974
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.379784961009803e-07,
-      "loss": 0.7366,
+      "loss": 0.8086,
       "step": 24975
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3763379395467043e-07,
-      "loss": 0.6775,
+      "loss": 0.7613,
       "step": 24976
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3728922447262454e-07,
-      "loss": 0.778,
+      "loss": 0.7568,
       "step": 24977
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3694478765962336e-07,
-      "loss": 0.7212,
+      "loss": 0.7456,
       "step": 24978
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.366004835204429e-07,
-      "loss": 0.6548,
+      "loss": 0.6922,
       "step": 24979
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.36256312059864e-07,
-      "loss": 0.7783,
+      "loss": 0.7833,
       "step": 24980
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3591227328265815e-07,
-      "loss": 0.8273,
+      "loss": 0.8418,
       "step": 24981
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3556836719360064e-07,
-      "loss": 0.7465,
+      "loss": 0.8785,
       "step": 24982
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3522459379746197e-07,
-      "loss": 0.7309,
+      "loss": 0.8175,
       "step": 24983
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.348809530990117e-07,
-      "loss": 0.8069,
+      "loss": 0.8762,
       "step": 24984
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.34537445103016e-07,
-      "loss": 0.8475,
+      "loss": 0.7368,
       "step": 24985
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3419406981424217e-07,
-      "loss": 0.8309,
+      "loss": 0.8793,
       "step": 24986
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3385082723745533e-07,
-      "loss": 0.7115,
+      "loss": 0.8049,
       "step": 24987
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.335077173774127e-07,
-      "loss": 0.6873,
+      "loss": 0.7179,
       "step": 24988
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.331647402388794e-07,
-      "loss": 0.7127,
+      "loss": 0.812,
       "step": 24989
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.328218958266106e-07,
-      "loss": 0.7993,
+      "loss": 0.8424,
       "step": 24990
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.324791841453646e-07,
-      "loss": 0.7477,
+      "loss": 0.8287,
       "step": 24991
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.321366051998943e-07,
-      "loss": 0.6272,
+      "loss": 0.7335,
       "step": 24992
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.317941589949559e-07,
-      "loss": 0.7441,
+      "loss": 0.7656,
       "step": 24993
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3145184553529564e-07,
-      "loss": 0.7141,
+      "loss": 0.7911,
       "step": 24994
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.311096648256663e-07,
-      "loss": 0.8826,
+      "loss": 0.8208,
       "step": 24995
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3076761687081416e-07,
-      "loss": 0.8706,
+      "loss": 0.853,
       "step": 24996
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3042570167548426e-07,
-      "loss": 0.6795,
+      "loss": 0.7349,
       "step": 24997
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.3008391924441837e-07,
-      "loss": 0.7254,
+      "loss": 0.7907,
       "step": 24998
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2974226958236163e-07,
-      "loss": 0.7163,
+      "loss": 0.756,
       "step": 24999
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2940075269405244e-07,
-      "loss": 0.6979,
+      "loss": 0.6868,
       "step": 25000
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2905936858422926e-07,
-      "loss": 0.7171,
+      "loss": 0.7592,
       "step": 25001
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2871811725762824e-07,
-      "loss": 0.7463,
+      "loss": 0.7882,
       "step": 25002
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.283769987189834e-07,
-      "loss": 0.6563,
+      "loss": 0.7603,
       "step": 25003
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2803601297302544e-07,
-      "loss": 0.9214,
+      "loss": 0.8936,
       "step": 25004
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.276951600244883e-07,
-      "loss": 0.6587,
+      "loss": 0.7105,
       "step": 25005
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2735443987809934e-07,
-      "loss": 0.7682,
+      "loss": 0.763,
       "step": 25006
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2701385253858697e-07,
-      "loss": 0.7399,
+      "loss": 0.8251,
       "step": 25007
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.266733980106741e-07,
-      "loss": 0.777,
+      "loss": 0.8226,
       "step": 25008
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.263330762990847e-07,
-      "loss": 0.6839,
+      "loss": 0.7542,
       "step": 25009
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2599288740854173e-07,
-      "loss": 0.8076,
+      "loss": 0.8246,
       "step": 25010
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2565283134376134e-07,
-      "loss": 0.71,
+      "loss": 0.8571,
       "step": 25011
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2531290810946646e-07,
-      "loss": 0.7612,
+      "loss": 0.8333,
       "step": 25012
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2497311771036884e-07,
-      "loss": 0.8347,
+      "loss": 0.8646,
       "step": 25013
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.246334601511837e-07,
-      "loss": 0.7876,
+      "loss": 0.7879,
       "step": 25014
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2429393543662487e-07,
-      "loss": 0.7531,
+      "loss": 0.778,
       "step": 25015
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2395454357140096e-07,
-      "loss": 0.6926,
+      "loss": 0.7751,
       "step": 25016
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2361528456021925e-07,
-      "loss": 0.6375,
+      "loss": 0.6857,
       "step": 25017
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2327615840779046e-07,
-      "loss": 0.673,
+      "loss": 0.8119,
       "step": 25018
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2293716511881743e-07,
-      "loss": 0.5925,
+      "loss": 0.6627,
       "step": 25019
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2259830469800314e-07,
-      "loss": 0.8158,
+      "loss": 0.8646,
       "step": 25020
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2225957715004704e-07,
-      "loss": 0.6793,
+      "loss": 0.7196,
       "step": 25021
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.219209824796533e-07,
-      "loss": 0.7067,
+      "loss": 0.7388,
       "step": 25022
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.215825206915136e-07,
-      "loss": 0.7384,
+      "loss": 0.7849,
       "step": 25023
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.212441917903276e-07,
-      "loss": 0.7333,
+      "loss": 0.8437,
       "step": 25024
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.2090599578078815e-07,
-      "loss": 0.7621,
+      "loss": 0.8358,
       "step": 25025
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.205679326675871e-07,
-      "loss": 0.5914,
+      "loss": 0.6517,
       "step": 25026
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.202300024554151e-07,
-      "loss": 0.808,
+      "loss": 0.782,
       "step": 25027
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1989220514895957e-07,
-      "loss": 0.8366,
+      "loss": 0.8711,
       "step": 25028
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.195545407529067e-07,
-      "loss": 0.6767,
+      "loss": 0.7651,
       "step": 25029
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1921700927194053e-07,
-      "loss": 0.7404,
+      "loss": 0.7914,
       "step": 25030
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.188796107107473e-07,
-      "loss": 0.7657,
+      "loss": 0.7503,
       "step": 25031
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.185423450740045e-07,
-      "loss": 0.6087,
+      "loss": 0.689,
       "step": 25032
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1820521236639266e-07,
-      "loss": 0.7846,
+      "loss": 0.7748,
       "step": 25033
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1786821259258816e-07,
-      "loss": 0.8109,
+      "loss": 0.7926,
       "step": 25034
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.175313457572683e-07,
-      "loss": 0.609,
+      "loss": 0.7416,
       "step": 25035
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.171946118651027e-07,
-      "loss": 0.636,
+      "loss": 0.7678,
       "step": 25036
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.168580109207676e-07,
-      "loss": 0.8057,
+      "loss": 0.8065,
       "step": 25037
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1652154292893045e-07,
-      "loss": 0.7742,
+      "loss": 0.798,
       "step": 25038
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1618520789426075e-07,
-      "loss": 0.7789,
+      "loss": 0.7914,
       "step": 25039
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1584900582142153e-07,
-      "loss": 0.7151,
+      "loss": 0.8222,
       "step": 25040
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.155129367150812e-07,
-      "loss": 0.8157,
+      "loss": 0.827,
       "step": 25041
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.151770005798994e-07,
-      "loss": 0.715,
+      "loss": 0.7244,
       "step": 25042
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1484119742053685e-07,
-      "loss": 0.687,
+      "loss": 0.8483,
       "step": 25043
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1450552724165426e-07,
-      "loss": 0.846,
+      "loss": 0.8215,
       "step": 25044
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1416999004790794e-07,
-      "loss": 0.6957,
+      "loss": 0.7634,
       "step": 25045
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1383458584395187e-07,
-      "loss": 0.6751,
+      "loss": 0.7668,
       "step": 25046
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1349931463444016e-07,
-      "loss": 0.7272,
+      "loss": 0.8297,
       "step": 25047
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.131641764240235e-07,
-      "loss": 0.7215,
+      "loss": 0.8092,
       "step": 25048
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1282917121735155e-07,
-      "loss": 0.8063,
+      "loss": 0.8286,
       "step": 25049
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.124942990190728e-07,
-      "loss": 0.7843,
+      "loss": 0.7482,
       "step": 25050
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1215955983383236e-07,
-      "loss": 0.6629,
+      "loss": 0.7809,
       "step": 25051
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.118249536662755e-07,
-      "loss": 0.6577,
+      "loss": 0.7298,
       "step": 25052
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.114904805210429e-07,
-      "loss": 0.8213,
+      "loss": 0.8191,
       "step": 25053
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.111561404027775e-07,
-      "loss": 0.7444,
+      "loss": 0.8183,
       "step": 25054
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.108219333161123e-07,
-      "loss": 0.7313,
+      "loss": 0.7849,
       "step": 25055
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1048785926569023e-07,
-      "loss": 0.6687,
+      "loss": 0.7984,
       "step": 25056
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.1015391825614316e-07,
-      "loss": 0.8029,
+      "loss": 0.8311,
       "step": 25057
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.09820110292104e-07,
-      "loss": 0.7601,
+      "loss": 0.7607,
       "step": 25058
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.094864353782024e-07,
-      "loss": 0.7616,
+      "loss": 0.8253,
       "step": 25059
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.091528935190725e-07,
-      "loss": 0.7209,
+      "loss": 0.7921,
       "step": 25060
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0881948471933607e-07,
-      "loss": 0.6221,
+      "loss": 0.7487,
       "step": 25061
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.084862089836217e-07,
-      "loss": 0.7883,
+      "loss": 0.8871,
       "step": 25062
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.081530663165534e-07,
-      "loss": 0.6785,
+      "loss": 0.7877,
       "step": 25063
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0782005672275193e-07,
-      "loss": 0.8353,
+      "loss": 0.8183,
       "step": 25064
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0748718020683697e-07,
-      "loss": 0.6423,
+      "loss": 0.7271,
       "step": 25065
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0715443677342815e-07,
-      "loss": 0.5776,
+      "loss": 0.6732,
       "step": 25066
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0682182642714177e-07,
-      "loss": 0.7441,
+      "loss": 0.887,
       "step": 25067
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0648934917258964e-07,
-      "loss": 0.7205,
+      "loss": 0.7692,
       "step": 25068
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0615700501438703e-07,
-      "loss": 0.7034,
+      "loss": 0.7679,
       "step": 25069
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.058247939571447e-07,
-      "loss": 0.716,
+      "loss": 0.8029,
       "step": 25070
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.054927160054711e-07,
-      "loss": 0.7252,
+      "loss": 0.7678,
       "step": 25071
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0516077116397155e-07,
-      "loss": 0.6247,
+      "loss": 0.7476,
       "step": 25072
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0482895943725565e-07,
-      "loss": 0.6911,
+      "loss": 0.7715,
       "step": 25073
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.044972808299219e-07,
-      "loss": 0.6579,
+      "loss": 0.7653,
       "step": 25074
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0416573534657553e-07,
-      "loss": 0.8873,
+      "loss": 0.8946,
       "step": 25075
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.038343229918151e-07,
-      "loss": 0.7,
+      "loss": 0.7563,
       "step": 25076
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.03503043770237e-07,
-      "loss": 0.7954,
+      "loss": 0.8514,
       "step": 25077
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.031718976864385e-07,
-      "loss": 0.7638,
+      "loss": 0.8401,
       "step": 25078
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0284088474501604e-07,
-      "loss": 0.6991,
+      "loss": 0.7917,
       "step": 25079
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0251000495055703e-07,
-      "loss": 0.6276,
+      "loss": 0.7623,
       "step": 25080
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.021792583076556e-07,
-      "loss": 0.6355,
+      "loss": 0.7303,
       "step": 25081
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0184864482090025e-07,
-      "loss": 0.6918,
+      "loss": 0.758,
       "step": 25082
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.015181644948762e-07,
-      "loss": 0.8469,
+      "loss": 0.8337,
       "step": 25083
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0118781733416877e-07,
-      "loss": 0.6281,
+      "loss": 0.6899,
       "step": 25084
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.008576033433609e-07,
-      "loss": 0.6619,
+      "loss": 0.7237,
       "step": 25085
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.0052752252703773e-07,
-      "loss": 0.6557,
+      "loss": 0.7653,
       "step": 25086
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.001975748897713e-07,
-      "loss": 0.7003,
+      "loss": 0.7541,
       "step": 25087
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.998677604361456e-07,
-      "loss": 0.6373,
+      "loss": 0.7513,
       "step": 25088
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9953807917073263e-07,
-      "loss": 0.5852,
+      "loss": 0.7088,
       "step": 25089
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9920853109810865e-07,
-      "loss": 0.7139,
+      "loss": 0.7697,
       "step": 25090
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9887911622284225e-07,
-      "loss": 0.6011,
+      "loss": 0.7168,
       "step": 25091
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.985498345495098e-07,
-      "loss": 0.6751,
+      "loss": 0.7968,
       "step": 25092
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.982206860826721e-07,
-      "loss": 0.8493,
+      "loss": 0.9015,
       "step": 25093
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9789167082690093e-07,
-      "loss": 0.7583,
+      "loss": 0.8101,
       "step": 25094
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.975627887867595e-07,
-      "loss": 0.7854,
+      "loss": 0.7544,
       "step": 25095
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9723403996680955e-07,
-      "loss": 0.7464,
+      "loss": 0.8028,
       "step": 25096
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9690542437161193e-07,
-      "loss": 0.7589,
+      "loss": 0.8261,
       "step": 25097
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9657694200572747e-07,
-      "loss": 0.7425,
+      "loss": 0.8134,
       "step": 25098
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9624859287371363e-07,
-      "loss": 0.6769,
+      "loss": 0.7697,
       "step": 25099
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.959203769801234e-07,
-      "loss": 0.6894,
+      "loss": 0.7219,
       "step": 25100
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.955922943295121e-07,
-      "loss": 0.6365,
+      "loss": 0.7634,
       "step": 25101
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.952643449264304e-07,
-      "loss": 0.7245,
+      "loss": 0.7772,
       "step": 25102
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.949365287754292e-07,
-      "loss": 0.6538,
+      "loss": 0.8042,
       "step": 25103
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9460884588105376e-07,
-      "loss": 0.7718,
+      "loss": 0.7859,
       "step": 25104
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.942812962478548e-07,
-      "loss": 0.717,
+      "loss": 0.8417,
       "step": 25105
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9395387988037216e-07,
-      "loss": 0.6738,
+      "loss": 0.7551,
       "step": 25106
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9362659678315205e-07,
-      "loss": 0.7945,
+      "loss": 0.8652,
       "step": 25107
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.932994469607321e-07,
-      "loss": 0.7049,
+      "loss": 0.7774,
       "step": 25108
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9297243041765297e-07,
-      "loss": 0.6006,
+      "loss": 0.7261,
       "step": 25109
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9264554715844895e-07,
-      "loss": 0.7209,
+      "loss": 0.7855,
       "step": 25110
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9231879718765966e-07,
-      "loss": 0.7322,
+      "loss": 0.8337,
       "step": 25111
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.919921805098137e-07,
-      "loss": 0.7914,
+      "loss": 0.8181,
       "step": 25112
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9166569712944524e-07,
-      "loss": 0.8009,
+      "loss": 0.8124,
       "step": 25113
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.913393470510818e-07,
-      "loss": 0.6817,
+      "loss": 0.7167,
       "step": 25114
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9101313027925303e-07,
-      "loss": 0.7627,
+      "loss": 0.8076,
       "step": 25115
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.90687046818482e-07,
-      "loss": 0.6747,
+      "loss": 0.8009,
       "step": 25116
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9036109667329623e-07,
-      "loss": 0.7323,
+      "loss": 0.8549,
       "step": 25117
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.9003527984821545e-07,
-      "loss": 0.6953,
+      "loss": 0.8014,
       "step": 25118
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.897095963477604e-07,
-      "loss": 0.7798,
+      "loss": 0.7875,
       "step": 25119
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.893840461764475e-07,
-      "loss": 0.7627,
+      "loss": 0.8718,
       "step": 25120
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.890586293387988e-07,
-      "loss": 0.7364,
+      "loss": 0.7986,
       "step": 25121
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8873334583932386e-07,
-      "loss": 0.7623,
+      "loss": 0.7686,
       "step": 25122
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8840819568253585e-07,
-      "loss": 0.8016,
+      "loss": 0.8163,
       "step": 25123
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8808317887295e-07,
-      "loss": 0.8005,
+      "loss": 0.914,
       "step": 25124
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8775829541506937e-07,
-      "loss": 0.6652,
+      "loss": 0.7495,
       "step": 25125
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.874335453134059e-07,
-      "loss": 0.6815,
+      "loss": 0.7737,
       "step": 25126
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8710892857246384e-07,
-      "loss": 0.7388,
+      "loss": 0.8452,
       "step": 25127
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8678444519674617e-07,
-      "loss": 0.6048,
+      "loss": 0.6618,
       "step": 25128
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.864600951907538e-07,
-      "loss": 0.6309,
+      "loss": 0.7465,
       "step": 25129
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.861358785589897e-07,
-      "loss": 0.6533,
+      "loss": 0.771,
       "step": 25130
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8581179530594924e-07,
-      "loss": 0.7682,
+      "loss": 0.849,
       "step": 25131
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.854878454361288e-07,
-      "loss": 0.7817,
+      "loss": 0.8403,
       "step": 25132
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8516402895402373e-07,
-      "loss": 0.7534,
+      "loss": 0.7843,
       "step": 25133
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.84840345864127e-07,
-      "loss": 0.6254,
+      "loss": 0.7767,
       "step": 25134
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.845167961709262e-07,
-      "loss": 0.8184,
+      "loss": 0.8221,
       "step": 25135
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8419337987891323e-07,
-      "loss": 0.7535,
+      "loss": 0.7846,
       "step": 25136
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.838700969925746e-07,
-      "loss": 0.7393,
+      "loss": 0.77,
       "step": 25137
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8354694751639555e-07,
-      "loss": 0.5988,
+      "loss": 0.8489,
       "step": 25138
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8322393145485694e-07,
-      "loss": 0.8412,
+      "loss": 0.7817,
       "step": 25139
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8290104881244405e-07,
-      "loss": 0.7116,
+      "loss": 0.7695,
       "step": 25140
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8257829959363224e-07,
-      "loss": 0.6813,
+      "loss": 0.7547,
       "step": 25141
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8225568380290125e-07,
-      "loss": 0.7841,
+      "loss": 0.7407,
       "step": 25142
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8193320144472745e-07,
-      "loss": 0.6671,
+      "loss": 0.7234,
       "step": 25143
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.81610852523584e-07,
-      "loss": 0.7853,
+      "loss": 0.8153,
       "step": 25144
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.8128863704394394e-07,
-      "loss": 0.7794,
+      "loss": 0.7523,
       "step": 25145
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.80966555010277e-07,
-      "loss": 0.7824,
+      "loss": 0.9004,
       "step": 25146
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.806446064270508e-07,
-      "loss": 0.7315,
+      "loss": 0.7886,
       "step": 25147
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.803227912987317e-07,
-      "loss": 0.6887,
+      "loss": 0.7804,
       "step": 25148
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.800011096297862e-07,
-      "loss": 0.7922,
+      "loss": 0.798,
       "step": 25149
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7967956142467624e-07,
-      "loss": 0.834,
+      "loss": 0.8612,
       "step": 25150
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7935814668786373e-07,
-      "loss": 0.7107,
+      "loss": 0.7905,
       "step": 25151
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7903686542380525e-07,
-      "loss": 0.8015,
+      "loss": 0.8107,
       "step": 25152
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.787157176369627e-07,
-      "loss": 0.6826,
+      "loss": 0.7944,
       "step": 25153
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.783947033317858e-07,
-      "loss": 0.6522,
+      "loss": 0.802,
       "step": 25154
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.780738225127323e-07,
-      "loss": 0.689,
+      "loss": 0.7005,
       "step": 25155
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7775307518425284e-07,
-      "loss": 0.699,
+      "loss": 0.9113,
       "step": 25156
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.774324613507985e-07,
-      "loss": 0.7369,
+      "loss": 0.7569,
       "step": 25157
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7711198101681335e-07,
-      "loss": 0.7322,
+      "loss": 0.864,
       "step": 25158
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.7679163418674946e-07,
-      "loss": 0.8012,
+      "loss": 0.8355,
       "step": 25159
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7647142086504664e-07,
-      "loss": 0.702,
+      "loss": 0.7824,
       "step": 25160
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7615134105614795e-07,
-      "loss": 0.7205,
+      "loss": 0.7649,
       "step": 25161
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.758313947644954e-07,
-      "loss": 0.7056,
+      "loss": 0.7894,
       "step": 25162
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.755115819945288e-07,
-      "loss": 0.6633,
+      "loss": 0.7541,
       "step": 25163
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.751919027506823e-07,
-      "loss": 0.6276,
+      "loss": 0.8079,
       "step": 25164
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.748723570373913e-07,
-      "loss": 0.8039,
+      "loss": 0.8682,
       "step": 25165
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7455294485909344e-07,
-      "loss": 0.7726,
+      "loss": 0.8308,
       "step": 25166
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7423366622021395e-07,
-      "loss": 0.761,
+      "loss": 0.8064,
       "step": 25167
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.73914521125186e-07,
-      "loss": 0.6876,
+      "loss": 0.7666,
       "step": 25168
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.73595509578436e-07,
-      "loss": 0.6774,
+      "loss": 0.761,
       "step": 25169
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7327663158439056e-07,
-      "loss": 0.7392,
+      "loss": 0.803,
       "step": 25170
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.729578871474726e-07,
-      "loss": 0.8202,
+      "loss": 0.8402,
       "step": 25171
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.726392762721076e-07,
-      "loss": 0.7418,
+      "loss": 0.7732,
       "step": 25172
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7232079896270977e-07,
-      "loss": 0.6701,
+      "loss": 0.7098,
       "step": 25173
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7200245522370226e-07,
-      "loss": 0.6694,
+      "loss": 0.7768,
       "step": 25174
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.716842450595015e-07,
-      "loss": 0.6759,
+      "loss": 0.715,
       "step": 25175
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7136616847452066e-07,
-      "loss": 0.696,
+      "loss": 0.7323,
       "step": 25176
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7104822547317065e-07,
-      "loss": 0.7984,
+      "loss": 0.8526,
       "step": 25177
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7073041605986793e-07,
-      "loss": 0.7407,
+      "loss": 0.7795,
       "step": 25178
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.7041274023901785e-07,
-      "loss": 0.6781,
+      "loss": 0.735,
       "step": 25179
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.700951980150269e-07,
-      "loss": 0.6072,
+      "loss": 0.7119,
       "step": 25180
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6977778939230267e-07,
-      "loss": 0.6924,
+      "loss": 0.771,
       "step": 25181
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6946051437524834e-07,
-      "loss": 0.7983,
+      "loss": 0.883,
       "step": 25182
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6914337296826585e-07,
-      "loss": 0.7682,
+      "loss": 0.8288,
       "step": 25183
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.68826365175754e-07,
-      "loss": 0.876,
+      "loss": 0.8295,
       "step": 25184
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.685094910021125e-07,
-      "loss": 0.6845,
+      "loss": 0.8394,
       "step": 25185
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.681927504517346e-07,
-      "loss": 0.8014,
+      "loss": 0.7398,
       "step": 25186
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.678761435290179e-07,
-      "loss": 0.6327,
+      "loss": 0.716,
       "step": 25187
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.675596702383533e-07,
-      "loss": 0.6991,
+      "loss": 0.7705,
       "step": 25188
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.672433305841328e-07,
-      "loss": 0.7967,
+      "loss": 0.8702,
       "step": 25189
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6692712457074177e-07,
-      "loss": 0.6571,
+      "loss": 0.7347,
       "step": 25190
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.666110522025712e-07,
-      "loss": 0.7351,
+      "loss": 0.8368,
       "step": 25191
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6629511348400315e-07,
-      "loss": 0.718,
+      "loss": 0.7912,
       "step": 25192
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6597930841942184e-07,
-      "loss": 0.7479,
+      "loss": 0.8123,
       "step": 25193
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.656636370132094e-07,
-      "loss": 0.7779,
+      "loss": 0.8669,
       "step": 25194
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.653480992697445e-07,
-      "loss": 0.7141,
+      "loss": 0.8151,
       "step": 25195
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.650326951934047e-07,
-      "loss": 0.6787,
+      "loss": 0.7377,
       "step": 25196
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.647174247885654e-07,
-      "loss": 0.7808,
+      "loss": 0.8704,
       "step": 25197
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6440228805960433e-07,
-      "loss": 0.7417,
+      "loss": 0.6991,
       "step": 25198
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6408728501088673e-07,
-      "loss": 0.853,
+      "loss": 0.8854,
       "step": 25199
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.637724156467881e-07,
-      "loss": 0.7284,
+      "loss": 0.804,
       "step": 25200
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.634576799716749e-07,
-      "loss": 0.702,
+      "loss": 0.7979,
       "step": 25201
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6314307798991364e-07,
-      "loss": 0.6003,
+      "loss": 0.6867,
       "step": 25202
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.628286097058686e-07,
-      "loss": 0.8232,
+      "loss": 0.8954,
       "step": 25203
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6251427512390523e-07,
-      "loss": 0.7577,
+      "loss": 0.7986,
       "step": 25204
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.622000742483811e-07,
-      "loss": 0.7957,
+      "loss": 0.8305,
       "step": 25205
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.618860070836572e-07,
-      "loss": 0.682,
+      "loss": 0.6924,
       "step": 25206
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6157207363408995e-07,
-      "loss": 0.7325,
+      "loss": 0.7421,
       "step": 25207
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.61258273904036e-07,
-      "loss": 0.6622,
+      "loss": 0.7905,
       "step": 25208
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.609446078978451e-07,
-      "loss": 0.6736,
+      "loss": 0.781,
       "step": 25209
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.606310756198761e-07,
-      "loss": 0.7758,
+      "loss": 0.844,
       "step": 25210
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.6031767707447096e-07,
-      "loss": 0.8019,
+      "loss": 0.8292,
       "step": 25211
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.600044122659818e-07,
-      "loss": 0.7312,
+      "loss": 0.8027,
       "step": 25212
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5969128119875517e-07,
-      "loss": 0.7683,
+      "loss": 0.8046,
       "step": 25213
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5937828387713316e-07,
-      "loss": 0.8501,
+      "loss": 0.8834,
       "step": 25214
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.590654203054589e-07,
-      "loss": 0.7634,
+      "loss": 0.7876,
       "step": 25215
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.587526904880745e-07,
-      "loss": 0.6653,
+      "loss": 0.7888,
       "step": 25216
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.584400944293176e-07,
-      "loss": 0.6809,
+      "loss": 0.765,
       "step": 25217
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5812763213352255e-07,
-      "loss": 0.7184,
+      "loss": 0.8333,
       "step": 25218
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5781530360502914e-07,
-      "loss": 0.7668,
+      "loss": 0.8474,
       "step": 25219
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.575031088481662e-07,
-      "loss": 0.7418,
+      "loss": 0.8109,
       "step": 25220
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.57191047867268e-07,
-      "loss": 0.801,
+      "loss": 0.8752,
       "step": 25221
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.568791206666611e-07,
-      "loss": 0.7642,
+      "loss": 0.8135,
       "step": 25222
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5656732725067645e-07,
-      "loss": 0.6778,
+      "loss": 0.755,
       "step": 25223
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5625566762363615e-07,
-      "loss": 0.6943,
+      "loss": 0.7312,
       "step": 25224
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.559441417898668e-07,
-      "loss": 0.7231,
+      "loss": 0.8333,
       "step": 25225
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.556327497536893e-07,
-      "loss": 0.7888,
+      "loss": 0.8799,
       "step": 25226
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.553214915194236e-07,
-      "loss": 0.7415,
+      "loss": 0.7899,
       "step": 25227
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5501036709138735e-07,
-      "loss": 0.6599,
+      "loss": 0.7446,
       "step": 25228
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.546993764738982e-07,
-      "loss": 0.9055,
+      "loss": 0.9294,
       "step": 25229
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.543885196712715e-07,
-      "loss": 0.6772,
+      "loss": 0.7871,
       "step": 25230
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.540777966878184e-07,
-      "loss": 0.7612,
+      "loss": 0.8203,
       "step": 25231
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5376720752784976e-07,
-      "loss": 0.8372,
+      "loss": 0.8538,
       "step": 25232
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.534567521956744e-07,
-      "loss": 0.761,
+      "loss": 0.8097,
       "step": 25233
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5314643069560006e-07,
-      "loss": 0.7678,
+      "loss": 0.8029,
       "step": 25234
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.528362430319321e-07,
-      "loss": 0.7057,
+      "loss": 0.7797,
       "step": 25235
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5252618920897376e-07,
-      "loss": 0.7313,
+      "loss": 0.766,
       "step": 25236
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5221626923102716e-07,
-      "loss": 0.8135,
+      "loss": 0.7975,
       "step": 25237
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5190648310239107e-07,
-      "loss": 0.7157,
+      "loss": 0.7904,
       "step": 25238
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.515968308273643e-07,
-      "loss": 0.7837,
+      "loss": 0.8393,
       "step": 25239
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5128731241024115e-07,
-      "loss": 0.8074,
+      "loss": 0.8472,
       "step": 25240
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.50977927855316e-07,
-      "loss": 0.6106,
+      "loss": 0.7759,
       "step": 25241
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5066867716688325e-07,
-      "loss": 0.739,
+      "loss": 0.7427,
       "step": 25242
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.503595603492327e-07,
-      "loss": 0.6593,
+      "loss": 0.7719,
       "step": 25243
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.5005057740665095e-07,
-      "loss": 0.6504,
+      "loss": 0.7637,
       "step": 25244
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.497417283434268e-07,
-      "loss": 0.7422,
+      "loss": 0.8269,
       "step": 25245
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.494330131638435e-07,
-      "loss": 0.6927,
+      "loss": 0.7311,
       "step": 25246
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4912443187218424e-07,
-      "loss": 0.6706,
+      "loss": 0.7733,
       "step": 25247
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.488159844727312e-07,
-      "loss": 0.7533,
+      "loss": 0.7444,
       "step": 25248
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.485076709697643e-07,
-      "loss": 0.768,
+      "loss": 0.8263,
       "step": 25249
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4819949136755906e-07,
-      "loss": 0.8009,
+      "loss": 0.812,
       "step": 25250
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.478914456703908e-07,
-      "loss": 0.7555,
+      "loss": 0.8598,
       "step": 25251
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4758353388253617e-07,
-      "loss": 0.6604,
+      "loss": 0.7595,
       "step": 25252
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.472757560082629e-07,
-      "loss": 0.711,
+      "loss": 0.7545,
       "step": 25253
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.469681120518442e-07,
-      "loss": 0.6196,
+      "loss": 0.7116,
       "step": 25254
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.466606020175467e-07,
-      "loss": 0.7019,
+      "loss": 0.8697,
       "step": 25255
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.46353225909638e-07,
-      "loss": 0.7051,
+      "loss": 0.768,
       "step": 25256
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.460459837323804e-07,
-      "loss": 0.6725,
+      "loss": 0.7246,
       "step": 25257
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4573887549003817e-07,
-      "loss": 0.7019,
+      "loss": 0.7771,
       "step": 25258
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.454319011868712e-07,
-      "loss": 0.6363,
+      "loss": 0.7131,
       "step": 25259
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4512506082713836e-07,
-      "loss": 0.6455,
+      "loss": 0.7319,
       "step": 25260
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.448183544150974e-07,
-      "loss": 0.6257,
+      "loss": 0.7353,
       "step": 25261
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4451178195500255e-07,
-      "loss": 0.7775,
+      "loss": 0.8138,
       "step": 25262
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.442053434511072e-07,
-      "loss": 0.8242,
+      "loss": 0.8733,
       "step": 25263
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.438990389076624e-07,
-      "loss": 0.83,
+      "loss": 0.8253,
       "step": 25264
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4359286832892025e-07,
-      "loss": 0.6796,
+      "loss": 0.7809,
       "step": 25265
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4328683171912403e-07,
-      "loss": 0.8488,
+      "loss": 0.8368,
       "step": 25266
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4298092908252367e-07,
-      "loss": 0.7261,
+      "loss": 0.7659,
       "step": 25267
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4267516042336026e-07,
-      "loss": 0.7666,
+      "loss": 0.7968,
       "step": 25268
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4236952574587814e-07,
-      "loss": 0.677,
+      "loss": 0.7496,
       "step": 25269
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.42064025054315e-07,
-      "loss": 0.6852,
+      "loss": 0.7984,
       "step": 25270
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.41758658352912e-07,
-      "loss": 0.748,
+      "loss": 0.8184,
       "step": 25271
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4145342564590345e-07,
-      "loss": 0.7703,
+      "loss": 0.9283,
       "step": 25272
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.41148326937526e-07,
-      "loss": 0.7583,
+      "loss": 0.7885,
       "step": 25273
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4084336223201063e-07,
-      "loss": 0.75,
+      "loss": 0.8673,
       "step": 25274
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4053853153358964e-07,
-      "loss": 0.7507,
+      "loss": 0.7653,
       "step": 25275
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.4023383484649064e-07,
-      "loss": 0.9175,
+      "loss": 0.8877,
       "step": 25276
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3992927217494254e-07,
-      "loss": 0.7906,
+      "loss": 0.8697,
       "step": 25277
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.396248435231697e-07,
-      "loss": 0.7049,
+      "loss": 0.8347,
       "step": 25278
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3932054889539543e-07,
-      "loss": 0.7018,
+      "loss": 0.7467,
       "step": 25279
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3901638829584193e-07,
-      "loss": 0.7331,
+      "loss": 0.8224,
       "step": 25280
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.387123617287291e-07,
-      "loss": 0.6464,
+      "loss": 0.7362,
       "step": 25281
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.384084691982759e-07,
-      "loss": 0.7004,
+      "loss": 0.8318,
       "step": 25282
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3810471070869434e-07,
-      "loss": 0.6812,
+      "loss": 0.7878,
       "step": 25283
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.378010862642045e-07,
-      "loss": 0.7837,
+      "loss": 0.8447,
       "step": 25284
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.374975958690141e-07,
-      "loss": 0.6893,
+      "loss": 0.7897,
       "step": 25285
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3719423952733533e-07,
-      "loss": 0.8401,
+      "loss": 0.8585,
       "step": 25286
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.36891017243377e-07,
-      "loss": 0.7526,
+      "loss": 0.7316,
       "step": 25287
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3658792902134584e-07,
-      "loss": 0.7182,
+      "loss": 0.7437,
       "step": 25288
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.362849748654451e-07,
-      "loss": 0.7342,
+      "loss": 0.8092,
       "step": 25289
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.359821547798825e-07,
-      "loss": 0.919,
+      "loss": 0.8681,
       "step": 25290
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.356794687688536e-07,
-      "loss": 0.6986,
+      "loss": 0.8585,
       "step": 25291
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.353769168365606e-07,
-      "loss": 0.7203,
+      "loss": 0.7913,
       "step": 25292
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3507449898720124e-07,
-      "loss": 0.8538,
+      "loss": 0.7813,
       "step": 25293
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3477221522496995e-07,
-      "loss": 0.8738,
+      "loss": 0.8968,
       "step": 25294
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3447006555406117e-07,
-      "loss": 0.6837,
+      "loss": 0.7603,
       "step": 25295
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3416804997866483e-07,
-      "loss": 0.7734,
+      "loss": 0.7639,
       "step": 25296
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.338661685029765e-07,
-      "loss": 0.6016,
+      "loss": 0.739,
       "step": 25297
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3356442113117727e-07,
-      "loss": 0.8369,
+      "loss": 0.8204,
       "step": 25298
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.332628078674582e-07,
-      "loss": 0.9123,
+      "loss": 0.8279,
       "step": 25299
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.329613287160027e-07,
-      "loss": 0.7163,
+      "loss": 0.8498,
       "step": 25300
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3265998368099295e-07,
-      "loss": 0.7515,
+      "loss": 0.8011,
       "step": 25301
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.323587727666078e-07,
-      "loss": 0.6551,
+      "loss": 0.7305,
       "step": 25302
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.320576959770316e-07,
-      "loss": 0.6863,
+      "loss": 0.6876,
       "step": 25303
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3175675331643454e-07,
-      "loss": 0.6566,
+      "loss": 0.7715,
       "step": 25304
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3145594478899646e-07,
-      "loss": 0.6608,
+      "loss": 0.7636,
       "step": 25305
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3115527039888963e-07,
-      "loss": 0.7613,
+      "loss": 0.8312,
       "step": 25306
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.308547301502851e-07,
-      "loss": 0.7061,
+      "loss": 0.7753,
       "step": 25307
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3055432404735075e-07,
-      "loss": 0.8679,
+      "loss": 0.8876,
       "step": 25308
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.3025405209425763e-07,
-      "loss": 0.679,
+      "loss": 0.7947,
       "step": 25309
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.299539142951702e-07,
-      "loss": 0.6784,
+      "loss": 0.7189,
       "step": 25310
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.296539106542518e-07,
-      "loss": 0.7345,
+      "loss": 0.8487,
       "step": 25311
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.293540411756657e-07,
-      "loss": 0.8189,
+      "loss": 0.8315,
       "step": 25312
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.290543058635709e-07,
-      "loss": 0.7169,
+      "loss": 0.804,
       "step": 25313
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2875470472212623e-07,
-      "loss": 0.777,
+      "loss": 0.7396,
       "step": 25314
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2845523775548727e-07,
-      "loss": 0.7573,
+      "loss": 0.7808,
       "step": 25315
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.281559049678118e-07,
-      "loss": 0.8492,
+      "loss": 0.8553,
       "step": 25316
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2785670636324874e-07,
-      "loss": 0.7284,
+      "loss": 0.7667,
       "step": 25317
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2755764194595144e-07,
-      "loss": 0.6225,
+      "loss": 0.7214,
       "step": 25318
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2725871172006876e-07,
-      "loss": 0.6945,
+      "loss": 0.7911,
       "step": 25319
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2695991568974737e-07,
-      "loss": 0.7483,
+      "loss": 0.8036,
       "step": 25320
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2666125385913295e-07,
-      "loss": 0.7688,
+      "loss": 0.8051,
       "step": 25321
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2636272623236984e-07,
-      "loss": 0.7574,
+      "loss": 0.7418,
       "step": 25322
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2606433281359705e-07,
-      "loss": 0.6913,
+      "loss": 0.7567,
       "step": 25323
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.257660736069579e-07,
-      "loss": 0.7755,
+      "loss": 0.8074,
       "step": 25324
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2546794861658793e-07,
-      "loss": 0.8639,
+      "loss": 0.8407,
       "step": 25325
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2516995784662277e-07,
-      "loss": 0.7073,
+      "loss": 0.7413,
       "step": 25326
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2487210130119795e-07,
-      "loss": 0.7726,
+      "loss": 0.821,
       "step": 25327
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.245743789844469e-07,
-      "loss": 0.7046,
+      "loss": 0.6921,
       "step": 25328
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2427679090049734e-07,
-      "loss": 0.6772,
+      "loss": 0.8109,
       "step": 25329
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.239793370534805e-07,
-      "loss": 0.7723,
+      "loss": 0.8298,
       "step": 25330
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.236820174475197e-07,
-      "loss": 0.7603,
+      "loss": 0.8274,
       "step": 25331
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.23384832086745e-07,
-      "loss": 0.7532,
+      "loss": 0.7919,
       "step": 25332
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.230877809752753e-07,
-      "loss": 0.7687,
+      "loss": 0.8023,
       "step": 25333
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.227908641172317e-07,
-      "loss": 0.6551,
+      "loss": 0.823,
       "step": 25334
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2249408151673653e-07,
-      "loss": 0.7439,
+      "loss": 0.8474,
       "step": 25335
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.221974331779032e-07,
-      "loss": 0.7693,
+      "loss": 0.7649,
       "step": 25336
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2190091910485054e-07,
-      "loss": 0.6869,
+      "loss": 0.7133,
       "step": 25337
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.21604539301692e-07,
-      "loss": 0.6188,
+      "loss": 0.6787,
       "step": 25338
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.213082937725376e-07,
-      "loss": 0.7032,
+      "loss": 0.7993,
       "step": 25339
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2101218252149845e-07,
-      "loss": 0.768,
+      "loss": 0.7757,
       "step": 25340
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2071620555268246e-07,
-      "loss": 0.8674,
+      "loss": 0.8412,
       "step": 25341
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.204203628701963e-07,
-      "loss": 0.7168,
+      "loss": 0.7518,
       "step": 25342
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.2012465447814447e-07,
-      "loss": 0.7273,
+      "loss": 0.8098,
       "step": 25343
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.198290803806292e-07,
-      "loss": 0.7562,
+      "loss": 0.8333,
       "step": 25344
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.195336405817506e-07,
-      "loss": 0.7508,
+      "loss": 0.7282,
       "step": 25345
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1923833508560765e-07,
-      "loss": 0.6899,
+      "loss": 0.7292,
       "step": 25346
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.189431638962981e-07,
-      "loss": 0.7355,
+      "loss": 0.7863,
       "step": 25347
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.186481270179165e-07,
-      "loss": 0.7559,
+      "loss": 0.7946,
       "step": 25348
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.183532244545562e-07,
-      "loss": 0.807,
+      "loss": 0.8434,
       "step": 25349
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.180584562103073e-07,
-      "loss": 0.8368,
+      "loss": 0.8712,
       "step": 25350
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1776382228926207e-07,
-      "loss": 0.7175,
+      "loss": 0.7439,
       "step": 25351
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1746932269550614e-07,
-      "loss": 0.7308,
+      "loss": 0.7057,
       "step": 25352
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.171749574331251e-07,
-      "loss": 0.7417,
+      "loss": 0.8288,
       "step": 25353
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.168807265062035e-07,
-      "loss": 0.703,
+      "loss": 0.7711,
       "step": 25354
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.165866299188236e-07,
-      "loss": 0.7908,
+      "loss": 0.881,
       "step": 25355
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.162926676750655e-07,
-      "loss": 0.7465,
+      "loss": 0.8208,
       "step": 25356
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.159988397790059e-07,
-      "loss": 0.7214,
+      "loss": 0.8124,
       "step": 25357
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.157051462347238e-07,
-      "loss": 0.8059,
+      "loss": 0.8965,
       "step": 25358
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.154115870462915e-07,
-      "loss": 0.7902,
+      "loss": 0.8477,
       "step": 25359
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.151181622177835e-07,
-      "loss": 0.6463,
+      "loss": 0.7154,
       "step": 25360
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.148248717532709e-07,
-      "loss": 0.751,
+      "loss": 0.7025,
       "step": 25361
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.145317156568206e-07,
-      "loss": 0.6066,
+      "loss": 0.705,
       "step": 25362
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.142386939325004e-07,
-      "loss": 0.6691,
+      "loss": 0.7295,
       "step": 25363
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.139458065843781e-07,
-      "loss": 0.7046,
+      "loss": 0.8023,
       "step": 25364
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.136530536165128e-07,
-      "loss": 0.7428,
+      "loss": 0.8013,
       "step": 25365
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.133604350329689e-07,
-      "loss": 0.6597,
+      "loss": 0.8107,
       "step": 25366
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.130679508378054e-07,
-      "loss": 0.7234,
+      "loss": 0.7386,
       "step": 25367
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.127756010350813e-07,
-      "loss": 0.7682,
+      "loss": 0.7652,
       "step": 25368
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1248338562884896e-07,
-      "loss": 0.6593,
+      "loss": 0.721,
       "step": 25369
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.121913046231673e-07,
-      "loss": 0.8096,
+      "loss": 0.8252,
       "step": 25370
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1189935802208525e-07,
-      "loss": 0.7365,
+      "loss": 0.781,
       "step": 25371
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.11607545829653e-07,
-      "loss": 0.7673,
+      "loss": 0.8437,
       "step": 25372
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1131586804992174e-07,
-      "loss": 0.6778,
+      "loss": 0.7477,
       "step": 25373
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.110243246869349e-07,
-      "loss": 0.6229,
+      "loss": 0.7258,
       "step": 25374
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.107329157447403e-07,
-      "loss": 0.6267,
+      "loss": 0.7499,
       "step": 25375
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.104416412273781e-07,
-      "loss": 0.7911,
+      "loss": 0.8823,
       "step": 25376
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.1015050113889056e-07,
-      "loss": 0.6185,
+      "loss": 0.6964,
       "step": 25377
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0985949548331564e-07,
-      "loss": 0.7059,
+      "loss": 0.827,
       "step": 25378
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0956862426469337e-07,
-      "loss": 0.7215,
+      "loss": 0.8125,
       "step": 25379
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.092778874870572e-07,
-      "loss": 0.7228,
+      "loss": 0.8202,
       "step": 25380
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0898728515444064e-07,
-      "loss": 0.5733,
+      "loss": 0.7525,
       "step": 25381
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0869681727087484e-07,
-      "loss": 0.8866,
+      "loss": 0.8497,
       "step": 25382
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.084064838403922e-07,
-      "loss": 0.7572,
+      "loss": 0.8103,
       "step": 25383
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0811628486701827e-07,
-      "loss": 0.6328,
+      "loss": 0.7661,
       "step": 25384
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0782622035477993e-07,
-      "loss": 0.6938,
+      "loss": 0.8452,
       "step": 25385
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.075362903077017e-07,
-      "loss": 0.7531,
+      "loss": 0.8572,
       "step": 25386
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0724649472980485e-07,
-      "loss": 0.6725,
+      "loss": 0.7455,
       "step": 25387
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.069568336251105e-07,
-      "loss": 0.7282,
+      "loss": 0.8608,
       "step": 25388
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0666730699764e-07,
-      "loss": 0.7587,
+      "loss": 0.8476,
       "step": 25389
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0637791485140567e-07,
-      "loss": 0.7848,
+      "loss": 0.8074,
       "step": 25390
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0608865719042313e-07,
-      "loss": 0.7349,
+      "loss": 0.819,
       "step": 25391
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.057995340187081e-07,
-      "loss": 0.8302,
+      "loss": 0.8067,
       "step": 25392
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.055105453402707e-07,
-      "loss": 0.8155,
+      "loss": 0.8808,
       "step": 25393
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.052216911591188e-07,
-      "loss": 0.639,
+      "loss": 0.7953,
       "step": 25394
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.049329714792604e-07,
-      "loss": 0.8077,
+      "loss": 0.7939,
       "step": 25395
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.046443863047033e-07,
-      "loss": 0.8057,
+      "loss": 0.8653,
       "step": 25396
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0435593563944765e-07,
-      "loss": 0.7989,
+      "loss": 0.8337,
       "step": 25397
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.040676194874981e-07,
-      "loss": 0.7334,
+      "loss": 0.7859,
       "step": 25398
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0377943785285355e-07,
-      "loss": 0.7218,
+      "loss": 0.7576,
       "step": 25399
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0349139073951205e-07,
-      "loss": 0.7425,
+      "loss": 0.7764,
       "step": 25400
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.032034781514681e-07,
-      "loss": 0.6867,
+      "loss": 0.8498,
       "step": 25401
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0291570009272187e-07,
-      "loss": 0.6568,
+      "loss": 0.7867,
       "step": 25402
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.026280565672579e-07,
-      "loss": 0.8655,
+      "loss": 0.9021,
       "step": 25403
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0234054757907303e-07,
-      "loss": 0.6953,
+      "loss": 0.7542,
       "step": 25404
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.020531731321541e-07,
-      "loss": 0.6479,
+      "loss": 0.7492,
       "step": 25405
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.017659332304879e-07,
-      "loss": 0.807,
+      "loss": 0.8374,
       "step": 25406
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.014788278780578e-07,
-      "loss": 0.8291,
+      "loss": 0.8472,
       "step": 25407
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.011918570788497e-07,
-      "loss": 0.8679,
+      "loss": 0.8135,
       "step": 25408
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.009050208368447e-07,
-      "loss": 0.6853,
+      "loss": 0.7603,
       "step": 25409
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.006183191560197e-07,
-      "loss": 0.8198,
+      "loss": 0.8798,
       "step": 25410
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.003317520403537e-07,
-      "loss": 0.7322,
+      "loss": 0.8481,
       "step": 25411
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.0004531949382355e-07,
-      "loss": 0.7656,
+      "loss": 0.7813,
       "step": 25412
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.997590215204027e-07,
-      "loss": 0.891,
+      "loss": 0.8921,
       "step": 25413
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.994728581240602e-07,
-      "loss": 0.7506,
+      "loss": 0.8435,
       "step": 25414
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9918682930877184e-07,
-      "loss": 0.7464,
+      "loss": 0.7717,
       "step": 25415
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9890093507849996e-07,
-      "loss": 0.7423,
+      "loss": 0.8006,
       "step": 25416
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.986151754372146e-07,
-      "loss": 0.6977,
+      "loss": 0.7803,
       "step": 25417
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9832955038887945e-07,
-      "loss": 0.6973,
+      "loss": 0.7979,
       "step": 25418
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.980440599374568e-07,
-      "loss": 0.7798,
+      "loss": 0.8211,
       "step": 25419
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.977587040869057e-07,
-      "loss": 0.652,
+      "loss": 0.6674,
       "step": 25420
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.974734828411907e-07,
-      "loss": 0.6533,
+      "loss": 0.7806,
       "step": 25421
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9718839620426097e-07,
-      "loss": 0.6824,
+      "loss": 0.7533,
       "step": 25422
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.969034441800789e-07,
-      "loss": 0.7364,
+      "loss": 0.8131,
       "step": 25423
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.966186267725935e-07,
-      "loss": 0.7765,
+      "loss": 0.8567,
       "step": 25424
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.963339439857571e-07,
-      "loss": 0.7149,
+      "loss": 0.8059,
       "step": 25425
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9604939582351887e-07,
-      "loss": 0.8356,
+      "loss": 0.8705,
       "step": 25426
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9576498228982785e-07,
-      "loss": 0.8483,
+      "loss": 0.9329,
       "step": 25427
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.954807033886309e-07,
-      "loss": 0.8043,
+      "loss": 0.8237,
       "step": 25428
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9519655912386704e-07,
-      "loss": 0.5923,
+      "loss": 0.7159,
       "step": 25429
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9491254949948423e-07,
-      "loss": 0.7454,
+      "loss": 0.77,
       "step": 25430
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9462867451941934e-07,
-      "loss": 0.8895,
+      "loss": 0.8864,
       "step": 25431
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9434493418761036e-07,
-      "loss": 0.8231,
+      "loss": 0.8045,
       "step": 25432
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.9406132850799517e-07,
-      "loss": 0.6902,
+      "loss": 0.8009,
       "step": 25433
     },
     {
       "epoch": 0.92,
       "learning_rate": 2.937778574845107e-07,
-      "loss": 0.5302,
+      "loss": 0.6473,
       "step": 25434
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.934945211210838e-07,
-      "loss": 0.8104,
+      "loss": 0.8325,
       "step": 25435
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.932113194216502e-07,
-      "loss": 0.8026,
+      "loss": 0.8373,
       "step": 25436
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.929282523901378e-07,
-      "loss": 0.7495,
+      "loss": 0.735,
       "step": 25437
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.926453200304724e-07,
-      "loss": 0.7368,
+      "loss": 0.8543,
       "step": 25438
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.923625223465798e-07,
-      "loss": 0.7509,
+      "loss": 0.7972,
       "step": 25439
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.920798593423846e-07,
-      "loss": 0.7769,
+      "loss": 0.8389,
       "step": 25440
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.917973310218081e-07,
-      "loss": 0.7356,
+      "loss": 0.7789,
       "step": 25441
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.9151493738876934e-07,
-      "loss": 0.7566,
+      "loss": 0.7998,
       "step": 25442
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.912326784471864e-07,
-      "loss": 0.6621,
+      "loss": 0.791,
       "step": 25443
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.9095055420097497e-07,
-      "loss": 0.7482,
+      "loss": 0.812,
       "step": 25444
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.906685646540486e-07,
-      "loss": 0.7715,
+      "loss": 0.8474,
       "step": 25445
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.903867098103208e-07,
-      "loss": 0.696,
+      "loss": 0.7972,
       "step": 25446
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.901049896737018e-07,
-      "loss": 0.7907,
+      "loss": 0.828,
       "step": 25447
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8982340424809854e-07,
-      "loss": 0.7141,
+      "loss": 0.7845,
       "step": 25448
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.895419535374189e-07,
-      "loss": 0.7701,
+      "loss": 0.8546,
       "step": 25449
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8926063754556753e-07,
-      "loss": 0.6552,
+      "loss": 0.7985,
       "step": 25450
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8897945627644696e-07,
-      "loss": 0.8041,
+      "loss": 0.8422,
       "step": 25451
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8869840973395737e-07,
-      "loss": 0.7375,
+      "loss": 0.7515,
       "step": 25452
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.884174979219989e-07,
-      "loss": 0.5674,
+      "loss": 0.7342,
       "step": 25453
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8813672084446966e-07,
-      "loss": 0.6692,
+      "loss": 0.7519,
       "step": 25454
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.878560785052642e-07,
-      "loss": 0.7156,
+      "loss": 0.6699,
       "step": 25455
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8757557090827614e-07,
-      "loss": 0.7717,
+      "loss": 0.8473,
       "step": 25456
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.872951980573968e-07,
-      "loss": 0.6896,
+      "loss": 0.8463,
       "step": 25457
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.870149599565142e-07,
-      "loss": 0.5823,
+      "loss": 0.8168,
       "step": 25458
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8673485660951963e-07,
-      "loss": 0.7243,
+      "loss": 0.7726,
       "step": 25459
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8645488802029774e-07,
-      "loss": 0.7203,
+      "loss": 0.7835,
       "step": 25460
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8617505419273217e-07,
-      "loss": 0.7509,
+      "loss": 0.8295,
       "step": 25461
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.858953551307053e-07,
-      "loss": 0.8298,
+      "loss": 0.8203,
       "step": 25462
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.856157908380997e-07,
-      "loss": 0.7179,
+      "loss": 0.7573,
       "step": 25463
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.853363613187898e-07,
-      "loss": 0.8623,
+      "loss": 0.9016,
       "step": 25464
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.85057066576655e-07,
-      "loss": 0.6333,
+      "loss": 0.7565,
       "step": 25465
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8477790661557205e-07,
-      "loss": 0.6593,
+      "loss": 0.7849,
       "step": 25466
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8449888143940893e-07,
-      "loss": 0.6778,
+      "loss": 0.7959,
       "step": 25467
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8421999105204044e-07,
-      "loss": 0.6929,
+      "loss": 0.678,
       "step": 25468
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8394123545733455e-07,
-      "loss": 0.7467,
+      "loss": 0.7135,
       "step": 25469
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.836626146591581e-07,
-      "loss": 0.7088,
+      "loss": 0.7374,
       "step": 25470
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8338412866137696e-07,
-      "loss": 0.7534,
+      "loss": 0.7287,
       "step": 25471
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.831057774678558e-07,
-      "loss": 0.6408,
+      "loss": 0.7732,
       "step": 25472
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.82827561082456e-07,
-      "loss": 0.6831,
+      "loss": 0.7292,
       "step": 25473
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8254947950903553e-07,
-      "loss": 0.8677,
+      "loss": 0.8198,
       "step": 25474
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.822715327514536e-07,
-      "loss": 0.6643,
+      "loss": 0.7944,
       "step": 25475
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8199372081356926e-07,
-      "loss": 0.8583,
+      "loss": 0.8893,
       "step": 25476
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8171604369923057e-07,
-      "loss": 0.7221,
+      "loss": 0.7407,
       "step": 25477
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8143850141229554e-07,
-      "loss": 0.8015,
+      "loss": 0.7717,
       "step": 25478
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.811610939566123e-07,
-      "loss": 0.7171,
+      "loss": 0.7601,
       "step": 25479
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8088382133602877e-07,
-      "loss": 0.805,
+      "loss": 0.8217,
       "step": 25480
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.806066835543919e-07,
-      "loss": 0.7307,
+      "loss": 0.8147,
       "step": 25481
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.803296806155498e-07,
-      "loss": 0.828,
+      "loss": 0.8502,
       "step": 25482
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.8005281252334037e-07,
-      "loss": 0.5549,
+      "loss": 0.7426,
       "step": 25483
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7977607928160955e-07,
-      "loss": 0.8482,
+      "loss": 0.8077,
       "step": 25484
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.794994808941931e-07,
-      "loss": 0.6926,
+      "loss": 0.7835,
       "step": 25485
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.792230173649302e-07,
-      "loss": 0.6927,
+      "loss": 0.7922,
       "step": 25486
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.789466886976566e-07,
-      "loss": 0.6628,
+      "loss": 0.7631,
       "step": 25487
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.786704948962049e-07,
-      "loss": 0.7039,
+      "loss": 0.755,
       "step": 25488
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.783944359644064e-07,
-      "loss": 0.6379,
+      "loss": 0.7225,
       "step": 25489
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7811851190609253e-07,
-      "loss": 0.6952,
+      "loss": 0.6925,
       "step": 25490
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7784272272509015e-07,
-      "loss": 0.8302,
+      "loss": 0.8636,
       "step": 25491
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.775670684252274e-07,
-      "loss": 0.5796,
+      "loss": 0.7009,
       "step": 25492
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7729154901032673e-07,
-      "loss": 0.6072,
+      "loss": 0.6718,
       "step": 25493
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7701616448420954e-07,
-      "loss": 0.6311,
+      "loss": 0.7821,
       "step": 25494
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7674091485070056e-07,
-      "loss": 0.7861,
+      "loss": 0.807,
       "step": 25495
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7646580011361443e-07,
-      "loss": 0.7268,
+      "loss": 0.8053,
       "step": 25496
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7619082027676934e-07,
-      "loss": 0.8179,
+      "loss": 0.8089,
       "step": 25497
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.759159753439811e-07,
-      "loss": 0.6237,
+      "loss": 0.7643,
       "step": 25498
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.75641265319061e-07,
-      "loss": 0.5731,
+      "loss": 0.7242,
       "step": 25499
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.753666902058216e-07,
-      "loss": 0.6379,
+      "loss": 0.7641,
       "step": 25500
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7509225000807325e-07,
-      "loss": 0.7972,
+      "loss": 0.7908,
       "step": 25501
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7481794472962066e-07,
-      "loss": 0.828,
+      "loss": 0.8005,
       "step": 25502
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.745437743742707e-07,
-      "loss": 0.7535,
+      "loss": 0.8106,
       "step": 25503
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.742697389458282e-07,
-      "loss": 0.7458,
+      "loss": 0.7686,
       "step": 25504
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7399583844809343e-07,
-      "loss": 0.8123,
+      "loss": 0.7952,
       "step": 25505
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7372207288486773e-07,
-      "loss": 0.6655,
+      "loss": 0.7206,
       "step": 25506
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7344844225994707e-07,
-      "loss": 0.599,
+      "loss": 0.6429,
       "step": 25507
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.731749465771316e-07,
-      "loss": 0.6365,
+      "loss": 0.7477,
       "step": 25508
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7290158584021067e-07,
-      "loss": 0.8016,
+      "loss": 0.7959,
       "step": 25509
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7262836005298e-07,
-      "loss": 0.8438,
+      "loss": 0.8972,
       "step": 25510
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7235526921922994e-07,
-      "loss": 0.8113,
+      "loss": 0.8921,
       "step": 25511
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.720823133427486e-07,
-      "loss": 0.7466,
+      "loss": 0.8544,
       "step": 25512
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.718094924273218e-07,
-      "loss": 0.7149,
+      "loss": 0.7596,
       "step": 25513
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7153680647673875e-07,
-      "loss": 0.7035,
+      "loss": 0.7376,
       "step": 25514
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7126425549477755e-07,
-      "loss": 0.684,
+      "loss": 0.7245,
       "step": 25515
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.709918394852218e-07,
-      "loss": 0.6144,
+      "loss": 0.775,
       "step": 25516
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.707195584518507e-07,
-      "loss": 0.8108,
+      "loss": 0.8497,
       "step": 25517
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.704474123984413e-07,
-      "loss": 0.7,
+      "loss": 0.7697,
       "step": 25518
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.7017540132876943e-07,
-      "loss": 0.8346,
+      "loss": 0.8484,
       "step": 25519
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.699035252466109e-07,
-      "loss": 0.6724,
+      "loss": 0.7102,
       "step": 25520
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6963178415573386e-07,
-      "loss": 0.6775,
+      "loss": 0.8016,
       "step": 25521
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.693601780599109e-07,
-      "loss": 0.8089,
+      "loss": 0.8065,
       "step": 25522
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6908870696291e-07,
-      "loss": 0.6826,
+      "loss": 0.7248,
       "step": 25523
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.68817370868496e-07,
-      "loss": 0.6983,
+      "loss": 0.7334,
       "step": 25524
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.685461697804348e-07,
-      "loss": 0.7469,
+      "loss": 0.7824,
       "step": 25525
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6827510370248777e-07,
-      "loss": 0.7462,
+      "loss": 0.7658,
       "step": 25526
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.680041726384175e-07,
-      "loss": 0.7423,
+      "loss": 0.7837,
       "step": 25527
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6773337659197983e-07,
-      "loss": 0.6865,
+      "loss": 0.7073,
       "step": 25528
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.67462715566934e-07,
-      "loss": 0.7419,
+      "loss": 0.7532,
       "step": 25529
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6719218956703376e-07,
-      "loss": 0.6845,
+      "loss": 0.7742,
       "step": 25530
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.669217985960337e-07,
-      "loss": 0.7041,
+      "loss": 0.8192,
       "step": 25531
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6665154265768324e-07,
-      "loss": 0.6899,
+      "loss": 0.7957,
       "step": 25532
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6638142175573476e-07,
-      "loss": 0.8639,
+      "loss": 0.8546,
       "step": 25533
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6611143589393206e-07,
-      "loss": 0.679,
+      "loss": 0.7674,
       "step": 25534
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.658415850760232e-07,
-      "loss": 0.715,
+      "loss": 0.7268,
       "step": 25535
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6557186930575186e-07,
-      "loss": 0.754,
+      "loss": 0.77,
       "step": 25536
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.653022885868584e-07,
-      "loss": 0.6865,
+      "loss": 0.7649,
       "step": 25537
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.650328429230842e-07,
-      "loss": 0.8156,
+      "loss": 0.7717,
       "step": 25538
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6476353231816744e-07,
-      "loss": 0.6337,
+      "loss": 0.7516,
       "step": 25539
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.64494356775844e-07,
-      "loss": 0.756,
+      "loss": 0.8428,
       "step": 25540
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.642253162998487e-07,
-      "loss": 0.6807,
+      "loss": 0.73,
       "step": 25541
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.639564108939141e-07,
-      "loss": 0.7383,
+      "loss": 0.7868,
       "step": 25542
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6368764056177053e-07,
-      "loss": 0.788,
+      "loss": 0.795,
       "step": 25543
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.634190053071473e-07,
-      "loss": 0.8575,
+      "loss": 0.8837,
       "step": 25544
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6315050513377017e-07,
-      "loss": 0.6368,
+      "loss": 0.7376,
       "step": 25545
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6288214004536625e-07,
-      "loss": 0.7784,
+      "loss": 0.869,
       "step": 25546
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6261391004565597e-07,
-      "loss": 0.6878,
+      "loss": 0.709,
       "step": 25547
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6234581513836286e-07,
-      "loss": 0.6999,
+      "loss": 0.758,
       "step": 25548
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.620778553272052e-07,
-      "loss": 0.8008,
+      "loss": 0.808,
       "step": 25549
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6181003061590216e-07,
-      "loss": 0.838,
+      "loss": 0.8652,
       "step": 25550
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.615423410081663e-07,
-      "loss": 0.8012,
+      "loss": 0.837,
       "step": 25551
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.612747865077148e-07,
-      "loss": 0.7605,
+      "loss": 0.8904,
       "step": 25552
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6100736711825783e-07,
-      "loss": 0.734,
+      "loss": 0.7647,
       "step": 25553
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.607400828435058e-07,
-      "loss": 0.787,
+      "loss": 0.7858,
       "step": 25554
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6047293368716697e-07,
-      "loss": 0.8054,
+      "loss": 0.7983,
       "step": 25555
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.6020591965294826e-07,
-      "loss": 0.6121,
+      "loss": 0.7581,
       "step": 25556
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5993904074455233e-07,
-      "loss": 0.8521,
+      "loss": 0.8842,
       "step": 25557
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5967229696568285e-07,
-      "loss": 0.7971,
+      "loss": 0.8386,
       "step": 25558
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5940568832004133e-07,
-      "loss": 0.6903,
+      "loss": 0.7288,
       "step": 25559
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.591392148113259e-07,
-      "loss": 0.7632,
+      "loss": 0.7783,
       "step": 25560
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5887287644323243e-07,
-      "loss": 0.6493,
+      "loss": 0.7205,
       "step": 25561
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.586066732194581e-07,
-      "loss": 0.8015,
+      "loss": 0.7573,
       "step": 25562
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5834060514369543e-07,
-      "loss": 0.6876,
+      "loss": 0.7557,
       "step": 25563
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5807467221963256e-07,
-      "loss": 0.7648,
+      "loss": 0.7706,
       "step": 25564
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5780887445096436e-07,
-      "loss": 0.7012,
+      "loss": 0.8587,
       "step": 25565
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.575432118413745e-07,
-      "loss": 0.8502,
+      "loss": 0.8871,
       "step": 25566
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.572776843945501e-07,
-      "loss": 0.7203,
+      "loss": 0.7856,
       "step": 25567
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.570122921141738e-07,
-      "loss": 0.7689,
+      "loss": 0.7957,
       "step": 25568
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.567470350039303e-07,
-      "loss": 0.8231,
+      "loss": 0.8263,
       "step": 25569
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.564819130674956e-07,
-      "loss": 0.8181,
+      "loss": 0.7958,
       "step": 25570
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.562169263085512e-07,
-      "loss": 0.7133,
+      "loss": 0.767,
       "step": 25571
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.55952074730772e-07,
-      "loss": 0.7053,
+      "loss": 0.76,
       "step": 25572
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.556873583378339e-07,
-      "loss": 0.7714,
+      "loss": 0.8372,
       "step": 25573
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.554227771334061e-07,
-      "loss": 0.8548,
+      "loss": 0.8698,
       "step": 25574
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.551583311211647e-07,
-      "loss": 0.6928,
+      "loss": 0.7714,
       "step": 25575
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.548940203047723e-07,
-      "loss": 0.7176,
+      "loss": 0.7529,
       "step": 25576
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.546298446879003e-07,
-      "loss": 0.866,
+      "loss": 0.8191,
       "step": 25577
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.543658042742125e-07,
-      "loss": 0.8017,
+      "loss": 0.807,
       "step": 25578
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.541018990673716e-07,
-      "loss": 0.7036,
+      "loss": 0.7264,
       "step": 25579
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.53838129071039e-07,
-      "loss": 0.8731,
+      "loss": 0.8632,
       "step": 25580
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5357449428887514e-07,
-      "loss": 0.6158,
+      "loss": 0.7337,
       "step": 25581
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5331099472453715e-07,
-      "loss": 0.6619,
+      "loss": 0.7602,
       "step": 25582
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.530476303816787e-07,
-      "loss": 0.8096,
+      "loss": 0.7827,
       "step": 25583
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5278440126395574e-07,
-      "loss": 0.6689,
+      "loss": 0.7681,
       "step": 25584
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.525213073750199e-07,
-      "loss": 0.7329,
+      "loss": 0.755,
       "step": 25585
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5225834871852153e-07,
-      "loss": 0.7642,
+      "loss": 0.8287,
       "step": 25586
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.519955252981088e-07,
-      "loss": 0.7962,
+      "loss": 0.8092,
       "step": 25587
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5173283711742657e-07,
-      "loss": 0.6093,
+      "loss": 0.7262,
       "step": 25588
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5147028418011974e-07,
-      "loss": 0.7705,
+      "loss": 0.8284,
       "step": 25589
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.512078664898321e-07,
-      "loss": 0.8261,
+      "loss": 0.8637,
       "step": 25590
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.50945584050204e-07,
-      "loss": 0.8311,
+      "loss": 0.8729,
       "step": 25591
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.506834368648736e-07,
-      "loss": 0.8231,
+      "loss": 0.8036,
       "step": 25592
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.5042142493747703e-07,
-      "loss": 0.8786,
+      "loss": 0.8779,
       "step": 25593
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.501595482716523e-07,
-      "loss": 0.7319,
+      "loss": 0.7597,
       "step": 25594
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.498978068710278e-07,
-      "loss": 0.7308,
+      "loss": 0.7864,
       "step": 25595
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4963620073923947e-07,
-      "loss": 0.8587,
+      "loss": 0.857,
       "step": 25596
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.493747298799154e-07,
-      "loss": 0.7144,
+      "loss": 0.7791,
       "step": 25597
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4911339429668167e-07,
-      "loss": 0.7017,
+      "loss": 0.7737,
       "step": 25598
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.488521939931643e-07,
-      "loss": 0.6033,
+      "loss": 0.668,
       "step": 25599
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4859112897299034e-07,
-      "loss": 0.8113,
+      "loss": 0.8034,
       "step": 25600
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4833019923977685e-07,
-      "loss": 0.7363,
+      "loss": 0.7882,
       "step": 25601
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.480694047971455e-07,
-      "loss": 0.8791,
+      "loss": 0.9057,
       "step": 25602
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.478087456487166e-07,
-      "loss": 0.7151,
+      "loss": 0.7669,
       "step": 25603
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.47548221798104e-07,
-      "loss": 0.8284,
+      "loss": 0.851,
       "step": 25604
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4728783324892256e-07,
-      "loss": 0.7954,
+      "loss": 0.812,
       "step": 25605
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.47027580004785e-07,
-      "loss": 0.7033,
+      "loss": 0.811,
       "step": 25606
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.467674620693039e-07,
-      "loss": 0.7376,
+      "loss": 0.8138,
       "step": 25607
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.465074794460842e-07,
-      "loss": 0.7241,
+      "loss": 0.753,
       "step": 25608
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.462476321387364e-07,
-      "loss": 0.7813,
+      "loss": 0.8556,
       "step": 25609
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4598792015086305e-07,
-      "loss": 0.7602,
+      "loss": 0.7702,
       "step": 25610
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.45728343486068e-07,
-      "loss": 0.7377,
+      "loss": 0.753,
       "step": 25611
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.454689021479528e-07,
-      "loss": 0.7604,
+      "loss": 0.7681,
       "step": 25612
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4520959614011796e-07,
-      "loss": 0.8469,
+      "loss": 0.8819,
       "step": 25613
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.449504254661572e-07,
-      "loss": 0.6757,
+      "loss": 0.7178,
       "step": 25614
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.44691390129671e-07,
-      "loss": 0.8955,
+      "loss": 0.9,
       "step": 25615
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4443249013424986e-07,
-      "loss": 0.8361,
+      "loss": 0.8666,
       "step": 25616
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.441737254834864e-07,
-      "loss": 0.6731,
+      "loss": 0.7938,
       "step": 25617
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.439150961809711e-07,
-      "loss": 0.7783,
+      "loss": 0.7894,
       "step": 25618
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4365660223029106e-07,
-      "loss": 0.716,
+      "loss": 0.7937,
       "step": 25619
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4339824363503574e-07,
-      "loss": 0.7077,
+      "loss": 0.7988,
       "step": 25620
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.431400203987844e-07,
-      "loss": 0.7881,
+      "loss": 0.8016,
       "step": 25621
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.428819325251219e-07,
-      "loss": 0.8425,
+      "loss": 0.8773,
       "step": 25622
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.426239800176311e-07,
-      "loss": 0.7534,
+      "loss": 0.7613,
       "step": 25623
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.423661628798868e-07,
-      "loss": 0.8134,
+      "loss": 0.8879,
       "step": 25624
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4210848111546836e-07,
-      "loss": 0.7241,
+      "loss": 0.7821,
       "step": 25625
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.418509347279507e-07,
-      "loss": 0.6467,
+      "loss": 0.7987,
       "step": 25626
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.415935237209055e-07,
-      "loss": 0.7594,
+      "loss": 0.7619,
       "step": 25627
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.413362480979053e-07,
-      "loss": 0.8288,
+      "loss": 0.7708,
       "step": 25628
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4107910786251854e-07,
-      "loss": 0.6299,
+      "loss": 0.7236,
       "step": 25629
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4082210301831334e-07,
-      "loss": 0.6262,
+      "loss": 0.7595,
       "step": 25630
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.4056523356885463e-07,
-      "loss": 0.8079,
+      "loss": 0.7862,
       "step": 25631
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.403084995177074e-07,
-      "loss": 0.5838,
+      "loss": 0.7533,
       "step": 25632
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.40051900868431e-07,
-      "loss": 0.8096,
+      "loss": 0.8964,
       "step": 25633
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.397954376245881e-07,
-      "loss": 0.8014,
+      "loss": 0.8349,
       "step": 25634
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.395391097897359e-07,
-      "loss": 0.7842,
+      "loss": 0.7803,
       "step": 25635
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3928291736742934e-07,
-      "loss": 0.6778,
+      "loss": 0.7237,
       "step": 25636
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.390268603612234e-07,
-      "loss": 0.7927,
+      "loss": 0.8295,
       "step": 25637
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3877093877467174e-07,
-      "loss": 0.7418,
+      "loss": 0.7907,
       "step": 25638
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3851515261132494e-07,
-      "loss": 0.694,
+      "loss": 0.754,
       "step": 25639
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3825950187472802e-07,
-      "loss": 0.7399,
+      "loss": 0.797,
       "step": 25640
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3800398656843248e-07,
-      "loss": 0.8014,
+      "loss": 0.8599,
       "step": 25641
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3774860669598e-07,
-      "loss": 0.6689,
+      "loss": 0.7392,
       "step": 25642
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3749336226091546e-07,
-      "loss": 0.7331,
+      "loss": 0.7125,
       "step": 25643
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.372382532667783e-07,
-      "loss": 0.7559,
+      "loss": 0.7692,
       "step": 25644
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3698327971711012e-07,
-      "loss": 0.846,
+      "loss": 0.8575,
       "step": 25645
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3672844161544584e-07,
-      "loss": 0.7023,
+      "loss": 0.6876,
       "step": 25646
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3647373896532378e-07,
-      "loss": 0.7821,
+      "loss": 0.8529,
       "step": 25647
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3621917177027442e-07,
-      "loss": 0.6995,
+      "loss": 0.7507,
       "step": 25648
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3596474003383163e-07,
-      "loss": 0.7361,
+      "loss": 0.8281,
       "step": 25649
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3571044375952367e-07,
-      "loss": 0.5809,
+      "loss": 0.5899,
       "step": 25650
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3545628295088107e-07,
-      "loss": 0.7739,
+      "loss": 0.79,
       "step": 25651
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.352022576114288e-07,
-      "loss": 0.7689,
+      "loss": 0.8539,
       "step": 25652
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3494836774468953e-07,
-      "loss": 0.6402,
+      "loss": 0.7127,
       "step": 25653
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.346946133541872e-07,
-      "loss": 0.7867,
+      "loss": 0.8638,
       "step": 25654
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3444099444344227e-07,
-      "loss": 0.6855,
+      "loss": 0.835,
       "step": 25655
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3418751101597192e-07,
-      "loss": 0.7843,
+      "loss": 0.8129,
       "step": 25656
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3393416307529558e-07,
-      "loss": 0.777,
+      "loss": 0.8272,
       "step": 25657
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3368095062492713e-07,
-      "loss": 0.63,
+      "loss": 0.7116,
       "step": 25658
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3342787366837594e-07,
-      "loss": 0.83,
+      "loss": 0.8404,
       "step": 25659
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.331749322091581e-07,
-      "loss": 0.7181,
+      "loss": 0.8455,
       "step": 25660
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.329221262507808e-07,
-      "loss": 0.8595,
+      "loss": 0.8307,
       "step": 25661
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3266945579675126e-07,
-      "loss": 0.748,
+      "loss": 0.7438,
       "step": 25662
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3241692085057332e-07,
-      "loss": 0.7407,
+      "loss": 0.7856,
       "step": 25663
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3216452141575418e-07,
-      "loss": 0.783,
+      "loss": 0.7893,
       "step": 25664
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3191225749579215e-07,
-      "loss": 0.6706,
+      "loss": 0.7858,
       "step": 25665
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3166012909419e-07,
-      "loss": 0.737,
+      "loss": 0.809,
       "step": 25666
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.314081362144427e-07,
-      "loss": 0.7853,
+      "loss": 0.8154,
       "step": 25667
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3115627886004855e-07,
-      "loss": 0.6472,
+      "loss": 0.7334,
       "step": 25668
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.309045570345003e-07,
-      "loss": 0.7184,
+      "loss": 0.79,
       "step": 25669
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3065297074129077e-07,
-      "loss": 0.9795,
+      "loss": 0.9068,
       "step": 25670
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3040151998391048e-07,
-      "loss": 0.7399,
+      "loss": 0.8108,
       "step": 25671
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.3015020476584881e-07,
-      "loss": 0.7959,
+      "loss": 0.8196,
       "step": 25672
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.298990250905897e-07,
-      "loss": 0.8327,
+      "loss": 0.8595,
       "step": 25673
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2964798096162144e-07,
-      "loss": 0.6283,
+      "loss": 0.6276,
       "step": 25674
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2939707238242349e-07,
-      "loss": 0.7986,
+      "loss": 0.8845,
       "step": 25675
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.291462993564797e-07,
-      "loss": 0.7474,
+      "loss": 0.798,
       "step": 25676
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2889566188726842e-07,
-      "loss": 0.8633,
+      "loss": 0.853,
       "step": 25677
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2864515997826465e-07,
-      "loss": 0.8164,
+      "loss": 0.8456,
       "step": 25678
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.283947936329467e-07,
-      "loss": 0.713,
+      "loss": 0.7069,
       "step": 25679
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2814456285478736e-07,
-      "loss": 0.7686,
+      "loss": 0.8714,
       "step": 25680
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.278944676472572e-07,
-      "loss": 0.7156,
+      "loss": 0.8091,
       "step": 25681
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2764450801382566e-07,
-      "loss": 0.6817,
+      "loss": 0.7372,
       "step": 25682
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2739468395796215e-07,
-      "loss": 0.6717,
+      "loss": 0.8315,
       "step": 25683
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2714499548313173e-07,
-      "loss": 0.6399,
+      "loss": 0.7669,
       "step": 25684
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2689544259279938e-07,
-      "loss": 0.8017,
+      "loss": 0.8285,
       "step": 25685
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2664602529042567e-07,
-      "loss": 0.6398,
+      "loss": 0.7102,
       "step": 25686
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2639674357947227e-07,
-      "loss": 0.7587,
+      "loss": 0.866,
       "step": 25687
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2614759746339644e-07,
-      "loss": 0.7103,
+      "loss": 0.7375,
       "step": 25688
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2589858694565536e-07,
-      "loss": 0.7346,
+      "loss": 0.7712,
       "step": 25689
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.256497120297052e-07,
-      "loss": 0.684,
+      "loss": 0.813,
       "step": 25690
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2540097271899653e-07,
-      "loss": 0.799,
+      "loss": 0.8431,
       "step": 25691
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.251523690169799e-07,
-      "loss": 0.592,
+      "loss": 0.6761,
       "step": 25692
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2490390092710702e-07,
-      "loss": 0.6866,
+      "loss": 0.7502,
       "step": 25693
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2465556845282177e-07,
-      "loss": 0.8618,
+      "loss": 0.8826,
       "step": 25694
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2440737159757253e-07,
-      "loss": 0.715,
+      "loss": 0.7138,
       "step": 25695
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.24159310364801e-07,
-      "loss": 0.7938,
+      "loss": 0.8049,
       "step": 25696
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2391138475794881e-07,
-      "loss": 0.5945,
+      "loss": 0.7383,
       "step": 25697
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2366359478045663e-07,
-      "loss": 0.7484,
+      "loss": 0.7318,
       "step": 25698
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2341594043576053e-07,
-      "loss": 0.6587,
+      "loss": 0.7435,
       "step": 25699
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2316842172729668e-07,
-      "loss": 0.7203,
+      "loss": 0.7727,
       "step": 25700
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2292103865849902e-07,
-      "loss": 0.6128,
+      "loss": 0.6675,
       "step": 25701
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2267379123280143e-07,
-      "loss": 0.6887,
+      "loss": 0.7725,
       "step": 25702
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2242667945363226e-07,
-      "loss": 0.8273,
+      "loss": 0.7931,
       "step": 25703
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2217970332441995e-07,
-      "loss": 0.7448,
+      "loss": 0.8383,
       "step": 25704
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2193286284859062e-07,
-      "loss": 0.6865,
+      "loss": 0.7632,
       "step": 25705
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.216861580295715e-07,
-      "loss": 0.7455,
+      "loss": 0.8021,
       "step": 25706
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.214395888707821e-07,
-      "loss": 0.7249,
+      "loss": 0.8052,
       "step": 25707
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2119315537564524e-07,
-      "loss": 0.7229,
+      "loss": 0.7271,
       "step": 25708
     },
     {
       "epoch": 0.93,
       "learning_rate": 2.2094685754757817e-07,
-      "loss": 0.7128,
+      "loss": 0.7992,
       "step": 25709
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.2070069538999926e-07,
-      "loss": 0.6336,
+      "loss": 0.7301,
       "step": 25710
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.204546689063225e-07,
-      "loss": 0.772,
+      "loss": 0.8106,
       "step": 25711
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.20208778099964e-07,
-      "loss": 0.8977,
+      "loss": 0.9053,
       "step": 25712
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1996302297432993e-07,
-      "loss": 0.7229,
+      "loss": 0.8289,
       "step": 25713
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1971740353283422e-07,
-      "loss": 0.7076,
+      "loss": 0.8264,
       "step": 25714
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1947191977888416e-07,
-      "loss": 0.7665,
+      "loss": 0.8144,
       "step": 25715
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1922657171588367e-07,
-      "loss": 0.8535,
+      "loss": 0.7877,
       "step": 25716
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1898135934723674e-07,
-      "loss": 0.6648,
+      "loss": 0.7418,
       "step": 25717
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1873628267634617e-07,
-      "loss": 0.6404,
+      "loss": 0.7565,
       "step": 25718
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.184913417066137e-07,
-      "loss": 0.8364,
+      "loss": 0.8578,
       "step": 25719
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1824653644143323e-07,
-      "loss": 0.8135,
+      "loss": 0.7877,
       "step": 25720
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1800186688420434e-07,
-      "loss": 0.7417,
+      "loss": 0.8127,
       "step": 25721
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1775733303832202e-07,
-      "loss": 0.6496,
+      "loss": 0.752,
       "step": 25722
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1751293490717586e-07,
-      "loss": 0.669,
+      "loss": 0.6967,
       "step": 25723
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.172686724941586e-07,
-      "loss": 0.7456,
+      "loss": 0.7359,
       "step": 25724
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1702454580265985e-07,
-      "loss": 0.8091,
+      "loss": 0.7985,
       "step": 25725
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1678055483606354e-07,
-      "loss": 0.6509,
+      "loss": 0.7334,
       "step": 25726
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1653669959775692e-07,
-      "loss": 0.6216,
+      "loss": 0.6715,
       "step": 25727
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1629298009112287e-07,
-      "loss": 0.6809,
+      "loss": 0.7669,
       "step": 25728
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1604939631954314e-07,
-      "loss": 0.8095,
+      "loss": 0.8162,
       "step": 25729
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.15805948286395e-07,
-      "loss": 0.6846,
+      "loss": 0.7137,
       "step": 25730
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.155626359950591e-07,
-      "loss": 0.8635,
+      "loss": 0.818,
       "step": 25731
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1531945944890832e-07,
-      "loss": 0.8359,
+      "loss": 0.8817,
       "step": 25732
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1507641865131767e-07,
-      "loss": 0.7892,
+      "loss": 0.898,
       "step": 25733
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1483351360565784e-07,
-      "loss": 0.5584,
+      "loss": 0.7534,
       "step": 25734
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.145907443153006e-07,
-      "loss": 0.834,
+      "loss": 0.9319,
       "step": 25735
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1434811078361317e-07,
-      "loss": 0.772,
+      "loss": 0.7635,
       "step": 25736
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1410561301396071e-07,
-      "loss": 0.6431,
+      "loss": 0.7141,
       "step": 25737
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1386325100971052e-07,
-      "loss": 0.8415,
+      "loss": 0.8309,
       "step": 25738
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1362102477422097e-07,
-      "loss": 0.882,
+      "loss": 0.8497,
       "step": 25739
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.13378934310855e-07,
-      "loss": 0.8364,
+      "loss": 0.8543,
       "step": 25740
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1313697962297207e-07,
-      "loss": 0.6996,
+      "loss": 0.7339,
       "step": 25741
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1289516071392736e-07,
-      "loss": 0.7152,
+      "loss": 0.6921,
       "step": 25742
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1265347758707478e-07,
-      "loss": 0.7208,
+      "loss": 0.787,
       "step": 25743
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1241193024577055e-07,
-      "loss": 0.6863,
+      "loss": 0.7454,
       "step": 25744
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1217051869336314e-07,
-      "loss": 0.7324,
+      "loss": 0.7455,
       "step": 25745
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1192924293320317e-07,
-      "loss": 0.6538,
+      "loss": 0.7243,
       "step": 25746
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1168810296863684e-07,
-      "loss": 0.706,
+      "loss": 0.7051,
       "step": 25747
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.114470988030104e-07,
-      "loss": 0.6566,
+      "loss": 0.7134,
       "step": 25748
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.112062304396667e-07,
-      "loss": 0.7093,
+      "loss": 0.7728,
       "step": 25749
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1096549788194975e-07,
-      "loss": 0.7915,
+      "loss": 0.7878,
       "step": 25750
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1072490113319687e-07,
-      "loss": 0.7845,
+      "loss": 0.8175,
       "step": 25751
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.104844401967465e-07,
-      "loss": 0.7658,
+      "loss": 0.8461,
       "step": 25752
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.10244115075936e-07,
-      "loss": 0.5682,
+      "loss": 0.7394,
       "step": 25753
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.1000392577409822e-07,
-      "loss": 0.7915,
+      "loss": 0.8351,
       "step": 25754
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0976387229456496e-07,
-      "loss": 0.6789,
+      "loss": 0.7263,
       "step": 25755
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0952395464066688e-07,
-      "loss": 0.7388,
+      "loss": 0.8416,
       "step": 25756
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0928417281573577e-07,
-      "loss": 0.7182,
+      "loss": 0.7985,
       "step": 25757
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.090445268230934e-07,
-      "loss": 0.8562,
+      "loss": 0.8067,
       "step": 25758
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.088050166660671e-07,
-      "loss": 0.7006,
+      "loss": 0.7478,
       "step": 25759
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0856564234797984e-07,
-      "loss": 0.8355,
+      "loss": 0.8647,
       "step": 25760
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.083264038721511e-07,
-      "loss": 0.7181,
+      "loss": 0.7902,
       "step": 25761
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0808730124190047e-07,
-      "loss": 0.7064,
+      "loss": 0.8372,
       "step": 25762
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0784833446054753e-07,
-      "loss": 0.8188,
+      "loss": 0.8175,
       "step": 25763
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0760950353140408e-07,
-      "loss": 0.7846,
+      "loss": 0.8331,
       "step": 25764
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0737080845778634e-07,
-      "loss": 0.7311,
+      "loss": 0.7975,
       "step": 25765
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0713224924300391e-07,
-      "loss": 0.756,
+      "loss": 0.7714,
       "step": 25766
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0689382589036856e-07,
-      "loss": 0.7607,
+      "loss": 0.7666,
       "step": 25767
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.066555384031843e-07,
-      "loss": 0.7953,
+      "loss": 0.8314,
       "step": 25768
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0641738678476188e-07,
-      "loss": 0.7036,
+      "loss": 0.7879,
       "step": 25769
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.061793710384019e-07,
-      "loss": 0.7168,
+      "loss": 0.7885,
       "step": 25770
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0594149116740737e-07,
-      "loss": 0.7868,
+      "loss": 0.8171,
       "step": 25771
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0570374717507891e-07,
-      "loss": 0.8196,
+      "loss": 0.8556,
       "step": 25772
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0546613906471612e-07,
-      "loss": 0.7584,
+      "loss": 0.7672,
       "step": 25773
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0522866683961196e-07,
-      "loss": 0.757,
+      "loss": 0.7562,
       "step": 25774
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0499133050306374e-07,
-      "loss": 0.8438,
+      "loss": 0.8879,
       "step": 25775
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.047541300583644e-07,
-      "loss": 0.7168,
+      "loss": 0.7851,
       "step": 25776
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0451706550880245e-07,
-      "loss": 0.7976,
+      "loss": 0.8274,
       "step": 25777
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0428013685766857e-07,
-      "loss": 0.793,
+      "loss": 0.7921,
       "step": 25778
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.040433441082501e-07,
-      "loss": 0.6787,
+      "loss": 0.708,
       "step": 25779
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0380668726383112e-07,
-      "loss": 0.766,
+      "loss": 0.8503,
       "step": 25780
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0357016632769567e-07,
-      "loss": 0.6953,
+      "loss": 0.7266,
       "step": 25781
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.033337813031244e-07,
-      "loss": 0.6819,
+      "loss": 0.879,
       "step": 25782
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0309753219339812e-07,
-      "loss": 0.6828,
+      "loss": 0.693,
       "step": 25783
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0286141900179303e-07,
-      "loss": 0.7706,
+      "loss": 0.8451,
       "step": 25784
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0262544173158428e-07,
-      "loss": 0.7063,
+      "loss": 0.807,
       "step": 25785
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0238960038604927e-07,
-      "loss": 0.7529,
+      "loss": 0.7693,
       "step": 25786
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.021538949684554e-07,
-      "loss": 0.7069,
+      "loss": 0.7637,
       "step": 25787
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0191832548207558e-07,
-      "loss": 0.6943,
+      "loss": 0.7178,
       "step": 25788
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.016828919301783e-07,
-      "loss": 0.6468,
+      "loss": 0.717,
       "step": 25789
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0144759431602768e-07,
-      "loss": 0.6663,
+      "loss": 0.736,
       "step": 25790
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0121243264288993e-07,
-      "loss": 0.6613,
+      "loss": 0.7238,
       "step": 25791
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0097740691402802e-07,
-      "loss": 0.6994,
+      "loss": 0.7751,
       "step": 25792
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0074251713270042e-07,
-      "loss": 0.7304,
+      "loss": 0.7768,
       "step": 25793
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0050776330216571e-07,
-      "loss": 0.7321,
+      "loss": 0.7719,
       "step": 25794
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0027314542568454e-07,
-      "loss": 0.6443,
+      "loss": 0.7251,
       "step": 25795
     },
     {
       "epoch": 0.94,
       "learning_rate": 2.0003866350650768e-07,
-      "loss": 0.6775,
+      "loss": 0.6958,
       "step": 25796
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9980431754789142e-07,
-      "loss": 0.735,
+      "loss": 0.7648,
       "step": 25797
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9957010755308538e-07,
-      "loss": 0.7169,
+      "loss": 0.8095,
       "step": 25798
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9933603352533914e-07,
-      "loss": 0.6853,
+      "loss": 0.7812,
       "step": 25799
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9910209546789905e-07,
-      "loss": 0.7919,
+      "loss": 0.9063,
       "step": 25800
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9886829338401248e-07,
-      "loss": 0.6414,
+      "loss": 0.6944,
       "step": 25801
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.986346272769224e-07,
-      "loss": 0.7615,
+      "loss": 0.84,
       "step": 25802
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9840109714987066e-07,
-      "loss": 0.7927,
+      "loss": 0.8494,
       "step": 25803
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.981677030060958e-07,
-      "loss": 0.7432,
+      "loss": 0.8279,
       "step": 25804
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9793444484883962e-07,
-      "loss": 0.6784,
+      "loss": 0.7782,
       "step": 25805
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.977013226813329e-07,
-      "loss": 0.7626,
+      "loss": 0.8007,
       "step": 25806
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9746833650681418e-07,
-      "loss": 0.7704,
+      "loss": 0.8449,
       "step": 25807
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9723548632851418e-07,
-      "loss": 0.6898,
+      "loss": 0.7857,
       "step": 25808
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9700277214966368e-07,
-      "loss": 0.6871,
+      "loss": 0.8125,
       "step": 25809
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9677019397349006e-07,
-      "loss": 0.5666,
+      "loss": 0.6987,
       "step": 25810
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9653775180322298e-07,
-      "loss": 0.7904,
+      "loss": 0.7986,
       "step": 25811
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.963054456420843e-07,
-      "loss": 0.7507,
+      "loss": 0.8651,
       "step": 25812
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.96073275493297e-07,
-      "loss": 0.5836,
+      "loss": 0.6978,
       "step": 25813
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.958412413600841e-07,
-      "loss": 0.8401,
+      "loss": 0.853,
       "step": 25814
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9560934324566294e-07,
-      "loss": 0.689,
+      "loss": 0.7561,
       "step": 25815
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9537758115325212e-07,
-      "loss": 0.8197,
+      "loss": 0.8294,
       "step": 25816
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.951459550860657e-07,
-      "loss": 0.605,
+      "loss": 0.7586,
       "step": 25817
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9491446504732004e-07,
-      "loss": 0.8306,
+      "loss": 0.8692,
       "step": 25818
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9468311104022253e-07,
-      "loss": 0.7798,
+      "loss": 0.7588,
       "step": 25819
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9445189306798507e-07,
-      "loss": 0.7895,
+      "loss": 0.8393,
       "step": 25820
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.942208111338162e-07,
-      "loss": 0.7543,
+      "loss": 0.7709,
       "step": 25821
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9398986524092e-07,
-      "loss": 0.7347,
+      "loss": 0.7892,
       "step": 25822
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9375905539250172e-07,
-      "loss": 0.8523,
+      "loss": 0.8589,
       "step": 25823
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9352838159176434e-07,
-      "loss": 0.741,
+      "loss": 0.7868,
       "step": 25824
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.932978438419053e-07,
-      "loss": 0.7433,
+      "loss": 0.7764,
       "step": 25825
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.930674421461265e-07,
-      "loss": 0.8554,
+      "loss": 0.927,
       "step": 25826
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9283717650762203e-07,
-      "loss": 0.7137,
+      "loss": 0.7688,
       "step": 25827
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.926070469295871e-07,
-      "loss": 0.713,
+      "loss": 0.7428,
       "step": 25828
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9237705341521474e-07,
-      "loss": 0.8099,
+      "loss": 0.84,
       "step": 25829
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.921471959676957e-07,
-      "loss": 0.8693,
+      "loss": 0.8692,
       "step": 25830
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.919174745902197e-07,
-      "loss": 0.7087,
+      "loss": 0.7357,
       "step": 25831
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9168788928597083e-07,
-      "loss": 0.7078,
+      "loss": 0.8029,
       "step": 25832
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9145844005813762e-07,
-      "loss": 0.7759,
+      "loss": 0.8,
       "step": 25833
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9122912690990093e-07,
-      "loss": 0.7267,
+      "loss": 0.8226,
       "step": 25834
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9099994984444481e-07,
-      "loss": 0.7508,
+      "loss": 0.7856,
       "step": 25835
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9077090886494565e-07,
-      "loss": 0.7213,
+      "loss": 0.7931,
       "step": 25836
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9054200397458422e-07,
-      "loss": 0.8515,
+      "loss": 0.8412,
       "step": 25837
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9031323517653355e-07,
-      "loss": 0.7534,
+      "loss": 0.7203,
       "step": 25838
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.9008460247396886e-07,
-      "loss": 0.6787,
+      "loss": 0.7703,
       "step": 25839
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.898561058700621e-07,
-      "loss": 0.7679,
+      "loss": 0.8376,
       "step": 25840
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8962774536798289e-07,
-      "loss": 0.6791,
+      "loss": 0.7793,
       "step": 25841
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8939952097089986e-07,
-      "loss": 0.6433,
+      "loss": 0.7422,
       "step": 25842
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8917143268197934e-07,
-      "loss": 0.708,
+      "loss": 0.754,
       "step": 25843
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8894348050438437e-07,
-      "loss": 0.83,
+      "loss": 0.7989,
       "step": 25844
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8871566444127908e-07,
-      "loss": 0.6774,
+      "loss": 0.805,
       "step": 25845
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.884879844958243e-07,
-      "loss": 0.7127,
+      "loss": 0.7507,
       "step": 25846
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8826044067117745e-07,
-      "loss": 0.8408,
+      "loss": 0.8012,
       "step": 25847
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8803303297049492e-07,
-      "loss": 0.7216,
+      "loss": 0.8175,
       "step": 25848
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8780576139693419e-07,
-      "loss": 0.7801,
+      "loss": 0.7882,
       "step": 25849
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8757862595364716e-07,
-      "loss": 0.8152,
+      "loss": 0.8575,
       "step": 25850
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.873516266437836e-07,
-      "loss": 0.6973,
+      "loss": 0.7864,
       "step": 25851
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8712476347049423e-07,
-      "loss": 0.6902,
+      "loss": 0.7632,
       "step": 25852
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.868980364369255e-07,
-      "loss": 0.7649,
+      "loss": 0.8197,
       "step": 25853
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8667144554622486e-07,
-      "loss": 0.7597,
+      "loss": 0.8076,
       "step": 25854
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.864449908015331e-07,
-      "loss": 0.8165,
+      "loss": 0.8652,
       "step": 25855
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8621867220599443e-07,
-      "loss": 0.7959,
+      "loss": 0.8455,
       "step": 25856
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8599248976274741e-07,
-      "loss": 0.838,
+      "loss": 0.8577,
       "step": 25857
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8576644347493067e-07,
-      "loss": 0.719,
+      "loss": 0.7683,
       "step": 25858
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8554053334567945e-07,
-      "loss": 0.7011,
+      "loss": 0.7988,
       "step": 25859
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8531475937812905e-07,
-      "loss": 0.7325,
+      "loss": 0.7584,
       "step": 25860
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8508912157541025e-07,
-      "loss": 0.7419,
+      "loss": 0.7859,
       "step": 25861
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8486361994065505e-07,
-      "loss": 0.5747,
+      "loss": 0.706,
       "step": 25862
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8463825447699092e-07,
-      "loss": 0.7405,
+      "loss": 0.8033,
       "step": 25863
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8441302518754533e-07,
-      "loss": 0.7721,
+      "loss": 0.8421,
       "step": 25864
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.841879320754425e-07,
-      "loss": 0.779,
+      "loss": 0.8013,
       "step": 25865
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8396297514380545e-07,
-      "loss": 0.667,
+      "loss": 0.7122,
       "step": 25866
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8373815439575283e-07,
-      "loss": 0.7202,
+      "loss": 0.72,
       "step": 25867
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8351346983440876e-07,
-      "loss": 0.704,
+      "loss": 0.754,
       "step": 25868
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8328892146288635e-07,
-      "loss": 0.8149,
+      "loss": 0.8319,
       "step": 25869
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8306450928430197e-07,
-      "loss": 0.8264,
+      "loss": 0.7875,
       "step": 25870
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8284023330176982e-07,
-      "loss": 0.758,
+      "loss": 0.8253,
       "step": 25871
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.826160935183996e-07,
-      "loss": 0.7614,
+      "loss": 0.7661,
       "step": 25872
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8239208993730327e-07,
-      "loss": 0.905,
+      "loss": 0.9274,
       "step": 25873
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8216822256158507e-07,
-      "loss": 0.7292,
+      "loss": 0.8199,
       "step": 25874
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8194449139435576e-07,
-      "loss": 0.8104,
+      "loss": 0.8813,
       "step": 25875
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8172089643871517e-07,
-      "loss": 0.9106,
+      "loss": 0.8952,
       "step": 25876
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.814974376977674e-07,
-      "loss": 0.7408,
+      "loss": 0.7503,
       "step": 25877
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8127411517461114e-07,
-      "loss": 0.5499,
+      "loss": 0.6655,
       "step": 25878
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8105092887234722e-07,
-      "loss": 0.7591,
+      "loss": 0.8704,
       "step": 25879
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.808278787940687e-07,
-      "loss": 0.727,
+      "loss": 0.776,
       "step": 25880
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8060496494287205e-07,
-      "loss": 0.697,
+      "loss": 0.7373,
       "step": 25881
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.803821873218503e-07,
-      "loss": 0.711,
+      "loss": 0.8339,
       "step": 25882
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.8015954593409323e-07,
-      "loss": 0.7698,
+      "loss": 0.8614,
       "step": 25883
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7993704078268948e-07,
-      "loss": 0.6829,
+      "loss": 0.7091,
       "step": 25884
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7971467187072765e-07,
-      "loss": 0.741,
+      "loss": 0.7887,
       "step": 25885
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.794924392012909e-07,
-      "loss": 0.7475,
+      "loss": 0.7894,
       "step": 25886
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7927034277746225e-07,
-      "loss": 0.7902,
+      "loss": 0.847,
       "step": 25887
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7904838260232594e-07,
-      "loss": 0.6265,
+      "loss": 0.7688,
       "step": 25888
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7882655867895726e-07,
-      "loss": 0.7082,
+      "loss": 0.7921,
       "step": 25889
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.78604871010436e-07,
-      "loss": 0.7887,
+      "loss": 0.8093,
       "step": 25890
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7838331959983857e-07,
-      "loss": 0.6317,
+      "loss": 0.7662,
       "step": 25891
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7816190445023695e-07,
-      "loss": 0.6509,
+      "loss": 0.6547,
       "step": 25892
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7794062556470205e-07,
-      "loss": 0.6323,
+      "loss": 0.78,
       "step": 25893
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.777194829463069e-07,
-      "loss": 0.7483,
+      "loss": 0.8461,
       "step": 25894
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.774984765981169e-07,
-      "loss": 0.8258,
+      "loss": 0.8548,
       "step": 25895
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.772776065232007e-07,
-      "loss": 0.732,
+      "loss": 0.803,
       "step": 25896
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7705687272462026e-07,
-      "loss": 0.6135,
+      "loss": 0.7576,
       "step": 25897
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7683627520543866e-07,
-      "loss": 0.7569,
+      "loss": 0.7776,
       "step": 25898
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7661581396871575e-07,
-      "loss": 0.6831,
+      "loss": 0.7801,
       "step": 25899
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7639548901751237e-07,
-      "loss": 0.5934,
+      "loss": 0.7148,
       "step": 25900
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7617530035488272e-07,
-      "loss": 0.706,
+      "loss": 0.7556,
       "step": 25901
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.759552479838833e-07,
-      "loss": 0.8196,
+      "loss": 0.8571,
       "step": 25902
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7573533190756498e-07,
-      "loss": 0.8389,
+      "loss": 0.8051,
       "step": 25903
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7551555212898198e-07,
-      "loss": 0.7266,
+      "loss": 0.8115,
       "step": 25904
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7529590865117963e-07,
-      "loss": 0.7279,
+      "loss": 0.7046,
       "step": 25905
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7507640147720883e-07,
-      "loss": 0.724,
+      "loss": 0.7313,
       "step": 25906
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7485703061011271e-07,
-      "loss": 0.7279,
+      "loss": 0.7977,
       "step": 25907
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.746377960529344e-07,
-      "loss": 0.6738,
+      "loss": 0.8178,
       "step": 25908
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.74418697808717e-07,
-      "loss": 0.8602,
+      "loss": 0.8471,
       "step": 25909
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.741997358805003e-07,
-      "loss": 0.711,
+      "loss": 0.7516,
       "step": 25910
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7398091027131968e-07,
-      "loss": 0.7608,
+      "loss": 0.8126,
       "step": 25911
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7376222098421268e-07,
-      "loss": 0.7945,
+      "loss": 0.8735,
       "step": 25912
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7354366802221468e-07,
-      "loss": 0.6881,
+      "loss": 0.7479,
       "step": 25913
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7332525138835543e-07,
-      "loss": 0.8258,
+      "loss": 0.8649,
       "step": 25914
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.73106971085667e-07,
-      "loss": 0.846,
+      "loss": 0.79,
       "step": 25915
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.728888271171747e-07,
-      "loss": 0.8621,
+      "loss": 0.8528,
       "step": 25916
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7267081948590948e-07,
-      "loss": 0.7007,
+      "loss": 0.7615,
       "step": 25917
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7245294819489223e-07,
-      "loss": 0.8187,
+      "loss": 0.8293,
       "step": 25918
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.722352132471472e-07,
-      "loss": 0.825,
+      "loss": 0.828,
       "step": 25919
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7201761464569422e-07,
-      "loss": 0.7077,
+      "loss": 0.7864,
       "step": 25920
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7180015239355418e-07,
-      "loss": 0.8135,
+      "loss": 0.8388,
       "step": 25921
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7158282649374026e-07,
-      "loss": 0.6573,
+      "loss": 0.7471,
       "step": 25922
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.713656369492722e-07,
-      "loss": 0.7942,
+      "loss": 0.7991,
       "step": 25923
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.711485837631599e-07,
-      "loss": 0.7027,
+      "loss": 0.784,
       "step": 25924
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7093166693841646e-07,
-      "loss": 0.7573,
+      "loss": 0.8034,
       "step": 25925
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7071488647804945e-07,
-      "loss": 0.8129,
+      "loss": 0.802,
       "step": 25926
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7049824238506764e-07,
-      "loss": 0.7029,
+      "loss": 0.7726,
       "step": 25927
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7028173466247744e-07,
-      "loss": 0.7679,
+      "loss": 0.7063,
       "step": 25928
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.7006536331327982e-07,
-      "loss": 0.6828,
+      "loss": 0.759,
       "step": 25929
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.698491283404813e-07,
-      "loss": 0.7406,
+      "loss": 0.7843,
       "step": 25930
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6963302974707607e-07,
-      "loss": 0.7599,
+      "loss": 0.8543,
       "step": 25931
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6941706753606623e-07,
-      "loss": 0.7337,
+      "loss": 0.7594,
       "step": 25932
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6920124171044717e-07,
-      "loss": 0.7351,
+      "loss": 0.8454,
       "step": 25933
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6898555227321312e-07,
-      "loss": 0.787,
+      "loss": 0.7899,
       "step": 25934
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6876999922735393e-07,
-      "loss": 0.8274,
+      "loss": 0.923,
       "step": 25935
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.68554582575865e-07,
-      "loss": 0.6825,
+      "loss": 0.7621,
       "step": 25936
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6833930232173056e-07,
-      "loss": 0.6881,
+      "loss": 0.6787,
       "step": 25937
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.681241584679394e-07,
-      "loss": 0.7663,
+      "loss": 0.8597,
       "step": 25938
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6790915101747685e-07,
-      "loss": 0.7885,
+      "loss": 0.7616,
       "step": 25939
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.676942799733239e-07,
-      "loss": 0.7193,
+      "loss": 0.7751,
       "step": 25940
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6747954533846257e-07,
-      "loss": 0.6343,
+      "loss": 0.7845,
       "step": 25941
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.672649471158727e-07,
-      "loss": 0.7221,
+      "loss": 0.8065,
       "step": 25942
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6705048530852973e-07,
-      "loss": 0.8748,
+      "loss": 0.9183,
       "step": 25943
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6683615991941127e-07,
-      "loss": 0.7044,
+      "loss": 0.7538,
       "step": 25944
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6662197095148936e-07,
-      "loss": 0.6091,
+      "loss": 0.7097,
       "step": 25945
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6640791840773608e-07,
-      "loss": 0.8111,
+      "loss": 0.8655,
       "step": 25946
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6619400229112016e-07,
-      "loss": 0.7527,
+      "loss": 0.7667,
       "step": 25947
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6598022260461034e-07,
-      "loss": 0.7359,
+      "loss": 0.7909,
       "step": 25948
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6576657935117312e-07,
-      "loss": 0.7681,
+      "loss": 0.8705,
       "step": 25949
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6555307253377062e-07,
-      "loss": 0.7569,
+      "loss": 0.8095,
       "step": 25950
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.65339702155366e-07,
-      "loss": 0.6866,
+      "loss": 0.6782,
       "step": 25951
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.651264682189202e-07,
-      "loss": 0.7687,
+      "loss": 0.8212,
       "step": 25952
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6491337072739087e-07,
-      "loss": 0.8548,
+      "loss": 0.8965,
       "step": 25953
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6470040968373234e-07,
-      "loss": 0.8571,
+      "loss": 0.8308,
       "step": 25954
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6448758509090446e-07,
-      "loss": 0.7315,
+      "loss": 0.8244,
       "step": 25955
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6427489695185372e-07,
-      "loss": 0.721,
+      "loss": 0.772,
       "step": 25956
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6406234526953558e-07,
-      "loss": 0.8685,
+      "loss": 0.8656,
       "step": 25957
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6384993004689544e-07,
-      "loss": 0.8439,
+      "loss": 0.8408,
       "step": 25958
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6363765128688313e-07,
-      "loss": 0.7906,
+      "loss": 0.7695,
       "step": 25959
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6342550899244191e-07,
-      "loss": 0.6962,
+      "loss": 0.732,
       "step": 25960
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6321350316651497e-07,
-      "loss": 0.605,
+      "loss": 0.6855,
       "step": 25961
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6300163381204547e-07,
-      "loss": 0.7084,
+      "loss": 0.7321,
       "step": 25962
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6278990093197e-07,
-      "loss": 0.7542,
+      "loss": 0.7809,
       "step": 25963
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6257830452922841e-07,
-      "loss": 0.7169,
+      "loss": 0.7187,
       "step": 25964
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.62366844606755e-07,
-      "loss": 0.6702,
+      "loss": 0.792,
       "step": 25965
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.621555211674841e-07,
-      "loss": 0.7193,
+      "loss": 0.7272,
       "step": 25966
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6194433421434675e-07,
-      "loss": 0.8915,
+      "loss": 0.8823,
       "step": 25967
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6173328375027497e-07,
-      "loss": 0.6938,
+      "loss": 0.7883,
       "step": 25968
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6152236977819312e-07,
-      "loss": 0.8284,
+      "loss": 0.7069,
       "step": 25969
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6131159230103e-07,
-      "loss": 0.8175,
+      "loss": 0.8711,
       "step": 25970
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6110095132170988e-07,
-      "loss": 0.6295,
+      "loss": 0.7037,
       "step": 25971
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6089044684315492e-07,
-      "loss": 0.8174,
+      "loss": 0.8005,
       "step": 25972
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6068007886828275e-07,
-      "loss": 0.7717,
+      "loss": 0.8933,
       "step": 25973
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6046984740001658e-07,
-      "loss": 0.9255,
+      "loss": 0.9743,
       "step": 25974
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6025975244126968e-07,
-      "loss": 0.8602,
+      "loss": 0.854,
       "step": 25975
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.6004979399495857e-07,
-      "loss": 0.688,
+      "loss": 0.8041,
       "step": 25976
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.5983997206399647e-07,
-      "loss": 0.709,
+      "loss": 0.73,
       "step": 25977
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.5963028665129222e-07,
-      "loss": 0.6953,
+      "loss": 0.8026,
       "step": 25978
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.5942073775975563e-07,
-      "loss": 0.6882,
+      "loss": 0.7153,
       "step": 25979
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.5921132539229444e-07,
-      "loss": 0.7413,
+      "loss": 0.7885,
       "step": 25980
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.590020495518141e-07,
-      "loss": 0.7382,
+      "loss": 0.7665,
       "step": 25981
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.587929102412189e-07,
-      "loss": 0.7341,
+      "loss": 0.7337,
       "step": 25982
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.585839074634077e-07,
-      "loss": 0.722,
+      "loss": 0.7354,
       "step": 25983
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.5837504122128254e-07,
-      "loss": 0.7032,
+      "loss": 0.7731,
       "step": 25984
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5816631151774008e-07,
-      "loss": 0.7569,
+      "loss": 0.8002,
       "step": 25985
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5795771835567575e-07,
-      "loss": 0.6219,
+      "loss": 0.675,
       "step": 25986
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5774926173798499e-07,
-      "loss": 0.8063,
+      "loss": 0.8745,
       "step": 25987
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5754094166755772e-07,
-      "loss": 0.6972,
+      "loss": 0.7337,
       "step": 25988
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.573327581472861e-07,
-      "loss": 0.7206,
+      "loss": 0.8014,
       "step": 25989
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.571247111800578e-07,
-      "loss": 0.7065,
+      "loss": 0.7725,
       "step": 25990
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5691680076875825e-07,
-      "loss": 0.8452,
+      "loss": 0.8471,
       "step": 25991
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5670902691627187e-07,
-      "loss": 0.7393,
+      "loss": 0.7569,
       "step": 25992
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.56501389625483e-07,
-      "loss": 0.6997,
+      "loss": 0.7501,
       "step": 25993
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5629388889927156e-07,
-      "loss": 0.7308,
+      "loss": 0.7417,
       "step": 25994
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5608652474051523e-07,
-      "loss": 0.6628,
+      "loss": 0.7545,
       "step": 25995
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5587929715209282e-07,
-      "loss": 0.6671,
+      "loss": 0.7679,
       "step": 25996
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.556722061368776e-07,
-      "loss": 0.5882,
+      "loss": 0.694,
       "step": 25997
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5546525169774173e-07,
-      "loss": 0.7251,
+      "loss": 0.7364,
       "step": 25998
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5525843383755955e-07,
-      "loss": 0.8132,
+      "loss": 0.7227,
       "step": 25999
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.550517525591988e-07,
-      "loss": 0.7653,
+      "loss": 0.7578,
       "step": 26000
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5484520786552605e-07,
-      "loss": 0.6348,
+      "loss": 0.7487,
       "step": 26001
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5463879975940786e-07,
-      "loss": 0.7599,
+      "loss": 0.813,
       "step": 26002
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.544325282437087e-07,
-      "loss": 0.7793,
+      "loss": 0.796,
       "step": 26003
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.542263933212884e-07,
-      "loss": 0.7962,
+      "loss": 0.8149,
       "step": 26004
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5402039499500587e-07,
-      "loss": 0.8169,
+      "loss": 0.8177,
       "step": 26005
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5381453326772322e-07,
-      "loss": 0.6244,
+      "loss": 0.809,
       "step": 26006
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5360880814229262e-07,
-      "loss": 0.7371,
+      "loss": 0.7434,
       "step": 26007
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5340321962157068e-07,
-      "loss": 0.6647,
+      "loss": 0.71,
       "step": 26008
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5319776770840732e-07,
-      "loss": 0.7979,
+      "loss": 0.8595,
       "step": 26009
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5299245240565474e-07,
-      "loss": 0.6244,
+      "loss": 0.6994,
       "step": 26010
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5278727371616065e-07,
-      "loss": 0.8395,
+      "loss": 0.8041,
       "step": 26011
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5258223164277164e-07,
-      "loss": 0.8017,
+      "loss": 0.8228,
       "step": 26012
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5237732618833323e-07,
-      "loss": 0.7816,
+      "loss": 0.8009,
       "step": 26013
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.521725573556876e-07,
-      "loss": 0.8113,
+      "loss": 0.7466,
       "step": 26014
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5196792514767354e-07,
-      "loss": 0.9227,
+      "loss": 0.9703,
       "step": 26015
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.517634295671344e-07,
-      "loss": 0.7539,
+      "loss": 0.8479,
       "step": 26016
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5155907061690344e-07,
-      "loss": 0.7665,
+      "loss": 0.7859,
       "step": 26017
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.513548482998173e-07,
-      "loss": 0.8883,
+      "loss": 0.7998,
       "step": 26018
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.511507626187103e-07,
-      "loss": 0.8216,
+      "loss": 0.8368,
       "step": 26019
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.509468135764114e-07,
-      "loss": 0.6979,
+      "loss": 0.7608,
       "step": 26020
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5074300117575157e-07,
-      "loss": 0.7621,
+      "loss": 0.9254,
       "step": 26021
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5053932541955973e-07,
-      "loss": 0.6537,
+      "loss": 0.7283,
       "step": 26022
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5033578631066026e-07,
-      "loss": 0.838,
+      "loss": 0.8824,
       "step": 26023
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.5013238385187535e-07,
-      "loss": 0.7233,
+      "loss": 0.7282,
       "step": 26024
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4992911804602938e-07,
-      "loss": 0.7121,
+      "loss": 0.7585,
       "step": 26025
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4972598889594125e-07,
-      "loss": 0.8266,
+      "loss": 0.8361,
       "step": 26026
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4952299640442869e-07,
-      "loss": 0.8309,
+      "loss": 0.8038,
       "step": 26027
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4932014057430832e-07,
-      "loss": 0.7233,
+      "loss": 0.7195,
       "step": 26028
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.491174214083968e-07,
-      "loss": 0.7226,
+      "loss": 0.7074,
       "step": 26029
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4891483890950298e-07,
-      "loss": 0.8068,
+      "loss": 0.8327,
       "step": 26030
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4871239308043905e-07,
-      "loss": 0.8637,
+      "loss": 0.8613,
       "step": 26031
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.485100839240139e-07,
-      "loss": 0.819,
+      "loss": 0.8455,
       "step": 26032
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4830791144303413e-07,
-      "loss": 0.617,
+      "loss": 0.6926,
       "step": 26033
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.481058756403031e-07,
-      "loss": 0.8403,
+      "loss": 0.8083,
       "step": 26034
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.479039765186263e-07,
-      "loss": 0.6842,
+      "loss": 0.7675,
       "step": 26035
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4770221408080265e-07,
-      "loss": 0.6736,
+      "loss": 0.7709,
       "step": 26036
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.475005883296321e-07,
-      "loss": 0.7928,
+      "loss": 0.8,
       "step": 26037
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4729909926791352e-07,
-      "loss": 0.7941,
+      "loss": 0.8391,
       "step": 26038
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4709774689843915e-07,
-      "loss": 0.7756,
+      "loss": 0.7905,
       "step": 26039
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.468965312240045e-07,
-      "loss": 0.8861,
+      "loss": 0.8936,
       "step": 26040
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4669545224740178e-07,
-      "loss": 0.8562,
+      "loss": 0.8846,
       "step": 26041
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.464945099714188e-07,
-      "loss": 0.6132,
+      "loss": 0.6996,
       "step": 26042
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4629370439884326e-07,
-      "loss": 0.7554,
+      "loss": 0.8817,
       "step": 26043
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4609303553246302e-07,
-      "loss": 0.8716,
+      "loss": 0.8163,
       "step": 26044
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4589250337506134e-07,
-      "loss": 0.6582,
+      "loss": 0.8148,
       "step": 26045
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4569210792942045e-07,
-      "loss": 0.7403,
+      "loss": 0.7862,
       "step": 26046
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4549184919831815e-07,
-      "loss": 0.8144,
+      "loss": 0.8497,
       "step": 26047
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4529172718453665e-07,
-      "loss": 0.8607,
+      "loss": 0.8951,
       "step": 26048
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.450917418908493e-07,
-      "loss": 0.9188,
+      "loss": 0.8984,
       "step": 26049
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4489189332003272e-07,
-      "loss": 0.8806,
+      "loss": 0.8585,
       "step": 26050
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4469218147485808e-07,
-      "loss": 0.7319,
+      "loss": 0.826,
       "step": 26051
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.444926063580965e-07,
-      "loss": 0.7672,
+      "loss": 0.7945,
       "step": 26052
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.442931679725157e-07,
-      "loss": 0.6818,
+      "loss": 0.7477,
       "step": 26053
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4409386632088574e-07,
-      "loss": 0.6381,
+      "loss": 0.6704,
       "step": 26054
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4389470140596884e-07,
-      "loss": 0.764,
+      "loss": 0.7578,
       "step": 26055
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4369567323052946e-07,
-      "loss": 0.7484,
+      "loss": 0.8484,
       "step": 26056
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4349678179732763e-07,
-      "loss": 0.633,
+      "loss": 0.7479,
       "step": 26057
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4329802710912332e-07,
-      "loss": 0.7801,
+      "loss": 0.8596,
       "step": 26058
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.430994091686744e-07,
-      "loss": 0.7064,
+      "loss": 0.7626,
       "step": 26059
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4290092797873634e-07,
-      "loss": 0.8279,
+      "loss": 0.7961,
       "step": 26060
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4270258354206257e-07,
-      "loss": 0.7556,
+      "loss": 0.7851,
       "step": 26061
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4250437586140419e-07,
-      "loss": 0.6309,
+      "loss": 0.7009,
       "step": 26062
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4230630493951126e-07,
-      "loss": 0.6007,
+      "loss": 0.6465,
       "step": 26063
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4210837077913376e-07,
-      "loss": 0.6967,
+      "loss": 0.7366,
       "step": 26064
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4191057338301506e-07,
-      "loss": 0.829,
+      "loss": 0.8989,
       "step": 26065
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4171291275389965e-07,
-      "loss": 0.6922,
+      "loss": 0.7672,
       "step": 26066
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4151538889453197e-07,
-      "loss": 0.7707,
+      "loss": 0.7869,
       "step": 26067
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4131800180764988e-07,
-      "loss": 0.7964,
+      "loss": 0.7674,
       "step": 26068
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4112075149599336e-07,
-      "loss": 0.7284,
+      "loss": 0.8473,
       "step": 26069
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4092363796229802e-07,
-      "loss": 0.6777,
+      "loss": 0.7899,
       "step": 26070
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4072666120929946e-07,
-      "loss": 0.7463,
+      "loss": 0.7873,
       "step": 26071
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4052982123972881e-07,
-      "loss": 0.7266,
+      "loss": 0.748,
       "step": 26072
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4033311805631943e-07,
-      "loss": 0.7878,
+      "loss": 0.8306,
       "step": 26073
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.4013655166179917e-07,
-      "loss": 0.7165,
+      "loss": 0.7574,
       "step": 26074
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3994012205889474e-07,
-      "loss": 0.8855,
+      "loss": 0.9007,
       "step": 26075
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3974382925033058e-07,
-      "loss": 0.7825,
+      "loss": 0.7892,
       "step": 26076
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3954767323883233e-07,
-      "loss": 0.7475,
+      "loss": 0.7215,
       "step": 26077
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3935165402711782e-07,
-      "loss": 0.7038,
+      "loss": 0.7347,
       "step": 26078
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3915577161791037e-07,
-      "loss": 0.7833,
+      "loss": 0.8606,
       "step": 26079
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3896002601392677e-07,
-      "loss": 0.8146,
+      "loss": 0.817,
       "step": 26080
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3876441721788037e-07,
-      "loss": 0.734,
+      "loss": 0.8072,
       "step": 26081
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3856894523248676e-07,
-      "loss": 0.787,
+      "loss": 0.7882,
       "step": 26082
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3837361006045714e-07,
-      "loss": 0.7435,
+      "loss": 0.8511,
       "step": 26083
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3817841170450262e-07,
-      "loss": 0.8338,
+      "loss": 0.7502,
       "step": 26084
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.379833501673289e-07,
-      "loss": 0.8122,
+      "loss": 0.8709,
       "step": 26085
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3778842545164596e-07,
-      "loss": 0.8215,
+      "loss": 0.8379,
       "step": 26086
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3759363756015497e-07,
-      "loss": 0.7966,
+      "loss": 0.8668,
       "step": 26087
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3739898649555938e-07,
-      "loss": 0.7225,
+      "loss": 0.8187,
       "step": 26088
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3720447226055923e-07,
-      "loss": 0.7626,
+      "loss": 0.8121,
       "step": 26089
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3701009485785454e-07,
-      "loss": 0.8146,
+      "loss": 0.8432,
       "step": 26090
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3681585429013989e-07,
-      "loss": 0.7892,
+      "loss": 0.8543,
       "step": 26091
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3662175056011196e-07,
-      "loss": 0.7771,
+      "loss": 0.7679,
       "step": 26092
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3642778367046193e-07,
-      "loss": 0.8426,
+      "loss": 0.8074,
       "step": 26093
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3623395362388324e-07,
-      "loss": 0.7792,
+      "loss": 0.8453,
       "step": 26094
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3604026042306263e-07,
-      "loss": 0.7438,
+      "loss": 0.7589,
       "step": 26095
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.35846704070689e-07,
-      "loss": 0.7262,
+      "loss": 0.7922,
       "step": 26096
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3565328456944583e-07,
-      "loss": 0.8314,
+      "loss": 0.8692,
       "step": 26097
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3546000192201868e-07,
-      "loss": 0.7262,
+      "loss": 0.701,
       "step": 26098
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.352668561310888e-07,
-      "loss": 0.7836,
+      "loss": 0.7979,
       "step": 26099
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3507384719933402e-07,
-      "loss": 0.5987,
+      "loss": 0.6894,
       "step": 26100
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3488097512943332e-07,
-      "loss": 0.7149,
+      "loss": 0.7438,
       "step": 26101
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3468823992406232e-07,
-      "loss": 0.7641,
+      "loss": 0.787,
       "step": 26102
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3449564158589445e-07,
-      "loss": 0.663,
+      "loss": 0.6545,
       "step": 26103
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3430318011760203e-07,
-      "loss": 0.752,
+      "loss": 0.8535,
       "step": 26104
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.341108555218562e-07,
-      "loss": 0.8282,
+      "loss": 0.8166,
       "step": 26105
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.339186678013238e-07,
-      "loss": 0.745,
+      "loss": 0.7332,
       "step": 26106
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.337266169586726e-07,
-      "loss": 0.779,
+      "loss": 0.7629,
       "step": 26107
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3353470299656612e-07,
-      "loss": 0.762,
+      "loss": 0.8645,
       "step": 26108
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3334292591766551e-07,
-      "loss": 0.8378,
+      "loss": 0.9102,
       "step": 26109
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.331512857246342e-07,
-      "loss": 0.6963,
+      "loss": 0.7751,
       "step": 26110
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3295978242012897e-07,
-      "loss": 0.7735,
+      "loss": 0.7373,
       "step": 26111
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3276841600680767e-07,
-      "loss": 0.7119,
+      "loss": 0.7631,
       "step": 26112
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3257718648732486e-07,
-      "loss": 0.7002,
+      "loss": 0.8008,
       "step": 26113
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3238609386433288e-07,
-      "loss": 0.7289,
+      "loss": 0.8115,
       "step": 26114
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.32195138140484e-07,
-      "loss": 0.774,
+      "loss": 0.7721,
       "step": 26115
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3200431931842617e-07,
-      "loss": 0.7726,
+      "loss": 0.7495,
       "step": 26116
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3181363740080942e-07,
-      "loss": 0.7407,
+      "loss": 0.7182,
       "step": 26117
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3162309239027615e-07,
-      "loss": 0.6745,
+      "loss": 0.6701,
       "step": 26118
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3143268428947086e-07,
-      "loss": 0.6934,
+      "loss": 0.7012,
       "step": 26119
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.312424131010359e-07,
-      "loss": 0.8428,
+      "loss": 0.8383,
       "step": 26120
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.310522788276103e-07,
-      "loss": 0.7447,
+      "loss": 0.8794,
       "step": 26121
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.30862281471833e-07,
-      "loss": 0.722,
+      "loss": 0.7604,
       "step": 26122
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.306724210363375e-07,
-      "loss": 0.7982,
+      "loss": 0.8233,
       "step": 26123
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3048269752376052e-07,
-      "loss": 0.8033,
+      "loss": 0.7179,
       "step": 26124
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3029311093673224e-07,
-      "loss": 0.7293,
+      "loss": 0.7731,
       "step": 26125
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3010366127788499e-07,
-      "loss": 0.7919,
+      "loss": 0.8026,
       "step": 26126
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2991434854984442e-07,
-      "loss": 0.7103,
+      "loss": 0.8134,
       "step": 26127
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2972517275523956e-07,
-      "loss": 0.8183,
+      "loss": 0.8795,
       "step": 26128
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2953613389669272e-07,
-      "loss": 0.8169,
+      "loss": 0.8615,
       "step": 26129
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2934723197682853e-07,
-      "loss": 0.7101,
+      "loss": 0.8148,
       "step": 26130
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2915846699826707e-07,
-      "loss": 0.7431,
+      "loss": 0.8106,
       "step": 26131
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2896983896362737e-07,
-      "loss": 0.6643,
+      "loss": 0.8249,
       "step": 26132
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.28781347875524e-07,
-      "loss": 0.7491,
+      "loss": 0.8674,
       "step": 26133
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2859299373657595e-07,
-      "loss": 0.894,
+      "loss": 0.9196,
       "step": 26134
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2840477654939343e-07,
-      "loss": 0.8,
+      "loss": 0.7904,
       "step": 26135
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2821669631658873e-07,
-      "loss": 0.6622,
+      "loss": 0.7453,
       "step": 26136
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.28028753040772e-07,
-      "loss": 0.7703,
+      "loss": 0.7976,
       "step": 26137
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2784094672454894e-07,
-      "loss": 0.8753,
+      "loss": 0.9124,
       "step": 26138
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2765327737052634e-07,
-      "loss": 0.7046,
+      "loss": 0.7449,
       "step": 26139
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2746574498130772e-07,
-      "loss": 0.6483,
+      "loss": 0.6681,
       "step": 26140
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2727834955949425e-07,
-      "loss": 0.7344,
+      "loss": 0.8133,
       "step": 26141
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2709109110768613e-07,
-      "loss": 0.8331,
+      "loss": 0.8463,
       "step": 26142
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2690396962848127e-07,
-      "loss": 0.6742,
+      "loss": 0.7334,
       "step": 26143
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.267169851244765e-07,
-      "loss": 0.7788,
+      "loss": 0.7736,
       "step": 26144
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2653013759826416e-07,
-      "loss": 0.7176,
+      "loss": 0.7018,
       "step": 26145
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2634342705243774e-07,
-      "loss": 0.7337,
+      "loss": 0.7462,
       "step": 26146
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2615685348958852e-07,
-      "loss": 0.7543,
+      "loss": 0.777,
       "step": 26147
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2597041691230217e-07,
-      "loss": 0.8223,
+      "loss": 0.8535,
       "step": 26148
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2578411732316664e-07,
-      "loss": 0.7069,
+      "loss": 0.7481,
       "step": 26149
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2559795472476877e-07,
-      "loss": 0.6345,
+      "loss": 0.7242,
       "step": 26150
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2541192911968757e-07,
-      "loss": 0.8321,
+      "loss": 0.861,
       "step": 26151
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2522604051050657e-07,
-      "loss": 0.7007,
+      "loss": 0.7538,
       "step": 26152
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2504028889980368e-07,
-      "loss": 0.7568,
+      "loss": 0.7681,
       "step": 26153
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2485467429015574e-07,
-      "loss": 0.8713,
+      "loss": 0.8442,
       "step": 26154
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2466919668413847e-07,
-      "loss": 0.8348,
+      "loss": 0.8578,
       "step": 26155
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2448385608432422e-07,
-      "loss": 0.7445,
+      "loss": 0.7997,
       "step": 26156
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2429865249328655e-07,
-      "loss": 0.6129,
+      "loss": 0.734,
       "step": 26157
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2411358591359224e-07,
-      "loss": 0.6097,
+      "loss": 0.7247,
       "step": 26158
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2392865634780927e-07,
-      "loss": 0.7993,
+      "loss": 0.8044,
       "step": 26159
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2374386379850445e-07,
-      "loss": 0.7303,
+      "loss": 0.8137,
       "step": 26160
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2355920826824018e-07,
-      "loss": 0.8343,
+      "loss": 0.8679,
       "step": 26161
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2337468975958e-07,
-      "loss": 0.6427,
+      "loss": 0.723,
       "step": 26162
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2319030827508294e-07,
-      "loss": 0.7665,
+      "loss": 0.8073,
       "step": 26163
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2300606381730696e-07,
-      "loss": 0.7137,
+      "loss": 0.7462,
       "step": 26164
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2282195638880778e-07,
-      "loss": 0.6953,
+      "loss": 0.7158,
       "step": 26165
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2263798599214117e-07,
-      "loss": 0.8613,
+      "loss": 0.8464,
       "step": 26166
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2245415262985726e-07,
-      "loss": 0.7287,
+      "loss": 0.7413,
       "step": 26167
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.222704563045074e-07,
-      "loss": 0.5862,
+      "loss": 0.7272,
       "step": 26168
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2208689701864062e-07,
-      "loss": 0.8673,
+      "loss": 0.9041,
       "step": 26169
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.219034747748038e-07,
-      "loss": 0.6489,
+      "loss": 0.7744,
       "step": 26170
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2172018957554044e-07,
-      "loss": 0.7198,
+      "loss": 0.8422,
       "step": 26171
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2153704142339405e-07,
-      "loss": 0.7234,
+      "loss": 0.7251,
       "step": 26172
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2135403032090488e-07,
-      "loss": 0.7385,
+      "loss": 0.7575,
       "step": 26173
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2117115627061417e-07,
-      "loss": 0.7782,
+      "loss": 0.7925,
       "step": 26174
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2098841927505657e-07,
-      "loss": 0.8031,
+      "loss": 0.8101,
       "step": 26175
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2080581933676781e-07,
-      "loss": 0.7953,
+      "loss": 0.797,
       "step": 26176
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.206233564582826e-07,
-      "loss": 0.7449,
+      "loss": 0.8237,
       "step": 26177
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2044103064212998e-07,
-      "loss": 0.6536,
+      "loss": 0.7587,
       "step": 26178
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.2025884189084236e-07,
-      "loss": 0.6671,
+      "loss": 0.809,
       "step": 26179
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.200767902069444e-07,
-      "loss": 0.7682,
+      "loss": 0.7689,
       "step": 26180
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.198948755929641e-07,
-      "loss": 0.6455,
+      "loss": 0.7016,
       "step": 26181
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1971309805142384e-07,
-      "loss": 0.8905,
+      "loss": 0.8453,
       "step": 26182
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1953145758484608e-07,
-      "loss": 0.7189,
+      "loss": 0.8002,
       "step": 26183
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.193499541957499e-07,
-      "loss": 0.6695,
+      "loss": 0.7477,
       "step": 26184
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1916858788665664e-07,
-      "loss": 0.7587,
+      "loss": 0.8253,
       "step": 26185
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.189873586600776e-07,
-      "loss": 0.7637,
+      "loss": 0.7521,
       "step": 26186
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1880626651853189e-07,
-      "loss": 0.6283,
+      "loss": 0.693,
       "step": 26187
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.186253114645286e-07,
-      "loss": 0.7077,
+      "loss": 0.7462,
       "step": 26188
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1844449350057907e-07,
-      "loss": 0.8276,
+      "loss": 0.8253,
       "step": 26189
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1826381262919129e-07,
-      "loss": 0.661,
+      "loss": 0.7904,
       "step": 26190
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1808326885287436e-07,
-      "loss": 0.825,
+      "loss": 0.8649,
       "step": 26191
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1790286217413071e-07,
-      "loss": 0.8285,
+      "loss": 0.8636,
       "step": 26192
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1772259259546504e-07,
-      "loss": 0.6395,
+      "loss": 0.756,
       "step": 26193
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1754246011937531e-07,
-      "loss": 0.8235,
+      "loss": 0.834,
       "step": 26194
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1736246474836399e-07,
-      "loss": 0.7193,
+      "loss": 0.8197,
       "step": 26195
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1718260648492685e-07,
-      "loss": 0.6837,
+      "loss": 0.7792,
       "step": 26196
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1700288533155857e-07,
-      "loss": 0.6674,
+      "loss": 0.7928,
       "step": 26197
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.168233012907538e-07,
-      "loss": 0.8744,
+      "loss": 0.8114,
       "step": 26198
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1664385436500281e-07,
-      "loss": 0.6391,
+      "loss": 0.7376,
       "step": 26199
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1646454455679579e-07,
-      "loss": 0.8068,
+      "loss": 0.8026,
       "step": 26200
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1628537186862077e-07,
-      "loss": 0.6883,
+      "loss": 0.7925,
       "step": 26201
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1610633630296243e-07,
-      "loss": 0.7952,
+      "loss": 0.8871,
       "step": 26202
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1592743786230432e-07,
-      "loss": 0.7838,
+      "loss": 0.7386,
       "step": 26203
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1574867654913113e-07,
-      "loss": 0.8501,
+      "loss": 0.8427,
       "step": 26204
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1557005236592089e-07,
-      "loss": 0.8152,
+      "loss": 0.8417,
       "step": 26205
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1539156531515161e-07,
-      "loss": 0.6522,
+      "loss": 0.7399,
       "step": 26206
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1521321539929909e-07,
-      "loss": 0.8239,
+      "loss": 0.8743,
       "step": 26207
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1503500262084021e-07,
-      "loss": 0.8157,
+      "loss": 0.873,
       "step": 26208
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1485692698224416e-07,
-      "loss": 0.8444,
+      "loss": 0.8552,
       "step": 26209
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1467898848598335e-07,
-      "loss": 0.7781,
+      "loss": 0.7348,
       "step": 26210
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1450118713452695e-07,
-      "loss": 0.7497,
+      "loss": 0.8417,
       "step": 26211
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1432352293034077e-07,
-      "loss": 0.6966,
+      "loss": 0.7437,
       "step": 26212
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1414599587588837e-07,
-      "loss": 0.697,
+      "loss": 0.7844,
       "step": 26213
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1396860597363668e-07,
-      "loss": 0.7509,
+      "loss": 0.7621,
       "step": 26214
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1379135322604263e-07,
-      "loss": 0.7391,
+      "loss": 0.7696,
       "step": 26215
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1361423763556645e-07,
-      "loss": 0.8411,
+      "loss": 0.8312,
       "step": 26216
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.134372592046662e-07,
-      "loss": 0.8303,
+      "loss": 0.851,
       "step": 26217
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1326041793579656e-07,
-      "loss": 0.6891,
+      "loss": 0.8202,
       "step": 26218
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1308371383141226e-07,
-      "loss": 0.7327,
+      "loss": 0.8017,
       "step": 26219
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1290714689396243e-07,
-      "loss": 0.7442,
+      "loss": 0.857,
       "step": 26220
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1273071712589845e-07,
-      "loss": 0.6756,
+      "loss": 0.7369,
       "step": 26221
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1255442452966724e-07,
-      "loss": 0.7441,
+      "loss": 0.8025,
       "step": 26222
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1237826910771576e-07,
-      "loss": 0.6813,
+      "loss": 0.7558,
       "step": 26223
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1220225086248649e-07,
-      "loss": 0.7873,
+      "loss": 0.7931,
       "step": 26224
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.120263697964219e-07,
-      "loss": 0.8739,
+      "loss": 0.9408,
       "step": 26225
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1185062591196227e-07,
-      "loss": 0.707,
+      "loss": 0.8107,
       "step": 26226
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1167501921154677e-07,
-      "loss": 0.57,
+      "loss": 0.6306,
       "step": 26227
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1149954969760901e-07,
-      "loss": 0.6725,
+      "loss": 0.7562,
       "step": 26228
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1132421737258592e-07,
-      "loss": 0.8637,
+      "loss": 0.9241,
       "step": 26229
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1114902223891e-07,
-      "loss": 0.754,
+      "loss": 0.8093,
       "step": 26230
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.109739642990093e-07,
-      "loss": 0.7853,
+      "loss": 0.7816,
       "step": 26231
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1079904355531523e-07,
-      "loss": 0.7655,
+      "loss": 0.8014,
       "step": 26232
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1062426001025361e-07,
-      "loss": 0.7002,
+      "loss": 0.7175,
       "step": 26233
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1044961366624918e-07,
-      "loss": 0.7567,
+      "loss": 0.7916,
       "step": 26234
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1027510452572331e-07,
-      "loss": 0.7004,
+      "loss": 0.7882,
       "step": 26235
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1010073259109966e-07,
-      "loss": 0.9556,
+      "loss": 0.937,
       "step": 26236
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0992649786479736e-07,
-      "loss": 0.721,
+      "loss": 0.7411,
       "step": 26237
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0975240034923229e-07,
-      "loss": 0.6892,
+      "loss": 0.7471,
       "step": 26238
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0957844004681916e-07,
-      "loss": 0.6918,
+      "loss": 0.7195,
       "step": 26239
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0940461695997385e-07,
-      "loss": 0.7934,
+      "loss": 0.8073,
       "step": 26240
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.092309310911055e-07,
-      "loss": 0.7685,
+      "loss": 0.8077,
       "step": 26241
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0905738244262554e-07,
-      "loss": 0.7169,
+      "loss": 0.7847,
       "step": 26242
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0888397101694092e-07,
-      "loss": 0.7186,
+      "loss": 0.7369,
       "step": 26243
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0871069681645752e-07,
-      "loss": 0.7455,
+      "loss": 0.7896,
       "step": 26244
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0853755984357895e-07,
-      "loss": 0.6891,
+      "loss": 0.7354,
       "step": 26245
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0836456010070773e-07,
-      "loss": 0.6967,
+      "loss": 0.6922,
       "step": 26246
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0819169759024417e-07,
-      "loss": 0.6469,
+      "loss": 0.7547,
       "step": 26247
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0801897231458524e-07,
-      "loss": 0.7527,
+      "loss": 0.8028,
       "step": 26248
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0784638427612904e-07,
-      "loss": 0.7668,
+      "loss": 0.7829,
       "step": 26249
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0767393347726918e-07,
-      "loss": 0.6608,
+      "loss": 0.7494,
       "step": 26250
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0750161992039709e-07,
-      "loss": 0.7085,
+      "loss": 0.7827,
       "step": 26251
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0732944360790643e-07,
-      "loss": 0.7403,
+      "loss": 0.7713,
       "step": 26252
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0715740454218304e-07,
-      "loss": 0.7205,
+      "loss": 0.7957,
       "step": 26253
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0698550272561281e-07,
-      "loss": 0.7098,
+      "loss": 0.7636,
       "step": 26254
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0681373816058382e-07,
-      "loss": 0.7083,
+      "loss": 0.7515,
       "step": 26255
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0664211084947751e-07,
-      "loss": 0.7716,
+      "loss": 0.8615,
       "step": 26256
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.064706207946753e-07,
-      "loss": 0.8194,
+      "loss": 0.8843,
       "step": 26257
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0629926799855528e-07,
-      "loss": 0.708,
+      "loss": 0.707,
       "step": 26258
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.0612805246349778e-07,
-      "loss": 0.7502,
+      "loss": 0.8277,
       "step": 26259
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0595697419187312e-07,
-      "loss": 0.7319,
+      "loss": 0.7422,
       "step": 26260
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.057860331860594e-07,
-      "loss": 0.7396,
+      "loss": 0.8231,
       "step": 26261
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0561522944842584e-07,
-      "loss": 0.8623,
+      "loss": 0.8949,
       "step": 26262
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0544456298134275e-07,
-      "loss": 0.8059,
+      "loss": 0.8167,
       "step": 26263
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0527403378717716e-07,
-      "loss": 0.6915,
+      "loss": 0.7426,
       "step": 26264
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0510364186829713e-07,
-      "loss": 0.7818,
+      "loss": 0.7275,
       "step": 26265
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0493338722706414e-07,
-      "loss": 0.8641,
+      "loss": 0.8666,
       "step": 26266
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0476326986584074e-07,
-      "loss": 0.8319,
+      "loss": 0.8734,
       "step": 26267
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0459328978698835e-07,
-      "loss": 0.8704,
+      "loss": 0.8091,
       "step": 26268
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.04423446992864e-07,
-      "loss": 0.7215,
+      "loss": 0.7277,
       "step": 26269
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0425374148582357e-07,
-      "loss": 0.6813,
+      "loss": 0.8188,
       "step": 26270
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0408417326822295e-07,
-      "loss": 0.7936,
+      "loss": 0.7158,
       "step": 26271
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0391474234241471e-07,
-      "loss": 0.7543,
+      "loss": 0.7239,
       "step": 26272
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.037454487107481e-07,
-      "loss": 0.737,
+      "loss": 0.7968,
       "step": 26273
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0357629237557232e-07,
-      "loss": 0.7614,
+      "loss": 0.7744,
       "step": 26274
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0340727333923329e-07,
-      "loss": 0.8199,
+      "loss": 0.7605,
       "step": 26275
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.03238391604078e-07,
-      "loss": 0.7607,
+      "loss": 0.8194,
       "step": 26276
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0306964717244794e-07,
-      "loss": 0.6777,
+      "loss": 0.8212,
       "step": 26277
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0290104004668567e-07,
-      "loss": 0.8178,
+      "loss": 0.8821,
       "step": 26278
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.027325702291282e-07,
-      "loss": 0.5855,
+      "loss": 0.7195,
       "step": 26279
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0256423772211477e-07,
-      "loss": 0.7879,
+      "loss": 0.7942,
       "step": 26280
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0239604252797908e-07,
-      "loss": 0.6779,
+      "loss": 0.7108,
       "step": 26281
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0222798464905593e-07,
-      "loss": 0.7092,
+      "loss": 0.7186,
       "step": 26282
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0206006408767566e-07,
-      "loss": 0.7239,
+      "loss": 0.78,
       "step": 26283
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0189228084616976e-07,
-      "loss": 0.7736,
+      "loss": 0.8857,
       "step": 26284
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0172463492686412e-07,
-      "loss": 0.6741,
+      "loss": 0.7352,
       "step": 26285
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0155712633208581e-07,
-      "loss": 0.7671,
+      "loss": 0.8482,
       "step": 26286
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0138975506415849e-07,
-      "loss": 0.6221,
+      "loss": 0.6973,
       "step": 26287
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0122252112540365e-07,
-      "loss": 0.7686,
+      "loss": 0.824,
       "step": 26288
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0105542451814054e-07,
-      "loss": 0.7181,
+      "loss": 0.739,
       "step": 26289
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0088846524469064e-07,
-      "loss": 0.6956,
+      "loss": 0.816,
       "step": 26290
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0072164330736767e-07,
-      "loss": 0.6916,
+      "loss": 0.7737,
       "step": 26291
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0055495870848641e-07,
-      "loss": 0.9263,
+      "loss": 0.9132,
       "step": 26292
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0038841145035948e-07,
-      "loss": 0.7417,
+      "loss": 0.7427,
       "step": 26293
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0022200153529838e-07,
-      "loss": 0.7287,
+      "loss": 0.7512,
       "step": 26294
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.0005572896561011e-07,
-      "loss": 0.8682,
+      "loss": 0.8434,
       "step": 26295
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.988959374360285e-08,
-      "loss": 0.7243,
+      "loss": 0.764,
       "step": 26296
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.972359587158142e-08,
-      "loss": 0.8847,
+      "loss": 0.8238,
       "step": 26297
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.955773535184732e-08,
-      "loss": 0.8133,
+      "loss": 0.8737,
       "step": 26298
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.939201218670424e-08,
-      "loss": 0.6658,
+      "loss": 0.6926,
       "step": 26299
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.922642637844925e-08,
-      "loss": 0.6852,
+      "loss": 0.7538,
       "step": 26300
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.906097792938052e-08,
-      "loss": 0.6817,
+      "loss": 0.7328,
       "step": 26301
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.889566684179175e-08,
-      "loss": 0.6181,
+      "loss": 0.7206,
       "step": 26302
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.873049311797889e-08,
-      "loss": 0.7653,
+      "loss": 0.7603,
       "step": 26303
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.856545676023233e-08,
-      "loss": 0.7505,
+      "loss": 0.8244,
       "step": 26304
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.840055777084246e-08,
-      "loss": 0.6528,
+      "loss": 0.7449,
       "step": 26305
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.823579615209522e-08,
-      "loss": 0.7759,
+      "loss": 0.7824,
       "step": 26306
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.807117190627769e-08,
-      "loss": 0.8224,
+      "loss": 0.8532,
       "step": 26307
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.790668503567357e-08,
-      "loss": 0.8226,
+      "loss": 0.7755,
       "step": 26308
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.774233554256551e-08,
-      "loss": 0.7454,
+      "loss": 0.7881,
       "step": 26309
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.757812342923278e-08,
-      "loss": 0.7284,
+      "loss": 0.8031,
       "step": 26310
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.741404869795356e-08,
-      "loss": 0.7124,
+      "loss": 0.7629,
       "step": 26311
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.725011135100493e-08,
-      "loss": 0.7355,
+      "loss": 0.7905,
       "step": 26312
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.70863113906606e-08,
-      "loss": 0.8081,
+      "loss": 0.7778,
       "step": 26313
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.692264881919433e-08,
-      "loss": 0.801,
+      "loss": 0.8173,
       "step": 26314
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.675912363887429e-08,
-      "loss": 0.6817,
+      "loss": 0.7627,
       "step": 26315
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.65957358519709e-08,
-      "loss": 0.7163,
+      "loss": 0.8,
       "step": 26316
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.643248546075123e-08,
-      "loss": 0.7247,
+      "loss": 0.7557,
       "step": 26317
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.626937246747903e-08,
-      "loss": 0.7719,
+      "loss": 0.843,
       "step": 26318
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.610639687441692e-08,
-      "loss": 0.7298,
+      "loss": 0.8512,
       "step": 26319
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.594355868382865e-08,
-      "loss": 0.7542,
+      "loss": 0.7274,
       "step": 26320
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.578085789796909e-08,
-      "loss": 0.755,
+      "loss": 0.8323,
       "step": 26321
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.561829451909976e-08,
-      "loss": 0.7655,
+      "loss": 0.8527,
       "step": 26322
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.54558685494733e-08,
-      "loss": 0.6699,
+      "loss": 0.7754,
       "step": 26323
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.529357999134459e-08,
-      "loss": 0.7625,
+      "loss": 0.8011,
       "step": 26324
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.513142884696292e-08,
-      "loss": 0.839,
+      "loss": 0.8263,
       "step": 26325
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.496941511858204e-08,
-      "loss": 0.9095,
+      "loss": 0.8139,
       "step": 26326
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.480753880844462e-08,
-      "loss": 0.742,
+      "loss": 0.726,
       "step": 26327
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.464579991880107e-08,
-      "loss": 0.8341,
+      "loss": 0.8876,
       "step": 26328
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.448419845189294e-08,
-      "loss": 0.7865,
+      "loss": 0.7255,
       "step": 26329
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.432273440996064e-08,
-      "loss": 0.7416,
+      "loss": 0.7604,
       "step": 26330
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.416140779524796e-08,
-      "loss": 0.7293,
+      "loss": 0.8015,
       "step": 26331
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.400021860998976e-08,
-      "loss": 0.7293,
+      "loss": 0.7333,
       "step": 26332
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.383916685642536e-08,
-      "loss": 0.6956,
+      "loss": 0.8274,
       "step": 26333
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.36782525367852e-08,
-      "loss": 0.8548,
+      "loss": 0.8804,
       "step": 26334
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.351747565330638e-08,
-      "loss": 0.7671,
+      "loss": 0.7626,
       "step": 26335
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.3356836208216e-08,
-      "loss": 0.9359,
+      "loss": 0.9269,
       "step": 26336
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.319633420374342e-08,
-      "loss": 0.6648,
+      "loss": 0.7859,
       "step": 26337
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.303596964211681e-08,
-      "loss": 0.7956,
+      "loss": 0.8803,
       "step": 26338
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.287574252555997e-08,
-      "loss": 0.6455,
+      "loss": 0.6929,
       "step": 26339
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.271565285629447e-08,
-      "loss": 0.61,
+      "loss": 0.7544,
       "step": 26340
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.255570063654406e-08,
-      "loss": 0.5713,
+      "loss": 0.7532,
       "step": 26341
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.239588586852587e-08,
-      "loss": 0.7462,
+      "loss": 0.7255,
       "step": 26342
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.223620855445814e-08,
-      "loss": 0.6401,
+      "loss": 0.7321,
       "step": 26343
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.207666869655463e-08,
-      "loss": 0.6918,
+      "loss": 0.7799,
       "step": 26344
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.191726629703135e-08,
-      "loss": 0.652,
+      "loss": 0.7307,
       "step": 26345
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.175800135809765e-08,
-      "loss": 0.7823,
+      "loss": 0.7946,
       "step": 26346
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.159887388196398e-08,
-      "loss": 0.7489,
+      "loss": 0.7402,
       "step": 26347
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.143988387083746e-08,
-      "loss": 0.6996,
+      "loss": 0.7687,
       "step": 26348
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.128103132692412e-08,
-      "loss": 0.7139,
+      "loss": 0.768,
       "step": 26349
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.112231625242774e-08,
-      "loss": 0.759,
+      "loss": 0.7332,
       "step": 26350
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.0963738649551e-08,
-      "loss": 0.6855,
+      "loss": 0.7497,
       "step": 26351
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.080529852049324e-08,
-      "loss": 0.7099,
+      "loss": 0.8622,
       "step": 26352
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.064699586745163e-08,
-      "loss": 0.7594,
+      "loss": 0.8412,
       "step": 26353
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.048883069262326e-08,
-      "loss": 0.7595,
+      "loss": 0.7943,
       "step": 26354
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.033080299820419e-08,
-      "loss": 0.7999,
+      "loss": 0.7958,
       "step": 26355
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.017291278638374e-08,
-      "loss": 0.6246,
+      "loss": 0.7221,
       "step": 26356
     },
     {
       "epoch": 0.96,
       "learning_rate": 9.001516005935462e-08,
-      "loss": 0.71,
+      "loss": 0.7956,
       "step": 26357
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.985754481930508e-08,
-      "loss": 0.8463,
+      "loss": 0.8344,
       "step": 26358
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.970006706842116e-08,
-      "loss": 0.7132,
+      "loss": 0.7411,
       "step": 26359
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.954272680888775e-08,
-      "loss": 0.7827,
+      "loss": 0.7885,
       "step": 26360
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.938552404288758e-08,
-      "loss": 0.7732,
+      "loss": 0.8068,
       "step": 26361
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.922845877260222e-08,
-      "loss": 0.8165,
+      "loss": 0.8545,
       "step": 26362
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.907153100021104e-08,
-      "loss": 0.7713,
+      "loss": 0.754,
       "step": 26363
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.891474072789008e-08,
-      "loss": 0.7368,
+      "loss": 0.8224,
       "step": 26364
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.875808795781537e-08,
-      "loss": 0.6572,
+      "loss": 0.7586,
       "step": 26365
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.860157269215963e-08,
-      "loss": 0.7561,
+      "loss": 0.7898,
       "step": 26366
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.844519493309444e-08,
-      "loss": 0.8154,
+      "loss": 0.8787,
       "step": 26367
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.82889546827903e-08,
-      "loss": 0.6018,
+      "loss": 0.6924,
       "step": 26368
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.813285194341325e-08,
-      "loss": 0.6908,
+      "loss": 0.7479,
       "step": 26369
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.797688671712934e-08,
-      "loss": 0.7027,
+      "loss": 0.8179,
       "step": 26370
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.78210590061046e-08,
-      "loss": 0.6938,
+      "loss": 0.8067,
       "step": 26371
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.766536881249732e-08,
-      "loss": 0.9102,
+      "loss": 0.9077,
       "step": 26372
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.750981613847021e-08,
-      "loss": 0.8048,
+      "loss": 0.8183,
       "step": 26373
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.735440098618042e-08,
-      "loss": 0.7161,
+      "loss": 0.7465,
       "step": 26374
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.719912335778403e-08,
-      "loss": 0.7707,
+      "loss": 0.8075,
       "step": 26375
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.704398325543484e-08,
-      "loss": 0.6393,
+      "loss": 0.749,
       "step": 26376
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.68889806812867e-08,
-      "loss": 0.7461,
+      "loss": 0.8058,
       "step": 26377
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.673411563748901e-08,
-      "loss": 0.7711,
+      "loss": 0.8112,
       "step": 26378
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.657938812619004e-08,
-      "loss": 0.7802,
+      "loss": 0.8562,
       "step": 26379
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.642479814953586e-08,
-      "loss": 0.7448,
+      "loss": 0.8424,
       "step": 26380
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.627034570967364e-08,
-      "loss": 0.6102,
+      "loss": 0.5948,
       "step": 26381
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.611603080874276e-08,
-      "loss": 0.8083,
+      "loss": 0.8112,
       "step": 26382
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.596185344888819e-08,
-      "loss": 0.6637,
+      "loss": 0.6927,
       "step": 26383
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.580781363224488e-08,
-      "loss": 0.7786,
+      "loss": 0.7253,
       "step": 26384
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.565391136095335e-08,
-      "loss": 0.7763,
+      "loss": 0.7782,
       "step": 26385
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.550014663714634e-08,
-      "loss": 0.8457,
+      "loss": 0.8415,
       "step": 26386
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.53465194629577e-08,
-      "loss": 0.7416,
+      "loss": 0.8298,
       "step": 26387
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.519302984051903e-08,
-      "loss": 0.7427,
+      "loss": 0.7939,
       "step": 26388
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.503967777195976e-08,
-      "loss": 0.7099,
+      "loss": 0.8012,
       "step": 26389
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.48864632594093e-08,
-      "loss": 0.6646,
+      "loss": 0.7602,
       "step": 26390
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.473338630498928e-08,
-      "loss": 0.8081,
+      "loss": 0.7815,
       "step": 26391
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.458044691082579e-08,
-      "loss": 0.6626,
+      "loss": 0.76,
       "step": 26392
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.442764507904156e-08,
-      "loss": 0.6901,
+      "loss": 0.8148,
       "step": 26393
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.42749808117549e-08,
-      "loss": 0.7412,
+      "loss": 0.8192,
       "step": 26394
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.4122454111083e-08,
-      "loss": 0.7648,
+      "loss": 0.8691,
       "step": 26395
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.397006497914528e-08,
-      "loss": 0.7653,
+      "loss": 0.75,
       "step": 26396
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.381781341805339e-08,
-      "loss": 0.6706,
+      "loss": 0.6916,
       "step": 26397
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.366569942991898e-08,
-      "loss": 0.7269,
+      "loss": 0.7432,
       "step": 26398
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.351372301685478e-08,
-      "loss": 0.734,
+      "loss": 0.8263,
       "step": 26399
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.336188418096803e-08,
-      "loss": 0.7189,
+      "loss": 0.7654,
       "step": 26400
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.321018292436478e-08,
-      "loss": 0.8592,
+      "loss": 0.8317,
       "step": 26401
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.305861924915004e-08,
-      "loss": 0.8288,
+      "loss": 0.8531,
       "step": 26402
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.290719315742657e-08,
-      "loss": 0.7409,
+      "loss": 0.7193,
       "step": 26403
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.275590465129602e-08,
-      "loss": 0.7221,
+      "loss": 0.861,
       "step": 26404
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.260475373285559e-08,
-      "loss": 0.636,
+      "loss": 0.7484,
       "step": 26405
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.245374040420473e-08,
-      "loss": 0.7393,
+      "loss": 0.8636,
       "step": 26406
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.23028646674362e-08,
-      "loss": 0.7176,
+      "loss": 0.7273,
       "step": 26407
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.215212652464277e-08,
-      "loss": 0.8235,
+      "loss": 0.7691,
       "step": 26408
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.200152597791944e-08,
-      "loss": 0.6814,
+      "loss": 0.803,
       "step": 26409
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.18510630293512e-08,
-      "loss": 0.6747,
+      "loss": 0.7155,
       "step": 26410
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.17007376810286e-08,
-      "loss": 0.6972,
+      "loss": 0.7293,
       "step": 26411
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.155054993503553e-08,
-      "loss": 0.7262,
+      "loss": 0.8666,
       "step": 26412
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.140049979345587e-08,
-      "loss": 0.6608,
+      "loss": 0.7552,
       "step": 26413
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.125058725837132e-08,
-      "loss": 0.7233,
+      "loss": 0.7923,
       "step": 26414
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.110081233186351e-08,
-      "loss": 0.7063,
+      "loss": 0.7281,
       "step": 26415
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.095117501600746e-08,
-      "loss": 0.7363,
+      "loss": 0.7192,
       "step": 26416
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.080167531288153e-08,
-      "loss": 0.7236,
+      "loss": 0.7642,
       "step": 26417
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.065231322455846e-08,
-      "loss": 0.6875,
+      "loss": 0.7293,
       "step": 26418
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.050308875311108e-08,
-      "loss": 0.8225,
+      "loss": 0.9014,
       "step": 26419
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.035400190060771e-08,
-      "loss": 0.7787,
+      "loss": 0.8158,
       "step": 26420
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.020505266912004e-08,
-      "loss": 0.8044,
+      "loss": 0.7904,
       "step": 26421
     },
     {
       "epoch": 0.96,
       "learning_rate": 8.005624106071308e-08,
-      "loss": 0.7881,
+      "loss": 0.7716,
       "step": 26422
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.990756707745073e-08,
-      "loss": 0.8219,
+      "loss": 0.861,
       "step": 26423
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.97590307213969e-08,
-      "loss": 0.7434,
+      "loss": 0.7581,
       "step": 26424
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.961063199461106e-08,
-      "loss": 0.8507,
+      "loss": 0.8395,
       "step": 26425
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.946237089915154e-08,
-      "loss": 0.7114,
+      "loss": 0.8104,
       "step": 26426
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.931424743707672e-08,
-      "loss": 0.7502,
+      "loss": 0.7614,
       "step": 26427
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.916626161044161e-08,
-      "loss": 0.7655,
+      "loss": 0.8911,
       "step": 26428
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.901841342129901e-08,
-      "loss": 0.6725,
+      "loss": 0.7008,
       "step": 26429
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.887070287169951e-08,
-      "loss": 0.7153,
+      "loss": 0.8116,
       "step": 26430
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.872312996369259e-08,
-      "loss": 0.7263,
+      "loss": 0.7568,
       "step": 26431
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.857569469932547e-08,
-      "loss": 0.7799,
+      "loss": 0.8703,
       "step": 26432
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.842839708064432e-08,
-      "loss": 0.8181,
+      "loss": 0.8036,
       "step": 26433
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.828123710969193e-08,
-      "loss": 0.6159,
+      "loss": 0.7663,
       "step": 26434
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.813421478851114e-08,
-      "loss": 0.7307,
+      "loss": 0.7789,
       "step": 26435
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.798733011914028e-08,
-      "loss": 0.7884,
+      "loss": 0.87,
       "step": 26436
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.784058310361775e-08,
-      "loss": 0.7546,
+      "loss": 0.8043,
       "step": 26437
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.769397374397969e-08,
-      "loss": 0.8431,
+      "loss": 0.8621,
       "step": 26438
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.754750204226003e-08,
-      "loss": 0.6524,
+      "loss": 0.716,
       "step": 26439
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.740116800049047e-08,
-      "loss": 0.7625,
+      "loss": 0.8015,
       "step": 26440
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.72549716207005e-08,
-      "loss": 0.7711,
+      "loss": 0.7891,
       "step": 26441
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.710891290492073e-08,
-      "loss": 0.8211,
+      "loss": 0.778,
       "step": 26442
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.696299185517397e-08,
-      "loss": 0.8174,
+      "loss": 0.8249,
       "step": 26443
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.681720847348862e-08,
-      "loss": 0.7426,
+      "loss": 0.8088,
       "step": 26444
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.667156276188525e-08,
-      "loss": 0.8081,
+      "loss": 0.8357,
       "step": 26445
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.65260547223834e-08,
-      "loss": 0.6954,
+      "loss": 0.7093,
       "step": 26446
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.638068435700363e-08,
-      "loss": 0.588,
+      "loss": 0.6888,
       "step": 26447
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.623545166776324e-08,
-      "loss": 0.6698,
+      "loss": 0.7551,
       "step": 26448
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.609035665667397e-08,
-      "loss": 0.7831,
+      "loss": 0.7921,
       "step": 26449
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.594539932575196e-08,
-      "loss": 0.7915,
+      "loss": 0.8829,
       "step": 26450
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.580057967700782e-08,
-      "loss": 0.734,
+      "loss": 0.7674,
       "step": 26451
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.565589771244775e-08,
-      "loss": 0.7549,
+      "loss": 0.8011,
       "step": 26452
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.551135343408345e-08,
-      "loss": 0.6658,
+      "loss": 0.6694,
       "step": 26453
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.536694684391776e-08,
-      "loss": 0.8111,
+      "loss": 0.7592,
       "step": 26454
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.522267794395355e-08,
-      "loss": 0.7293,
+      "loss": 0.7993,
       "step": 26455
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.507854673619363e-08,
-      "loss": 0.7392,
+      "loss": 0.7748,
       "step": 26456
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.493455322263754e-08,
-      "loss": 0.7401,
+      "loss": 0.7423,
       "step": 26457
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.479069740528255e-08,
-      "loss": 0.858,
+      "loss": 0.911,
       "step": 26458
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.464697928612485e-08,
-      "loss": 0.7615,
+      "loss": 0.8341,
       "step": 26459
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.450339886715729e-08,
-      "loss": 0.6522,
+      "loss": 0.7535,
       "step": 26460
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.435995615037384e-08,
-      "loss": 0.6899,
+      "loss": 0.7535,
       "step": 26461
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.421665113776177e-08,
-      "loss": 0.8507,
+      "loss": 0.8504,
       "step": 26462
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.407348383131286e-08,
-      "loss": 0.8391,
+      "loss": 0.7872,
       "step": 26463
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.393045423300993e-08,
-      "loss": 0.7446,
+      "loss": 0.8057,
       "step": 26464
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.378756234483808e-08,
-      "loss": 0.7758,
+      "loss": 0.7703,
       "step": 26465
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.364480816878239e-08,
-      "loss": 0.6612,
+      "loss": 0.7638,
       "step": 26466
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.350219170682015e-08,
-      "loss": 0.7546,
+      "loss": 0.7842,
       "step": 26467
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.335971296093092e-08,
-      "loss": 0.7314,
+      "loss": 0.7643,
       "step": 26468
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.321737193309087e-08,
-      "loss": 0.7134,
+      "loss": 0.8215,
       "step": 26469
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.30751686252773e-08,
-      "loss": 0.7096,
+      "loss": 0.7423,
       "step": 26470
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.293310303946088e-08,
-      "loss": 0.7594,
+      "loss": 0.6986,
       "step": 26471
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.279117517761225e-08,
-      "loss": 0.6681,
+      "loss": 0.7928,
       "step": 26472
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.264938504170205e-08,
-      "loss": 0.6729,
+      "loss": 0.7721,
       "step": 26473
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.250773263369537e-08,
-      "loss": 0.7246,
+      "loss": 0.7885,
       "step": 26474
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.236621795555954e-08,
-      "loss": 0.6766,
+      "loss": 0.8279,
       "step": 26475
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.222484100925741e-08,
-      "loss": 0.7606,
+      "loss": 0.8167,
       "step": 26476
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.208360179674855e-08,
-      "loss": 0.8597,
+      "loss": 0.8323,
       "step": 26477
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.194250031999472e-08,
-      "loss": 0.7341,
+      "loss": 0.7688,
       "step": 26478
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.180153658095213e-08,
-      "loss": 0.6877,
+      "loss": 0.6858,
       "step": 26479
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.166071058157809e-08,
-      "loss": 0.7612,
+      "loss": 0.7936,
       "step": 26480
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.152002232382327e-08,
-      "loss": 0.712,
+      "loss": 0.8002,
       "step": 26481
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.137947180964278e-08,
-      "loss": 0.758,
+      "loss": 0.7853,
       "step": 26482
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.123905904098616e-08,
-      "loss": 0.7838,
+      "loss": 0.869,
       "step": 26483
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.109878401979853e-08,
-      "loss": 0.7378,
+      "loss": 0.7333,
       "step": 26484
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.095864674802944e-08,
-      "loss": 0.7719,
+      "loss": 0.8557,
       "step": 26485
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.081864722762177e-08,
-      "loss": 0.7585,
+      "loss": 0.8785,
       "step": 26486
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.067878546051732e-08,
-      "loss": 0.6778,
+      "loss": 0.757,
       "step": 26487
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.053906144865563e-08,
-      "loss": 0.8338,
+      "loss": 0.838,
       "step": 26488
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.03994751939785e-08,
-      "loss": 0.7852,
+      "loss": 0.8534,
       "step": 26489
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.02600266984188e-08,
-      "loss": 0.746,
+      "loss": 0.7893,
       "step": 26490
     },
     {
       "epoch": 0.96,
       "learning_rate": 7.01207159639128e-08,
-      "loss": 0.6646,
+      "loss": 0.7733,
       "step": 26491
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.998154299239446e-08,
-      "loss": 0.689,
+      "loss": 0.7498,
       "step": 26492
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.984250778579226e-08,
-      "loss": 0.7761,
+      "loss": 0.7882,
       "step": 26493
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.970361034603579e-08,
-      "loss": 0.7058,
+      "loss": 0.7627,
       "step": 26494
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.956485067505458e-08,
-      "loss": 0.8648,
+      "loss": 0.8159,
       "step": 26495
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.942622877476935e-08,
-      "loss": 0.8268,
+      "loss": 0.8089,
       "step": 26496
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.928774464710631e-08,
-      "loss": 0.6845,
+      "loss": 0.7685,
       "step": 26497
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.914939829398615e-08,
-      "loss": 0.8134,
+      "loss": 0.8731,
       "step": 26498
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.901118971732845e-08,
-      "loss": 0.7277,
+      "loss": 0.7465,
       "step": 26499
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.887311891904946e-08,
-      "loss": 0.7044,
+      "loss": 0.7508,
       "step": 26500
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.873518590106543e-08,
-      "loss": 0.6884,
+      "loss": 0.7779,
       "step": 26501
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.859739066529148e-08,
-      "loss": 0.864,
+      "loss": 0.8518,
       "step": 26502
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.84597332136372e-08,
-      "loss": 0.8229,
+      "loss": 0.841,
       "step": 26503
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.832221354801216e-08,
-      "loss": 0.6304,
+      "loss": 0.7449,
       "step": 26504
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.818483167032597e-08,
-      "loss": 0.7636,
+      "loss": 0.7544,
       "step": 26505
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.804758758248486e-08,
-      "loss": 0.6785,
+      "loss": 0.7518,
       "step": 26506
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.791048128638956e-08,
-      "loss": 0.8049,
+      "loss": 0.7647,
       "step": 26507
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.777351278394628e-08,
-      "loss": 0.6306,
+      "loss": 0.7005,
       "step": 26508
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.763668207705354e-08,
-      "loss": 0.7566,
+      "loss": 0.7956,
       "step": 26509
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.74999891676087e-08,
-      "loss": 0.7043,
+      "loss": 0.754,
       "step": 26510
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.736343405751022e-08,
-      "loss": 0.742,
+      "loss": 0.8133,
       "step": 26511
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.722701674865107e-08,
-      "loss": 0.7929,
+      "loss": 0.8295,
       "step": 26512
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.709073724292303e-08,
-      "loss": 0.8139,
+      "loss": 0.7879,
       "step": 26513
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.695459554221906e-08,
-      "loss": 0.7303,
+      "loss": 0.7717,
       "step": 26514
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.681859164842763e-08,
-      "loss": 0.8183,
+      "loss": 0.839,
       "step": 26515
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.668272556343391e-08,
-      "loss": 0.8742,
+      "loss": 0.8686,
       "step": 26516
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.654699728912417e-08,
-      "loss": 0.6689,
+      "loss": 0.7427,
       "step": 26517
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.641140682738135e-08,
-      "loss": 0.661,
+      "loss": 0.7884,
       "step": 26518
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.627595418008615e-08,
-      "loss": 0.7496,
+      "loss": 0.7891,
       "step": 26519
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.614063934911818e-08,
-      "loss": 0.8395,
+      "loss": 0.7501,
       "step": 26520
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.600546233635485e-08,
-      "loss": 0.667,
+      "loss": 0.7798,
       "step": 26521
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.58704231436702e-08,
-      "loss": 0.6624,
+      "loss": 0.7085,
       "step": 26522
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.57355217729394e-08,
-      "loss": 0.7271,
+      "loss": 0.7387,
       "step": 26523
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.560075822603318e-08,
-      "loss": 0.8155,
+      "loss": 0.8317,
       "step": 26524
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.546613250482225e-08,
-      "loss": 0.7386,
+      "loss": 0.7431,
       "step": 26525
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.53316446111718e-08,
-      "loss": 0.811,
+      "loss": 0.8491,
       "step": 26526
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.519729454695034e-08,
-      "loss": 0.8265,
+      "loss": 0.8187,
       "step": 26527
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.506308231402081e-08,
-      "loss": 0.7474,
+      "loss": 0.8116,
       "step": 26528
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.492900791424506e-08,
-      "loss": 0.675,
+      "loss": 0.7547,
       "step": 26529
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.479507134948271e-08,
-      "loss": 0.7345,
+      "loss": 0.8563,
       "step": 26530
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.466127262159339e-08,
-      "loss": 0.7879,
+      "loss": 0.78,
       "step": 26531
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.452761173243227e-08,
-      "loss": 0.848,
+      "loss": 0.8462,
       "step": 26532
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.439408868385344e-08,
-      "loss": 0.7762,
+      "loss": 0.8427,
       "step": 26533
     },
     {
       "epoch": 0.96,
       "learning_rate": 6.426070347770986e-08,
-      "loss": 0.7235,
+      "loss": 0.7413,
       "step": 26534
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.412745611585336e-08,
-      "loss": 0.7182,
+      "loss": 0.7137,
       "step": 26535
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.399434660012915e-08,
-      "loss": 0.7973,
+      "loss": 0.8812,
       "step": 26536
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.386137493238798e-08,
-      "loss": 0.729,
+      "loss": 0.7927,
       "step": 26537
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.372854111447169e-08,
-      "loss": 0.6446,
+      "loss": 0.6855,
       "step": 26538
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.359584514822436e-08,
-      "loss": 0.9102,
+      "loss": 0.8653,
       "step": 26539
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.346328703548787e-08,
-      "loss": 0.7158,
+      "loss": 0.748,
       "step": 26540
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.333086677809852e-08,
-      "loss": 0.7069,
+      "loss": 0.7802,
       "step": 26541
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.319858437789595e-08,
-      "loss": 0.749,
+      "loss": 0.7851,
       "step": 26542
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.306643983671423e-08,
-      "loss": 0.6969,
+      "loss": 0.7971,
       "step": 26543
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.293443315638748e-08,
-      "loss": 0.7158,
+      "loss": 0.7913,
       "step": 26544
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.280256433874531e-08,
-      "loss": 0.8623,
+      "loss": 0.8392,
       "step": 26545
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.267083338561963e-08,
-      "loss": 0.662,
+      "loss": 0.6689,
       "step": 26546
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.253924029883673e-08,
-      "loss": 0.8186,
+      "loss": 0.8541,
       "step": 26547
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.240778508022294e-08,
-      "loss": 0.6658,
+      "loss": 0.7982,
       "step": 26548
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.227646773160013e-08,
-      "loss": 0.7234,
+      "loss": 0.7488,
       "step": 26549
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.214528825479239e-08,
-      "loss": 0.6994,
+      "loss": 0.7738,
       "step": 26550
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.201424665161937e-08,
-      "loss": 0.7991,
+      "loss": 0.7959,
       "step": 26551
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.188334292389741e-08,
-      "loss": 0.6837,
+      "loss": 0.6839,
       "step": 26552
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.175257707344506e-08,
-      "loss": 0.6665,
+      "loss": 0.7827,
       "step": 26553
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.16219491020742e-08,
-      "loss": 0.6964,
+      "loss": 0.7659,
       "step": 26554
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.149145901159782e-08,
-      "loss": 0.7755,
+      "loss": 0.7889,
       "step": 26555
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.136110680382779e-08,
-      "loss": 0.7495,
+      "loss": 0.8034,
       "step": 26556
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.123089248056935e-08,
-      "loss": 0.7392,
+      "loss": 0.7875,
       "step": 26557
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.110081604363327e-08,
-      "loss": 0.7296,
+      "loss": 0.838,
       "step": 26558
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.097087749482033e-08,
-      "loss": 0.698,
+      "loss": 0.6558,
       "step": 26559
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.084107683593576e-08,
-      "loss": 0.7261,
+      "loss": 0.8107,
       "step": 26560
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.071141406877923e-08,
-      "loss": 0.728,
+      "loss": 0.7908,
       "step": 26561
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.058188919514928e-08,
-      "loss": 0.7353,
+      "loss": 0.7472,
       "step": 26562
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.04525022168434e-08,
-      "loss": 0.7761,
+      "loss": 0.7418,
       "step": 26563
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.03232531356568e-08,
-      "loss": 0.6895,
+      "loss": 0.7425,
       "step": 26564
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.019414195338358e-08,
-      "loss": 0.7763,
+      "loss": 0.7808,
       "step": 26565
     },
     {
       "epoch": 0.97,
       "learning_rate": 6.006516867181234e-08,
-      "loss": 0.8273,
+      "loss": 0.8359,
       "step": 26566
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.993633329273496e-08,
-      "loss": 0.7265,
+      "loss": 0.7545,
       "step": 26567
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.98076358179378e-08,
-      "loss": 0.7759,
+      "loss": 0.7732,
       "step": 26568
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.96790762492061e-08,
-      "loss": 0.6797,
+      "loss": 0.6593,
       "step": 26569
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.955065458832399e-08,
-      "loss": 0.7596,
+      "loss": 0.8119,
       "step": 26570
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.942237083707225e-08,
-      "loss": 0.623,
+      "loss": 0.7039,
       "step": 26571
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.9294224997231695e-08,
-      "loss": 0.63,
+      "loss": 0.7589,
       "step": 26572
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.916621707057979e-08,
-      "loss": 0.7067,
+      "loss": 0.7861,
       "step": 26573
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.903834705889289e-08,
-      "loss": 0.6926,
+      "loss": 0.7171,
       "step": 26574
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.8910614963944015e-08,
-      "loss": 0.696,
+      "loss": 0.8146,
       "step": 26575
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.8783020787506196e-08,
-      "loss": 0.7377,
+      "loss": 0.7965,
       "step": 26576
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.865556453134802e-08,
-      "loss": 0.7065,
+      "loss": 0.7886,
       "step": 26577
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.852824619724029e-08,
-      "loss": 0.7396,
+      "loss": 0.7254,
       "step": 26578
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.840106578694826e-08,
-      "loss": 0.7353,
+      "loss": 0.8286,
       "step": 26579
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.8274023302234975e-08,
-      "loss": 0.7266,
+      "loss": 0.7541,
       "step": 26580
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.8147118744864563e-08,
-      "loss": 0.7117,
+      "loss": 0.7352,
       "step": 26581
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.802035211659784e-08,
-      "loss": 0.6716,
+      "loss": 0.7455,
       "step": 26582
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.789372341919231e-08,
-      "loss": 0.7894,
+      "loss": 0.8196,
       "step": 26583
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.7767232654405424e-08,
-      "loss": 0.6757,
+      "loss": 0.7242,
       "step": 26584
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.764087982399136e-08,
-      "loss": 0.9025,
+      "loss": 0.8986,
       "step": 26585
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.751466492970537e-08,
-      "loss": 0.6691,
+      "loss": 0.7594,
       "step": 26586
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.7388587973294943e-08,
-      "loss": 0.6613,
+      "loss": 0.7713,
       "step": 26587
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.726264895651312e-08,
-      "loss": 0.7317,
+      "loss": 0.7376,
       "step": 26588
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.7136847881102966e-08,
-      "loss": 0.7073,
+      "loss": 0.7386,
       "step": 26589
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.7011184748814176e-08,
-      "loss": 0.7869,
+      "loss": 0.8028,
       "step": 26590
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.6885659561386476e-08,
-      "loss": 0.8224,
+      "loss": 0.8242,
       "step": 26591
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.6760272320562915e-08,
-      "loss": 0.7976,
+      "loss": 0.8035,
       "step": 26592
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.663502302808321e-08,
-      "loss": 0.7379,
+      "loss": 0.7905,
       "step": 26593
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.650991168568487e-08,
-      "loss": 0.7503,
+      "loss": 0.7791,
       "step": 26594
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.638493829510427e-08,
-      "loss": 0.8191,
+      "loss": 0.8406,
       "step": 26595
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.6260102858073375e-08,
-      "loss": 0.6368,
+      "loss": 0.7285,
       "step": 26596
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.6135405376325224e-08,
-      "loss": 0.7253,
+      "loss": 0.7512,
       "step": 26597
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.6010845851590664e-08,
-      "loss": 0.778,
+      "loss": 0.8734,
       "step": 26598
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.5886424285596095e-08,
-      "loss": 0.7147,
+      "loss": 0.7859,
       "step": 26599
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.576214068006902e-08,
-      "loss": 0.7453,
+      "loss": 0.7941,
       "step": 26600
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.5637995036733615e-08,
-      "loss": 0.6689,
+      "loss": 0.746,
       "step": 26601
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.5513987357310727e-08,
-      "loss": 0.6768,
+      "loss": 0.7754,
       "step": 26602
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.539011764352231e-08,
-      "loss": 0.732,
+      "loss": 0.7545,
       "step": 26603
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.5266385897086996e-08,
-      "loss": 0.6698,
+      "loss": 0.8101,
       "step": 26604
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.514279211972118e-08,
-      "loss": 0.7402,
+      "loss": 0.7954,
       "step": 26605
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.501933631313794e-08,
-      "loss": 0.7386,
+      "loss": 0.7845,
       "step": 26606
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.489601847905257e-08,
-      "loss": 0.7267,
+      "loss": 0.7589,
       "step": 26607
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.4772838619173704e-08,
-      "loss": 0.5784,
+      "loss": 0.6959,
       "step": 26608
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.464979673521109e-08,
-      "loss": 0.766,
+      "loss": 0.7354,
       "step": 26609
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.4526892828872244e-08,
-      "loss": 0.6837,
+      "loss": 0.7697,
       "step": 26610
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.440412690186248e-08,
-      "loss": 0.836,
+      "loss": 0.8732,
       "step": 26611
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.428149895588375e-08,
-      "loss": 0.6204,
+      "loss": 0.7301,
       "step": 26612
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.415900899263915e-08,
-      "loss": 0.7488,
+      "loss": 0.8117,
       "step": 26613
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.403665701382621e-08,
-      "loss": 0.8812,
+      "loss": 0.8994,
       "step": 26614
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.391444302114357e-08,
-      "loss": 0.8558,
+      "loss": 0.8483,
       "step": 26615
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.379236701628654e-08,
-      "loss": 0.8585,
+      "loss": 0.7916,
       "step": 26616
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.367042900094821e-08,
-      "loss": 0.7251,
+      "loss": 0.8719,
       "step": 26617
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.354862897682056e-08,
-      "loss": 0.8315,
+      "loss": 0.736,
       "step": 26618
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.342696694559335e-08,
-      "loss": 0.8915,
+      "loss": 0.8927,
       "step": 26619
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.3305442908955226e-08,
-      "loss": 0.7748,
+      "loss": 0.8087,
       "step": 26620
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.3184056868590405e-08,
-      "loss": 0.7616,
+      "loss": 0.7654,
       "step": 26621
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.306280882618531e-08,
-      "loss": 0.7803,
+      "loss": 0.7962,
       "step": 26622
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.2941698783420814e-08,
-      "loss": 0.8576,
+      "loss": 0.8971,
       "step": 26623
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.2820726741976694e-08,
-      "loss": 0.727,
+      "loss": 0.7853,
       "step": 26624
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.26998927035316e-08,
-      "loss": 0.8264,
+      "loss": 0.7529,
       "step": 26625
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.257919666976308e-08,
-      "loss": 0.7487,
+      "loss": 0.8451,
       "step": 26626
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.2458638642344237e-08,
-      "loss": 0.6524,
+      "loss": 0.6504,
       "step": 26627
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.233821862294819e-08,
-      "loss": 0.6472,
+      "loss": 0.7849,
       "step": 26628
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.2217936613244705e-08,
-      "loss": 0.7863,
+      "loss": 0.82,
       "step": 26629
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.209779261490466e-08,
-      "loss": 0.7358,
+      "loss": 0.7136,
       "step": 26630
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.19777866295923e-08,
-      "loss": 0.7372,
+      "loss": 0.8325,
       "step": 26631
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.185791865897405e-08,
-      "loss": 0.7314,
+      "loss": 0.8367,
       "step": 26632
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.173818870471303e-08,
-      "loss": 0.6387,
+      "loss": 0.7062,
       "step": 26633
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.161859676846903e-08,
-      "loss": 0.8037,
+      "loss": 0.8145,
       "step": 26634
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.149914285190294e-08,
-      "loss": 0.7569,
+      "loss": 0.8013,
       "step": 26635
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.137982695667121e-08,
-      "loss": 0.8553,
+      "loss": 0.9088,
       "step": 26636
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.126064908442918e-08,
-      "loss": 0.7804,
+      "loss": 0.8113,
       "step": 26637
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.114160923682887e-08,
-      "loss": 0.7775,
+      "loss": 0.7782,
       "step": 26638
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.102270741552451e-08,
-      "loss": 0.71,
+      "loss": 0.7113,
       "step": 26639
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.090394362216367e-08,
-      "loss": 0.8353,
+      "loss": 0.8463,
       "step": 26640
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.078531785839502e-08,
-      "loss": 0.7378,
+      "loss": 0.7567,
       "step": 26641
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.066683012586393e-08,
-      "loss": 0.787,
+      "loss": 0.8731,
       "step": 26642
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.054848042621463e-08,
-      "loss": 0.688,
+      "loss": 0.7432,
       "step": 26643
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.043026876108803e-08,
-      "loss": 0.6387,
+      "loss": 0.7018,
       "step": 26644
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.031219513212504e-08,
-      "loss": 0.7904,
+      "loss": 0.7803,
       "step": 26645
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.0194259540964354e-08,
-      "loss": 0.7616,
+      "loss": 0.8167,
       "step": 26646
     },
     {
       "epoch": 0.97,
       "learning_rate": 5.007646198924132e-08,
-      "loss": 0.7837,
+      "loss": 0.7748,
       "step": 26647
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.995880247858909e-08,
-      "loss": 0.7499,
+      "loss": 0.7912,
       "step": 26648
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.9841281010641893e-08,
-      "loss": 0.6875,
+      "loss": 0.8477,
       "step": 26649
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.972389758702956e-08,
-      "loss": 0.7916,
+      "loss": 0.7773,
       "step": 26650
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.9606652209380765e-08,
-      "loss": 0.8283,
+      "loss": 0.7998,
       "step": 26651
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.9489544879322004e-08,
-      "loss": 0.7493,
+      "loss": 0.7661,
       "step": 26652
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.937257559847752e-08,
-      "loss": 0.7542,
+      "loss": 0.8079,
       "step": 26653
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.925574436847047e-08,
-      "loss": 0.8075,
+      "loss": 0.859,
       "step": 26654
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.913905119092177e-08,
-      "loss": 0.7392,
+      "loss": 0.8026,
       "step": 26655
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.9022496067450133e-08,
-      "loss": 0.7169,
+      "loss": 0.8437,
       "step": 26656
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.890607899967315e-08,
-      "loss": 0.7688,
+      "loss": 0.8383,
       "step": 26657
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.878979998920508e-08,
-      "loss": 0.7731,
+      "loss": 0.8313,
       "step": 26658
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.8673659037660193e-08,
-      "loss": 0.7407,
+      "loss": 0.7686,
       "step": 26659
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.8557656146648316e-08,
-      "loss": 0.8263,
+      "loss": 0.794,
       "step": 26660
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.844179131778037e-08,
-      "loss": 0.8128,
+      "loss": 0.7234,
       "step": 26661
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.8326064552662865e-08,
-      "loss": 0.7168,
+      "loss": 0.7793,
       "step": 26662
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.8210475852902285e-08,
-      "loss": 0.7854,
+      "loss": 0.7744,
       "step": 26663
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.809502522010068e-08,
-      "loss": 0.6963,
+      "loss": 0.7877,
       "step": 26664
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.797971265586121e-08,
-      "loss": 0.7027,
+      "loss": 0.7743,
       "step": 26665
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.7864538161783714e-08,
-      "loss": 0.6251,
+      "loss": 0.7079,
       "step": 26666
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.774950173946358e-08,
-      "loss": 0.7025,
+      "loss": 0.7764,
       "step": 26667
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.763460339050063e-08,
-      "loss": 0.6482,
+      "loss": 0.753,
       "step": 26668
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.751984311648694e-08,
-      "loss": 0.7225,
+      "loss": 0.7599,
       "step": 26669
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.7405220919013454e-08,
-      "loss": 0.6518,
+      "loss": 0.7216,
       "step": 26670
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.729073679967222e-08,
-      "loss": 0.733,
+      "loss": 0.7559,
       "step": 26671
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.7176390760051985e-08,
-      "loss": 0.8363,
+      "loss": 0.8294,
       "step": 26672
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.706218280173702e-08,
-      "loss": 0.8107,
+      "loss": 0.8643,
       "step": 26673
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.694811292631274e-08,
-      "loss": 0.6729,
+      "loss": 0.7163,
       "step": 26674
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.683418113536231e-08,
-      "loss": 0.6704,
+      "loss": 0.7145,
       "step": 26675
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.672038743046558e-08,
-      "loss": 0.6498,
+      "loss": 0.6476,
       "step": 26676
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.660673181320241e-08,
-      "loss": 0.7604,
+      "loss": 0.7599,
       "step": 26677
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.649321428514819e-08,
-      "loss": 0.6896,
+      "loss": 0.7763,
       "step": 26678
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.6379834847878335e-08,
-      "loss": 0.7952,
+      "loss": 0.8075,
       "step": 26679
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.6266593502966026e-08,
-      "loss": 0.6649,
+      "loss": 0.8043,
       "step": 26680
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.615349025198335e-08,
-      "loss": 0.6575,
+      "loss": 0.7819,
       "step": 26681
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.604052509649681e-08,
-      "loss": 0.8013,
+      "loss": 0.7943,
       "step": 26682
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.592769803807517e-08,
-      "loss": 0.6321,
+      "loss": 0.6138,
       "step": 26683
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.581500907828496e-08,
-      "loss": 0.7749,
+      "loss": 0.7852,
       "step": 26684
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.570245821868713e-08,
-      "loss": 0.8173,
+      "loss": 0.8245,
       "step": 26685
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.559004546084489e-08,
-      "loss": 0.8314,
+      "loss": 0.8875,
       "step": 26686
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.54777708063181e-08,
-      "loss": 0.7984,
+      "loss": 0.8502,
       "step": 26687
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.5365634256662186e-08,
-      "loss": 0.7337,
+      "loss": 0.8078,
       "step": 26688
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.525363581343478e-08,
-      "loss": 0.7218,
+      "loss": 0.6872,
       "step": 26689
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.51417754781891e-08,
-      "loss": 0.6022,
+      "loss": 0.6751,
       "step": 26690
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.5030053252478334e-08,
-      "loss": 0.7278,
+      "loss": 0.73,
       "step": 26691
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.491846913785014e-08,
-      "loss": 0.614,
+      "loss": 0.6323,
       "step": 26692
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.4807023135854386e-08,
-      "loss": 0.8208,
+      "loss": 0.879,
       "step": 26693
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.469571524803651e-08,
-      "loss": 0.7234,
+      "loss": 0.8213,
       "step": 26694
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.458454547594082e-08,
-      "loss": 0.7186,
+      "loss": 0.668,
       "step": 26695
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.447351382110942e-08,
-      "loss": 0.7161,
+      "loss": 0.7868,
       "step": 26696
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.4362620285083314e-08,
-      "loss": 0.7717,
+      "loss": 0.7571,
       "step": 26697
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.4251864869400144e-08,
-      "loss": 0.8157,
+      "loss": 0.818,
       "step": 26698
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.4141247575596456e-08,
-      "loss": 0.7594,
+      "loss": 0.7199,
       "step": 26699
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.4030768405208814e-08,
-      "loss": 0.7044,
+      "loss": 0.7651,
       "step": 26700
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.3920427359768204e-08,
-      "loss": 0.7283,
+      "loss": 0.8566,
       "step": 26701
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.3810224440805626e-08,
-      "loss": 0.617,
+      "loss": 0.6823,
       "step": 26702
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.370015964985097e-08,
-      "loss": 0.7773,
+      "loss": 0.7906,
       "step": 26703
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.359023298842968e-08,
-      "loss": 0.6771,
+      "loss": 0.8062,
       "step": 26704
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.3480444458067204e-08,
-      "loss": 0.6901,
+      "loss": 0.7465,
       "step": 26705
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.337079406028788e-08,
-      "loss": 0.8379,
+      "loss": 0.9123,
       "step": 26706
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.32612817966116e-08,
-      "loss": 0.6861,
+      "loss": 0.7447,
       "step": 26707
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.3151907668558256e-08,
-      "loss": 0.7221,
+      "loss": 0.8341,
       "step": 26708
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.304267167764553e-08,
-      "loss": 0.6909,
+      "loss": 0.7504,
       "step": 26709
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.293357382538887e-08,
-      "loss": 0.7948,
+      "loss": 0.79,
       "step": 26710
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.282461411330152e-08,
-      "loss": 0.8178,
+      "loss": 0.8479,
       "step": 26711
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.271579254289559e-08,
-      "loss": 0.694,
+      "loss": 0.7682,
       "step": 26712
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.260710911567989e-08,
-      "loss": 0.657,
+      "loss": 0.7321,
       "step": 26713
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.24985638331632e-08,
-      "loss": 0.7757,
+      "loss": 0.8489,
       "step": 26714
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.239015669685098e-08,
-      "loss": 0.6645,
+      "loss": 0.7219,
       "step": 26715
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.228188770824759e-08,
-      "loss": 0.7859,
+      "loss": 0.8566,
       "step": 26716
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.217375686885406e-08,
-      "loss": 0.8236,
+      "loss": 0.889,
       "step": 26717
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.2065764180172496e-08,
-      "loss": 0.8355,
+      "loss": 0.843,
       "step": 26718
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.1957909643699504e-08,
-      "loss": 0.8105,
+      "loss": 0.8467,
       "step": 26719
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.185019326093165e-08,
-      "loss": 0.6802,
+      "loss": 0.7177,
       "step": 26720
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.1742615033363296e-08,
-      "loss": 0.7224,
+      "loss": 0.8355,
       "step": 26721
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.16351749624877e-08,
-      "loss": 0.823,
+      "loss": 0.6863,
       "step": 26722
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.1527873049795885e-08,
-      "loss": 0.766,
+      "loss": 0.8029,
       "step": 26723
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.142070929677333e-08,
-      "loss": 0.7875,
+      "loss": 0.7344,
       "step": 26724
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.1313683704911066e-08,
-      "loss": 0.7562,
+      "loss": 0.7687,
       "step": 26725
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.120679627569124e-08,
-      "loss": 0.7529,
+      "loss": 0.808,
       "step": 26726
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.110004701059711e-08,
-      "loss": 0.7129,
+      "loss": 0.7283,
       "step": 26727
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.099343591111082e-08,
-      "loss": 0.782,
+      "loss": 0.7639,
       "step": 26728
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.088696297871009e-08,
-      "loss": 0.6491,
+      "loss": 0.7503,
       "step": 26729
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.078062821487261e-08,
-      "loss": 0.684,
+      "loss": 0.7663,
       "step": 26730
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.067443162107387e-08,
-      "loss": 0.8475,
+      "loss": 0.8013,
       "step": 26731
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.056837319878715e-08,
-      "loss": 0.8245,
+      "loss": 0.8436,
       "step": 26732
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.046245294948237e-08,
-      "loss": 0.7854,
+      "loss": 0.8049,
       "step": 26733
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.0356670874631685e-08,
-      "loss": 0.7151,
+      "loss": 0.74,
       "step": 26734
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.02510269757006e-08,
-      "loss": 0.912,
+      "loss": 0.8897,
       "step": 26735
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.014552125415572e-08,
-      "loss": 0.9314,
+      "loss": 0.8929,
       "step": 26736
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.004015371146031e-08,
-      "loss": 0.8058,
+      "loss": 0.8305,
       "step": 26737
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.993492434907764e-08,
-      "loss": 0.7582,
+      "loss": 0.7538,
       "step": 26738
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.9829833168465446e-08,
-      "loss": 0.8505,
+      "loss": 0.8743,
       "step": 26739
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.972488017108256e-08,
-      "loss": 0.8537,
+      "loss": 0.8281,
       "step": 26740
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.962006535838559e-08,
-      "loss": 0.6143,
+      "loss": 0.7203,
       "step": 26741
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.951538873182781e-08,
-      "loss": 0.8425,
+      "loss": 0.8658,
       "step": 26742
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.94108502928614e-08,
-      "loss": 0.6109,
+      "loss": 0.6793,
       "step": 26743
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.9306450042938535e-08,
-      "loss": 0.8694,
+      "loss": 0.8706,
       "step": 26744
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.9202187983505836e-08,
-      "loss": 0.7846,
+      "loss": 0.7867,
       "step": 26745
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.9098064116009915e-08,
-      "loss": 0.8411,
+      "loss": 0.8907,
       "step": 26746
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.899407844189518e-08,
-      "loss": 0.7642,
+      "loss": 0.8069,
       "step": 26747
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.889023096260603e-08,
-      "loss": 0.7091,
+      "loss": 0.7467,
       "step": 26748
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.87865216795813e-08,
-      "loss": 0.7507,
+      "loss": 0.8123,
       "step": 26749
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.868295059426097e-08,
-      "loss": 0.756,
+      "loss": 0.7522,
       "step": 26750
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.857951770808166e-08,
-      "loss": 0.8268,
+      "loss": 0.8512,
       "step": 26751
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.847622302247667e-08,
-      "loss": 0.66,
+      "loss": 0.7147,
       "step": 26752
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.837306653888151e-08,
-      "loss": 0.6755,
+      "loss": 0.8084,
       "step": 26753
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.8270048258726154e-08,
-      "loss": 0.7778,
+      "loss": 0.8116,
       "step": 26754
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.816716818344057e-08,
-      "loss": 0.739,
+      "loss": 0.7784,
       "step": 26755
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.80644263144514e-08,
-      "loss": 0.6598,
+      "loss": 0.7132,
       "step": 26756
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.796182265318305e-08,
-      "loss": 0.8749,
+      "loss": 0.8817,
       "step": 26757
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7859357201061044e-08,
-      "loss": 0.6942,
+      "loss": 0.7468,
       "step": 26758
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.775702995950647e-08,
-      "loss": 0.7024,
+      "loss": 0.7037,
       "step": 26759
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7654840929937095e-08,
-      "loss": 0.7275,
+      "loss": 0.8686,
       "step": 26760
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.755279011377399e-08,
-      "loss": 0.761,
+      "loss": 0.7627,
       "step": 26761
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7450877512429376e-08,
-      "loss": 0.7495,
+      "loss": 0.7865,
       "step": 26762
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7349103127319874e-08,
-      "loss": 0.8547,
+      "loss": 0.8298,
       "step": 26763
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.724746695985659e-08,
-      "loss": 0.665,
+      "loss": 0.7933,
       "step": 26764
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.71459690114484e-08,
-      "loss": 0.8114,
+      "loss": 0.8096,
       "step": 26765
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7044609283504174e-08,
-      "loss": 0.7085,
+      "loss": 0.8365,
       "step": 26766
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.694338777743167e-08,
-      "loss": 0.6904,
+      "loss": 0.7257,
       "step": 26767
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6842304494633106e-08,
-      "loss": 0.9021,
+      "loss": 0.981,
       "step": 26768
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6741359436512914e-08,
-      "loss": 0.6408,
+      "loss": 0.7298,
       "step": 26769
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6640552604468856e-08,
-      "loss": 0.6613,
+      "loss": 0.7708,
       "step": 26770
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6539883999902046e-08,
-      "loss": 0.7096,
+      "loss": 0.8174,
       "step": 26771
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.643935362420692e-08,
-      "loss": 0.7039,
+      "loss": 0.7852,
       "step": 26772
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6338961478780135e-08,
-      "loss": 0.7472,
+      "loss": 0.7798,
       "step": 26773
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.623870756501391e-08,
-      "loss": 0.6666,
+      "loss": 0.715,
       "step": 26774
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.6138591884298243e-08,
-      "loss": 0.6544,
+      "loss": 0.7929,
       "step": 26775
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.603861443802314e-08,
-      "loss": 0.7035,
+      "loss": 0.8255,
       "step": 26776
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.593877522757527e-08,
-      "loss": 0.7816,
+      "loss": 0.8169,
       "step": 26777
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.583907425433908e-08,
-      "loss": 0.819,
+      "loss": 0.8709,
       "step": 26778
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.573951151969901e-08,
-      "loss": 0.6956,
+      "loss": 0.7801,
       "step": 26779
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.564008702503619e-08,
-      "loss": 0.7545,
+      "loss": 0.791,
       "step": 26780
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.554080077172839e-08,
-      "loss": 0.7055,
+      "loss": 0.7398,
       "step": 26781
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.544165276115452e-08,
-      "loss": 0.6929,
+      "loss": 0.6959,
       "step": 26782
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.534264299469126e-08,
-      "loss": 0.7983,
+      "loss": 0.8526,
       "step": 26783
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.524377147370972e-08,
-      "loss": 0.612,
+      "loss": 0.702,
       "step": 26784
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.514503819958215e-08,
-      "loss": 0.7027,
+      "loss": 0.6997,
       "step": 26785
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.5046443173679666e-08,
-      "loss": 0.8353,
+      "loss": 0.8326,
       "step": 26786
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.494798639736896e-08,
-      "loss": 0.8618,
+      "loss": 0.8704,
       "step": 26787
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4849667872016714e-08,
-      "loss": 0.6584,
+      "loss": 0.7374,
       "step": 26788
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4751487598987387e-08,
-      "loss": 0.6516,
+      "loss": 0.738,
       "step": 26789
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.465344557964101e-08,
-      "loss": 0.9293,
+      "loss": 0.9474,
       "step": 26790
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4555541815339824e-08,
-      "loss": 0.8521,
+      "loss": 0.8009,
       "step": 26791
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4457776307441624e-08,
-      "loss": 0.7775,
+      "loss": 0.7317,
       "step": 26792
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4360149057302006e-08,
-      "loss": 0.6433,
+      "loss": 0.773,
       "step": 26793
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4262660066275436e-08,
-      "loss": 0.7601,
+      "loss": 0.7854,
       "step": 26794
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.416530933571638e-08,
-      "loss": 0.6398,
+      "loss": 0.7856,
       "step": 26795
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.4068096866972654e-08,
-      "loss": 0.8085,
+      "loss": 0.7979,
       "step": 26796
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.39710226613954e-08,
-      "loss": 0.6036,
+      "loss": 0.6547,
       "step": 26797
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.3874086720329104e-08,
-      "loss": 0.7013,
+      "loss": 0.7793,
       "step": 26798
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.3777289045120454e-08,
-      "loss": 0.8134,
+      "loss": 0.774,
       "step": 26799
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.368062963711061e-08,
-      "loss": 0.6855,
+      "loss": 0.7594,
       "step": 26800
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.3584108497642934e-08,
-      "loss": 0.7583,
+      "loss": 0.8289,
       "step": 26801
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.348772562805525e-08,
-      "loss": 0.8566,
+      "loss": 0.8469,
       "step": 26802
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.339148102968426e-08,
-      "loss": 0.7252,
+      "loss": 0.7806,
       "step": 26803
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.329537470386557e-08,
-      "loss": 0.6916,
+      "loss": 0.7193,
       "step": 26804
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.319940665193366e-08,
-      "loss": 0.8421,
+      "loss": 0.8167,
       "step": 26805
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.310357687521748e-08,
-      "loss": 0.8176,
+      "loss": 0.7904,
       "step": 26806
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.300788537504929e-08,
-      "loss": 0.7649,
+      "loss": 0.8214,
       "step": 26807
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.2912332152754686e-08,
-      "loss": 0.5992,
+      "loss": 0.7362,
       "step": 26808
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.2816917209660406e-08,
-      "loss": 0.7177,
+      "loss": 0.7401,
       "step": 26809
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.2721640547088705e-08,
-      "loss": 0.7005,
+      "loss": 0.7237,
       "step": 26810
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.262650216636409e-08,
-      "loss": 0.7913,
+      "loss": 0.8367,
       "step": 26811
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.25315020688044e-08,
-      "loss": 0.8094,
+      "loss": 0.8186,
       "step": 26812
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.243664025572746e-08,
-      "loss": 0.8176,
+      "loss": 0.8398,
       "step": 26813
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.2341916728449995e-08,
-      "loss": 0.7933,
+      "loss": 0.7395,
       "step": 26814
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.2247331488286516e-08,
-      "loss": 0.6976,
+      "loss": 0.7474,
       "step": 26815
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.21528845365493e-08,
-      "loss": 0.5935,
+      "loss": 0.6982,
       "step": 26816
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.205857587454841e-08,
-      "loss": 0.7418,
+      "loss": 0.7309,
       "step": 26817
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.19644055035917e-08,
-      "loss": 0.7826,
+      "loss": 0.8105,
       "step": 26818
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.187037342498589e-08,
-      "loss": 0.7964,
+      "loss": 0.8189,
       "step": 26819
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.177647964003661e-08,
-      "loss": 0.7698,
+      "loss": 0.7886,
       "step": 26820
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.168272415004503e-08,
-      "loss": 0.784,
+      "loss": 0.8209,
       "step": 26821
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.158910695631234e-08,
-      "loss": 0.6853,
+      "loss": 0.7636,
       "step": 26822
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.1495628060136395e-08,
-      "loss": 0.5643,
+      "loss": 0.7317,
       "step": 26823
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.140228746281726e-08,
-      "loss": 0.6833,
+      "loss": 0.7543,
       "step": 26824
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.1309085165646124e-08,
-      "loss": 0.7095,
+      "loss": 0.7861,
       "step": 26825
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.121602116991862e-08,
-      "loss": 0.6722,
+      "loss": 0.8042,
       "step": 26826
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.1123095476924827e-08,
-      "loss": 0.7787,
+      "loss": 0.8358,
       "step": 26827
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.103030808795371e-08,
-      "loss": 0.7109,
+      "loss": 0.7253,
       "step": 26828
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.093765900429313e-08,
-      "loss": 0.608,
+      "loss": 0.7509,
       "step": 26829
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.084514822722873e-08,
-      "loss": 0.6794,
+      "loss": 0.7208,
       "step": 26830
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.075277575804392e-08,
-      "loss": 0.7741,
+      "loss": 0.7349,
       "step": 26831
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.066054159801879e-08,
-      "loss": 0.7419,
+      "loss": 0.7425,
       "step": 26832
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.056844574843454e-08,
-      "loss": 0.7949,
+      "loss": 0.8268,
       "step": 26833
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.047648821056792e-08,
-      "loss": 0.7339,
+      "loss": 0.7831,
       "step": 26834
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.038466898569459e-08,
-      "loss": 0.7343,
+      "loss": 0.8112,
       "step": 26835
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.029298807509018e-08,
-      "loss": 0.671,
+      "loss": 0.806,
       "step": 26836
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.0201445480024794e-08,
-      "loss": 0.6845,
+      "loss": 0.7153,
       "step": 26837
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.011004120176853e-08,
-      "loss": 0.7769,
+      "loss": 0.8089,
       "step": 26838
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.001877524159147e-08,
-      "loss": 0.6862,
+      "loss": 0.7937,
       "step": 26839
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.992764760075706e-08,
-      "loss": 0.7769,
+      "loss": 0.7971,
       "step": 26840
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.983665828053206e-08,
-      "loss": 0.7478,
+      "loss": 0.7838,
       "step": 26841
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9745807282176574e-08,
-      "loss": 0.7536,
+      "loss": 0.6897,
       "step": 26842
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9655094606951818e-08,
-      "loss": 0.6956,
+      "loss": 0.7514,
       "step": 26843
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.956452025611789e-08,
-      "loss": 0.7724,
+      "loss": 0.8082,
       "step": 26844
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9474084230928234e-08,
-      "loss": 0.6521,
+      "loss": 0.729,
       "step": 26845
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9383786532640736e-08,
-      "loss": 0.8077,
+      "loss": 0.7896,
       "step": 26846
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.929362716250661e-08,
-      "loss": 0.6775,
+      "loss": 0.7515,
       "step": 26847
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9203606121777085e-08,
-      "loss": 0.6801,
+      "loss": 0.73,
       "step": 26848
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9113723411700046e-08,
-      "loss": 0.6914,
+      "loss": 0.7194,
       "step": 26849
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.9023979033523386e-08,
-      "loss": 0.7161,
+      "loss": 0.7201,
       "step": 26850
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8934372988491667e-08,
-      "loss": 0.674,
+      "loss": 0.7379,
       "step": 26851
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8844905277849446e-08,
-      "loss": 0.6396,
+      "loss": 0.6883,
       "step": 26852
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.875557590283684e-08,
-      "loss": 0.8118,
+      "loss": 0.781,
       "step": 26853
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8666384864692863e-08,
-      "loss": 0.7243,
+      "loss": 0.7252,
       "step": 26854
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8577332164654303e-08,
-      "loss": 0.6968,
+      "loss": 0.7946,
       "step": 26855
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8488417803959056e-08,
-      "loss": 0.7752,
+      "loss": 0.8333,
       "step": 26856
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.839964178383947e-08,
-      "loss": 0.705,
+      "loss": 0.7923,
       "step": 26857
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8311004105525673e-08,
-      "loss": 0.6997,
+      "loss": 0.7989,
       "step": 26858
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8222504770250014e-08,
-      "loss": 0.7239,
+      "loss": 0.7879,
       "step": 26859
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.8134143779238175e-08,
-      "loss": 0.6361,
+      "loss": 0.7614,
       "step": 26860
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.804592113371696e-08,
-      "loss": 0.539,
+      "loss": 0.6767,
       "step": 26861
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7957836834910935e-08,
-      "loss": 0.7462,
+      "loss": 0.7225,
       "step": 26862
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.786989088404135e-08,
-      "loss": 0.8436,
+      "loss": 0.8702,
       "step": 26863
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7782083282327232e-08,
-      "loss": 0.7653,
+      "loss": 0.8255,
       "step": 26864
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7694414030989825e-08,
-      "loss": 0.7987,
+      "loss": 0.8099,
       "step": 26865
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.76068831312426e-08,
-      "loss": 0.8647,
+      "loss": 0.8282,
       "step": 26866
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7519490584300146e-08,
-      "loss": 0.7867,
+      "loss": 0.7518,
       "step": 26867
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7432236391375934e-08,
-      "loss": 0.782,
+      "loss": 0.7433,
       "step": 26868
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7345120553681214e-08,
-      "loss": 0.8356,
+      "loss": 0.8684,
       "step": 26869
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7258143072422803e-08,
-      "loss": 0.7136,
+      "loss": 0.8132,
       "step": 26870
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.7171303948808624e-08,
-      "loss": 0.8156,
+      "loss": 0.8113,
       "step": 26871
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.708460318404327e-08,
-      "loss": 0.6755,
+      "loss": 0.7722,
       "step": 26872
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.6998040779329105e-08,
-      "loss": 0.7479,
+      "loss": 0.793,
       "step": 26873
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.6911616735866285e-08,
-      "loss": 0.7565,
+      "loss": 0.8707,
       "step": 26874
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.6825331054856074e-08,
-      "loss": 0.672,
+      "loss": 0.7068,
       "step": 26875
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.673918373749307e-08,
-      "loss": 0.7732,
+      "loss": 0.791,
       "step": 26876
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.66531747849752e-08,
-      "loss": 0.7254,
+      "loss": 0.8058,
       "step": 26877
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.656730419849263e-08,
-      "loss": 0.7641,
+      "loss": 0.7777,
       "step": 26878
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.648157197923995e-08,
-      "loss": 0.7003,
+      "loss": 0.7136,
       "step": 26879
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.639597812840289e-08,
-      "loss": 0.7841,
+      "loss": 0.851,
       "step": 26880
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.631052264717271e-08,
-      "loss": 0.8528,
+      "loss": 0.8586,
       "step": 26881
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.6225205536732913e-08,
-      "loss": 0.907,
+      "loss": 0.8665,
       "step": 26882
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.6140026798266994e-08,
-      "loss": 0.8915,
+      "loss": 0.8554,
       "step": 26883
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.605498643295734e-08,
-      "loss": 0.7793,
+      "loss": 0.7517,
       "step": 26884
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5970084441983013e-08,
-      "loss": 0.6633,
+      "loss": 0.6975,
       "step": 26885
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5885320826523063e-08,
-      "loss": 0.7341,
+      "loss": 0.7937,
       "step": 26886
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5800695587753223e-08,
-      "loss": 0.7534,
+      "loss": 0.7354,
       "step": 26887
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.571620872684588e-08,
-      "loss": 0.7269,
+      "loss": 0.8366,
       "step": 26888
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.563186024497566e-08,
-      "loss": 0.8166,
+      "loss": 0.8365,
       "step": 26889
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.554765014331162e-08,
-      "loss": 0.6669,
+      "loss": 0.8127,
       "step": 26890
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.546357842302061e-08,
-      "loss": 0.8018,
+      "loss": 0.7763,
       "step": 26891
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5379645085272798e-08,
-      "loss": 0.7681,
+      "loss": 0.7717,
       "step": 26892
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.529585013122837e-08,
-      "loss": 0.7746,
+      "loss": 0.8102,
       "step": 26893
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5212193562051958e-08,
-      "loss": 0.8466,
+      "loss": 0.8629,
       "step": 26894
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.512867537890484e-08,
-      "loss": 0.7222,
+      "loss": 0.8494,
       "step": 26895
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.5045295582944993e-08,
-      "loss": 0.6332,
+      "loss": 0.7765,
       "step": 26896
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4962054175328153e-08,
-      "loss": 0.7264,
+      "loss": 0.7822,
       "step": 26897
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4878951157210064e-08,
-      "loss": 0.7554,
+      "loss": 0.7078,
       "step": 26898
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4795986529744243e-08,
-      "loss": 0.6507,
+      "loss": 0.7404,
       "step": 26899
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4713160294080885e-08,
-      "loss": 0.7309,
+      "loss": 0.7702,
       "step": 26900
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.463047245136907e-08,
-      "loss": 0.7105,
+      "loss": 0.8092,
       "step": 26901
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.454792300275677e-08,
-      "loss": 0.8733,
+      "loss": 0.8388,
       "step": 26902
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4465511949388622e-08,
-      "loss": 0.8769,
+      "loss": 0.7691,
       "step": 26903
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4383239292408155e-08,
-      "loss": 0.7089,
+      "loss": 0.755,
       "step": 26904
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4301105032956683e-08,
-      "loss": 0.6626,
+      "loss": 0.68,
       "step": 26905
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4219109172173295e-08,
-      "loss": 0.7867,
+      "loss": 0.774,
       "step": 26906
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4137251711195965e-08,
-      "loss": 0.6595,
+      "loss": 0.7393,
       "step": 26907
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.4055532651160453e-08,
-      "loss": 0.71,
+      "loss": 0.7566,
       "step": 26908
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3973951993201404e-08,
-      "loss": 0.8016,
+      "loss": 0.7427,
       "step": 26909
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3892509738447922e-08,
-      "loss": 0.7549,
+      "loss": 0.7207,
       "step": 26910
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.381120588803354e-08,
-      "loss": 0.8652,
+      "loss": 0.8359,
       "step": 26911
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3730040443082912e-08,
-      "loss": 0.7245,
+      "loss": 0.7432,
       "step": 26912
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.364901340472403e-08,
-      "loss": 0.8279,
+      "loss": 0.8405,
       "step": 26913
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3568124774080438e-08,
-      "loss": 0.664,
+      "loss": 0.7257,
       "step": 26914
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3487374552274568e-08,
-      "loss": 0.7222,
+      "loss": 0.8139,
       "step": 26915
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3406762740425525e-08,
-      "loss": 0.775,
+      "loss": 0.7764,
       "step": 26916
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.332628933965353e-08,
-      "loss": 0.6635,
+      "loss": 0.8384,
       "step": 26917
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3245954351074352e-08,
-      "loss": 0.7024,
+      "loss": 0.7135,
       "step": 26918
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3165757775802654e-08,
-      "loss": 0.7433,
+      "loss": 0.7146,
       "step": 26919
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3085699614949773e-08,
-      "loss": 0.8102,
+      "loss": 0.7874,
       "step": 26920
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.3005779869628153e-08,
-      "loss": 0.8515,
+      "loss": 0.866,
       "step": 26921
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2925998540945792e-08,
-      "loss": 0.6978,
+      "loss": 0.7582,
       "step": 26922
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2846355630009586e-08,
-      "loss": 0.7102,
+      "loss": 0.7483,
       "step": 26923
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2766851137924208e-08,
-      "loss": 0.7327,
+      "loss": 0.7925,
       "step": 26924
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2687485065793214e-08,
-      "loss": 0.7802,
+      "loss": 0.7764,
       "step": 26925
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2608257414717948e-08,
-      "loss": 0.7971,
+      "loss": 0.784,
       "step": 26926
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2529168185796426e-08,
-      "loss": 0.749,
+      "loss": 0.7069,
       "step": 26927
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2450217380126647e-08,
-      "loss": 0.7701,
+      "loss": 0.8101,
       "step": 26928
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2371404998803304e-08,
-      "loss": 0.7168,
+      "loss": 0.725,
       "step": 26929
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.229273104292107e-08,
-      "loss": 0.7669,
+      "loss": 0.7712,
       "step": 26930
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.221419551356907e-08,
-      "loss": 0.787,
+      "loss": 0.8402,
       "step": 26931
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.213579841183977e-08,
-      "loss": 0.7452,
+      "loss": 0.7875,
       "step": 26932
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.2057539738820078e-08,
-      "loss": 0.7614,
+      "loss": 0.8412,
       "step": 26933
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.197941949559468e-08,
-      "loss": 0.7544,
+      "loss": 0.7788,
       "step": 26934
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1901437683247152e-08,
-      "loss": 0.6726,
+      "loss": 0.8067,
       "step": 26935
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1823594302861074e-08,
-      "loss": 0.6673,
+      "loss": 0.8558,
       "step": 26936
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.174588935551447e-08,
-      "loss": 0.7364,
+      "loss": 0.7701,
       "step": 26937
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1668322842287593e-08,
-      "loss": 0.8224,
+      "loss": 0.7912,
       "step": 26938
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1590894764254023e-08,
-      "loss": 0.694,
+      "loss": 0.6829,
       "step": 26939
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1513605122490678e-08,
-      "loss": 0.6987,
+      "loss": 0.7771,
       "step": 26940
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.14364539180667e-08,
-      "loss": 0.9137,
+      "loss": 0.828,
       "step": 26941
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1359441152055683e-08,
-      "loss": 0.7447,
+      "loss": 0.8483,
       "step": 26942
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.128256682552454e-08,
-      "loss": 0.6507,
+      "loss": 0.6881,
       "step": 26943
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.1205830939539096e-08,
-      "loss": 0.7956,
+      "loss": 0.8044,
       "step": 26944
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.112923349516516e-08,
-      "loss": 0.8432,
+      "loss": 0.8478,
       "step": 26945
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.105277449346521e-08,
-      "loss": 0.8237,
+      "loss": 0.813,
       "step": 26946
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.097645393549841e-08,
-      "loss": 0.7627,
+      "loss": 0.7727,
       "step": 26947
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0900271822326123e-08,
-      "loss": 0.804,
+      "loss": 0.8079,
       "step": 26948
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.082422815500418e-08,
-      "loss": 0.7357,
+      "loss": 0.7776,
       "step": 26949
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0748322934586172e-08,
-      "loss": 0.7347,
+      "loss": 0.8293,
       "step": 26950
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0672556162127933e-08,
-      "loss": 0.6654,
+      "loss": 0.7453,
       "step": 26951
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0596927838678614e-08,
-      "loss": 0.717,
+      "loss": 0.7952,
       "step": 26952
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0521437965288493e-08,
-      "loss": 0.7192,
+      "loss": 0.7288,
       "step": 26953
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0446086543003396e-08,
-      "loss": 0.7859,
+      "loss": 0.7874,
       "step": 26954
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0370873572870263e-08,
-      "loss": 0.7444,
+      "loss": 0.7908,
       "step": 26955
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0295799055931598e-08,
-      "loss": 0.8268,
+      "loss": 0.7943,
       "step": 26956
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0220862993229896e-08,
-      "loss": 0.7699,
+      "loss": 0.8233,
       "step": 26957
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0146065385804325e-08,
-      "loss": 0.8,
+      "loss": 0.8749,
       "step": 26958
     },
     {
       "epoch": 0.98,
       "learning_rate": 2.0071406234691838e-08,
-      "loss": 0.7777,
+      "loss": 0.7784,
       "step": 26959
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.999688554092938e-08,
-      "loss": 0.7085,
+      "loss": 0.7531,
       "step": 26960
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9922503305550565e-08,
-      "loss": 0.7063,
+      "loss": 0.7581,
       "step": 26961
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.98482595295868e-08,
-      "loss": 0.8892,
+      "loss": 0.8385,
       "step": 26962
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9774154214068365e-08,
-      "loss": 0.6895,
+      "loss": 0.7597,
       "step": 26963
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9700187360024437e-08,
-      "loss": 0.6477,
+      "loss": 0.7164,
       "step": 26964
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9626358968479752e-08,
-      "loss": 0.6678,
+      "loss": 0.7276,
       "step": 26965
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.955266904045794e-08,
-      "loss": 0.8016,
+      "loss": 0.7665,
       "step": 26966
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.947911757698373e-08,
-      "loss": 0.8152,
+      "loss": 0.8119,
       "step": 26967
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9405704579075203e-08,
-      "loss": 0.7437,
+      "loss": 0.8065,
       "step": 26968
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.933243004775265e-08,
-      "loss": 0.7085,
+      "loss": 0.7874,
       "step": 26969
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9259293984031925e-08,
-      "loss": 0.682,
+      "loss": 0.8328,
       "step": 26970
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9186296388927773e-08,
-      "loss": 0.7735,
+      "loss": 0.7985,
       "step": 26971
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.9113437263452718e-08,
-      "loss": 0.6885,
+      "loss": 0.7286,
       "step": 26972
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.904071660861706e-08,
-      "loss": 0.8574,
+      "loss": 0.8115,
       "step": 26973
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8968134425431107e-08,
-      "loss": 0.7046,
+      "loss": 0.7109,
       "step": 26974
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8895690714900717e-08,
-      "loss": 0.734,
+      "loss": 0.7498,
       "step": 26975
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8823385478030643e-08,
-      "loss": 0.6921,
+      "loss": 0.7291,
       "step": 26976
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8751218715825635e-08,
-      "loss": 0.8988,
+      "loss": 0.8848,
       "step": 26977
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8679190429286008e-08,
-      "loss": 0.702,
+      "loss": 0.7514,
       "step": 26978
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.860730061940985e-08,
-      "loss": 0.7803,
+      "loss": 0.8002,
       "step": 26979
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8535549287197472e-08,
-      "loss": 0.7686,
+      "loss": 0.809,
       "step": 26980
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8463936433641418e-08,
-      "loss": 0.6775,
+      "loss": 0.7476,
       "step": 26981
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8392462059736438e-08,
-      "loss": 0.7325,
+      "loss": 0.7717,
       "step": 26982
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8321126166472857e-08,
-      "loss": 0.8019,
+      "loss": 0.8564,
       "step": 26983
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8249928754843216e-08,
-      "loss": 0.6613,
+      "loss": 0.7289,
       "step": 26984
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.817886982583228e-08,
-      "loss": 0.7469,
+      "loss": 0.761,
       "step": 26985
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.810794938042815e-08,
-      "loss": 0.8957,
+      "loss": 0.8619,
       "step": 26986
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.8037167419613367e-08,
-      "loss": 0.7419,
+      "loss": 0.7189,
       "step": 26987
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7966523944370484e-08,
-      "loss": 0.8844,
+      "loss": 0.8402,
       "step": 26988
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.789601895567872e-08,
-      "loss": 0.8421,
+      "loss": 0.7952,
       "step": 26989
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7825652454517282e-08,
-      "loss": 0.719,
+      "loss": 0.7829,
       "step": 26990
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.775542444186318e-08,
-      "loss": 0.865,
+      "loss": 0.8506,
       "step": 26991
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7685334918688957e-08,
-      "loss": 0.7522,
+      "loss": 0.7955,
       "step": 26992
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7615383885967173e-08,
-      "loss": 0.6939,
+      "loss": 0.7517,
       "step": 26993
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7545571344669276e-08,
-      "loss": 0.7499,
+      "loss": 0.7924,
       "step": 26994
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7475897295762267e-08,
-      "loss": 0.7304,
+      "loss": 0.7426,
       "step": 26995
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7406361740215372e-08,
-      "loss": 0.7331,
+      "loss": 0.8026,
       "step": 26996
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7336964678991153e-08,
-      "loss": 0.7103,
+      "loss": 0.7315,
       "step": 26997
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7267706113052172e-08,
-      "loss": 0.8148,
+      "loss": 0.819,
       "step": 26998
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7198586043360997e-08,
-      "loss": 0.7507,
+      "loss": 0.7473,
       "step": 26999
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.7129604470875748e-08,
-      "loss": 0.8662,
+      "loss": 0.8822,
       "step": 27000
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.706076139655344e-08,
-      "loss": 0.666,
+      "loss": 0.6864,
       "step": 27001
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6992056821348856e-08,
-      "loss": 0.7191,
+      "loss": 0.7949,
       "step": 27002
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6923490746214576e-08,
-      "loss": 0.7922,
+      "loss": 0.8488,
       "step": 27003
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.685506317210428e-08,
-      "loss": 0.6236,
+      "loss": 0.7337,
       "step": 27004
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6786774099964986e-08,
-      "loss": 0.7449,
+      "loss": 0.6955,
       "step": 27005
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6718623530744827e-08,
-      "loss": 0.7551,
+      "loss": 0.7598,
       "step": 27006
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.665061146538971e-08,
-      "loss": 0.7405,
+      "loss": 0.7961,
       "step": 27007
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6582737904842217e-08,
-      "loss": 0.8097,
+      "loss": 0.8309,
       "step": 27008
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6515002850044927e-08,
-      "loss": 0.6589,
+      "loss": 0.7589,
       "step": 27009
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6447406301938197e-08,
-      "loss": 0.6925,
+      "loss": 0.6775,
       "step": 27010
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6379948261457946e-08,
-      "loss": 0.8424,
+      "loss": 0.8396,
       "step": 27011
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.63126287295412e-08,
-      "loss": 0.7305,
+      "loss": 0.7549,
       "step": 27012
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.6245447707121664e-08,
-      "loss": 0.7354,
+      "loss": 0.7253,
       "step": 27013
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.617840519513081e-08,
-      "loss": 0.758,
+      "loss": 0.8092,
       "step": 27014
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.611150119450011e-08,
-      "loss": 0.7113,
+      "loss": 0.7455,
       "step": 27015
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.604473570615772e-08,
-      "loss": 0.6189,
+      "loss": 0.7082,
       "step": 27016
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5978108731028454e-08,
-      "loss": 0.7279,
+      "loss": 0.7715,
       "step": 27017
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.591162027003712e-08,
-      "loss": 0.6556,
+      "loss": 0.699,
       "step": 27018
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5845270324106322e-08,
-      "loss": 0.6883,
+      "loss": 0.7171,
       "step": 27019
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.577905889415754e-08,
-      "loss": 0.7019,
+      "loss": 0.7592,
       "step": 27020
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.571298598110893e-08,
-      "loss": 0.7083,
+      "loss": 0.7231,
       "step": 27021
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5647051585875316e-08,
-      "loss": 0.7914,
+      "loss": 0.8019,
       "step": 27022
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5581255709374854e-08,
-      "loss": 0.7252,
+      "loss": 0.7318,
       "step": 27023
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5515598352516814e-08,
-      "loss": 0.8247,
+      "loss": 0.8144,
       "step": 27024
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5450079516213802e-08,
-      "loss": 0.7696,
+      "loss": 0.7456,
       "step": 27025
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5384699201375087e-08,
-      "loss": 0.6959,
+      "loss": 0.7437,
       "step": 27026
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5319457408907723e-08,
-      "loss": 0.7556,
+      "loss": 0.7594,
       "step": 27027
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5254354139716542e-08,
-      "loss": 0.8294,
+      "loss": 0.8443,
       "step": 27028
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5189389394704158e-08,
-      "loss": 0.7657,
+      "loss": 0.7735,
       "step": 27029
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.5124563174772068e-08,
-      "loss": 0.7998,
+      "loss": 0.7736,
       "step": 27030
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.505987548082066e-08,
-      "loss": 0.856,
+      "loss": 0.8914,
       "step": 27031
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4995326313747007e-08,
-      "loss": 0.6848,
+      "loss": 0.7346,
       "step": 27032
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4930915674445934e-08,
-      "loss": 0.7183,
+      "loss": 0.7598,
       "step": 27033
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4866643563811178e-08,
-      "loss": 0.8062,
+      "loss": 0.8054,
       "step": 27034
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4802509982734247e-08,
-      "loss": 0.777,
+      "loss": 0.8037,
       "step": 27035
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4738514932106651e-08,
-      "loss": 0.6406,
+      "loss": 0.7491,
       "step": 27036
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4674658412814347e-08,
-      "loss": 0.8377,
+      "loss": 0.863,
       "step": 27037
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4610940425743292e-08,
-      "loss": 0.7986,
+      "loss": 0.7663,
       "step": 27038
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4547360971778335e-08,
-      "loss": 0.8432,
+      "loss": 0.8115,
       "step": 27039
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4483920051800993e-08,
-      "loss": 0.7406,
+      "loss": 0.7628,
       "step": 27040
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4420617666692783e-08,
-      "loss": 0.6794,
+      "loss": 0.7275,
       "step": 27041
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.435745381732967e-08,
-      "loss": 0.5903,
+      "loss": 0.6948,
       "step": 27042
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4294428504589841e-08,
-      "loss": 0.8751,
+      "loss": 0.84,
       "step": 27043
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4231541729345932e-08,
-      "loss": 0.8087,
+      "loss": 0.8121,
       "step": 27044
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4168793492472798e-08,
-      "loss": 0.7092,
+      "loss": 0.8331,
       "step": 27045
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4106183794838634e-08,
-      "loss": 0.7728,
+      "loss": 0.7811,
       "step": 27046
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.4043712637312745e-08,
-      "loss": 0.8091,
+      "loss": 0.8064,
       "step": 27047
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3981380020763324e-08,
-      "loss": 0.7074,
+      "loss": 0.7351,
       "step": 27048
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3919185946053016e-08,
-      "loss": 0.6602,
+      "loss": 0.7195,
       "step": 27049
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3857130414046683e-08,
-      "loss": 0.745,
+      "loss": 0.7448,
       "step": 27050
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3795213425603638e-08,
-      "loss": 0.7312,
+      "loss": 0.8036,
       "step": 27051
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3733434981583194e-08,
-      "loss": 0.8071,
+      "loss": 0.7771,
       "step": 27052
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3671795082842444e-08,
-      "loss": 0.7407,
+      "loss": 0.7411,
       "step": 27053
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3610293730236257e-08,
-      "loss": 0.7844,
+      "loss": 0.7234,
       "step": 27054
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3548930924619508e-08,
-      "loss": 0.7599,
+      "loss": 0.7352,
       "step": 27055
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3487706666842626e-08,
-      "loss": 0.7789,
+      "loss": 0.828,
       "step": 27056
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3426620957753821e-08,
-      "loss": 0.8157,
+      "loss": 0.7547,
       "step": 27057
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3365673798202416e-08,
-      "loss": 0.7606,
+      "loss": 0.809,
       "step": 27058
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.330486518903329e-08,
-      "loss": 0.7601,
+      "loss": 0.8328,
       "step": 27059
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3244195131089099e-08,
-      "loss": 0.7488,
+      "loss": 0.814,
       "step": 27060
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3183663625212506e-08,
-      "loss": 0.7975,
+      "loss": 0.7887,
       "step": 27061
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3123270672242838e-08,
-      "loss": 0.7716,
+      "loss": 0.8057,
       "step": 27062
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3063016273018314e-08,
-      "loss": 0.7989,
+      "loss": 0.7604,
       "step": 27063
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3002900428376043e-08,
-      "loss": 0.7192,
+      "loss": 0.7706,
       "step": 27064
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2942923139147579e-08,
-      "loss": 0.749,
+      "loss": 0.7438,
       "step": 27065
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2883084406166702e-08,
-      "loss": 0.6878,
+      "loss": 0.7761,
       "step": 27066
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2823384230262747e-08,
-      "loss": 0.8354,
+      "loss": 0.8282,
       "step": 27067
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2763822612265053e-08,
-      "loss": 0.7766,
+      "loss": 0.7717,
       "step": 27068
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2704399552998515e-08,
-      "loss": 0.6535,
+      "loss": 0.6795,
       "step": 27069
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2645115053289136e-08,
-      "loss": 0.8438,
+      "loss": 0.8546,
       "step": 27070
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2585969113958485e-08,
-      "loss": 0.718,
+      "loss": 0.7541,
       "step": 27071
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2526961735827015e-08,
-      "loss": 0.8008,
+      "loss": 0.764,
       "step": 27072
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2468092919712959e-08,
-      "loss": 0.8388,
+      "loss": 0.8409,
       "step": 27073
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2409362666434555e-08,
-      "loss": 0.6842,
+      "loss": 0.7448,
       "step": 27074
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2350770976805593e-08,
-      "loss": 0.7426,
+      "loss": 0.8579,
       "step": 27075
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2292317851638757e-08,
-      "loss": 0.7048,
+      "loss": 0.7724,
       "step": 27076
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2234003291745623e-08,
-      "loss": 0.6251,
+      "loss": 0.7468,
       "step": 27077
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2175827297934428e-08,
-      "loss": 0.7189,
+      "loss": 0.7772,
       "step": 27078
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.211778987101342e-08,
-      "loss": 0.7625,
+      "loss": 0.7619,
       "step": 27079
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2059891011786396e-08,
-      "loss": 0.7853,
+      "loss": 0.8127,
       "step": 27080
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.2002130721057159e-08,
-      "loss": 0.7889,
+      "loss": 0.7787,
       "step": 27081
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.1944508999628402e-08,
-      "loss": 0.6135,
+      "loss": 0.6679,
       "step": 27082
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.1887025848297262e-08,
-      "loss": 0.7119,
+      "loss": 0.7461,
       "step": 27083
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.1829681267863102e-08,
-      "loss": 0.6717,
+      "loss": 0.7308,
       "step": 27084
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1772475259120842e-08,
-      "loss": 0.7713,
+      "loss": 0.8017,
       "step": 27085
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1715407822863178e-08,
-      "loss": 0.7973,
+      "loss": 0.8464,
       "step": 27086
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1658478959883924e-08,
-      "loss": 0.8498,
+      "loss": 0.836,
       "step": 27087
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1601688670970224e-08,
-      "loss": 0.7336,
+      "loss": 0.7035,
       "step": 27088
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.154503695691256e-08,
-      "loss": 0.6679,
+      "loss": 0.7291,
       "step": 27089
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1488523818495856e-08,
-      "loss": 0.7399,
+      "loss": 0.735,
       "step": 27090
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1432149256502822e-08,
-      "loss": 0.7815,
+      "loss": 0.7683,
       "step": 27091
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1375913271717276e-08,
-      "loss": 0.758,
+      "loss": 0.7876,
       "step": 27092
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1319815864919703e-08,
-      "loss": 0.7776,
+      "loss": 0.7231,
       "step": 27093
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.126385703688726e-08,
-      "loss": 0.7558,
+      "loss": 0.8416,
       "step": 27094
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1208036788395993e-08,
-      "loss": 0.8739,
+      "loss": 0.7937,
       "step": 27095
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1152355120221947e-08,
-      "loss": 0.8519,
+      "loss": 0.7985,
       "step": 27096
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.1096812033135618e-08,
-      "loss": 0.723,
+      "loss": 0.7328,
       "step": 27097
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.104140752790861e-08,
-      "loss": 0.7843,
+      "loss": 0.7499,
       "step": 27098
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0986141605310308e-08,
-      "loss": 0.7661,
+      "loss": 0.8244,
       "step": 27099
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0931014266106766e-08,
-      "loss": 0.8179,
+      "loss": 0.7085,
       "step": 27100
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0876025511062926e-08,
-      "loss": 0.7564,
+      "loss": 0.8221,
       "step": 27101
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0821175340940404e-08,
-      "loss": 0.7503,
+      "loss": 0.7785,
       "step": 27102
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0766463756503031e-08,
-      "loss": 0.7789,
+      "loss": 0.7766,
       "step": 27103
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.071189075850687e-08,
-      "loss": 0.9369,
+      "loss": 0.841,
       "step": 27104
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0657456347710205e-08,
-      "loss": 0.8116,
+      "loss": 0.7349,
       "step": 27105
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0603160524869093e-08,
-      "loss": 0.7413,
+      "loss": 0.8039,
       "step": 27106
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.054900329073516e-08,
-      "loss": 0.7533,
+      "loss": 0.7519,
       "step": 27107
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0494984646061135e-08,
-      "loss": 0.7012,
+      "loss": 0.8428,
       "step": 27108
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0441104591595308e-08,
-      "loss": 0.7317,
+      "loss": 0.7686,
       "step": 27109
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.038736312808597e-08,
-      "loss": 0.7675,
+      "loss": 0.8266,
       "step": 27110
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.033376025627919e-08,
-      "loss": 0.8161,
+      "loss": 0.8599,
       "step": 27111
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0280295976917709e-08,
-      "loss": 0.746,
+      "loss": 0.7203,
       "step": 27112
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0226970290744264e-08,
-      "loss": 0.7283,
+      "loss": 0.744,
       "step": 27113
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0173783198497156e-08,
-      "loss": 0.8181,
+      "loss": 0.8239,
       "step": 27114
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0120734700914681e-08,
-      "loss": 0.6605,
+      "loss": 0.6891,
       "step": 27115
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0067824798734027e-08,
-      "loss": 0.7321,
+      "loss": 0.7428,
       "step": 27116
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0015053492687943e-08,
-      "loss": 0.9727,
+      "loss": 0.831,
       "step": 27117
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.962420783509175e-09,
-      "loss": 0.8672,
+      "loss": 0.8176,
       "step": 27118
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.90992667192825e-09,
-      "loss": 0.8373,
+      "loss": 0.843,
       "step": 27119
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.857571158673473e-09,
-      "loss": 0.7989,
+      "loss": 0.7654,
       "step": 27120
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.805354244469823e-09,
-      "loss": 0.7971,
+      "loss": 0.8459,
       "step": 27121
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.753275930042272e-09,
-      "loss": 0.8143,
+      "loss": 0.8057,
       "step": 27122
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.701336216114688e-09,
-      "loss": 0.7595,
+      "loss": 0.8014,
       "step": 27123
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.649535103405383e-09,
-      "loss": 0.836,
+      "loss": 0.8322,
       "step": 27124
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.597872592636003e-09,
-      "loss": 0.6855,
+      "loss": 0.7595,
       "step": 27125
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.546348684520423e-09,
-      "loss": 0.675,
+      "loss": 0.6733,
       "step": 27126
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.494963379774736e-09,
-      "loss": 0.7072,
+      "loss": 0.6914,
       "step": 27127
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.443716679112813e-09,
-      "loss": 0.6136,
+      "loss": 0.7705,
       "step": 27128
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.39260858324298e-09,
-      "loss": 0.8244,
+      "loss": 0.8474,
       "step": 27129
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.341639092876886e-09,
-      "loss": 0.7821,
+      "loss": 0.8102,
       "step": 27130
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.290808208720636e-09,
-      "loss": 0.6835,
+      "loss": 0.7672,
       "step": 27131
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.240115931479221e-09,
-      "loss": 0.9523,
+      "loss": 0.9701,
       "step": 27132
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.189562261856521e-09,
-      "loss": 0.7844,
+      "loss": 0.8416,
       "step": 27133
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.139147200553088e-09,
-      "loss": 0.7772,
+      "loss": 0.7767,
       "step": 27134
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.088870748269474e-09,
-      "loss": 0.893,
+      "loss": 0.9008,
       "step": 27135
     },
     {
       "epoch": 0.99,
       "learning_rate": 9.038732905701786e-09,
-      "loss": 0.6525,
+      "loss": 0.7189,
       "step": 27136
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.988733673546135e-09,
-      "loss": 0.7555,
+      "loss": 0.8023,
       "step": 27137
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.938873052497521e-09,
-      "loss": 0.7037,
+      "loss": 0.7807,
       "step": 27138
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.889151043245392e-09,
-      "loss": 0.6929,
+      "loss": 0.6618,
       "step": 27139
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.839567646481417e-09,
-      "loss": 0.7287,
+      "loss": 0.7059,
       "step": 27140
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.790122862892824e-09,
-      "loss": 0.7555,
+      "loss": 0.7806,
       "step": 27141
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.740816693165733e-09,
-      "loss": 0.7311,
+      "loss": 0.6969,
       "step": 27142
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.691649137984038e-09,
-      "loss": 0.7126,
+      "loss": 0.77,
       "step": 27143
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.642620198029417e-09,
-      "loss": 0.8043,
+      "loss": 0.8217,
       "step": 27144
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.593729873982437e-09,
-      "loss": 0.7889,
+      "loss": 0.8072,
       "step": 27145
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.544978166522556e-09,
-      "loss": 0.9376,
+      "loss": 0.873,
       "step": 27146
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.496365076323677e-09,
-      "loss": 0.7235,
+      "loss": 0.8151,
       "step": 27147
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.447890604063036e-09,
-      "loss": 0.7684,
+      "loss": 0.8571,
       "step": 27148
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.39955475041121e-09,
-      "loss": 0.8477,
+      "loss": 0.8156,
       "step": 27149
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.351357516038772e-09,
-      "loss": 0.6817,
+      "loss": 0.7065,
       "step": 27150
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.303298901616297e-09,
-      "loss": 0.8789,
+      "loss": 0.838,
       "step": 27151
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.255378907807698e-09,
-      "loss": 0.6605,
+      "loss": 0.6956,
       "step": 27152
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.207597535280221e-09,
-      "loss": 0.8282,
+      "loss": 0.7755,
       "step": 27153
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.159954784695556e-09,
-      "loss": 0.811,
+      "loss": 0.7936,
       "step": 27154
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.112450656714289e-09,
-      "loss": 0.7807,
+      "loss": 0.7535,
       "step": 27155
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.065085151996999e-09,
-      "loss": 0.6984,
+      "loss": 0.7623,
       "step": 27156
     },
     {
       "epoch": 0.99,
       "learning_rate": 8.01785827119983e-09,
-      "loss": 0.7774,
+      "loss": 0.7776,
       "step": 27157
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.970770014977814e-09,
-      "loss": 0.7023,
+      "loss": 0.7557,
       "step": 27158
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.923820383984871e-09,
-      "loss": 0.7801,
+      "loss": 0.7409,
       "step": 27159
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.877009378871592e-09,
-      "loss": 0.7815,
+      "loss": 0.7566,
       "step": 27160
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.830337000288568e-09,
-      "loss": 0.8858,
+      "loss": 0.8824,
       "step": 27161
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.78380324888195e-09,
-      "loss": 0.8064,
+      "loss": 0.8185,
       "step": 27162
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.737408125297884e-09,
-      "loss": 0.7015,
+      "loss": 0.753,
       "step": 27163
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.691151630180305e-09,
-      "loss": 0.7106,
+      "loss": 0.7773,
       "step": 27164
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.645033764170917e-09,
-      "loss": 0.8092,
+      "loss": 0.8392,
       "step": 27165
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.599054527909211e-09,
-      "loss": 0.7871,
+      "loss": 0.7903,
       "step": 27166
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.553213922033564e-09,
-      "loss": 0.8615,
+      "loss": 0.867,
       "step": 27167
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.507511947180136e-09,
-      "loss": 0.7588,
+      "loss": 0.7737,
       "step": 27168
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.46194860398064e-09,
-      "loss": 0.6364,
+      "loss": 0.6914,
       "step": 27169
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.4165238930712365e-09,
-      "loss": 0.7992,
+      "loss": 0.741,
       "step": 27170
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.371237815078091e-09,
-      "loss": 0.7941,
+      "loss": 0.8582,
       "step": 27171
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.3260903706329215e-09,
-      "loss": 0.6672,
+      "loss": 0.778,
       "step": 27172
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.281081560359671e-09,
-      "loss": 0.7672,
+      "loss": 0.8427,
       "step": 27173
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.236211384883396e-09,
-      "loss": 0.854,
+      "loss": 0.8222,
       "step": 27174
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.191479844826932e-09,
-      "loss": 0.6913,
+      "loss": 0.7624,
       "step": 27175
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.146886940810893e-09,
-      "loss": 0.8104,
+      "loss": 0.876,
       "step": 27176
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.102432673453674e-09,
-      "loss": 0.6607,
+      "loss": 0.6676,
       "step": 27177
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.058117043372559e-09,
-      "loss": 0.6337,
+      "loss": 0.7139,
       "step": 27178
     },
     {
       "epoch": 0.99,
       "learning_rate": 7.0139400511815e-09,
-      "loss": 0.7667,
+      "loss": 0.763,
       "step": 27179
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.969901697494452e-09,
-      "loss": 0.7462,
+      "loss": 0.8138,
       "step": 27180
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.9260019829209265e-09,
-      "loss": 0.7729,
+      "loss": 0.8127,
       "step": 27181
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.882240908070436e-09,
-      "loss": 0.7025,
+      "loss": 0.7919,
       "step": 27182
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.838618473551384e-09,
-      "loss": 0.7628,
+      "loss": 0.7668,
       "step": 27183
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.79513467996773e-09,
-      "loss": 0.7556,
+      "loss": 0.7961,
       "step": 27184
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.751789527923436e-09,
-      "loss": 0.6784,
+      "loss": 0.7374,
       "step": 27185
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.708583018019132e-09,
-      "loss": 0.7069,
+      "loss": 0.6813,
       "step": 27186
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.66551515085545e-09,
-      "loss": 0.8055,
+      "loss": 0.8119,
       "step": 27187
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.622585927027469e-09,
-      "loss": 0.6807,
+      "loss": 0.7378,
       "step": 27188
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.579795347133599e-09,
-      "loss": 0.8338,
+      "loss": 0.8797,
       "step": 27189
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.53714341176559e-09,
-      "loss": 0.7064,
+      "loss": 0.7661,
       "step": 27190
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.494630121516299e-09,
-      "loss": 0.7888,
+      "loss": 0.8469,
       "step": 27191
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.452255476975256e-09,
-      "loss": 0.7413,
+      "loss": 0.7872,
       "step": 27192
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.410019478729767e-09,
-      "loss": 0.7205,
+      "loss": 0.7662,
       "step": 27193
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.367922127366033e-09,
-      "loss": 0.6874,
+      "loss": 0.7876,
       "step": 27194
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.3259634234691395e-09,
-      "loss": 0.7261,
+      "loss": 0.7734,
       "step": 27195
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.2841433676197325e-09,
-      "loss": 0.7302,
+      "loss": 0.7694,
       "step": 27196
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.242461960398461e-09,
-      "loss": 0.7064,
+      "loss": 0.7609,
       "step": 27197
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.200919202383749e-09,
-      "loss": 0.7135,
+      "loss": 0.7765,
       "step": 27198
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.159515094151802e-09,
-      "loss": 0.7095,
+      "loss": 0.7424,
       "step": 27199
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.118249636277718e-09,
-      "loss": 0.8684,
+      "loss": 0.7965,
       "step": 27200
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.0771228293332595e-09,
-      "loss": 0.6909,
+      "loss": 0.773,
       "step": 27201
     },
     {
       "epoch": 0.99,
       "learning_rate": 6.036134673889082e-09,
-      "loss": 0.707,
+      "loss": 0.6826,
       "step": 27202
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.995285170513621e-09,
-      "loss": 0.6642,
+      "loss": 0.6776,
       "step": 27203
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.954574319775308e-09,
-      "loss": 0.8036,
+      "loss": 0.7802,
       "step": 27204
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.9140021222359176e-09,
-      "loss": 0.738,
+      "loss": 0.7384,
       "step": 27205
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.8735685784605535e-09,
-      "loss": 0.7179,
+      "loss": 0.775,
       "step": 27206
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.833273689009877e-09,
-      "loss": 0.828,
+      "loss": 0.7778,
       "step": 27207
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.793117454442332e-09,
-      "loss": 0.6998,
+      "loss": 0.7843,
       "step": 27208
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.753099875316359e-09,
-      "loss": 0.7028,
+      "loss": 0.7415,
       "step": 27209
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.71322095218485e-09,
-      "loss": 0.8001,
+      "loss": 0.8091,
       "step": 27210
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.6734806856029164e-09,
-      "loss": 0.6777,
+      "loss": 0.8023,
       "step": 27211
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.633879076121229e-09,
-      "loss": 0.7775,
+      "loss": 0.8113,
       "step": 27212
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.594416124289348e-09,
-      "loss": 0.8907,
+      "loss": 0.8501,
       "step": 27213
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.555091830654613e-09,
-      "loss": 0.7389,
+      "loss": 0.8385,
       "step": 27214
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.515906195763254e-09,
-      "loss": 0.8136,
+      "loss": 0.8804,
       "step": 27215
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.4768592201581705e-09,
-      "loss": 0.754,
+      "loss": 0.8058,
       "step": 27216
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.437950904381151e-09,
-      "loss": 0.7765,
+      "loss": 0.8544,
       "step": 27217
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.399181248971763e-09,
-      "loss": 0.848,
+      "loss": 0.8687,
       "step": 27218
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.360550254468467e-09,
-      "loss": 0.7483,
+      "loss": 0.7329,
       "step": 27219
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.322057921406387e-09,
-      "loss": 0.6529,
+      "loss": 0.8078,
       "step": 27220
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.283704250321764e-09,
-      "loss": 0.7568,
+      "loss": 0.7457,
       "step": 27221
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.245489241744173e-09,
-      "loss": 0.7073,
+      "loss": 0.7385,
       "step": 27222
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.207412896204301e-09,
-      "loss": 0.8473,
+      "loss": 0.9126,
       "step": 27223
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.169475214231723e-09,
-      "loss": 0.6503,
+      "loss": 0.7926,
       "step": 27224
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.131676196351576e-09,
-      "loss": 0.7979,
+      "loss": 0.8454,
       "step": 27225
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.094015843087885e-09,
-      "loss": 0.684,
+      "loss": 0.7709,
       "step": 27226
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.0564941549646754e-09,
-      "loss": 0.7218,
+      "loss": 0.7426,
       "step": 27227
     },
     {
       "epoch": 0.99,
       "learning_rate": 5.019111132501531e-09,
-      "loss": 0.8134,
+      "loss": 0.883,
       "step": 27228
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.981866776216926e-09,
-      "loss": 0.8343,
+      "loss": 0.8428,
       "step": 27229
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.944761086627115e-09,
-      "loss": 0.755,
+      "loss": 0.7784,
       "step": 27230
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.90779406424835e-09,
-      "loss": 0.7612,
+      "loss": 0.8364,
       "step": 27231
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.870965709593556e-09,
-      "loss": 0.7423,
+      "loss": 0.7892,
       "step": 27232
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.834276023171214e-09,
-      "loss": 0.7138,
+      "loss": 0.8165,
       "step": 27233
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.797725005493137e-09,
-      "loss": 0.7592,
+      "loss": 0.9081,
       "step": 27234
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.7613126570655865e-09,
-      "loss": 0.609,
+      "loss": 0.7101,
       "step": 27235
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.725038978392604e-09,
-      "loss": 0.7539,
+      "loss": 0.8568,
       "step": 27236
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.688903969978231e-09,
-      "loss": 0.7468,
+      "loss": 0.7195,
       "step": 27237
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.652907632324288e-09,
-      "loss": 0.6716,
+      "loss": 0.7625,
       "step": 27238
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.6170499659292655e-09,
-      "loss": 0.8365,
+      "loss": 0.8886,
       "step": 27239
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.581330971291653e-09,
-      "loss": 0.7343,
+      "loss": 0.8097,
       "step": 27240
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.545750648906611e-09,
-      "loss": 0.7684,
+      "loss": 0.7879,
       "step": 27241
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.510308999267077e-09,
-      "loss": 0.7601,
+      "loss": 0.8327,
       "step": 27242
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.475006022864881e-09,
-      "loss": 0.795,
+      "loss": 0.8063,
       "step": 27243
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.4398417201907405e-09,
-      "loss": 0.6638,
+      "loss": 0.7507,
       "step": 27244
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.404816091732045e-09,
-      "loss": 0.7894,
+      "loss": 0.7603,
       "step": 27245
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.36992913797396e-09,
-      "loss": 0.7739,
+      "loss": 0.7947,
       "step": 27246
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.335180859401656e-09,
-      "loss": 0.7649,
+      "loss": 0.8494,
       "step": 27247
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.300571256496966e-09,
-      "loss": 0.7299,
+      "loss": 0.7146,
       "step": 27248
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.2661003297395084e-09,
-      "loss": 0.7157,
+      "loss": 0.8235,
       "step": 27249
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.23176807960779e-09,
-      "loss": 0.7562,
+      "loss": 0.7599,
       "step": 27250
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.197574506578095e-09,
-      "loss": 0.7059,
+      "loss": 0.7061,
       "step": 27251
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.163519611124489e-09,
-      "loss": 0.7606,
+      "loss": 0.861,
       "step": 27252
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.129603393721038e-09,
-      "loss": 0.7754,
+      "loss": 0.7796,
       "step": 27253
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.095825854836255e-09,
-      "loss": 0.7736,
+      "loss": 0.8137,
       "step": 27254
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.062186994939765e-09,
-      "loss": 0.7626,
+      "loss": 0.7806,
       "step": 27255
     },
     {
       "epoch": 0.99,
       "learning_rate": 4.028686814497862e-09,
-      "loss": 0.7279,
+      "loss": 0.7006,
       "step": 27256
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.995325313975728e-09,
-      "loss": 0.7688,
+      "loss": 0.8239,
       "step": 27257
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.962102493835218e-09,
-      "loss": 0.7856,
+      "loss": 0.8577,
       "step": 27258
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.929018354538183e-09,
-      "loss": 0.7047,
+      "loss": 0.7997,
       "step": 27259
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.896072896544256e-09,
-      "loss": 0.7166,
+      "loss": 0.7208,
       "step": 27260
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.863266120309739e-09,
-      "loss": 0.7751,
+      "loss": 0.7259,
       "step": 27261
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.830598026288712e-09,
-      "loss": 0.6682,
+      "loss": 0.7648,
       "step": 27262
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.798068614936368e-09,
-      "loss": 0.7745,
+      "loss": 0.7817,
       "step": 27263
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.765677886702346e-09,
-      "loss": 0.6625,
+      "loss": 0.7499,
       "step": 27264
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.733425842037397e-09,
-      "loss": 0.7787,
+      "loss": 0.7726,
       "step": 27265
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.701312481387831e-09,
-      "loss": 0.8139,
+      "loss": 0.8979,
       "step": 27266
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.6693378051999574e-09,
-      "loss": 0.6679,
+      "loss": 0.717,
       "step": 27267
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.637501813916755e-09,
-      "loss": 0.6537,
+      "loss": 0.6958,
       "step": 27268
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.605804507980093e-09,
-      "loss": 0.6349,
+      "loss": 0.7117,
       "step": 27269
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.5742458878307294e-09,
-      "loss": 0.783,
+      "loss": 0.7937,
       "step": 27270
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.5428259539049826e-09,
-      "loss": 0.7796,
+      "loss": 0.7741,
       "step": 27271
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.5115447066402795e-09,
-      "loss": 0.7927,
+      "loss": 0.81,
       "step": 27272
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.480402146468498e-09,
-      "loss": 0.7594,
+      "loss": 0.8107,
       "step": 27273
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.4493982738237342e-09,
-      "loss": 0.8195,
+      "loss": 0.8344,
       "step": 27274
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.4185330891345347e-09,
-      "loss": 0.7692,
+      "loss": 0.8232,
       "step": 27275
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.387806592830556e-09,
-      "loss": 0.8159,
+      "loss": 0.7188,
       "step": 27276
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.3572187853370132e-09,
-      "loss": 0.8082,
+      "loss": 0.8261,
       "step": 27277
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.3267696670791215e-09,
-      "loss": 0.7977,
+      "loss": 0.8397,
       "step": 27278
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.2964592384787664e-09,
-      "loss": 0.688,
+      "loss": 0.7528,
       "step": 27279
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.266287499956722e-09,
-      "loss": 0.6864,
+      "loss": 0.7745,
       "step": 27280
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.2362544519304316e-09,
-      "loss": 0.7886,
+      "loss": 0.8273,
       "step": 27281
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.20636009481845e-09,
-      "loss": 0.6776,
+      "loss": 0.796,
       "step": 27282
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.176604429034891e-09,
-      "loss": 0.8188,
+      "loss": 0.8419,
       "step": 27283
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.1469874549916455e-09,
-      "loss": 0.702,
+      "loss": 0.7831,
       "step": 27284
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.117509173100608e-09,
-      "loss": 0.7258,
+      "loss": 0.7106,
       "step": 27285
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.0881695837703396e-09,
-      "loss": 0.7544,
+      "loss": 0.8039,
       "step": 27286
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.0589686874082926e-09,
-      "loss": 0.7957,
+      "loss": 0.8223,
       "step": 27287
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.029906484418588e-09,
-      "loss": 0.732,
+      "loss": 0.8,
       "step": 27288
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.0009829752053466e-09,
-      "loss": 0.8382,
+      "loss": 0.7581,
       "step": 27289
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.9721981601693597e-09,
-      "loss": 0.6525,
+      "loss": 0.7744,
       "step": 27290
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.9435520397103067e-09,
-      "loss": 0.6881,
+      "loss": 0.7568,
       "step": 27291
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.9150446142256484e-09,
-      "loss": 0.5678,
+      "loss": 0.7091,
       "step": 27292
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.8866758841106237e-09,
-      "loss": 0.7665,
+      "loss": 0.7538,
       "step": 27293
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.8584458497593613e-09,
-      "loss": 0.7007,
+      "loss": 0.6985,
       "step": 27294
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.8303545115626607e-09,
-      "loss": 0.7939,
+      "loss": 0.801,
       "step": 27295
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.8024018699113196e-09,
-      "loss": 0.84,
+      "loss": 0.8466,
       "step": 27296
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.7745879251916964e-09,
-      "loss": 0.7849,
+      "loss": 0.7682,
       "step": 27297
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.7469126777901476e-09,
-      "loss": 0.8887,
+      "loss": 0.8245,
       "step": 27298
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.719376128091922e-09,
-      "loss": 0.8212,
+      "loss": 0.8306,
       "step": 27299
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.6919782764778247e-09,
-      "loss": 0.7184,
+      "loss": 0.7204,
       "step": 27300
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.6647191233275525e-09,
-      "loss": 0.7171,
+      "loss": 0.7765,
       "step": 27301
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.6375986690208023e-09,
-      "loss": 0.843,
+      "loss": 0.8237,
       "step": 27302
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.6106169139328285e-09,
-      "loss": 0.7304,
+      "loss": 0.7667,
       "step": 27303
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.583773858437777e-09,
-      "loss": 0.7321,
+      "loss": 0.7597,
       "step": 27304
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.557069502908682e-09,
-      "loss": 0.7397,
+      "loss": 0.7831,
       "step": 27305
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.530503847715249e-09,
-      "loss": 0.7573,
+      "loss": 0.7787,
       "step": 27306
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.5040768932260705e-09,
-      "loss": 0.865,
+      "loss": 0.7952,
       "step": 27307
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.4777886398097416e-09,
-      "loss": 0.7302,
+      "loss": 0.7988,
       "step": 27308
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.451639087828195e-09,
-      "loss": 0.7314,
+      "loss": 0.7246,
       "step": 27309
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.425628237645583e-09,
-      "loss": 0.797,
+      "loss": 0.7592,
       "step": 27310
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.3997560896238394e-09,
-      "loss": 0.7495,
+      "loss": 0.7586,
       "step": 27311
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.3740226441193448e-09,
-      "loss": 0.7387,
+      "loss": 0.7309,
       "step": 27312
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.348427901490702e-09,
-      "loss": 0.7763,
+      "loss": 0.7996,
       "step": 27313
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.322971862094292e-09,
-      "loss": 0.8371,
+      "loss": 0.8983,
       "step": 27314
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.2976545262798354e-09,
-      "loss": 0.7353,
+      "loss": 0.7342,
       "step": 27315
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.2724758944014935e-09,
-      "loss": 0.7757,
+      "loss": 0.756,
       "step": 27316
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.247435966807876e-09,
-      "loss": 0.6962,
+      "loss": 0.7426,
       "step": 27317
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.2225347438453726e-09,
-      "loss": 0.8389,
+      "loss": 0.865,
       "step": 27318
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.1977722258614832e-09,
-      "loss": 0.6718,
+      "loss": 0.7474,
       "step": 27319
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.173148413197046e-09,
-      "loss": 0.7372,
+      "loss": 0.8073,
       "step": 27320
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.14866330619623e-09,
-      "loss": 0.7129,
+      "loss": 0.771,
       "step": 27321
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.124316905196544e-09,
-      "loss": 0.8285,
+      "loss": 0.9076,
       "step": 27322
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.100109210537715e-09,
-      "loss": 0.8158,
+      "loss": 0.8771,
       "step": 27323
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.0760402225539213e-09,
-      "loss": 0.6619,
+      "loss": 0.7314,
       "step": 27324
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.052109941580449e-09,
-      "loss": 0.6096,
+      "loss": 0.7209,
       "step": 27325
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.0283183679492557e-09,
-      "loss": 0.7308,
+      "loss": 0.7785,
       "step": 27326
     },
     {
       "epoch": 0.99,
       "learning_rate": 2.004665501988967e-09,
-      "loss": 0.8189,
+      "loss": 0.7982,
       "step": 27327
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.9811513440282092e-09,
-      "loss": 0.8917,
+      "loss": 0.8737,
       "step": 27328
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.957775894394498e-09,
-      "loss": 0.7463,
+      "loss": 0.7991,
       "step": 27329
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.934539153410908e-09,
-      "loss": 0.7531,
+      "loss": 0.8387,
       "step": 27330
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.9114411214005146e-09,
-      "loss": 0.6968,
+      "loss": 0.7672,
       "step": 27331
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.8884817986830617e-09,
-      "loss": 0.7114,
+      "loss": 0.747,
       "step": 27332
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.8656611855782937e-09,
-      "loss": 0.6372,
+      "loss": 0.7059,
       "step": 27333
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.8429792824015136e-09,
-      "loss": 0.8004,
+      "loss": 0.8102,
       "step": 27334
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.8204360894680252e-09,
-      "loss": 0.7256,
+      "loss": 0.6818,
       "step": 27335
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.7980316070898007e-09,
-      "loss": 0.6877,
+      "loss": 0.7235,
       "step": 27336
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.7757658355788131e-09,
-      "loss": 0.6661,
+      "loss": 0.6949,
       "step": 27337
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.7536387752425943e-09,
-      "loss": 0.7802,
+      "loss": 0.8525,
       "step": 27338
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.7316504263897859e-09,
-      "loss": 0.7271,
+      "loss": 0.7721,
       "step": 27339
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.7098007893245893e-09,
-      "loss": 0.7517,
+      "loss": 0.8003,
       "step": 27340
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.6880898643500954e-09,
-      "loss": 0.6988,
+      "loss": 0.7604,
       "step": 27341
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.6665176517682847e-09,
-      "loss": 0.7824,
+      "loss": 0.8503,
       "step": 27342
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.645084151876697e-09,
-      "loss": 0.9103,
+      "loss": 0.8699,
       "step": 27343
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.6237893649739823e-09,
-      "loss": 0.7508,
+      "loss": 0.8124,
       "step": 27344
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.6026332913554598e-09,
-      "loss": 0.7332,
+      "loss": 0.7967,
       "step": 27345
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.581615931315339e-09,
-      "loss": 0.6421,
+      "loss": 0.732,
       "step": 27346
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.5607372851433878e-09,
-      "loss": 0.8173,
+      "loss": 0.8122,
       "step": 27347
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.5399973531304845e-09,
-      "loss": 0.7551,
+      "loss": 0.8819,
       "step": 27348
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.5193961355641773e-09,
-      "loss": 0.727,
+      "loss": 0.7729,
       "step": 27349
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.4989336327297932e-09,
-      "loss": 0.6878,
+      "loss": 0.7408,
       "step": 27350
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.4786098449126596e-09,
-      "loss": 0.7995,
+      "loss": 0.7419,
       "step": 27351
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.458424772393663e-09,
-      "loss": 0.7651,
+      "loss": 0.836,
       "step": 27352
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.4383784154525793e-09,
-      "loss": 0.8474,
+      "loss": 0.8341,
       "step": 27353
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.4184707743680748e-09,
-      "loss": 0.7921,
+      "loss": 0.794,
       "step": 27354
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3987018494154847e-09,
-      "loss": 0.7527,
+      "loss": 0.7869,
       "step": 27355
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3790716408701443e-09,
-      "loss": 0.7369,
+      "loss": 0.7133,
       "step": 27356
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3595801490040582e-09,
-      "loss": 0.7513,
+      "loss": 0.7734,
       "step": 27357
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.340227374088121e-09,
-      "loss": 0.717,
+      "loss": 0.7065,
       "step": 27358
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3210133163898964e-09,
-      "loss": 0.697,
+      "loss": 0.7839,
       "step": 27359
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3019379761758378e-09,
-      "loss": 0.7002,
+      "loss": 0.7711,
       "step": 27360
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2830013537112883e-09,
-      "loss": 0.7973,
+      "loss": 0.8624,
       "step": 27361
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2642034492593713e-09,
-      "loss": 0.8016,
+      "loss": 0.7352,
       "step": 27362
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2455442630798786e-09,
-      "loss": 0.5715,
+      "loss": 0.6434,
       "step": 27363
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2270237954314924e-09,
-      "loss": 0.7258,
+      "loss": 0.74,
       "step": 27364
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2086420465728943e-09,
-      "loss": 0.7571,
+      "loss": 0.823,
       "step": 27365
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1903990167572156e-09,
-      "loss": 0.8684,
+      "loss": 0.8521,
       "step": 27366
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1722947062386968e-09,
-      "loss": 0.8897,
+      "loss": 0.8112,
       "step": 27367
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.154329115268249e-09,
-      "loss": 0.6309,
+      "loss": 0.6779,
       "step": 27368
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1365022440945616e-09,
-      "loss": 0.588,
+      "loss": 0.6385,
       "step": 27369
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1188140929652146e-09,
-      "loss": 0.8092,
+      "loss": 0.7634,
       "step": 27370
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1012646621255674e-09,
-      "loss": 0.6813,
+      "loss": 0.68,
       "step": 27371
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0838539518198686e-09,
-      "loss": 0.8204,
+      "loss": 0.8636,
       "step": 27372
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0665819622890372e-09,
-      "loss": 0.7882,
+      "loss": 0.8326,
       "step": 27373
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0494486937728809e-09,
-      "loss": 0.8504,
+      "loss": 0.783,
       "step": 27374
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0324541465089876e-09,
-      "loss": 0.7532,
+      "loss": 0.7762,
       "step": 27375
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0155983207327247e-09,
-      "loss": 0.7972,
+      "loss": 0.7837,
       "step": 27376
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.98881216678349e-10,
-      "loss": 0.7627,
+      "loss": 0.8564,
       "step": 27377
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.823028345778973e-10,
-      "loss": 0.6242,
+      "loss": 0.6445,
       "step": 27378
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.658631746611858e-10,
-      "loss": 0.7554,
+      "loss": 0.7476,
       "step": 27379
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.4956223715581e-10,
-      "loss": 0.8958,
+      "loss": 0.8859,
       "step": 27380
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.334000222893658e-10,
-      "loss": 0.7752,
+      "loss": 0.7577,
       "step": 27381
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.173765302850079e-10,
-      "loss": 0.6736,
+      "loss": 0.7604,
       "step": 27382
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.014917613647812e-10,
-      "loss": 0.7239,
+      "loss": 0.7843,
       "step": 27383
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.857457157496196e-10,
-      "loss": 0.7465,
+      "loss": 0.761,
       "step": 27384
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.701383936582375e-10,
-      "loss": 0.7143,
+      "loss": 0.7927,
       "step": 27385
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.546697953071282e-10,
-      "loss": 0.7436,
+      "loss": 0.778,
       "step": 27386
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.393399209094543e-10,
-      "loss": 0.6685,
+      "loss": 0.7512,
       "step": 27387
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.241487706794893e-10,
-      "loss": 0.731,
+      "loss": 0.7523,
       "step": 27388
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.090963448270651e-10,
-      "loss": 0.8115,
+      "loss": 0.8146,
       "step": 27389
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.941826435620136e-10,
-      "loss": 0.6561,
+      "loss": 0.7234,
       "step": 27390
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.794076670908369e-10,
-      "loss": 0.7198,
+      "loss": 0.7931,
       "step": 27391
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.647714156178154e-10,
-      "loss": 0.8459,
+      "loss": 0.9319,
       "step": 27392
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.502738893472305e-10,
-      "loss": 0.6538,
+      "loss": 0.7253,
       "step": 27393
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.359150884789223e-10,
-      "loss": 0.7901,
+      "loss": 0.8376,
       "step": 27394
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.216950132127309e-10,
-      "loss": 0.7521,
+      "loss": 0.7721,
       "step": 27395
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.076136637451659e-10,
-      "loss": 0.8066,
+      "loss": 0.7712,
       "step": 27396
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.936710402738466e-10,
-      "loss": 0.774,
+      "loss": 0.8128,
       "step": 27397
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.798671429897319e-10,
-      "loss": 0.8091,
+      "loss": 0.7067,
       "step": 27398
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.6620197208489e-10,
-      "loss": 0.8206,
+      "loss": 0.7905,
       "step": 27399
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.526755277502795e-10,
-      "loss": 0.9483,
+      "loss": 0.8656,
       "step": 27400
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.392878101724176e-10,
-      "loss": 0.7167,
+      "loss": 0.771,
       "step": 27401
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.260388195367117e-10,
-      "loss": 0.7259,
+      "loss": 0.8118,
       "step": 27402
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.12928556027459e-10,
-      "loss": 0.791,
+      "loss": 0.7953,
       "step": 27403
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.999570198267357e-10,
-      "loss": 0.7283,
+      "loss": 0.8044,
       "step": 27404
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.871242111143982e-10,
-      "loss": 0.7834,
+      "loss": 0.8007,
       "step": 27405
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.744301300691923e-10,
-      "loss": 0.7103,
+      "loss": 0.7733,
       "step": 27406
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.618747768654231e-10,
-      "loss": 0.7192,
+      "loss": 0.8513,
       "step": 27407
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.494581516785058e-10,
-      "loss": 0.672,
+      "loss": 0.7426,
       "step": 27408
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.37180254680525e-10,
-      "loss": 0.877,
+      "loss": 0.8833,
       "step": 27409
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.250410860413447e-10,
-      "loss": 0.7823,
+      "loss": 0.7975,
       "step": 27410
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.13040645930829e-10,
-      "loss": 0.729,
+      "loss": 0.7129,
       "step": 27411
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.011789345132911e-10,
-      "loss": 0.6898,
+      "loss": 0.7662,
       "step": 27412
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.894559519552644e-10,
-      "loss": 0.7237,
+      "loss": 0.7189,
       "step": 27413
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.778716984177311e-10,
-      "loss": 0.6914,
+      "loss": 0.7127,
       "step": 27414
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.664261740638942e-10,
-      "loss": 0.6766,
+      "loss": 0.7468,
       "step": 27415
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.551193790491848e-10,
-      "loss": 0.8691,
+      "loss": 0.804,
       "step": 27416
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.4395131353347497e-10,
-      "loss": 0.8958,
+      "loss": 0.889,
       "step": 27417
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.3292197766997554e-10,
-      "loss": 0.7466,
+      "loss": 0.7354,
       "step": 27418
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.2203137161189733e-10,
-      "loss": 0.7928,
+      "loss": 0.8561,
       "step": 27419
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.112794955102306e-10,
-      "loss": 0.7319,
+      "loss": 0.769,
       "step": 27420
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.006663495148555e-10,
-      "loss": 0.7303,
+      "loss": 0.7904,
       "step": 27421
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.901919337734317e-10,
-      "loss": 0.8102,
+      "loss": 0.8741,
       "step": 27422
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.798562484291779e-10,
-      "loss": 0.8987,
+      "loss": 0.8842,
       "step": 27423
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.696592936275334e-10,
-      "loss": 0.8112,
+      "loss": 0.8553,
       "step": 27424
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.5960106950949646e-10,
-      "loss": 0.7824,
+      "loss": 0.8036,
       "step": 27425
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.49681576213845e-10,
-      "loss": 0.7879,
+      "loss": 0.7751,
       "step": 27426
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.399008138793569e-10,
-      "loss": 0.7031,
+      "loss": 0.7365,
       "step": 27427
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.3025878264036915e-10,
-      "loss": 0.7999,
+      "loss": 0.8155,
       "step": 27428
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.207554826323289e-10,
-      "loss": 0.7115,
+      "loss": 0.7757,
       "step": 27429
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.113909139851323e-10,
-      "loss": 0.8026,
+      "loss": 0.8201,
       "step": 27430
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.021650768297857e-10,
-      "loss": 0.7984,
+      "loss": 0.7415,
       "step": 27431
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.9307797129507485e-10,
-      "loss": 0.6942,
+      "loss": 0.6867,
       "step": 27432
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.841295975053449e-10,
-      "loss": 0.7269,
+      "loss": 0.7631,
       "step": 27433
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.753199555849406e-10,
-      "loss": 0.7855,
+      "loss": 0.8341,
       "step": 27434
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.6664904565820715e-10,
-      "loss": 0.6636,
+      "loss": 0.7908,
       "step": 27435
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.581168678428281e-10,
-      "loss": 0.6222,
+      "loss": 0.7284,
       "step": 27436
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.4972342225870747e-10,
-      "loss": 0.7312,
+      "loss": 0.7085,
       "step": 27437
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.414687090224188e-10,
-      "loss": 0.7204,
+      "loss": 0.7586,
       "step": 27438
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.333527282472048e-10,
-      "loss": 0.8097,
+      "loss": 0.7684,
       "step": 27439
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.2537548004630816e-10,
-      "loss": 0.8816,
+      "loss": 0.8239,
       "step": 27440
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.1753696453075125e-10,
-      "loss": 0.8405,
+      "loss": 0.8289,
       "step": 27441
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.098371818093359e-10,
-      "loss": 0.7378,
+      "loss": 0.7958,
       "step": 27442
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.022761319886435e-10,
-      "loss": 0.8642,
+      "loss": 0.8026,
       "step": 27443
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.9485381517303503e-10,
-      "loss": 0.8958,
+      "loss": 0.8773,
       "step": 27444
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.8757023146576125e-10,
-      "loss": 0.8183,
+      "loss": 0.8196,
       "step": 27445
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.8042538096896266e-10,
-      "loss": 0.7815,
+      "loss": 0.7908,
       "step": 27446
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.734192637803389e-10,
-      "loss": 0.7856,
+      "loss": 0.8337,
       "step": 27447
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.665518799975896e-10,
-      "loss": 0.8703,
+      "loss": 0.8656,
       "step": 27448
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.598232297161939e-10,
-      "loss": 0.7542,
+      "loss": 0.7679,
       "step": 27449
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.5323331302941058e-10,
-      "loss": 0.7708,
+      "loss": 0.8227,
       "step": 27450
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.467821300282779e-10,
-      "loss": 0.734,
+      "loss": 0.7357,
       "step": 27451
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.4046968080272394e-10,
-      "loss": 0.8134,
+      "loss": 0.8127,
       "step": 27452
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3429596544045631e-10,
-      "loss": 0.8087,
+      "loss": 0.7828,
       "step": 27453
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.282609840269622e-10,
-      "loss": 0.8258,
+      "loss": 0.8003,
       "step": 27454
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.223647366455083e-10,
-      "loss": 0.8099,
+      "loss": 0.8412,
       "step": 27455
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1660722337825114e-10,
-      "loss": 0.7648,
+      "loss": 0.7974,
       "step": 27456
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1098844430512678e-10,
-      "loss": 0.81,
+      "loss": 0.7574,
       "step": 27457
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0550839950385084e-10,
-      "loss": 0.7832,
+      "loss": 0.7838,
       "step": 27458
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0016708905102868e-10,
-      "loss": 0.7079,
+      "loss": 0.7275,
       "step": 27459
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.496451301993503e-11,
-      "loss": 0.6807,
+      "loss": 0.7367,
       "step": 27460
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.990067148384463e-11,
-      "loss": 0.6958,
+      "loss": 0.8018,
       "step": 27461
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.497556451159128e-11,
-      "loss": 0.7996,
+      "loss": 0.8143,
       "step": 27462
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.018919217311905e-11,
-      "loss": 0.7871,
+      "loss": 0.7587,
       "step": 27463
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.554155453393109e-11,
-      "loss": 0.7888,
+      "loss": 0.7864,
       "step": 27464
     },
     {
       "epoch": 1.0,
       "learning_rate": 7.103265165842032e-11,
-      "loss": 0.8564,
+      "loss": 0.823,
       "step": 27465
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.666248360875927e-11,
-      "loss": 0.8303,
+      "loss": 0.744,
       "step": 27466
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.243105044601016e-11,
-      "loss": 0.7809,
+      "loss": 0.7662,
       "step": 27467
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.833835223012508e-11,
-      "loss": 0.7202,
+      "loss": 0.8576,
       "step": 27468
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.438438901550491e-11,
-      "loss": 0.6951,
+      "loss": 0.7585,
       "step": 27469
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.056916085877106e-11,
-      "loss": 0.701,
+      "loss": 0.7608,
       "step": 27470
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.689266781321422e-11,
-      "loss": 0.8242,
+      "loss": 0.8556,
       "step": 27471
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.335490992768421e-11,
-      "loss": 0.836,
+      "loss": 0.7085,
       "step": 27472
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.995588725325128e-11,
-      "loss": 0.8395,
+      "loss": 0.821,
       "step": 27473
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.669559983543458e-11,
-      "loss": 0.7651,
+      "loss": 0.7807,
       "step": 27474
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.357404772086348e-11,
-      "loss": 0.6765,
+      "loss": 0.7559,
       "step": 27475
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.059123095283667e-11,
-      "loss": 0.8102,
+      "loss": 0.7939,
       "step": 27476
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.7747149571322185e-11,
-      "loss": 0.7555,
+      "loss": 0.6885,
       "step": 27477
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.5041803617398276e-11,
-      "loss": 0.7961,
+      "loss": 0.8223,
       "step": 27478
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.2475193127702298e-11,
-      "loss": 0.732,
+      "loss": 0.7477,
       "step": 27479
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.0047318137761397e-11,
-      "loss": 0.6978,
+      "loss": 0.7576,
       "step": 27480
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.7758178681992476e-11,
-      "loss": 0.6731,
+      "loss": 0.763,
       "step": 27481
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.5607774791481788e-11,
-      "loss": 0.78,
+      "loss": 0.862,
       "step": 27482
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3596106496205352e-11,
-      "loss": 0.6521,
+      "loss": 0.6631,
       "step": 27483
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1723173825028966e-11,
-      "loss": 0.7742,
+      "loss": 0.8071,
       "step": 27484
     },
     {
       "epoch": 1.0,
       "learning_rate": 9.988976802377537e-12,
-      "loss": 0.773,
+      "loss": 0.8067,
       "step": 27485
     },
     {
       "epoch": 1.0,
       "learning_rate": 8.393515452675971e-12,
-      "loss": 0.7106,
+      "loss": 0.7664,
       "step": 27486
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.9367897992389525e-12,
-      "loss": 0.7934,
+      "loss": 0.7766,
       "step": 27487
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.6187998609402716e-12,
-      "loss": 0.7016,
+      "loss": 0.7795,
       "step": 27488
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.439545655543498e-12,
-      "loss": 0.7468,
+      "loss": 0.7442,
       "step": 27489
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.3990272019224226e-12,
-      "loss": 0.7642,
+      "loss": 0.7121,
       "step": 27490
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.4972445122894984e-12,
-      "loss": 0.6418,
+      "loss": 0.7366,
       "step": 27491
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.7341975999674021e-12,
-      "loss": 0.7595,
+      "loss": 0.9398,
       "step": 27492
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.1098864760583639e-12,
-      "loss": 0.7219,
+      "loss": 0.8125,
       "step": 27493
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.243111472237218e-13,
-      "loss": 0.7283,
+      "loss": 0.7825,
       "step": 27494
     },
     {
       "epoch": 1.0,
       "learning_rate": 2.7747162234526006e-13,
-      "loss": 0.7736,
+      "loss": 0.7417,
       "step": 27495
     },
     {
       "epoch": 1.0,
       "learning_rate": 6.936790586387076e-14,
-      "loss": 0.8315,
+      "loss": 0.801,
       "step": 27496
     },
     {
       "epoch": 1.0,
       "learning_rate": 0.0,
-      "loss": 0.6452,
+      "loss": 0.6782,
       "step": 27497
     },
     {
       "epoch": 1.0,
       "step": 27497,
-      "total_flos": 9.990854113551712e+18,
-      "train_loss": 0.7777426339138307,
-      "train_runtime": 259312.2881,
-      "train_samples_per_second": 13.572,
-      "train_steps_per_second": 0.106
+      "total_flos": 4.329334477492637e+19,
+      "train_loss": 0.8715659759259421,
+      "train_runtime": 253524.1542,
+      "train_samples_per_second": 13.882,
+      "train_steps_per_second": 0.108
     }
   ],
   "max_steps": 27497,
   "num_train_epochs": 1,
-  "total_flos": 9.990854113551712e+18,
+  "total_flos": 4.329334477492637e+19,
   "trial_name": null,
   "trial_params": null
 }