diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -11,13519 +11,13519 @@
     {
       "epoch": 0.0,
       "learning_rate": 2.9411764705882356e-07,
-      "loss": 1.1113,
+      "loss": 4.0234,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 5.882352941176471e-07,
-      "loss": 1.1318,
+      "loss": 3.9707,
       "step": 2
     },
     {
       "epoch": 0.0,
       "learning_rate": 8.823529411764707e-07,
-      "loss": 1.0889,
+      "loss": 4.1055,
       "step": 3
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.1764705882352942e-06,
-      "loss": 1.0835,
+      "loss": 3.8262,
       "step": 4
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.4705882352941177e-06,
-      "loss": 1.0273,
+      "loss": 3.9375,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.7647058823529414e-06,
-      "loss": 0.9458,
+      "loss": 3.8047,
       "step": 6
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.058823529411765e-06,
-      "loss": 0.7324,
+      "loss": 3.3633,
       "step": 7
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.3529411764705885e-06,
-      "loss": 0.73,
+      "loss": 3.1504,
       "step": 8
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.647058823529412e-06,
-      "loss": 0.4011,
+      "loss": 2.0312,
       "step": 9
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.9411764705882355e-06,
-      "loss": 0.4395,
+      "loss": 1.8184,
       "step": 10
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.2352941176470594e-06,
-      "loss": 0.3674,
+      "loss": 1.6484,
       "step": 11
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.529411764705883e-06,
-      "loss": 0.2308,
+      "loss": 0.8643,
       "step": 12
     },
     {
       "epoch": 0.01,
       "learning_rate": 3.8235294117647055e-06,
-      "loss": 0.2202,
+      "loss": 0.7646,
       "step": 13
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.11764705882353e-06,
-      "loss": 0.2196,
+      "loss": 0.6567,
       "step": 14
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.411764705882353e-06,
-      "loss": 0.1855,
+      "loss": 0.439,
       "step": 15
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.705882352941177e-06,
-      "loss": 0.1979,
+      "loss": 0.2462,
       "step": 16
     },
     {
       "epoch": 0.02,
       "learning_rate": 5e-06,
-      "loss": 0.2407,
+      "loss": 0.4292,
       "step": 17
     },
     {
       "epoch": 0.02,
       "learning_rate": 5.294117647058824e-06,
-      "loss": 0.2654,
+      "loss": 0.3247,
       "step": 18
     },
     {
       "epoch": 0.02,
       "learning_rate": 5.588235294117647e-06,
-      "loss": 0.1726,
+      "loss": 0.2642,
       "step": 19
     },
     {
       "epoch": 0.02,
       "learning_rate": 5.882352941176471e-06,
-      "loss": 0.1804,
+      "loss": 0.2363,
       "step": 20
     },
     {
       "epoch": 0.02,
       "learning_rate": 6.176470588235295e-06,
-      "loss": 0.1951,
+      "loss": 0.1917,
       "step": 21
     },
     {
       "epoch": 0.02,
       "learning_rate": 6.470588235294119e-06,
-      "loss": 0.1919,
+      "loss": 0.1836,
       "step": 22
     },
     {
       "epoch": 0.02,
       "learning_rate": 6.764705882352942e-06,
-      "loss": 0.2084,
+      "loss": 0.1716,
       "step": 23
     },
     {
       "epoch": 0.02,
       "learning_rate": 7.058823529411766e-06,
-      "loss": 0.1992,
+      "loss": 0.172,
       "step": 24
     },
     {
       "epoch": 0.02,
       "learning_rate": 7.352941176470589e-06,
-      "loss": 0.1842,
+      "loss": 0.167,
       "step": 25
     },
     {
       "epoch": 0.02,
       "learning_rate": 7.647058823529411e-06,
-      "loss": 0.1824,
+      "loss": 0.1665,
       "step": 26
     },
     {
       "epoch": 0.02,
       "learning_rate": 7.941176470588236e-06,
-      "loss": 0.1721,
+      "loss": 0.1688,
       "step": 27
     },
     {
       "epoch": 0.02,
       "learning_rate": 8.23529411764706e-06,
-      "loss": 0.1655,
+      "loss": 0.1661,
       "step": 28
     },
     {
       "epoch": 0.03,
       "learning_rate": 8.529411764705883e-06,
-      "loss": 0.1689,
+      "loss": 0.1644,
       "step": 29
     },
     {
       "epoch": 0.03,
       "learning_rate": 8.823529411764707e-06,
-      "loss": 0.1698,
+      "loss": 0.1643,
       "step": 30
     },
     {
       "epoch": 0.03,
       "learning_rate": 9.11764705882353e-06,
-      "loss": 0.1692,
+      "loss": 0.1724,
       "step": 31
     },
     {
       "epoch": 0.03,
       "learning_rate": 9.411764705882354e-06,
-      "loss": 0.1622,
+      "loss": 0.1621,
       "step": 32
     },
     {
       "epoch": 0.03,
       "learning_rate": 9.705882352941177e-06,
-      "loss": 0.1528,
+      "loss": 0.1617,
       "step": 33
     },
     {
       "epoch": 0.03,
       "learning_rate": 1e-05,
-      "loss": 0.1501,
+      "loss": 0.1595,
       "step": 34
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.0294117647058823e-05,
-      "loss": 0.1526,
+      "loss": 0.1687,
       "step": 35
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.0588235294117648e-05,
-      "loss": 0.1638,
+      "loss": 0.1625,
       "step": 36
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.0882352941176471e-05,
-      "loss": 0.1587,
+      "loss": 0.161,
       "step": 37
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.1176470588235295e-05,
-      "loss": 0.1531,
+      "loss": 0.1581,
       "step": 38
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.1470588235294118e-05,
-      "loss": 0.1321,
+      "loss": 0.1583,
       "step": 39
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.1764705882352942e-05,
-      "loss": 0.1366,
+      "loss": 0.1707,
       "step": 40
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.2058823529411765e-05,
-      "loss": 0.1357,
+      "loss": 0.1569,
       "step": 41
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.235294117647059e-05,
-      "loss": 0.1288,
+      "loss": 0.1621,
       "step": 42
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.2647058823529412e-05,
-      "loss": 0.1329,
+      "loss": 0.1658,
       "step": 43
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.2941176470588238e-05,
-      "loss": 0.1343,
+      "loss": 0.1627,
       "step": 44
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.323529411764706e-05,
-      "loss": 0.1306,
+      "loss": 0.1561,
       "step": 45
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.3529411764705885e-05,
-      "loss": 0.1265,
+      "loss": 0.1732,
       "step": 46
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.3823529411764706e-05,
-      "loss": 0.1289,
+      "loss": 0.1625,
       "step": 47
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.4117647058823532e-05,
-      "loss": 0.1187,
+      "loss": 0.1649,
       "step": 48
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.4411764705882353e-05,
-      "loss": 0.1292,
+      "loss": 0.1625,
       "step": 49
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.4705882352941179e-05,
-      "loss": 0.1156,
+      "loss": 0.1609,
       "step": 50
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.1257,
+      "loss": 0.1648,
       "step": 51
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.5294117647058822e-05,
-      "loss": 0.1259,
+      "loss": 0.162,
       "step": 52
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.558823529411765e-05,
-      "loss": 0.1531,
+      "loss": 0.161,
       "step": 53
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.5882352941176473e-05,
-      "loss": 0.1203,
+      "loss": 0.1622,
       "step": 54
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.6176470588235296e-05,
-      "loss": 0.1143,
+      "loss": 0.1617,
       "step": 55
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.647058823529412e-05,
-      "loss": 0.1235,
+      "loss": 0.1616,
       "step": 56
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.6764705882352943e-05,
-      "loss": 0.0975,
+      "loss": 0.1697,
       "step": 57
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.7058823529411767e-05,
-      "loss": 0.1176,
+      "loss": 0.1587,
       "step": 58
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.735294117647059e-05,
-      "loss": 0.1244,
+      "loss": 0.1639,
       "step": 59
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.7647058823529414e-05,
-      "loss": 0.1138,
+      "loss": 0.163,
       "step": 60
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.7941176470588237e-05,
-      "loss": 0.1202,
+      "loss": 0.1635,
       "step": 61
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.823529411764706e-05,
-      "loss": 0.1124,
+      "loss": 0.1628,
       "step": 62
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.8529411764705884e-05,
-      "loss": 0.1094,
+      "loss": 0.1592,
       "step": 63
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.8823529411764708e-05,
-      "loss": 0.0987,
+      "loss": 0.1581,
       "step": 64
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.911764705882353e-05,
-      "loss": 0.1213,
+      "loss": 0.1635,
       "step": 65
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9411764705882355e-05,
-      "loss": 0.1146,
+      "loss": 0.1558,
       "step": 66
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9705882352941178e-05,
-      "loss": 0.1152,
+      "loss": 0.1544,
       "step": 67
     },
     {
       "epoch": 0.06,
       "learning_rate": 2e-05,
-      "loss": 0.1021,
+      "loss": 0.1622,
       "step": 68
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.999998963521776e-05,
-      "loss": 0.1083,
+      "loss": 0.162,
       "step": 69
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9999958540892526e-05,
-      "loss": 0.1056,
+      "loss": 0.1615,
       "step": 70
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.999990671708875e-05,
-      "loss": 0.0933,
+      "loss": 0.1615,
       "step": 71
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9999834163913866e-05,
-      "loss": 0.1111,
+      "loss": 0.1608,
       "step": 72
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.999974088151827e-05,
-      "loss": 0.1053,
+      "loss": 0.16,
       "step": 73
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9999626870095335e-05,
-      "loss": 0.1038,
+      "loss": 0.1633,
       "step": 74
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.99994921298814e-05,
-      "loss": 0.0944,
+      "loss": 0.1569,
       "step": 75
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.999933666115578e-05,
-      "loss": 0.0914,
+      "loss": 0.1573,
       "step": 76
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9999160464240748e-05,
-      "loss": 0.0939,
+      "loss": 0.165,
       "step": 77
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9998963539501557e-05,
-      "loss": 0.0862,
+      "loss": 0.1573,
       "step": 78
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9998745887346422e-05,
-      "loss": 0.0941,
+      "loss": 0.1602,
       "step": 79
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9998507508226525e-05,
-      "loss": 0.0933,
+      "loss": 0.1614,
       "step": 80
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.999824840263602e-05,
-      "loss": 0.1001,
+      "loss": 0.1569,
       "step": 81
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9997968571112015e-05,
-      "loss": 0.0979,
+      "loss": 0.1573,
       "step": 82
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9997668014234594e-05,
-      "loss": 0.0981,
+      "loss": 0.161,
       "step": 83
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9997346732626797e-05,
-      "loss": 0.0901,
+      "loss": 0.1552,
       "step": 84
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9997004726954625e-05,
-      "loss": 0.1023,
+      "loss": 0.1536,
       "step": 85
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9996641997927045e-05,
-      "loss": 0.1096,
+      "loss": 0.1503,
       "step": 86
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9996258546295975e-05,
-      "loss": 0.1008,
+      "loss": 0.1512,
       "step": 87
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.999585437285629e-05,
-      "loss": 0.0793,
+      "loss": 0.162,
       "step": 88
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9995429478445833e-05,
-      "loss": 0.088,
+      "loss": 0.1506,
       "step": 89
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9994983863945387e-05,
-      "loss": 0.0865,
+      "loss": 0.1445,
       "step": 90
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9994517530278688e-05,
-      "loss": 0.0845,
+      "loss": 0.1611,
       "step": 91
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9994030478412432e-05,
-      "loss": 0.0948,
+      "loss": 0.157,
       "step": 92
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.999352270935625e-05,
-      "loss": 0.0925,
+      "loss": 0.1444,
       "step": 93
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.999299422416273e-05,
-      "loss": 0.097,
+      "loss": 0.1492,
       "step": 94
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9992445023927395e-05,
-      "loss": 0.0943,
+      "loss": 0.136,
       "step": 95
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.999187510978872e-05,
-      "loss": 0.0852,
+      "loss": 0.1389,
       "step": 96
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9991284482928102e-05,
-      "loss": 0.0984,
+      "loss": 0.1465,
       "step": 97
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.999067314456989e-05,
-      "loss": 0.0991,
+      "loss": 0.1495,
       "step": 98
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9990041095981364e-05,
-      "loss": 0.0904,
+      "loss": 0.1471,
       "step": 99
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9989388338472732e-05,
-      "loss": 0.0864,
+      "loss": 0.1504,
       "step": 100
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9988714873397128e-05,
-      "loss": 0.093,
+      "loss": 0.1324,
       "step": 101
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.998802070215062e-05,
-      "loss": 0.0924,
+      "loss": 0.1526,
       "step": 102
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9987305826172194e-05,
-      "loss": 0.0903,
+      "loss": 0.1407,
       "step": 103
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9986570246943756e-05,
-      "loss": 0.0854,
+      "loss": 0.1439,
       "step": 104
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.998581396599013e-05,
-      "loss": 0.0886,
+      "loss": 0.1386,
       "step": 105
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9985036984879053e-05,
-      "loss": 0.0816,
+      "loss": 0.1309,
       "step": 106
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.998423930522117e-05,
-      "loss": 0.0915,
+      "loss": 0.1326,
       "step": 107
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9983420928670044e-05,
-      "loss": 0.0808,
+      "loss": 0.1222,
       "step": 108
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9982581856922127e-05,
-      "loss": 0.0787,
+      "loss": 0.124,
       "step": 109
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9981722091716784e-05,
-      "loss": 0.0775,
+      "loss": 0.1226,
       "step": 110
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9980841634836265e-05,
-      "loss": 0.081,
+      "loss": 0.1211,
       "step": 111
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9979940488105722e-05,
-      "loss": 0.0869,
+      "loss": 0.1274,
       "step": 112
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.997901865339319e-05,
-      "loss": 0.0735,
+      "loss": 0.1238,
       "step": 113
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.99780761326096e-05,
-      "loss": 0.0854,
+      "loss": 0.1177,
       "step": 114
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9977112927708748e-05,
-      "loss": 0.0934,
+      "loss": 0.1139,
       "step": 115
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.997612904068732e-05,
-      "loss": 0.0938,
+      "loss": 0.1232,
       "step": 116
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9975124473584862e-05,
-      "loss": 0.0928,
+      "loss": 0.1063,
       "step": 117
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9974099228483813e-05,
-      "loss": 0.0948,
+      "loss": 0.1132,
       "step": 118
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9973053307509455e-05,
-      "loss": 0.0913,
+      "loss": 0.121,
       "step": 119
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.997198671282993e-05,
-      "loss": 0.0837,
+      "loss": 0.113,
       "step": 120
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9970899446656254e-05,
-      "loss": 0.0768,
+      "loss": 0.1038,
       "step": 121
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9969791511242272e-05,
-      "loss": 0.0737,
+      "loss": 0.1065,
       "step": 122
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9968662908884693e-05,
-      "loss": 0.0793,
+      "loss": 0.1024,
       "step": 123
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9967513641923056e-05,
-      "loss": 0.0791,
+      "loss": 0.1069,
       "step": 124
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9966343712739743e-05,
-      "loss": 0.0849,
+      "loss": 0.1113,
       "step": 125
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.996515312375997e-05,
-      "loss": 0.089,
+      "loss": 0.1125,
       "step": 126
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9963941877451763e-05,
-      "loss": 0.0911,
+      "loss": 0.0963,
       "step": 127
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9962709976326004e-05,
-      "loss": 0.0887,
+      "loss": 0.1058,
       "step": 128
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9961457422936352e-05,
-      "loss": 0.0828,
+      "loss": 0.1046,
       "step": 129
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9960184219879303e-05,
-      "loss": 0.095,
+      "loss": 0.0927,
       "step": 130
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9958890369794153e-05,
-      "loss": 0.0677,
+      "loss": 0.1146,
       "step": 131
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9957575875362993e-05,
-      "loss": 0.0702,
+      "loss": 0.0983,
       "step": 132
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9956240739310716e-05,
-      "loss": 0.085,
+      "loss": 0.1125,
       "step": 133
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9954884964404995e-05,
-      "loss": 0.0919,
+      "loss": 0.1066,
       "step": 134
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.99535085534563e-05,
-      "loss": 0.0775,
+      "loss": 0.1081,
       "step": 135
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.995211150931787e-05,
-      "loss": 0.0961,
+      "loss": 0.1025,
       "step": 136
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.995069383488571e-05,
-      "loss": 0.0837,
+      "loss": 0.1,
       "step": 137
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9949255533098602e-05,
-      "loss": 0.0764,
+      "loss": 0.1055,
       "step": 138
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9947796606938085e-05,
-      "loss": 0.0745,
+      "loss": 0.1041,
       "step": 139
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9946317059428447e-05,
-      "loss": 0.0746,
+      "loss": 0.0826,
       "step": 140
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9944816893636723e-05,
-      "loss": 0.0671,
+      "loss": 0.0953,
       "step": 141
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9943296112672697e-05,
-      "loss": 0.0804,
+      "loss": 0.0947,
       "step": 142
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9941754719688876e-05,
-      "loss": 0.0753,
+      "loss": 0.0979,
       "step": 143
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.99401927178805e-05,
-      "loss": 0.0752,
+      "loss": 0.097,
       "step": 144
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.993861011048554e-05,
-      "loss": 0.0876,
+      "loss": 0.0931,
       "step": 145
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9937006900784664e-05,
-      "loss": 0.0714,
+      "loss": 0.1028,
       "step": 146
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9935383092101256e-05,
-      "loss": 0.0822,
+      "loss": 0.1025,
       "step": 147
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9933738687801405e-05,
-      "loss": 0.0733,
+      "loss": 0.0974,
       "step": 148
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9932073691293886e-05,
-      "loss": 0.0947,
+      "loss": 0.0991,
       "step": 149
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9930388106030167e-05,
-      "loss": 0.0909,
+      "loss": 0.1064,
       "step": 150
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.992868193550439e-05,
-      "loss": 0.0753,
+      "loss": 0.0891,
       "step": 151
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9926955183253374e-05,
-      "loss": 0.0661,
+      "loss": 0.0983,
       "step": 152
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.99252078528566e-05,
-      "loss": 0.095,
+      "loss": 0.0958,
       "step": 153
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9923439947936204e-05,
-      "loss": 0.077,
+      "loss": 0.1131,
       "step": 154
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9921651472156987e-05,
-      "loss": 0.0802,
+      "loss": 0.0922,
       "step": 155
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9919842429226374e-05,
-      "loss": 0.0723,
+      "loss": 0.0908,
       "step": 156
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9918012822894427e-05,
-      "loss": 0.082,
+      "loss": 0.0999,
       "step": 157
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.991616265695385e-05,
-      "loss": 0.089,
+      "loss": 0.0976,
       "step": 158
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9914291935239952e-05,
-      "loss": 0.0822,
+      "loss": 0.0992,
       "step": 159
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9912400661630658e-05,
-      "loss": 0.0654,
+      "loss": 0.0981,
       "step": 160
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9910488840046497e-05,
-      "loss": 0.0766,
+      "loss": 0.0889,
       "step": 161
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.990855647445059e-05,
-      "loss": 0.062,
+      "loss": 0.0993,
       "step": 162
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.990660356884865e-05,
-      "loss": 0.0948,
+      "loss": 0.0989,
       "step": 163
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9904630127288958e-05,
-      "loss": 0.0641,
+      "loss": 0.089,
       "step": 164
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9902636153862383e-05,
-      "loss": 0.0806,
+      "loss": 0.0869,
       "step": 165
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9900621652702337e-05,
-      "loss": 0.0768,
+      "loss": 0.0913,
       "step": 166
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9898586627984794e-05,
-      "loss": 0.0722,
+      "loss": 0.0818,
       "step": 167
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9896531083928273e-05,
-      "loss": 0.0756,
+      "loss": 0.0869,
       "step": 168
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.989445502479383e-05,
-      "loss": 0.0714,
+      "loss": 0.0958,
       "step": 169
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9892358454885042e-05,
-      "loss": 0.0755,
+      "loss": 0.0926,
       "step": 170
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9890241378548006e-05,
-      "loss": 0.0662,
+      "loss": 0.0934,
       "step": 171
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9888103800171336e-05,
-      "loss": 0.08,
+      "loss": 0.0889,
       "step": 172
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9885945724186128e-05,
-      "loss": 0.0686,
+      "loss": 0.1038,
       "step": 173
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9883767155065987e-05,
-      "loss": 0.0673,
+      "loss": 0.0833,
       "step": 174
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9881568097326988e-05,
-      "loss": 0.0672,
+      "loss": 0.0833,
       "step": 175
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.987934855552769e-05,
-      "loss": 0.0782,
+      "loss": 0.0838,
       "step": 176
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9877108534269093e-05,
-      "loss": 0.0822,
+      "loss": 0.0807,
       "step": 177
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9874848038194673e-05,
-      "loss": 0.0749,
+      "loss": 0.0803,
       "step": 178
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.987256707199034e-05,
-      "loss": 0.0629,
+      "loss": 0.0681,
       "step": 179
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9870265640384436e-05,
-      "loss": 0.0573,
+      "loss": 0.0835,
       "step": 180
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9867943748147727e-05,
-      "loss": 0.0744,
+      "loss": 0.0898,
       "step": 181
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9865601400093395e-05,
-      "loss": 0.1124,
+      "loss": 0.0813,
       "step": 182
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9863238601077026e-05,
-      "loss": 0.0617,
+      "loss": 0.0718,
       "step": 183
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.98608553559966e-05,
-      "loss": 0.0814,
+      "loss": 0.0838,
       "step": 184
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.9858451669792478e-05,
-      "loss": 0.0666,
+      "loss": 0.0847,
       "step": 185
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.98560275474474e-05,
-      "loss": 0.079,
+      "loss": 0.0847,
       "step": 186
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9853582993986465e-05,
-      "loss": 0.0827,
+      "loss": 0.0844,
       "step": 187
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9851118014477128e-05,
-      "loss": 0.0777,
+      "loss": 0.0856,
       "step": 188
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9848632614029175e-05,
-      "loss": 0.0905,
+      "loss": 0.0894,
       "step": 189
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9846126797794743e-05,
-      "loss": 0.0689,
+      "loss": 0.0806,
       "step": 190
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9843600570968275e-05,
-      "loss": 0.0662,
+      "loss": 0.0827,
       "step": 191
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9841053938786535e-05,
-      "loss": 0.0679,
+      "loss": 0.0914,
       "step": 192
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9838486906528574e-05,
-      "loss": 0.0675,
+      "loss": 0.0854,
       "step": 193
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9835899479515738e-05,
-      "loss": 0.0687,
+      "loss": 0.0779,
       "step": 194
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.9833291663111653e-05,
-      "loss": 0.0649,
+      "loss": 0.0673,
       "step": 195
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.983066346272221e-05,
-      "loss": 0.0699,
+      "loss": 0.0762,
       "step": 196
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.982801488379555e-05,
-      "loss": 0.071,
+      "loss": 0.0711,
       "step": 197
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9825345931822068e-05,
-      "loss": 0.0644,
+      "loss": 0.0747,
       "step": 198
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.982265661233438e-05,
-      "loss": 0.0706,
+      "loss": 0.0696,
       "step": 199
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.981994693090733e-05,
-      "loss": 0.0603,
+      "loss": 0.0731,
       "step": 200
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9817216893157968e-05,
-      "loss": 0.0606,
+      "loss": 0.0771,
       "step": 201
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.981446650474555e-05,
-      "loss": 0.0663,
+      "loss": 0.0767,
       "step": 202
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.98116957713715e-05,
-      "loss": 0.0739,
+      "loss": 0.076,
       "step": 203
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9808904698779434e-05,
-      "loss": 0.0626,
+      "loss": 0.0833,
       "step": 204
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9806093292755126e-05,
-      "loss": 0.0702,
+      "loss": 0.0832,
       "step": 205
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9803261559126493e-05,
-      "loss": 0.0711,
+      "loss": 0.0883,
       "step": 206
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.98004095037636e-05,
-      "loss": 0.0688,
+      "loss": 0.076,
       "step": 207
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.9797537132578633e-05,
-      "loss": 0.0776,
+      "loss": 0.0826,
       "step": 208
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9794644451525887e-05,
-      "loss": 0.0574,
+      "loss": 0.0815,
       "step": 209
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9791731466601774e-05,
-      "loss": 0.0815,
+      "loss": 0.0788,
       "step": 210
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9788798183844778e-05,
-      "loss": 0.0617,
+      "loss": 0.0762,
       "step": 211
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.978584460933546e-05,
-      "loss": 0.0716,
+      "loss": 0.0772,
       "step": 212
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.978287074919647e-05,
-      "loss": 0.0688,
+      "loss": 0.0758,
       "step": 213
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9779876609592473e-05,
-      "loss": 0.0609,
+      "loss": 0.0916,
       "step": 214
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9776862196730195e-05,
-      "loss": 0.0642,
+      "loss": 0.0753,
       "step": 215
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9773827516858387e-05,
-      "loss": 0.0839,
+      "loss": 0.0784,
       "step": 216
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.97707725762678e-05,
-      "loss": 0.0751,
+      "loss": 0.0771,
       "step": 217
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9767697381291203e-05,
-      "loss": 0.0776,
+      "loss": 0.0718,
       "step": 218
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.9764601938303332e-05,
-      "loss": 0.0657,
+      "loss": 0.0765,
       "step": 219
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9761486253720915e-05,
-      "loss": 0.07,
+      "loss": 0.0786,
       "step": 220
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.975835033400262e-05,
-      "loss": 0.0708,
+      "loss": 0.0877,
       "step": 221
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.975519418564908e-05,
-      "loss": 0.0634,
+      "loss": 0.0737,
       "step": 222
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9752017815202848e-05,
-      "loss": 0.0628,
+      "loss": 0.0687,
       "step": 223
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9748821229248404e-05,
-      "loss": 0.0598,
+      "loss": 0.0681,
       "step": 224
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.974560443441213e-05,
-      "loss": 0.0703,
+      "loss": 0.0722,
       "step": 225
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9742367437362307e-05,
-      "loss": 0.0656,
+      "loss": 0.079,
       "step": 226
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9739110244809082e-05,
-      "loss": 0.0749,
+      "loss": 0.0717,
       "step": 227
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9735832863504474e-05,
-      "loss": 0.056,
+      "loss": 0.0762,
       "step": 228
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9732535300242358e-05,
-      "loss": 0.059,
+      "loss": 0.0878,
       "step": 229
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.9729217561858435e-05,
-      "loss": 0.0624,
+      "loss": 0.0742,
       "step": 230
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9725879655230227e-05,
-      "loss": 0.0575,
+      "loss": 0.0711,
       "step": 231
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9722521587277077e-05,
-      "loss": 0.0566,
+      "loss": 0.0806,
       "step": 232
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9719143364960107e-05,
-      "loss": 0.0708,
+      "loss": 0.0865,
       "step": 233
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9715744995282235e-05,
-      "loss": 0.065,
+      "loss": 0.0777,
       "step": 234
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.971232648528812e-05,
-      "loss": 0.0669,
+      "loss": 0.0774,
       "step": 235
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9708887842064196e-05,
-      "loss": 0.0472,
+      "loss": 0.0794,
       "step": 236
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.970542907273861e-05,
-      "loss": 0.0665,
+      "loss": 0.0714,
       "step": 237
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.970195018448125e-05,
-      "loss": 0.0557,
+      "loss": 0.0738,
       "step": 238
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9698451184503695e-05,
-      "loss": 0.0822,
+      "loss": 0.0813,
       "step": 239
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.969493208005922e-05,
-      "loss": 0.0714,
+      "loss": 0.0771,
       "step": 240
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.9691392878442776e-05,
-      "loss": 0.0643,
+      "loss": 0.0828,
       "step": 241
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9687833586990975e-05,
-      "loss": 0.0749,
+      "loss": 0.0721,
       "step": 242
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9684254213082066e-05,
-      "loss": 0.0602,
+      "loss": 0.0713,
       "step": 243
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9680654764135944e-05,
-      "loss": 0.0738,
+      "loss": 0.0789,
       "step": 244
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9677035247614108e-05,
-      "loss": 0.0716,
+      "loss": 0.08,
       "step": 245
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9673395671019653e-05,
-      "loss": 0.0594,
+      "loss": 0.0646,
       "step": 246
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.966973604189727e-05,
-      "loss": 0.061,
+      "loss": 0.0782,
       "step": 247
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9666056367833203e-05,
-      "loss": 0.0571,
+      "loss": 0.0718,
       "step": 248
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9662356656455267e-05,
-      "loss": 0.0744,
+      "loss": 0.0725,
       "step": 249
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.965863691543279e-05,
-      "loss": 0.0596,
+      "loss": 0.073,
       "step": 250
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9654897152476643e-05,
-      "loss": 0.0627,
+      "loss": 0.0737,
       "step": 251
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.965113737533918e-05,
-      "loss": 0.0549,
+      "loss": 0.0698,
       "step": 252
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.9647357591814273e-05,
-      "loss": 0.0501,
+      "loss": 0.0708,
       "step": 253
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.964355780973723e-05,
-      "loss": 0.0597,
+      "loss": 0.0595,
       "step": 254
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.963973803698485e-05,
-      "loss": 0.0678,
+      "loss": 0.0729,
       "step": 255
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9635898281475344e-05,
-      "loss": 0.0522,
+      "loss": 0.0674,
       "step": 256
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9632038551168362e-05,
-      "loss": 0.0555,
+      "loss": 0.0821,
       "step": 257
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.962815885406496e-05,
-      "loss": 0.0546,
+      "loss": 0.0654,
       "step": 258
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9624259198207572e-05,
-      "loss": 0.051,
+      "loss": 0.067,
       "step": 259
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9620339591680024e-05,
-      "loss": 0.0604,
+      "loss": 0.0654,
       "step": 260
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9616400042607486e-05,
-      "loss": 0.0584,
+      "loss": 0.0612,
       "step": 261
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9612440559156472e-05,
-      "loss": 0.0701,
+      "loss": 0.0741,
       "step": 262
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9608461149534817e-05,
-      "loss": 0.0709,
+      "loss": 0.057,
       "step": 263
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.9604461821991667e-05,
-      "loss": 0.0667,
+      "loss": 0.0611,
       "step": 264
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9600442584817454e-05,
-      "loss": 0.0733,
+      "loss": 0.0813,
       "step": 265
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9596403446343875e-05,
-      "loss": 0.0587,
+      "loss": 0.0861,
       "step": 266
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9592344414943896e-05,
-      "loss": 0.0491,
+      "loss": 0.0736,
       "step": 267
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9588265499031708e-05,
-      "loss": 0.0695,
+      "loss": 0.0753,
       "step": 268
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.958416670706273e-05,
-      "loss": 0.056,
+      "loss": 0.0742,
       "step": 269
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9580048047533578e-05,
-      "loss": 0.0581,
+      "loss": 0.0836,
       "step": 270
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9575909528982048e-05,
-      "loss": 0.0551,
+      "loss": 0.0794,
       "step": 271
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9571751159987116e-05,
-      "loss": 0.0484,
+      "loss": 0.0836,
       "step": 272
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9567572949168894e-05,
-      "loss": 0.0699,
+      "loss": 0.0753,
       "step": 273
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.9563374905188636e-05,
-      "loss": 0.0874,
+      "loss": 0.0663,
       "step": 274
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.95591570367487e-05,
-      "loss": 0.0667,
+      "loss": 0.0658,
       "step": 275
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9554919352592552e-05,
-      "loss": 0.0883,
+      "loss": 0.0855,
       "step": 276
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9550661861504718e-05,
-      "loss": 0.0651,
+      "loss": 0.0718,
       "step": 277
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.954638457231079e-05,
-      "loss": 0.0559,
+      "loss": 0.0869,
       "step": 278
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.954208749387741e-05,
-      "loss": 0.0725,
+      "loss": 0.0705,
       "step": 279
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9537770635112232e-05,
-      "loss": 0.0705,
+      "loss": 0.0728,
       "step": 280
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9533434004963913e-05,
-      "loss": 0.0573,
+      "loss": 0.0795,
       "step": 281
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9529077612422103e-05,
-      "loss": 0.058,
+      "loss": 0.0705,
       "step": 282
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.952470146651741e-05,
-      "loss": 0.0822,
+      "loss": 0.0858,
       "step": 283
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.9520305576321394e-05,
-      "loss": 0.0618,
+      "loss": 0.0682,
       "step": 284
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.951588995094655e-05,
-      "loss": 0.0675,
+      "loss": 0.0744,
       "step": 285
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.951145459954627e-05,
-      "loss": 0.0597,
+      "loss": 0.0681,
       "step": 286
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9506999531314842e-05,
-      "loss": 0.0543,
+      "loss": 0.0824,
       "step": 287
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.950252475548744e-05,
-      "loss": 0.0516,
+      "loss": 0.0699,
       "step": 288
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.949803028134007e-05,
-      "loss": 0.065,
+      "loss": 0.069,
       "step": 289
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9493516118189583e-05,
-      "loss": 0.0744,
+      "loss": 0.0747,
       "step": 290
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9488982275393644e-05,
-      "loss": 0.0695,
+      "loss": 0.0714,
       "step": 291
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.948442876235071e-05,
-      "loss": 0.0588,
+      "loss": 0.0733,
       "step": 292
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9479855588500016e-05,
-      "loss": 0.0622,
+      "loss": 0.0689,
       "step": 293
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9475262763321554e-05,
-      "loss": 0.0586,
+      "loss": 0.067,
       "step": 294
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.947065029633605e-05,
-      "loss": 0.0714,
+      "loss": 0.0776,
       "step": 295
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9466018197104945e-05,
-      "loss": 0.067,
+      "loss": 0.0705,
       "step": 296
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.946136647523038e-05,
-      "loss": 0.0632,
+      "loss": 0.075,
       "step": 297
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.9456695140355175e-05,
-      "loss": 0.0712,
+      "loss": 0.077,
       "step": 298
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9452004202162794e-05,
-      "loss": 0.0648,
+      "loss": 0.076,
       "step": 299
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9447293670377362e-05,
-      "loss": 0.0652,
+      "loss": 0.0791,
       "step": 300
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9442563554763593e-05,
-      "loss": 0.0452,
+      "loss": 0.0762,
       "step": 301
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.943781386512682e-05,
-      "loss": 0.0562,
+      "loss": 0.0731,
       "step": 302
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9433044611312937e-05,
-      "loss": 0.0635,
+      "loss": 0.0762,
       "step": 303
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.94282558032084e-05,
-      "loss": 0.0611,
+      "loss": 0.0945,
       "step": 304
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9423447450740204e-05,
-      "loss": 0.073,
+      "loss": 0.0828,
       "step": 305
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.941861956387585e-05,
-      "loss": 0.0508,
+      "loss": 0.0779,
       "step": 306
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9413772152623337e-05,
-      "loss": 0.0634,
+      "loss": 0.0804,
       "step": 307
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9408905227031142e-05,
-      "loss": 0.0777,
+      "loss": 0.0814,
       "step": 308
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.9404018797188184e-05,
-      "loss": 0.0483,
+      "loss": 0.0833,
       "step": 309
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9399112873223824e-05,
-      "loss": 0.0436,
+      "loss": 0.0724,
       "step": 310
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9394187465307828e-05,
-      "loss": 0.0584,
+      "loss": 0.0699,
       "step": 311
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9389242583650348e-05,
-      "loss": 0.0558,
+      "loss": 0.0736,
       "step": 312
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9384278238501913e-05,
-      "loss": 0.0565,
+      "loss": 0.0598,
       "step": 313
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9379294440153393e-05,
-      "loss": 0.0556,
+      "loss": 0.0622,
       "step": 314
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.937429119893598e-05,
-      "loss": 0.0465,
+      "loss": 0.0665,
       "step": 315
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9369268525221187e-05,
-      "loss": 0.0511,
+      "loss": 0.0825,
       "step": 316
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9364226429420782e-05,
-      "loss": 0.0713,
+      "loss": 0.0743,
       "step": 317
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9359164921986826e-05,
-      "loss": 0.0556,
+      "loss": 0.0704,
       "step": 318
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.935408401341159e-05,
-      "loss": 0.0465,
+      "loss": 0.0908,
       "step": 319
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.9348983714227584e-05,
-      "loss": 0.0691,
+      "loss": 0.0823,
       "step": 320
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9343864035007507e-05,
-      "loss": 0.0471,
+      "loss": 0.0779,
       "step": 321
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9338724986364223e-05,
-      "loss": 0.0585,
+      "loss": 0.0723,
       "step": 322
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9333566578950764e-05,
-      "loss": 0.0551,
+      "loss": 0.0672,
       "step": 323
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.932838882346028e-05,
-      "loss": 0.0582,
+      "loss": 0.0611,
       "step": 324
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9323191730626036e-05,
-      "loss": 0.0633,
+      "loss": 0.0917,
       "step": 325
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9317975311221374e-05,
-      "loss": 0.0657,
+      "loss": 0.0687,
       "step": 326
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.931273957605971e-05,
-      "loss": 0.0477,
+      "loss": 0.0765,
       "step": 327
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.930748453599449e-05,
-      "loss": 0.0575,
+      "loss": 0.0728,
       "step": 328
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9302210201919187e-05,
-      "loss": 0.0596,
+      "loss": 0.0792,
       "step": 329
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9296916584767263e-05,
-      "loss": 0.0561,
+      "loss": 0.0823,
       "step": 330
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.9291603695512155e-05,
-      "loss": 0.063,
+      "loss": 0.0668,
       "step": 331
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9286271545167253e-05,
-      "loss": 0.045,
+      "loss": 0.0732,
       "step": 332
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9280920144785878e-05,
-      "loss": 0.0627,
+      "loss": 0.0681,
       "step": 333
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.927554950546124e-05,
-      "loss": 0.06,
+      "loss": 0.0662,
       "step": 334
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9270159638326443e-05,
-      "loss": 0.0496,
+      "loss": 0.0816,
       "step": 335
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.926475055455445e-05,
-      "loss": 0.0632,
+      "loss": 0.0732,
       "step": 336
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.925932226535806e-05,
-      "loss": 0.0457,
+      "loss": 0.0859,
       "step": 337
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9253874781989864e-05,
-      "loss": 0.0438,
+      "loss": 0.0707,
       "step": 338
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9248408115742277e-05,
-      "loss": 0.0654,
+      "loss": 0.0728,
       "step": 339
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.924292227794745e-05,
-      "loss": 0.056,
+      "loss": 0.0949,
       "step": 340
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9237417279977287e-05,
-      "loss": 0.0505,
+      "loss": 0.0752,
       "step": 341
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.9231893133243408e-05,
-      "loss": 0.0571,
+      "loss": 0.0567,
       "step": 342
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.922634984919713e-05,
-      "loss": 0.0497,
+      "loss": 0.0587,
       "step": 343
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.922078743932944e-05,
-      "loss": 0.0551,
+      "loss": 0.0679,
       "step": 344
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.921520591517097e-05,
-      "loss": 0.0549,
+      "loss": 0.0847,
       "step": 345
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.920960528829198e-05,
-      "loss": 0.0454,
+      "loss": 0.0627,
       "step": 346
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.920398557030232e-05,
-      "loss": 0.0437,
+      "loss": 0.0721,
       "step": 347
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9198346772851423e-05,
-      "loss": 0.0494,
+      "loss": 0.0693,
       "step": 348
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9192688907628275e-05,
-      "loss": 0.0561,
+      "loss": 0.0639,
       "step": 349
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9187011986361374e-05,
-      "loss": 0.0531,
+      "loss": 0.0655,
       "step": 350
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.918131602081874e-05,
-      "loss": 0.0663,
+      "loss": 0.0806,
       "step": 351
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9175601022807862e-05,
-      "loss": 0.0529,
+      "loss": 0.0777,
       "step": 352
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9169867004175677e-05,
-      "loss": 0.0483,
+      "loss": 0.0729,
       "step": 353
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.9164113976808553e-05,
-      "loss": 0.0557,
+      "loss": 0.0748,
       "step": 354
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9158341952632272e-05,
-      "loss": 0.0439,
+      "loss": 0.0854,
       "step": 355
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9152550943611986e-05,
-      "loss": 0.0612,
+      "loss": 0.0757,
       "step": 356
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9146740961752203e-05,
-      "loss": 0.0613,
+      "loss": 0.0771,
       "step": 357
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9140912019096766e-05,
-      "loss": 0.049,
+      "loss": 0.0838,
       "step": 358
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.913506412772882e-05,
-      "loss": 0.0507,
+      "loss": 0.0764,
       "step": 359
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9129197299770782e-05,
-      "loss": 0.0598,
+      "loss": 0.0605,
       "step": 360
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9123311547384333e-05,
-      "loss": 0.0659,
+      "loss": 0.0808,
       "step": 361
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.911740688277039e-05,
-      "loss": 0.045,
+      "loss": 0.0795,
       "step": 362
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9111483318169056e-05,
-      "loss": 0.0514,
+      "loss": 0.0683,
       "step": 363
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9105540865859625e-05,
-      "loss": 0.0531,
+      "loss": 0.064,
       "step": 364
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.9099579538160542e-05,
-      "loss": 0.0457,
+      "loss": 0.0823,
       "step": 365
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.909359934742938e-05,
-      "loss": 0.0488,
+      "loss": 0.0729,
       "step": 366
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9087600306062816e-05,
-      "loss": 0.0546,
+      "loss": 0.0723,
       "step": 367
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9081582426496598e-05,
-      "loss": 0.0559,
+      "loss": 0.0729,
       "step": 368
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.907554572120553e-05,
-      "loss": 0.0596,
+      "loss": 0.0662,
       "step": 369
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9069490202703438e-05,
-      "loss": 0.071,
+      "loss": 0.0727,
       "step": 370
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.906341588354315e-05,
-      "loss": 0.0527,
+      "loss": 0.0725,
       "step": 371
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.905732277631647e-05,
-      "loss": 0.0632,
+      "loss": 0.0659,
       "step": 372
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9051210893654127e-05,
-      "loss": 0.0398,
+      "loss": 0.0756,
       "step": 373
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9045080248225804e-05,
-      "loss": 0.0554,
+      "loss": 0.0668,
       "step": 374
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.903893085274006e-05,
-      "loss": 0.0516,
+      "loss": 0.0614,
       "step": 375
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.9032762719944315e-05,
-      "loss": 0.0487,
+      "loss": 0.0701,
       "step": 376
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.9026575862624845e-05,
-      "loss": 0.0656,
+      "loss": 0.0621,
       "step": 377
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.902037029360674e-05,
-      "loss": 0.0494,
+      "loss": 0.0714,
       "step": 378
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.901414602575387e-05,
-      "loss": 0.0413,
+      "loss": 0.0572,
       "step": 379
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.900790307196887e-05,
-      "loss": 0.0607,
+      "loss": 0.0645,
       "step": 380
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.900164144519311e-05,
-      "loss": 0.0538,
+      "loss": 0.0691,
       "step": 381
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.899536115840668e-05,
-      "loss": 0.0515,
+      "loss": 0.0695,
       "step": 382
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8989062224628327e-05,
-      "loss": 0.0645,
+      "loss": 0.0768,
       "step": 383
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8982744656915475e-05,
-      "loss": 0.0621,
+      "loss": 0.0674,
       "step": 384
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8976408468364163e-05,
-      "loss": 0.039,
+      "loss": 0.0603,
       "step": 385
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8970053672109037e-05,
-      "loss": 0.0441,
+      "loss": 0.0699,
       "step": 386
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8963680281323313e-05,
-      "loss": 0.0536,
+      "loss": 0.0676,
       "step": 387
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.8957288309218745e-05,
-      "loss": 0.045,
+      "loss": 0.0622,
       "step": 388
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8950877769045623e-05,
-      "loss": 0.0462,
+      "loss": 0.0575,
       "step": 389
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8944448674092715e-05,
-      "loss": 0.0533,
+      "loss": 0.0685,
       "step": 390
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.893800103768725e-05,
-      "loss": 0.0403,
+      "loss": 0.0625,
       "step": 391
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.89315348731949e-05,
-      "loss": 0.0451,
+      "loss": 0.0668,
       "step": 392
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8925050194019747e-05,
-      "loss": 0.0473,
+      "loss": 0.0762,
       "step": 393
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.891854701360424e-05,
-      "loss": 0.048,
+      "loss": 0.0673,
       "step": 394
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8912025345429197e-05,
-      "loss": 0.0532,
+      "loss": 0.0637,
       "step": 395
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8905485203013746e-05,
-      "loss": 0.0575,
+      "loss": 0.0683,
       "step": 396
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.889892659991532e-05,
-      "loss": 0.0542,
+      "loss": 0.058,
       "step": 397
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8892349549729617e-05,
-      "loss": 0.0548,
+      "loss": 0.063,
       "step": 398
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.8885754066090575e-05,
-      "loss": 0.0739,
+      "loss": 0.0692,
       "step": 399
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8879140162670346e-05,
-      "loss": 0.0483,
+      "loss": 0.0787,
       "step": 400
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8872507853179264e-05,
-      "loss": 0.0444,
+      "loss": 0.071,
       "step": 401
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8865857151365815e-05,
-      "loss": 0.0595,
+      "loss": 0.0608,
       "step": 402
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8859188071016616e-05,
-      "loss": 0.0568,
+      "loss": 0.0659,
       "step": 403
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8852500625956382e-05,
-      "loss": 0.0472,
+      "loss": 0.0568,
       "step": 404
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8845794830047892e-05,
-      "loss": 0.0604,
+      "loss": 0.0627,
       "step": 405
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.883907069719197e-05,
-      "loss": 0.0484,
+      "loss": 0.0533,
       "step": 406
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8832328241327448e-05,
-      "loss": 0.0416,
+      "loss": 0.0581,
       "step": 407
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.882556747643115e-05,
-      "loss": 0.0468,
+      "loss": 0.0653,
       "step": 408
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8818788416517844e-05,
-      "loss": 0.0525,
+      "loss": 0.0557,
       "step": 409
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.8811991075640223e-05,
-      "loss": 0.0487,
+      "loss": 0.0789,
       "step": 410
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.880517546788888e-05,
-      "loss": 0.0483,
+      "loss": 0.0646,
       "step": 411
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8798341607392278e-05,
-      "loss": 0.0536,
+      "loss": 0.0617,
       "step": 412
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8791489508316704e-05,
-      "loss": 0.0534,
+      "loss": 0.0596,
       "step": 413
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8784619184866267e-05,
-      "loss": 0.0377,
+      "loss": 0.067,
       "step": 414
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8777730651282847e-05,
-      "loss": 0.0491,
+      "loss": 0.059,
       "step": 415
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.877082392184607e-05,
-      "loss": 0.0499,
+      "loss": 0.0626,
       "step": 416
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8763899010873293e-05,
-      "loss": 0.0469,
+      "loss": 0.0619,
       "step": 417
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8756955932719546e-05,
-      "loss": 0.0461,
+      "loss": 0.0522,
       "step": 418
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8749994701777536e-05,
-      "loss": 0.0481,
+      "loss": 0.0669,
       "step": 419
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.8743015332477587e-05,
-      "loss": 0.0434,
+      "loss": 0.0649,
       "step": 420
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.873601783928763e-05,
-      "loss": 0.0406,
+      "loss": 0.0762,
       "step": 421
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8729002236713158e-05,
-      "loss": 0.0628,
+      "loss": 0.082,
       "step": 422
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8721968539297218e-05,
-      "loss": 0.0437,
+      "loss": 0.0788,
       "step": 423
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.871491676162035e-05,
-      "loss": 0.0428,
+      "loss": 0.066,
       "step": 424
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8707846918300588e-05,
-      "loss": 0.0453,
+      "loss": 0.0603,
       "step": 425
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.870075902399341e-05,
-      "loss": 0.039,
+      "loss": 0.0695,
       "step": 426
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8693653093391706e-05,
-      "loss": 0.0547,
+      "loss": 0.0701,
       "step": 427
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8686529141225767e-05,
-      "loss": 0.0468,
+      "loss": 0.0607,
       "step": 428
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8679387182263232e-05,
-      "loss": 0.0423,
+      "loss": 0.0647,
       "step": 429
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.867222723130907e-05,
-      "loss": 0.0486,
+      "loss": 0.0774,
       "step": 430
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.866504930320555e-05,
-      "loss": 0.0545,
+      "loss": 0.0613,
       "step": 431
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8657853412832204e-05,
-      "loss": 0.0437,
+      "loss": 0.0587,
       "step": 432
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.8650639575105797e-05,
-      "loss": 0.0604,
+      "loss": 0.0761,
       "step": 433
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8643407804980303e-05,
-      "loss": 0.0415,
+      "loss": 0.0683,
       "step": 434
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.863615811744687e-05,
-      "loss": 0.0521,
+      "loss": 0.0671,
       "step": 435
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8628890527533776e-05,
-      "loss": 0.0513,
+      "loss": 0.0783,
       "step": 436
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.862160505030642e-05,
-      "loss": 0.0517,
+      "loss": 0.0662,
       "step": 437
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8614301700867287e-05,
-      "loss": 0.0469,
+      "loss": 0.0634,
       "step": 438
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8606980494355897e-05,
-      "loss": 0.057,
+      "loss": 0.0594,
       "step": 439
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8599641445948792e-05,
-      "loss": 0.0368,
+      "loss": 0.0571,
       "step": 440
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.85922845708595e-05,
-      "loss": 0.0629,
+      "loss": 0.0618,
       "step": 441
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.85849098843385e-05,
-      "loss": 0.0519,
+      "loss": 0.0584,
       "step": 442
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.85775174016732e-05,
-      "loss": 0.0466,
+      "loss": 0.0741,
       "step": 443
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.8570107138187895e-05,
-      "loss": 0.0695,
+      "loss": 0.0725,
       "step": 444
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8562679109243733e-05,
-      "loss": 0.0397,
+      "loss": 0.0455,
       "step": 445
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8555233330238703e-05,
-      "loss": 0.0387,
+      "loss": 0.0641,
       "step": 446
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8547769816607572e-05,
-      "loss": 0.0534,
+      "loss": 0.0719,
       "step": 447
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.854028858382188e-05,
-      "loss": 0.057,
+      "loss": 0.0622,
       "step": 448
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8532789647389907e-05,
-      "loss": 0.0447,
+      "loss": 0.0676,
       "step": 449
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8525273022856608e-05,
-      "loss": 0.0506,
+      "loss": 0.0629,
       "step": 450
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8517738725803628e-05,
-      "loss": 0.0518,
+      "loss": 0.054,
       "step": 451
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.851018677184923e-05,
-      "loss": 0.0422,
+      "loss": 0.0531,
       "step": 452
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.850261717664829e-05,
-      "loss": 0.0504,
+      "loss": 0.0622,
       "step": 453
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8495029955892247e-05,
-      "loss": 0.0485,
+      "loss": 0.0808,
       "step": 454
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.848742512530908e-05,
-      "loss": 0.0425,
+      "loss": 0.0628,
       "step": 455
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.847980270066327e-05,
-      "loss": 0.0333,
+      "loss": 0.0554,
       "step": 456
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8472162697755776e-05,
-      "loss": 0.0552,
+      "loss": 0.0671,
       "step": 457
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8464505132423985e-05,
-      "loss": 0.0615,
+      "loss": 0.0621,
       "step": 458
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8456830020541702e-05,
-      "loss": 0.0543,
+      "loss": 0.0598,
       "step": 459
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8449137378019095e-05,
-      "loss": 0.0431,
+      "loss": 0.0627,
       "step": 460
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.844142722080268e-05,
-      "loss": 0.0358,
+      "loss": 0.0612,
       "step": 461
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8433699564875275e-05,
-      "loss": 0.0603,
+      "loss": 0.0626,
       "step": 462
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.842595442625598e-05,
-      "loss": 0.0489,
+      "loss": 0.0608,
       "step": 463
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.841819182100012e-05,
-      "loss": 0.0482,
+      "loss": 0.0599,
       "step": 464
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.8410411765199247e-05,
-      "loss": 0.058,
+      "loss": 0.059,
       "step": 465
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.840261427498107e-05,
-      "loss": 0.0475,
+      "loss": 0.0644,
       "step": 466
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.839479936650945e-05,
-      "loss": 0.0425,
+      "loss": 0.0653,
       "step": 467
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.838696705598436e-05,
-      "loss": 0.0467,
+      "loss": 0.0625,
       "step": 468
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8379117359641826e-05,
-      "loss": 0.0677,
+      "loss": 0.0632,
       "step": 469
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.837125029375393e-05,
-      "loss": 0.0457,
+      "loss": 0.0505,
       "step": 470
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.836336587462876e-05,
-      "loss": 0.0405,
+      "loss": 0.0605,
       "step": 471
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8355464118610375e-05,
-      "loss": 0.0569,
+      "loss": 0.0547,
       "step": 472
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8347545042078764e-05,
-      "loss": 0.047,
+      "loss": 0.0501,
       "step": 473
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8339608661449833e-05,
-      "loss": 0.0477,
+      "loss": 0.0596,
       "step": 474
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8331654993175353e-05,
-      "loss": 0.0368,
+      "loss": 0.0502,
       "step": 475
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.832368405374293e-05,
-      "loss": 0.0363,
+      "loss": 0.058,
       "step": 476
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8315695859675976e-05,
-      "loss": 0.0326,
+      "loss": 0.0643,
       "step": 477
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.8307690427533673e-05,
-      "loss": 0.0474,
+      "loss": 0.0547,
       "step": 478
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8299667773910924e-05,
-      "loss": 0.043,
+      "loss": 0.0645,
       "step": 479
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.829162791543835e-05,
-      "loss": 0.0532,
+      "loss": 0.0499,
       "step": 480
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8283570868782223e-05,
-      "loss": 0.0556,
+      "loss": 0.0609,
       "step": 481
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8275496650644446e-05,
-      "loss": 0.0477,
+      "loss": 0.0579,
       "step": 482
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8267405277762528e-05,
-      "loss": 0.0455,
+      "loss": 0.0677,
       "step": 483
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8259296766909528e-05,
-      "loss": 0.043,
+      "loss": 0.0774,
       "step": 484
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.825117113489404e-05,
-      "loss": 0.0363,
+      "loss": 0.0679,
       "step": 485
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8243028398560143e-05,
-      "loss": 0.0471,
+      "loss": 0.0508,
       "step": 486
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8234868574787376e-05,
-      "loss": 0.0546,
+      "loss": 0.0582,
       "step": 487
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.8226691680490696e-05,
-      "loss": 0.0364,
+      "loss": 0.0734,
       "step": 488
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.821849773262045e-05,
-      "loss": 0.0411,
+      "loss": 0.0526,
       "step": 489
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8210286748162334e-05,
-      "loss": 0.0467,
+      "loss": 0.0675,
       "step": 490
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8202058744137365e-05,
-      "loss": 0.0342,
+      "loss": 0.0623,
       "step": 491
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8193813737601834e-05,
-      "loss": 0.0505,
+      "loss": 0.0709,
       "step": 492
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.818555174564728e-05,
-      "loss": 0.0461,
+      "loss": 0.0835,
       "step": 493
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8177272785400455e-05,
-      "loss": 0.0577,
+      "loss": 0.0573,
       "step": 494
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8168976874023282e-05,
-      "loss": 0.0483,
+      "loss": 0.0626,
       "step": 495
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.816066402871282e-05,
-      "loss": 0.0316,
+      "loss": 0.0558,
       "step": 496
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8152334266701238e-05,
-      "loss": 0.0467,
+      "loss": 0.0726,
       "step": 497
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8143987605255774e-05,
-      "loss": 0.0508,
+      "loss": 0.0615,
       "step": 498
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8135624061678688e-05,
-      "loss": 0.0534,
+      "loss": 0.0616,
       "step": 499
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.8127243653307248e-05,
-      "loss": 0.055,
+      "loss": 0.0496,
       "step": 500
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8118846397513666e-05,
-      "loss": 0.0473,
+      "loss": 0.0605,
       "step": 501
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8110432311705095e-05,
-      "loss": 0.0499,
+      "loss": 0.0771,
       "step": 502
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8102001413323567e-05,
-      "loss": 0.0417,
+      "loss": 0.063,
       "step": 503
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8093553719845967e-05,
-      "loss": 0.0459,
+      "loss": 0.0554,
       "step": 504
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8085089248783992e-05,
-      "loss": 0.0273,
+      "loss": 0.0572,
       "step": 505
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8076608017684128e-05,
-      "loss": 0.0484,
+      "loss": 0.0549,
       "step": 506
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8068110044127592e-05,
-      "loss": 0.0501,
+      "loss": 0.0607,
       "step": 507
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.805959534573032e-05,
-      "loss": 0.0426,
+      "loss": 0.0563,
       "step": 508
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.80510639401429e-05,
-      "loss": 0.0463,
+      "loss": 0.0657,
       "step": 509
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8042515845050578e-05,
-      "loss": 0.028,
+      "loss": 0.0482,
       "step": 510
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.8033951078173172e-05,
-      "loss": 0.0338,
+      "loss": 0.0677,
       "step": 511
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.8025369657265075e-05,
-      "loss": 0.0295,
+      "loss": 0.0581,
       "step": 512
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.80167716001152e-05,
-      "loss": 0.037,
+      "loss": 0.0564,
       "step": 513
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.8008156924546943e-05,
-      "loss": 0.0334,
+      "loss": 0.0677,
       "step": 514
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.799952564841815e-05,
-      "loss": 0.0455,
+      "loss": 0.0617,
       "step": 515
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.7990877789621082e-05,
-      "loss": 0.035,
+      "loss": 0.0576,
       "step": 516
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.7982213366082372e-05,
-      "loss": 0.0462,
+      "loss": 0.0714,
       "step": 517
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.7973532395763e-05,
-      "loss": 0.0304,
+      "loss": 0.0607,
       "step": 518
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.796483489665823e-05,
-      "loss": 0.0464,
+      "loss": 0.0627,
       "step": 519
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.7956120886797604e-05,
-      "loss": 0.0451,
+      "loss": 0.0619,
       "step": 520
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.794739038424488e-05,
-      "loss": 0.0412,
+      "loss": 0.0742,
       "step": 521
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.793864340709802e-05,
-      "loss": 0.0361,
+      "loss": 0.0639,
       "step": 522
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.7929879973489113e-05,
-      "loss": 0.0417,
+      "loss": 0.0615,
       "step": 523
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7921100101584387e-05,
-      "loss": 0.034,
+      "loss": 0.0591,
       "step": 524
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7912303809584126e-05,
-      "loss": 0.0451,
+      "loss": 0.0719,
       "step": 525
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7903491115722665e-05,
-      "loss": 0.0532,
+      "loss": 0.0709,
       "step": 526
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7894662038268335e-05,
-      "loss": 0.0326,
+      "loss": 0.0558,
       "step": 527
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7885816595523425e-05,
-      "loss": 0.0406,
+      "loss": 0.0625,
       "step": 528
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.787695480582415e-05,
-      "loss": 0.0334,
+      "loss": 0.0641,
       "step": 529
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7868076687540625e-05,
-      "loss": 0.0363,
+      "loss": 0.0677,
       "step": 530
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7859182259076795e-05,
-      "loss": 0.0391,
+      "loss": 0.0648,
       "step": 531
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7850271538870423e-05,
-      "loss": 0.0389,
+      "loss": 0.0752,
       "step": 532
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7841344545393044e-05,
-      "loss": 0.0411,
+      "loss": 0.0672,
       "step": 533
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.7832401297149927e-05,
-      "loss": 0.0444,
+      "loss": 0.0644,
       "step": 534
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.782344181268004e-05,
-      "loss": 0.0275,
+      "loss": 0.0538,
       "step": 535
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7814466110555998e-05,
-      "loss": 0.0337,
+      "loss": 0.0637,
       "step": 536
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7805474209384047e-05,
-      "loss": 0.053,
+      "loss": 0.0621,
       "step": 537
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7796466127804e-05,
-      "loss": 0.0413,
+      "loss": 0.0732,
       "step": 538
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.778744188448922e-05,
-      "loss": 0.0326,
+      "loss": 0.0593,
       "step": 539
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.777840149814657e-05,
-      "loss": 0.0373,
+      "loss": 0.0639,
       "step": 540
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.776934498751638e-05,
-      "loss": 0.0504,
+      "loss": 0.0614,
       "step": 541
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7760272371372404e-05,
-      "loss": 0.0267,
+      "loss": 0.0625,
       "step": 542
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.775118366852177e-05,
-      "loss": 0.0455,
+      "loss": 0.0541,
       "step": 543
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7742078897804977e-05,
-      "loss": 0.0448,
+      "loss": 0.0512,
       "step": 544
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.7732958078095807e-05,
-      "loss": 0.0324,
+      "loss": 0.0774,
       "step": 545
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7723821228301324e-05,
-      "loss": 0.0308,
+      "loss": 0.0631,
       "step": 546
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7714668367361825e-05,
-      "loss": 0.0339,
+      "loss": 0.0425,
       "step": 547
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7705499514250784e-05,
-      "loss": 0.0407,
+      "loss": 0.0525,
       "step": 548
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7696314687974844e-05,
-      "loss": 0.0364,
+      "loss": 0.0695,
       "step": 549
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7687113907573744e-05,
-      "loss": 0.0442,
+      "loss": 0.0659,
       "step": 550
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.76778971921203e-05,
-      "loss": 0.037,
+      "loss": 0.0652,
       "step": 551
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7668664560720362e-05,
-      "loss": 0.0308,
+      "loss": 0.0567,
       "step": 552
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7659416032512776e-05,
-      "loss": 0.0473,
+      "loss": 0.0613,
       "step": 553
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.765015162666934e-05,
-      "loss": 0.0354,
+      "loss": 0.0738,
       "step": 554
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7640871362394752e-05,
-      "loss": 0.054,
+      "loss": 0.0614,
       "step": 555
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.7631575258926612e-05,
-      "loss": 0.0491,
+      "loss": 0.0555,
       "step": 556
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7622263335535325e-05,
-      "loss": 0.0384,
+      "loss": 0.0538,
       "step": 557
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.761293561152411e-05,
-      "loss": 0.0388,
+      "loss": 0.0576,
       "step": 558
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.760359210622893e-05,
-      "loss": 0.0381,
+      "loss": 0.0644,
       "step": 559
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7594232839018463e-05,
-      "loss": 0.0442,
+      "loss": 0.0597,
       "step": 560
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7584857829294062e-05,
-      "loss": 0.0464,
+      "loss": 0.0578,
       "step": 561
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.757546709648972e-05,
-      "loss": 0.0359,
+      "loss": 0.0675,
       "step": 562
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7566060660072008e-05,
-      "loss": 0.0443,
+      "loss": 0.0623,
       "step": 563
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7556638539540066e-05,
-      "loss": 0.0554,
+      "loss": 0.0522,
       "step": 564
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.754720075442554e-05,
-      "loss": 0.0352,
+      "loss": 0.0545,
       "step": 565
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7537747324292542e-05,
-      "loss": 0.0486,
+      "loss": 0.0483,
       "step": 566
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7528278268737624e-05,
-      "loss": 0.0439,
+      "loss": 0.0506,
       "step": 567
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.7518793607389725e-05,
-      "loss": 0.0416,
+      "loss": 0.0553,
       "step": 568
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7509293359910136e-05,
-      "loss": 0.0449,
+      "loss": 0.0548,
       "step": 569
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7499777545992452e-05,
-      "loss": 0.0308,
+      "loss": 0.0475,
       "step": 570
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.749024618536255e-05,
-      "loss": 0.0344,
+      "loss": 0.0588,
       "step": 571
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.748069929777852e-05,
-      "loss": 0.0493,
+      "loss": 0.064,
       "step": 572
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.747113690303064e-05,
-      "loss": 0.0287,
+      "loss": 0.0515,
       "step": 573
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7461559020941345e-05,
-      "loss": 0.0457,
+      "loss": 0.0662,
       "step": 574
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.745196567136516e-05,
-      "loss": 0.0311,
+      "loss": 0.0498,
       "step": 575
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7442356874188692e-05,
-      "loss": 0.0435,
+      "loss": 0.0611,
       "step": 576
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.743273264933055e-05,
-      "loss": 0.0395,
+      "loss": 0.0592,
       "step": 577
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7423093016741332e-05,
-      "loss": 0.0386,
+      "loss": 0.0657,
       "step": 578
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.7413437996403582e-05,
-      "loss": 0.0446,
+      "loss": 0.0563,
       "step": 579
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7403767608331735e-05,
-      "loss": 0.0404,
+      "loss": 0.0576,
       "step": 580
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7394081872572083e-05,
-      "loss": 0.0355,
+      "loss": 0.0577,
       "step": 581
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7384380809202736e-05,
-      "loss": 0.0414,
+      "loss": 0.0652,
       "step": 582
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7374664438333577e-05,
-      "loss": 0.048,
+      "loss": 0.0554,
       "step": 583
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7364932780106212e-05,
-      "loss": 0.0284,
+      "loss": 0.0558,
       "step": 584
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7355185854693955e-05,
-      "loss": 0.0396,
+      "loss": 0.0605,
       "step": 585
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7345423682301754e-05,
-      "loss": 0.0396,
+      "loss": 0.055,
       "step": 586
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7335646283166167e-05,
-      "loss": 0.0398,
+      "loss": 0.0576,
       "step": 587
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7325853677555313e-05,
-      "loss": 0.048,
+      "loss": 0.0529,
       "step": 588
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.731604588576884e-05,
-      "loss": 0.0308,
+      "loss": 0.0557,
       "step": 589
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.7306222928137874e-05,
-      "loss": 0.0288,
+      "loss": 0.0581,
       "step": 590
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.729638482502498e-05,
-      "loss": 0.0413,
+      "loss": 0.0618,
       "step": 591
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7286531596824113e-05,
-      "loss": 0.0528,
+      "loss": 0.0518,
       "step": 592
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7276663263960588e-05,
-      "loss": 0.0553,
+      "loss": 0.0603,
       "step": 593
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.726677984689103e-05,
-      "loss": 0.0418,
+      "loss": 0.0528,
       "step": 594
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.725688136610333e-05,
-      "loss": 0.0357,
+      "loss": 0.0735,
       "step": 595
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7246967842116608e-05,
-      "loss": 0.0408,
+      "loss": 0.0637,
       "step": 596
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7237039295481167e-05,
-      "loss": 0.0358,
+      "loss": 0.0587,
       "step": 597
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7227095746778456e-05,
-      "loss": 0.0464,
+      "loss": 0.064,
       "step": 598
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7217137216621015e-05,
-      "loss": 0.0428,
+      "loss": 0.0625,
       "step": 599
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7207163725652444e-05,
-      "loss": 0.0443,
+      "loss": 0.0669,
       "step": 600
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.7197175294547356e-05,
-      "loss": 0.0412,
+      "loss": 0.0644,
       "step": 601
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.718717194401133e-05,
-      "loss": 0.0464,
+      "loss": 0.0672,
       "step": 602
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.717715369478088e-05,
-      "loss": 0.0302,
+      "loss": 0.0565,
       "step": 603
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.71671205676234e-05,
-      "loss": 0.0325,
+      "loss": 0.0671,
       "step": 604
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.7157072583337126e-05,
-      "loss": 0.0316,
+      "loss": 0.0508,
       "step": 605
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.7147009762751087e-05,
-      "loss": 0.0376,
+      "loss": 0.0683,
       "step": 606
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.713693212672508e-05,
-      "loss": 0.0475,
+      "loss": 0.0599,
       "step": 607
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.7126839696149596e-05,
-      "loss": 0.0324,
+      "loss": 0.0595,
       "step": 608
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.711673249194581e-05,
-      "loss": 0.0315,
+      "loss": 0.0565,
       "step": 609
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.710661053506552e-05,
-      "loss": 0.0555,
+      "loss": 0.0625,
       "step": 610
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.709647384649109e-05,
-      "loss": 0.0471,
+      "loss": 0.0632,
       "step": 611
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.708632244723545e-05,
-      "loss": 0.0348,
+      "loss": 0.0667,
       "step": 612
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.707615635834199e-05,
-      "loss": 0.0313,
+      "loss": 0.0598,
       "step": 613
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.7065975600884584e-05,
-      "loss": 0.046,
+      "loss": 0.0649,
       "step": 614
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.7055780195967492e-05,
-      "loss": 0.0288,
+      "loss": 0.0543,
       "step": 615
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.704557016472535e-05,
-      "loss": 0.0321,
+      "loss": 0.0605,
       "step": 616
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.70353455283231e-05,
-      "loss": 0.0378,
+      "loss": 0.0532,
       "step": 617
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.7025106307955976e-05,
-      "loss": 0.0424,
+      "loss": 0.0502,
       "step": 618
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.7014852524849425e-05,
-      "loss": 0.0391,
+      "loss": 0.0576,
       "step": 619
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.7004584200259107e-05,
-      "loss": 0.0249,
+      "loss": 0.049,
       "step": 620
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.69943013554708e-05,
-      "loss": 0.0331,
+      "loss": 0.0538,
       "step": 621
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6984004011800402e-05,
-      "loss": 0.0398,
+      "loss": 0.0597,
       "step": 622
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6973692190593853e-05,
-      "loss": 0.0312,
+      "loss": 0.0643,
       "step": 623
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.696336591322711e-05,
-      "loss": 0.0318,
+      "loss": 0.072,
       "step": 624
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6953025201106096e-05,
-      "loss": 0.0399,
+      "loss": 0.0777,
       "step": 625
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6942670075666657e-05,
-      "loss": 0.0379,
+      "loss": 0.0646,
       "step": 626
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6932300558374514e-05,
-      "loss": 0.036,
+      "loss": 0.0586,
       "step": 627
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6921916670725233e-05,
-      "loss": 0.0289,
+      "loss": 0.0483,
       "step": 628
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6911518434244158e-05,
-      "loss": 0.0394,
+      "loss": 0.0551,
       "step": 629
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.690110587048637e-05,
-      "loss": 0.0316,
+      "loss": 0.0581,
       "step": 630
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6890679001036677e-05,
-      "loss": 0.0399,
+      "loss": 0.0706,
       "step": 631
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6880237847509514e-05,
-      "loss": 0.0396,
+      "loss": 0.0605,
       "step": 632
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.686978243154894e-05,
-      "loss": 0.0458,
+      "loss": 0.0592,
       "step": 633
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6859312774828578e-05,
-      "loss": 0.052,
+      "loss": 0.0526,
       "step": 634
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.684882889905157e-05,
-      "loss": 0.0428,
+      "loss": 0.0639,
       "step": 635
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.683833082595053e-05,
-      "loss": 0.0459,
+      "loss": 0.0638,
       "step": 636
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.682781857728751e-05,
-      "loss": 0.0515,
+      "loss": 0.0479,
       "step": 637
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.681729217485395e-05,
-      "loss": 0.0224,
+      "loss": 0.0696,
       "step": 638
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.680675164047061e-05,
-      "loss": 0.0338,
+      "loss": 0.0534,
       "step": 639
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.679619699598757e-05,
-      "loss": 0.0456,
+      "loss": 0.0576,
       "step": 640
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6785628263284143e-05,
-      "loss": 0.0358,
+      "loss": 0.0564,
       "step": 641
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6775045464268856e-05,
-      "loss": 0.0341,
+      "loss": 0.0524,
       "step": 642
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6764448620879382e-05,
-      "loss": 0.0396,
+      "loss": 0.0678,
       "step": 643
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6753837755082527e-05,
-      "loss": 0.0438,
+      "loss": 0.0736,
       "step": 644
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6743212888874144e-05,
-      "loss": 0.0267,
+      "loss": 0.0652,
       "step": 645
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.673257404427912e-05,
-      "loss": 0.0375,
+      "loss": 0.0615,
       "step": 646
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6721921243351323e-05,
-      "loss": 0.0369,
+      "loss": 0.0583,
       "step": 647
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6711254508173542e-05,
-      "loss": 0.0504,
+      "loss": 0.0589,
       "step": 648
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6700573860857453e-05,
-      "loss": 0.0302,
+      "loss": 0.0567,
       "step": 649
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6689879323543567e-05,
-      "loss": 0.0235,
+      "loss": 0.0543,
       "step": 650
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6679170918401205e-05,
-      "loss": 0.0379,
+      "loss": 0.0591,
       "step": 651
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6668448667628417e-05,
-      "loss": 0.0394,
+      "loss": 0.0729,
       "step": 652
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.665771259345197e-05,
-      "loss": 0.0526,
+      "loss": 0.0559,
       "step": 653
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6646962718127263e-05,
-      "loss": 0.0383,
+      "loss": 0.065,
       "step": 654
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6636199063938335e-05,
-      "loss": 0.0303,
+      "loss": 0.0482,
       "step": 655
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6625421653197768e-05,
-      "loss": 0.0353,
+      "loss": 0.0588,
       "step": 656
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6614630508246657e-05,
-      "loss": 0.0337,
+      "loss": 0.0573,
       "step": 657
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.660382565145459e-05,
-      "loss": 0.0423,
+      "loss": 0.0596,
       "step": 658
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.659300710521955e-05,
-      "loss": 0.0365,
+      "loss": 0.0567,
       "step": 659
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.658217489196792e-05,
-      "loss": 0.0436,
+      "loss": 0.0526,
       "step": 660
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6571329034154406e-05,
-      "loss": 0.0411,
+      "loss": 0.064,
       "step": 661
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6560469554262e-05,
-      "loss": 0.0291,
+      "loss": 0.0593,
       "step": 662
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6549596474801927e-05,
-      "loss": 0.0309,
+      "loss": 0.0573,
       "step": 663
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6538709818313607e-05,
-      "loss": 0.0267,
+      "loss": 0.055,
       "step": 664
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6527809607364608e-05,
-      "loss": 0.0308,
+      "loss": 0.0598,
       "step": 665
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6516895864550592e-05,
-      "loss": 0.044,
+      "loss": 0.0691,
       "step": 666
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.650596861249527e-05,
-      "loss": 0.0294,
+      "loss": 0.0536,
       "step": 667
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6495027873850364e-05,
-      "loss": 0.0384,
+      "loss": 0.051,
       "step": 668
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6484073671295545e-05,
-      "loss": 0.0423,
+      "loss": 0.0641,
       "step": 669
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6473106027538396e-05,
-      "loss": 0.0266,
+      "loss": 0.0548,
       "step": 670
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.646212496531437e-05,
-      "loss": 0.0321,
+      "loss": 0.0609,
       "step": 671
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.645113050738673e-05,
-      "loss": 0.0301,
+      "loss": 0.0511,
       "step": 672
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6440122676546505e-05,
-      "loss": 0.047,
+      "loss": 0.0598,
       "step": 673
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6429101495612453e-05,
-      "loss": 0.0302,
+      "loss": 0.0603,
       "step": 674
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6418066987430998e-05,
-      "loss": 0.0296,
+      "loss": 0.0623,
       "step": 675
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.64070191748762e-05,
-      "loss": 0.0419,
+      "loss": 0.0619,
       "step": 676
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6395958080849686e-05,
-      "loss": 0.0265,
+      "loss": 0.0569,
       "step": 677
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.638488372828063e-05,
-      "loss": 0.0244,
+      "loss": 0.0545,
       "step": 678
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.637379614012567e-05,
-      "loss": 0.0288,
+      "loss": 0.0559,
       "step": 679
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6362695339368912e-05,
-      "loss": 0.0326,
+      "loss": 0.0664,
       "step": 680
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6351581349021822e-05,
-      "loss": 0.0173,
+      "loss": 0.0576,
       "step": 681
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6340454192123215e-05,
-      "loss": 0.0342,
+      "loss": 0.0612,
       "step": 682
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.632931389173921e-05,
-      "loss": 0.0342,
+      "loss": 0.0587,
       "step": 683
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.631816047096316e-05,
-      "loss": 0.0377,
+      "loss": 0.0672,
       "step": 684
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.630699395291562e-05,
-      "loss": 0.0281,
+      "loss": 0.0528,
       "step": 685
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6295814360744294e-05,
-      "loss": 0.038,
+      "loss": 0.0696,
       "step": 686
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6284621717624e-05,
-      "loss": 0.0227,
+      "loss": 0.0485,
       "step": 687
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6273416046756584e-05,
-      "loss": 0.0321,
+      "loss": 0.0537,
       "step": 688
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6262197371370927e-05,
-      "loss": 0.0247,
+      "loss": 0.0562,
       "step": 689
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6250965714722852e-05,
-      "loss": 0.0178,
+      "loss": 0.0639,
       "step": 690
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.6239721100095083e-05,
-      "loss": 0.0345,
+      "loss": 0.066,
       "step": 691
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6228463550797236e-05,
-      "loss": 0.0422,
+      "loss": 0.0479,
       "step": 692
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.62171930901657e-05,
-      "loss": 0.0366,
+      "loss": 0.0633,
       "step": 693
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6205909741563662e-05,
-      "loss": 0.0278,
+      "loss": 0.0514,
       "step": 694
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.619461352838101e-05,
-      "loss": 0.0364,
+      "loss": 0.0517,
       "step": 695
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6183304474034304e-05,
-      "loss": 0.0367,
+      "loss": 0.0479,
       "step": 696
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6171982601966713e-05,
-      "loss": 0.0289,
+      "loss": 0.0529,
       "step": 697
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6160647935647987e-05,
-      "loss": 0.0364,
+      "loss": 0.0519,
       "step": 698
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6149300498574402e-05,
-      "loss": 0.034,
+      "loss": 0.0496,
       "step": 699
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6137940314268698e-05,
-      "loss": 0.0352,
+      "loss": 0.0524,
       "step": 700
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.612656740628004e-05,
-      "loss": 0.0394,
+      "loss": 0.0641,
       "step": 701
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.611518179818397e-05,
-      "loss": 0.0373,
+      "loss": 0.045,
       "step": 702
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.6103783513582364e-05,
-      "loss": 0.0357,
+      "loss": 0.0509,
       "step": 703
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6092372576103362e-05,
-      "loss": 0.0386,
+      "loss": 0.0558,
       "step": 704
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6080949009401345e-05,
-      "loss": 0.0363,
+      "loss": 0.0523,
       "step": 705
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6069512837156867e-05,
-      "loss": 0.04,
+      "loss": 0.0575,
       "step": 706
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6058064083076618e-05,
-      "loss": 0.0276,
+      "loss": 0.0573,
       "step": 707
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6046602770893364e-05,
-      "loss": 0.0327,
+      "loss": 0.058,
       "step": 708
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.603512892436591e-05,
-      "loss": 0.0345,
+      "loss": 0.074,
       "step": 709
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.602364256727903e-05,
-      "loss": 0.0445,
+      "loss": 0.0589,
       "step": 710
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6012143723443454e-05,
-      "loss": 0.0338,
+      "loss": 0.0498,
       "step": 711
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.6000632416695783e-05,
-      "loss": 0.0316,
+      "loss": 0.0676,
       "step": 712
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5989108670898444e-05,
-      "loss": 0.0307,
+      "loss": 0.0574,
       "step": 713
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5977572509939673e-05,
-      "loss": 0.035,
+      "loss": 0.0584,
       "step": 714
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5966023957733417e-05,
-      "loss": 0.0329,
+      "loss": 0.0598,
       "step": 715
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.595446303821933e-05,
-      "loss": 0.0357,
+      "loss": 0.0451,
       "step": 716
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5942889775362697e-05,
-      "loss": 0.0213,
+      "loss": 0.0485,
       "step": 717
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5931304193154377e-05,
-      "loss": 0.0347,
+      "loss": 0.0581,
       "step": 718
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5919706315610787e-05,
-      "loss": 0.024,
+      "loss": 0.0643,
       "step": 719
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5908096166773818e-05,
-      "loss": 0.0306,
+      "loss": 0.0714,
       "step": 720
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.589647377071081e-05,
-      "loss": 0.0275,
+      "loss": 0.0458,
       "step": 721
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.588483915151447e-05,
-      "loss": 0.026,
+      "loss": 0.0451,
       "step": 722
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5873192333302868e-05,
-      "loss": 0.0371,
+      "loss": 0.0606,
       "step": 723
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.586153334021935e-05,
-      "loss": 0.0313,
+      "loss": 0.0454,
       "step": 724
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5849862196432492e-05,
-      "loss": 0.0379,
+      "loss": 0.0512,
       "step": 725
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5838178926136077e-05,
-      "loss": 0.0328,
+      "loss": 0.0581,
       "step": 726
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5826483553549016e-05,
-      "loss": 0.0306,
+      "loss": 0.0682,
       "step": 727
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5814776102915293e-05,
-      "loss": 0.0203,
+      "loss": 0.0659,
       "step": 728
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.580305659850396e-05,
-      "loss": 0.031,
+      "loss": 0.0603,
       "step": 729
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.579132506460903e-05,
-      "loss": 0.0374,
+      "loss": 0.0509,
       "step": 730
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5779581525549468e-05,
-      "loss": 0.0239,
+      "loss": 0.0551,
       "step": 731
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.576782600566911e-05,
-      "loss": 0.0237,
+      "loss": 0.0556,
       "step": 732
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5756058529336647e-05,
-      "loss": 0.0205,
+      "loss": 0.0674,
       "step": 733
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5744279120945534e-05,
-      "loss": 0.0294,
+      "loss": 0.0535,
       "step": 734
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.573248780491398e-05,
-      "loss": 0.0404,
+      "loss": 0.0603,
       "step": 735
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5720684605684868e-05,
-      "loss": 0.027,
+      "loss": 0.0539,
       "step": 736
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.570886954772571e-05,
-      "loss": 0.0236,
+      "loss": 0.0468,
       "step": 737
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5697042655528617e-05,
-      "loss": 0.0299,
+      "loss": 0.0562,
       "step": 738
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5685203953610216e-05,
-      "loss": 0.0329,
+      "loss": 0.0572,
       "step": 739
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5673353466511618e-05,
-      "loss": 0.0276,
+      "loss": 0.0518,
       "step": 740
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.566149121879837e-05,
-      "loss": 0.0257,
+      "loss": 0.0533,
       "step": 741
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5649617235060394e-05,
-      "loss": 0.0261,
+      "loss": 0.0548,
       "step": 742
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.563773153991194e-05,
-      "loss": 0.0353,
+      "loss": 0.0592,
       "step": 743
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5625834157991538e-05,
-      "loss": 0.0315,
+      "loss": 0.0551,
       "step": 744
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5613925113961942e-05,
-      "loss": 0.0236,
+      "loss": 0.0489,
       "step": 745
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5602004432510085e-05,
-      "loss": 0.0326,
+      "loss": 0.0678,
       "step": 746
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.559007213834701e-05,
-      "loss": 0.0278,
+      "loss": 0.0493,
       "step": 747
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.557812825620786e-05,
-      "loss": 0.023,
+      "loss": 0.058,
       "step": 748
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5566172810851766e-05,
-      "loss": 0.0398,
+      "loss": 0.0522,
       "step": 749
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5554205827061853e-05,
-      "loss": 0.0236,
+      "loss": 0.0637,
       "step": 750
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5542227329645157e-05,
-      "loss": 0.0306,
+      "loss": 0.0634,
       "step": 751
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.553023734343258e-05,
-      "loss": 0.0314,
+      "loss": 0.0571,
       "step": 752
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5518235893278845e-05,
-      "loss": 0.0424,
+      "loss": 0.0517,
       "step": 753
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5506223004062428e-05,
-      "loss": 0.0345,
+      "loss": 0.0653,
       "step": 754
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5494198700685535e-05,
-      "loss": 0.0353,
+      "loss": 0.05,
       "step": 755
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5482163008074017e-05,
-      "loss": 0.0332,
+      "loss": 0.0535,
       "step": 756
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5470115951177343e-05,
-      "loss": 0.0374,
+      "loss": 0.0484,
       "step": 757
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5458057554968532e-05,
-      "loss": 0.0349,
+      "loss": 0.0614,
       "step": 758
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5445987844444122e-05,
-      "loss": 0.0364,
+      "loss": 0.0511,
       "step": 759
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.543390684462409e-05,
-      "loss": 0.0419,
+      "loss": 0.0378,
       "step": 760
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5421814580551825e-05,
-      "loss": 0.0293,
+      "loss": 0.0686,
       "step": 761
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5409711077294066e-05,
-      "loss": 0.0369,
+      "loss": 0.0573,
       "step": 762
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5397596359940853e-05,
-      "loss": 0.0289,
+      "loss": 0.0511,
       "step": 763
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5385470453605457e-05,
-      "loss": 0.0275,
+      "loss": 0.0607,
       "step": 764
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5373333383424352e-05,
-      "loss": 0.0309,
+      "loss": 0.0478,
       "step": 765
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.536118517455717e-05,
-      "loss": 0.029,
+      "loss": 0.0617,
       "step": 766
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5349025852186604e-05,
-      "loss": 0.0375,
+      "loss": 0.0557,
       "step": 767
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5336855441518413e-05,
-      "loss": 0.0177,
+      "loss": 0.0554,
       "step": 768
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5324673967781318e-05,
-      "loss": 0.0209,
+      "loss": 0.049,
       "step": 769
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5312481456226986e-05,
-      "loss": 0.0253,
+      "loss": 0.0555,
       "step": 770
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5300277932129967e-05,
-      "loss": 0.0299,
+      "loss": 0.0583,
       "step": 771
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.528806342078763e-05,
-      "loss": 0.0421,
+      "loss": 0.0597,
       "step": 772
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.527583794752013e-05,
-      "loss": 0.0334,
+      "loss": 0.0481,
       "step": 773
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5263601537670335e-05,
-      "loss": 0.029,
+      "loss": 0.0699,
       "step": 774
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5251354216603792e-05,
-      "loss": 0.0244,
+      "loss": 0.05,
       "step": 775
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5239096009708665e-05,
-      "loss": 0.0261,
+      "loss": 0.0602,
       "step": 776
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.522682694239568e-05,
-      "loss": 0.0333,
+      "loss": 0.0651,
       "step": 777
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5214547040098082e-05,
-      "loss": 0.0298,
+      "loss": 0.0421,
       "step": 778
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5202256328271574e-05,
-      "loss": 0.0263,
+      "loss": 0.0565,
       "step": 779
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5189954832394267e-05,
-      "loss": 0.0254,
+      "loss": 0.0529,
       "step": 780
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5177642577966625e-05,
-      "loss": 0.0287,
+      "loss": 0.0484,
       "step": 781
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5165319590511414e-05,
-      "loss": 0.0241,
+      "loss": 0.0572,
       "step": 782
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5152985895573649e-05,
-      "loss": 0.0269,
+      "loss": 0.0484,
       "step": 783
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5140641518720547e-05,
-      "loss": 0.0392,
+      "loss": 0.0475,
       "step": 784
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.512828648554146e-05,
-      "loss": 0.0303,
+      "loss": 0.0519,
       "step": 785
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5115920821647834e-05,
-      "loss": 0.0191,
+      "loss": 0.0461,
       "step": 786
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5103544552673154e-05,
-      "loss": 0.0367,
+      "loss": 0.0519,
       "step": 787
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5091157704272886e-05,
-      "loss": 0.0246,
+      "loss": 0.0425,
       "step": 788
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5078760302124424e-05,
-      "loss": 0.0319,
+      "loss": 0.0498,
       "step": 789
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5066352371927047e-05,
-      "loss": 0.0259,
+      "loss": 0.0384,
       "step": 790
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5053933939401853e-05,
-      "loss": 0.0369,
+      "loss": 0.039,
       "step": 791
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.504150503029171e-05,
-      "loss": 0.0231,
+      "loss": 0.0497,
       "step": 792
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5029065670361205e-05,
-      "loss": 0.0261,
+      "loss": 0.0578,
       "step": 793
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5016615885396591e-05,
-      "loss": 0.0225,
+      "loss": 0.0546,
       "step": 794
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5004155701205732e-05,
-      "loss": 0.0262,
+      "loss": 0.0571,
       "step": 795
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4991685143618043e-05,
-      "loss": 0.0242,
+      "loss": 0.0661,
       "step": 796
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4979204238484445e-05,
-      "loss": 0.0284,
+      "loss": 0.0439,
       "step": 797
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4966713011677319e-05,
-      "loss": 0.0263,
+      "loss": 0.0621,
       "step": 798
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4954211489090427e-05,
-      "loss": 0.0246,
+      "loss": 0.0556,
       "step": 799
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4941699696638887e-05,
-      "loss": 0.0267,
+      "loss": 0.0488,
       "step": 800
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4929177660259094e-05,
-      "loss": 0.0269,
+      "loss": 0.0721,
       "step": 801
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4916645405908683e-05,
-      "loss": 0.0283,
+      "loss": 0.0587,
       "step": 802
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.4904102959566477e-05,
-      "loss": 0.0258,
+      "loss": 0.0577,
       "step": 803
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.489155034723242e-05,
-      "loss": 0.0246,
+      "loss": 0.0448,
       "step": 804
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4878987594927528e-05,
-      "loss": 0.0254,
+      "loss": 0.0504,
       "step": 805
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4866414728693839e-05,
-      "loss": 0.0202,
+      "loss": 0.054,
       "step": 806
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4853831774594362e-05,
-      "loss": 0.0226,
+      "loss": 0.0477,
       "step": 807
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4841238758713008e-05,
-      "loss": 0.0228,
+      "loss": 0.0408,
       "step": 808
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4828635707154555e-05,
-      "loss": 0.0309,
+      "loss": 0.0526,
       "step": 809
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4816022646044572e-05,
-      "loss": 0.0222,
+      "loss": 0.0577,
       "step": 810
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4803399601529396e-05,
-      "loss": 0.024,
+      "loss": 0.0453,
       "step": 811
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4790766599776038e-05,
-      "loss": 0.0245,
+      "loss": 0.0657,
       "step": 812
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4778123666972172e-05,
-      "loss": 0.0221,
+      "loss": 0.0593,
       "step": 813
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4765470829326034e-05,
-      "loss": 0.0243,
+      "loss": 0.061,
       "step": 814
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.4752808113066414e-05,
-      "loss": 0.0209,
+      "loss": 0.0482,
       "step": 815
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.474013554444257e-05,
-      "loss": 0.025,
+      "loss": 0.052,
       "step": 816
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4727453149724185e-05,
-      "loss": 0.0364,
+      "loss": 0.047,
       "step": 817
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4714760955201308e-05,
-      "loss": 0.0316,
+      "loss": 0.0574,
       "step": 818
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4702058987184309e-05,
-      "loss": 0.0301,
+      "loss": 0.0649,
       "step": 819
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4689347272003813e-05,
-      "loss": 0.0276,
+      "loss": 0.0489,
       "step": 820
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4676625836010651e-05,
-      "loss": 0.0238,
+      "loss": 0.0506,
       "step": 821
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.466389470557581e-05,
-      "loss": 0.0177,
+      "loss": 0.0532,
       "step": 822
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.465115390709036e-05,
-      "loss": 0.0215,
+      "loss": 0.0463,
       "step": 823
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.463840346696543e-05,
-      "loss": 0.0247,
+      "loss": 0.0477,
       "step": 824
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4625643411632127e-05,
-      "loss": 0.018,
+      "loss": 0.0477,
       "step": 825
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.4612873767541485e-05,
-      "loss": 0.0331,
+      "loss": 0.0546,
       "step": 826
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.460009456116442e-05,
-      "loss": 0.0319,
+      "loss": 0.0503,
       "step": 827
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4587305818991673e-05,
-      "loss": 0.0357,
+      "loss": 0.0597,
       "step": 828
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4574507567533752e-05,
-      "loss": 0.032,
+      "loss": 0.0557,
       "step": 829
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4561699833320868e-05,
-      "loss": 0.0327,
+      "loss": 0.0607,
       "step": 830
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4548882642902899e-05,
-      "loss": 0.0283,
+      "loss": 0.0476,
       "step": 831
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4536056022849328e-05,
-      "loss": 0.042,
+      "loss": 0.0587,
       "step": 832
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4523219999749175e-05,
-      "loss": 0.031,
+      "loss": 0.0454,
       "step": 833
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4510374600210955e-05,
-      "loss": 0.0378,
+      "loss": 0.0705,
       "step": 834
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4497519850862625e-05,
-      "loss": 0.0219,
+      "loss": 0.0551,
       "step": 835
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.448465577835152e-05,
-      "loss": 0.0234,
+      "loss": 0.0731,
       "step": 836
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4471782409344301e-05,
-      "loss": 0.0393,
+      "loss": 0.0594,
       "step": 837
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.4458899770526902e-05,
-      "loss": 0.0204,
+      "loss": 0.0561,
       "step": 838
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4446007888604472e-05,
-      "loss": 0.0225,
+      "loss": 0.0563,
       "step": 839
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.443310679030132e-05,
-      "loss": 0.021,
+      "loss": 0.046,
       "step": 840
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4420196502360864e-05,
-      "loss": 0.0313,
+      "loss": 0.0655,
       "step": 841
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4407277051545565e-05,
-      "loss": 0.0331,
+      "loss": 0.0462,
       "step": 842
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4394348464636882e-05,
-      "loss": 0.0289,
+      "loss": 0.05,
       "step": 843
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4381410768435215e-05,
-      "loss": 0.0248,
+      "loss": 0.059,
       "step": 844
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4368463989759847e-05,
-      "loss": 0.024,
+      "loss": 0.051,
       "step": 845
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.435550815544888e-05,
-      "loss": 0.0183,
+      "loss": 0.0532,
       "step": 846
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4342543292359196e-05,
-      "loss": 0.035,
+      "loss": 0.0536,
       "step": 847
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4329569427366394e-05,
-      "loss": 0.0204,
+      "loss": 0.0632,
       "step": 848
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4316586587364729e-05,
-      "loss": 0.0242,
+      "loss": 0.0589,
       "step": 849
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4303594799267067e-05,
-      "loss": 0.0327,
+      "loss": 0.0542,
       "step": 850
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4290594090004812e-05,
-      "loss": 0.0208,
+      "loss": 0.0472,
       "step": 851
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4277584486527873e-05,
-      "loss": 0.0195,
+      "loss": 0.0483,
       "step": 852
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4264566015804592e-05,
-      "loss": 0.0275,
+      "loss": 0.053,
       "step": 853
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.425153870482169e-05,
-      "loss": 0.0225,
+      "loss": 0.058,
       "step": 854
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4238502580584212e-05,
-      "loss": 0.0437,
+      "loss": 0.0601,
       "step": 855
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4225457670115483e-05,
-      "loss": 0.039,
+      "loss": 0.071,
       "step": 856
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4212404000457027e-05,
-      "loss": 0.0267,
+      "loss": 0.0452,
       "step": 857
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4199341598668539e-05,
-      "loss": 0.0345,
+      "loss": 0.0492,
       "step": 858
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4186270491827805e-05,
-      "loss": 0.0267,
+      "loss": 0.0426,
       "step": 859
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.417319070703066e-05,
-      "loss": 0.0259,
+      "loss": 0.0571,
       "step": 860
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4160102271390932e-05,
-      "loss": 0.0277,
+      "loss": 0.0602,
       "step": 861
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4147005212040374e-05,
-      "loss": 0.0304,
+      "loss": 0.0663,
       "step": 862
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4133899556128624e-05,
-      "loss": 0.0325,
+      "loss": 0.0518,
       "step": 863
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4120785330823128e-05,
-      "loss": 0.0219,
+      "loss": 0.0483,
       "step": 864
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4107662563309114e-05,
-      "loss": 0.025,
+      "loss": 0.0464,
       "step": 865
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4094531280789501e-05,
-      "loss": 0.0259,
+      "loss": 0.0483,
       "step": 866
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4081391510484866e-05,
-      "loss": 0.0216,
+      "loss": 0.06,
       "step": 867
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4068243279633382e-05,
-      "loss": 0.0397,
+      "loss": 0.0599,
       "step": 868
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.405508661549076e-05,
-      "loss": 0.0409,
+      "loss": 0.0587,
       "step": 869
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4041921545330192e-05,
-      "loss": 0.0259,
+      "loss": 0.0581,
       "step": 870
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4028748096442294e-05,
-      "loss": 0.0231,
+      "loss": 0.0604,
       "step": 871
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4015566296135048e-05,
-      "loss": 0.0248,
+      "loss": 0.0518,
       "step": 872
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4002376171733758e-05,
-      "loss": 0.0188,
+      "loss": 0.0402,
       "step": 873
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3989177750580975e-05,
-      "loss": 0.0368,
+      "loss": 0.0446,
       "step": 874
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.397597106003645e-05,
-      "loss": 0.0243,
+      "loss": 0.048,
       "step": 875
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3962756127477077e-05,
-      "loss": 0.0262,
+      "loss": 0.046,
       "step": 876
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3949532980296842e-05,
-      "loss": 0.025,
+      "loss": 0.0546,
       "step": 877
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3936301645906744e-05,
-      "loss": 0.0188,
+      "loss": 0.064,
       "step": 878
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.392306215173477e-05,
-      "loss": 0.0313,
+      "loss": 0.0461,
       "step": 879
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3909814525225812e-05,
-      "loss": 0.0139,
+      "loss": 0.0436,
       "step": 880
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3896558793841623e-05,
-      "loss": 0.0227,
+      "loss": 0.0398,
       "step": 881
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3883294985060756e-05,
-      "loss": 0.0332,
+      "loss": 0.0427,
       "step": 882
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.3870023126378509e-05,
-      "loss": 0.0193,
+      "loss": 0.0669,
       "step": 883
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3856743245306867e-05,
-      "loss": 0.0257,
+      "loss": 0.0555,
       "step": 884
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3843455369374449e-05,
-      "loss": 0.0149,
+      "loss": 0.061,
       "step": 885
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.383015952612644e-05,
-      "loss": 0.0249,
+      "loss": 0.0429,
       "step": 886
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3816855743124539e-05,
-      "loss": 0.0226,
+      "loss": 0.0425,
       "step": 887
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3803544047946918e-05,
-      "loss": 0.0237,
+      "loss": 0.0432,
       "step": 888
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3790224468188135e-05,
-      "loss": 0.0249,
+      "loss": 0.0522,
       "step": 889
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3776897031459104e-05,
-      "loss": 0.0255,
+      "loss": 0.0648,
       "step": 890
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3763561765387014e-05,
-      "loss": 0.0294,
+      "loss": 0.0417,
       "step": 891
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3750218697615297e-05,
-      "loss": 0.0316,
+      "loss": 0.0656,
       "step": 892
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.3736867855803549e-05,
-      "loss": 0.0257,
+      "loss": 0.0555,
       "step": 893
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.372350926762748e-05,
-      "loss": 0.0167,
+      "loss": 0.0514,
       "step": 894
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.371014296077887e-05,
-      "loss": 0.0201,
+      "loss": 0.0447,
       "step": 895
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3696768962965481e-05,
-      "loss": 0.016,
+      "loss": 0.0675,
       "step": 896
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3683387301911038e-05,
-      "loss": 0.023,
+      "loss": 0.0485,
       "step": 897
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3669998005355131e-05,
-      "loss": 0.0323,
+      "loss": 0.0577,
       "step": 898
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3656601101053199e-05,
-      "loss": 0.0248,
+      "loss": 0.0518,
       "step": 899
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3643196616776433e-05,
-      "loss": 0.0169,
+      "loss": 0.0478,
       "step": 900
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3629784580311751e-05,
-      "loss": 0.0167,
+      "loss": 0.0658,
       "step": 901
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3616365019461717e-05,
-      "loss": 0.0368,
+      "loss": 0.0536,
       "step": 902
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3602937962044496e-05,
-      "loss": 0.0267,
+      "loss": 0.0516,
       "step": 903
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3589503435893794e-05,
-      "loss": 0.0197,
+      "loss": 0.0478,
       "step": 904
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.3576061468858802e-05,
-      "loss": 0.0291,
+      "loss": 0.0437,
       "step": 905
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3562612088804128e-05,
-      "loss": 0.0242,
+      "loss": 0.0455,
       "step": 906
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.354915532360975e-05,
-      "loss": 0.0223,
+      "loss": 0.0557,
       "step": 907
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3535691201170959e-05,
-      "loss": 0.0235,
+      "loss": 0.0377,
       "step": 908
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3522219749398295e-05,
-      "loss": 0.0123,
+      "loss": 0.0541,
       "step": 909
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3508740996217493e-05,
-      "loss": 0.0236,
+      "loss": 0.0471,
       "step": 910
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3495254969569414e-05,
-      "loss": 0.0332,
+      "loss": 0.0663,
       "step": 911
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.348176169741001e-05,
-      "loss": 0.0247,
+      "loss": 0.055,
       "step": 912
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3468261207710242e-05,
-      "loss": 0.0385,
+      "loss": 0.0551,
       "step": 913
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.345475352845604e-05,
-      "loss": 0.018,
+      "loss": 0.0598,
       "step": 914
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3441238687648236e-05,
-      "loss": 0.0318,
+      "loss": 0.0416,
       "step": 915
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.3427716713302505e-05,
-      "loss": 0.0265,
+      "loss": 0.0543,
       "step": 916
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3414187633449311e-05,
-      "loss": 0.0233,
+      "loss": 0.0508,
       "step": 917
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3400651476133846e-05,
-      "loss": 0.0162,
+      "loss": 0.0486,
       "step": 918
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3387108269415977e-05,
-      "loss": 0.0296,
+      "loss": 0.0489,
       "step": 919
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3373558041370179e-05,
-      "loss": 0.017,
+      "loss": 0.0537,
       "step": 920
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3360000820085484e-05,
-      "loss": 0.0227,
+      "loss": 0.0433,
       "step": 921
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3346436633665429e-05,
-      "loss": 0.0239,
+      "loss": 0.045,
       "step": 922
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.333286551022797e-05,
-      "loss": 0.0263,
+      "loss": 0.0495,
       "step": 923
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3319287477905465e-05,
-      "loss": 0.0199,
+      "loss": 0.0571,
       "step": 924
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3305702564844577e-05,
-      "loss": 0.014,
+      "loss": 0.044,
       "step": 925
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3292110799206243e-05,
-      "loss": 0.0282,
+      "loss": 0.0573,
       "step": 926
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3278512209165602e-05,
-      "loss": 0.0285,
+      "loss": 0.0507,
       "step": 927
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.3264906822911931e-05,
-      "loss": 0.0182,
+      "loss": 0.0495,
       "step": 928
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3251294668648613e-05,
-      "loss": 0.0271,
+      "loss": 0.0522,
       "step": 929
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3237675774593047e-05,
-      "loss": 0.0331,
+      "loss": 0.0466,
       "step": 930
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3224050168976606e-05,
-      "loss": 0.0304,
+      "loss": 0.0436,
       "step": 931
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3210417880044579e-05,
-      "loss": 0.0376,
+      "loss": 0.0472,
       "step": 932
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3196778936056105e-05,
-      "loss": 0.029,
+      "loss": 0.0526,
       "step": 933
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3183133365284122e-05,
-      "loss": 0.02,
+      "loss": 0.0576,
       "step": 934
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3169481196015306e-05,
-      "loss": 0.0247,
+      "loss": 0.0527,
       "step": 935
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3155822456550007e-05,
-      "loss": 0.0191,
+      "loss": 0.0615,
       "step": 936
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3142157175202197e-05,
-      "loss": 0.0239,
+      "loss": 0.0474,
       "step": 937
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.312848538029941e-05,
-      "loss": 0.0258,
+      "loss": 0.0587,
       "step": 938
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.3114807100182679e-05,
-      "loss": 0.0145,
+      "loss": 0.0518,
       "step": 939
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3101122363206489e-05,
-      "loss": 0.0202,
+      "loss": 0.048,
       "step": 940
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3087431197738694e-05,
-      "loss": 0.023,
+      "loss": 0.0522,
       "step": 941
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3073733632160494e-05,
-      "loss": 0.031,
+      "loss": 0.0466,
       "step": 942
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3060029694866341e-05,
-      "loss": 0.0357,
+      "loss": 0.0508,
       "step": 943
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3046319414263902e-05,
-      "loss": 0.0207,
+      "loss": 0.0461,
       "step": 944
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3032602818773985e-05,
-      "loss": 0.0191,
+      "loss": 0.0608,
       "step": 945
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3018879936830504e-05,
-      "loss": 0.0196,
+      "loss": 0.0547,
       "step": 946
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.3005150796880391e-05,
-      "loss": 0.0223,
+      "loss": 0.0526,
       "step": 947
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2991415427383555e-05,
-      "loss": 0.0111,
+      "loss": 0.0537,
       "step": 948
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.297767385681282e-05,
-      "loss": 0.0169,
+      "loss": 0.0459,
       "step": 949
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.2963926113653864e-05,
-      "loss": 0.022,
+      "loss": 0.0437,
       "step": 950
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.295017222640516e-05,
-      "loss": 0.0183,
+      "loss": 0.0564,
       "step": 951
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2936412223577914e-05,
-      "loss": 0.0148,
+      "loss": 0.0406,
       "step": 952
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2922646133696017e-05,
-      "loss": 0.0176,
+      "loss": 0.0384,
       "step": 953
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2908873985295969e-05,
-      "loss": 0.0138,
+      "loss": 0.0486,
       "step": 954
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2895095806926836e-05,
-      "loss": 0.029,
+      "loss": 0.0465,
       "step": 955
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2881311627150185e-05,
-      "loss": 0.0094,
+      "loss": 0.0618,
       "step": 956
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2867521474540016e-05,
-      "loss": 0.0182,
+      "loss": 0.0476,
       "step": 957
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2853725377682718e-05,
-      "loss": 0.0233,
+      "loss": 0.0554,
       "step": 958
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2839923365176993e-05,
-      "loss": 0.0274,
+      "loss": 0.0575,
       "step": 959
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2826115465633822e-05,
-      "loss": 0.0155,
+      "loss": 0.0448,
       "step": 960
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.2812301707676366e-05,
-      "loss": 0.0386,
+      "loss": 0.0575,
       "step": 961
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2798482119939957e-05,
-      "loss": 0.023,
+      "loss": 0.0598,
       "step": 962
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.278465673107199e-05,
-      "loss": 0.0242,
+      "loss": 0.052,
       "step": 963
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2770825569731895e-05,
-      "loss": 0.0195,
+      "loss": 0.0558,
       "step": 964
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.275698866459107e-05,
-      "loss": 0.0098,
+      "loss": 0.0421,
       "step": 965
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2743146044332818e-05,
-      "loss": 0.0195,
+      "loss": 0.0549,
       "step": 966
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2729297737652281e-05,
-      "loss": 0.0241,
+      "loss": 0.0566,
       "step": 967
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2715443773256403e-05,
-      "loss": 0.016,
+      "loss": 0.0473,
       "step": 968
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2701584179863843e-05,
-      "loss": 0.0213,
+      "loss": 0.044,
       "step": 969
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.268771898620494e-05,
-      "loss": 0.0276,
+      "loss": 0.0527,
       "step": 970
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2673848221021633e-05,
-      "loss": 0.0234,
+      "loss": 0.0519,
       "step": 971
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2659971913067415e-05,
-      "loss": 0.0207,
+      "loss": 0.0504,
       "step": 972
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.2646090091107265e-05,
-      "loss": 0.0305,
+      "loss": 0.0531,
       "step": 973
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.26322027839176e-05,
-      "loss": 0.0229,
+      "loss": 0.0483,
       "step": 974
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2618310020286202e-05,
-      "loss": 0.0287,
+      "loss": 0.0436,
       "step": 975
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2604411829012166e-05,
-      "loss": 0.0208,
+      "loss": 0.0584,
       "step": 976
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2590508238905834e-05,
-      "loss": 0.0218,
+      "loss": 0.0549,
       "step": 977
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2576599278788741e-05,
-      "loss": 0.0173,
+      "loss": 0.0509,
       "step": 978
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2562684977493562e-05,
-      "loss": 0.0167,
+      "loss": 0.0487,
       "step": 979
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2548765363864038e-05,
-      "loss": 0.0261,
+      "loss": 0.0473,
       "step": 980
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2534840466754912e-05,
-      "loss": 0.0187,
+      "loss": 0.0489,
       "step": 981
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.2520910315031895e-05,
-      "loss": 0.0125,
+      "loss": 0.0542,
       "step": 982
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.250697493757159e-05,
-      "loss": 0.0128,
+      "loss": 0.0464,
       "step": 983
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.249303436326142e-05,
-      "loss": 0.0147,
+      "loss": 0.0538,
       "step": 984
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2479088620999592e-05,
-      "loss": 0.0237,
+      "loss": 0.0369,
       "step": 985
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2465137739695021e-05,
-      "loss": 0.0187,
+      "loss": 0.0573,
       "step": 986
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2451181748267281e-05,
-      "loss": 0.017,
+      "loss": 0.0538,
       "step": 987
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2437220675646525e-05,
-      "loss": 0.0211,
+      "loss": 0.0559,
       "step": 988
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2423254550773458e-05,
-      "loss": 0.0211,
+      "loss": 0.0583,
       "step": 989
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2409283402599239e-05,
-      "loss": 0.0197,
+      "loss": 0.043,
       "step": 990
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2395307260085458e-05,
-      "loss": 0.0293,
+      "loss": 0.0549,
       "step": 991
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2381326152204049e-05,
-      "loss": 0.0213,
+      "loss": 0.0448,
       "step": 992
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2367340107937236e-05,
-      "loss": 0.0254,
+      "loss": 0.0559,
       "step": 993
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.235334915627748e-05,
-      "loss": 0.0195,
+      "loss": 0.0594,
       "step": 994
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.2339353326227418e-05,
-      "loss": 0.0219,
+      "loss": 0.0373,
       "step": 995
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2325352646799796e-05,
-      "loss": 0.0207,
+      "loss": 0.0528,
       "step": 996
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2311347147017407e-05,
-      "loss": 0.013,
+      "loss": 0.0673,
       "step": 997
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2297336855913046e-05,
-      "loss": 0.0151,
+      "loss": 0.0582,
       "step": 998
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2283321802529437e-05,
-      "loss": 0.0216,
+      "loss": 0.0449,
       "step": 999
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2269302015919173e-05,
-      "loss": 0.0164,
+      "loss": 0.0406,
       "step": 1000
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2255277525144664e-05,
-      "loss": 0.0416,
+      "loss": 0.0509,
       "step": 1001
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2241248359278064e-05,
-      "loss": 0.0134,
+      "loss": 0.0498,
       "step": 1002
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.222721454740123e-05,
-      "loss": 0.0191,
+      "loss": 0.0507,
       "step": 1003
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2213176118605636e-05,
-      "loss": 0.0294,
+      "loss": 0.0491,
       "step": 1004
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.219913310199234e-05,
-      "loss": 0.0222,
+      "loss": 0.0557,
       "step": 1005
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.2185085526671893e-05,
-      "loss": 0.0143,
+      "loss": 0.0484,
       "step": 1006
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.217103342176432e-05,
-      "loss": 0.0126,
+      "loss": 0.044,
       "step": 1007
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2156976816399013e-05,
-      "loss": 0.0305,
+      "loss": 0.0544,
       "step": 1008
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2142915739714708e-05,
-      "loss": 0.0218,
+      "loss": 0.0483,
       "step": 1009
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2128850220859399e-05,
-      "loss": 0.0152,
+      "loss": 0.0449,
       "step": 1010
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2114780288990298e-05,
-      "loss": 0.015,
+      "loss": 0.0495,
       "step": 1011
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2100705973273765e-05,
-      "loss": 0.029,
+      "loss": 0.0475,
       "step": 1012
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2086627302885234e-05,
-      "loss": 0.0244,
+      "loss": 0.0519,
       "step": 1013
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2072544307009184e-05,
-      "loss": 0.0163,
+      "loss": 0.0527,
       "step": 1014
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2058457014839047e-05,
-      "loss": 0.0226,
+      "loss": 0.0426,
       "step": 1015
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2044365455577172e-05,
-      "loss": 0.0366,
+      "loss": 0.0426,
       "step": 1016
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2030269658434738e-05,
-      "loss": 0.0282,
+      "loss": 0.0474,
       "step": 1017
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.2016169652631725e-05,
-      "loss": 0.0196,
+      "loss": 0.0524,
       "step": 1018
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.2002065467396833e-05,
-      "loss": 0.0256,
+      "loss": 0.0397,
       "step": 1019
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1987957131967418e-05,
-      "loss": 0.0121,
+      "loss": 0.0515,
       "step": 1020
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.197384467558945e-05,
-      "loss": 0.0141,
+      "loss": 0.0582,
       "step": 1021
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1959728127517434e-05,
-      "loss": 0.0363,
+      "loss": 0.0522,
       "step": 1022
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1945607517014358e-05,
-      "loss": 0.0145,
+      "loss": 0.0457,
       "step": 1023
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1931482873351631e-05,
-      "loss": 0.0249,
+      "loss": 0.0516,
       "step": 1024
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1917354225809034e-05,
-      "loss": 0.0315,
+      "loss": 0.0635,
       "step": 1025
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1903221603674629e-05,
-      "loss": 0.0398,
+      "loss": 0.0541,
       "step": 1026
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1889085036244723e-05,
-      "loss": 0.0122,
+      "loss": 0.0453,
       "step": 1027
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1874944552823818e-05,
-      "loss": 0.0199,
+      "loss": 0.0465,
       "step": 1028
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.1860800182724507e-05,
-      "loss": 0.0251,
+      "loss": 0.0408,
       "step": 1029
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1846651955267463e-05,
-      "loss": 0.019,
+      "loss": 0.059,
       "step": 1030
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1832499899781334e-05,
-      "loss": 0.0199,
+      "loss": 0.0515,
       "step": 1031
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1818344045602728e-05,
-      "loss": 0.0306,
+      "loss": 0.0554,
       "step": 1032
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1804184422076105e-05,
-      "loss": 0.0104,
+      "loss": 0.047,
       "step": 1033
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1790021058553751e-05,
-      "loss": 0.018,
+      "loss": 0.048,
       "step": 1034
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.17758539843957e-05,
-      "loss": 0.0098,
+      "loss": 0.039,
       "step": 1035
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1761683228969682e-05,
-      "loss": 0.018,
+      "loss": 0.0415,
       "step": 1036
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1747508821651059e-05,
-      "loss": 0.0253,
+      "loss": 0.0522,
       "step": 1037
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.173333079182275e-05,
-      "loss": 0.0213,
+      "loss": 0.053,
       "step": 1038
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1719149168875203e-05,
-      "loss": 0.0192,
+      "loss": 0.048,
       "step": 1039
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.1704963982206298e-05,
-      "loss": 0.0193,
+      "loss": 0.0391,
       "step": 1040
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1690775261221317e-05,
-      "loss": 0.0133,
+      "loss": 0.0617,
       "step": 1041
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1676583035332854e-05,
-      "loss": 0.0275,
+      "loss": 0.038,
       "step": 1042
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1662387333960774e-05,
-      "loss": 0.0153,
+      "loss": 0.0563,
       "step": 1043
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1648188186532154e-05,
-      "loss": 0.0301,
+      "loss": 0.0433,
       "step": 1044
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1633985622481202e-05,
-      "loss": 0.022,
+      "loss": 0.0544,
       "step": 1045
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1619779671249222e-05,
-      "loss": 0.0149,
+      "loss": 0.0618,
       "step": 1046
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1605570362284524e-05,
-      "loss": 0.0188,
+      "loss": 0.049,
       "step": 1047
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1591357725042394e-05,
-      "loss": 0.0234,
+      "loss": 0.0516,
       "step": 1048
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1577141788985005e-05,
-      "loss": 0.0149,
+      "loss": 0.0418,
       "step": 1049
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1562922583581375e-05,
-      "loss": 0.0303,
+      "loss": 0.058,
       "step": 1050
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.1548700138307294e-05,
-      "loss": 0.0259,
+      "loss": 0.056,
       "step": 1051
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1534474482645282e-05,
-      "loss": 0.0183,
+      "loss": 0.0359,
       "step": 1052
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.152024564608449e-05,
-      "loss": 0.0163,
+      "loss": 0.059,
       "step": 1053
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1506013658120687e-05,
-      "loss": 0.0215,
+      "loss": 0.0587,
       "step": 1054
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1491778548256158e-05,
-      "loss": 0.0129,
+      "loss": 0.0488,
       "step": 1055
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.147754034599967e-05,
-      "loss": 0.0226,
+      "loss": 0.0421,
       "step": 1056
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1463299080866396e-05,
-      "loss": 0.0166,
+      "loss": 0.0528,
       "step": 1057
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1449054782377854e-05,
-      "loss": 0.0185,
+      "loss": 0.0409,
       "step": 1058
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1434807480061857e-05,
-      "loss": 0.0189,
+      "loss": 0.0445,
       "step": 1059
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1420557203452445e-05,
-      "loss": 0.0165,
+      "loss": 0.0634,
       "step": 1060
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1406303982089817e-05,
-      "loss": 0.0144,
+      "loss": 0.0548,
       "step": 1061
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1392047845520283e-05,
-      "loss": 0.0196,
+      "loss": 0.0542,
       "step": 1062
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.1377788823296187e-05,
-      "loss": 0.0238,
+      "loss": 0.0544,
       "step": 1063
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1363526944975867e-05,
-      "loss": 0.0172,
+      "loss": 0.0462,
       "step": 1064
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1349262240123575e-05,
-      "loss": 0.0262,
+      "loss": 0.05,
       "step": 1065
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1334994738309424e-05,
-      "loss": 0.0138,
+      "loss": 0.0435,
       "step": 1066
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.132072446910932e-05,
-      "loss": 0.0433,
+      "loss": 0.0536,
       "step": 1067
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1306451462104913e-05,
-      "loss": 0.0283,
+      "loss": 0.0619,
       "step": 1068
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.129217574688352e-05,
-      "loss": 0.0247,
+      "loss": 0.0482,
       "step": 1069
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1277897353038085e-05,
-      "loss": 0.0135,
+      "loss": 0.0487,
       "step": 1070
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.126361631016709e-05,
-      "loss": 0.0268,
+      "loss": 0.0498,
       "step": 1071
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1249332647874514e-05,
-      "loss": 0.0171,
+      "loss": 0.043,
       "step": 1072
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1235046395769772e-05,
-      "loss": 0.0138,
+      "loss": 0.0472,
       "step": 1073
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.1220757583467643e-05,
-      "loss": 0.0133,
+      "loss": 0.043,
       "step": 1074
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1206466240588208e-05,
-      "loss": 0.0202,
+      "loss": 0.0571,
       "step": 1075
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1192172396756798e-05,
-      "loss": 0.0179,
+      "loss": 0.0413,
       "step": 1076
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1177876081603933e-05,
-      "loss": 0.0156,
+      "loss": 0.047,
       "step": 1077
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1163577324765249e-05,
-      "loss": 0.014,
+      "loss": 0.0555,
       "step": 1078
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1149276155881446e-05,
-      "loss": 0.0194,
+      "loss": 0.0524,
       "step": 1079
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1134972604598224e-05,
-      "loss": 0.0134,
+      "loss": 0.0577,
       "step": 1080
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1120666700566225e-05,
-      "loss": 0.0185,
+      "loss": 0.0436,
       "step": 1081
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1106358473440964e-05,
-      "loss": 0.0256,
+      "loss": 0.0533,
       "step": 1082
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1092047952882767e-05,
-      "loss": 0.0237,
+      "loss": 0.0504,
       "step": 1083
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.1077735168556729e-05,
-      "loss": 0.0235,
+      "loss": 0.049,
       "step": 1084
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.106342015013262e-05,
-      "loss": 0.0275,
+      "loss": 0.0484,
       "step": 1085
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.1049102927284858e-05,
-      "loss": 0.0171,
+      "loss": 0.0475,
       "step": 1086
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.1034783529692416e-05,
-      "loss": 0.0186,
+      "loss": 0.05,
       "step": 1087
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.1020461987038781e-05,
-      "loss": 0.0126,
+      "loss": 0.0473,
       "step": 1088
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.1006138329011895e-05,
-      "loss": 0.0099,
+      "loss": 0.0374,
       "step": 1089
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.099181258530407e-05,
-      "loss": 0.0179,
+      "loss": 0.0527,
       "step": 1090
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.0977484785611951e-05,
-      "loss": 0.0145,
+      "loss": 0.044,
       "step": 1091
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.096315495963644e-05,
-      "loss": 0.011,
+      "loss": 0.0447,
       "step": 1092
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.0948823137082649e-05,
-      "loss": 0.0206,
+      "loss": 0.038,
       "step": 1093
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.0934489347659816e-05,
-      "loss": 0.0181,
+      "loss": 0.0392,
       "step": 1094
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.0920153621081265e-05,
-      "loss": 0.0218,
+      "loss": 0.0373,
       "step": 1095
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.0905815987064328e-05,
-      "loss": 0.0179,
+      "loss": 0.0485,
       "step": 1096
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0891476475330302e-05,
-      "loss": 0.0211,
+      "loss": 0.0313,
       "step": 1097
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0877135115604372e-05,
-      "loss": 0.0252,
+      "loss": 0.0477,
       "step": 1098
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0862791937615544e-05,
-      "loss": 0.0249,
+      "loss": 0.0529,
       "step": 1099
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0848446971096606e-05,
-      "loss": 0.0201,
+      "loss": 0.0451,
       "step": 1100
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.083410024578405e-05,
-      "loss": 0.0161,
+      "loss": 0.0522,
       "step": 1101
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0819751791418011e-05,
-      "loss": 0.0198,
+      "loss": 0.0377,
       "step": 1102
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0805401637742208e-05,
-      "loss": 0.026,
+      "loss": 0.0488,
       "step": 1103
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0791049814503888e-05,
-      "loss": 0.0196,
+      "loss": 0.0498,
       "step": 1104
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0776696351453755e-05,
-      "loss": 0.0197,
+      "loss": 0.0436,
       "step": 1105
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.076234127834591e-05,
-      "loss": 0.0238,
+      "loss": 0.0457,
       "step": 1106
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0747984624937799e-05,
-      "loss": 0.0197,
+      "loss": 0.0395,
       "step": 1107
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.0733626420990135e-05,
-      "loss": 0.0186,
+      "loss": 0.047,
       "step": 1108
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.071926669626685e-05,
-      "loss": 0.0203,
+      "loss": 0.0484,
       "step": 1109
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.070490548053503e-05,
-      "loss": 0.0129,
+      "loss": 0.0584,
       "step": 1110
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0690542803564851e-05,
-      "loss": 0.0198,
+      "loss": 0.0492,
       "step": 1111
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0676178695129514e-05,
-      "loss": 0.0291,
+      "loss": 0.0472,
       "step": 1112
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0661813185005188e-05,
-      "loss": 0.0285,
+      "loss": 0.0485,
       "step": 1113
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0647446302970954e-05,
-      "loss": 0.0148,
+      "loss": 0.0541,
       "step": 1114
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0633078078808731e-05,
-      "loss": 0.0182,
+      "loss": 0.0468,
       "step": 1115
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0618708542303226e-05,
-      "loss": 0.016,
+      "loss": 0.0466,
       "step": 1116
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0604337723241852e-05,
-      "loss": 0.0147,
+      "loss": 0.0406,
       "step": 1117
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0589965651414704e-05,
-      "loss": 0.0212,
+      "loss": 0.0509,
       "step": 1118
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.0575592356614453e-05,
-      "loss": 0.0127,
+      "loss": 0.0368,
       "step": 1119
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0561217868636314e-05,
-      "loss": 0.0145,
+      "loss": 0.0512,
       "step": 1120
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0546842217277975e-05,
-      "loss": 0.0209,
+      "loss": 0.0378,
       "step": 1121
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0532465432339539e-05,
-      "loss": 0.0109,
+      "loss": 0.047,
       "step": 1122
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0518087543623448e-05,
-      "loss": 0.0261,
+      "loss": 0.0455,
       "step": 1123
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0503708580934443e-05,
-      "loss": 0.0106,
+      "loss": 0.0423,
       "step": 1124
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0489328574079487e-05,
-      "loss": 0.0133,
+      "loss": 0.0495,
       "step": 1125
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0474947552867707e-05,
-      "loss": 0.0075,
+      "loss": 0.0388,
       "step": 1126
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.046056554711034e-05,
-      "loss": 0.0128,
+      "loss": 0.0323,
       "step": 1127
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0446182586620647e-05,
-      "loss": 0.0105,
+      "loss": 0.0379,
       "step": 1128
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0431798701213886e-05,
-      "loss": 0.005,
+      "loss": 0.0345,
       "step": 1129
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0417413920707223e-05,
-      "loss": 0.0113,
+      "loss": 0.0384,
       "step": 1130
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0403028274919681e-05,
-      "loss": 0.0205,
+      "loss": 0.0463,
       "step": 1131
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0388641793672079e-05,
-      "loss": 0.015,
+      "loss": 0.0535,
       "step": 1132
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0374254506786962e-05,
-      "loss": 0.0077,
+      "loss": 0.0367,
       "step": 1133
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0359866444088555e-05,
-      "loss": 0.0199,
+      "loss": 0.0349,
       "step": 1134
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.034547763540268e-05,
-      "loss": 0.0105,
+      "loss": 0.0369,
       "step": 1135
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0331088110556718e-05,
-      "loss": 0.0094,
+      "loss": 0.0312,
       "step": 1136
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0316697899379515e-05,
-      "loss": 0.0107,
+      "loss": 0.0476,
       "step": 1137
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0302307031701364e-05,
-      "loss": 0.0184,
+      "loss": 0.038,
       "step": 1138
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0287915537353902e-05,
-      "loss": 0.0195,
+      "loss": 0.0431,
       "step": 1139
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.027352344617007e-05,
-      "loss": 0.0122,
+      "loss": 0.0532,
       "step": 1140
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.0259130787984044e-05,
-      "loss": 0.0066,
+      "loss": 0.0335,
       "step": 1141
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0244737592631182e-05,
-      "loss": 0.0064,
+      "loss": 0.0404,
       "step": 1142
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0230343889947951e-05,
-      "loss": 0.0086,
+      "loss": 0.0496,
       "step": 1143
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0215949709771867e-05,
-      "loss": 0.0059,
+      "loss": 0.0428,
       "step": 1144
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0201555081941438e-05,
-      "loss": 0.0098,
+      "loss": 0.043,
       "step": 1145
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0187160036296104e-05,
-      "loss": 0.0083,
+      "loss": 0.037,
       "step": 1146
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0172764602676166e-05,
-      "loss": 0.0201,
+      "loss": 0.0571,
       "step": 1147
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.015836881092273e-05,
-      "loss": 0.0159,
+      "loss": 0.0353,
       "step": 1148
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0143972690877646e-05,
-      "loss": 0.0068,
+      "loss": 0.0407,
       "step": 1149
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0129576272383446e-05,
-      "loss": 0.0096,
+      "loss": 0.0336,
       "step": 1150
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.0115179585283272e-05,
-      "loss": 0.013,
+      "loss": 0.0412,
       "step": 1151
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.010078265942084e-05,
-      "loss": 0.0136,
+      "loss": 0.0387,
       "step": 1152
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.008638552464034e-05,
-      "loss": 0.009,
+      "loss": 0.0513,
       "step": 1153
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.0071988210786412e-05,
-      "loss": 0.0093,
+      "loss": 0.0408,
       "step": 1154
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.0057590747704058e-05,
-      "loss": 0.0117,
+      "loss": 0.0333,
       "step": 1155
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.0043193165238591e-05,
-      "loss": 0.0087,
+      "loss": 0.0463,
       "step": 1156
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.0028795493235575e-05,
-      "loss": 0.0164,
+      "loss": 0.0328,
       "step": 1157
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.0014397761540755e-05,
-      "loss": 0.0133,
+      "loss": 0.0335,
       "step": 1158
     },
     {
       "epoch": 1.03,
       "learning_rate": 1e-05,
-      "loss": 0.0133,
+      "loss": 0.0321,
       "step": 1159
     },
     {
       "epoch": 1.03,
       "learning_rate": 9.985602238459248e-06,
-      "loss": 0.0138,
+      "loss": 0.0494,
       "step": 1160
     },
     {
       "epoch": 1.03,
       "learning_rate": 9.971204506764428e-06,
-      "loss": 0.0224,
+      "loss": 0.0404,
       "step": 1161
     },
     {
       "epoch": 1.03,
       "learning_rate": 9.956806834761412e-06,
-      "loss": 0.0179,
+      "loss": 0.0447,
       "step": 1162
     },
     {
       "epoch": 1.03,
       "learning_rate": 9.942409252295946e-06,
-      "loss": 0.0145,
+      "loss": 0.0374,
       "step": 1163
     },
     {
       "epoch": 1.03,
       "learning_rate": 9.928011789213591e-06,
-      "loss": 0.0099,
+      "loss": 0.0463,
       "step": 1164
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.913614475359662e-06,
-      "loss": 0.0064,
+      "loss": 0.0432,
       "step": 1165
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.899217340579165e-06,
-      "loss": 0.0179,
+      "loss": 0.0338,
       "step": 1166
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.88482041471673e-06,
-      "loss": 0.019,
+      "loss": 0.0435,
       "step": 1167
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.870423727616558e-06,
-      "loss": 0.0111,
+      "loss": 0.0295,
       "step": 1168
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.856027309122356e-06,
-      "loss": 0.0149,
+      "loss": 0.0385,
       "step": 1169
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.84163118907727e-06,
-      "loss": 0.0085,
+      "loss": 0.042,
       "step": 1170
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.827235397323839e-06,
-      "loss": 0.0132,
+      "loss": 0.0341,
       "step": 1171
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.8128399637039e-06,
-      "loss": 0.0144,
+      "loss": 0.0419,
       "step": 1172
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.798444918058564e-06,
-      "loss": 0.0048,
+      "loss": 0.0387,
       "step": 1173
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.784050290228135e-06,
-      "loss": 0.014,
+      "loss": 0.0407,
       "step": 1174
     },
     {
       "epoch": 1.04,
       "learning_rate": 9.769656110052054e-06,
-      "loss": 0.0058,
+      "loss": 0.0472,
       "step": 1175
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.755262407368821e-06,
-      "loss": 0.0164,
+      "loss": 0.0341,
       "step": 1176
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.740869212015959e-06,
-      "loss": 0.026,
+      "loss": 0.0392,
       "step": 1177
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.726476553829933e-06,
-      "loss": 0.01,
+      "loss": 0.0384,
       "step": 1178
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.712084462646098e-06,
-      "loss": 0.017,
+      "loss": 0.0413,
       "step": 1179
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.69769296829864e-06,
-      "loss": 0.018,
+      "loss": 0.0552,
       "step": 1180
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.683302100620487e-06,
-      "loss": 0.0117,
+      "loss": 0.0434,
       "step": 1181
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.668911889443285e-06,
-      "loss": 0.0157,
+      "loss": 0.0343,
       "step": 1182
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.65452236459732e-06,
-      "loss": 0.014,
+      "loss": 0.0509,
       "step": 1183
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.64013355591145e-06,
-      "loss": 0.0175,
+      "loss": 0.0475,
       "step": 1184
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.625745493213041e-06,
-      "loss": 0.0168,
+      "loss": 0.0349,
       "step": 1185
     },
     {
       "epoch": 1.05,
       "learning_rate": 9.611358206327923e-06,
-      "loss": 0.0137,
+      "loss": 0.0354,
       "step": 1186
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.59697172508032e-06,
-      "loss": 0.0106,
+      "loss": 0.0396,
       "step": 1187
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.582586079292779e-06,
-      "loss": 0.0123,
+      "loss": 0.0418,
       "step": 1188
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.568201298786119e-06,
-      "loss": 0.0137,
+      "loss": 0.0355,
       "step": 1189
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.553817413379356e-06,
-      "loss": 0.0094,
+      "loss": 0.0334,
       "step": 1190
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.539434452889664e-06,
-      "loss": 0.0061,
+      "loss": 0.0482,
       "step": 1191
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.525052447132294e-06,
-      "loss": 0.0132,
+      "loss": 0.0365,
       "step": 1192
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.510671425920517e-06,
-      "loss": 0.0099,
+      "loss": 0.0471,
       "step": 1193
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.49629141906556e-06,
-      "loss": 0.0086,
+      "loss": 0.0412,
       "step": 1194
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.481912456376557e-06,
-      "loss": 0.0064,
+      "loss": 0.0482,
       "step": 1195
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.467534567660466e-06,
-      "loss": 0.0205,
+      "loss": 0.0345,
       "step": 1196
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.453157782722027e-06,
-      "loss": 0.0082,
+      "loss": 0.0384,
       "step": 1197
     },
     {
       "epoch": 1.06,
       "learning_rate": 9.43878213136369e-06,
-      "loss": 0.0145,
+      "loss": 0.0447,
       "step": 1198
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.42440764338555e-06,
-      "loss": 0.0149,
+      "loss": 0.0352,
       "step": 1199
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.410034348585299e-06,
-      "loss": 0.0076,
+      "loss": 0.0453,
       "step": 1200
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.395662276758147e-06,
-      "loss": 0.0046,
+      "loss": 0.0464,
       "step": 1201
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.381291457696779e-06,
-      "loss": 0.0132,
+      "loss": 0.0418,
       "step": 1202
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.36692192119127e-06,
-      "loss": 0.0038,
+      "loss": 0.0411,
       "step": 1203
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.352553697029049e-06,
-      "loss": 0.0117,
+      "loss": 0.0443,
       "step": 1204
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.338186814994814e-06,
-      "loss": 0.0083,
+      "loss": 0.0354,
       "step": 1205
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.32382130487049e-06,
-      "loss": 0.0076,
+      "loss": 0.0476,
       "step": 1206
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.309457196435154e-06,
-      "loss": 0.0134,
+      "loss": 0.0404,
       "step": 1207
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.295094519464971e-06,
-      "loss": 0.0086,
+      "loss": 0.0383,
       "step": 1208
     },
     {
       "epoch": 1.07,
       "learning_rate": 9.280733303733151e-06,
-      "loss": 0.0141,
+      "loss": 0.0348,
       "step": 1209
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.266373579009867e-06,
-      "loss": 0.0045,
+      "loss": 0.0425,
       "step": 1210
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.252015375062206e-06,
-      "loss": 0.0071,
+      "loss": 0.0391,
       "step": 1211
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.237658721654092e-06,
-      "loss": 0.0181,
+      "loss": 0.033,
       "step": 1212
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.22330364854625e-06,
-      "loss": 0.0193,
+      "loss": 0.0411,
       "step": 1213
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.208950185496115e-06,
-      "loss": 0.0116,
+      "loss": 0.0349,
       "step": 1214
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.194598362257792e-06,
-      "loss": 0.0096,
+      "loss": 0.0435,
       "step": 1215
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.180248208581994e-06,
-      "loss": 0.0041,
+      "loss": 0.0383,
       "step": 1216
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.165899754215955e-06,
-      "loss": 0.0105,
+      "loss": 0.0561,
       "step": 1217
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.151553028903397e-06,
-      "loss": 0.0139,
+      "loss": 0.0387,
       "step": 1218
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.137208062384456e-06,
-      "loss": 0.0243,
+      "loss": 0.0294,
       "step": 1219
     },
     {
       "epoch": 1.08,
       "learning_rate": 9.122864884395633e-06,
-      "loss": 0.0199,
+      "loss": 0.0377,
       "step": 1220
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.1085235246697e-06,
-      "loss": 0.0124,
+      "loss": 0.037,
       "step": 1221
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.094184012935675e-06,
-      "loss": 0.0099,
+      "loss": 0.0377,
       "step": 1222
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.079846378918737e-06,
-      "loss": 0.0071,
+      "loss": 0.0402,
       "step": 1223
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.065510652340185e-06,
-      "loss": 0.0132,
+      "loss": 0.0299,
       "step": 1224
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.051176862917356e-06,
-      "loss": 0.0121,
+      "loss": 0.0424,
       "step": 1225
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.036845040363564e-06,
-      "loss": 0.0137,
+      "loss": 0.0341,
       "step": 1226
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.022515214388052e-06,
-      "loss": 0.0202,
+      "loss": 0.031,
       "step": 1227
     },
     {
       "epoch": 1.09,
       "learning_rate": 9.008187414695932e-06,
-      "loss": 0.0171,
+      "loss": 0.0398,
       "step": 1228
     },
     {
       "epoch": 1.09,
       "learning_rate": 8.993861670988112e-06,
-      "loss": 0.0117,
+      "loss": 0.0322,
       "step": 1229
     },
     {
       "epoch": 1.09,
       "learning_rate": 8.979538012961222e-06,
-      "loss": 0.0081,
+      "loss": 0.0225,
       "step": 1230
     },
     {
       "epoch": 1.09,
       "learning_rate": 8.965216470307587e-06,
-      "loss": 0.0189,
+      "loss": 0.0446,
       "step": 1231
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.950897072715145e-06,
-      "loss": 0.0142,
+      "loss": 0.0321,
       "step": 1232
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.93657984986738e-06,
-      "loss": 0.0084,
+      "loss": 0.0315,
       "step": 1233
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.922264831443274e-06,
-      "loss": 0.0127,
+      "loss": 0.0486,
       "step": 1234
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.907952047117235e-06,
-      "loss": 0.0048,
+      "loss": 0.0303,
       "step": 1235
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.89364152655904e-06,
-      "loss": 0.0121,
+      "loss": 0.0496,
       "step": 1236
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.879333299433777e-06,
-      "loss": 0.0087,
+      "loss": 0.0477,
       "step": 1237
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.86502739540178e-06,
-      "loss": 0.0035,
+      "loss": 0.0292,
       "step": 1238
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.850723844118557e-06,
-      "loss": 0.0116,
+      "loss": 0.0287,
       "step": 1239
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.836422675234754e-06,
-      "loss": 0.0198,
+      "loss": 0.0359,
       "step": 1240
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.82212391839607e-06,
-      "loss": 0.0085,
+      "loss": 0.0484,
       "step": 1241
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.807827603243204e-06,
-      "loss": 0.0101,
+      "loss": 0.0331,
       "step": 1242
     },
     {
       "epoch": 1.1,
       "learning_rate": 8.793533759411797e-06,
-      "loss": 0.005,
+      "loss": 0.0495,
       "step": 1243
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.77924241653236e-06,
-      "loss": 0.005,
+      "loss": 0.0368,
       "step": 1244
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.76495360423023e-06,
-      "loss": 0.0087,
+      "loss": 0.0263,
       "step": 1245
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.750667352125486e-06,
-      "loss": 0.0122,
+      "loss": 0.0432,
       "step": 1246
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.736383689832914e-06,
-      "loss": 0.0127,
+      "loss": 0.0455,
       "step": 1247
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.72210264696192e-06,
-      "loss": 0.0102,
+      "loss": 0.0572,
       "step": 1248
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.707824253116481e-06,
-      "loss": 0.0106,
+      "loss": 0.0424,
       "step": 1249
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.693548537895092e-06,
-      "loss": 0.0251,
+      "loss": 0.048,
       "step": 1250
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.679275530890681e-06,
-      "loss": 0.0117,
+      "loss": 0.0395,
       "step": 1251
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.66500526169058e-06,
-      "loss": 0.0211,
+      "loss": 0.0382,
       "step": 1252
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.650737759876428e-06,
-      "loss": 0.0185,
+      "loss": 0.0435,
       "step": 1253
     },
     {
       "epoch": 1.11,
       "learning_rate": 8.636473055024135e-06,
-      "loss": 0.0148,
+      "loss": 0.0385,
       "step": 1254
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.622211176703816e-06,
-      "loss": 0.0113,
+      "loss": 0.0348,
       "step": 1255
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.607952154479724e-06,
-      "loss": 0.0113,
+      "loss": 0.0344,
       "step": 1256
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.593696017910188e-06,
-      "loss": 0.0126,
+      "loss": 0.0389,
       "step": 1257
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.579442796547559e-06,
-      "loss": 0.0151,
+      "loss": 0.0405,
       "step": 1258
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.565192519938144e-06,
-      "loss": 0.0044,
+      "loss": 0.0386,
       "step": 1259
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.550945217622147e-06,
-      "loss": 0.0157,
+      "loss": 0.0502,
       "step": 1260
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.536700919133609e-06,
-      "loss": 0.0084,
+      "loss": 0.0321,
       "step": 1261
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.522459654000332e-06,
-      "loss": 0.0098,
+      "loss": 0.0379,
       "step": 1262
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.508221451743845e-06,
-      "loss": 0.0159,
+      "loss": 0.0507,
       "step": 1263
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.493986341879315e-06,
-      "loss": 0.006,
+      "loss": 0.0343,
       "step": 1264
     },
     {
       "epoch": 1.12,
       "learning_rate": 8.479754353915514e-06,
-      "loss": 0.009,
+      "loss": 0.0296,
       "step": 1265
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.465525517354724e-06,
-      "loss": 0.0184,
+      "loss": 0.0275,
       "step": 1266
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.451299861692708e-06,
-      "loss": 0.0049,
+      "loss": 0.0387,
       "step": 1267
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.437077416418627e-06,
-      "loss": 0.0049,
+      "loss": 0.0449,
       "step": 1268
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.422858211014996e-06,
-      "loss": 0.0061,
+      "loss": 0.0331,
       "step": 1269
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.408642274957612e-06,
-      "loss": 0.0147,
+      "loss": 0.0361,
       "step": 1270
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.39442963771548e-06,
-      "loss": 0.0098,
+      "loss": 0.0497,
       "step": 1271
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.380220328750781e-06,
-      "loss": 0.0042,
+      "loss": 0.0364,
       "step": 1272
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.3660143775188e-06,
-      "loss": 0.0074,
+      "loss": 0.0449,
       "step": 1273
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.35181181346785e-06,
-      "loss": 0.0083,
+      "loss": 0.0427,
       "step": 1274
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.337612666039227e-06,
-      "loss": 0.0156,
+      "loss": 0.0339,
       "step": 1275
     },
     {
       "epoch": 1.13,
       "learning_rate": 8.323416964667151e-06,
-      "loss": 0.0069,
+      "loss": 0.0289,
       "step": 1276
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.309224738778685e-06,
-      "loss": 0.0099,
+      "loss": 0.0378,
       "step": 1277
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.295036017793702e-06,
-      "loss": 0.0091,
+      "loss": 0.0358,
       "step": 1278
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.2808508311248e-06,
-      "loss": 0.0115,
+      "loss": 0.031,
       "step": 1279
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.266669208177251e-06,
-      "loss": 0.012,
+      "loss": 0.0294,
       "step": 1280
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.252491178348945e-06,
-      "loss": 0.0073,
+      "loss": 0.0356,
       "step": 1281
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.238316771030318e-06,
-      "loss": 0.0118,
+      "loss": 0.0297,
       "step": 1282
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.224146015604302e-06,
-      "loss": 0.0072,
+      "loss": 0.0387,
       "step": 1283
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.209978941446252e-06,
-      "loss": 0.0143,
+      "loss": 0.058,
       "step": 1284
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.195815577923898e-06,
-      "loss": 0.012,
+      "loss": 0.0416,
       "step": 1285
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.181655954397275e-06,
-      "loss": 0.0104,
+      "loss": 0.0339,
       "step": 1286
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.167500100218666e-06,
-      "loss": 0.014,
+      "loss": 0.0367,
       "step": 1287
     },
     {
       "epoch": 1.14,
       "learning_rate": 8.153348044732543e-06,
-      "loss": 0.0026,
+      "loss": 0.0386,
       "step": 1288
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.139199817275496e-06,
-      "loss": 0.0092,
+      "loss": 0.0534,
       "step": 1289
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.125055447176187e-06,
-      "loss": 0.0061,
+      "loss": 0.0411,
       "step": 1290
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.110914963755277e-06,
-      "loss": 0.0149,
+      "loss": 0.0382,
       "step": 1291
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.096778396325378e-06,
-      "loss": 0.0132,
+      "loss": 0.0402,
       "step": 1292
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.082645774190971e-06,
-      "loss": 0.0094,
+      "loss": 0.0404,
       "step": 1293
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.06851712664837e-06,
-      "loss": 0.0132,
+      "loss": 0.0378,
       "step": 1294
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.054392482985647e-06,
-      "loss": 0.0116,
+      "loss": 0.0321,
       "step": 1295
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.04027187248257e-06,
-      "loss": 0.0095,
+      "loss": 0.0423,
       "step": 1296
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.026155324410556e-06,
-      "loss": 0.0051,
+      "loss": 0.0269,
       "step": 1297
     },
     {
       "epoch": 1.15,
       "learning_rate": 8.012042868032585e-06,
-      "loss": 0.0227,
+      "loss": 0.0489,
       "step": 1298
     },
     {
       "epoch": 1.15,
       "learning_rate": 7.997934532603172e-06,
-      "loss": 0.0119,
+      "loss": 0.0401,
       "step": 1299
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.983830347368276e-06,
-      "loss": 0.0125,
+      "loss": 0.0428,
       "step": 1300
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.969730341565263e-06,
-      "loss": 0.005,
+      "loss": 0.0401,
       "step": 1301
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.955634544422834e-06,
-      "loss": 0.0067,
+      "loss": 0.0456,
       "step": 1302
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.941542985160955e-06,
-      "loss": 0.005,
+      "loss": 0.0349,
       "step": 1303
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.927455692990819e-06,
-      "loss": 0.0179,
+      "loss": 0.0525,
       "step": 1304
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.913372697114764e-06,
-      "loss": 0.0097,
+      "loss": 0.0346,
       "step": 1305
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.89929402672624e-06,
-      "loss": 0.0083,
+      "loss": 0.0368,
       "step": 1306
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.885219711009704e-06,
-      "loss": 0.0067,
+      "loss": 0.0374,
       "step": 1307
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.871149779140604e-06,
-      "loss": 0.0118,
+      "loss": 0.029,
       "step": 1308
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.857084260285294e-06,
-      "loss": 0.013,
+      "loss": 0.0346,
       "step": 1309
     },
     {
       "epoch": 1.16,
       "learning_rate": 7.843023183600989e-06,
-      "loss": 0.006,
+      "loss": 0.0348,
       "step": 1310
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.828966578235686e-06,
-      "loss": 0.005,
+      "loss": 0.0362,
       "step": 1311
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.81491447332811e-06,
-      "loss": 0.0161,
+      "loss": 0.0328,
       "step": 1312
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.800866898007663e-06,
-      "loss": 0.0099,
+      "loss": 0.0374,
       "step": 1313
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.786823881394364e-06,
-      "loss": 0.0116,
+      "loss": 0.0391,
       "step": 1314
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.772785452598774e-06,
-      "loss": 0.0141,
+      "loss": 0.036,
       "step": 1315
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.758751640721937e-06,
-      "loss": 0.0179,
+      "loss": 0.0462,
       "step": 1316
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.744722474855338e-06,
-      "loss": 0.0112,
+      "loss": 0.0269,
       "step": 1317
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.730697984080827e-06,
-      "loss": 0.0147,
+      "loss": 0.0404,
       "step": 1318
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.716678197470565e-06,
-      "loss": 0.0041,
+      "loss": 0.0368,
       "step": 1319
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.702663144086957e-06,
-      "loss": 0.0085,
+      "loss": 0.0351,
       "step": 1320
     },
     {
       "epoch": 1.17,
       "learning_rate": 7.688652852982595e-06,
-      "loss": 0.0095,
+      "loss": 0.0394,
       "step": 1321
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.674647353200208e-06,
-      "loss": 0.01,
+      "loss": 0.0336,
       "step": 1322
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.660646673772583e-06,
-      "loss": 0.0067,
+      "loss": 0.046,
       "step": 1323
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.646650843722521e-06,
-      "loss": 0.0049,
+      "loss": 0.0253,
       "step": 1324
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.632659892062766e-06,
-      "loss": 0.0191,
+      "loss": 0.0366,
       "step": 1325
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.618673847795954e-06,
-      "loss": 0.0084,
+      "loss": 0.0522,
       "step": 1326
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.604692739914542e-06,
-      "loss": 0.0129,
+      "loss": 0.0367,
       "step": 1327
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.590716597400761e-06,
-      "loss": 0.0091,
+      "loss": 0.0487,
       "step": 1328
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.576745449226548e-06,
-      "loss": 0.0116,
+      "loss": 0.0428,
       "step": 1329
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.562779324353478e-06,
-      "loss": 0.0118,
+      "loss": 0.0362,
       "step": 1330
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.548818251732723e-06,
-      "loss": 0.013,
+      "loss": 0.0308,
       "step": 1331
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.5348622603049805e-06,
-      "loss": 0.0028,
+      "loss": 0.0448,
       "step": 1332
     },
     {
       "epoch": 1.18,
       "learning_rate": 7.520911379000411e-06,
-      "loss": 0.0084,
+      "loss": 0.0458,
       "step": 1333
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.506965636738582e-06,
-      "loss": 0.0056,
+      "loss": 0.0344,
       "step": 1334
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.493025062428413e-06,
-      "loss": 0.0063,
+      "loss": 0.035,
       "step": 1335
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.479089684968106e-06,
-      "loss": 0.0187,
+      "loss": 0.0433,
       "step": 1336
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.46515953324509e-06,
-      "loss": 0.009,
+      "loss": 0.0259,
       "step": 1337
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.451234636135969e-06,
-      "loss": 0.028,
+      "loss": 0.0377,
       "step": 1338
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.43731502250644e-06,
-      "loss": 0.0087,
+      "loss": 0.0453,
       "step": 1339
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.4234007212112605e-06,
-      "loss": 0.0132,
+      "loss": 0.0426,
       "step": 1340
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.40949176109417e-06,
-      "loss": 0.0126,
+      "loss": 0.0392,
       "step": 1341
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.395588170987839e-06,
-      "loss": 0.0024,
+      "loss": 0.0295,
       "step": 1342
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.3816899797138e-06,
-      "loss": 0.0084,
+      "loss": 0.0427,
       "step": 1343
     },
     {
       "epoch": 1.19,
       "learning_rate": 7.367797216082402e-06,
-      "loss": 0.016,
+      "loss": 0.0396,
       "step": 1344
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.353909908892737e-06,
-      "loss": 0.0158,
+      "loss": 0.0442,
       "step": 1345
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.340028086932587e-06,
-      "loss": 0.015,
+      "loss": 0.0359,
       "step": 1346
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.326151778978372e-06,
-      "loss": 0.0215,
+      "loss": 0.0349,
       "step": 1347
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.312281013795064e-06,
-      "loss": 0.0093,
+      "loss": 0.0356,
       "step": 1348
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.2984158201361585e-06,
-      "loss": 0.0077,
+      "loss": 0.0364,
       "step": 1349
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.2845562267435974e-06,
-      "loss": 0.0139,
+      "loss": 0.0373,
       "step": 1350
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.270702262347722e-06,
-      "loss": 0.0073,
+      "loss": 0.0306,
       "step": 1351
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.2568539556671865e-06,
-      "loss": 0.0117,
+      "loss": 0.0445,
       "step": 1352
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.2430113354089315e-06,
-      "loss": 0.0071,
+      "loss": 0.0514,
       "step": 1353
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.2291744302681045e-06,
-      "loss": 0.0137,
+      "loss": 0.0441,
       "step": 1354
     },
     {
       "epoch": 1.2,
       "learning_rate": 7.215343268928011e-06,
-      "loss": 0.0098,
+      "loss": 0.0344,
       "step": 1355
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.201517880060048e-06,
-      "loss": 0.0107,
+      "loss": 0.0525,
       "step": 1356
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.187698292323637e-06,
-      "loss": 0.0175,
+      "loss": 0.0441,
       "step": 1357
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.1738845343661826e-06,
-      "loss": 0.011,
+      "loss": 0.0317,
       "step": 1358
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.160076634823007e-06,
-      "loss": 0.0056,
+      "loss": 0.0267,
       "step": 1359
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.1462746223172885e-06,
-      "loss": 0.0067,
+      "loss": 0.0418,
       "step": 1360
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.1324785254599844e-06,
-      "loss": 0.0128,
+      "loss": 0.0322,
       "step": 1361
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.118688372849815e-06,
-      "loss": 0.0134,
+      "loss": 0.0331,
       "step": 1362
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.104904193073163e-06,
-      "loss": 0.0069,
+      "loss": 0.0383,
       "step": 1363
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.091126014704032e-06,
-      "loss": 0.0098,
+      "loss": 0.0264,
       "step": 1364
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.077353866303986e-06,
-      "loss": 0.0045,
+      "loss": 0.0301,
       "step": 1365
     },
     {
       "epoch": 1.21,
       "learning_rate": 7.063587776422088e-06,
-      "loss": 0.0071,
+      "loss": 0.0344,
       "step": 1366
     },
     {
       "epoch": 1.22,
       "learning_rate": 7.049827773594842e-06,
-      "loss": 0.0132,
+      "loss": 0.0309,
       "step": 1367
     },
     {
       "epoch": 1.22,
       "learning_rate": 7.036073886346137e-06,
-      "loss": 0.0147,
+      "loss": 0.0363,
       "step": 1368
     },
     {
       "epoch": 1.22,
       "learning_rate": 7.022326143187183e-06,
-      "loss": 0.0092,
+      "loss": 0.0375,
       "step": 1369
     },
     {
       "epoch": 1.22,
       "learning_rate": 7.008584572616448e-06,
-      "loss": 0.0036,
+      "loss": 0.0295,
       "step": 1370
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.9948492031196114e-06,
-      "loss": 0.0088,
+      "loss": 0.0348,
       "step": 1371
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.9811200631694985e-06,
-      "loss": 0.0052,
+      "loss": 0.0359,
       "step": 1372
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.967397181226017e-06,
-      "loss": 0.0054,
+      "loss": 0.0311,
       "step": 1373
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.953680585736105e-06,
-      "loss": 0.0055,
+      "loss": 0.0294,
       "step": 1374
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.939970305133663e-06,
-      "loss": 0.0086,
+      "loss": 0.0331,
       "step": 1375
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.9262663678395085e-06,
-      "loss": 0.0128,
+      "loss": 0.031,
       "step": 1376
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.912568802261307e-06,
-      "loss": 0.0032,
+      "loss": 0.0398,
       "step": 1377
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.898877636793517e-06,
-      "loss": 0.0114,
+      "loss": 0.0377,
       "step": 1378
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.885192899817324e-06,
-      "loss": 0.0094,
+      "loss": 0.0434,
       "step": 1379
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.8715146197005945e-06,
-      "loss": 0.0075,
+      "loss": 0.0346,
       "step": 1380
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.857842824797807e-06,
-      "loss": 0.0176,
+      "loss": 0.0346,
       "step": 1381
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.844177543449997e-06,
-      "loss": 0.0052,
+      "loss": 0.0307,
       "step": 1382
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.8305188039846985e-06,
-      "loss": 0.0104,
+      "loss": 0.0368,
       "step": 1383
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.816866634715881e-06,
-      "loss": 0.016,
+      "loss": 0.031,
       "step": 1384
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.803221063943899e-06,
-      "loss": 0.0061,
+      "loss": 0.0386,
       "step": 1385
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.789582119955425e-06,
-      "loss": 0.0145,
+      "loss": 0.0364,
       "step": 1386
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.775949831023398e-06,
-      "loss": 0.0084,
+      "loss": 0.0435,
       "step": 1387
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.7623242254069575e-06,
-      "loss": 0.0142,
+      "loss": 0.0448,
       "step": 1388
     },
     {
       "epoch": 1.23,
       "learning_rate": 6.748705331351389e-06,
-      "loss": 0.0121,
+      "loss": 0.0454,
       "step": 1389
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.7350931770880705e-06,
-      "loss": 0.0088,
+      "loss": 0.0403,
       "step": 1390
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.7214877908344e-06,
-      "loss": 0.0073,
+      "loss": 0.0481,
       "step": 1391
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.70788920079376e-06,
-      "loss": 0.0078,
+      "loss": 0.0344,
       "step": 1392
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.694297435155426e-06,
-      "loss": 0.0075,
+      "loss": 0.0356,
       "step": 1393
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.680712522094538e-06,
-      "loss": 0.0089,
+      "loss": 0.0299,
       "step": 1394
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.667134489772029e-06,
-      "loss": 0.0065,
+      "loss": 0.0486,
       "step": 1395
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.653563366334577e-06,
-      "loss": 0.0054,
+      "loss": 0.0393,
       "step": 1396
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.639999179914518e-06,
-      "loss": 0.0101,
+      "loss": 0.0404,
       "step": 1397
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.6264419586298255e-06,
-      "loss": 0.0107,
+      "loss": 0.0472,
       "step": 1398
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.612891730584026e-06,
-      "loss": 0.0088,
+      "loss": 0.0322,
       "step": 1399
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.599348523866155e-06,
-      "loss": 0.0085,
+      "loss": 0.0434,
       "step": 1400
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.585812366550695e-06,
-      "loss": 0.0045,
+      "loss": 0.0441,
       "step": 1401
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.572283286697496e-06,
-      "loss": 0.0036,
+      "loss": 0.0413,
       "step": 1402
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.558761312351764e-06,
-      "loss": 0.015,
+      "loss": 0.0309,
       "step": 1403
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.5452464715439605e-06,
-      "loss": 0.0078,
+      "loss": 0.0398,
       "step": 1404
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.531738792289764e-06,
-      "loss": 0.0052,
+      "loss": 0.0301,
       "step": 1405
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.518238302589994e-06,
-      "loss": 0.0092,
+      "loss": 0.0444,
       "step": 1406
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.504745030430588e-06,
-      "loss": 0.0073,
+      "loss": 0.0377,
       "step": 1407
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.491259003782511e-06,
-      "loss": 0.012,
+      "loss": 0.0372,
       "step": 1408
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.477780250601704e-06,
-      "loss": 0.0049,
+      "loss": 0.0408,
       "step": 1409
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.464308798829042e-06,
-      "loss": 0.0079,
+      "loss": 0.0331,
       "step": 1410
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.450844676390252e-06,
-      "loss": 0.0167,
+      "loss": 0.0393,
       "step": 1411
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.437387911195876e-06,
-      "loss": 0.0041,
+      "loss": 0.0324,
       "step": 1412
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.423938531141201e-06,
-      "loss": 0.0157,
+      "loss": 0.035,
       "step": 1413
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.4104965641062074e-06,
-      "loss": 0.0033,
+      "loss": 0.0321,
       "step": 1414
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.397062037955507e-06,
-      "loss": 0.0142,
+      "loss": 0.0359,
       "step": 1415
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.3836349805382865e-06,
-      "loss": 0.0071,
+      "loss": 0.0412,
       "step": 1416
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.370215419688253e-06,
-      "loss": 0.0083,
+      "loss": 0.0331,
       "step": 1417
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.356803383223568e-06,
-      "loss": 0.0054,
+      "loss": 0.0453,
       "step": 1418
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.343398898946805e-06,
-      "loss": 0.0066,
+      "loss": 0.0363,
       "step": 1419
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.330001994644871e-06,
-      "loss": 0.0064,
+      "loss": 0.0347,
       "step": 1420
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.316612698088966e-06,
-      "loss": 0.0023,
+      "loss": 0.0402,
       "step": 1421
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.303231037034521e-06,
-      "loss": 0.0067,
+      "loss": 0.0353,
       "step": 1422
     },
     {
       "epoch": 1.26,
       "learning_rate": 6.289857039221135e-06,
-      "loss": 0.0097,
+      "loss": 0.0316,
       "step": 1423
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.276490732372522e-06,
-      "loss": 0.0124,
+      "loss": 0.0392,
       "step": 1424
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.263132144196456e-06,
-      "loss": 0.0021,
+      "loss": 0.0302,
       "step": 1425
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.2497813023847054e-06,
-      "loss": 0.0069,
+      "loss": 0.0447,
       "step": 1426
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.236438234612987e-06,
-      "loss": 0.0046,
+      "loss": 0.0455,
       "step": 1427
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.2231029685409e-06,
-      "loss": 0.0192,
+      "loss": 0.0469,
       "step": 1428
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.209775531811868e-06,
-      "loss": 0.0124,
+      "loss": 0.0505,
       "step": 1429
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.196455952053085e-06,
-      "loss": 0.0015,
+      "loss": 0.0351,
       "step": 1430
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.183144256875462e-06,
-      "loss": 0.0047,
+      "loss": 0.0265,
       "step": 1431
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.169840473873567e-06,
-      "loss": 0.0103,
+      "loss": 0.0334,
       "step": 1432
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.156544630625555e-06,
-      "loss": 0.0089,
+      "loss": 0.0439,
       "step": 1433
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.143256754693134e-06,
-      "loss": 0.0179,
+      "loss": 0.0314,
       "step": 1434
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.129976873621495e-06,
-      "loss": 0.0058,
+      "loss": 0.0334,
       "step": 1435
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.116705014939246e-06,
-      "loss": 0.0084,
+      "loss": 0.0411,
       "step": 1436
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.103441206158383e-06,
-      "loss": 0.0068,
+      "loss": 0.0367,
       "step": 1437
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.090185474774192e-06,
-      "loss": 0.0058,
+      "loss": 0.0265,
       "step": 1438
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.076937848265233e-06,
-      "loss": 0.0112,
+      "loss": 0.0439,
       "step": 1439
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.063698354093255e-06,
-      "loss": 0.0087,
+      "loss": 0.0348,
       "step": 1440
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.050467019703163e-06,
-      "loss": 0.0055,
+      "loss": 0.0398,
       "step": 1441
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.0372438725229246e-06,
-      "loss": 0.014,
+      "loss": 0.0379,
       "step": 1442
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.024028939963552e-06,
-      "loss": 0.0114,
+      "loss": 0.0427,
       "step": 1443
     },
     {
       "epoch": 1.28,
       "learning_rate": 6.010822249419027e-06,
-      "loss": 0.0039,
+      "loss": 0.0384,
       "step": 1444
     },
     {
       "epoch": 1.28,
       "learning_rate": 5.997623828266243e-06,
-      "loss": 0.0068,
+      "loss": 0.0344,
       "step": 1445
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.984433703864956e-06,
-      "loss": 0.0109,
+      "loss": 0.0418,
       "step": 1446
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.971251903557711e-06,
-      "loss": 0.0065,
+      "loss": 0.0423,
       "step": 1447
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.95807845466981e-06,
-      "loss": 0.0078,
+      "loss": 0.0298,
       "step": 1448
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.944913384509241e-06,
-      "loss": 0.0076,
+      "loss": 0.0331,
       "step": 1449
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.9317567203666215e-06,
-      "loss": 0.0105,
+      "loss": 0.0327,
       "step": 1450
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.918608489515136e-06,
-      "loss": 0.0091,
+      "loss": 0.0317,
       "step": 1451
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.905468719210503e-06,
-      "loss": 0.0093,
+      "loss": 0.0418,
       "step": 1452
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.892337436690889e-06,
-      "loss": 0.0079,
+      "loss": 0.0343,
       "step": 1453
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.8792146691768725e-06,
-      "loss": 0.0038,
+      "loss": 0.0366,
       "step": 1454
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.8661004438713805e-06,
-      "loss": 0.0092,
+      "loss": 0.0225,
       "step": 1455
     },
     {
       "epoch": 1.29,
       "learning_rate": 5.8529947879596286e-06,
-      "loss": 0.0062,
+      "loss": 0.0416,
       "step": 1456
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.839897728609071e-06,
-      "loss": 0.0034,
+      "loss": 0.0411,
       "step": 1457
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.826809292969341e-06,
-      "loss": 0.0073,
+      "loss": 0.0356,
       "step": 1458
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.813729508172199e-06,
-      "loss": 0.004,
+      "loss": 0.033,
       "step": 1459
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.800658401331467e-06,
-      "loss": 0.0171,
+      "loss": 0.035,
       "step": 1460
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.787595999542974e-06,
-      "loss": 0.0066,
+      "loss": 0.0403,
       "step": 1461
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.7745423298845185e-06,
-      "loss": 0.0072,
+      "loss": 0.0393,
       "step": 1462
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.761497419415789e-06,
-      "loss": 0.004,
+      "loss": 0.0397,
       "step": 1463
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.748461295178315e-06,
-      "loss": 0.009,
+      "loss": 0.0306,
       "step": 1464
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.735433984195413e-06,
-      "loss": 0.0101,
+      "loss": 0.0283,
       "step": 1465
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.722415513472128e-06,
-      "loss": 0.0145,
+      "loss": 0.0416,
       "step": 1466
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.709405909995187e-06,
-      "loss": 0.0103,
+      "loss": 0.0358,
       "step": 1467
     },
     {
       "epoch": 1.3,
       "learning_rate": 5.69640520073294e-06,
-      "loss": 0.0091,
+      "loss": 0.0356,
       "step": 1468
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.683413412635273e-06,
-      "loss": 0.006,
+      "loss": 0.038,
       "step": 1469
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.670430572633607e-06,
-      "loss": 0.0108,
+      "loss": 0.042,
       "step": 1470
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.657456707640806e-06,
-      "loss": 0.0099,
+      "loss": 0.0415,
       "step": 1471
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.644491844551121e-06,
-      "loss": 0.0074,
+      "loss": 0.0342,
       "step": 1472
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.6315360102401594e-06,
-      "loss": 0.0056,
+      "loss": 0.028,
       "step": 1473
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.618589231564786e-06,
-      "loss": 0.0098,
+      "loss": 0.0283,
       "step": 1474
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.605651535363117e-06,
-      "loss": 0.0169,
+      "loss": 0.04,
       "step": 1475
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.592722948454438e-06,
-      "loss": 0.0079,
+      "loss": 0.026,
       "step": 1476
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.57980349763914e-06,
-      "loss": 0.0035,
+      "loss": 0.0433,
       "step": 1477
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.5668932096986805e-06,
-      "loss": 0.0065,
+      "loss": 0.0355,
       "step": 1478
     },
     {
       "epoch": 1.31,
       "learning_rate": 5.553992111395532e-06,
-      "loss": 0.0065,
+      "loss": 0.0245,
       "step": 1479
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.5411002294731e-06,
-      "loss": 0.013,
+      "loss": 0.0356,
       "step": 1480
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.528217590655702e-06,
-      "loss": 0.0117,
+      "loss": 0.0391,
       "step": 1481
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.515344221648484e-06,
-      "loss": 0.0052,
+      "loss": 0.0275,
       "step": 1482
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.502480149137377e-06,
-      "loss": 0.0072,
+      "loss": 0.0341,
       "step": 1483
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.489625399789048e-06,
-      "loss": 0.0038,
+      "loss": 0.0412,
       "step": 1484
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.476780000250828e-06,
-      "loss": 0.0071,
+      "loss": 0.034,
       "step": 1485
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.4639439771506745e-06,
-      "loss": 0.0097,
+      "loss": 0.0378,
       "step": 1486
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.451117357097103e-06,
-      "loss": 0.013,
+      "loss": 0.0402,
       "step": 1487
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.438300166679135e-06,
-      "loss": 0.0154,
+      "loss": 0.0315,
       "step": 1488
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.425492432466254e-06,
-      "loss": 0.0233,
+      "loss": 0.0452,
       "step": 1489
     },
     {
       "epoch": 1.32,
       "learning_rate": 5.412694181008329e-06,
-      "loss": 0.0082,
+      "loss": 0.0317,
       "step": 1490
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.399905438835585e-06,
-      "loss": 0.0053,
+      "loss": 0.0301,
       "step": 1491
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.387126232458522e-06,
-      "loss": 0.0095,
+      "loss": 0.0226,
       "step": 1492
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.374356588367877e-06,
-      "loss": 0.0107,
+      "loss": 0.027,
       "step": 1493
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.361596533034572e-06,
-      "loss": 0.0077,
+      "loss": 0.0404,
       "step": 1494
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.348846092909643e-06,
-      "loss": 0.0068,
+      "loss": 0.0299,
       "step": 1495
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.3361052944241945e-06,
-      "loss": 0.0044,
+      "loss": 0.033,
       "step": 1496
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.323374163989352e-06,
-      "loss": 0.0021,
+      "loss": 0.0302,
       "step": 1497
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.310652727996189e-06,
-      "loss": 0.0085,
+      "loss": 0.0365,
       "step": 1498
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.297941012815691e-06,
-      "loss": 0.0057,
+      "loss": 0.0275,
       "step": 1499
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.285239044798696e-06,
-      "loss": 0.0052,
+      "loss": 0.0306,
       "step": 1500
     },
     {
       "epoch": 1.33,
       "learning_rate": 5.272546850275817e-06,
-      "loss": 0.0067,
+      "loss": 0.0395,
       "step": 1501
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.259864455557433e-06,
-      "loss": 0.0103,
+      "loss": 0.0456,
       "step": 1502
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.247191886933587e-06,
-      "loss": 0.0099,
+      "loss": 0.0383,
       "step": 1503
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.234529170673969e-06,
-      "loss": 0.0108,
+      "loss": 0.0281,
       "step": 1504
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.2218763330278356e-06,
-      "loss": 0.0063,
+      "loss": 0.0229,
       "step": 1505
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.209233400223963e-06,
-      "loss": 0.0035,
+      "loss": 0.0384,
       "step": 1506
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.196600398470605e-06,
-      "loss": 0.0226,
+      "loss": 0.037,
       "step": 1507
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.183977353955427e-06,
-      "loss": 0.0115,
+      "loss": 0.0267,
       "step": 1508
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.171364292845449e-06,
-      "loss": 0.0146,
+      "loss": 0.0415,
       "step": 1509
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.158761241286996e-06,
-      "loss": 0.01,
+      "loss": 0.0436,
       "step": 1510
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.14616822540564e-06,
-      "loss": 0.009,
+      "loss": 0.0302,
       "step": 1511
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.13358527130616e-06,
-      "loss": 0.0037,
+      "loss": 0.0312,
       "step": 1512
     },
     {
       "epoch": 1.34,
       "learning_rate": 5.121012405072479e-06,
-      "loss": 0.0107,
+      "loss": 0.0208,
       "step": 1513
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.108449652767584e-06,
-      "loss": 0.0115,
+      "loss": 0.0278,
       "step": 1514
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.095897040433524e-06,
-      "loss": 0.0057,
+      "loss": 0.0454,
       "step": 1515
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.083354594091318e-06,
-      "loss": 0.01,
+      "loss": 0.0456,
       "step": 1516
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.070822339740909e-06,
-      "loss": 0.0049,
+      "loss": 0.0334,
       "step": 1517
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.058300303361117e-06,
-      "loss": 0.0056,
+      "loss": 0.0416,
       "step": 1518
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.045788510909575e-06,
-      "loss": 0.0068,
+      "loss": 0.029,
       "step": 1519
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.033286988322682e-06,
-      "loss": 0.0098,
+      "loss": 0.0253,
       "step": 1520
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.0207957615155555e-06,
-      "loss": 0.0066,
+      "loss": 0.037,
       "step": 1521
     },
     {
       "epoch": 1.35,
       "learning_rate": 5.008314856381962e-06,
-      "loss": 0.0048,
+      "loss": 0.0325,
       "step": 1522
     },
     {
       "epoch": 1.35,
       "learning_rate": 4.995844298794271e-06,
-      "loss": 0.0042,
+      "loss": 0.0357,
       "step": 1523
     },
     {
       "epoch": 1.35,
       "learning_rate": 4.983384114603412e-06,
-      "loss": 0.0046,
+      "loss": 0.0408,
       "step": 1524
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.9709343296387954e-06,
-      "loss": 0.0056,
+      "loss": 0.0349,
       "step": 1525
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.958494969708294e-06,
-      "loss": 0.01,
+      "loss": 0.0443,
       "step": 1526
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.946066060598151e-06,
-      "loss": 0.002,
+      "loss": 0.0353,
       "step": 1527
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.933647628072954e-06,
-      "loss": 0.0154,
+      "loss": 0.0431,
       "step": 1528
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.921239697875579e-06,
-      "loss": 0.002,
+      "loss": 0.0279,
       "step": 1529
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.908842295727116e-06,
-      "loss": 0.0025,
+      "loss": 0.0327,
       "step": 1530
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.896455447326849e-06,
-      "loss": 0.0039,
+      "loss": 0.0297,
       "step": 1531
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.88407917835217e-06,
-      "loss": 0.0119,
+      "loss": 0.0487,
       "step": 1532
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.871713514458543e-06,
-      "loss": 0.0041,
+      "loss": 0.0364,
       "step": 1533
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.859358481279458e-06,
-      "loss": 0.0024,
+      "loss": 0.027,
       "step": 1534
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.847014104426353e-06,
-      "loss": 0.0052,
+      "loss": 0.0371,
       "step": 1535
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.83468040948859e-06,
-      "loss": 0.0168,
+      "loss": 0.0462,
       "step": 1536
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.82235742203338e-06,
-      "loss": 0.0103,
+      "loss": 0.0441,
       "step": 1537
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.810045167605735e-06,
-      "loss": 0.0096,
+      "loss": 0.0402,
       "step": 1538
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.7977436717284275e-06,
-      "loss": 0.0083,
+      "loss": 0.0376,
       "step": 1539
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.785452959901922e-06,
-      "loss": 0.0098,
+      "loss": 0.0306,
       "step": 1540
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.773173057604322e-06,
-      "loss": 0.0025,
+      "loss": 0.0333,
       "step": 1541
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.76090399029134e-06,
-      "loss": 0.0064,
+      "loss": 0.0288,
       "step": 1542
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.748645783396209e-06,
-      "loss": 0.0082,
+      "loss": 0.0419,
       "step": 1543
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.736398462329664e-06,
-      "loss": 0.0101,
+      "loss": 0.0349,
       "step": 1544
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.724162052479875e-06,
-      "loss": 0.0032,
+      "loss": 0.0346,
       "step": 1545
     },
     {
       "epoch": 1.37,
       "learning_rate": 4.711936579212369e-06,
-      "loss": 0.0077,
+      "loss": 0.0329,
       "step": 1546
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.699722067870035e-06,
-      "loss": 0.0102,
+      "loss": 0.0288,
       "step": 1547
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.687518543773013e-06,
-      "loss": 0.0095,
+      "loss": 0.0343,
       "step": 1548
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.675326032218685e-06,
-      "loss": 0.0069,
+      "loss": 0.0364,
       "step": 1549
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.663144558481593e-06,
-      "loss": 0.0058,
+      "loss": 0.0304,
       "step": 1550
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.650974147813396e-06,
-      "loss": 0.0024,
+      "loss": 0.0387,
       "step": 1551
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.6388148254428325e-06,
-      "loss": 0.0036,
+      "loss": 0.0271,
       "step": 1552
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.626666616575648e-06,
-      "loss": 0.0091,
+      "loss": 0.0506,
       "step": 1553
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.6145295463945495e-06,
-      "loss": 0.0082,
+      "loss": 0.0393,
       "step": 1554
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.602403640059154e-06,
-      "loss": 0.0032,
+      "loss": 0.0388,
       "step": 1555
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.590288922705934e-06,
-      "loss": 0.0088,
+      "loss": 0.035,
       "step": 1556
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.578185419448174e-06,
-      "loss": 0.0025,
+      "loss": 0.0272,
       "step": 1557
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.566093155375916e-06,
-      "loss": 0.0053,
+      "loss": 0.0275,
       "step": 1558
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.554012155555883e-06,
-      "loss": 0.0072,
+      "loss": 0.0308,
       "step": 1559
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.54194244503147e-06,
-      "loss": 0.0166,
+      "loss": 0.0333,
       "step": 1560
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.5298840488226615e-06,
-      "loss": 0.0074,
+      "loss": 0.0354,
       "step": 1561
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.517836991925983e-06,
-      "loss": 0.0025,
+      "loss": 0.0258,
       "step": 1562
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.505801299314466e-06,
-      "loss": 0.0081,
+      "loss": 0.0426,
       "step": 1563
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.493776995937573e-06,
-      "loss": 0.0087,
+      "loss": 0.035,
       "step": 1564
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.481764106721157e-06,
-      "loss": 0.0077,
+      "loss": 0.0497,
       "step": 1565
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.469762656567422e-06,
-      "loss": 0.0082,
+      "loss": 0.0394,
       "step": 1566
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.457772670354848e-06,
-      "loss": 0.0112,
+      "loss": 0.0366,
       "step": 1567
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.4457941729381494e-06,
-      "loss": 0.0019,
+      "loss": 0.024,
       "step": 1568
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.4338271891482385e-06,
-      "loss": 0.0019,
+      "loss": 0.0252,
       "step": 1569
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.421871743792144e-06,
-      "loss": 0.0132,
+      "loss": 0.0364,
       "step": 1570
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.4099278616529905e-06,
-      "loss": 0.0117,
+      "loss": 0.0291,
       "step": 1571
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.397995567489921e-06,
-      "loss": 0.0076,
+      "loss": 0.0227,
       "step": 1572
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.386074886038059e-06,
-      "loss": 0.0032,
+      "loss": 0.0217,
       "step": 1573
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.374165842008464e-06,
-      "loss": 0.0079,
+      "loss": 0.045,
       "step": 1574
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.36226846008806e-06,
-      "loss": 0.0022,
+      "loss": 0.0418,
       "step": 1575
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.350382764939609e-06,
-      "loss": 0.0182,
+      "loss": 0.0359,
       "step": 1576
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.338508781201635e-06,
-      "loss": 0.0024,
+      "loss": 0.039,
       "step": 1577
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.326646533488384e-06,
-      "loss": 0.0038,
+      "loss": 0.0319,
       "step": 1578
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.3147960463897885e-06,
-      "loss": 0.0073,
+      "loss": 0.0314,
       "step": 1579
     },
     {
       "epoch": 1.4,
       "learning_rate": 4.3029573444713836e-06,
-      "loss": 0.0087,
+      "loss": 0.0283,
       "step": 1580
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.291130452274291e-06,
-      "loss": 0.0086,
+      "loss": 0.0325,
       "step": 1581
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.279315394315138e-06,
-      "loss": 0.0081,
+      "loss": 0.028,
       "step": 1582
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.267512195086022e-06,
-      "loss": 0.0085,
+      "loss": 0.0258,
       "step": 1583
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.255720879054469e-06,
-      "loss": 0.0094,
+      "loss": 0.0382,
       "step": 1584
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.243941470663357e-06,
-      "loss": 0.0033,
+      "loss": 0.0454,
       "step": 1585
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.232173994330893e-06,
-      "loss": 0.0139,
+      "loss": 0.0315,
       "step": 1586
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.220418474450537e-06,
-      "loss": 0.0058,
+      "loss": 0.0341,
       "step": 1587
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.20867493539097e-06,
-      "loss": 0.0085,
+      "loss": 0.038,
       "step": 1588
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.196943401496038e-06,
-      "loss": 0.0071,
+      "loss": 0.0489,
       "step": 1589
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.185223897084709e-06,
-      "loss": 0.0023,
+      "loss": 0.0334,
       "step": 1590
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.173516446450989e-06,
-      "loss": 0.0033,
+      "loss": 0.0339,
       "step": 1591
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.1618210738639255e-06,
-      "loss": 0.0068,
+      "loss": 0.0387,
       "step": 1592
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.150137803567509e-06,
-      "loss": 0.0081,
+      "loss": 0.0317,
       "step": 1593
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.138466659780652e-06,
-      "loss": 0.0053,
+      "loss": 0.0344,
       "step": 1594
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.126807666697137e-06,
-      "loss": 0.0036,
+      "loss": 0.0286,
       "step": 1595
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.115160848485532e-06,
-      "loss": 0.0117,
+      "loss": 0.0378,
       "step": 1596
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.103526229289193e-06,
-      "loss": 0.0071,
+      "loss": 0.0276,
       "step": 1597
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.091903833226182e-06,
-      "loss": 0.0068,
+      "loss": 0.0324,
       "step": 1598
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.080293684389216e-06,
-      "loss": 0.0135,
+      "loss": 0.0199,
       "step": 1599
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.068695806845624e-06,
-      "loss": 0.0036,
+      "loss": 0.0347,
       "step": 1600
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.057110224637309e-06,
-      "loss": 0.0064,
+      "loss": 0.033,
       "step": 1601
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.045536961780671e-06,
-      "loss": 0.0053,
+      "loss": 0.0251,
       "step": 1602
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.033976042266585e-06,
-      "loss": 0.0037,
+      "loss": 0.0325,
       "step": 1603
     },
     {
       "epoch": 1.43,
       "learning_rate": 4.022427490060334e-06,
-      "loss": 0.0035,
+      "loss": 0.0198,
       "step": 1604
     },
     {
       "epoch": 1.43,
       "learning_rate": 4.010891329101557e-06,
-      "loss": 0.005,
+      "loss": 0.0283,
       "step": 1605
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.999367583304222e-06,
-      "loss": 0.0127,
+      "loss": 0.0372,
       "step": 1606
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.9878562765565444e-06,
-      "loss": 0.0137,
+      "loss": 0.0325,
       "step": 1607
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.9763574327209685e-06,
-      "loss": 0.0041,
+      "loss": 0.0407,
       "step": 1608
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.964871075634095e-06,
-      "loss": 0.0021,
+      "loss": 0.0257,
       "step": 1609
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.953397229106636e-06,
-      "loss": 0.0124,
+      "loss": 0.0301,
       "step": 1610
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.941935916923384e-06,
-      "loss": 0.016,
+      "loss": 0.0382,
       "step": 1611
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.930487162843132e-06,
-      "loss": 0.0047,
+      "loss": 0.0365,
       "step": 1612
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.919050990598657e-06,
-      "loss": 0.0121,
+      "loss": 0.04,
       "step": 1613
     },
     {
       "epoch": 1.43,
       "learning_rate": 3.907627423896642e-06,
-      "loss": 0.006,
+      "loss": 0.0334,
       "step": 1614
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.896216486417639e-06,
-      "loss": 0.0014,
+      "loss": 0.0277,
       "step": 1615
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.8848182018160314e-06,
-      "loss": 0.0128,
+      "loss": 0.0344,
       "step": 1616
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.873432593719965e-06,
-      "loss": 0.0031,
+      "loss": 0.0293,
       "step": 1617
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.8620596857313046e-06,
-      "loss": 0.0066,
+      "loss": 0.036,
       "step": 1618
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.8506995014256e-06,
-      "loss": 0.0092,
+      "loss": 0.036,
       "step": 1619
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.839352064352012e-06,
-      "loss": 0.0123,
+      "loss": 0.0307,
       "step": 1620
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.828017398033291e-06,
-      "loss": 0.005,
+      "loss": 0.0317,
       "step": 1621
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.816695525965702e-06,
-      "loss": 0.0099,
+      "loss": 0.0322,
       "step": 1622
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.8053864716189903e-06,
-      "loss": 0.0046,
+      "loss": 0.0227,
       "step": 1623
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.7940902584363383e-06,
-      "loss": 0.0073,
+      "loss": 0.0397,
       "step": 1624
     },
     {
       "epoch": 1.44,
       "learning_rate": 3.782806909834299e-06,
-      "loss": 0.0025,
+      "loss": 0.0349,
       "step": 1625
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7715364492027685e-06,
-      "loss": 0.0053,
+      "loss": 0.0272,
       "step": 1626
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7602788999049167e-06,
-      "loss": 0.0036,
+      "loss": 0.0385,
       "step": 1627
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7490342852771523e-06,
-      "loss": 0.0056,
+      "loss": 0.0247,
       "step": 1628
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7378026286290747e-06,
-      "loss": 0.0046,
+      "loss": 0.0317,
       "step": 1629
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7265839532434154e-06,
-      "loss": 0.005,
+      "loss": 0.0289,
       "step": 1630
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7153782823760054e-06,
-      "loss": 0.0091,
+      "loss": 0.0271,
       "step": 1631
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.7041856392557084e-06,
-      "loss": 0.0082,
+      "loss": 0.0356,
       "step": 1632
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.693006047084383e-06,
-      "loss": 0.0021,
+      "loss": 0.0293,
       "step": 1633
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.6818395290368415e-06,
-      "loss": 0.0036,
+      "loss": 0.0346,
       "step": 1634
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.670686108260796e-06,
-      "loss": 0.0043,
+      "loss": 0.0336,
       "step": 1635
     },
     {
       "epoch": 1.45,
       "learning_rate": 3.6595458078767864e-06,
-      "loss": 0.0048,
+      "loss": 0.0319,
       "step": 1636
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.6484186509781826e-06,
-      "loss": 0.0045,
+      "loss": 0.0336,
       "step": 1637
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.637304660631088e-06,
-      "loss": 0.0049,
+      "loss": 0.046,
       "step": 1638
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.6262038598743254e-06,
-      "loss": 0.0019,
+      "loss": 0.0419,
       "step": 1639
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.6151162717193776e-06,
-      "loss": 0.0064,
+      "loss": 0.0191,
       "step": 1640
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.6040419191503173e-06,
-      "loss": 0.0129,
+      "loss": 0.0294,
       "step": 1641
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.592980825123803e-06,
-      "loss": 0.0022,
+      "loss": 0.0457,
       "step": 1642
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.5819330125690044e-06,
-      "loss": 0.0095,
+      "loss": 0.0321,
       "step": 1643
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.570898504387551e-06,
-      "loss": 0.0021,
+      "loss": 0.0242,
       "step": 1644
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.5598773234534955e-06,
-      "loss": 0.0097,
+      "loss": 0.0361,
       "step": 1645
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.548869492613273e-06,
-      "loss": 0.0082,
+      "loss": 0.0378,
       "step": 1646
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.5378750346856294e-06,
-      "loss": 0.0058,
+      "loss": 0.034,
       "step": 1647
     },
     {
       "epoch": 1.46,
       "learning_rate": 3.5268939724616056e-06,
-      "loss": 0.0057,
+      "loss": 0.0366,
       "step": 1648
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.51592632870446e-06,
-      "loss": 0.0017,
+      "loss": 0.038,
       "step": 1649
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.5049721261496396e-06,
-      "loss": 0.0054,
+      "loss": 0.0389,
       "step": 1650
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.4940313875047325e-06,
-      "loss": 0.0045,
+      "loss": 0.0307,
       "step": 1651
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.48310413544941e-06,
-      "loss": 0.0088,
+      "loss": 0.0259,
       "step": 1652
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.472190392635395e-06,
-      "loss": 0.0037,
+      "loss": 0.029,
       "step": 1653
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.4612901816863975e-06,
-      "loss": 0.0019,
+      "loss": 0.0354,
       "step": 1654
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.450403525198076e-06,
-      "loss": 0.0053,
+      "loss": 0.0395,
       "step": 1655
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.439530445738004e-06,
-      "loss": 0.0019,
+      "loss": 0.0338,
       "step": 1656
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.428670965845594e-06,
-      "loss": 0.0123,
+      "loss": 0.0453,
       "step": 1657
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.417825108032081e-06,
-      "loss": 0.0041,
+      "loss": 0.0308,
       "step": 1658
     },
     {
       "epoch": 1.47,
       "learning_rate": 3.406992894780453e-06,
-      "loss": 0.0042,
+      "loss": 0.0236,
       "step": 1659
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3961743485454133e-06,
-      "loss": 0.0069,
+      "loss": 0.0272,
       "step": 1660
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.385369491753344e-06,
-      "loss": 0.0018,
+      "loss": 0.0315,
       "step": 1661
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3745783468022387e-06,
-      "loss": 0.013,
+      "loss": 0.0358,
       "step": 1662
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3638009360616664e-06,
-      "loss": 0.0057,
+      "loss": 0.0302,
       "step": 1663
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3530372818727395e-06,
-      "loss": 0.0026,
+      "loss": 0.0238,
       "step": 1664
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.342287406548035e-06,
-      "loss": 0.0118,
+      "loss": 0.0247,
       "step": 1665
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3315513323715854e-06,
-      "loss": 0.0085,
+      "loss": 0.0338,
       "step": 1666
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.3208290815987997e-06,
-      "loss": 0.0035,
+      "loss": 0.0302,
       "step": 1667
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.310120676456434e-06,
-      "loss": 0.0039,
+      "loss": 0.0498,
       "step": 1668
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.299426139142553e-06,
-      "loss": 0.0038,
+      "loss": 0.0423,
       "step": 1669
     },
     {
       "epoch": 1.48,
       "learning_rate": 3.2887454918264593e-06,
-      "loss": 0.0067,
+      "loss": 0.0311,
       "step": 1670
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.2780787566486773e-06,
-      "loss": 0.0064,
+      "loss": 0.0271,
       "step": 1671
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.267425955720881e-06,
-      "loss": 0.003,
+      "loss": 0.0292,
       "step": 1672
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.2567871111258575e-06,
-      "loss": 0.0105,
+      "loss": 0.0272,
       "step": 1673
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.246162244917478e-06,
-      "loss": 0.0077,
+      "loss": 0.0262,
       "step": 1674
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.2355513791206182e-06,
-      "loss": 0.0078,
+      "loss": 0.0311,
       "step": 1675
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.2249545357311486e-06,
-      "loss": 0.0016,
+      "loss": 0.0203,
       "step": 1676
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.21437173671586e-06,
-      "loss": 0.0062,
+      "loss": 0.0294,
       "step": 1677
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.2038030040124324e-06,
-      "loss": 0.0041,
+      "loss": 0.029,
       "step": 1678
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.1932483595293896e-06,
-      "loss": 0.0009,
+      "loss": 0.0265,
       "step": 1679
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.1827078251460564e-06,
-      "loss": 0.004,
+      "loss": 0.0291,
       "step": 1680
     },
     {
       "epoch": 1.49,
       "learning_rate": 3.1721814227124903e-06,
-      "loss": 0.0073,
+      "loss": 0.0245,
       "step": 1681
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.1616691740494743e-06,
-      "loss": 0.0059,
+      "loss": 0.0271,
       "step": 1682
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.151171100948435e-06,
-      "loss": 0.0082,
+      "loss": 0.0288,
       "step": 1683
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.1406872251714238e-06,
-      "loss": 0.0041,
+      "loss": 0.0277,
       "step": 1684
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.130217568451065e-06,
-      "loss": 0.0174,
+      "loss": 0.0283,
       "step": 1685
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.1197621524904885e-06,
-      "loss": 0.0035,
+      "loss": 0.0303,
       "step": 1686
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.109320998963323e-06,
-      "loss": 0.0104,
+      "loss": 0.036,
       "step": 1687
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.098894129513629e-06,
-      "loss": 0.0077,
+      "loss": 0.0234,
       "step": 1688
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.0884815657558474e-06,
-      "loss": 0.0038,
+      "loss": 0.032,
       "step": 1689
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.078083329274767e-06,
-      "loss": 0.0029,
+      "loss": 0.038,
       "step": 1690
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.067699441625486e-06,
-      "loss": 0.0051,
+      "loss": 0.0268,
       "step": 1691
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.0573299243333452e-06,
-      "loss": 0.0054,
+      "loss": 0.0325,
       "step": 1692
     },
     {
       "epoch": 1.5,
       "learning_rate": 3.046974798893908e-06,
-      "loss": 0.0047,
+      "loss": 0.0324,
       "step": 1693
     },
     {
       "epoch": 1.51,
       "learning_rate": 3.0366340867728947e-06,
-      "loss": 0.0053,
+      "loss": 0.0288,
       "step": 1694
     },
     {
       "epoch": 1.51,
       "learning_rate": 3.0263078094061493e-06,
-      "loss": 0.0033,
+      "loss": 0.0429,
       "step": 1695
     },
     {
       "epoch": 1.51,
       "learning_rate": 3.0159959881996e-06,
-      "loss": 0.0047,
+      "loss": 0.0356,
       "step": 1696
     },
     {
       "epoch": 1.51,
       "learning_rate": 3.005698644529199e-06,
-      "loss": 0.0014,
+      "loss": 0.0394,
       "step": 1697
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9954157997408963e-06,
-      "loss": 0.002,
+      "loss": 0.0275,
       "step": 1698
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.985147475150577e-06,
-      "loss": 0.0109,
+      "loss": 0.0323,
       "step": 1699
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9748936920440284e-06,
-      "loss": 0.0137,
+      "loss": 0.0265,
       "step": 1700
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9646544716769033e-06,
-      "loss": 0.0101,
+      "loss": 0.0261,
       "step": 1701
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9544298352746525e-06,
-      "loss": 0.0034,
+      "loss": 0.0375,
       "step": 1702
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9442198040325087e-06,
-      "loss": 0.0087,
+      "loss": 0.0429,
       "step": 1703
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.9340243991154195e-06,
-      "loss": 0.0061,
+      "loss": 0.0311,
       "step": 1704
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.923843641658011e-06,
-      "loss": 0.0021,
+      "loss": 0.0267,
       "step": 1705
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.9136775527645557e-06,
-      "loss": 0.0073,
+      "loss": 0.0325,
       "step": 1706
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.903526153508912e-06,
-      "loss": 0.0043,
+      "loss": 0.0292,
       "step": 1707
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.893389464934484e-06,
-      "loss": 0.0038,
+      "loss": 0.0412,
       "step": 1708
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.8832675080541918e-06,
-      "loss": 0.0097,
+      "loss": 0.0405,
       "step": 1709
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.8731603038504054e-06,
-      "loss": 0.0066,
+      "loss": 0.0373,
       "step": 1710
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.8630678732749253e-06,
-      "loss": 0.001,
+      "loss": 0.0295,
       "step": 1711
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.8529902372489172e-06,
-      "loss": 0.0054,
+      "loss": 0.0284,
       "step": 1712
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.8429274166628785e-06,
-      "loss": 0.0069,
+      "loss": 0.0257,
       "step": 1713
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.832879432376604e-06,
-      "loss": 0.0066,
+      "loss": 0.04,
       "step": 1714
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.822846305219121e-06,
-      "loss": 0.0048,
+      "loss": 0.0315,
       "step": 1715
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.812828055988673e-06,
-      "loss": 0.0114,
+      "loss": 0.0195,
       "step": 1716
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.8028247054526504e-06,
-      "loss": 0.0039,
+      "loss": 0.0253,
       "step": 1717
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.792836274347559e-06,
-      "loss": 0.0037,
+      "loss": 0.0264,
       "step": 1718
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.782862783378989e-06,
-      "loss": 0.001,
+      "loss": 0.0245,
       "step": 1719
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.7729042532215455e-06,
-      "loss": 0.01,
+      "loss": 0.0321,
       "step": 1720
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.7629607045188346e-06,
-      "loss": 0.0098,
+      "loss": 0.0239,
       "step": 1721
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.7530321578833976e-06,
-      "loss": 0.0113,
+      "loss": 0.0303,
       "step": 1722
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.7431186338966744e-06,
-      "loss": 0.0048,
+      "loss": 0.0502,
       "step": 1723
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.733220153108973e-06,
-      "loss": 0.0028,
+      "loss": 0.0248,
       "step": 1724
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.723336736039418e-06,
-      "loss": 0.0038,
+      "loss": 0.0384,
       "step": 1725
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.713468403175891e-06,
-      "loss": 0.006,
+      "loss": 0.0327,
       "step": 1726
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.7036151749750226e-06,
-      "loss": 0.0051,
+      "loss": 0.0287,
       "step": 1727
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6937770718621282e-06,
-      "loss": 0.0082,
+      "loss": 0.0297,
       "step": 1728
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6839541142311598e-06,
-      "loss": 0.007,
+      "loss": 0.0335,
       "step": 1729
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6741463224446928e-06,
-      "loss": 0.0062,
+      "loss": 0.0314,
       "step": 1730
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6643537168338375e-06,
-      "loss": 0.0052,
+      "loss": 0.0274,
       "step": 1731
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6545763176982477e-06,
-      "loss": 0.0059,
+      "loss": 0.0294,
       "step": 1732
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.644814145306046e-06,
-      "loss": 0.007,
+      "loss": 0.0284,
       "step": 1733
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.635067219893789e-06,
-      "loss": 0.0043,
+      "loss": 0.0276,
       "step": 1734
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6253355616664254e-06,
-      "loss": 0.0017,
+      "loss": 0.0377,
       "step": 1735
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6156191907972663e-06,
-      "loss": 0.0019,
+      "loss": 0.0262,
       "step": 1736
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.6059181274279178e-06,
-      "loss": 0.0112,
+      "loss": 0.0301,
       "step": 1737
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.5962323916682676e-06,
-      "loss": 0.0056,
+      "loss": 0.0216,
       "step": 1738
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.586562003596421e-06,
-      "loss": 0.0059,
+      "loss": 0.0243,
       "step": 1739
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.576906983258669e-06,
-      "loss": 0.0073,
+      "loss": 0.0265,
       "step": 1740
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.5672673506694558e-06,
-      "loss": 0.0076,
+      "loss": 0.0308,
       "step": 1741
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.55764312581131e-06,
-      "loss": 0.0077,
+      "loss": 0.04,
       "step": 1742
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.5480343286348406e-06,
-      "loss": 0.0021,
+      "loss": 0.0324,
       "step": 1743
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.5384409790586594e-06,
-      "loss": 0.0066,
+      "loss": 0.0357,
       "step": 1744
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.528863096969362e-06,
-      "loss": 0.0031,
+      "loss": 0.0406,
       "step": 1745
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.5193007022214855e-06,
-      "loss": 0.0071,
+      "loss": 0.0328,
       "step": 1746
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.5097538146374512e-06,
-      "loss": 0.0059,
+      "loss": 0.03,
       "step": 1747
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.500222454007548e-06,
-      "loss": 0.0061,
+      "loss": 0.0305,
       "step": 1748
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.490706640089868e-06,
-      "loss": 0.0088,
+      "loss": 0.0254,
       "step": 1749
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.481206392610278e-06,
-      "loss": 0.003,
+      "loss": 0.0373,
       "step": 1750
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.4717217312623796e-06,
-      "loss": 0.0038,
+      "loss": 0.0354,
       "step": 1751
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.4622526757074628e-06,
-      "loss": 0.0035,
+      "loss": 0.0255,
       "step": 1752
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.452799245574463e-06,
-      "loss": 0.0017,
+      "loss": 0.0305,
       "step": 1753
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.4433614604599365e-06,
-      "loss": 0.0058,
+      "loss": 0.026,
       "step": 1754
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.433939339927992e-06,
-      "loss": 0.0096,
+      "loss": 0.0245,
       "step": 1755
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.424532903510284e-06,
-      "loss": 0.0024,
+      "loss": 0.0357,
       "step": 1756
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.415142170705941e-06,
-      "loss": 0.002,
+      "loss": 0.0214,
       "step": 1757
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.4057671609815403e-06,
-      "loss": 0.0076,
+      "loss": 0.0317,
       "step": 1758
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.3964078937710746e-06,
-      "loss": 0.0117,
+      "loss": 0.0346,
       "step": 1759
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.3870643884758915e-06,
-      "loss": 0.0027,
+      "loss": 0.0353,
       "step": 1760
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.377736664464676e-06,
-      "loss": 0.0076,
+      "loss": 0.0358,
       "step": 1761
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.368424741073392e-06,
-      "loss": 0.004,
+      "loss": 0.029,
       "step": 1762
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.359128637605247e-06,
-      "loss": 0.0104,
+      "loss": 0.0315,
       "step": 1763
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.3498483733306655e-06,
-      "loss": 0.0044,
+      "loss": 0.0263,
       "step": 1764
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.3405839674872244e-06,
-      "loss": 0.0054,
+      "loss": 0.0249,
       "step": 1765
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.3313354392796394e-06,
-      "loss": 0.0015,
+      "loss": 0.0338,
       "step": 1766
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.3221028078797046e-06,
-      "loss": 0.0033,
+      "loss": 0.0291,
       "step": 1767
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.31288609242626e-06,
-      "loss": 0.002,
+      "loss": 0.0198,
       "step": 1768
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.3036853120251566e-06,
-      "loss": 0.0044,
+      "loss": 0.0231,
       "step": 1769
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.294500485749218e-06,
-      "loss": 0.0074,
+      "loss": 0.0271,
       "step": 1770
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.285331632638179e-06,
-      "loss": 0.0039,
+      "loss": 0.038,
       "step": 1771
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.276178771698676e-06,
-      "loss": 0.0068,
+      "loss": 0.0256,
       "step": 1772
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.2670419219041962e-06,
-      "loss": 0.0058,
+      "loss": 0.0361,
       "step": 1773
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.2579211021950244e-06,
-      "loss": 0.0036,
+      "loss": 0.0188,
       "step": 1774
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.248816331478232e-06,
-      "loss": 0.0089,
+      "loss": 0.0333,
       "step": 1775
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.239727628627599e-06,
-      "loss": 0.0015,
+      "loss": 0.027,
       "step": 1776
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.230655012483619e-06,
-      "loss": 0.0019,
+      "loss": 0.0295,
       "step": 1777
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.22159850185343e-06,
-      "loss": 0.0016,
+      "loss": 0.0284,
       "step": 1778
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.2125581155107833e-06,
-      "loss": 0.0048,
+      "loss": 0.0274,
       "step": 1779
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.2035338721960032e-06,
-      "loss": 0.0052,
+      "loss": 0.032,
       "step": 1780
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.1945257906159578e-06,
-      "loss": 0.0012,
+      "loss": 0.0255,
       "step": 1781
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.1855338894440025e-06,
-      "loss": 0.0021,
+      "loss": 0.0294,
       "step": 1782
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.1765581873199626e-06,
-      "loss": 0.0104,
+      "loss": 0.0331,
       "step": 1783
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.1675987028500746e-06,
-      "loss": 0.004,
+      "loss": 0.0289,
       "step": 1784
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.1586554546069573e-06,
-      "loss": 0.0007,
+      "loss": 0.027,
       "step": 1785
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.14972846112958e-06,
-      "loss": 0.0105,
+      "loss": 0.029,
       "step": 1786
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.140817740923207e-06,
-      "loss": 0.0029,
+      "loss": 0.0247,
       "step": 1787
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.131923312459376e-06,
-      "loss": 0.002,
+      "loss": 0.0218,
       "step": 1788
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.12304519417585e-06,
-      "loss": 0.003,
+      "loss": 0.0312,
       "step": 1789
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.1141834044765775e-06,
-      "loss": 0.0023,
+      "loss": 0.0284,
       "step": 1790
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.1053379617316683e-06,
-      "loss": 0.0048,
+      "loss": 0.0306,
       "step": 1791
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.096508884277334e-06,
-      "loss": 0.0086,
+      "loss": 0.0303,
       "step": 1792
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.0876961904158744e-06,
-      "loss": 0.0023,
+      "loss": 0.0238,
       "step": 1793
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.078899898415616e-06,
-      "loss": 0.0022,
+      "loss": 0.0201,
       "step": 1794
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.0701200265108865e-06,
-      "loss": 0.0094,
+      "loss": 0.0276,
       "step": 1795
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.061356592901984e-06,
-      "loss": 0.0012,
+      "loss": 0.0229,
       "step": 1796
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.0526096157551213e-06,
-      "loss": 0.0078,
+      "loss": 0.0353,
       "step": 1797
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.0438791132023985e-06,
-      "loss": 0.0045,
+      "loss": 0.0304,
       "step": 1798
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.035165103341773e-06,
-      "loss": 0.0104,
+      "loss": 0.0339,
       "step": 1799
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.0264676042370026e-06,
-      "loss": 0.0024,
+      "loss": 0.0247,
       "step": 1800
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.017786633917629e-06,
-      "loss": 0.0049,
+      "loss": 0.0203,
       "step": 1801
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.009122210378922e-06,
-      "loss": 0.0085,
+      "loss": 0.0246,
       "step": 1802
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.0004743515818536e-06,
-      "loss": 0.0084,
+      "loss": 0.023,
       "step": 1803
     },
     {
       "epoch": 1.6,
       "learning_rate": 1.9918430754530625e-06,
-      "loss": 0.0041,
+      "loss": 0.033,
       "step": 1804
     },
     {
       "epoch": 1.6,
       "learning_rate": 1.9832283998848035e-06,
-      "loss": 0.0078,
+      "loss": 0.0385,
       "step": 1805
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.974630342734928e-06,
-      "loss": 0.0014,
+      "loss": 0.0235,
       "step": 1806
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.966048921826832e-06,
-      "loss": 0.0031,
+      "loss": 0.0357,
       "step": 1807
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.9574841549494263e-06,
-      "loss": 0.0014,
+      "loss": 0.0318,
       "step": 1808
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.9489360598571018e-06,
-      "loss": 0.0051,
+      "loss": 0.0266,
       "step": 1809
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.940404654269684e-06,
-      "loss": 0.0037,
+      "loss": 0.0269,
       "step": 1810
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.9318899558724102e-06,
-      "loss": 0.0098,
+      "loss": 0.0313,
       "step": 1811
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.9233919823158777e-06,
-      "loss": 0.0027,
+      "loss": 0.0346,
       "step": 1812
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.91491075121601e-06,
-      "loss": 0.0091,
+      "loss": 0.0305,
       "step": 1813
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.9064462801540362e-06,
-      "loss": 0.0021,
+      "loss": 0.0363,
       "step": 1814
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.8979985866764384e-06,
-      "loss": 0.0126,
+      "loss": 0.0226,
       "step": 1815
     },
     {
       "epoch": 1.61,
       "learning_rate": 1.8895676882949076e-06,
-      "loss": 0.0095,
+      "loss": 0.0292,
       "step": 1816
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8811536024863353e-06,
-      "loss": 0.0011,
+      "loss": 0.0336,
       "step": 1817
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8727563466927568e-06,
-      "loss": 0.0008,
+      "loss": 0.0309,
       "step": 1818
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8643759383213112e-06,
-      "loss": 0.0072,
+      "loss": 0.0368,
       "step": 1819
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.85601239474423e-06,
-      "loss": 0.002,
+      "loss": 0.0282,
       "step": 1820
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8476657332987636e-06,
-      "loss": 0.0031,
+      "loss": 0.0321,
       "step": 1821
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8393359712871817e-06,
-      "loss": 0.0057,
+      "loss": 0.0305,
       "step": 1822
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.831023125976722e-06,
-      "loss": 0.0019,
+      "loss": 0.0289,
       "step": 1823
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.822727214599549e-06,
-      "loss": 0.0051,
+      "loss": 0.0304,
       "step": 1824
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.814448254352721e-06,
-      "loss": 0.0049,
+      "loss": 0.0203,
       "step": 1825
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.8061862623981685e-06,
-      "loss": 0.0037,
+      "loss": 0.0294,
       "step": 1826
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.7979412558626353e-06,
-      "loss": 0.0046,
+      "loss": 0.0353,
       "step": 1827
     },
     {
       "epoch": 1.62,
       "learning_rate": 1.789713251837667e-06,
-      "loss": 0.0048,
+      "loss": 0.0296,
       "step": 1828
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7815022673795534e-06,
-      "loss": 0.0023,
+      "loss": 0.0314,
       "step": 1829
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.773308319509306e-06,
-      "loss": 0.0047,
+      "loss": 0.0283,
       "step": 1830
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7651314252126284e-06,
-      "loss": 0.0009,
+      "loss": 0.0359,
       "step": 1831
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7569716014398586e-06,
-      "loss": 0.0033,
+      "loss": 0.027,
       "step": 1832
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7488288651059626e-06,
-      "loss": 0.0074,
+      "loss": 0.0334,
       "step": 1833
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7407032330904749e-06,
-      "loss": 0.0058,
+      "loss": 0.0302,
       "step": 1834
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.732594722237475e-06,
-      "loss": 0.008,
+      "loss": 0.0304,
       "step": 1835
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7245033493555575e-06,
-      "loss": 0.0077,
+      "loss": 0.0367,
       "step": 1836
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7164291312177816e-06,
-      "loss": 0.0059,
+      "loss": 0.0239,
       "step": 1837
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7083720845616536e-06,
-      "loss": 0.002,
+      "loss": 0.0198,
       "step": 1838
     },
     {
       "epoch": 1.63,
       "learning_rate": 1.7003322260890776e-06,
-      "loss": 0.0043,
+      "loss": 0.0316,
       "step": 1839
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6923095724663296e-06,
-      "loss": 0.003,
+      "loss": 0.026,
       "step": 1840
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6843041403240245e-06,
-      "loss": 0.006,
+      "loss": 0.0218,
       "step": 1841
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.67631594625707e-06,
-      "loss": 0.0121,
+      "loss": 0.0168,
       "step": 1842
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6683450068246488e-06,
-      "loss": 0.0074,
+      "loss": 0.0208,
       "step": 1843
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6603913385501703e-06,
-      "loss": 0.0042,
+      "loss": 0.0227,
       "step": 1844
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.652454957921238e-06,
-      "loss": 0.0025,
+      "loss": 0.0385,
       "step": 1845
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6445358813896296e-06,
-      "loss": 0.004,
+      "loss": 0.0354,
       "step": 1846
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6366341253712437e-06,
-      "loss": 0.0056,
+      "loss": 0.0414,
       "step": 1847
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6287497062460722e-06,
-      "loss": 0.0118,
+      "loss": 0.0362,
       "step": 1848
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.620882640358179e-06,
-      "loss": 0.001,
+      "loss": 0.0224,
       "step": 1849
     },
     {
       "epoch": 1.64,
       "learning_rate": 1.6130329440156433e-06,
-      "loss": 0.0059,
+      "loss": 0.0504,
       "step": 1850
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.6052006334905468e-06,
-      "loss": 0.0023,
+      "loss": 0.0192,
       "step": 1851
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5973857250189328e-06,
-      "loss": 0.0076,
+      "loss": 0.0255,
       "step": 1852
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5895882348007562e-06,
-      "loss": 0.0054,
+      "loss": 0.0287,
       "step": 1853
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5818081789998807e-06,
-      "loss": 0.0108,
+      "loss": 0.0194,
       "step": 1854
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.574045573744023e-06,
-      "loss": 0.0068,
+      "loss": 0.0349,
       "step": 1855
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5663004351247257e-06,
-      "loss": 0.0016,
+      "loss": 0.0362,
       "step": 1856
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5585727791973238e-06,
-      "loss": 0.0023,
+      "loss": 0.0273,
       "step": 1857
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5508626219809064e-06,
-      "loss": 0.0055,
+      "loss": 0.0284,
       "step": 1858
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5431699794583e-06,
-      "loss": 0.0022,
+      "loss": 0.026,
       "step": 1859
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5354948675760161e-06,
-      "loss": 0.0062,
+      "loss": 0.0389,
       "step": 1860
     },
     {
       "epoch": 1.65,
       "learning_rate": 1.5278373022442271e-06,
-      "loss": 0.0027,
+      "loss": 0.0298,
       "step": 1861
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.5201972993367309e-06,
-      "loss": 0.0102,
+      "loss": 0.0326,
       "step": 1862
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.512574874690923e-06,
-      "loss": 0.0048,
+      "loss": 0.017,
       "step": 1863
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.5049700441077542e-06,
-      "loss": 0.0028,
+      "loss": 0.0276,
       "step": 1864
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.497382823351715e-06,
-      "loss": 0.004,
+      "loss": 0.0267,
       "step": 1865
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4898132281507727e-06,
-      "loss": 0.0016,
+      "loss": 0.0269,
       "step": 1866
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4822612741963737e-06,
-      "loss": 0.0153,
+      "loss": 0.0301,
       "step": 1867
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4747269771433935e-06,
-      "loss": 0.0028,
+      "loss": 0.0241,
       "step": 1868
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4672103526100945e-06,
-      "loss": 0.0115,
+      "loss": 0.0259,
       "step": 1869
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.459711416178119e-06,
-      "loss": 0.0037,
+      "loss": 0.0347,
       "step": 1870
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4522301833924314e-06,
-      "loss": 0.0012,
+      "loss": 0.0295,
       "step": 1871
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4447666697613005e-06,
-      "loss": 0.0068,
+      "loss": 0.0461,
       "step": 1872
     },
     {
       "epoch": 1.66,
       "learning_rate": 1.4373208907562687e-06,
-      "loss": 0.0074,
+      "loss": 0.0335,
       "step": 1873
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.4298928618121099e-06,
-      "loss": 0.0067,
+      "loss": 0.0305,
       "step": 1874
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.422482598326802e-06,
-      "loss": 0.0017,
+      "loss": 0.0339,
       "step": 1875
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.4150901156615027e-06,
-      "loss": 0.0033,
+      "loss": 0.0304,
       "step": 1876
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.4077154291405027e-06,
-      "loss": 0.0021,
+      "loss": 0.0414,
       "step": 1877
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.400358554051211e-06,
-      "loss": 0.0026,
+      "loss": 0.0344,
       "step": 1878
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.3930195056441054e-06,
-      "loss": 0.0016,
+      "loss": 0.027,
       "step": 1879
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.385698299132713e-06,
-      "loss": 0.0017,
+      "loss": 0.0284,
       "step": 1880
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.378394949693579e-06,
-      "loss": 0.0109,
+      "loss": 0.0185,
       "step": 1881
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.3711094724662255e-06,
-      "loss": 0.007,
+      "loss": 0.0271,
       "step": 1882
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.3638418825531342e-06,
-      "loss": 0.0058,
+      "loss": 0.0314,
       "step": 1883
     },
     {
       "epoch": 1.67,
       "learning_rate": 1.356592195019698e-06,
-      "loss": 0.0026,
+      "loss": 0.0265,
       "step": 1884
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3493604248942037e-06,
-      "loss": 0.0066,
+      "loss": 0.0332,
       "step": 1885
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3421465871677997e-06,
-      "loss": 0.0006,
+      "loss": 0.021,
       "step": 1886
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3349506967944515e-06,
-      "loss": 0.0077,
+      "loss": 0.0347,
       "step": 1887
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3277727686909325e-06,
-      "loss": 0.0036,
+      "loss": 0.034,
       "step": 1888
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3206128177367727e-06,
-      "loss": 0.0066,
+      "loss": 0.0317,
       "step": 1889
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3134708587742361e-06,
-      "loss": 0.0054,
+      "loss": 0.0336,
       "step": 1890
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.3063469066082956e-06,
-      "loss": 0.0052,
+      "loss": 0.0298,
       "step": 1891
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.299240976006594e-06,
-      "loss": 0.0075,
+      "loss": 0.0304,
       "step": 1892
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.292153081699412e-06,
-      "loss": 0.0016,
+      "loss": 0.0317,
       "step": 1893
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.2850832383796518e-06,
-      "loss": 0.0071,
+      "loss": 0.0279,
       "step": 1894
     },
     {
       "epoch": 1.68,
       "learning_rate": 1.2780314607027844e-06,
-      "loss": 0.0047,
+      "loss": 0.0303,
       "step": 1895
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.270997763286841e-06,
-      "loss": 0.0027,
+      "loss": 0.0279,
       "step": 1896
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2639821607123748e-06,
-      "loss": 0.0109,
+      "loss": 0.0276,
       "step": 1897
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.256984667522414e-06,
-      "loss": 0.0033,
+      "loss": 0.0268,
       "step": 1898
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2500052982224641e-06,
-      "loss": 0.0037,
+      "loss": 0.0244,
       "step": 1899
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2430440672804544e-06,
-      "loss": 0.0011,
+      "loss": 0.0285,
       "step": 1900
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2361009891267106e-06,
-      "loss": 0.0036,
+      "loss": 0.0275,
       "step": 1901
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.229176078153933e-06,
-      "loss": 0.006,
+      "loss": 0.0249,
       "step": 1902
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2222693487171567e-06,
-      "loss": 0.0037,
+      "loss": 0.018,
       "step": 1903
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2153808151337342e-06,
-      "loss": 0.004,
+      "loss": 0.0199,
       "step": 1904
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2085104916832979e-06,
-      "loss": 0.0089,
+      "loss": 0.0297,
       "step": 1905
     },
     {
       "epoch": 1.69,
       "learning_rate": 1.2016583926077264e-06,
-      "loss": 0.0063,
+      "loss": 0.0291,
       "step": 1906
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1948245321111208e-06,
-      "loss": 0.0017,
+      "loss": 0.0279,
       "step": 1907
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1880089243597792e-06,
-      "loss": 0.0025,
+      "loss": 0.0398,
       "step": 1908
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.181211583482158e-06,
-      "loss": 0.0094,
+      "loss": 0.0294,
       "step": 1909
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1744325235688536e-06,
-      "loss": 0.0101,
+      "loss": 0.0356,
       "step": 1910
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1676717586725527e-06,
-      "loss": 0.0108,
+      "loss": 0.0236,
       "step": 1911
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1609293028080325e-06,
-      "loss": 0.0027,
+      "loss": 0.032,
       "step": 1912
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1542051699521117e-06,
-      "loss": 0.0012,
+      "loss": 0.0291,
       "step": 1913
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1474993740436202e-06,
-      "loss": 0.0074,
+      "loss": 0.0545,
       "step": 1914
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1408119289833851e-06,
-      "loss": 0.0076,
+      "loss": 0.044,
       "step": 1915
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1341428486341876e-06,
-      "loss": 0.0017,
+      "loss": 0.025,
       "step": 1916
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.127492146820739e-06,
-      "loss": 0.0134,
+      "loss": 0.0294,
       "step": 1917
     },
     {
       "epoch": 1.7,
       "learning_rate": 1.1208598373296564e-06,
-      "loss": 0.0016,
+      "loss": 0.024,
       "step": 1918
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.114245933909428e-06,
-      "loss": 0.0018,
+      "loss": 0.0292,
       "step": 1919
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.1076504502703867e-06,
-      "loss": 0.0072,
+      "loss": 0.0172,
       "step": 1920
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.1010734000846847e-06,
-      "loss": 0.0038,
+      "loss": 0.0203,
       "step": 1921
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0945147969862567e-06,
-      "loss": 0.0051,
+      "loss": 0.0324,
       "step": 1922
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.087974654570807e-06,
-      "loss": 0.0026,
+      "loss": 0.0427,
       "step": 1923
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0814529863957625e-06,
-      "loss": 0.0087,
+      "loss": 0.0251,
       "step": 1924
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0749498059802554e-06,
-      "loss": 0.0086,
+      "loss": 0.0304,
       "step": 1925
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0684651268051005e-06,
-      "loss": 0.0055,
+      "loss": 0.0278,
       "step": 1926
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0619989623127512e-06,
-      "loss": 0.0065,
+      "loss": 0.0287,
       "step": 1927
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.0555513259072869e-06,
-      "loss": 0.0016,
+      "loss": 0.0204,
       "step": 1928
     },
     {
       "epoch": 1.71,
       "learning_rate": 1.049122230954378e-06,
-      "loss": 0.0047,
+      "loss": 0.03,
       "step": 1929
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.042711690781254e-06,
-      "loss": 0.0039,
+      "loss": 0.0268,
       "step": 1930
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.0363197186766905e-06,
-      "loss": 0.0026,
+      "loss": 0.0238,
       "step": 1931
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.0299463278909638e-06,
-      "loss": 0.0091,
+      "loss": 0.0178,
       "step": 1932
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.0235915316358381e-06,
-      "loss": 0.0086,
+      "loss": 0.0293,
       "step": 1933
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.017255343084529e-06,
-      "loss": 0.0043,
+      "loss": 0.0379,
       "step": 1934
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.0109377753716754e-06,
-      "loss": 0.0059,
+      "loss": 0.0272,
       "step": 1935
     },
     {
       "epoch": 1.72,
       "learning_rate": 1.0046388415933227e-06,
-      "loss": 0.0025,
+      "loss": 0.0352,
       "step": 1936
     },
     {
       "epoch": 1.72,
       "learning_rate": 9.983585548068908e-07,
-      "loss": 0.0042,
+      "loss": 0.0218,
       "step": 1937
     },
     {
       "epoch": 1.72,
       "learning_rate": 9.920969280311332e-07,
-      "loss": 0.004,
+      "loss": 0.0348,
       "step": 1938
     },
     {
       "epoch": 1.72,
       "learning_rate": 9.858539742461337e-07,
-      "loss": 0.0047,
+      "loss": 0.0351,
       "step": 1939
     },
     {
       "epoch": 1.72,
       "learning_rate": 9.796297063932614e-07,
-      "loss": 0.0073,
+      "loss": 0.0219,
       "step": 1940
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.734241373751541e-07,
-      "loss": 0.0059,
+      "loss": 0.021,
       "step": 1941
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.672372800556883e-07,
-      "loss": 0.0014,
+      "loss": 0.0367,
       "step": 1942
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.610691472599442e-07,
-      "loss": 0.0031,
+      "loss": 0.0269,
       "step": 1943
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.549197517741959e-07,
-      "loss": 0.0054,
+      "loss": 0.0357,
       "step": 1944
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.487891063458743e-07,
-      "loss": 0.0033,
+      "loss": 0.0279,
       "step": 1945
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.426772236835369e-07,
-      "loss": 0.0071,
+      "loss": 0.0321,
       "step": 1946
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.365841164568523e-07,
-      "loss": 0.0085,
+      "loss": 0.0354,
       "step": 1947
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.305097972965626e-07,
-      "loss": 0.0036,
+      "loss": 0.0302,
       "step": 1948
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.24454278794471e-07,
-      "loss": 0.0074,
+      "loss": 0.0262,
       "step": 1949
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.18417573503404e-07,
-      "loss": 0.0015,
+      "loss": 0.0313,
       "step": 1950
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.123996939371871e-07,
-      "loss": 0.0035,
+      "loss": 0.0301,
       "step": 1951
     },
     {
       "epoch": 1.74,
       "learning_rate": 9.064006525706215e-07,
-      "loss": 0.004,
+      "loss": 0.0304,
       "step": 1952
     },
     {
       "epoch": 1.74,
       "learning_rate": 9.004204618394608e-07,
-      "loss": 0.0062,
+      "loss": 0.0272,
       "step": 1953
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.944591341403775e-07,
-      "loss": 0.0006,
+      "loss": 0.026,
       "step": 1954
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.885166818309499e-07,
-      "loss": 0.0058,
+      "loss": 0.0353,
       "step": 1955
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.825931172296143e-07,
-      "loss": 0.0021,
+      "loss": 0.0283,
       "step": 1956
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.766884526156671e-07,
-      "loss": 0.0037,
+      "loss": 0.0261,
       "step": 1957
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.708027002292219e-07,
-      "loss": 0.0022,
+      "loss": 0.0394,
       "step": 1958
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.649358722711832e-07,
-      "loss": 0.0011,
+      "loss": 0.025,
       "step": 1959
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.590879809032348e-07,
-      "loss": 0.0045,
+      "loss": 0.0368,
       "step": 1960
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.532590382477979e-07,
-      "loss": 0.0013,
+      "loss": 0.0335,
       "step": 1961
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.474490563880155e-07,
-      "loss": 0.0051,
+      "loss": 0.0239,
       "step": 1962
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.416580473677305e-07,
-      "loss": 0.0051,
+      "loss": 0.0307,
       "step": 1963
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.358860231914501e-07,
-      "loss": 0.0051,
+      "loss": 0.0231,
       "step": 1964
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.301329958243276e-07,
-      "loss": 0.0081,
+      "loss": 0.0328,
       "step": 1965
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.243989771921413e-07,
-      "loss": 0.0097,
+      "loss": 0.031,
       "step": 1966
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.186839791812595e-07,
-      "loss": 0.002,
+      "loss": 0.0293,
       "step": 1967
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.129880136386259e-07,
-      "loss": 0.0107,
+      "loss": 0.0358,
       "step": 1968
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.073110923717298e-07,
-      "loss": 0.0059,
+      "loss": 0.0301,
       "step": 1969
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.016532271485788e-07,
-      "loss": 0.0034,
+      "loss": 0.0239,
       "step": 1970
     },
     {
       "epoch": 1.75,
       "learning_rate": 7.960144296976835e-07,
-      "loss": 0.0016,
+      "loss": 0.0253,
       "step": 1971
     },
     {
       "epoch": 1.75,
       "learning_rate": 7.903947117080235e-07,
-      "loss": 0.0046,
+      "loss": 0.0254,
       "step": 1972
     },
     {
       "epoch": 1.75,
       "learning_rate": 7.847940848290326e-07,
-      "loss": 0.003,
+      "loss": 0.0203,
       "step": 1973
     },
     {
       "epoch": 1.75,
       "learning_rate": 7.792125606705636e-07,
-      "loss": 0.0021,
+      "loss": 0.0297,
       "step": 1974
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.736501508028727e-07,
-      "loss": 0.006,
+      "loss": 0.02,
       "step": 1975
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.681068667565961e-07,
-      "loss": 0.0048,
+      "loss": 0.0284,
       "step": 1976
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.625827200227165e-07,
-      "loss": 0.0055,
+      "loss": 0.0298,
       "step": 1977
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.570777220525538e-07,
-      "loss": 0.002,
+      "loss": 0.0242,
       "step": 1978
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.515918842577263e-07,
-      "loss": 0.0028,
+      "loss": 0.0339,
       "step": 1979
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.461252180101352e-07,
-      "loss": 0.008,
+      "loss": 0.0187,
       "step": 1980
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.406777346419436e-07,
-      "loss": 0.0046,
+      "loss": 0.0213,
       "step": 1981
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.352494454455517e-07,
-      "loss": 0.0027,
+      "loss": 0.03,
       "step": 1982
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.298403616735583e-07,
-      "loss": 0.005,
+      "loss": 0.048,
       "step": 1983
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.244504945387631e-07,
-      "loss": 0.0017,
+      "loss": 0.0345,
       "step": 1984
     },
     {
       "epoch": 1.76,
       "learning_rate": 7.190798552141254e-07,
-      "loss": 0.0088,
+      "loss": 0.029,
       "step": 1985
     },
     {
       "epoch": 1.77,
       "learning_rate": 7.13728454832745e-07,
-      "loss": 0.0015,
+      "loss": 0.0299,
       "step": 1986
     },
     {
       "epoch": 1.77,
       "learning_rate": 7.083963044878473e-07,
-      "loss": 0.0025,
+      "loss": 0.0386,
       "step": 1987
     },
     {
       "epoch": 1.77,
       "learning_rate": 7.030834152327393e-07,
-      "loss": 0.0103,
+      "loss": 0.0405,
       "step": 1988
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.977897980808146e-07,
-      "loss": 0.0011,
+      "loss": 0.0321,
       "step": 1989
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.925154640055109e-07,
-      "loss": 0.0026,
+      "loss": 0.0265,
       "step": 1990
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.872604239402924e-07,
-      "loss": 0.0046,
+      "loss": 0.0279,
       "step": 1991
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.820246887786275e-07,
-      "loss": 0.0069,
+      "loss": 0.023,
       "step": 1992
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.768082693739664e-07,
-      "loss": 0.0018,
+      "loss": 0.0246,
       "step": 1993
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.716111765397193e-07,
-      "loss": 0.0019,
+      "loss": 0.0279,
       "step": 1994
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.664334210492374e-07,
-      "loss": 0.009,
+      "loss": 0.0279,
       "step": 1995
     },
     {
       "epoch": 1.77,
       "learning_rate": 6.612750136357782e-07,
-      "loss": 0.0119,
+      "loss": 0.0195,
       "step": 1996
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.561359649924959e-07,
-      "loss": 0.0007,
+      "loss": 0.0279,
       "step": 1997
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.510162857724167e-07,
-      "loss": 0.0092,
+      "loss": 0.0392,
       "step": 1998
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.459159865884102e-07,
-      "loss": 0.0004,
+      "loss": 0.0246,
       "step": 1999
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.408350780131778e-07,
-      "loss": 0.0005,
+      "loss": 0.0299,
       "step": 2000
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.35773570579219e-07,
-      "loss": 0.0077,
+      "loss": 0.0259,
       "step": 2001
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.307314747788163e-07,
-      "loss": 0.0045,
+      "loss": 0.0252,
       "step": 2002
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.2570880106402e-07,
-      "loss": 0.0016,
+      "loss": 0.0274,
       "step": 2003
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.207055598466094e-07,
-      "loss": 0.0054,
+      "loss": 0.0404,
       "step": 2004
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.157217614980881e-07,
-      "loss": 0.0023,
+      "loss": 0.0219,
       "step": 2005
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.107574163496532e-07,
-      "loss": 0.0067,
+      "loss": 0.0254,
       "step": 2006
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.05812534692174e-07,
-      "loss": 0.0047,
+      "loss": 0.0284,
       "step": 2007
     },
     {
       "epoch": 1.78,
       "learning_rate": 6.008871267761762e-07,
-      "loss": 0.0022,
+      "loss": 0.0248,
       "step": 2008
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.959812028118172e-07,
-      "loss": 0.0046,
+      "loss": 0.0402,
       "step": 2009
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.910947729688598e-07,
-      "loss": 0.0038,
+      "loss": 0.0242,
       "step": 2010
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.862278473766636e-07,
-      "loss": 0.003,
+      "loss": 0.0243,
       "step": 2011
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.813804361241516e-07,
-      "loss": 0.0071,
+      "loss": 0.0367,
       "step": 2012
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.76552549259799e-07,
-      "loss": 0.0011,
+      "loss": 0.0167,
       "step": 2013
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.717441967916027e-07,
-      "loss": 0.0044,
+      "loss": 0.0358,
       "step": 2014
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.66955388687066e-07,
-      "loss": 0.0056,
+      "loss": 0.0294,
       "step": 2015
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.621861348731838e-07,
-      "loss": 0.0066,
+      "loss": 0.0205,
       "step": 2016
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.574364452364079e-07,
-      "loss": 0.0015,
+      "loss": 0.0268,
       "step": 2017
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.527063296226421e-07,
-      "loss": 0.0009,
+      "loss": 0.0219,
       "step": 2018
     },
     {
       "epoch": 1.79,
       "learning_rate": 5.479957978372064e-07,
-      "loss": 0.008,
+      "loss": 0.023,
       "step": 2019
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.433048596448287e-07,
-      "loss": 0.0047,
+      "loss": 0.0333,
       "step": 2020
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.386335247696217e-07,
-      "loss": 0.0022,
+      "loss": 0.0226,
       "step": 2021
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.339818028950561e-07,
-      "loss": 0.0023,
+      "loss": 0.0305,
       "step": 2022
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.293497036639517e-07,
-      "loss": 0.0047,
+      "loss": 0.0353,
       "step": 2023
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.247372366784476e-07,
-      "loss": 0.0151,
+      "loss": 0.0382,
       "step": 2024
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.201444114999843e-07,
-      "loss": 0.0055,
+      "loss": 0.0304,
       "step": 2025
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.155712376492905e-07,
-      "loss": 0.0025,
+      "loss": 0.0334,
       "step": 2026
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.110177246063596e-07,
-      "loss": 0.0014,
+      "loss": 0.0204,
       "step": 2027
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.064838818104201e-07,
-      "loss": 0.0028,
+      "loss": 0.0297,
       "step": 2028
     },
     {
       "epoch": 1.8,
       "learning_rate": 5.019697186599337e-07,
-      "loss": 0.0006,
+      "loss": 0.0202,
       "step": 2029
     },
     {
       "epoch": 1.8,
       "learning_rate": 4.97475244512563e-07,
-      "loss": 0.0036,
+      "loss": 0.0272,
       "step": 2030
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.930004686851575e-07,
-      "loss": 0.0076,
+      "loss": 0.0239,
       "step": 2031
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.88545400453736e-07,
-      "loss": 0.0063,
+      "loss": 0.0419,
       "step": 2032
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.841100490534545e-07,
-      "loss": 0.0032,
+      "loss": 0.0379,
       "step": 2033
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.796944236786061e-07,
-      "loss": 0.0016,
+      "loss": 0.0249,
       "step": 2034
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.7529853348259215e-07,
-      "loss": 0.0116,
+      "loss": 0.0276,
       "step": 2035
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.709223875779001e-07,
-      "loss": 0.0031,
+      "loss": 0.0189,
       "step": 2036
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.66565995036089e-07,
-      "loss": 0.0009,
+      "loss": 0.0258,
       "step": 2037
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.622293648877696e-07,
-      "loss": 0.0069,
+      "loss": 0.0274,
       "step": 2038
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.5791250612258974e-07,
-      "loss": 0.0027,
+      "loss": 0.0282,
       "step": 2039
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.536154276892102e-07,
-      "loss": 0.005,
+      "loss": 0.0289,
       "step": 2040
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.493381384952855e-07,
-      "loss": 0.0051,
+      "loss": 0.0378,
       "step": 2041
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.450806474074498e-07,
-      "loss": 0.0035,
+      "loss": 0.0291,
       "step": 2042
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.408429632512989e-07,
-      "loss": 0.0019,
+      "loss": 0.0285,
       "step": 2043
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.366250948113648e-07,
-      "loss": 0.0067,
+      "loss": 0.028,
       "step": 2044
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.3242705083110683e-07,
-      "loss": 0.0059,
+      "loss": 0.0272,
       "step": 2045
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.2824884001288704e-07,
-      "loss": 0.0026,
+      "loss": 0.0199,
       "step": 2046
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.2409047101795386e-07,
-      "loss": 0.0009,
+      "loss": 0.0287,
       "step": 2047
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.199519524664253e-07,
-      "loss": 0.0027,
+      "loss": 0.027,
       "step": 2048
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.158332929372699e-07,
-      "loss": 0.0025,
+      "loss": 0.0175,
       "step": 2049
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.117345009682916e-07,
-      "loss": 0.0051,
+      "loss": 0.0371,
       "step": 2050
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.076555850561048e-07,
-      "loss": 0.0108,
+      "loss": 0.0232,
       "step": 2051
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.0359655365612483e-07,
-      "loss": 0.0067,
+      "loss": 0.0269,
       "step": 2052
     },
     {
       "epoch": 1.82,
       "learning_rate": 3.9955741518254985e-07,
-      "loss": 0.0075,
+      "loss": 0.0245,
       "step": 2053
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.955381780083356e-07,
-      "loss": 0.0048,
+      "loss": 0.0263,
       "step": 2054
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.91538850465184e-07,
-      "loss": 0.0066,
+      "loss": 0.0279,
       "step": 2055
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.8755944084353125e-07,
-      "loss": 0.0044,
+      "loss": 0.0285,
       "step": 2056
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.835999573925153e-07,
-      "loss": 0.0052,
+      "loss": 0.0201,
       "step": 2057
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.796604083199784e-07,
-      "loss": 0.0059,
+      "loss": 0.0301,
       "step": 2058
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.7574080179243023e-07,
-      "loss": 0.0134,
+      "loss": 0.0251,
       "step": 2059
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.718411459350446e-07,
-      "loss": 0.0018,
+      "loss": 0.0244,
       "step": 2060
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.6796144883163964e-07,
-      "loss": 0.0074,
+      "loss": 0.034,
       "step": 2061
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.641017185246576e-07,
-      "loss": 0.0024,
+      "loss": 0.0247,
       "step": 2062
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.6026196301515273e-07,
-      "loss": 0.0038,
+      "loss": 0.0246,
       "step": 2063
     },
     {
       "epoch": 1.83,
       "learning_rate": 3.5644219026276903e-07,
-      "loss": 0.0074,
+      "loss": 0.0241,
       "step": 2064
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.5264240818573046e-07,
-      "loss": 0.002,
+      "loss": 0.0252,
       "step": 2065
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.488626246608184e-07,
-      "loss": 0.0031,
+      "loss": 0.0292,
       "step": 2066
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.451028475233609e-07,
-      "loss": 0.0024,
+      "loss": 0.0305,
       "step": 2067
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.413630845672122e-07,
-      "loss": 0.003,
+      "loss": 0.0243,
       "step": 2068
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.3764334354473795e-07,
-      "loss": 0.0035,
+      "loss": 0.0133,
       "step": 2069
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.339436321667966e-07,
-      "loss": 0.002,
+      "loss": 0.0281,
       "step": 2070
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.302639581027312e-07,
-      "loss": 0.0072,
+      "loss": 0.0268,
       "step": 2071
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.266043289803489e-07,
-      "loss": 0.0045,
+      "loss": 0.0251,
       "step": 2072
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.2296475238589477e-07,
-      "loss": 0.0018,
+      "loss": 0.0205,
       "step": 2073
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.193452358640581e-07,
-      "loss": 0.0045,
+      "loss": 0.0256,
       "step": 2074
     },
     {
       "epoch": 1.84,
       "learning_rate": 3.1574578691793476e-07,
-      "loss": 0.0062,
+      "loss": 0.031,
       "step": 2075
     },
     {
       "epoch": 1.85,
       "learning_rate": 3.121664130090285e-07,
-      "loss": 0.0036,
+      "loss": 0.023,
       "step": 2076
     },
     {
       "epoch": 1.85,
       "learning_rate": 3.086071215572262e-07,
-      "loss": 0.0029,
+      "loss": 0.0243,
       "step": 2077
     },
     {
       "epoch": 1.85,
       "learning_rate": 3.050679199407813e-07,
-      "loss": 0.008,
+      "loss": 0.0382,
       "step": 2078
     },
     {
       "epoch": 1.85,
       "learning_rate": 3.0154881549630534e-07,
-      "loss": 0.0062,
+      "loss": 0.021,
       "step": 2079
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.980498155187506e-07,
-      "loss": 0.0041,
+      "loss": 0.0289,
       "step": 2080
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.945709272613906e-07,
-      "loss": 0.0021,
+      "loss": 0.0222,
       "step": 2081
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.9111215793580894e-07,
-      "loss": 0.0059,
+      "loss": 0.0199,
       "step": 2082
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.8767351471188145e-07,
-      "loss": 0.0016,
+      "loss": 0.0277,
       "step": 2083
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.842550047177683e-07,
-      "loss": 0.0067,
+      "loss": 0.0269,
       "step": 2084
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.8085663503989336e-07,
-      "loss": 0.0037,
+      "loss": 0.0315,
       "step": 2085
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.774784127229269e-07,
-      "loss": 0.0084,
+      "loss": 0.0339,
       "step": 2086
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.741203447697749e-07,
-      "loss": 0.0017,
+      "loss": 0.0311,
       "step": 2087
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.707824381415702e-07,
-      "loss": 0.0013,
+      "loss": 0.0323,
       "step": 2088
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.6746469975764354e-07,
-      "loss": 0.0068,
+      "loss": 0.0311,
       "step": 2089
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.641671364955256e-07,
-      "loss": 0.0053,
+      "loss": 0.0347,
       "step": 2090
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.6088975519092066e-07,
-      "loss": 0.001,
+      "loss": 0.0209,
       "step": 2091
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.576325626376952e-07,
-      "loss": 0.0082,
+      "loss": 0.0246,
       "step": 2092
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.5439556558786935e-07,
-      "loss": 0.0088,
+      "loss": 0.0276,
       "step": 2093
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.511787707515967e-07,
-      "loss": 0.0019,
+      "loss": 0.0311,
       "step": 2094
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.479821847971542e-07,
-      "loss": 0.0034,
+      "loss": 0.0255,
       "step": 2095
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.4480581435092357e-07,
-      "loss": 0.0027,
+      "loss": 0.0197,
       "step": 2096
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.4164966599738103e-07,
-      "loss": 0.0055,
+      "loss": 0.0362,
       "step": 2097
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.3851374627908875e-07,
-      "loss": 0.0035,
+      "loss": 0.0205,
       "step": 2098
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.3539806169666891e-07,
-      "loss": 0.0071,
+      "loss": 0.0218,
       "step": 2099
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.3230261870879956e-07,
-      "loss": 0.0038,
+      "loss": 0.0307,
       "step": 2100
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.2922742373220118e-07,
-      "loss": 0.0104,
+      "loss": 0.0341,
       "step": 2101
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.2617248314161676e-07,
-      "loss": 0.0031,
+      "loss": 0.024,
       "step": 2102
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.2313780326980728e-07,
-      "loss": 0.0009,
+      "loss": 0.0338,
       "step": 2103
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.2012339040753172e-07,
-      "loss": 0.0068,
+      "loss": 0.0313,
       "step": 2104
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.1712925080353385e-07,
-      "loss": 0.0065,
+      "loss": 0.032,
       "step": 2105
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.1415539066453884e-07,
-      "loss": 0.0034,
+      "loss": 0.023,
       "step": 2106
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.112018161552254e-07,
-      "loss": 0.0063,
+      "loss": 0.0257,
       "step": 2107
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.0826853339822817e-07,
-      "loss": 0.0015,
+      "loss": 0.0294,
       "step": 2108
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.053555484741121e-07,
-      "loss": 0.0109,
+      "loss": 0.029,
       "step": 2109
     },
     {
       "epoch": 1.88,
       "learning_rate": 2.024628674213691e-07,
-      "loss": 0.0051,
+      "loss": 0.021,
       "step": 2110
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.9959049623640147e-07,
-      "loss": 0.0008,
+      "loss": 0.0232,
       "step": 2111
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.9673844087350736e-07,
-      "loss": 0.004,
+      "loss": 0.0226,
       "step": 2112
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.9390670724487636e-07,
-      "loss": 0.0059,
+      "loss": 0.0395,
       "step": 2113
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.9109530122056742e-07,
-      "loss": 0.0022,
+      "loss": 0.0402,
       "step": 2114
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.8830422862850305e-07,
-      "loss": 0.0007,
+      "loss": 0.0193,
       "step": 2115
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.8553349525445406e-07,
-      "loss": 0.0027,
+      "loss": 0.0267,
       "step": 2116
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.827831068420316e-07,
-      "loss": 0.0079,
+      "loss": 0.0269,
       "step": 2117
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.8005306909267161e-07,
-      "loss": 0.0027,
+      "loss": 0.0276,
       "step": 2118
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.7734338766562165e-07,
-      "loss": 0.0007,
+      "loss": 0.0328,
       "step": 2119
     },
     {
       "epoch": 1.88,
       "learning_rate": 1.7465406817793296e-07,
-      "loss": 0.0025,
+      "loss": 0.0205,
       "step": 2120
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.7198511620444947e-07,
-      "loss": 0.0062,
+      "loss": 0.0262,
       "step": 2121
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.693365372777922e-07,
-      "loss": 0.0023,
+      "loss": 0.0232,
       "step": 2122
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.667083368883482e-07,
-      "loss": 0.0036,
+      "loss": 0.0344,
       "step": 2123
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.6410052048426382e-07,
-      "loss": 0.009,
+      "loss": 0.0276,
       "step": 2124
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.615130934714304e-07,
-      "loss": 0.0023,
+      "loss": 0.0218,
       "step": 2125
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.5894606121346745e-07,
-      "loss": 0.0049,
+      "loss": 0.0263,
       "step": 2126
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.56399429031725e-07,
-      "loss": 0.0046,
+      "loss": 0.0205,
       "step": 2127
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.5387320220525804e-07,
-      "loss": 0.0009,
+      "loss": 0.0258,
       "step": 2128
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.5136738597082644e-07,
-      "loss": 0.0009,
+      "loss": 0.0226,
       "step": 2129
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.4888198552287624e-07,
-      "loss": 0.0041,
+      "loss": 0.0277,
       "step": 2130
     },
     {
       "epoch": 1.89,
       "learning_rate": 1.4641700601353726e-07,
-      "loss": 0.002,
+      "loss": 0.0276,
       "step": 2131
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.4397245255260095e-07,
-      "loss": 0.0035,
+      "loss": 0.0261,
       "step": 2132
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.4154833020752267e-07,
-      "loss": 0.0038,
+      "loss": 0.0206,
       "step": 2133
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.3914464400340166e-07,
-      "loss": 0.0055,
+      "loss": 0.0395,
       "step": 2134
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.3676139892297546e-07,
-      "loss": 0.0015,
+      "loss": 0.0297,
       "step": 2135
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.3439859990660775e-07,
-      "loss": 0.0125,
+      "loss": 0.0239,
       "step": 2136
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.32056251852275e-07,
-      "loss": 0.0073,
+      "loss": 0.0307,
       "step": 2137
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.2973435961556647e-07,
-      "loss": 0.0026,
+      "loss": 0.0189,
       "step": 2138
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.274329280096609e-07,
-      "loss": 0.0076,
+      "loss": 0.0344,
       "step": 2139
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.2515196180532653e-07,
-      "loss": 0.0017,
+      "loss": 0.0234,
       "step": 2140
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.228914657309088e-07,
-      "loss": 0.0045,
+      "loss": 0.0235,
       "step": 2141
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.2065144447231392e-07,
-      "loss": 0.0008,
+      "loss": 0.0359,
       "step": 2142
     },
     {
       "epoch": 1.9,
       "learning_rate": 1.1843190267301297e-07,
-      "loss": 0.0011,
+      "loss": 0.0177,
       "step": 2143
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.1623284493401444e-07,
-      "loss": 0.0025,
+      "loss": 0.0214,
       "step": 2144
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.1405427581387407e-07,
-      "loss": 0.0137,
+      "loss": 0.0239,
       "step": 2145
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.1189619982866829e-07,
-      "loss": 0.0016,
+      "loss": 0.022,
       "step": 2146
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.09758621451993e-07,
-      "loss": 0.0071,
+      "loss": 0.0338,
       "step": 2147
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.0764154511495928e-07,
-      "loss": 0.0053,
+      "loss": 0.0358,
       "step": 2148
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.0554497520616991e-07,
-      "loss": 0.002,
+      "loss": 0.023,
       "step": 2149
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.0346891607172615e-07,
-      "loss": 0.0007,
+      "loss": 0.0276,
       "step": 2150
     },
     {
       "epoch": 1.91,
       "learning_rate": 1.0141337201520662e-07,
-      "loss": 0.0031,
+      "loss": 0.027,
       "step": 2151
     },
     {
       "epoch": 1.91,
       "learning_rate": 9.937834729766394e-08,
-      "loss": 0.0176,
+      "loss": 0.0241,
       "step": 2152
     },
     {
       "epoch": 1.91,
       "learning_rate": 9.736384613761807e-08,
-      "loss": 0.0015,
+      "loss": 0.0253,
       "step": 2153
     },
     {
       "epoch": 1.91,
       "learning_rate": 9.536987271104081e-08,
-      "loss": 0.0062,
+      "loss": 0.0279,
       "step": 2154
     },
     {
       "epoch": 1.92,
       "learning_rate": 9.339643115135133e-08,
-      "loss": 0.0011,
+      "loss": 0.0159,
       "step": 2155
     },
     {
       "epoch": 1.92,
       "learning_rate": 9.144352554940949e-08,
-      "loss": 0.0016,
+      "loss": 0.0259,
       "step": 2156
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.951115995350367e-08,
-      "loss": 0.0112,
+      "loss": 0.0326,
       "step": 2157
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.759933836934187e-08,
-      "loss": 0.0067,
+      "loss": 0.0271,
       "step": 2158
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.570806476004945e-08,
-      "loss": 0.0046,
+      "loss": 0.0241,
       "step": 2159
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.383734304615144e-08,
-      "loss": 0.0021,
+      "loss": 0.034,
       "step": 2160
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.198717710557358e-08,
-      "loss": 0.0018,
+      "loss": 0.026,
       "step": 2161
     },
     {
       "epoch": 1.92,
       "learning_rate": 8.015757077363129e-08,
-      "loss": 0.0008,
+      "loss": 0.0229,
       "step": 2162
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.834852784301628e-08,
-      "loss": 0.0046,
+      "loss": 0.0272,
       "step": 2163
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.656005206379657e-08,
-      "loss": 0.0016,
+      "loss": 0.0261,
       "step": 2164
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.479214714340432e-08,
-      "loss": 0.0066,
+      "loss": 0.0248,
       "step": 2165
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.304481674662911e-08,
-      "loss": 0.0096,
+      "loss": 0.0251,
       "step": 2166
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.131806449561241e-08,
-      "loss": 0.0047,
+      "loss": 0.0267,
       "step": 2167
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.961189396983537e-08,
-      "loss": 0.0025,
+      "loss": 0.0316,
       "step": 2168
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.792630870611438e-08,
-      "loss": 0.0122,
+      "loss": 0.0257,
       "step": 2169
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.626131219859555e-08,
-      "loss": 0.0089,
+      "loss": 0.0241,
       "step": 2170
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.461690789874354e-08,
-      "loss": 0.0074,
+      "loss": 0.0259,
       "step": 2171
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.299309921533714e-08,
-      "loss": 0.003,
+      "loss": 0.0345,
       "step": 2172
     },
     {
       "epoch": 1.93,
       "learning_rate": 6.138988951446046e-08,
-      "loss": 0.0042,
+      "loss": 0.0226,
       "step": 2173
     },
     {
       "epoch": 1.93,
       "learning_rate": 5.980728211949838e-08,
-      "loss": 0.0085,
+      "loss": 0.0233,
       "step": 2174
     },
     {
       "epoch": 1.93,
       "learning_rate": 5.824528031112553e-08,
-      "loss": 0.0055,
+      "loss": 0.0375,
       "step": 2175
     },
     {
       "epoch": 1.93,
       "learning_rate": 5.670388732730625e-08,
-      "loss": 0.0058,
+      "loss": 0.0217,
       "step": 2176
     },
     {
       "epoch": 1.94,
       "learning_rate": 5.518310636327795e-08,
-      "loss": 0.0014,
+      "loss": 0.0236,
       "step": 2177
     },
     {
       "epoch": 1.94,
       "learning_rate": 5.3682940571555544e-08,
-      "loss": 0.0081,
+      "loss": 0.0302,
       "step": 2178
     },
     {
       "epoch": 1.94,
       "learning_rate": 5.220339306191591e-08,
-      "loss": 0.0011,
+      "loss": 0.0219,
       "step": 2179
     },
     {
       "epoch": 1.94,
       "learning_rate": 5.0744466901397895e-08,
-      "loss": 0.0009,
+      "loss": 0.0222,
       "step": 2180
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.93061651142912e-08,
-      "loss": 0.0024,
+      "loss": 0.0174,
       "step": 2181
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.788849068213197e-08,
-      "loss": 0.001,
+      "loss": 0.0264,
       "step": 2182
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.649144654370052e-08,
-      "loss": 0.0012,
+      "loss": 0.0283,
       "step": 2183
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.5115035595004744e-08,
-      "loss": 0.0094,
+      "loss": 0.0374,
       "step": 2184
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.375926068928671e-08,
-      "loss": 0.0014,
+      "loss": 0.0279,
       "step": 2185
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.2424124637009404e-08,
-      "loss": 0.005,
+      "loss": 0.0281,
       "step": 2186
     },
     {
       "epoch": 1.94,
       "learning_rate": 4.110963020584891e-08,
-      "loss": 0.0058,
+      "loss": 0.0227,
       "step": 2187
     },
     {
       "epoch": 1.94,
       "learning_rate": 3.9815780120697755e-08,
-      "loss": 0.0028,
+      "loss": 0.0208,
       "step": 2188
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.854257706364939e-08,
-      "loss": 0.003,
+      "loss": 0.0264,
       "step": 2189
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.729002367399925e-08,
-      "loss": 0.005,
+      "loss": 0.0344,
       "step": 2190
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.605812254823593e-08,
-      "loss": 0.0021,
+      "loss": 0.0335,
       "step": 2191
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.484687624003447e-08,
-      "loss": 0.0013,
+      "loss": 0.0233,
       "step": 2192
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.3656287260258605e-08,
-      "loss": 0.0022,
+      "loss": 0.025,
       "step": 2193
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.248635807694744e-08,
-      "loss": 0.0088,
+      "loss": 0.0303,
       "step": 2194
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.1337091115309914e-08,
-      "loss": 0.0015,
+      "loss": 0.0257,
       "step": 2195
     },
     {
       "epoch": 1.95,
       "learning_rate": 3.02084887577303e-08,
-      "loss": 0.0168,
+      "loss": 0.0274,
       "step": 2196
     },
     {
       "epoch": 1.95,
       "learning_rate": 2.9100553343749393e-08,
-      "loss": 0.0032,
+      "loss": 0.0275,
       "step": 2197
     },
     {
       "epoch": 1.95,
       "learning_rate": 2.8013287170070014e-08,
-      "loss": 0.0006,
+      "loss": 0.0191,
       "step": 2198
     },
     {
       "epoch": 1.95,
       "learning_rate": 2.6946692490549266e-08,
-      "loss": 0.0037,
+      "loss": 0.0324,
       "step": 2199
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.5900771516188527e-08,
-      "loss": 0.0024,
+      "loss": 0.032,
       "step": 2200
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.4875526415137907e-08,
-      "loss": 0.0059,
+      "loss": 0.0146,
       "step": 2201
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.3870959312685127e-08,
-      "loss": 0.0014,
+      "loss": 0.0277,
       "step": 2202
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.288707229125553e-08,
-      "loss": 0.0016,
+      "loss": 0.0154,
       "step": 2203
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.19238673904032e-08,
-      "loss": 0.0021,
+      "loss": 0.0331,
       "step": 2204
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.098134660680984e-08,
-      "loss": 0.0044,
+      "loss": 0.0255,
       "step": 2205
     },
     {
       "epoch": 1.96,
       "learning_rate": 2.005951189427924e-08,
-      "loss": 0.0016,
+      "loss": 0.031,
       "step": 2206
     },
     {
       "epoch": 1.96,
       "learning_rate": 1.9158365163736146e-08,
-      "loss": 0.0015,
+      "loss": 0.0304,
       "step": 2207
     },
     {
       "epoch": 1.96,
       "learning_rate": 1.8277908283217404e-08,
-      "loss": 0.004,
+      "loss": 0.0171,
       "step": 2208
     },
     {
       "epoch": 1.96,
       "learning_rate": 1.741814307787304e-08,
-      "loss": 0.0072,
+      "loss": 0.0226,
       "step": 2209
     },
     {
       "epoch": 1.96,
       "learning_rate": 1.65790713299574e-08,
-      "loss": 0.0013,
+      "loss": 0.0341,
       "step": 2210
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.5760694778830244e-08,
-      "loss": 0.0085,
+      "loss": 0.0242,
       "step": 2211
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.496301512095011e-08,
-      "loss": 0.0106,
+      "loss": 0.0323,
       "step": 2212
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.4186034009873173e-08,
-      "loss": 0.0012,
+      "loss": 0.0191,
       "step": 2213
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.3429753056246608e-08,
-      "loss": 0.0041,
+      "loss": 0.0447,
       "step": 2214
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.2694173827807465e-08,
-      "loss": 0.0111,
+      "loss": 0.0247,
       "step": 2215
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.1979297849381566e-08,
-      "loss": 0.0057,
+      "loss": 0.0236,
       "step": 2216
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.1285126602873508e-08,
-      "loss": 0.001,
+      "loss": 0.0307,
       "step": 2217
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.0611661527271111e-08,
-      "loss": 0.0065,
+      "loss": 0.0246,
       "step": 2218
     },
     {
       "epoch": 1.97,
       "learning_rate": 9.958904018637639e-09,
-      "loss": 0.0035,
+      "loss": 0.0431,
       "step": 2219
     },
     {
       "epoch": 1.97,
       "learning_rate": 9.326855430110693e-09,
-      "loss": 0.0025,
+      "loss": 0.0356,
       "step": 2220
     },
     {
       "epoch": 1.97,
       "learning_rate": 8.715517071899993e-09,
-      "loss": 0.0019,
+      "loss": 0.02,
       "step": 2221
     },
     {
       "epoch": 1.98,
       "learning_rate": 8.124890211284042e-09,
-      "loss": 0.0024,
+      "loss": 0.0225,
       "step": 2222
     },
     {
       "epoch": 1.98,
       "learning_rate": 7.554976072605692e-09,
-      "loss": 0.0015,
+      "loss": 0.0198,
       "step": 2223
     },
     {
       "epoch": 1.98,
       "learning_rate": 7.005775837272133e-09,
-      "loss": 0.0057,
+      "loss": 0.0384,
       "step": 2224
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.477290643751577e-09,
-      "loss": 0.0081,
+      "loss": 0.0263,
       "step": 2225
     },
     {
       "epoch": 1.98,
       "learning_rate": 5.969521587571025e-09,
-      "loss": 0.0016,
+      "loss": 0.0242,
       "step": 2226
     },
     {
       "epoch": 1.98,
       "learning_rate": 5.4824697213129445e-09,
-      "loss": 0.0016,
+      "loss": 0.0381,
       "step": 2227
     },
     {
       "epoch": 1.98,
       "learning_rate": 5.016136054616372e-09,
-      "loss": 0.0007,
+      "loss": 0.0253,
       "step": 2228
     },
     {
       "epoch": 1.98,
       "learning_rate": 4.5705215541680396e-09,
-      "loss": 0.0089,
+      "loss": 0.0436,
       "step": 2229
     },
     {
       "epoch": 1.98,
       "learning_rate": 4.14562714371014e-09,
-      "loss": 0.0055,
+      "loss": 0.0302,
       "step": 2230
     },
     {
       "epoch": 1.98,
       "learning_rate": 3.74145370402812e-09,
-      "loss": 0.0056,
+      "loss": 0.0414,
       "step": 2231
     },
     {
       "epoch": 1.98,
       "learning_rate": 3.3580020729573338e-09,
-      "loss": 0.0044,
+      "loss": 0.03,
       "step": 2232
     },
     {
       "epoch": 1.98,
       "learning_rate": 2.99527304537639e-09,
-      "loss": 0.0012,
+      "loss": 0.0313,
       "step": 2233
     },
     {
       "epoch": 1.99,
       "learning_rate": 2.6532673732049265e-09,
-      "loss": 0.0011,
+      "loss": 0.0198,
       "step": 2234
     },
     {
       "epoch": 1.99,
       "learning_rate": 2.3319857654080515e-09,
-      "loss": 0.015,
+      "loss": 0.0281,
       "step": 2235
     },
     {
       "epoch": 1.99,
       "learning_rate": 2.0314288879874635e-09,
-      "loss": 0.0038,
+      "loss": 0.035,
       "step": 2236
     },
     {
       "epoch": 1.99,
       "learning_rate": 1.7515973639836702e-09,
-      "loss": 0.0013,
+      "loss": 0.0222,
       "step": 2237
     },
     {
       "epoch": 1.99,
       "learning_rate": 1.4924917734771006e-09,
-      "loss": 0.0026,
+      "loss": 0.023,
       "step": 2238
     },
     {
       "epoch": 1.99,
       "learning_rate": 1.2541126535814407e-09,
-      "loss": 0.0031,
+      "loss": 0.0337,
       "step": 2239
     },
     {
       "epoch": 1.99,
       "learning_rate": 1.036460498445857e-09,
-      "loss": 0.0051,
+      "loss": 0.03,
       "step": 2240
     },
     {
       "epoch": 1.99,
       "learning_rate": 8.395357592538844e-10,
-      "loss": 0.0011,
+      "loss": 0.0225,
       "step": 2241
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.633388442223166e-10,
-      "loss": 0.0017,
+      "loss": 0.0258,
       "step": 2242
     },
     {
       "epoch": 1.99,
       "learning_rate": 5.078701186000956e-10,
-      "loss": 0.0008,
+      "loss": 0.0351,
       "step": 2243
     },
     {
       "epoch": 1.99,
       "learning_rate": 3.7312990466609187e-10,
-      "loss": 0.0006,
+      "loss": 0.0211,
       "step": 2244
     },
     {
       "epoch": 2.0,
       "learning_rate": 2.5911848173132416e-10,
-      "loss": 0.005,
+      "loss": 0.0228,
       "step": 2245
     },
     {
       "epoch": 2.0,
       "learning_rate": 1.6583608613562895e-10,
-      "loss": 0.005,
+      "loss": 0.0224,
       "step": 2246
     },
     {
       "epoch": 2.0,
       "learning_rate": 9.328291125099142e-11,
-      "loss": 0.0104,
+      "loss": 0.026,
       "step": 2247
     },
     {
       "epoch": 2.0,
       "learning_rate": 4.1459107477104245e-11,
-      "loss": 0.0021,
+      "loss": 0.0219,
       "step": 2248
     },
     {
       "epoch": 2.0,
       "learning_rate": 1.0364782240257499e-11,
-      "loss": 0.0021,
+      "loss": 0.0264,
       "step": 2249
     },
     {
       "epoch": 2.0,
       "learning_rate": 0.0,
-      "loss": 0.0027,
+      "loss": 0.03,
       "step": 2250
     },
     {
       "epoch": 2.0,
       "step": 2250,
-      "total_flos": 124165768839168.0,
-      "train_loss": 0.03172136057747735,
-      "train_runtime": 7604.0402,
-      "train_samples_per_second": 37.86,
-      "train_steps_per_second": 0.296
+      "total_flos": 124203837456384.0,
+      "train_loss": 0.06862774997287327,
+      "train_runtime": 7572.783,
+      "train_samples_per_second": 38.016,
+      "train_steps_per_second": 0.297
     }
   ],
   "logging_steps": 1.0,
   "max_steps": 2250,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
-  "save_steps": 1126,
-  "total_flos": 124165768839168.0,
+  "save_steps": 2500,
+  "total_flos": 124203837456384.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null