diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -100,463 +100,463 @@
     {
       "epoch": 0.29,
       "learning_rate": 1.2e-05,
-      "loss": 14.8434,
+      "loss": 14.8436,
       "step": 16
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.3e-05,
-      "loss": 13.3847,
+      "loss": 13.3845,
       "step": 17
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.4e-05,
-      "loss": 12.7353,
+      "loss": 12.7349,
       "step": 18
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.5e-05,
-      "loss": 14.4867,
+      "loss": 14.4859,
       "step": 19
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.6e-05,
-      "loss": 15.2679,
+      "loss": 15.2667,
       "step": 20
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 13.9622,
+      "loss": 13.9612,
       "step": 21
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.8e-05,
-      "loss": 13.1765,
+      "loss": 13.1742,
       "step": 22
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.9e-05,
-      "loss": 13.5087,
+      "loss": 13.5057,
       "step": 23
     },
     {
       "epoch": 0.44,
       "learning_rate": 2e-05,
-      "loss": 11.9876,
+      "loss": 11.9828,
       "step": 24
     },
     {
       "epoch": 0.45,
       "learning_rate": 2.1000000000000002e-05,
-      "loss": 13.563,
+      "loss": 13.5504,
       "step": 25
     },
     {
       "epoch": 0.47,
       "learning_rate": 2.2e-05,
-      "loss": 13.9001,
+      "loss": 13.8855,
       "step": 26
     },
     {
       "epoch": 0.49,
       "learning_rate": 2.3e-05,
-      "loss": 13.0176,
+      "loss": 12.9999,
       "step": 27
     },
     {
       "epoch": 0.51,
       "learning_rate": 2.4e-05,
-      "loss": 11.5002,
+      "loss": 11.4835,
       "step": 28
     },
     {
       "epoch": 0.53,
       "learning_rate": 2.5e-05,
-      "loss": 11.752,
+      "loss": 11.7251,
       "step": 29
     },
     {
       "epoch": 0.55,
       "learning_rate": 2.6e-05,
-      "loss": 10.5043,
+      "loss": 10.4768,
       "step": 30
     },
     {
       "epoch": 0.56,
       "learning_rate": 2.7e-05,
-      "loss": 10.0646,
+      "loss": 10.0401,
       "step": 31
     },
     {
       "epoch": 0.58,
       "learning_rate": 2.8e-05,
-      "loss": 11.345,
+      "loss": 11.3126,
       "step": 32
     },
     {
       "epoch": 0.6,
       "learning_rate": 2.9000000000000004e-05,
-      "loss": 9.9701,
+      "loss": 9.9391,
       "step": 33
     },
     {
       "epoch": 0.62,
       "learning_rate": 3e-05,
-      "loss": 8.84,
+      "loss": 8.8129,
       "step": 34
     },
     {
       "epoch": 0.64,
       "learning_rate": 3.1e-05,
-      "loss": 8.3918,
+      "loss": 8.3672,
       "step": 35
     },
     {
       "epoch": 0.65,
       "learning_rate": 3.2e-05,
-      "loss": 7.7157,
+      "loss": 7.6982,
       "step": 36
     },
     {
       "epoch": 0.67,
       "learning_rate": 3.3e-05,
-      "loss": 7.0621,
+      "loss": 7.0469,
       "step": 37
     },
     {
       "epoch": 0.69,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 7.8151,
+      "loss": 7.788,
       "step": 38
     },
     {
       "epoch": 0.71,
       "learning_rate": 3.5000000000000004e-05,
-      "loss": 7.3067,
+      "loss": 7.2927,
       "step": 39
     },
     {
       "epoch": 0.73,
       "learning_rate": 3.6e-05,
-      "loss": 7.1476,
+      "loss": 7.133,
       "step": 40
     },
     {
       "epoch": 0.75,
       "learning_rate": 3.7e-05,
-      "loss": 6.7575,
+      "loss": 6.7386,
       "step": 41
     },
     {
       "epoch": 0.76,
       "learning_rate": 3.8e-05,
-      "loss": 6.1275,
+      "loss": 6.118,
       "step": 42
     },
     {
       "epoch": 0.78,
       "learning_rate": 3.9e-05,
-      "loss": 6.0518,
+      "loss": 6.0401,
       "step": 43
     },
     {
       "epoch": 0.8,
       "learning_rate": 4e-05,
-      "loss": 7.0841,
+      "loss": 7.0736,
       "step": 44
     },
     {
       "epoch": 0.82,
       "learning_rate": 4.1e-05,
-      "loss": 5.9306,
+      "loss": 5.9223,
       "step": 45
     },
     {
       "epoch": 0.84,
       "learning_rate": 4.2000000000000004e-05,
-      "loss": 5.75,
+      "loss": 5.7344,
       "step": 46
     },
     {
       "epoch": 0.85,
       "learning_rate": 4.2999999999999995e-05,
-      "loss": 5.48,
+      "loss": 5.4695,
       "step": 47
     },
     {
       "epoch": 0.87,
       "learning_rate": 4.4e-05,
-      "loss": 5.2013,
+      "loss": 5.1942,
       "step": 48
     },
     {
       "epoch": 0.89,
       "learning_rate": 4.4999999999999996e-05,
-      "loss": 5.2336,
+      "loss": 5.2193,
       "step": 49
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.6e-05,
-      "loss": 5.1942,
+      "loss": 5.1935,
       "step": 50
     },
     {
       "epoch": 0.93,
       "learning_rate": 4.7000000000000004e-05,
-      "loss": 5.2704,
+      "loss": 5.2608,
       "step": 51
     },
     {
       "epoch": 0.95,
       "learning_rate": 4.8e-05,
-      "loss": 5.0845,
+      "loss": 5.0793,
       "step": 52
     },
     {
       "epoch": 0.96,
       "learning_rate": 4.9000000000000005e-05,
-      "loss": 4.8139,
+      "loss": 4.806,
       "step": 53
     },
     {
       "epoch": 0.98,
       "learning_rate": 5e-05,
-      "loss": 4.8668,
+      "loss": 4.8593,
       "step": 54
     },
     {
       "epoch": 1.0,
       "learning_rate": 5.1e-05,
-      "loss": 5.4899,
+      "loss": 5.4801,
       "step": 55
     },
     {
       "epoch": 1.02,
       "learning_rate": 5.2e-05,
-      "loss": 4.8183,
+      "loss": 4.8135,
       "step": 56
     },
     {
       "epoch": 1.04,
       "learning_rate": 5.3e-05,
-      "loss": 4.7891,
+      "loss": 4.7826,
       "step": 57
     },
     {
       "epoch": 1.05,
       "learning_rate": 5.4e-05,
-      "loss": 4.5955,
+      "loss": 4.5923,
       "step": 58
     },
     {
       "epoch": 1.07,
       "learning_rate": 5.5e-05,
-      "loss": 4.4249,
+      "loss": 4.4197,
       "step": 59
     },
     {
       "epoch": 1.09,
       "learning_rate": 5.6e-05,
-      "loss": 4.3802,
+      "loss": 4.3772,
       "step": 60
     },
     {
       "epoch": 1.11,
       "learning_rate": 5.7e-05,
-      "loss": 4.7067,
+      "loss": 4.7035,
       "step": 61
     },
     {
       "epoch": 1.13,
       "learning_rate": 5.800000000000001e-05,
-      "loss": 4.7273,
+      "loss": 4.7221,
       "step": 62
     },
     {
       "epoch": 1.15,
       "learning_rate": 5.9e-05,
-      "loss": 4.4545,
+      "loss": 4.45,
       "step": 63
     },
     {
       "epoch": 1.16,
       "learning_rate": 6e-05,
-      "loss": 4.3566,
+      "loss": 4.3542,
       "step": 64
     },
     {
       "epoch": 1.18,
       "learning_rate": 6.1e-05,
-      "loss": 4.2668,
+      "loss": 4.2637,
       "step": 65
     },
     {
       "epoch": 1.2,
       "learning_rate": 6.2e-05,
-      "loss": 4.1737,
+      "loss": 4.1704,
       "step": 66
     },
     {
       "epoch": 1.22,
       "learning_rate": 6.3e-05,
-      "loss": 4.1134,
+      "loss": 4.112,
       "step": 67
     },
     {
       "epoch": 1.24,
       "learning_rate": 6.4e-05,
-      "loss": 4.3,
+      "loss": 4.2965,
       "step": 68
     },
     {
       "epoch": 1.25,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 4.2145,
+      "loss": 4.2112,
       "step": 69
     },
     {
       "epoch": 1.27,
       "learning_rate": 6.6e-05,
-      "loss": 4.1075,
+      "loss": 4.1061,
       "step": 70
     },
     {
       "epoch": 1.29,
       "learning_rate": 6.7e-05,
-      "loss": 4.0525,
+      "loss": 4.0507,
       "step": 71
     },
     {
       "epoch": 1.31,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 3.9147,
+      "loss": 3.9127,
       "step": 72
     },
     {
       "epoch": 1.33,
       "learning_rate": 6.900000000000001e-05,
-      "loss": 3.9069,
+      "loss": 3.9066,
       "step": 73
     },
     {
       "epoch": 1.35,
       "learning_rate": 7.000000000000001e-05,
-      "loss": 4.0344,
+      "loss": 4.0324,
       "step": 74
     },
     {
       "epoch": 1.36,
       "learning_rate": 7.099999999999999e-05,
-      "loss": 4.0426,
+      "loss": 4.0413,
       "step": 75
     },
     {
       "epoch": 1.38,
       "learning_rate": 7.2e-05,
-      "loss": 3.9238,
+      "loss": 3.9241,
       "step": 76
     },
     {
       "epoch": 1.4,
       "learning_rate": 7.3e-05,
-      "loss": 3.8795,
+      "loss": 3.878,
       "step": 77
     },
     {
       "epoch": 1.42,
       "learning_rate": 7.4e-05,
-      "loss": 3.8475,
+      "loss": 3.8476,
       "step": 78
     },
     {
       "epoch": 1.44,
       "learning_rate": 7.5e-05,
-      "loss": 3.826,
+      "loss": 3.8257,
       "step": 79
     },
     {
       "epoch": 1.45,
       "learning_rate": 7.6e-05,
-      "loss": 3.9443,
+      "loss": 3.9435,
       "step": 80
     },
     {
       "epoch": 1.47,
       "learning_rate": 7.7e-05,
-      "loss": 3.8657,
+      "loss": 3.8658,
       "step": 81
     },
     {
       "epoch": 1.49,
       "learning_rate": 7.8e-05,
-      "loss": 3.7568,
+      "loss": 3.7561,
       "step": 82
     },
     {
       "epoch": 1.51,
       "learning_rate": 7.9e-05,
-      "loss": 3.7977,
+      "loss": 3.7974,
       "step": 83
     },
     {
       "epoch": 1.53,
       "learning_rate": 8e-05,
-      "loss": 3.7231,
+      "loss": 3.7229,
       "step": 84
     },
     {
       "epoch": 1.55,
       "learning_rate": 8.1e-05,
-      "loss": 3.63,
+      "loss": 3.6296,
       "step": 85
     },
     {
       "epoch": 1.56,
       "learning_rate": 8.2e-05,
-      "loss": 3.6694,
+      "loss": 3.6695,
       "step": 86
     },
     {
       "epoch": 1.58,
       "learning_rate": 8.300000000000001e-05,
-      "loss": 3.7353,
+      "loss": 3.7347,
       "step": 87
     },
     {
       "epoch": 1.6,
       "learning_rate": 8.400000000000001e-05,
-      "loss": 3.6386,
+      "loss": 3.6385,
       "step": 88
     },
     {
       "epoch": 1.62,
       "learning_rate": 8.5e-05,
-      "loss": 3.6016,
+      "loss": 3.6013,
       "step": 89
     },
     {
       "epoch": 1.64,
       "learning_rate": 8.599999999999999e-05,
-      "loss": 3.5628,
+      "loss": 3.5625,
       "step": 90
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.7e-05,
-      "loss": 3.5266,
+      "loss": 3.5267,
       "step": 91
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.8e-05,
-      "loss": 3.6114,
+      "loss": 3.6111,
       "step": 92
     },
     {
@@ -568,19 +568,19 @@
     {
       "epoch": 1.71,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 3.5156,
+      "loss": 3.5158,
       "step": 94
     },
     {
       "epoch": 1.73,
       "learning_rate": 9.1e-05,
-      "loss": 3.5022,
+      "loss": 3.5016,
       "step": 95
     },
     {
       "epoch": 1.75,
       "learning_rate": 9.2e-05,
-      "loss": 3.474,
+      "loss": 3.4743,
       "step": 96
     },
     {
@@ -592,13 +592,13 @@
     {
       "epoch": 1.78,
       "learning_rate": 9.400000000000001e-05,
-      "loss": 3.4268,
+      "loss": 3.4267,
       "step": 98
     },
     {
       "epoch": 1.8,
       "learning_rate": 9.5e-05,
-      "loss": 3.5077,
+      "loss": 3.508,
       "step": 99
     },
     {
@@ -610,13 +610,13 @@
     {
       "epoch": 1.84,
       "learning_rate": 9.7e-05,
-      "loss": 3.3916,
+      "loss": 3.3915,
       "step": 101
     },
     {
       "epoch": 1.85,
       "learning_rate": 9.800000000000001e-05,
-      "loss": 3.3406,
+      "loss": 3.3407,
       "step": 102
     },
     {
@@ -628,157 +628,157 @@
     {
       "epoch": 1.89,
       "learning_rate": 0.0001,
-      "loss": 3.3517,
+      "loss": 3.3518,
       "step": 104
     },
     {
       "epoch": 1.91,
       "learning_rate": 0.000101,
-      "loss": 3.3858,
+      "loss": 3.3859,
       "step": 105
     },
     {
       "epoch": 1.93,
       "learning_rate": 0.000102,
-      "loss": 3.3011,
+      "loss": 3.3012,
       "step": 106
     },
     {
       "epoch": 1.95,
       "learning_rate": 0.000103,
-      "loss": 3.3005,
+      "loss": 3.3006,
       "step": 107
     },
     {
       "epoch": 1.96,
       "learning_rate": 0.000104,
-      "loss": 3.2927,
+      "loss": 3.2928,
       "step": 108
     },
     {
       "epoch": 1.98,
       "learning_rate": 0.000105,
-      "loss": 3.3217,
+      "loss": 3.3218,
       "step": 109
     },
     {
       "epoch": 2.0,
       "learning_rate": 0.000106,
-      "loss": 3.3627,
+      "loss": 3.363,
       "step": 110
     },
     {
       "epoch": 2.02,
       "learning_rate": 0.000107,
-      "loss": 3.2978,
+      "loss": 3.2979,
       "step": 111
     },
     {
       "epoch": 2.04,
       "learning_rate": 0.000108,
-      "loss": 3.2557,
+      "loss": 3.2555,
       "step": 112
     },
     {
       "epoch": 2.05,
       "learning_rate": 0.000109,
-      "loss": 3.2299,
+      "loss": 3.2301,
       "step": 113
     },
     {
       "epoch": 2.07,
       "learning_rate": 0.00011,
-      "loss": 3.2368,
+      "loss": 3.2371,
       "step": 114
     },
     {
       "epoch": 2.09,
       "learning_rate": 0.000111,
-      "loss": 3.2166,
+      "loss": 3.2167,
       "step": 115
     },
     {
       "epoch": 2.11,
       "learning_rate": 0.000112,
-      "loss": 3.2651,
+      "loss": 3.2661,
       "step": 116
     },
     {
       "epoch": 2.13,
       "learning_rate": 0.00011300000000000001,
-      "loss": 3.2589,
+      "loss": 3.2592,
       "step": 117
     },
     {
       "epoch": 2.15,
       "learning_rate": 0.000114,
-      "loss": 3.2091,
+      "loss": 3.2096,
       "step": 118
     },
     {
       "epoch": 2.16,
       "learning_rate": 0.000115,
-      "loss": 3.2079,
+      "loss": 3.2083,
       "step": 119
     },
     {
       "epoch": 2.18,
       "learning_rate": 0.00011600000000000001,
-      "loss": 3.2058,
+      "loss": 3.2065,
       "step": 120
     },
     {
       "epoch": 2.2,
       "learning_rate": 0.00011700000000000001,
-      "loss": 3.194,
+      "loss": 3.1948,
       "step": 121
     },
     {
       "epoch": 2.22,
       "learning_rate": 0.000118,
-      "loss": 3.2257,
+      "loss": 3.2264,
       "step": 122
     },
     {
       "epoch": 2.24,
       "learning_rate": 0.00011899999999999999,
-      "loss": 3.3546,
+      "loss": 3.3564,
       "step": 123
     },
     {
       "epoch": 2.25,
       "learning_rate": 0.00012,
-      "loss": 3.1913,
+      "loss": 3.1916,
       "step": 124
     },
     {
       "epoch": 2.27,
       "learning_rate": 0.000121,
-      "loss": 3.1801,
+      "loss": 3.1813,
       "step": 125
     },
     {
       "epoch": 2.29,
       "learning_rate": 0.000122,
-      "loss": 3.1912,
+      "loss": 3.192,
       "step": 126
     },
     {
       "epoch": 2.31,
       "learning_rate": 0.000123,
-      "loss": 3.1784,
+      "loss": 3.179,
       "step": 127
     },
     {
       "epoch": 2.33,
       "learning_rate": 0.000124,
-      "loss": 3.1679,
+      "loss": 3.1696,
       "step": 128
     },
     {
       "epoch": 2.35,
       "learning_rate": 0.000125,
-      "loss": 3.238,
+      "loss": 3.2385,
       "step": 129
     },
     {
@@ -790,7857 +790,7857 @@
     {
       "epoch": 2.38,
       "learning_rate": 0.000127,
-      "loss": 3.1622,
+      "loss": 3.1604,
       "step": 131
     },
     {
       "epoch": 2.4,
       "learning_rate": 0.000128,
-      "loss": 3.1427,
+      "loss": 3.145,
       "step": 132
     },
     {
       "epoch": 2.42,
       "learning_rate": 0.00012900000000000002,
-      "loss": 3.1565,
+      "loss": 3.1575,
       "step": 133
     },
     {
       "epoch": 2.44,
       "learning_rate": 0.00013000000000000002,
-      "loss": 3.1401,
+      "loss": 3.1403,
       "step": 134
     },
     {
       "epoch": 2.45,
       "learning_rate": 0.000131,
-      "loss": 3.19,
+      "loss": 3.1664,
       "step": 135
     },
     {
       "epoch": 2.47,
       "learning_rate": 0.000132,
-      "loss": 3.1376,
+      "loss": 3.1662,
       "step": 136
     },
     {
       "epoch": 2.49,
       "learning_rate": 0.000133,
-      "loss": 3.1071,
+      "loss": 3.1023,
       "step": 137
     },
     {
       "epoch": 2.51,
       "learning_rate": 0.000134,
-      "loss": 3.125,
+      "loss": 3.1125,
       "step": 138
     },
     {
       "epoch": 2.53,
       "learning_rate": 0.000135,
-      "loss": 3.1312,
+      "loss": 3.1349,
       "step": 139
     },
     {
       "epoch": 2.55,
       "learning_rate": 0.00013600000000000003,
-      "loss": 3.1184,
+      "loss": 3.1101,
       "step": 140
     },
     {
       "epoch": 2.56,
       "learning_rate": 0.00013700000000000002,
-      "loss": 3.159,
+      "loss": 3.1658,
       "step": 141
     },
     {
       "epoch": 2.58,
       "learning_rate": 0.00013800000000000002,
-      "loss": 3.1468,
+      "loss": 3.1445,
       "step": 142
     },
     {
       "epoch": 2.6,
       "learning_rate": 0.00013900000000000002,
-      "loss": 3.1065,
+      "loss": 3.102,
       "step": 143
     },
     {
       "epoch": 2.62,
       "learning_rate": 0.00014000000000000001,
-      "loss": 3.1063,
+      "loss": 3.1052,
       "step": 144
     },
     {
       "epoch": 2.64,
       "learning_rate": 0.00014099999999999998,
-      "loss": 3.0925,
+      "loss": 3.1026,
       "step": 145
     },
     {
       "epoch": 2.65,
       "learning_rate": 0.00014199999999999998,
-      "loss": 3.1254,
+      "loss": 3.1702,
       "step": 146
     },
     {
       "epoch": 2.67,
       "learning_rate": 0.00014299999999999998,
-      "loss": 3.104,
+      "loss": 3.1051,
       "step": 147
     },
     {
       "epoch": 2.69,
       "learning_rate": 0.000144,
-      "loss": 3.1528,
+      "loss": 3.1723,
       "step": 148
     },
     {
       "epoch": 2.71,
       "learning_rate": 0.000145,
-      "loss": 3.0855,
+      "loss": 3.0944,
       "step": 149
     },
     {
       "epoch": 2.73,
       "learning_rate": 0.000146,
-      "loss": 3.0862,
+      "loss": 3.096,
       "step": 150
     },
     {
       "epoch": 2.75,
       "learning_rate": 0.000147,
-      "loss": 3.0729,
+      "loss": 3.0751,
       "step": 151
     },
     {
       "epoch": 2.76,
       "learning_rate": 0.000148,
-      "loss": 3.0882,
+      "loss": 3.1086,
       "step": 152
     },
     {
       "epoch": 2.78,
       "learning_rate": 0.000149,
-      "loss": 3.0835,
+      "loss": 3.1018,
       "step": 153
     },
     {
       "epoch": 2.8,
       "learning_rate": 0.00015,
-      "loss": 3.0749,
+      "loss": 3.0904,
       "step": 154
     },
     {
       "epoch": 2.82,
       "learning_rate": 0.000151,
-      "loss": 3.0826,
+      "loss": 3.0853,
       "step": 155
     },
     {
       "epoch": 2.84,
       "learning_rate": 0.000152,
-      "loss": 3.0554,
+      "loss": 3.0608,
       "step": 156
     },
     {
       "epoch": 2.85,
       "learning_rate": 0.000153,
-      "loss": 3.055,
+      "loss": 3.0627,
       "step": 157
     },
     {
       "epoch": 2.87,
       "learning_rate": 0.000154,
-      "loss": 3.0583,
+      "loss": 3.0684,
       "step": 158
     },
     {
       "epoch": 2.89,
       "learning_rate": 0.000155,
-      "loss": 3.0761,
+      "loss": 3.0799,
       "step": 159
     },
     {
       "epoch": 2.91,
       "learning_rate": 0.000156,
-      "loss": 3.1005,
+      "loss": 3.1102,
       "step": 160
     },
     {
       "epoch": 2.93,
       "learning_rate": 0.000157,
-      "loss": 3.0543,
+      "loss": 3.0573,
       "step": 161
     },
     {
       "epoch": 2.95,
       "learning_rate": 0.000158,
-      "loss": 3.0623,
+      "loss": 3.0716,
       "step": 162
     },
     {
       "epoch": 2.96,
       "learning_rate": 0.00015900000000000002,
-      "loss": 3.037,
+      "loss": 3.0455,
       "step": 163
     },
     {
       "epoch": 2.98,
       "learning_rate": 0.00016,
-      "loss": 3.024,
+      "loss": 3.032,
       "step": 164
     },
     {
       "epoch": 3.0,
       "learning_rate": 0.000161,
-      "loss": 3.0615,
+      "loss": 3.0731,
       "step": 165
     },
     {
       "epoch": 3.02,
       "learning_rate": 0.000162,
-      "loss": 3.0502,
+      "loss": 3.0543,
       "step": 166
     },
     {
       "epoch": 3.04,
       "learning_rate": 0.000163,
-      "loss": 3.0797,
+      "loss": 3.0656,
       "step": 167
     },
     {
       "epoch": 3.05,
       "learning_rate": 0.000164,
-      "loss": 3.0755,
+      "loss": 3.0728,
       "step": 168
     },
     {
       "epoch": 3.07,
       "learning_rate": 0.000165,
-      "loss": 3.0561,
+      "loss": 3.0593,
       "step": 169
     },
     {
       "epoch": 3.09,
       "learning_rate": 0.00016600000000000002,
-      "loss": 3.0959,
+      "loss": 3.0952,
       "step": 170
     },
     {
       "epoch": 3.11,
       "learning_rate": 0.00016700000000000002,
-      "loss": 3.1054,
+      "loss": 3.1384,
       "step": 171
     },
     {
       "epoch": 3.13,
       "learning_rate": 0.00016800000000000002,
-      "loss": 3.0588,
+      "loss": 3.0624,
       "step": 172
     },
     {
       "epoch": 3.15,
       "learning_rate": 0.00016900000000000002,
-      "loss": 3.0331,
+      "loss": 3.0304,
       "step": 173
     },
     {
       "epoch": 3.16,
       "learning_rate": 0.00017,
-      "loss": 3.0499,
+      "loss": 3.0632,
       "step": 174
     },
     {
       "epoch": 3.18,
       "learning_rate": 0.000171,
-      "loss": 3.0592,
+      "loss": 3.0636,
       "step": 175
     },
     {
       "epoch": 3.2,
       "learning_rate": 0.00017199999999999998,
-      "loss": 3.034,
+      "loss": 3.039,
       "step": 176
     },
     {
       "epoch": 3.22,
       "learning_rate": 0.000173,
-      "loss": 3.0323,
+      "loss": 3.0391,
       "step": 177
     },
     {
       "epoch": 3.24,
       "learning_rate": 0.000174,
-      "loss": 3.0774,
+      "loss": 3.1005,
       "step": 178
     },
     {
       "epoch": 3.25,
       "learning_rate": 0.000175,
-      "loss": 3.065,
+      "loss": 3.0673,
       "step": 179
     },
     {
       "epoch": 3.27,
       "learning_rate": 0.000176,
-      "loss": 3.002,
+      "loss": 2.9988,
       "step": 180
     },
     {
       "epoch": 3.29,
       "learning_rate": 0.000177,
-      "loss": 3.0421,
+      "loss": 3.0357,
       "step": 181
     },
     {
       "epoch": 3.31,
       "learning_rate": 0.000178,
-      "loss": 3.0703,
+      "loss": 3.0817,
       "step": 182
     },
     {
       "epoch": 3.33,
       "learning_rate": 0.000179,
-      "loss": 3.0061,
+      "loss": 3.0074,
       "step": 183
     },
     {
       "epoch": 3.35,
       "learning_rate": 0.00017999999999999998,
-      "loss": 3.0408,
+      "loss": 3.0521,
       "step": 184
     },
     {
       "epoch": 3.36,
       "learning_rate": 0.000181,
-      "loss": 3.0361,
+      "loss": 3.0422,
       "step": 185
     },
     {
       "epoch": 3.38,
       "learning_rate": 0.000182,
-      "loss": 3.059,
+      "loss": 3.0688,
       "step": 186
     },
     {
       "epoch": 3.4,
       "learning_rate": 0.000183,
-      "loss": 3.0361,
+      "loss": 3.0449,
       "step": 187
     },
     {
       "epoch": 3.42,
       "learning_rate": 0.000184,
-      "loss": 3.0309,
+      "loss": 3.0332,
       "step": 188
     },
     {
       "epoch": 3.44,
       "learning_rate": 0.000185,
-      "loss": 3.0325,
+      "loss": 3.0424,
       "step": 189
     },
     {
       "epoch": 3.45,
       "learning_rate": 0.000186,
-      "loss": 3.0286,
+      "loss": 3.0461,
       "step": 190
     },
     {
       "epoch": 3.47,
       "learning_rate": 0.000187,
-      "loss": 3.0525,
+      "loss": 3.0643,
       "step": 191
     },
     {
       "epoch": 3.49,
       "learning_rate": 0.00018800000000000002,
-      "loss": 3.0472,
+      "loss": 3.0508,
       "step": 192
     },
     {
       "epoch": 3.51,
       "learning_rate": 0.000189,
-      "loss": 3.0264,
+      "loss": 3.032,
       "step": 193
     },
     {
       "epoch": 3.53,
       "learning_rate": 0.00019,
-      "loss": 3.0299,
+      "loss": 3.0343,
       "step": 194
     },
     {
       "epoch": 3.55,
       "learning_rate": 0.000191,
-      "loss": 3.0292,
+      "loss": 3.0387,
       "step": 195
     },
     {
       "epoch": 3.56,
       "learning_rate": 0.000192,
-      "loss": 3.0072,
+      "loss": 3.0167,
       "step": 196
     },
     {
       "epoch": 3.58,
       "learning_rate": 0.000193,
-      "loss": 3.0249,
+      "loss": 3.0325,
       "step": 197
     },
     {
       "epoch": 3.6,
       "learning_rate": 0.000194,
-      "loss": 3.0343,
+      "loss": 3.0403,
       "step": 198
     },
     {
       "epoch": 3.62,
       "learning_rate": 0.00019500000000000002,
-      "loss": 3.0072,
+      "loss": 3.0195,
       "step": 199
     },
     {
       "epoch": 3.64,
       "learning_rate": 0.00019600000000000002,
-      "loss": 3.0279,
+      "loss": 3.0344,
       "step": 200
     },
     {
       "epoch": 3.65,
       "learning_rate": 0.00019700000000000002,
-      "loss": 3.0084,
+      "loss": 3.0141,
       "step": 201
     },
     {
       "epoch": 3.67,
       "learning_rate": 0.00019800000000000002,
-      "loss": 2.9773,
+      "loss": 2.9858,
       "step": 202
     },
     {
       "epoch": 3.69,
       "learning_rate": 0.000199,
-      "loss": 3.0364,
+      "loss": 3.047,
       "step": 203
     },
     {
       "epoch": 3.71,
       "learning_rate": 0.0002,
-      "loss": 3.0133,
+      "loss": 3.0179,
       "step": 204
     },
     {
       "epoch": 3.73,
       "learning_rate": 0.000201,
-      "loss": 2.9928,
+      "loss": 3.0001,
       "step": 205
     },
     {
       "epoch": 3.75,
       "learning_rate": 0.000202,
-      "loss": 2.9999,
+      "loss": 3.0088,
       "step": 206
     },
     {
       "epoch": 3.76,
       "learning_rate": 0.00020300000000000003,
-      "loss": 2.9836,
+      "loss": 2.9936,
       "step": 207
     },
     {
       "epoch": 3.78,
       "learning_rate": 0.000204,
-      "loss": 2.985,
+      "loss": 2.9983,
       "step": 208
     },
     {
       "epoch": 3.8,
       "learning_rate": 0.000205,
-      "loss": 2.96,
+      "loss": 2.9797,
       "step": 209
     },
     {
       "epoch": 3.82,
       "learning_rate": 0.000206,
-      "loss": 2.9872,
+      "loss": 3.002,
       "step": 210
     },
     {
       "epoch": 3.84,
       "learning_rate": 0.000207,
-      "loss": 2.9954,
+      "loss": 3.0194,
       "step": 211
     },
     {
       "epoch": 3.85,
       "learning_rate": 0.000208,
-      "loss": 2.9774,
+      "loss": 3.0002,
       "step": 212
     },
     {
       "epoch": 3.87,
       "learning_rate": 0.00020899999999999998,
-      "loss": 2.9585,
+      "loss": 2.9841,
       "step": 213
     },
     {
       "epoch": 3.89,
       "learning_rate": 0.00021,
-      "loss": 2.9698,
+      "loss": 2.985,
       "step": 214
     },
     {
       "epoch": 3.91,
       "learning_rate": 0.000211,
-      "loss": 3.0113,
+      "loss": 3.0376,
       "step": 215
     },
     {
       "epoch": 3.93,
       "learning_rate": 0.000212,
-      "loss": 3.0041,
+      "loss": 3.0038,
       "step": 216
     },
     {
       "epoch": 3.95,
       "learning_rate": 0.000213,
-      "loss": 2.989,
+      "loss": 2.9834,
       "step": 217
     },
     {
       "epoch": 3.96,
       "learning_rate": 0.000214,
-      "loss": 2.9904,
+      "loss": 2.9885,
       "step": 218
     },
     {
       "epoch": 3.98,
       "learning_rate": 0.000215,
-      "loss": 2.9334,
+      "loss": 2.9501,
       "step": 219
     },
     {
       "epoch": 4.0,
       "learning_rate": 0.000216,
-      "loss": 2.9714,
+      "loss": 2.9682,
       "step": 220
     },
     {
       "epoch": 4.02,
       "learning_rate": 0.00021700000000000002,
-      "loss": 2.9956,
+      "loss": 2.9893,
       "step": 221
     },
     {
       "epoch": 4.04,
       "learning_rate": 0.000218,
-      "loss": 2.9623,
+      "loss": 2.9759,
       "step": 222
     },
     {
       "epoch": 4.05,
       "learning_rate": 0.000219,
-      "loss": 2.9466,
+      "loss": 2.9426,
       "step": 223
     },
     {
       "epoch": 4.07,
       "learning_rate": 0.00022,
-      "loss": 2.9743,
+      "loss": 2.9926,
       "step": 224
     },
     {
       "epoch": 4.09,
       "learning_rate": 0.000221,
-      "loss": 2.9319,
+      "loss": 2.9561,
       "step": 225
     },
     {
       "epoch": 4.11,
       "learning_rate": 0.000222,
-      "loss": 2.9228,
+      "loss": 2.9403,
       "step": 226
     },
     {
       "epoch": 4.13,
       "learning_rate": 0.000223,
-      "loss": 2.9784,
+      "loss": 2.9725,
       "step": 227
     },
     {
       "epoch": 4.15,
       "learning_rate": 0.000224,
-      "loss": 2.9447,
+      "loss": 2.9766,
       "step": 228
     },
     {
       "epoch": 4.16,
       "learning_rate": 0.00022500000000000002,
-      "loss": 2.9417,
+      "loss": 2.9548,
       "step": 229
     },
     {
       "epoch": 4.18,
       "learning_rate": 0.00022600000000000002,
-      "loss": 2.9148,
+      "loss": 2.9513,
       "step": 230
     },
     {
       "epoch": 4.2,
       "learning_rate": 0.00022700000000000002,
-      "loss": 2.8922,
+      "loss": 2.9077,
       "step": 231
     },
     {
       "epoch": 4.22,
       "learning_rate": 0.000228,
-      "loss": 2.8836,
+      "loss": 2.9143,
       "step": 232
     },
     {
       "epoch": 4.24,
       "learning_rate": 0.000229,
-      "loss": 2.9092,
+      "loss": 2.9356,
       "step": 233
     },
     {
       "epoch": 4.25,
       "learning_rate": 0.00023,
-      "loss": 2.9154,
+      "loss": 2.9378,
       "step": 234
     },
     {
       "epoch": 4.27,
       "learning_rate": 0.000231,
-      "loss": 2.8459,
+      "loss": 2.8849,
       "step": 235
     },
     {
       "epoch": 4.29,
       "learning_rate": 0.00023200000000000003,
-      "loss": 2.8951,
+      "loss": 2.9252,
       "step": 236
     },
     {
       "epoch": 4.31,
       "learning_rate": 0.00023300000000000003,
-      "loss": 2.8659,
+      "loss": 2.9062,
       "step": 237
     },
     {
       "epoch": 4.33,
       "learning_rate": 0.00023400000000000002,
-      "loss": 2.8649,
+      "loss": 2.901,
       "step": 238
     },
     {
       "epoch": 4.35,
       "learning_rate": 0.000235,
-      "loss": 2.8476,
+      "loss": 2.8922,
       "step": 239
     },
     {
       "epoch": 4.36,
       "learning_rate": 0.000236,
-      "loss": 2.8395,
+      "loss": 2.9049,
       "step": 240
     },
     {
       "epoch": 4.38,
       "learning_rate": 0.000237,
-      "loss": 2.8178,
+      "loss": 2.8723,
       "step": 241
     },
     {
       "epoch": 4.4,
       "learning_rate": 0.00023799999999999998,
-      "loss": 2.9221,
+      "loss": 3.0104,
       "step": 242
     },
     {
       "epoch": 4.42,
       "learning_rate": 0.00023899999999999998,
-      "loss": 2.7778,
+      "loss": 2.8856,
       "step": 243
     },
     {
       "epoch": 4.44,
       "learning_rate": 0.00024,
-      "loss": 2.798,
+      "loss": 2.8642,
       "step": 244
     },
     {
       "epoch": 4.45,
       "learning_rate": 0.000241,
-      "loss": 2.7857,
+      "loss": 2.8782,
       "step": 245
     },
     {
       "epoch": 4.47,
       "learning_rate": 0.000242,
-      "loss": 2.8069,
+      "loss": 2.9113,
       "step": 246
     },
     {
       "epoch": 4.49,
       "learning_rate": 0.000243,
-      "loss": 2.758,
+      "loss": 2.8437,
       "step": 247
     },
     {
       "epoch": 4.51,
       "learning_rate": 0.000244,
-      "loss": 2.7808,
+      "loss": 2.8635,
       "step": 248
     },
     {
       "epoch": 4.53,
       "learning_rate": 0.000245,
-      "loss": 2.7353,
+      "loss": 2.8396,
       "step": 249
     },
     {
       "epoch": 4.55,
       "learning_rate": 0.000246,
-      "loss": 2.6982,
+      "loss": 2.8323,
       "step": 250
     },
     {
       "epoch": 4.56,
       "learning_rate": 0.000247,
-      "loss": 2.6893,
+      "loss": 2.8041,
       "step": 251
     },
     {
       "epoch": 4.58,
       "learning_rate": 0.000248,
-      "loss": 2.6735,
+      "loss": 2.8148,
       "step": 252
     },
     {
       "epoch": 4.6,
       "learning_rate": 0.000249,
-      "loss": 2.6579,
+      "loss": 2.7792,
       "step": 253
     },
     {
       "epoch": 4.62,
       "learning_rate": 0.00025,
-      "loss": 2.6189,
+      "loss": 2.7734,
       "step": 254
     },
     {
       "epoch": 4.64,
       "learning_rate": 0.00025100000000000003,
-      "loss": 2.6484,
+      "loss": 2.7825,
       "step": 255
     },
     {
       "epoch": 4.65,
       "learning_rate": 0.000252,
-      "loss": 2.6139,
+      "loss": 2.7286,
       "step": 256
     },
     {
       "epoch": 4.67,
       "learning_rate": 0.000253,
-      "loss": 2.4748,
+      "loss": 2.6544,
       "step": 257
     },
     {
       "epoch": 4.69,
       "learning_rate": 0.000254,
-      "loss": 2.5707,
+      "loss": 2.7085,
       "step": 258
     },
     {
       "epoch": 4.71,
       "learning_rate": 0.000255,
-      "loss": 2.5199,
+      "loss": 2.6757,
       "step": 259
     },
     {
       "epoch": 4.73,
       "learning_rate": 0.000256,
-      "loss": 2.4575,
+      "loss": 2.6221,
       "step": 260
     },
     {
       "epoch": 4.75,
       "learning_rate": 0.000257,
-      "loss": 2.397,
+      "loss": 2.6726,
       "step": 261
     },
     {
       "epoch": 4.76,
       "learning_rate": 0.00025800000000000004,
-      "loss": 2.3821,
+      "loss": 2.5808,
       "step": 262
     },
     {
       "epoch": 4.78,
       "learning_rate": 0.000259,
-      "loss": 2.3241,
+      "loss": 2.6096,
       "step": 263
     },
     {
       "epoch": 4.8,
       "learning_rate": 0.00026000000000000003,
-      "loss": 2.3209,
+      "loss": 2.576,
       "step": 264
     },
     {
       "epoch": 4.82,
       "learning_rate": 0.000261,
-      "loss": 2.2574,
+      "loss": 2.546,
       "step": 265
     },
     {
       "epoch": 4.84,
       "learning_rate": 0.000262,
-      "loss": 2.2083,
+      "loss": 2.5267,
       "step": 266
     },
     {
       "epoch": 4.85,
       "learning_rate": 0.000263,
-      "loss": 2.1363,
+      "loss": 2.4527,
       "step": 267
     },
     {
       "epoch": 4.87,
       "learning_rate": 0.000264,
-      "loss": 2.2447,
+      "loss": 2.4464,
       "step": 268
     },
     {
       "epoch": 4.89,
       "learning_rate": 0.00026500000000000004,
-      "loss": 2.2039,
+      "loss": 2.4142,
       "step": 269
     },
     {
       "epoch": 4.91,
       "learning_rate": 0.000266,
-      "loss": 2.0992,
+      "loss": 2.3872,
       "step": 270
     },
     {
       "epoch": 4.93,
       "learning_rate": 0.00026700000000000004,
-      "loss": 2.0228,
+      "loss": 2.3191,
       "step": 271
     },
     {
       "epoch": 4.95,
       "learning_rate": 0.000268,
-      "loss": 1.9831,
+      "loss": 2.2963,
       "step": 272
     },
     {
       "epoch": 4.96,
       "learning_rate": 0.00026900000000000003,
-      "loss": 1.9159,
+      "loss": 2.2208,
       "step": 273
     },
     {
       "epoch": 4.98,
       "learning_rate": 0.00027,
-      "loss": 1.8765,
+      "loss": 2.1489,
       "step": 274
     },
     {
       "epoch": 5.0,
       "learning_rate": 0.00027100000000000003,
-      "loss": 1.8392,
+      "loss": 2.072,
       "step": 275
     },
     {
       "epoch": 5.02,
       "learning_rate": 0.00027200000000000005,
-      "loss": 1.7968,
+      "loss": 2.0953,
       "step": 276
     },
     {
       "epoch": 5.04,
       "learning_rate": 0.000273,
-      "loss": 1.7849,
+      "loss": 2.0457,
       "step": 277
     },
     {
       "epoch": 5.05,
       "learning_rate": 0.00027400000000000005,
-      "loss": 1.7099,
+      "loss": 1.9927,
       "step": 278
     },
     {
       "epoch": 5.07,
       "learning_rate": 0.000275,
-      "loss": 1.7709,
+      "loss": 2.0044,
       "step": 279
     },
     {
       "epoch": 5.09,
       "learning_rate": 0.00027600000000000004,
-      "loss": 1.6975,
+      "loss": 1.8778,
       "step": 280
     },
     {
       "epoch": 5.11,
       "learning_rate": 0.000277,
-      "loss": 1.6633,
+      "loss": 1.9065,
       "step": 281
     },
     {
       "epoch": 5.13,
       "learning_rate": 0.00027800000000000004,
-      "loss": 1.5916,
+      "loss": 1.8387,
       "step": 282
     },
     {
       "epoch": 5.15,
       "learning_rate": 0.000279,
-      "loss": 1.5169,
+      "loss": 1.7364,
       "step": 283
     },
     {
       "epoch": 5.16,
       "learning_rate": 0.00028000000000000003,
-      "loss": 1.5347,
+      "loss": 1.7542,
       "step": 284
     },
     {
       "epoch": 5.18,
       "learning_rate": 0.00028100000000000005,
-      "loss": 1.5109,
+      "loss": 1.7038,
       "step": 285
     },
     {
       "epoch": 5.2,
       "learning_rate": 0.00028199999999999997,
-      "loss": 1.4776,
+      "loss": 1.6512,
       "step": 286
     },
     {
       "epoch": 5.22,
       "learning_rate": 0.000283,
-      "loss": 1.4569,
+      "loss": 1.6501,
       "step": 287
     },
     {
       "epoch": 5.24,
       "learning_rate": 0.00028399999999999996,
-      "loss": 1.609,
+      "loss": 1.7553,
       "step": 288
     },
     {
       "epoch": 5.25,
       "learning_rate": 0.000285,
-      "loss": 1.524,
+      "loss": 1.5803,
       "step": 289
     },
     {
       "epoch": 5.27,
       "learning_rate": 0.00028599999999999996,
-      "loss": 1.4046,
+      "loss": 1.5767,
       "step": 290
     },
     {
       "epoch": 5.29,
       "learning_rate": 0.000287,
-      "loss": 1.4208,
+      "loss": 1.5539,
       "step": 291
     },
     {
       "epoch": 5.31,
       "learning_rate": 0.000288,
-      "loss": 1.399,
+      "loss": 1.5134,
       "step": 292
     },
     {
       "epoch": 5.33,
       "learning_rate": 0.000289,
-      "loss": 1.4494,
+      "loss": 1.5507,
       "step": 293
     },
     {
       "epoch": 5.35,
       "learning_rate": 0.00029,
-      "loss": 1.4386,
+      "loss": 1.5343,
       "step": 294
     },
     {
       "epoch": 5.36,
       "learning_rate": 0.00029099999999999997,
-      "loss": 1.3206,
+      "loss": 1.4206,
       "step": 295
     },
     {
       "epoch": 5.38,
       "learning_rate": 0.000292,
-      "loss": 1.3173,
+      "loss": 1.4263,
       "step": 296
     },
     {
       "epoch": 5.4,
       "learning_rate": 0.00029299999999999997,
-      "loss": 1.3283,
+      "loss": 1.4438,
       "step": 297
     },
     {
       "epoch": 5.42,
       "learning_rate": 0.000294,
-      "loss": 1.2425,
+      "loss": 1.3487,
       "step": 298
     },
     {
       "epoch": 5.44,
       "learning_rate": 0.000295,
-      "loss": 1.2792,
+      "loss": 1.3797,
       "step": 299
     },
     {
       "epoch": 5.45,
       "learning_rate": 0.000296,
-      "loss": 1.3056,
+      "loss": 1.407,
       "step": 300
     },
     {
       "epoch": 5.47,
       "learning_rate": 0.000297,
-      "loss": 1.3157,
+      "loss": 1.4154,
       "step": 301
     },
     {
       "epoch": 5.49,
       "learning_rate": 0.000298,
-      "loss": 1.2396,
+      "loss": 1.3328,
       "step": 302
     },
     {
       "epoch": 5.51,
       "learning_rate": 0.000299,
-      "loss": 1.2202,
+      "loss": 1.2731,
       "step": 303
     },
     {
       "epoch": 5.53,
       "learning_rate": 0.0003,
-      "loss": 1.2631,
+      "loss": 1.2848,
       "step": 304
     },
     {
       "epoch": 5.55,
       "learning_rate": 0.000301,
-      "loss": 1.185,
+      "loss": 1.2333,
       "step": 305
     },
     {
       "epoch": 5.56,
       "learning_rate": 0.000302,
-      "loss": 1.2753,
+      "loss": 1.3581,
       "step": 306
     },
     {
       "epoch": 5.58,
       "learning_rate": 0.000303,
-      "loss": 1.3373,
+      "loss": 1.3254,
       "step": 307
     },
     {
       "epoch": 5.6,
       "learning_rate": 0.000304,
-      "loss": 1.2104,
+      "loss": 1.3037,
       "step": 308
     },
     {
       "epoch": 5.62,
       "learning_rate": 0.000305,
-      "loss": 1.1768,
+      "loss": 1.2515,
       "step": 309
     },
     {
       "epoch": 5.64,
       "learning_rate": 0.000306,
-      "loss": 1.2198,
+      "loss": 1.2897,
       "step": 310
     },
     {
       "epoch": 5.65,
       "learning_rate": 0.000307,
-      "loss": 1.2675,
+      "loss": 1.3121,
       "step": 311
     },
     {
       "epoch": 5.67,
       "learning_rate": 0.000308,
-      "loss": 1.1568,
+      "loss": 1.2436,
       "step": 312
     },
     {
       "epoch": 5.69,
       "learning_rate": 0.00030900000000000003,
-      "loss": 1.1475,
+      "loss": 1.243,
       "step": 313
     },
     {
       "epoch": 5.71,
       "learning_rate": 0.00031,
-      "loss": 1.1269,
+      "loss": 1.1976,
       "step": 314
     },
     {
       "epoch": 5.73,
       "learning_rate": 0.000311,
-      "loss": 1.096,
+      "loss": 1.1662,
       "step": 315
     },
     {
       "epoch": 5.75,
       "learning_rate": 0.000312,
-      "loss": 1.1111,
+      "loss": 1.1186,
       "step": 316
     },
     {
       "epoch": 5.76,
       "learning_rate": 0.000313,
-      "loss": 1.0966,
+      "loss": 1.1656,
       "step": 317
     },
     {
       "epoch": 5.78,
       "learning_rate": 0.000314,
-      "loss": 1.0762,
+      "loss": 1.1693,
       "step": 318
     },
     {
       "epoch": 5.8,
       "learning_rate": 0.000315,
-      "loss": 1.0873,
+      "loss": 1.1409,
       "step": 319
     },
     {
       "epoch": 5.82,
       "learning_rate": 0.000316,
-      "loss": 1.1246,
+      "loss": 1.1492,
       "step": 320
     },
     {
       "epoch": 5.84,
       "learning_rate": 0.000317,
-      "loss": 1.0277,
+      "loss": 1.1108,
       "step": 321
     },
     {
       "epoch": 5.85,
       "learning_rate": 0.00031800000000000003,
-      "loss": 1.0823,
+      "loss": 1.1233,
       "step": 322
     },
     {
       "epoch": 5.87,
       "learning_rate": 0.000319,
-      "loss": 1.0894,
+      "loss": 1.1032,
       "step": 323
     },
     {
       "epoch": 5.89,
       "learning_rate": 0.00032,
-      "loss": 1.1316,
+      "loss": 1.1293,
       "step": 324
     },
     {
       "epoch": 5.91,
       "learning_rate": 0.000321,
-      "loss": 1.0886,
+      "loss": 1.1135,
       "step": 325
     },
     {
       "epoch": 5.93,
       "learning_rate": 0.000322,
-      "loss": 1.127,
+      "loss": 1.1731,
       "step": 326
     },
     {
       "epoch": 5.95,
       "learning_rate": 0.000323,
-      "loss": 1.0624,
+      "loss": 1.1077,
       "step": 327
     },
     {
       "epoch": 5.96,
       "learning_rate": 0.000324,
-      "loss": 1.0792,
+      "loss": 1.1239,
       "step": 328
     },
     {
       "epoch": 5.98,
       "learning_rate": 0.00032500000000000004,
-      "loss": 1.1297,
+      "loss": 1.1974,
       "step": 329
     },
     {
       "epoch": 6.0,
       "learning_rate": 0.000326,
-      "loss": 1.1379,
+      "loss": 1.0912,
       "step": 330
     },
     {
       "epoch": 6.02,
       "learning_rate": 0.00032700000000000003,
-      "loss": 0.9843,
+      "loss": 1.0077,
       "step": 331
     },
     {
       "epoch": 6.04,
       "learning_rate": 0.000328,
-      "loss": 0.9679,
+      "loss": 0.988,
       "step": 332
     },
     {
       "epoch": 6.05,
       "learning_rate": 0.00032900000000000003,
-      "loss": 0.9631,
+      "loss": 0.9951,
       "step": 333
     },
     {
       "epoch": 6.07,
       "learning_rate": 0.00033,
-      "loss": 0.9441,
+      "loss": 0.9454,
       "step": 334
     },
     {
       "epoch": 6.09,
       "learning_rate": 0.000331,
-      "loss": 0.9771,
+      "loss": 1.0206,
       "step": 335
     },
     {
       "epoch": 6.11,
       "learning_rate": 0.00033200000000000005,
-      "loss": 0.9051,
+      "loss": 0.9259,
       "step": 336
     },
     {
       "epoch": 6.13,
       "learning_rate": 0.000333,
-      "loss": 0.9915,
+      "loss": 1.0508,
       "step": 337
     },
     {
       "epoch": 6.15,
       "learning_rate": 0.00033400000000000004,
-      "loss": 0.9406,
+      "loss": 0.9649,
       "step": 338
     },
     {
       "epoch": 6.16,
       "learning_rate": 0.000335,
-      "loss": 0.9389,
+      "loss": 0.9552,
       "step": 339
     },
     {
       "epoch": 6.18,
       "learning_rate": 0.00033600000000000004,
-      "loss": 0.9042,
+      "loss": 0.8884,
       "step": 340
     },
     {
       "epoch": 6.2,
       "learning_rate": 0.000337,
-      "loss": 0.9552,
+      "loss": 0.9564,
       "step": 341
     },
     {
       "epoch": 6.22,
       "learning_rate": 0.00033800000000000003,
-      "loss": 0.935,
+      "loss": 0.9433,
       "step": 342
     },
     {
       "epoch": 6.24,
       "learning_rate": 0.00033900000000000005,
-      "loss": 0.9283,
+      "loss": 0.9568,
       "step": 343
     },
     {
       "epoch": 6.25,
       "learning_rate": 0.00034,
-      "loss": 1.0168,
+      "loss": 1.0052,
       "step": 344
     },
     {
       "epoch": 6.27,
       "learning_rate": 0.00034100000000000005,
-      "loss": 0.9277,
+      "loss": 0.9352,
       "step": 345
     },
     {
       "epoch": 6.29,
       "learning_rate": 0.000342,
-      "loss": 0.898,
+      "loss": 0.9091,
       "step": 346
     },
     {
       "epoch": 6.31,
       "learning_rate": 0.00034300000000000004,
-      "loss": 0.8657,
+      "loss": 0.8731,
       "step": 347
     },
     {
       "epoch": 6.33,
       "learning_rate": 0.00034399999999999996,
-      "loss": 0.8812,
+      "loss": 0.9038,
       "step": 348
     },
     {
       "epoch": 6.35,
       "learning_rate": 0.000345,
-      "loss": 1.0223,
+      "loss": 1.009,
       "step": 349
     },
     {
       "epoch": 6.36,
       "learning_rate": 0.000346,
-      "loss": 0.9549,
+      "loss": 0.9972,
       "step": 350
     },
     {
       "epoch": 6.38,
       "learning_rate": 0.000347,
-      "loss": 0.8708,
+      "loss": 0.891,
       "step": 351
     },
     {
       "epoch": 6.4,
       "learning_rate": 0.000348,
-      "loss": 0.8748,
+      "loss": 0.9216,
       "step": 352
     },
     {
       "epoch": 6.42,
       "learning_rate": 0.00034899999999999997,
-      "loss": 0.8755,
+      "loss": 0.9082,
       "step": 353
     },
     {
       "epoch": 6.44,
       "learning_rate": 0.00035,
-      "loss": 0.9167,
+      "loss": 0.9258,
       "step": 354
     },
     {
       "epoch": 6.45,
       "learning_rate": 0.00035099999999999997,
-      "loss": 0.9,
+      "loss": 0.9059,
       "step": 355
     },
     {
       "epoch": 6.47,
       "learning_rate": 0.000352,
-      "loss": 0.9362,
+      "loss": 0.9364,
       "step": 356
     },
     {
       "epoch": 6.49,
       "learning_rate": 0.00035299999999999996,
-      "loss": 0.8727,
+      "loss": 0.8629,
       "step": 357
     },
     {
       "epoch": 6.51,
       "learning_rate": 0.000354,
-      "loss": 0.9506,
+      "loss": 0.9437,
       "step": 358
     },
     {
       "epoch": 6.53,
       "learning_rate": 0.000355,
-      "loss": 0.9629,
+      "loss": 0.9303,
       "step": 359
     },
     {
       "epoch": 6.55,
       "learning_rate": 0.000356,
-      "loss": 0.9842,
+      "loss": 0.9673,
       "step": 360
     },
     {
       "epoch": 6.56,
       "learning_rate": 0.000357,
-      "loss": 0.9201,
+      "loss": 0.9108,
       "step": 361
     },
     {
       "epoch": 6.58,
       "learning_rate": 0.000358,
-      "loss": 0.9487,
+      "loss": 0.96,
       "step": 362
     },
     {
       "epoch": 6.6,
       "learning_rate": 0.000359,
-      "loss": 0.905,
+      "loss": 0.9282,
       "step": 363
     },
     {
       "epoch": 6.62,
       "learning_rate": 0.00035999999999999997,
-      "loss": 0.9275,
+      "loss": 0.9115,
       "step": 364
     },
     {
       "epoch": 6.64,
       "learning_rate": 0.000361,
-      "loss": 0.8715,
+      "loss": 0.9303,
       "step": 365
     },
     {
       "epoch": 6.65,
       "learning_rate": 0.000362,
-      "loss": 0.9092,
+      "loss": 0.9551,
       "step": 366
     },
     {
       "epoch": 6.67,
       "learning_rate": 0.000363,
-      "loss": 0.9046,
+      "loss": 0.8899,
       "step": 367
     },
     {
       "epoch": 6.69,
       "learning_rate": 0.000364,
-      "loss": 0.8449,
+      "loss": 0.888,
       "step": 368
     },
     {
       "epoch": 6.71,
       "learning_rate": 0.000365,
-      "loss": 0.8786,
+      "loss": 0.9022,
       "step": 369
     },
     {
       "epoch": 6.73,
       "learning_rate": 0.000366,
-      "loss": 0.8749,
+      "loss": 0.8779,
       "step": 370
     },
     {
       "epoch": 6.75,
       "learning_rate": 0.000367,
-      "loss": 0.7888,
+      "loss": 0.8326,
       "step": 371
     },
     {
       "epoch": 6.76,
       "learning_rate": 0.000368,
-      "loss": 0.9058,
+      "loss": 0.9408,
       "step": 372
     },
     {
       "epoch": 6.78,
       "learning_rate": 0.000369,
-      "loss": 0.9306,
+      "loss": 0.952,
       "step": 373
     },
     {
       "epoch": 6.8,
       "learning_rate": 0.00037,
-      "loss": 0.8398,
+      "loss": 0.9119,
       "step": 374
     },
     {
       "epoch": 6.82,
       "learning_rate": 0.000371,
-      "loss": 0.9304,
+      "loss": 0.9009,
       "step": 375
     },
     {
       "epoch": 6.84,
       "learning_rate": 0.000372,
-      "loss": 0.8315,
+      "loss": 0.8858,
       "step": 376
     },
     {
       "epoch": 6.85,
       "learning_rate": 0.000373,
-      "loss": 0.8105,
+      "loss": 0.8048,
       "step": 377
     },
     {
       "epoch": 6.87,
       "learning_rate": 0.000374,
-      "loss": 0.913,
+      "loss": 0.8992,
       "step": 378
     },
     {
       "epoch": 6.89,
       "learning_rate": 0.000375,
-      "loss": 0.872,
+      "loss": 0.9188,
       "step": 379
     },
     {
       "epoch": 6.91,
       "learning_rate": 0.00037600000000000003,
-      "loss": 0.9358,
+      "loss": 0.9319,
       "step": 380
     },
     {
       "epoch": 6.93,
       "learning_rate": 0.000377,
-      "loss": 0.9119,
+      "loss": 0.9227,
       "step": 381
     },
     {
       "epoch": 6.95,
       "learning_rate": 0.000378,
-      "loss": 0.8272,
+      "loss": 0.8448,
       "step": 382
     },
     {
       "epoch": 6.96,
       "learning_rate": 0.000379,
-      "loss": 0.8889,
+      "loss": 0.9143,
       "step": 383
     },
     {
       "epoch": 6.98,
       "learning_rate": 0.00038,
-      "loss": 0.8541,
+      "loss": 0.9387,
       "step": 384
     },
     {
       "epoch": 7.0,
       "learning_rate": 0.000381,
-      "loss": 0.8628,
+      "loss": 0.7939,
       "step": 385
     },
     {
       "epoch": 7.02,
       "learning_rate": 0.000382,
-      "loss": 0.7676,
+      "loss": 0.7773,
       "step": 386
     },
     {
       "epoch": 7.04,
       "learning_rate": 0.00038300000000000004,
-      "loss": 0.7832,
+      "loss": 0.8103,
       "step": 387
     },
     {
       "epoch": 7.05,
       "learning_rate": 0.000384,
-      "loss": 0.7545,
+      "loss": 0.7278,
       "step": 388
     },
     {
       "epoch": 7.07,
       "learning_rate": 0.00038500000000000003,
-      "loss": 0.7516,
+      "loss": 0.7585,
       "step": 389
     },
     {
       "epoch": 7.09,
       "learning_rate": 0.000386,
-      "loss": 0.7653,
+      "loss": 0.7952,
       "step": 390
     },
     {
       "epoch": 7.11,
       "learning_rate": 0.00038700000000000003,
-      "loss": 0.7342,
+      "loss": 0.7367,
       "step": 391
     },
     {
       "epoch": 7.13,
       "learning_rate": 0.000388,
-      "loss": 0.8099,
+      "loss": 0.8654,
       "step": 392
     },
     {
       "epoch": 7.15,
       "learning_rate": 0.000389,
-      "loss": 0.7758,
+      "loss": 0.807,
       "step": 393
     },
     {
       "epoch": 7.16,
       "learning_rate": 0.00039000000000000005,
-      "loss": 0.765,
+      "loss": 0.7647,
       "step": 394
     },
     {
       "epoch": 7.18,
       "learning_rate": 0.000391,
-      "loss": 0.7228,
+      "loss": 0.7434,
       "step": 395
     },
     {
       "epoch": 7.2,
       "learning_rate": 0.00039200000000000004,
-      "loss": 0.7156,
+      "loss": 0.7349,
       "step": 396
     },
     {
       "epoch": 7.22,
       "learning_rate": 0.000393,
-      "loss": 0.7644,
+      "loss": 0.7495,
       "step": 397
     },
     {
       "epoch": 7.24,
       "learning_rate": 0.00039400000000000004,
-      "loss": 0.8986,
+      "loss": 0.9258,
       "step": 398
     },
     {
       "epoch": 7.25,
       "learning_rate": 0.000395,
-      "loss": 0.7974,
+      "loss": 0.8328,
       "step": 399
     },
     {
       "epoch": 7.27,
       "learning_rate": 0.00039600000000000003,
-      "loss": 0.7186,
+      "loss": 0.7282,
       "step": 400
     },
     {
       "epoch": 7.29,
       "learning_rate": 0.00039700000000000005,
-      "loss": 0.8556,
+      "loss": 0.8341,
       "step": 401
     },
     {
       "epoch": 7.31,
       "learning_rate": 0.000398,
-      "loss": 0.8353,
+      "loss": 0.8429,
       "step": 402
     },
     {
       "epoch": 7.33,
       "learning_rate": 0.00039900000000000005,
-      "loss": 0.8063,
+      "loss": 0.8084,
       "step": 403
     },
     {
       "epoch": 7.35,
       "learning_rate": 0.0004,
-      "loss": 0.7679,
+      "loss": 0.7493,
       "step": 404
     },
     {
       "epoch": 7.36,
       "learning_rate": 0.00040100000000000004,
-      "loss": 0.8583,
+      "loss": 0.8158,
       "step": 405
     },
     {
       "epoch": 7.38,
       "learning_rate": 0.000402,
-      "loss": 0.703,
+      "loss": 0.7475,
       "step": 406
     },
     {
       "epoch": 7.4,
       "learning_rate": 0.00040300000000000004,
-      "loss": 0.7997,
+      "loss": 0.7616,
       "step": 407
     },
     {
       "epoch": 7.42,
       "learning_rate": 0.000404,
-      "loss": 0.7391,
+      "loss": 0.7755,
       "step": 408
     },
     {
       "epoch": 7.44,
       "learning_rate": 0.00040500000000000003,
-      "loss": 0.819,
+      "loss": 0.8576,
       "step": 409
     },
     {
       "epoch": 7.45,
       "learning_rate": 0.00040600000000000006,
-      "loss": 0.8636,
+      "loss": 0.8022,
       "step": 410
     },
     {
       "epoch": 7.47,
       "learning_rate": 0.00040699999999999997,
-      "loss": 0.878,
+      "loss": 0.8344,
       "step": 411
     },
     {
       "epoch": 7.49,
       "learning_rate": 0.000408,
-      "loss": 0.7885,
+      "loss": 0.8271,
       "step": 412
     },
     {
       "epoch": 7.51,
       "learning_rate": 0.00040899999999999997,
-      "loss": 0.7586,
+      "loss": 0.8034,
       "step": 413
     },
     {
       "epoch": 7.53,
       "learning_rate": 0.00041,
-      "loss": 0.835,
+      "loss": 0.8157,
       "step": 414
     },
     {
       "epoch": 7.55,
       "learning_rate": 0.00041099999999999996,
-      "loss": 0.7157,
+      "loss": 0.7078,
       "step": 415
     },
     {
       "epoch": 7.56,
       "learning_rate": 0.000412,
-      "loss": 0.8006,
+      "loss": 0.846,
       "step": 416
     },
     {
       "epoch": 7.58,
       "learning_rate": 0.000413,
-      "loss": 0.7423,
+      "loss": 0.7693,
       "step": 417
     },
     {
       "epoch": 7.6,
       "learning_rate": 0.000414,
-      "loss": 0.8513,
+      "loss": 0.8231,
       "step": 418
     },
     {
       "epoch": 7.62,
       "learning_rate": 0.000415,
-      "loss": 0.7697,
+      "loss": 0.8055,
       "step": 419
     },
     {
       "epoch": 7.64,
       "learning_rate": 0.000416,
-      "loss": 0.7427,
+      "loss": 0.7489,
       "step": 420
     },
     {
       "epoch": 7.65,
       "learning_rate": 0.000417,
-      "loss": 0.7706,
+      "loss": 0.777,
       "step": 421
     },
     {
       "epoch": 7.67,
       "learning_rate": 0.00041799999999999997,
-      "loss": 0.8148,
+      "loss": 0.85,
       "step": 422
     },
     {
       "epoch": 7.69,
       "learning_rate": 0.000419,
-      "loss": 0.7451,
+      "loss": 0.798,
       "step": 423
     },
     {
       "epoch": 7.71,
       "learning_rate": 0.00042,
-      "loss": 0.7464,
+      "loss": 0.7454,
       "step": 424
     },
     {
       "epoch": 7.73,
       "learning_rate": 0.000421,
-      "loss": 0.6747,
+      "loss": 0.6989,
       "step": 425
     },
     {
       "epoch": 7.75,
       "learning_rate": 0.000422,
-      "loss": 0.7873,
+      "loss": 0.8041,
       "step": 426
     },
     {
       "epoch": 7.76,
       "learning_rate": 0.000423,
-      "loss": 0.7332,
+      "loss": 0.7557,
       "step": 427
     },
     {
       "epoch": 7.78,
       "learning_rate": 0.000424,
-      "loss": 0.798,
+      "loss": 0.8455,
       "step": 428
     },
     {
       "epoch": 7.8,
       "learning_rate": 0.000425,
-      "loss": 0.7679,
+      "loss": 0.805,
       "step": 429
     },
     {
       "epoch": 7.82,
       "learning_rate": 0.000426,
-      "loss": 0.7798,
+      "loss": 0.8144,
       "step": 430
     },
     {
       "epoch": 7.84,
       "learning_rate": 0.000427,
-      "loss": 0.721,
+      "loss": 0.7609,
       "step": 431
     },
     {
       "epoch": 7.85,
       "learning_rate": 0.000428,
-      "loss": 0.6922,
+      "loss": 0.7507,
       "step": 432
     },
     {
       "epoch": 7.87,
       "learning_rate": 0.000429,
-      "loss": 0.7274,
+      "loss": 0.7703,
       "step": 433
     },
     {
       "epoch": 7.89,
       "learning_rate": 0.00043,
-      "loss": 0.8091,
+      "loss": 0.8438,
       "step": 434
     },
     {
       "epoch": 7.91,
       "learning_rate": 0.000431,
-      "loss": 0.7234,
+      "loss": 0.7396,
       "step": 435
     },
     {
       "epoch": 7.93,
       "learning_rate": 0.000432,
-      "loss": 0.7867,
+      "loss": 0.8496,
       "step": 436
     },
     {
       "epoch": 7.95,
       "learning_rate": 0.000433,
-      "loss": 0.7866,
+      "loss": 0.7814,
       "step": 437
     },
     {
       "epoch": 7.96,
       "learning_rate": 0.00043400000000000003,
-      "loss": 0.7933,
+      "loss": 0.8737,
       "step": 438
     },
     {
       "epoch": 7.98,
       "learning_rate": 0.000435,
-      "loss": 0.7825,
+      "loss": 0.8889,
       "step": 439
     },
     {
       "epoch": 8.0,
       "learning_rate": 0.000436,
-      "loss": 0.9476,
+      "loss": 0.8235,
       "step": 440
     },
     {
       "epoch": 8.02,
       "learning_rate": 0.000437,
-      "loss": 0.7144,
+      "loss": 0.7383,
       "step": 441
     },
     {
       "epoch": 8.04,
       "learning_rate": 0.000438,
-      "loss": 0.7117,
+      "loss": 0.7883,
       "step": 442
     },
     {
       "epoch": 8.05,
       "learning_rate": 0.000439,
-      "loss": 0.6701,
+      "loss": 0.6692,
       "step": 443
     },
     {
       "epoch": 8.07,
       "learning_rate": 0.00044,
-      "loss": 0.652,
+      "loss": 0.7105,
       "step": 444
     },
     {
       "epoch": 8.09,
       "learning_rate": 0.000441,
-      "loss": 0.7025,
+      "loss": 0.7386,
       "step": 445
     },
     {
       "epoch": 8.11,
       "learning_rate": 0.000442,
-      "loss": 0.674,
+      "loss": 0.7373,
       "step": 446
     },
     {
       "epoch": 8.13,
       "learning_rate": 0.00044300000000000003,
-      "loss": 0.7302,
+      "loss": 0.7563,
       "step": 447
     },
     {
       "epoch": 8.15,
       "learning_rate": 0.000444,
-      "loss": 0.606,
+      "loss": 0.6812,
       "step": 448
     },
     {
       "epoch": 8.16,
       "learning_rate": 0.00044500000000000003,
-      "loss": 0.6163,
+      "loss": 0.6341,
       "step": 449
     },
     {
       "epoch": 8.18,
       "learning_rate": 0.000446,
-      "loss": 0.8106,
+      "loss": 0.8025,
       "step": 450
     },
     {
       "epoch": 8.2,
       "learning_rate": 0.000447,
-      "loss": 0.6955,
+      "loss": 0.7297,
       "step": 451
     },
     {
       "epoch": 8.22,
       "learning_rate": 0.000448,
-      "loss": 0.7647,
+      "loss": 0.8337,
       "step": 452
     },
     {
       "epoch": 8.24,
       "learning_rate": 0.000449,
-      "loss": 0.673,
+      "loss": 0.6924,
       "step": 453
     },
     {
       "epoch": 8.25,
       "learning_rate": 0.00045000000000000004,
-      "loss": 0.6991,
+      "loss": 0.705,
       "step": 454
     },
     {
       "epoch": 8.27,
       "learning_rate": 0.000451,
-      "loss": 0.6467,
+      "loss": 0.6914,
       "step": 455
     },
     {
       "epoch": 8.29,
       "learning_rate": 0.00045200000000000004,
-      "loss": 0.6427,
+      "loss": 0.6697,
       "step": 456
     },
     {
       "epoch": 8.31,
       "learning_rate": 0.000453,
-      "loss": 0.6386,
+      "loss": 0.6377,
       "step": 457
     },
     {
       "epoch": 8.33,
       "learning_rate": 0.00045400000000000003,
-      "loss": 0.6997,
+      "loss": 0.7195,
       "step": 458
     },
     {
       "epoch": 8.35,
       "learning_rate": 0.000455,
-      "loss": 0.702,
+      "loss": 0.6524,
       "step": 459
     },
     {
       "epoch": 8.36,
       "learning_rate": 0.000456,
-      "loss": 0.7721,
+      "loss": 0.7129,
       "step": 460
     },
     {
       "epoch": 8.38,
       "learning_rate": 0.00045700000000000005,
-      "loss": 0.6717,
+      "loss": 0.6511,
       "step": 461
     },
     {
       "epoch": 8.4,
       "learning_rate": 0.000458,
-      "loss": 0.7151,
+      "loss": 0.6281,
       "step": 462
     },
     {
       "epoch": 8.42,
       "learning_rate": 0.00045900000000000004,
-      "loss": 0.7974,
+      "loss": 0.7598,
       "step": 463
     },
     {
       "epoch": 8.44,
       "learning_rate": 0.00046,
-      "loss": 0.6968,
+      "loss": 0.6855,
       "step": 464
     },
     {
       "epoch": 8.45,
       "learning_rate": 0.00046100000000000004,
-      "loss": 0.7488,
+      "loss": 0.7156,
       "step": 465
     },
     {
       "epoch": 8.47,
       "learning_rate": 0.000462,
-      "loss": 0.6855,
+      "loss": 0.6734,
       "step": 466
     },
     {
       "epoch": 8.49,
       "learning_rate": 0.00046300000000000003,
-      "loss": 0.65,
+      "loss": 0.6415,
       "step": 467
     },
     {
       "epoch": 8.51,
       "learning_rate": 0.00046400000000000006,
-      "loss": 0.6773,
+      "loss": 0.7233,
       "step": 468
     },
     {
       "epoch": 8.53,
       "learning_rate": 0.000465,
-      "loss": 0.7257,
+      "loss": 0.6445,
       "step": 469
     },
     {
       "epoch": 8.55,
       "learning_rate": 0.00046600000000000005,
-      "loss": 0.6388,
+      "loss": 0.6717,
       "step": 470
     },
     {
       "epoch": 8.56,
       "learning_rate": 0.000467,
-      "loss": 0.6525,
+      "loss": 0.6664,
       "step": 471
     },
     {
       "epoch": 8.58,
       "learning_rate": 0.00046800000000000005,
-      "loss": 0.7266,
+      "loss": 0.7695,
       "step": 472
     },
     {
       "epoch": 8.6,
       "learning_rate": 0.00046899999999999996,
-      "loss": 0.6672,
+      "loss": 0.6321,
       "step": 473
     },
     {
       "epoch": 8.62,
       "learning_rate": 0.00047,
-      "loss": 0.6363,
+      "loss": 0.6171,
       "step": 474
     },
     {
       "epoch": 8.64,
       "learning_rate": 0.000471,
-      "loss": 0.6711,
+      "loss": 0.6478,
       "step": 475
     },
     {
       "epoch": 8.65,
       "learning_rate": 0.000472,
-      "loss": 0.7273,
+      "loss": 0.6986,
       "step": 476
     },
     {
       "epoch": 8.67,
       "learning_rate": 0.000473,
-      "loss": 0.6637,
+      "loss": 0.6651,
       "step": 477
     },
     {
       "epoch": 8.69,
       "learning_rate": 0.000474,
-      "loss": 0.7166,
+      "loss": 0.7636,
       "step": 478
     },
     {
       "epoch": 8.71,
       "learning_rate": 0.000475,
-      "loss": 0.7003,
+      "loss": 0.6767,
       "step": 479
     },
     {
       "epoch": 8.73,
       "learning_rate": 0.00047599999999999997,
-      "loss": 0.6248,
+      "loss": 0.6398,
       "step": 480
     },
     {
       "epoch": 8.75,
       "learning_rate": 0.000477,
-      "loss": 0.7101,
+      "loss": 0.7357,
       "step": 481
     },
     {
       "epoch": 8.76,
       "learning_rate": 0.00047799999999999996,
-      "loss": 0.7664,
+      "loss": 0.7647,
       "step": 482
     },
     {
       "epoch": 8.78,
       "learning_rate": 0.000479,
-      "loss": 0.7346,
+      "loss": 0.7029,
       "step": 483
     },
     {
       "epoch": 8.8,
       "learning_rate": 0.00048,
-      "loss": 0.6917,
+      "loss": 0.7036,
       "step": 484
     },
     {
       "epoch": 8.82,
       "learning_rate": 0.000481,
-      "loss": 0.6965,
+      "loss": 0.6937,
       "step": 485
     },
     {
       "epoch": 8.84,
       "learning_rate": 0.000482,
-      "loss": 0.6306,
+      "loss": 0.5841,
       "step": 486
     },
     {
       "epoch": 8.85,
       "learning_rate": 0.000483,
-      "loss": 0.6554,
+      "loss": 0.6718,
       "step": 487
     },
     {
       "epoch": 8.87,
       "learning_rate": 0.000484,
-      "loss": 0.6587,
+      "loss": 0.6661,
       "step": 488
     },
     {
       "epoch": 8.89,
       "learning_rate": 0.00048499999999999997,
-      "loss": 0.7198,
+      "loss": 0.7289,
       "step": 489
     },
     {
       "epoch": 8.91,
       "learning_rate": 0.000486,
-      "loss": 0.6692,
+      "loss": 0.6756,
       "step": 490
     },
     {
       "epoch": 8.93,
       "learning_rate": 0.000487,
-      "loss": 0.6517,
+      "loss": 0.6873,
       "step": 491
     },
     {
       "epoch": 8.95,
       "learning_rate": 0.000488,
-      "loss": 0.6776,
+      "loss": 0.7093,
       "step": 492
     },
     {
       "epoch": 8.96,
       "learning_rate": 0.000489,
-      "loss": 0.6747,
+      "loss": 0.697,
       "step": 493
     },
     {
       "epoch": 8.98,
       "learning_rate": 0.00049,
-      "loss": 0.6023,
+      "loss": 0.6658,
       "step": 494
     },
     {
       "epoch": 9.0,
       "learning_rate": 0.000491,
-      "loss": 0.6629,
+      "loss": 0.6464,
       "step": 495
     },
     {
       "epoch": 9.02,
       "learning_rate": 0.000492,
-      "loss": 0.644,
+      "loss": 0.6988,
       "step": 496
     },
     {
       "epoch": 9.04,
       "learning_rate": 0.0004930000000000001,
-      "loss": 0.598,
+      "loss": 0.6106,
       "step": 497
     },
     {
       "epoch": 9.05,
       "learning_rate": 0.000494,
-      "loss": 0.5491,
+      "loss": 0.5882,
       "step": 498
     },
     {
       "epoch": 9.07,
       "learning_rate": 0.000495,
-      "loss": 0.6013,
+      "loss": 0.6346,
       "step": 499
     },
     {
       "epoch": 9.09,
       "learning_rate": 0.000496,
-      "loss": 0.6379,
+      "loss": 0.6186,
       "step": 500
     },
     {
       "epoch": 9.09,
-      "eval_cer": 0.1326106838183481,
-      "eval_loss": 0.44991427659988403,
-      "eval_runtime": 14.3712,
-      "eval_samples_per_second": 114.604,
-      "eval_steps_per_second": 1.809,
-      "eval_wer": 0.5322234705341641,
+      "eval_cer": 0.14061680085158698,
+      "eval_loss": 0.4850134551525116,
+      "eval_runtime": 14.1748,
+      "eval_samples_per_second": 116.192,
+      "eval_steps_per_second": 1.834,
+      "eval_wer": 0.5648044122152998,
       "step": 500
     },
     {
       "epoch": 9.11,
       "learning_rate": 0.000497,
-      "loss": 0.669,
+      "loss": 0.6565,
       "step": 501
     },
     {
       "epoch": 9.13,
       "learning_rate": 0.000498,
-      "loss": 0.7005,
+      "loss": 0.7209,
       "step": 502
     },
     {
       "epoch": 9.15,
       "learning_rate": 0.000499,
-      "loss": 0.5879,
+      "loss": 0.6171,
       "step": 503
     },
     {
       "epoch": 9.16,
       "learning_rate": 0.0005,
-      "loss": 0.6639,
+      "loss": 0.6914,
       "step": 504
     },
     {
       "epoch": 9.18,
       "learning_rate": 0.0004999000000000001,
-      "loss": 0.6349,
+      "loss": 0.6711,
       "step": 505
     },
     {
       "epoch": 9.2,
       "learning_rate": 0.0004998,
-      "loss": 0.6,
+      "loss": 0.5971,
       "step": 506
     },
     {
       "epoch": 9.22,
       "learning_rate": 0.0004997,
-      "loss": 0.6223,
+      "loss": 0.6016,
       "step": 507
     },
     {
       "epoch": 9.24,
       "learning_rate": 0.0004996,
-      "loss": 0.6347,
+      "loss": 0.6493,
       "step": 508
     },
     {
       "epoch": 9.25,
       "learning_rate": 0.0004995,
-      "loss": 0.5993,
+      "loss": 0.648,
       "step": 509
     },
     {
       "epoch": 9.27,
       "learning_rate": 0.0004994,
-      "loss": 0.5989,
+      "loss": 0.5974,
       "step": 510
     },
     {
       "epoch": 9.29,
       "learning_rate": 0.0004993,
-      "loss": 0.7038,
+      "loss": 0.7255,
       "step": 511
     },
     {
       "epoch": 9.31,
       "learning_rate": 0.0004992,
-      "loss": 0.5757,
+      "loss": 0.6515,
       "step": 512
     },
     {
       "epoch": 9.33,
       "learning_rate": 0.0004991,
-      "loss": 0.6195,
+      "loss": 0.696,
       "step": 513
     },
     {
       "epoch": 9.35,
       "learning_rate": 0.000499,
-      "loss": 0.6632,
+      "loss": 0.6686,
       "step": 514
     },
     {
       "epoch": 9.36,
       "learning_rate": 0.0004989,
-      "loss": 0.645,
+      "loss": 0.6267,
       "step": 515
     },
     {
       "epoch": 9.38,
       "learning_rate": 0.0004988,
-      "loss": 0.5848,
+      "loss": 0.5963,
       "step": 516
     },
     {
       "epoch": 9.4,
       "learning_rate": 0.0004987,
-      "loss": 0.6648,
+      "loss": 0.6507,
       "step": 517
     },
     {
       "epoch": 9.42,
       "learning_rate": 0.0004986,
-      "loss": 0.6707,
+      "loss": 0.679,
       "step": 518
     },
     {
       "epoch": 9.44,
       "learning_rate": 0.0004985,
-      "loss": 0.6364,
+      "loss": 0.6961,
       "step": 519
     },
     {
       "epoch": 9.45,
       "learning_rate": 0.0004984,
-      "loss": 0.6663,
+      "loss": 0.7018,
       "step": 520
     },
     {
       "epoch": 9.47,
       "learning_rate": 0.0004983,
-      "loss": 0.5879,
+      "loss": 0.636,
       "step": 521
     },
     {
       "epoch": 9.49,
       "learning_rate": 0.0004982,
-      "loss": 0.542,
+      "loss": 0.5408,
       "step": 522
     },
     {
       "epoch": 9.51,
       "learning_rate": 0.0004981,
-      "loss": 0.6116,
+      "loss": 0.617,
       "step": 523
     },
     {
       "epoch": 9.53,
       "learning_rate": 0.000498,
-      "loss": 0.595,
+      "loss": 0.6706,
       "step": 524
     },
     {
       "epoch": 9.55,
       "learning_rate": 0.0004979,
-      "loss": 0.5651,
+      "loss": 0.6,
       "step": 525
     },
     {
       "epoch": 9.56,
       "learning_rate": 0.0004978000000000001,
-      "loss": 0.5499,
+      "loss": 0.6015,
       "step": 526
     },
     {
       "epoch": 9.58,
       "learning_rate": 0.0004977,
-      "loss": 0.6309,
+      "loss": 0.6721,
       "step": 527
     },
     {
       "epoch": 9.6,
       "learning_rate": 0.0004976,
-      "loss": 0.6289,
+      "loss": 0.676,
       "step": 528
     },
     {
       "epoch": 9.62,
       "learning_rate": 0.0004975,
-      "loss": 0.6147,
+      "loss": 0.6183,
       "step": 529
     },
     {
       "epoch": 9.64,
       "learning_rate": 0.0004974000000000001,
-      "loss": 0.498,
+      "loss": 0.4935,
       "step": 530
     },
     {
       "epoch": 9.65,
       "learning_rate": 0.0004973,
-      "loss": 0.6418,
+      "loss": 0.6566,
       "step": 531
     },
     {
       "epoch": 9.67,
       "learning_rate": 0.0004971999999999999,
-      "loss": 0.5467,
+      "loss": 0.5903,
       "step": 532
     },
     {
       "epoch": 9.69,
       "learning_rate": 0.0004971,
-      "loss": 0.6759,
+      "loss": 0.6101,
       "step": 533
     },
     {
       "epoch": 9.71,
       "learning_rate": 0.000497,
-      "loss": 0.6149,
+      "loss": 0.6384,
       "step": 534
     },
     {
       "epoch": 9.73,
       "learning_rate": 0.0004969,
-      "loss": 0.6703,
+      "loss": 0.6865,
       "step": 535
     },
     {
       "epoch": 9.75,
       "learning_rate": 0.0004968,
-      "loss": 0.6418,
+      "loss": 0.7039,
       "step": 536
     },
     {
       "epoch": 9.76,
       "learning_rate": 0.0004967,
-      "loss": 0.6284,
+      "loss": 0.6155,
       "step": 537
     },
     {
       "epoch": 9.78,
       "learning_rate": 0.0004966,
-      "loss": 0.5858,
+      "loss": 0.6206,
       "step": 538
     },
     {
       "epoch": 9.8,
       "learning_rate": 0.0004965,
-      "loss": 0.6288,
+      "loss": 0.6284,
       "step": 539
     },
     {
       "epoch": 9.82,
       "learning_rate": 0.0004964,
-      "loss": 0.6389,
+      "loss": 0.6212,
       "step": 540
     },
     {
       "epoch": 9.84,
       "learning_rate": 0.0004963000000000001,
-      "loss": 0.6295,
+      "loss": 0.6078,
       "step": 541
     },
     {
       "epoch": 9.85,
       "learning_rate": 0.0004962,
-      "loss": 0.61,
+      "loss": 0.6361,
       "step": 542
     },
     {
       "epoch": 9.87,
       "learning_rate": 0.0004961,
-      "loss": 0.6137,
+      "loss": 0.6112,
       "step": 543
     },
     {
       "epoch": 9.89,
       "learning_rate": 0.000496,
-      "loss": 0.6743,
+      "loss": 0.6614,
       "step": 544
     },
     {
       "epoch": 9.91,
       "learning_rate": 0.0004959,
-      "loss": 0.5383,
+      "loss": 0.5387,
       "step": 545
     },
     {
       "epoch": 9.93,
       "learning_rate": 0.0004958,
-      "loss": 0.5885,
+      "loss": 0.6218,
       "step": 546
     },
     {
       "epoch": 9.95,
       "learning_rate": 0.0004957,
-      "loss": 0.5691,
+      "loss": 0.5706,
       "step": 547
     },
     {
       "epoch": 9.96,
       "learning_rate": 0.0004956,
-      "loss": 0.6694,
+      "loss": 0.6294,
       "step": 548
     },
     {
       "epoch": 9.98,
       "learning_rate": 0.0004955,
-      "loss": 0.6446,
+      "loss": 0.6553,
       "step": 549
     },
     {
       "epoch": 10.0,
       "learning_rate": 0.0004954,
-      "loss": 0.6694,
+      "loss": 0.63,
       "step": 550
     },
     {
       "epoch": 10.02,
       "learning_rate": 0.0004953000000000001,
-      "loss": 0.6288,
+      "loss": 0.5943,
       "step": 551
     },
     {
       "epoch": 10.04,
       "learning_rate": 0.0004952,
-      "loss": 0.5483,
+      "loss": 0.531,
       "step": 552
     },
     {
       "epoch": 10.05,
       "learning_rate": 0.0004950999999999999,
-      "loss": 0.544,
+      "loss": 0.4986,
       "step": 553
     },
     {
       "epoch": 10.07,
       "learning_rate": 0.000495,
-      "loss": 0.507,
+      "loss": 0.5037,
       "step": 554
     },
     {
       "epoch": 10.09,
       "learning_rate": 0.0004949,
-      "loss": 0.6401,
+      "loss": 0.5576,
       "step": 555
     },
     {
       "epoch": 10.11,
       "learning_rate": 0.0004948,
-      "loss": 0.5179,
+      "loss": 0.5024,
       "step": 556
     },
     {
       "epoch": 10.13,
       "learning_rate": 0.0004946999999999999,
-      "loss": 0.5646,
+      "loss": 0.5863,
       "step": 557
     },
     {
       "epoch": 10.15,
       "learning_rate": 0.0004946,
-      "loss": 0.508,
+      "loss": 0.5547,
       "step": 558
     },
     {
       "epoch": 10.16,
       "learning_rate": 0.0004945,
-      "loss": 0.6319,
+      "loss": 0.6244,
       "step": 559
     },
     {
       "epoch": 10.18,
       "learning_rate": 0.0004944,
-      "loss": 0.5346,
+      "loss": 0.5539,
       "step": 560
     },
     {
       "epoch": 10.2,
       "learning_rate": 0.0004943,
-      "loss": 0.6215,
+      "loss": 0.5775,
       "step": 561
     },
     {
       "epoch": 10.22,
       "learning_rate": 0.0004942,
-      "loss": 0.544,
+      "loss": 0.5248,
       "step": 562
     },
     {
       "epoch": 10.24,
       "learning_rate": 0.0004941,
-      "loss": 0.5856,
+      "loss": 0.6014,
       "step": 563
     },
     {
       "epoch": 10.25,
       "learning_rate": 0.000494,
-      "loss": 0.5669,
+      "loss": 0.5424,
       "step": 564
     },
     {
       "epoch": 10.27,
       "learning_rate": 0.0004939,
-      "loss": 0.5542,
+      "loss": 0.5063,
       "step": 565
     },
     {
       "epoch": 10.29,
       "learning_rate": 0.0004938000000000001,
-      "loss": 0.5771,
+      "loss": 0.5975,
       "step": 566
     },
     {
       "epoch": 10.31,
       "learning_rate": 0.0004937,
-      "loss": 0.5502,
+      "loss": 0.5557,
       "step": 567
     },
     {
       "epoch": 10.33,
       "learning_rate": 0.0004936,
-      "loss": 0.5873,
+      "loss": 0.5836,
       "step": 568
     },
     {
       "epoch": 10.35,
       "learning_rate": 0.0004935,
-      "loss": 0.5898,
+      "loss": 0.5737,
       "step": 569
     },
     {
       "epoch": 10.36,
       "learning_rate": 0.0004934000000000001,
-      "loss": 0.5462,
+      "loss": 0.5678,
       "step": 570
     },
     {
       "epoch": 10.38,
       "learning_rate": 0.0004933,
-      "loss": 0.5782,
+      "loss": 0.5683,
       "step": 571
     },
     {
       "epoch": 10.4,
       "learning_rate": 0.0004932000000000001,
-      "loss": 0.5475,
+      "loss": 0.574,
       "step": 572
     },
     {
       "epoch": 10.42,
       "learning_rate": 0.0004931,
-      "loss": 0.5705,
+      "loss": 0.5189,
       "step": 573
     },
     {
       "epoch": 10.44,
       "learning_rate": 0.0004930000000000001,
-      "loss": 0.513,
+      "loss": 0.5244,
       "step": 574
     },
     {
       "epoch": 10.45,
       "learning_rate": 0.0004929,
-      "loss": 0.5506,
+      "loss": 0.5585,
       "step": 575
     },
     {
       "epoch": 10.47,
       "learning_rate": 0.0004928,
-      "loss": 0.6261,
+      "loss": 0.6163,
       "step": 576
     },
     {
       "epoch": 10.49,
       "learning_rate": 0.0004927,
-      "loss": 0.5692,
+      "loss": 0.5612,
       "step": 577
     },
     {
       "epoch": 10.51,
       "learning_rate": 0.0004925999999999999,
-      "loss": 0.5327,
+      "loss": 0.5495,
       "step": 578
     },
     {
       "epoch": 10.53,
       "learning_rate": 0.0004925,
-      "loss": 0.5387,
+      "loss": 0.5648,
       "step": 579
     },
     {
       "epoch": 10.55,
       "learning_rate": 0.0004924,
-      "loss": 0.5725,
+      "loss": 0.5591,
       "step": 580
     },
     {
       "epoch": 10.56,
       "learning_rate": 0.0004923,
-      "loss": 0.5402,
+      "loss": 0.5879,
       "step": 581
     },
     {
       "epoch": 10.58,
       "learning_rate": 0.0004922,
-      "loss": 0.5952,
+      "loss": 0.6037,
       "step": 582
     },
     {
       "epoch": 10.6,
       "learning_rate": 0.0004921,
-      "loss": 0.5377,
+      "loss": 0.5504,
       "step": 583
     },
     {
       "epoch": 10.62,
       "learning_rate": 0.000492,
-      "loss": 0.5528,
+      "loss": 0.5755,
       "step": 584
     },
     {
       "epoch": 10.64,
       "learning_rate": 0.0004919,
-      "loss": 0.5415,
+      "loss": 0.5572,
       "step": 585
     },
     {
       "epoch": 10.65,
       "learning_rate": 0.0004918,
-      "loss": 0.4959,
+      "loss": 0.495,
       "step": 586
     },
     {
       "epoch": 10.67,
       "learning_rate": 0.0004917000000000001,
-      "loss": 0.5722,
+      "loss": 0.5135,
       "step": 587
     },
     {
       "epoch": 10.69,
       "learning_rate": 0.0004916,
-      "loss": 0.5293,
+      "loss": 0.5386,
       "step": 588
     },
     {
       "epoch": 10.71,
       "learning_rate": 0.0004915,
-      "loss": 0.5286,
+      "loss": 0.4893,
       "step": 589
     },
     {
       "epoch": 10.73,
       "learning_rate": 0.0004914,
-      "loss": 0.5519,
+      "loss": 0.6136,
       "step": 590
     },
     {
       "epoch": 10.75,
       "learning_rate": 0.0004913000000000001,
-      "loss": 0.5613,
+      "loss": 0.5494,
       "step": 591
     },
     {
       "epoch": 10.76,
       "learning_rate": 0.0004912,
-      "loss": 0.561,
+      "loss": 0.6001,
       "step": 592
     },
     {
       "epoch": 10.78,
       "learning_rate": 0.0004911,
-      "loss": 0.5504,
+      "loss": 0.5776,
       "step": 593
     },
     {
       "epoch": 10.8,
       "learning_rate": 0.000491,
-      "loss": 0.5782,
+      "loss": 0.5916,
       "step": 594
     },
     {
       "epoch": 10.82,
       "learning_rate": 0.0004909000000000001,
-      "loss": 0.5191,
+      "loss": 0.5155,
       "step": 595
     },
     {
       "epoch": 10.84,
       "learning_rate": 0.0004908,
-      "loss": 0.4624,
+      "loss": 0.468,
       "step": 596
     },
     {
       "epoch": 10.85,
       "learning_rate": 0.0004907000000000001,
-      "loss": 0.5432,
+      "loss": 0.5504,
       "step": 597
     },
     {
       "epoch": 10.87,
       "learning_rate": 0.0004906,
-      "loss": 0.6582,
+      "loss": 0.6338,
       "step": 598
     },
     {
       "epoch": 10.89,
       "learning_rate": 0.0004905,
-      "loss": 0.5511,
+      "loss": 0.5549,
       "step": 599
     },
     {
       "epoch": 10.91,
       "learning_rate": 0.0004904,
-      "loss": 0.5949,
+      "loss": 0.6139,
       "step": 600
     },
     {
       "epoch": 10.93,
       "learning_rate": 0.0004903,
-      "loss": 0.5639,
+      "loss": 0.5674,
       "step": 601
     },
     {
       "epoch": 10.95,
       "learning_rate": 0.0004902,
-      "loss": 0.5864,
+      "loss": 0.5453,
       "step": 602
     },
     {
       "epoch": 10.96,
       "learning_rate": 0.0004901,
-      "loss": 0.5613,
+      "loss": 0.5715,
       "step": 603
     },
     {
       "epoch": 10.98,
       "learning_rate": 0.00049,
-      "loss": 0.563,
+      "loss": 0.5696,
       "step": 604
     },
     {
       "epoch": 11.0,
       "learning_rate": 0.0004899,
-      "loss": 0.496,
+      "loss": 0.4809,
       "step": 605
     },
     {
       "epoch": 11.02,
       "learning_rate": 0.0004898,
-      "loss": 0.514,
+      "loss": 0.5267,
       "step": 606
     },
     {
       "epoch": 11.04,
       "learning_rate": 0.0004897,
-      "loss": 0.5086,
+      "loss": 0.5044,
       "step": 607
     },
     {
       "epoch": 11.05,
       "learning_rate": 0.0004896,
-      "loss": 0.5547,
+      "loss": 0.5343,
       "step": 608
     },
     {
       "epoch": 11.07,
       "learning_rate": 0.0004895,
-      "loss": 0.4905,
+      "loss": 0.511,
       "step": 609
     },
     {
       "epoch": 11.09,
       "learning_rate": 0.0004894,
-      "loss": 0.5223,
+      "loss": 0.5137,
       "step": 610
     },
     {
       "epoch": 11.11,
       "learning_rate": 0.0004893,
-      "loss": 0.5418,
+      "loss": 0.5335,
       "step": 611
     },
     {
       "epoch": 11.13,
       "learning_rate": 0.0004892000000000001,
-      "loss": 0.478,
+      "loss": 0.5215,
       "step": 612
     },
     {
       "epoch": 11.15,
       "learning_rate": 0.0004891,
-      "loss": 0.5178,
+      "loss": 0.5199,
       "step": 613
     },
     {
       "epoch": 11.16,
       "learning_rate": 0.000489,
-      "loss": 0.5454,
+      "loss": 0.514,
       "step": 614
     },
     {
       "epoch": 11.18,
       "learning_rate": 0.0004889,
-      "loss": 0.5171,
+      "loss": 0.5686,
       "step": 615
     },
     {
       "epoch": 11.2,
       "learning_rate": 0.0004888000000000001,
-      "loss": 0.4091,
+      "loss": 0.4755,
       "step": 616
     },
     {
       "epoch": 11.22,
       "learning_rate": 0.0004887,
-      "loss": 0.5566,
+      "loss": 0.5259,
       "step": 617
     },
     {
       "epoch": 11.24,
       "learning_rate": 0.0004886,
-      "loss": 0.5307,
+      "loss": 0.5826,
       "step": 618
     },
     {
       "epoch": 11.25,
       "learning_rate": 0.0004885,
-      "loss": 0.5302,
+      "loss": 0.5446,
       "step": 619
     },
     {
       "epoch": 11.27,
       "learning_rate": 0.0004884,
-      "loss": 0.4643,
+      "loss": 0.4707,
       "step": 620
     },
     {
       "epoch": 11.29,
       "learning_rate": 0.0004883,
-      "loss": 0.4802,
+      "loss": 0.5207,
       "step": 621
     },
     {
       "epoch": 11.31,
       "learning_rate": 0.00048820000000000005,
-      "loss": 0.4858,
+      "loss": 0.5107,
       "step": 622
     },
     {
       "epoch": 11.33,
       "learning_rate": 0.0004881,
-      "loss": 0.485,
+      "loss": 0.5114,
       "step": 623
     },
     {
       "epoch": 11.35,
       "learning_rate": 0.000488,
-      "loss": 0.508,
+      "loss": 0.5366,
       "step": 624
     },
     {
       "epoch": 11.36,
       "learning_rate": 0.0004879,
-      "loss": 0.4802,
+      "loss": 0.5115,
       "step": 625
     },
     {
       "epoch": 11.38,
       "learning_rate": 0.00048780000000000004,
-      "loss": 0.4866,
+      "loss": 0.4901,
       "step": 626
     },
     {
       "epoch": 11.4,
       "learning_rate": 0.00048770000000000004,
-      "loss": 0.4803,
+      "loss": 0.5212,
       "step": 627
     },
     {
       "epoch": 11.42,
       "learning_rate": 0.0004876,
-      "loss": 0.5341,
+      "loss": 0.5702,
       "step": 628
     },
     {
       "epoch": 11.44,
       "learning_rate": 0.0004875,
-      "loss": 0.5524,
+      "loss": 0.542,
       "step": 629
     },
     {
       "epoch": 11.45,
       "learning_rate": 0.00048740000000000003,
-      "loss": 0.5688,
+      "loss": 0.5351,
       "step": 630
     },
     {
       "epoch": 11.47,
       "learning_rate": 0.0004873,
-      "loss": 0.5096,
+      "loss": 0.5018,
       "step": 631
     },
     {
       "epoch": 11.49,
       "learning_rate": 0.0004872,
-      "loss": 0.5204,
+      "loss": 0.5686,
       "step": 632
     },
     {
       "epoch": 11.51,
       "learning_rate": 0.00048709999999999997,
-      "loss": 0.5251,
+      "loss": 0.4944,
       "step": 633
     },
     {
       "epoch": 11.53,
       "learning_rate": 0.000487,
-      "loss": 0.4937,
+      "loss": 0.5056,
       "step": 634
     },
     {
       "epoch": 11.55,
       "learning_rate": 0.0004869,
-      "loss": 0.4988,
+      "loss": 0.5581,
       "step": 635
     },
     {
       "epoch": 11.56,
       "learning_rate": 0.0004868,
-      "loss": 0.5359,
+      "loss": 0.5188,
       "step": 636
     },
     {
       "epoch": 11.58,
       "learning_rate": 0.0004867,
-      "loss": 0.4938,
+      "loss": 0.4836,
       "step": 637
     },
     {
       "epoch": 11.6,
       "learning_rate": 0.0004866,
-      "loss": 0.5262,
+      "loss": 0.5185,
       "step": 638
     },
     {
       "epoch": 11.62,
       "learning_rate": 0.0004865,
-      "loss": 0.4674,
+      "loss": 0.4818,
       "step": 639
     },
     {
       "epoch": 11.64,
       "learning_rate": 0.0004864,
-      "loss": 0.5259,
+      "loss": 0.5939,
       "step": 640
     },
     {
       "epoch": 11.65,
       "learning_rate": 0.0004863,
-      "loss": 0.4851,
+      "loss": 0.5271,
       "step": 641
     },
     {
       "epoch": 11.67,
       "learning_rate": 0.00048620000000000005,
-      "loss": 0.5074,
+      "loss": 0.5486,
       "step": 642
     },
     {
       "epoch": 11.69,
       "learning_rate": 0.0004861,
-      "loss": 0.5729,
+      "loss": 0.5914,
       "step": 643
     },
     {
       "epoch": 11.71,
       "learning_rate": 0.000486,
-      "loss": 0.4968,
+      "loss": 0.5242,
       "step": 644
     },
     {
       "epoch": 11.73,
       "learning_rate": 0.0004859,
-      "loss": 0.434,
+      "loss": 0.4934,
       "step": 645
     },
     {
       "epoch": 11.75,
       "learning_rate": 0.00048580000000000004,
-      "loss": 0.4604,
+      "loss": 0.4685,
       "step": 646
     },
     {
       "epoch": 11.76,
       "learning_rate": 0.00048570000000000004,
-      "loss": 0.4788,
+      "loss": 0.5231,
       "step": 647
     },
     {
       "epoch": 11.78,
       "learning_rate": 0.0004856,
-      "loss": 0.4907,
+      "loss": 0.5449,
       "step": 648
     },
     {
       "epoch": 11.8,
       "learning_rate": 0.0004855,
-      "loss": 0.4719,
+      "loss": 0.4761,
       "step": 649
     },
     {
       "epoch": 11.82,
       "learning_rate": 0.00048540000000000003,
-      "loss": 0.433,
+      "loss": 0.4965,
       "step": 650
     },
     {
       "epoch": 11.84,
       "learning_rate": 0.00048530000000000003,
-      "loss": 0.4787,
+      "loss": 0.5438,
       "step": 651
     },
     {
       "epoch": 11.85,
       "learning_rate": 0.00048520000000000003,
-      "loss": 0.4799,
+      "loss": 0.5265,
       "step": 652
     },
     {
       "epoch": 11.87,
       "learning_rate": 0.0004851,
-      "loss": 0.5041,
+      "loss": 0.5324,
       "step": 653
     },
     {
       "epoch": 11.89,
       "learning_rate": 0.00048499999999999997,
-      "loss": 0.4319,
+      "loss": 0.507,
       "step": 654
     },
     {
       "epoch": 11.91,
       "learning_rate": 0.0004849,
-      "loss": 0.4815,
+      "loss": 0.488,
       "step": 655
     },
     {
       "epoch": 11.93,
       "learning_rate": 0.0004848,
-      "loss": 0.576,
+      "loss": 0.5924,
       "step": 656
     },
     {
       "epoch": 11.95,
       "learning_rate": 0.0004847,
-      "loss": 0.4478,
+      "loss": 0.4447,
       "step": 657
     },
     {
       "epoch": 11.96,
       "learning_rate": 0.00048459999999999996,
-      "loss": 0.4963,
+      "loss": 0.5597,
       "step": 658
     },
     {
       "epoch": 11.98,
       "learning_rate": 0.0004845,
-      "loss": 0.5134,
+      "loss": 0.5632,
       "step": 659
     },
     {
       "epoch": 12.0,
       "learning_rate": 0.0004844,
-      "loss": 0.6803,
+      "loss": 0.6697,
       "step": 660
     },
     {
       "epoch": 12.02,
       "learning_rate": 0.0004843,
-      "loss": 0.4257,
+      "loss": 0.4586,
       "step": 661
     },
     {
       "epoch": 12.04,
       "learning_rate": 0.0004842,
-      "loss": 0.5202,
+      "loss": 0.5939,
       "step": 662
     },
     {
       "epoch": 12.05,
       "learning_rate": 0.0004841,
-      "loss": 0.5052,
+      "loss": 0.5086,
       "step": 663
     },
     {
       "epoch": 12.07,
       "learning_rate": 0.000484,
-      "loss": 0.5119,
+      "loss": 0.5124,
       "step": 664
     },
     {
       "epoch": 12.09,
       "learning_rate": 0.0004839,
-      "loss": 0.5105,
+      "loss": 0.545,
       "step": 665
     },
     {
       "epoch": 12.11,
       "learning_rate": 0.0004838,
-      "loss": 0.5607,
+      "loss": 0.5435,
       "step": 666
     },
     {
       "epoch": 12.13,
       "learning_rate": 0.00048370000000000005,
-      "loss": 0.528,
+      "loss": 0.5455,
       "step": 667
     },
     {
       "epoch": 12.15,
       "learning_rate": 0.0004836,
-      "loss": 0.4855,
+      "loss": 0.4939,
       "step": 668
     },
     {
       "epoch": 12.16,
       "learning_rate": 0.0004835,
-      "loss": 0.443,
+      "loss": 0.4727,
       "step": 669
     },
     {
       "epoch": 12.18,
       "learning_rate": 0.0004834,
-      "loss": 0.4621,
+      "loss": 0.5372,
       "step": 670
     },
     {
       "epoch": 12.2,
       "learning_rate": 0.00048330000000000004,
-      "loss": 0.5105,
+      "loss": 0.5384,
       "step": 671
     },
     {
       "epoch": 12.22,
       "learning_rate": 0.00048320000000000004,
-      "loss": 0.469,
+      "loss": 0.4706,
       "step": 672
     },
     {
       "epoch": 12.24,
       "learning_rate": 0.0004831,
-      "loss": 0.5068,
+      "loss": 0.5244,
       "step": 673
     },
     {
       "epoch": 12.25,
       "learning_rate": 0.000483,
-      "loss": 0.4629,
+      "loss": 0.5011,
       "step": 674
     },
     {
       "epoch": 12.27,
       "learning_rate": 0.00048290000000000003,
-      "loss": 0.4486,
+      "loss": 0.464,
       "step": 675
     },
     {
       "epoch": 12.29,
       "learning_rate": 0.0004828,
-      "loss": 0.4116,
+      "loss": 0.4599,
       "step": 676
     },
     {
       "epoch": 12.31,
       "learning_rate": 0.0004827,
-      "loss": 0.5185,
+      "loss": 0.5447,
       "step": 677
     },
     {
       "epoch": 12.33,
       "learning_rate": 0.00048259999999999997,
-      "loss": 0.4477,
+      "loss": 0.4902,
       "step": 678
     },
     {
       "epoch": 12.35,
       "learning_rate": 0.0004825,
-      "loss": 0.5133,
+      "loss": 0.5715,
       "step": 679
     },
     {
       "epoch": 12.36,
       "learning_rate": 0.0004824,
-      "loss": 0.5182,
+      "loss": 0.5425,
       "step": 680
     },
     {
       "epoch": 12.38,
       "learning_rate": 0.0004823,
-      "loss": 0.4201,
+      "loss": 0.4433,
       "step": 681
     },
     {
       "epoch": 12.4,
       "learning_rate": 0.0004822,
-      "loss": 0.4785,
+      "loss": 0.4743,
       "step": 682
     },
     {
       "epoch": 12.42,
       "learning_rate": 0.0004821,
-      "loss": 0.4115,
+      "loss": 0.4508,
       "step": 683
     },
     {
       "epoch": 12.44,
       "learning_rate": 0.000482,
-      "loss": 0.482,
+      "loss": 0.5134,
       "step": 684
     },
     {
       "epoch": 12.45,
       "learning_rate": 0.0004819,
-      "loss": 0.5093,
+      "loss": 0.5254,
       "step": 685
     },
     {
       "epoch": 12.47,
       "learning_rate": 0.0004818,
-      "loss": 0.4328,
+      "loss": 0.4521,
       "step": 686
     },
     {
       "epoch": 12.49,
       "learning_rate": 0.00048170000000000005,
-      "loss": 0.4905,
+      "loss": 0.5301,
       "step": 687
     },
     {
       "epoch": 12.51,
       "learning_rate": 0.0004816,
-      "loss": 0.455,
+      "loss": 0.5277,
       "step": 688
     },
     {
       "epoch": 12.53,
       "learning_rate": 0.0004815,
-      "loss": 0.5405,
+      "loss": 0.5284,
       "step": 689
     },
     {
       "epoch": 12.55,
       "learning_rate": 0.0004814,
-      "loss": 0.4527,
+      "loss": 0.4782,
       "step": 690
     },
     {
       "epoch": 12.56,
       "learning_rate": 0.0004813,
-      "loss": 0.4278,
+      "loss": 0.478,
       "step": 691
     },
     {
       "epoch": 12.58,
       "learning_rate": 0.00048120000000000004,
-      "loss": 0.5085,
+      "loss": 0.4817,
       "step": 692
     },
     {
       "epoch": 12.6,
       "learning_rate": 0.00048110000000000004,
-      "loss": 0.4322,
+      "loss": 0.4573,
       "step": 693
     },
     {
       "epoch": 12.62,
       "learning_rate": 0.000481,
-      "loss": 0.4235,
+      "loss": 0.4363,
       "step": 694
     },
     {
       "epoch": 12.64,
       "learning_rate": 0.0004809,
-      "loss": 0.4405,
+      "loss": 0.4547,
       "step": 695
     },
     {
       "epoch": 12.65,
       "learning_rate": 0.00048080000000000003,
-      "loss": 0.4304,
+      "loss": 0.4538,
       "step": 696
     },
     {
       "epoch": 12.67,
       "learning_rate": 0.00048070000000000003,
-      "loss": 0.5083,
+      "loss": 0.5336,
       "step": 697
     },
     {
       "epoch": 12.69,
       "learning_rate": 0.0004806,
-      "loss": 0.4768,
+      "loss": 0.4692,
       "step": 698
     },
     {
       "epoch": 12.71,
       "learning_rate": 0.00048049999999999997,
-      "loss": 0.4678,
+      "loss": 0.4442,
       "step": 699
     },
     {
       "epoch": 12.73,
       "learning_rate": 0.0004804,
-      "loss": 0.4899,
+      "loss": 0.4465,
       "step": 700
     },
     {
       "epoch": 12.75,
       "learning_rate": 0.0004803,
-      "loss": 0.4548,
+      "loss": 0.466,
       "step": 701
     },
     {
       "epoch": 12.76,
       "learning_rate": 0.0004802,
-      "loss": 0.5146,
+      "loss": 0.5164,
       "step": 702
     },
     {
       "epoch": 12.78,
       "learning_rate": 0.0004801,
-      "loss": 0.5317,
+      "loss": 0.5154,
       "step": 703
     },
     {
       "epoch": 12.8,
       "learning_rate": 0.00048,
-      "loss": 0.4467,
+      "loss": 0.4651,
       "step": 704
     },
     {
       "epoch": 12.82,
       "learning_rate": 0.0004799,
-      "loss": 0.5015,
+      "loss": 0.4708,
       "step": 705
     },
     {
       "epoch": 12.84,
       "learning_rate": 0.0004798,
-      "loss": 0.4327,
+      "loss": 0.4605,
       "step": 706
     },
     {
       "epoch": 12.85,
       "learning_rate": 0.0004797,
-      "loss": 0.4404,
+      "loss": 0.4437,
       "step": 707
     },
     {
       "epoch": 12.87,
       "learning_rate": 0.00047960000000000006,
-      "loss": 0.4888,
+      "loss": 0.52,
       "step": 708
     },
     {
       "epoch": 12.89,
       "learning_rate": 0.0004795,
-      "loss": 0.4985,
+      "loss": 0.542,
       "step": 709
     },
     {
       "epoch": 12.91,
       "learning_rate": 0.0004794,
-      "loss": 0.5821,
+      "loss": 0.5538,
       "step": 710
     },
     {
       "epoch": 12.93,
       "learning_rate": 0.0004793,
-      "loss": 0.5135,
+      "loss": 0.5205,
       "step": 711
     },
     {
       "epoch": 12.95,
       "learning_rate": 0.00047920000000000005,
-      "loss": 0.4602,
+      "loss": 0.512,
       "step": 712
     },
     {
       "epoch": 12.96,
       "learning_rate": 0.00047910000000000004,
-      "loss": 0.4352,
+      "loss": 0.4464,
       "step": 713
     },
     {
       "epoch": 12.98,
       "learning_rate": 0.000479,
-      "loss": 0.4881,
+      "loss": 0.507,
       "step": 714
     },
     {
       "epoch": 13.0,
       "learning_rate": 0.0004789,
-      "loss": 0.4305,
+      "loss": 0.3977,
       "step": 715
     },
     {
       "epoch": 13.02,
       "learning_rate": 0.00047880000000000004,
-      "loss": 0.4821,
+      "loss": 0.51,
       "step": 716
     },
     {
       "epoch": 13.04,
       "learning_rate": 0.00047870000000000003,
-      "loss": 0.5015,
+      "loss": 0.4797,
       "step": 717
     },
     {
       "epoch": 13.05,
       "learning_rate": 0.00047860000000000003,
-      "loss": 0.4939,
+      "loss": 0.4516,
       "step": 718
     },
     {
       "epoch": 13.07,
       "learning_rate": 0.0004785,
-      "loss": 0.3873,
+      "loss": 0.4235,
       "step": 719
     },
     {
       "epoch": 13.09,
       "learning_rate": 0.00047840000000000003,
-      "loss": 0.4241,
+      "loss": 0.4637,
       "step": 720
     },
     {
       "epoch": 13.11,
       "learning_rate": 0.0004783,
-      "loss": 0.4635,
+      "loss": 0.4418,
       "step": 721
     },
     {
       "epoch": 13.13,
       "learning_rate": 0.0004782,
-      "loss": 0.4057,
+      "loss": 0.3978,
       "step": 722
     },
     {
       "epoch": 13.15,
       "learning_rate": 0.0004781,
-      "loss": 0.3931,
+      "loss": 0.4018,
       "step": 723
     },
     {
       "epoch": 13.16,
       "learning_rate": 0.00047799999999999996,
-      "loss": 0.3991,
+      "loss": 0.3643,
       "step": 724
     },
     {
       "epoch": 13.18,
       "learning_rate": 0.0004779,
-      "loss": 0.4133,
+      "loss": 0.4326,
       "step": 725
     },
     {
       "epoch": 13.2,
       "learning_rate": 0.0004778,
-      "loss": 0.372,
+      "loss": 0.4295,
       "step": 726
     },
     {
       "epoch": 13.22,
       "learning_rate": 0.0004777,
-      "loss": 0.5033,
+      "loss": 0.4783,
       "step": 727
     },
     {
       "epoch": 13.24,
       "learning_rate": 0.0004776,
-      "loss": 0.432,
+      "loss": 0.44,
       "step": 728
     },
     {
       "epoch": 13.25,
       "learning_rate": 0.0004775,
-      "loss": 0.4081,
+      "loss": 0.4517,
       "step": 729
     },
     {
       "epoch": 13.27,
       "learning_rate": 0.0004774,
-      "loss": 0.3948,
+      "loss": 0.4032,
       "step": 730
     },
     {
       "epoch": 13.29,
       "learning_rate": 0.0004773,
-      "loss": 0.5025,
+      "loss": 0.53,
       "step": 731
     },
     {
       "epoch": 13.31,
       "learning_rate": 0.0004772,
-      "loss": 0.4607,
+      "loss": 0.4789,
       "step": 732
     },
     {
       "epoch": 13.33,
       "learning_rate": 0.00047710000000000005,
-      "loss": 0.4459,
+      "loss": 0.4153,
       "step": 733
     },
     {
       "epoch": 13.35,
       "learning_rate": 0.000477,
-      "loss": 0.3747,
+      "loss": 0.3972,
       "step": 734
     },
     {
       "epoch": 13.36,
       "learning_rate": 0.0004769,
-      "loss": 0.4448,
+      "loss": 0.4638,
       "step": 735
     },
     {
       "epoch": 13.38,
       "learning_rate": 0.0004768,
-      "loss": 0.4473,
+      "loss": 0.4548,
       "step": 736
     },
     {
       "epoch": 13.4,
       "learning_rate": 0.00047670000000000004,
-      "loss": 0.4873,
+      "loss": 0.4857,
       "step": 737
     },
     {
       "epoch": 13.42,
       "learning_rate": 0.00047660000000000004,
-      "loss": 0.4113,
+      "loss": 0.4479,
       "step": 738
     },
     {
       "epoch": 13.44,
       "learning_rate": 0.0004765,
-      "loss": 0.4827,
+      "loss": 0.4521,
       "step": 739
     },
     {
       "epoch": 13.45,
       "learning_rate": 0.0004764,
-      "loss": 0.4519,
+      "loss": 0.4712,
       "step": 740
     },
     {
       "epoch": 13.47,
       "learning_rate": 0.00047630000000000003,
-      "loss": 0.4936,
+      "loss": 0.5109,
       "step": 741
     },
     {
       "epoch": 13.49,
       "learning_rate": 0.00047620000000000003,
-      "loss": 0.5027,
+      "loss": 0.5164,
       "step": 742
     },
     {
       "epoch": 13.51,
       "learning_rate": 0.0004761,
-      "loss": 0.4508,
+      "loss": 0.4526,
       "step": 743
     },
     {
       "epoch": 13.53,
       "learning_rate": 0.00047599999999999997,
-      "loss": 0.471,
+      "loss": 0.5007,
       "step": 744
     },
     {
       "epoch": 13.55,
       "learning_rate": 0.0004759,
-      "loss": 0.556,
+      "loss": 0.5024,
       "step": 745
     },
     {
       "epoch": 13.56,
       "learning_rate": 0.0004758,
-      "loss": 0.4707,
+      "loss": 0.4552,
       "step": 746
     },
     {
       "epoch": 13.58,
       "learning_rate": 0.0004757,
-      "loss": 0.4974,
+      "loss": 0.4774,
       "step": 747
     },
     {
       "epoch": 13.6,
       "learning_rate": 0.0004756,
-      "loss": 0.4646,
+      "loss": 0.4626,
       "step": 748
     },
     {
       "epoch": 13.62,
       "learning_rate": 0.0004755,
-      "loss": 0.4296,
+      "loss": 0.4588,
       "step": 749
     },
     {
       "epoch": 13.64,
       "learning_rate": 0.0004754,
-      "loss": 0.4317,
+      "loss": 0.4281,
       "step": 750
     },
     {
       "epoch": 13.65,
       "learning_rate": 0.0004753,
-      "loss": 0.4724,
+      "loss": 0.4458,
       "step": 751
     },
     {
       "epoch": 13.67,
       "learning_rate": 0.0004752,
-      "loss": 0.5089,
+      "loss": 0.4801,
       "step": 752
     },
     {
       "epoch": 13.69,
       "learning_rate": 0.00047510000000000006,
-      "loss": 0.476,
+      "loss": 0.4563,
       "step": 753
     },
     {
       "epoch": 13.71,
       "learning_rate": 0.000475,
-      "loss": 0.4784,
+      "loss": 0.4872,
       "step": 754
     },
     {
       "epoch": 13.73,
       "learning_rate": 0.0004749,
-      "loss": 0.4691,
+      "loss": 0.4392,
       "step": 755
     },
     {
       "epoch": 13.75,
       "learning_rate": 0.0004748,
-      "loss": 0.4322,
+      "loss": 0.3963,
       "step": 756
     },
     {
       "epoch": 13.76,
       "learning_rate": 0.00047470000000000005,
-      "loss": 0.4276,
+      "loss": 0.4395,
       "step": 757
     },
     {
       "epoch": 13.78,
       "learning_rate": 0.00047460000000000004,
-      "loss": 0.4829,
+      "loss": 0.4807,
       "step": 758
     },
     {
       "epoch": 13.8,
       "learning_rate": 0.0004745,
-      "loss": 0.4955,
+      "loss": 0.5278,
       "step": 759
     },
     {
       "epoch": 13.82,
       "learning_rate": 0.0004744,
-      "loss": 0.4627,
+      "loss": 0.4538,
       "step": 760
     },
     {
       "epoch": 13.84,
       "learning_rate": 0.00047430000000000004,
-      "loss": 0.4473,
+      "loss": 0.41,
       "step": 761
     },
     {
       "epoch": 13.85,
       "learning_rate": 0.00047420000000000003,
-      "loss": 0.447,
+      "loss": 0.4449,
       "step": 762
     },
     {
       "epoch": 13.87,
       "learning_rate": 0.00047410000000000003,
-      "loss": 0.4992,
+      "loss": 0.4729,
       "step": 763
     },
     {
       "epoch": 13.89,
       "learning_rate": 0.000474,
-      "loss": 0.4564,
+      "loss": 0.4793,
       "step": 764
     },
     {
       "epoch": 13.91,
       "learning_rate": 0.00047389999999999997,
-      "loss": 0.4567,
+      "loss": 0.4704,
       "step": 765
     },
     {
       "epoch": 13.93,
       "learning_rate": 0.0004738,
-      "loss": 0.4817,
+      "loss": 0.4986,
       "step": 766
     },
     {
       "epoch": 13.95,
       "learning_rate": 0.0004737,
-      "loss": 0.435,
+      "loss": 0.4432,
       "step": 767
     },
     {
       "epoch": 13.96,
       "learning_rate": 0.0004736,
-      "loss": 0.4435,
+      "loss": 0.4668,
       "step": 768
     },
     {
       "epoch": 13.98,
       "learning_rate": 0.00047349999999999996,
-      "loss": 0.4298,
+      "loss": 0.4632,
       "step": 769
     },
     {
       "epoch": 14.0,
       "learning_rate": 0.0004734,
-      "loss": 0.3984,
+      "loss": 0.5058,
       "step": 770
     },
     {
       "epoch": 14.02,
       "learning_rate": 0.0004733,
-      "loss": 0.4217,
+      "loss": 0.4696,
       "step": 771
     },
     {
       "epoch": 14.04,
       "learning_rate": 0.0004732,
-      "loss": 0.4287,
+      "loss": 0.4626,
       "step": 772
     },
     {
       "epoch": 14.05,
       "learning_rate": 0.0004731,
-      "loss": 0.4066,
+      "loss": 0.385,
       "step": 773
     },
     {
       "epoch": 14.07,
       "learning_rate": 0.000473,
-      "loss": 0.4993,
+      "loss": 0.4641,
       "step": 774
     },
     {
       "epoch": 14.09,
       "learning_rate": 0.0004729,
-      "loss": 0.4259,
+      "loss": 0.4675,
       "step": 775
     },
     {
       "epoch": 14.11,
       "learning_rate": 0.0004728,
-      "loss": 0.3932,
+      "loss": 0.384,
       "step": 776
     },
     {
       "epoch": 14.13,
       "learning_rate": 0.0004727,
-      "loss": 0.4387,
+      "loss": 0.4745,
       "step": 777
     },
     {
       "epoch": 14.15,
       "learning_rate": 0.00047260000000000005,
-      "loss": 0.4305,
+      "loss": 0.4304,
       "step": 778
     },
     {
       "epoch": 14.16,
       "learning_rate": 0.0004725,
-      "loss": 0.4242,
+      "loss": 0.4297,
       "step": 779
     },
     {
       "epoch": 14.18,
       "learning_rate": 0.0004724,
-      "loss": 0.4751,
+      "loss": 0.434,
       "step": 780
     },
     {
       "epoch": 14.2,
       "learning_rate": 0.0004723,
-      "loss": 0.3488,
+      "loss": 0.3381,
       "step": 781
     },
     {
       "epoch": 14.22,
       "learning_rate": 0.00047220000000000004,
-      "loss": 0.4514,
+      "loss": 0.4681,
       "step": 782
     },
     {
       "epoch": 14.24,
       "learning_rate": 0.00047210000000000004,
-      "loss": 0.4041,
+      "loss": 0.4142,
       "step": 783
     },
     {
       "epoch": 14.25,
       "learning_rate": 0.000472,
-      "loss": 0.4633,
+      "loss": 0.4495,
       "step": 784
     },
     {
       "epoch": 14.27,
       "learning_rate": 0.0004719,
-      "loss": 0.4714,
+      "loss": 0.4505,
       "step": 785
     },
     {
       "epoch": 14.29,
       "learning_rate": 0.00047180000000000003,
-      "loss": 0.4296,
+      "loss": 0.4383,
       "step": 786
     },
     {
       "epoch": 14.31,
       "learning_rate": 0.00047170000000000003,
-      "loss": 0.4693,
+      "loss": 0.4857,
       "step": 787
     },
     {
       "epoch": 14.33,
       "learning_rate": 0.0004716,
-      "loss": 0.3888,
+      "loss": 0.4294,
       "step": 788
     },
     {
       "epoch": 14.35,
       "learning_rate": 0.00047149999999999997,
-      "loss": 0.3832,
+      "loss": 0.3837,
       "step": 789
     },
     {
       "epoch": 14.36,
       "learning_rate": 0.0004714,
-      "loss": 0.393,
+      "loss": 0.4052,
       "step": 790
     },
     {
       "epoch": 14.38,
       "learning_rate": 0.0004713,
-      "loss": 0.3626,
+      "loss": 0.3382,
       "step": 791
     },
     {
       "epoch": 14.4,
       "learning_rate": 0.0004712,
-      "loss": 0.4448,
+      "loss": 0.3979,
       "step": 792
     },
     {
       "epoch": 14.42,
       "learning_rate": 0.0004711,
-      "loss": 0.443,
+      "loss": 0.4207,
       "step": 793
     },
     {
       "epoch": 14.44,
       "learning_rate": 0.000471,
-      "loss": 0.4363,
+      "loss": 0.4441,
       "step": 794
     },
     {
       "epoch": 14.45,
       "learning_rate": 0.0004709,
-      "loss": 0.4391,
+      "loss": 0.3998,
       "step": 795
     },
     {
       "epoch": 14.47,
       "learning_rate": 0.0004708,
-      "loss": 0.4167,
+      "loss": 0.4034,
       "step": 796
     },
     {
       "epoch": 14.49,
       "learning_rate": 0.0004707,
-      "loss": 0.4363,
+      "loss": 0.3841,
       "step": 797
     },
     {
       "epoch": 14.51,
       "learning_rate": 0.00047060000000000005,
-      "loss": 0.3832,
+      "loss": 0.3904,
       "step": 798
     },
     {
       "epoch": 14.53,
       "learning_rate": 0.0004705,
-      "loss": 0.4361,
+      "loss": 0.4555,
       "step": 799
     },
     {
       "epoch": 14.55,
       "learning_rate": 0.0004704,
-      "loss": 0.5075,
+      "loss": 0.4743,
       "step": 800
     },
     {
       "epoch": 14.56,
       "learning_rate": 0.0004703,
-      "loss": 0.4114,
+      "loss": 0.4051,
       "step": 801
     },
     {
       "epoch": 14.58,
       "learning_rate": 0.0004702,
-      "loss": 0.484,
+      "loss": 0.4958,
       "step": 802
     },
     {
       "epoch": 14.6,
       "learning_rate": 0.00047010000000000004,
-      "loss": 0.3975,
+      "loss": 0.4218,
       "step": 803
     },
     {
       "epoch": 14.62,
       "learning_rate": 0.00047,
-      "loss": 0.323,
+      "loss": 0.3639,
       "step": 804
     },
     {
       "epoch": 14.64,
       "learning_rate": 0.0004699,
-      "loss": 0.4046,
+      "loss": 0.407,
       "step": 805
     },
     {
       "epoch": 14.65,
       "learning_rate": 0.0004698,
-      "loss": 0.4392,
+      "loss": 0.4076,
       "step": 806
     },
     {
       "epoch": 14.67,
       "learning_rate": 0.00046970000000000003,
-      "loss": 0.4076,
+      "loss": 0.4453,
       "step": 807
     },
     {
       "epoch": 14.69,
       "learning_rate": 0.00046960000000000003,
-      "loss": 0.4758,
+      "loss": 0.4739,
       "step": 808
     },
     {
       "epoch": 14.71,
       "learning_rate": 0.0004695,
-      "loss": 0.4483,
+      "loss": 0.5016,
       "step": 809
     },
     {
       "epoch": 14.73,
       "learning_rate": 0.00046939999999999997,
-      "loss": 0.3258,
+      "loss": 0.3691,
       "step": 810
     },
     {
       "epoch": 14.75,
       "learning_rate": 0.0004693,
-      "loss": 0.403,
+      "loss": 0.4561,
       "step": 811
     },
     {
       "epoch": 14.76,
       "learning_rate": 0.0004692,
-      "loss": 0.411,
+      "loss": 0.4821,
       "step": 812
     },
     {
       "epoch": 14.78,
       "learning_rate": 0.0004691,
-      "loss": 0.454,
+      "loss": 0.4547,
       "step": 813
     },
     {
       "epoch": 14.8,
       "learning_rate": 0.00046899999999999996,
-      "loss": 0.4148,
+      "loss": 0.4429,
       "step": 814
     },
     {
       "epoch": 14.82,
       "learning_rate": 0.0004689,
-      "loss": 0.3877,
+      "loss": 0.4317,
       "step": 815
     },
     {
       "epoch": 14.84,
       "learning_rate": 0.0004688,
-      "loss": 0.4235,
+      "loss": 0.4655,
       "step": 816
     },
     {
       "epoch": 14.85,
       "learning_rate": 0.0004687,
-      "loss": 0.4216,
+      "loss": 0.4211,
       "step": 817
     },
     {
       "epoch": 14.87,
       "learning_rate": 0.0004686,
-      "loss": 0.4704,
+      "loss": 0.5014,
       "step": 818
     },
     {
       "epoch": 14.89,
       "learning_rate": 0.00046850000000000006,
-      "loss": 0.3945,
+      "loss": 0.4227,
       "step": 819
     },
     {
       "epoch": 14.91,
       "learning_rate": 0.0004684,
-      "loss": 0.3041,
+      "loss": 0.3408,
       "step": 820
     },
     {
       "epoch": 14.93,
       "learning_rate": 0.0004683,
-      "loss": 0.4635,
+      "loss": 0.4763,
       "step": 821
     },
     {
       "epoch": 14.95,
       "learning_rate": 0.0004682,
-      "loss": 0.4086,
+      "loss": 0.4346,
       "step": 822
     },
     {
       "epoch": 14.96,
       "learning_rate": 0.00046810000000000005,
-      "loss": 0.4071,
+      "loss": 0.452,
       "step": 823
     },
     {
       "epoch": 14.98,
       "learning_rate": 0.00046800000000000005,
-      "loss": 0.4817,
+      "loss": 0.5679,
       "step": 824
     },
     {
       "epoch": 15.0,
       "learning_rate": 0.0004679,
-      "loss": 0.5053,
+      "loss": 0.5487,
       "step": 825
     },
     {
       "epoch": 15.02,
       "learning_rate": 0.0004678,
-      "loss": 0.4377,
+      "loss": 0.4361,
       "step": 826
     },
     {
       "epoch": 15.04,
       "learning_rate": 0.00046770000000000004,
-      "loss": 0.4462,
+      "loss": 0.4498,
       "step": 827
     },
     {
       "epoch": 15.05,
       "learning_rate": 0.00046760000000000004,
-      "loss": 0.4807,
+      "loss": 0.4462,
       "step": 828
     },
     {
       "epoch": 15.07,
       "learning_rate": 0.00046750000000000003,
-      "loss": 0.3732,
+      "loss": 0.3829,
       "step": 829
     },
     {
       "epoch": 15.09,
       "learning_rate": 0.0004674,
-      "loss": 0.4345,
+      "loss": 0.4502,
       "step": 830
     },
     {
       "epoch": 15.11,
       "learning_rate": 0.00046730000000000003,
-      "loss": 0.4419,
+      "loss": 0.4112,
       "step": 831
     },
     {
       "epoch": 15.13,
       "learning_rate": 0.0004672,
-      "loss": 0.4186,
+      "loss": 0.4412,
       "step": 832
     },
     {
       "epoch": 15.15,
       "learning_rate": 0.0004671,
-      "loss": 0.404,
+      "loss": 0.4155,
       "step": 833
     },
     {
       "epoch": 15.16,
       "learning_rate": 0.000467,
-      "loss": 0.3669,
+      "loss": 0.3564,
       "step": 834
     },
     {
       "epoch": 15.18,
       "learning_rate": 0.00046689999999999996,
-      "loss": 0.4041,
+      "loss": 0.4118,
       "step": 835
     },
     {
       "epoch": 15.2,
       "learning_rate": 0.0004668,
-      "loss": 0.3873,
+      "loss": 0.4277,
       "step": 836
     },
     {
       "epoch": 15.22,
       "learning_rate": 0.0004667,
-      "loss": 0.3554,
+      "loss": 0.3728,
       "step": 837
     },
     {
       "epoch": 15.24,
       "learning_rate": 0.0004666,
-      "loss": 0.4048,
+      "loss": 0.4069,
       "step": 838
     },
     {
       "epoch": 15.25,
       "learning_rate": 0.0004665,
-      "loss": 0.3957,
+      "loss": 0.3851,
       "step": 839
     },
     {
       "epoch": 15.27,
       "learning_rate": 0.0004664,
-      "loss": 0.4257,
+      "loss": 0.4222,
       "step": 840
     },
     {
       "epoch": 15.29,
       "learning_rate": 0.0004663,
-      "loss": 0.4327,
+      "loss": 0.4079,
       "step": 841
     },
     {
       "epoch": 15.31,
       "learning_rate": 0.0004662,
-      "loss": 0.3629,
+      "loss": 0.3584,
       "step": 842
     },
     {
       "epoch": 15.33,
       "learning_rate": 0.0004661,
-      "loss": 0.4288,
+      "loss": 0.4431,
       "step": 843
     },
     {
       "epoch": 15.35,
       "learning_rate": 0.00046600000000000005,
-      "loss": 0.4345,
+      "loss": 0.4033,
       "step": 844
     },
     {
       "epoch": 15.36,
       "learning_rate": 0.0004659,
-      "loss": 0.428,
+      "loss": 0.4178,
       "step": 845
     },
     {
       "epoch": 15.38,
       "learning_rate": 0.0004658,
-      "loss": 0.33,
+      "loss": 0.339,
       "step": 846
     },
     {
       "epoch": 15.4,
       "learning_rate": 0.0004657,
-      "loss": 0.3568,
+      "loss": 0.3512,
       "step": 847
     },
     {
       "epoch": 15.42,
       "learning_rate": 0.00046560000000000004,
-      "loss": 0.3958,
+      "loss": 0.4709,
       "step": 848
     },
     {
       "epoch": 15.44,
       "learning_rate": 0.00046550000000000004,
-      "loss": 0.4098,
+      "loss": 0.4103,
       "step": 849
     },
     {
       "epoch": 15.45,
       "learning_rate": 0.0004654,
-      "loss": 0.4429,
+      "loss": 0.4784,
       "step": 850
     },
     {
       "epoch": 15.47,
       "learning_rate": 0.0004653,
-      "loss": 0.4183,
+      "loss": 0.4237,
       "step": 851
     },
     {
       "epoch": 15.49,
       "learning_rate": 0.00046520000000000003,
-      "loss": 0.396,
+      "loss": 0.4152,
       "step": 852
     },
     {
       "epoch": 15.51,
       "learning_rate": 0.00046510000000000003,
-      "loss": 0.3926,
+      "loss": 0.3917,
       "step": 853
     },
     {
       "epoch": 15.53,
       "learning_rate": 0.000465,
-      "loss": 0.4016,
+      "loss": 0.4306,
       "step": 854
     },
     {
       "epoch": 15.55,
       "learning_rate": 0.00046489999999999997,
-      "loss": 0.356,
+      "loss": 0.3592,
       "step": 855
     },
     {
       "epoch": 15.56,
       "learning_rate": 0.0004648,
-      "loss": 0.3756,
+      "loss": 0.3514,
       "step": 856
     },
     {
       "epoch": 15.58,
       "learning_rate": 0.0004647,
-      "loss": 0.438,
+      "loss": 0.3984,
       "step": 857
     },
     {
       "epoch": 15.6,
       "learning_rate": 0.0004646,
-      "loss": 0.4162,
+      "loss": 0.3987,
       "step": 858
     },
     {
       "epoch": 15.62,
       "learning_rate": 0.0004645,
-      "loss": 0.364,
+      "loss": 0.3622,
       "step": 859
     },
     {
       "epoch": 15.64,
       "learning_rate": 0.0004644,
-      "loss": 0.4304,
+      "loss": 0.4109,
       "step": 860
     },
     {
       "epoch": 15.65,
       "learning_rate": 0.0004643,
-      "loss": 0.4872,
+      "loss": 0.4762,
       "step": 861
     },
     {
       "epoch": 15.67,
       "learning_rate": 0.0004642,
-      "loss": 0.4661,
+      "loss": 0.4687,
       "step": 862
     },
     {
       "epoch": 15.69,
       "learning_rate": 0.0004641,
-      "loss": 0.3978,
+      "loss": 0.4476,
       "step": 863
     },
     {
       "epoch": 15.71,
       "learning_rate": 0.00046400000000000006,
-      "loss": 0.3959,
+      "loss": 0.381,
       "step": 864
     },
     {
       "epoch": 15.73,
       "learning_rate": 0.0004639,
-      "loss": 0.4164,
+      "loss": 0.4083,
       "step": 865
     },
     {
       "epoch": 15.75,
       "learning_rate": 0.0004638,
-      "loss": 0.3817,
+      "loss": 0.3831,
       "step": 866
     },
     {
       "epoch": 15.76,
       "learning_rate": 0.0004637,
-      "loss": 0.3903,
+      "loss": 0.4072,
       "step": 867
     },
     {
       "epoch": 15.78,
       "learning_rate": 0.00046360000000000005,
-      "loss": 0.4044,
+      "loss": 0.4591,
       "step": 868
     },
     {
       "epoch": 15.8,
       "learning_rate": 0.00046350000000000004,
-      "loss": 0.4517,
+      "loss": 0.464,
       "step": 869
     },
     {
       "epoch": 15.82,
       "learning_rate": 0.0004634,
-      "loss": 0.3871,
+      "loss": 0.3949,
       "step": 870
     },
     {
       "epoch": 15.84,
       "learning_rate": 0.0004633,
-      "loss": 0.3383,
+      "loss": 0.3513,
       "step": 871
     },
     {
       "epoch": 15.85,
       "learning_rate": 0.0004632,
-      "loss": 0.428,
+      "loss": 0.4138,
       "step": 872
     },
     {
       "epoch": 15.87,
       "learning_rate": 0.00046310000000000004,
-      "loss": 0.4155,
+      "loss": 0.4504,
       "step": 873
     },
     {
       "epoch": 15.89,
       "learning_rate": 0.00046300000000000003,
-      "loss": 0.3814,
+      "loss": 0.3864,
       "step": 874
     },
     {
       "epoch": 15.91,
       "learning_rate": 0.0004629,
-      "loss": 0.3733,
+      "loss": 0.419,
       "step": 875
     },
     {
       "epoch": 15.93,
       "learning_rate": 0.0004628,
-      "loss": 0.43,
+      "loss": 0.4583,
       "step": 876
     },
     {
       "epoch": 15.95,
       "learning_rate": 0.0004627,
-      "loss": 0.4018,
+      "loss": 0.3794,
       "step": 877
     },
     {
       "epoch": 15.96,
       "learning_rate": 0.0004626,
-      "loss": 0.403,
+      "loss": 0.4445,
       "step": 878
     },
     {
       "epoch": 15.98,
       "learning_rate": 0.0004625,
-      "loss": 0.416,
+      "loss": 0.4373,
       "step": 879
     },
     {
       "epoch": 16.0,
       "learning_rate": 0.00046239999999999996,
-      "loss": 0.3848,
+      "loss": 0.3229,
       "step": 880
     },
     {
       "epoch": 16.02,
       "learning_rate": 0.0004623,
-      "loss": 0.4475,
+      "loss": 0.4578,
       "step": 881
     },
     {
       "epoch": 16.04,
       "learning_rate": 0.0004622,
-      "loss": 0.3788,
+      "loss": 0.368,
       "step": 882
     },
     {
       "epoch": 16.05,
       "learning_rate": 0.0004621,
-      "loss": 0.393,
+      "loss": 0.3934,
       "step": 883
     },
     {
       "epoch": 16.07,
       "learning_rate": 0.000462,
-      "loss": 0.4003,
+      "loss": 0.3974,
       "step": 884
     },
     {
       "epoch": 16.09,
       "learning_rate": 0.0004619,
-      "loss": 0.4749,
+      "loss": 0.4941,
       "step": 885
     },
     {
       "epoch": 16.11,
       "learning_rate": 0.0004618,
-      "loss": 0.3598,
+      "loss": 0.3781,
       "step": 886
     },
     {
       "epoch": 16.13,
       "learning_rate": 0.0004617,
-      "loss": 0.4773,
+      "loss": 0.454,
       "step": 887
     },
     {
       "epoch": 16.15,
       "learning_rate": 0.0004616,
-      "loss": 0.4021,
+      "loss": 0.4032,
       "step": 888
     },
     {
       "epoch": 16.16,
       "learning_rate": 0.00046150000000000005,
-      "loss": 0.4102,
+      "loss": 0.4443,
       "step": 889
     },
     {
       "epoch": 16.18,
       "learning_rate": 0.0004614,
-      "loss": 0.384,
+      "loss": 0.3861,
       "step": 890
     },
     {
       "epoch": 16.2,
       "learning_rate": 0.0004613,
-      "loss": 0.3589,
+      "loss": 0.3852,
       "step": 891
     },
     {
       "epoch": 16.22,
       "learning_rate": 0.0004612,
-      "loss": 0.3689,
+      "loss": 0.3563,
       "step": 892
     },
     {
       "epoch": 16.24,
       "learning_rate": 0.00046110000000000004,
-      "loss": 0.4342,
+      "loss": 0.4246,
       "step": 893
     },
     {
       "epoch": 16.25,
       "learning_rate": 0.00046100000000000004,
-      "loss": 0.346,
+      "loss": 0.3562,
       "step": 894
     },
     {
       "epoch": 16.27,
       "learning_rate": 0.0004609,
-      "loss": 0.3801,
+      "loss": 0.3848,
       "step": 895
     },
     {
       "epoch": 16.29,
       "learning_rate": 0.0004608,
-      "loss": 0.3625,
+      "loss": 0.3647,
       "step": 896
     },
     {
       "epoch": 16.31,
       "learning_rate": 0.00046070000000000003,
-      "loss": 0.4399,
+      "loss": 0.453,
       "step": 897
     },
     {
       "epoch": 16.33,
       "learning_rate": 0.00046060000000000003,
-      "loss": 0.3358,
+      "loss": 0.3745,
       "step": 898
     },
     {
       "epoch": 16.35,
       "learning_rate": 0.0004605,
-      "loss": 0.381,
+      "loss": 0.3984,
       "step": 899
     },
     {
       "epoch": 16.36,
       "learning_rate": 0.00046039999999999997,
-      "loss": 0.3901,
+      "loss": 0.3951,
       "step": 900
     },
     {
       "epoch": 16.38,
       "learning_rate": 0.0004603,
-      "loss": 0.3933,
+      "loss": 0.3879,
       "step": 901
     },
     {
       "epoch": 16.4,
       "learning_rate": 0.0004602,
-      "loss": 0.3801,
+      "loss": 0.4049,
       "step": 902
     },
     {
       "epoch": 16.42,
       "learning_rate": 0.0004601,
-      "loss": 0.4153,
+      "loss": 0.4013,
       "step": 903
     },
     {
       "epoch": 16.44,
       "learning_rate": 0.00046,
-      "loss": 0.392,
+      "loss": 0.3784,
       "step": 904
     },
     {
       "epoch": 16.45,
       "learning_rate": 0.0004599,
-      "loss": 0.4114,
+      "loss": 0.4029,
       "step": 905
     },
     {
       "epoch": 16.47,
       "learning_rate": 0.0004598,
-      "loss": 0.3866,
+      "loss": 0.3869,
       "step": 906
     },
     {
       "epoch": 16.49,
       "learning_rate": 0.0004597,
-      "loss": 0.3283,
+      "loss": 0.3168,
       "step": 907
     },
     {
       "epoch": 16.51,
       "learning_rate": 0.0004596,
-      "loss": 0.4027,
+      "loss": 0.3527,
       "step": 908
     },
     {
       "epoch": 16.53,
       "learning_rate": 0.00045950000000000006,
-      "loss": 0.3246,
+      "loss": 0.3266,
       "step": 909
     },
     {
       "epoch": 16.55,
       "learning_rate": 0.0004594,
-      "loss": 0.3861,
+      "loss": 0.3978,
       "step": 910
     },
     {
       "epoch": 16.56,
       "learning_rate": 0.0004593,
-      "loss": 0.3506,
+      "loss": 0.2975,
       "step": 911
     },
     {
       "epoch": 16.58,
       "learning_rate": 0.0004592,
-      "loss": 0.4523,
+      "loss": 0.4139,
       "step": 912
     },
     {
       "epoch": 16.6,
       "learning_rate": 0.0004591,
-      "loss": 0.3496,
+      "loss": 0.3422,
       "step": 913
     },
     {
       "epoch": 16.62,
       "learning_rate": 0.00045900000000000004,
-      "loss": 0.4023,
+      "loss": 0.402,
       "step": 914
     },
     {
       "epoch": 16.64,
       "learning_rate": 0.0004589,
-      "loss": 0.4181,
+      "loss": 0.3585,
       "step": 915
     },
     {
       "epoch": 16.65,
       "learning_rate": 0.0004588,
-      "loss": 0.3829,
+      "loss": 0.3884,
       "step": 916
     },
     {
       "epoch": 16.67,
       "learning_rate": 0.0004587,
-      "loss": 0.4659,
+      "loss": 0.4622,
       "step": 917
     },
     {
       "epoch": 16.69,
       "learning_rate": 0.00045860000000000003,
-      "loss": 0.3699,
+      "loss": 0.3619,
       "step": 918
     },
     {
       "epoch": 16.71,
       "learning_rate": 0.00045850000000000003,
-      "loss": 0.3768,
+      "loss": 0.3956,
       "step": 919
     },
     {
       "epoch": 16.73,
       "learning_rate": 0.0004584,
-      "loss": 0.3402,
+      "loss": 0.3232,
       "step": 920
     },
     {
       "epoch": 16.75,
       "learning_rate": 0.00045829999999999997,
-      "loss": 0.3644,
+      "loss": 0.3538,
       "step": 921
     },
     {
       "epoch": 16.76,
       "learning_rate": 0.0004582,
-      "loss": 0.3344,
+      "loss": 0.333,
       "step": 922
     },
     {
       "epoch": 16.78,
       "learning_rate": 0.0004581,
-      "loss": 0.3918,
+      "loss": 0.3749,
       "step": 923
     },
     {
       "epoch": 16.8,
       "learning_rate": 0.000458,
-      "loss": 0.4197,
+      "loss": 0.4145,
       "step": 924
     },
     {
       "epoch": 16.82,
       "learning_rate": 0.00045789999999999996,
-      "loss": 0.3541,
+      "loss": 0.3458,
       "step": 925
     },
     {
       "epoch": 16.84,
       "learning_rate": 0.0004578,
-      "loss": 0.3557,
+      "loss": 0.3799,
       "step": 926
     },
     {
       "epoch": 16.85,
       "learning_rate": 0.0004577,
-      "loss": 0.3574,
+      "loss": 0.3833,
       "step": 927
     },
     {
       "epoch": 16.87,
       "learning_rate": 0.0004576,
-      "loss": 0.3882,
+      "loss": 0.3965,
       "step": 928
     },
     {
       "epoch": 16.89,
       "learning_rate": 0.0004575,
-      "loss": 0.3612,
+      "loss": 0.398,
       "step": 929
     },
     {
       "epoch": 16.91,
       "learning_rate": 0.0004574,
-      "loss": 0.4016,
+      "loss": 0.4141,
       "step": 930
     },
     {
       "epoch": 16.93,
       "learning_rate": 0.0004573,
-      "loss": 0.389,
+      "loss": 0.3747,
       "step": 931
     },
     {
       "epoch": 16.95,
       "learning_rate": 0.0004572,
-      "loss": 0.3604,
+      "loss": 0.3767,
       "step": 932
     },
     {
       "epoch": 16.96,
       "learning_rate": 0.0004571,
-      "loss": 0.3427,
+      "loss": 0.3581,
       "step": 933
     },
     {
       "epoch": 16.98,
       "learning_rate": 0.00045700000000000005,
-      "loss": 0.3929,
+      "loss": 0.3719,
       "step": 934
     },
     {
       "epoch": 17.0,
       "learning_rate": 0.0004569,
-      "loss": 0.2854,
+      "loss": 0.315,
       "step": 935
     },
     {
       "epoch": 17.02,
       "learning_rate": 0.0004568,
-      "loss": 0.3459,
+      "loss": 0.3555,
       "step": 936
     },
     {
       "epoch": 17.04,
       "learning_rate": 0.0004567,
-      "loss": 0.3617,
+      "loss": 0.3549,
       "step": 937
     },
     {
       "epoch": 17.05,
       "learning_rate": 0.00045660000000000004,
-      "loss": 0.3368,
+      "loss": 0.3112,
       "step": 938
     },
     {
       "epoch": 17.07,
       "learning_rate": 0.00045650000000000004,
-      "loss": 0.409,
+      "loss": 0.4323,
       "step": 939
     },
     {
       "epoch": 17.09,
       "learning_rate": 0.0004564,
-      "loss": 0.3244,
+      "loss": 0.3475,
       "step": 940
     },
     {
       "epoch": 17.11,
       "learning_rate": 0.0004563,
-      "loss": 0.3385,
+      "loss": 0.3766,
       "step": 941
     },
     {
       "epoch": 17.13,
       "learning_rate": 0.00045620000000000003,
-      "loss": 0.3948,
+      "loss": 0.3863,
       "step": 942
     },
     {
       "epoch": 17.15,
       "learning_rate": 0.00045610000000000003,
-      "loss": 0.3292,
+      "loss": 0.3439,
       "step": 943
     },
     {
       "epoch": 17.16,
       "learning_rate": 0.000456,
-      "loss": 0.3406,
+      "loss": 0.3336,
       "step": 944
     },
     {
       "epoch": 17.18,
       "learning_rate": 0.0004559,
-      "loss": 0.4171,
+      "loss": 0.3932,
       "step": 945
     },
     {
       "epoch": 17.2,
       "learning_rate": 0.00045579999999999997,
-      "loss": 0.355,
+      "loss": 0.3525,
       "step": 946
     },
     {
       "epoch": 17.22,
       "learning_rate": 0.0004557,
-      "loss": 0.3894,
+      "loss": 0.3724,
       "step": 947
     },
     {
       "epoch": 17.24,
       "learning_rate": 0.0004556,
-      "loss": 0.4125,
+      "loss": 0.3848,
       "step": 948
     },
     {
       "epoch": 17.25,
       "learning_rate": 0.0004555,
-      "loss": 0.4161,
+      "loss": 0.4098,
       "step": 949
     },
     {
       "epoch": 17.27,
       "learning_rate": 0.0004554,
-      "loss": 0.3694,
+      "loss": 0.4136,
       "step": 950
     },
     {
       "epoch": 17.29,
       "learning_rate": 0.0004553,
-      "loss": 0.4054,
+      "loss": 0.3882,
       "step": 951
     },
     {
       "epoch": 17.31,
       "learning_rate": 0.0004552,
-      "loss": 0.4189,
+      "loss": 0.4027,
       "step": 952
     },
     {
       "epoch": 17.33,
       "learning_rate": 0.0004551,
-      "loss": 0.4222,
+      "loss": 0.4026,
       "step": 953
     },
     {
       "epoch": 17.35,
       "learning_rate": 0.000455,
-      "loss": 0.4141,
+      "loss": 0.3854,
       "step": 954
     },
     {
       "epoch": 17.36,
       "learning_rate": 0.00045490000000000005,
-      "loss": 0.4172,
+      "loss": 0.4148,
       "step": 955
     },
     {
       "epoch": 17.38,
       "learning_rate": 0.0004548,
-      "loss": 0.3585,
+      "loss": 0.3728,
       "step": 956
     },
     {
       "epoch": 17.4,
       "learning_rate": 0.0004547,
-      "loss": 0.3692,
+      "loss": 0.4045,
       "step": 957
     },
     {
       "epoch": 17.42,
       "learning_rate": 0.0004546,
-      "loss": 0.3642,
+      "loss": 0.3971,
       "step": 958
     },
     {
       "epoch": 17.44,
       "learning_rate": 0.00045450000000000004,
-      "loss": 0.4368,
+      "loss": 0.3817,
       "step": 959
     },
     {
       "epoch": 17.45,
       "learning_rate": 0.00045440000000000004,
-      "loss": 0.3622,
+      "loss": 0.3572,
       "step": 960
     },
     {
       "epoch": 17.47,
       "learning_rate": 0.0004543,
-      "loss": 0.3878,
+      "loss": 0.4081,
       "step": 961
     },
     {
       "epoch": 17.49,
       "learning_rate": 0.0004542,
-      "loss": 0.3829,
+      "loss": 0.3578,
       "step": 962
     },
     {
       "epoch": 17.51,
       "learning_rate": 0.00045410000000000003,
-      "loss": 0.3673,
+      "loss": 0.3346,
       "step": 963
     },
     {
       "epoch": 17.53,
       "learning_rate": 0.00045400000000000003,
-      "loss": 0.3802,
+      "loss": 0.4014,
       "step": 964
     },
     {
       "epoch": 17.55,
       "learning_rate": 0.00045390000000000003,
-      "loss": 0.3703,
+      "loss": 0.3588,
       "step": 965
     },
     {
       "epoch": 17.56,
       "learning_rate": 0.00045379999999999997,
-      "loss": 0.3799,
+      "loss": 0.3787,
       "step": 966
     },
     {
       "epoch": 17.58,
       "learning_rate": 0.0004537,
-      "loss": 0.3745,
+      "loss": 0.3622,
       "step": 967
     },
     {
       "epoch": 17.6,
       "learning_rate": 0.0004536,
-      "loss": 0.4199,
+      "loss": 0.4251,
       "step": 968
     },
     {
       "epoch": 17.62,
       "learning_rate": 0.0004535,
-      "loss": 0.3754,
+      "loss": 0.3362,
       "step": 969
     },
     {
       "epoch": 17.64,
       "learning_rate": 0.0004534,
-      "loss": 0.3768,
+      "loss": 0.3668,
       "step": 970
     },
     {
       "epoch": 17.65,
       "learning_rate": 0.0004533,
-      "loss": 0.4004,
+      "loss": 0.3781,
       "step": 971
     },
     {
       "epoch": 17.67,
       "learning_rate": 0.0004532,
-      "loss": 0.4333,
+      "loss": 0.3764,
       "step": 972
     },
     {
       "epoch": 17.69,
       "learning_rate": 0.0004531,
-      "loss": 0.3486,
+      "loss": 0.3298,
       "step": 973
     },
     {
       "epoch": 17.71,
       "learning_rate": 0.000453,
-      "loss": 0.4398,
+      "loss": 0.4089,
       "step": 974
     },
     {
       "epoch": 17.73,
       "learning_rate": 0.00045290000000000006,
-      "loss": 0.4017,
+      "loss": 0.3739,
       "step": 975
     },
     {
       "epoch": 17.75,
       "learning_rate": 0.0004528,
-      "loss": 0.4257,
+      "loss": 0.421,
       "step": 976
     },
     {
       "epoch": 17.76,
       "learning_rate": 0.0004527,
-      "loss": 0.3479,
+      "loss": 0.3648,
       "step": 977
     },
     {
       "epoch": 17.78,
       "learning_rate": 0.0004526,
-      "loss": 0.3427,
+      "loss": 0.3239,
       "step": 978
     },
     {
       "epoch": 17.8,
       "learning_rate": 0.00045250000000000005,
-      "loss": 0.352,
+      "loss": 0.3456,
       "step": 979
     },
     {
       "epoch": 17.82,
       "learning_rate": 0.00045240000000000005,
-      "loss": 0.4049,
+      "loss": 0.4288,
       "step": 980
     },
     {
       "epoch": 17.84,
       "learning_rate": 0.0004523,
-      "loss": 0.3402,
+      "loss": 0.3183,
       "step": 981
     },
     {
       "epoch": 17.85,
       "learning_rate": 0.0004522,
-      "loss": 0.3577,
+      "loss": 0.3747,
       "step": 982
     },
     {
       "epoch": 17.87,
       "learning_rate": 0.0004521,
-      "loss": 0.4368,
+      "loss": 0.433,
       "step": 983
     },
     {
       "epoch": 17.89,
       "learning_rate": 0.00045200000000000004,
-      "loss": 0.4011,
+      "loss": 0.455,
       "step": 984
     },
     {
       "epoch": 17.91,
       "learning_rate": 0.00045190000000000003,
-      "loss": 0.3773,
+      "loss": 0.4036,
       "step": 985
     },
     {
       "epoch": 17.93,
       "learning_rate": 0.0004518,
-      "loss": 0.4003,
+      "loss": 0.3885,
       "step": 986
     },
     {
       "epoch": 17.95,
       "learning_rate": 0.0004517,
-      "loss": 0.4235,
+      "loss": 0.411,
       "step": 987
     },
     {
       "epoch": 17.96,
       "learning_rate": 0.0004516,
-      "loss": 0.4038,
+      "loss": 0.4243,
       "step": 988
     },
     {
       "epoch": 17.98,
       "learning_rate": 0.0004515,
-      "loss": 0.4205,
+      "loss": 0.3715,
       "step": 989
     },
     {
       "epoch": 18.0,
       "learning_rate": 0.0004514,
-      "loss": 0.2484,
+      "loss": 0.2816,
       "step": 990
     },
     {
       "epoch": 18.02,
       "learning_rate": 0.00045129999999999997,
-      "loss": 0.4144,
+      "loss": 0.3843,
       "step": 991
     },
     {
       "epoch": 18.04,
       "learning_rate": 0.0004512,
-      "loss": 0.3577,
+      "loss": 0.3858,
       "step": 992
     },
     {
       "epoch": 18.05,
       "learning_rate": 0.0004511,
-      "loss": 0.3223,
+      "loss": 0.2982,
       "step": 993
     },
     {
       "epoch": 18.07,
       "learning_rate": 0.000451,
-      "loss": 0.3276,
+      "loss": 0.2946,
       "step": 994
     },
     {
       "epoch": 18.09,
       "learning_rate": 0.0004509,
-      "loss": 0.4354,
+      "loss": 0.3862,
       "step": 995
     },
     {
       "epoch": 18.11,
       "learning_rate": 0.0004508,
-      "loss": 0.3701,
+      "loss": 0.3474,
       "step": 996
     },
     {
       "epoch": 18.13,
       "learning_rate": 0.0004507,
-      "loss": 0.3483,
+      "loss": 0.3082,
       "step": 997
     },
     {
       "epoch": 18.15,
       "learning_rate": 0.0004506,
-      "loss": 0.3456,
+      "loss": 0.3685,
       "step": 998
     },
     {
       "epoch": 18.16,
       "learning_rate": 0.0004505,
-      "loss": 0.356,
+      "loss": 0.3373,
       "step": 999
     },
     {
       "epoch": 18.18,
       "learning_rate": 0.00045040000000000005,
-      "loss": 0.3508,
+      "loss": 0.3496,
       "step": 1000
     },
     {
       "epoch": 18.18,
-      "eval_cer": 0.10494909968665198,
-      "eval_loss": 0.4063118100166321,
-      "eval_runtime": 13.7664,
-      "eval_samples_per_second": 119.639,
-      "eval_steps_per_second": 1.889,
-      "eval_wer": 0.4138494535798182,
+      "eval_cer": 0.10631343798257845,
+      "eval_loss": 0.378420889377594,
+      "eval_runtime": 14.1977,
+      "eval_samples_per_second": 116.005,
+      "eval_steps_per_second": 1.831,
+      "eval_wer": 0.424471453375549,
       "step": 1000
     },
     {
       "epoch": 18.2,
       "learning_rate": 0.0004503,
-      "loss": 0.405,
+      "loss": 0.4036,
       "step": 1001
     },
     {
       "epoch": 18.22,
       "learning_rate": 0.0004502,
-      "loss": 0.3682,
+      "loss": 0.3628,
       "step": 1002
     },
     {
       "epoch": 18.24,
       "learning_rate": 0.0004501,
-      "loss": 0.4202,
+      "loss": 0.3836,
       "step": 1003
     },
     {
       "epoch": 18.25,
       "learning_rate": 0.00045000000000000004,
-      "loss": 0.4004,
+      "loss": 0.3265,
       "step": 1004
     },
     {
       "epoch": 18.27,
       "learning_rate": 0.00044990000000000004,
-      "loss": 0.3519,
+      "loss": 0.3495,
       "step": 1005
     },
     {
       "epoch": 18.29,
       "learning_rate": 0.0004498,
-      "loss": 0.3743,
+      "loss": 0.3677,
       "step": 1006
     },
     {
       "epoch": 18.31,
       "learning_rate": 0.0004497,
-      "loss": 0.3548,
+      "loss": 0.3567,
       "step": 1007
     },
     {
       "epoch": 18.33,
       "learning_rate": 0.00044960000000000003,
-      "loss": 0.3466,
+      "loss": 0.3418,
       "step": 1008
     },
     {
       "epoch": 18.35,
       "learning_rate": 0.00044950000000000003,
-      "loss": 0.3866,
+      "loss": 0.386,
       "step": 1009
     },
     {
       "epoch": 18.36,
       "learning_rate": 0.00044940000000000003,
-      "loss": 0.3629,
+      "loss": 0.3654,
       "step": 1010
     },
     {
       "epoch": 18.38,
       "learning_rate": 0.00044929999999999997,
-      "loss": 0.3474,
+      "loss": 0.3811,
       "step": 1011
     },
     {
       "epoch": 18.4,
       "learning_rate": 0.0004492,
-      "loss": 0.3618,
+      "loss": 0.3512,
       "step": 1012
     },
     {
       "epoch": 18.42,
       "learning_rate": 0.0004491,
-      "loss": 0.3647,
+      "loss": 0.3608,
       "step": 1013
     },
     {
       "epoch": 18.44,
       "learning_rate": 0.000449,
-      "loss": 0.3402,
+      "loss": 0.3357,
       "step": 1014
     },
     {
       "epoch": 18.45,
       "learning_rate": 0.0004489,
-      "loss": 0.3585,
+      "loss": 0.3509,
       "step": 1015
     },
     {
       "epoch": 18.47,
       "learning_rate": 0.0004488,
-      "loss": 0.3468,
+      "loss": 0.3422,
       "step": 1016
     },
     {
       "epoch": 18.49,
       "learning_rate": 0.0004487,
-      "loss": 0.3332,
+      "loss": 0.3292,
       "step": 1017
     },
     {
       "epoch": 18.51,
       "learning_rate": 0.0004486,
-      "loss": 0.4107,
+      "loss": 0.4493,
       "step": 1018
     },
     {
       "epoch": 18.53,
       "learning_rate": 0.0004485,
-      "loss": 0.3345,
+      "loss": 0.3236,
       "step": 1019
     },
     {
       "epoch": 18.55,
       "learning_rate": 0.00044840000000000006,
-      "loss": 0.3365,
+      "loss": 0.3369,
       "step": 1020
     },
     {
       "epoch": 18.56,
       "learning_rate": 0.0004483,
-      "loss": 0.3596,
+      "loss": 0.3482,
       "step": 1021
     },
     {
       "epoch": 18.58,
       "learning_rate": 0.0004482,
-      "loss": 0.361,
+      "loss": 0.3549,
       "step": 1022
     },
     {
       "epoch": 18.6,
       "learning_rate": 0.0004481,
-      "loss": 0.4482,
+      "loss": 0.4565,
       "step": 1023
     },
     {
       "epoch": 18.62,
       "learning_rate": 0.000448,
-      "loss": 0.3427,
+      "loss": 0.3394,
       "step": 1024
     },
     {
       "epoch": 18.64,
       "learning_rate": 0.00044790000000000005,
-      "loss": 0.3479,
+      "loss": 0.3224,
       "step": 1025
     },
     {
       "epoch": 18.65,
       "learning_rate": 0.0004478,
-      "loss": 0.3826,
+      "loss": 0.3824,
       "step": 1026
     },
     {
       "epoch": 18.67,
       "learning_rate": 0.0004477,
-      "loss": 0.3624,
+      "loss": 0.3569,
       "step": 1027
     },
     {
       "epoch": 18.69,
       "learning_rate": 0.0004476,
-      "loss": 0.3663,
+      "loss": 0.408,
       "step": 1028
     },
     {
       "epoch": 18.71,
       "learning_rate": 0.00044750000000000004,
-      "loss": 0.3728,
+      "loss": 0.36,
       "step": 1029
     },
     {
       "epoch": 18.73,
       "learning_rate": 0.00044740000000000003,
-      "loss": 0.3582,
+      "loss": 0.3266,
       "step": 1030
     },
     {
       "epoch": 18.75,
       "learning_rate": 0.0004473,
-      "loss": 0.3635,
+      "loss": 0.3436,
       "step": 1031
     },
     {
       "epoch": 18.76,
       "learning_rate": 0.0004472,
-      "loss": 0.3407,
+      "loss": 0.3729,
       "step": 1032
     },
     {
       "epoch": 18.78,
       "learning_rate": 0.0004471,
-      "loss": 0.3696,
+      "loss": 0.3652,
       "step": 1033
     },
     {
       "epoch": 18.8,
       "learning_rate": 0.000447,
-      "loss": 0.4112,
+      "loss": 0.399,
       "step": 1034
     },
     {
       "epoch": 18.82,
       "learning_rate": 0.0004469,
-      "loss": 0.3805,
+      "loss": 0.3714,
       "step": 1035
     },
     {
       "epoch": 18.84,
       "learning_rate": 0.00044679999999999996,
-      "loss": 0.4104,
+      "loss": 0.3756,
       "step": 1036
     },
     {
       "epoch": 18.85,
       "learning_rate": 0.0004467,
-      "loss": 0.3233,
+      "loss": 0.3231,
       "step": 1037
     },
     {
       "epoch": 18.87,
       "learning_rate": 0.0004466,
-      "loss": 0.3304,
+      "loss": 0.3303,
       "step": 1038
     },
     {
       "epoch": 18.89,
       "learning_rate": 0.0004465,
-      "loss": 0.317,
+      "loss": 0.2955,
       "step": 1039
     },
     {
       "epoch": 18.91,
       "learning_rate": 0.0004464,
-      "loss": 0.3784,
+      "loss": 0.383,
       "step": 1040
     },
     {
       "epoch": 18.93,
       "learning_rate": 0.0004463,
-      "loss": 0.4286,
+      "loss": 0.4117,
       "step": 1041
     },
     {
       "epoch": 18.95,
       "learning_rate": 0.0004462,
-      "loss": 0.3168,
+      "loss": 0.3063,
       "step": 1042
     },
     {
       "epoch": 18.96,
       "learning_rate": 0.0004461,
-      "loss": 0.3587,
+      "loss": 0.3597,
       "step": 1043
     },
     {
       "epoch": 18.98,
       "learning_rate": 0.000446,
-      "loss": 0.3841,
+      "loss": 0.351,
       "step": 1044
     },
     {
       "epoch": 19.0,
       "learning_rate": 0.00044590000000000005,
-      "loss": 0.2978,
+      "loss": 0.307,
       "step": 1045
     },
     {
       "epoch": 19.02,
       "learning_rate": 0.0004458,
-      "loss": 0.3759,
+      "loss": 0.4005,
       "step": 1046
     },
     {
       "epoch": 19.04,
       "learning_rate": 0.0004457,
-      "loss": 0.3243,
+      "loss": 0.3173,
       "step": 1047
     },
     {
       "epoch": 19.05,
       "learning_rate": 0.0004456,
-      "loss": 0.3602,
+      "loss": 0.3396,
       "step": 1048
     },
     {
       "epoch": 19.07,
       "learning_rate": 0.00044550000000000004,
-      "loss": 0.3058,
+      "loss": 0.3037,
       "step": 1049
     },
     {
       "epoch": 19.09,
       "learning_rate": 0.00044540000000000004,
-      "loss": 0.3067,
+      "loss": 0.3392,
       "step": 1050
     },
     {
       "epoch": 19.11,
       "learning_rate": 0.0004453,
-      "loss": 0.3812,
+      "loss": 0.3393,
       "step": 1051
     },
     {
       "epoch": 19.13,
       "learning_rate": 0.0004452,
-      "loss": 0.4018,
+      "loss": 0.3557,
       "step": 1052
     },
     {
       "epoch": 19.15,
       "learning_rate": 0.00044510000000000003,
-      "loss": 0.409,
+      "loss": 0.3862,
       "step": 1053
     },
     {
       "epoch": 19.16,
       "learning_rate": 0.00044500000000000003,
-      "loss": 0.3834,
+      "loss": 0.3254,
       "step": 1054
     },
     {
       "epoch": 19.18,
       "learning_rate": 0.0004449,
-      "loss": 0.346,
+      "loss": 0.3221,
       "step": 1055
     },
     {
       "epoch": 19.2,
       "learning_rate": 0.00044479999999999997,
-      "loss": 0.3922,
+      "loss": 0.3977,
       "step": 1056
     },
     {
       "epoch": 19.22,
       "learning_rate": 0.00044469999999999997,
-      "loss": 0.3902,
+      "loss": 0.3543,
       "step": 1057
     },
     {
       "epoch": 19.24,
       "learning_rate": 0.0004446,
-      "loss": 0.3028,
+      "loss": 0.2924,
       "step": 1058
     },
     {
       "epoch": 19.25,
       "learning_rate": 0.0004445,
-      "loss": 0.396,
+      "loss": 0.3828,
       "step": 1059
     },
     {
       "epoch": 19.27,
       "learning_rate": 0.0004444,
-      "loss": 0.3451,
+      "loss": 0.3329,
       "step": 1060
     },
     {
       "epoch": 19.29,
       "learning_rate": 0.00044429999999999996,
-      "loss": 0.3517,
+      "loss": 0.3176,
       "step": 1061
     },
     {
       "epoch": 19.31,
       "learning_rate": 0.0004442,
-      "loss": 0.3714,
+      "loss": 0.3445,
       "step": 1062
     },
     {
       "epoch": 19.33,
       "learning_rate": 0.0004441,
-      "loss": 0.3871,
+      "loss": 0.3669,
       "step": 1063
     },
     {
       "epoch": 19.35,
       "learning_rate": 0.000444,
-      "loss": 0.3892,
+      "loss": 0.3737,
       "step": 1064
     },
     {
       "epoch": 19.36,
       "learning_rate": 0.0004439,
-      "loss": 0.3617,
+      "loss": 0.3406,
       "step": 1065
     },
     {
       "epoch": 19.38,
       "learning_rate": 0.0004438,
-      "loss": 0.3313,
+      "loss": 0.3297,
       "step": 1066
     },
     {
       "epoch": 19.4,
       "learning_rate": 0.0004437,
-      "loss": 0.321,
+      "loss": 0.3118,
       "step": 1067
     },
     {
       "epoch": 19.42,
       "learning_rate": 0.0004436,
-      "loss": 0.383,
+      "loss": 0.3728,
       "step": 1068
     },
     {
       "epoch": 19.44,
       "learning_rate": 0.0004435,
-      "loss": 0.2945,
+      "loss": 0.3277,
       "step": 1069
     },
     {
       "epoch": 19.45,
       "learning_rate": 0.00044340000000000004,
-      "loss": 0.3153,
+      "loss": 0.3427,
       "step": 1070
     },
     {
       "epoch": 19.47,
       "learning_rate": 0.00044330000000000004,
-      "loss": 0.3795,
+      "loss": 0.3486,
       "step": 1071
     },
     {
       "epoch": 19.49,
       "learning_rate": 0.0004432,
-      "loss": 0.3282,
+      "loss": 0.3237,
       "step": 1072
     },
     {
       "epoch": 19.51,
       "learning_rate": 0.0004431,
-      "loss": 0.3799,
+      "loss": 0.3384,
       "step": 1073
     },
     {
       "epoch": 19.53,
       "learning_rate": 0.00044300000000000003,
-      "loss": 0.3869,
+      "loss": 0.3583,
       "step": 1074
     },
     {
       "epoch": 19.55,
       "learning_rate": 0.00044290000000000003,
-      "loss": 0.3683,
+      "loss": 0.3046,
       "step": 1075
     },
     {
       "epoch": 19.56,
       "learning_rate": 0.00044280000000000003,
-      "loss": 0.3967,
+      "loss": 0.3742,
       "step": 1076
     },
     {
       "epoch": 19.58,
       "learning_rate": 0.00044269999999999997,
-      "loss": 0.3465,
+      "loss": 0.3408,
       "step": 1077
     },
     {
       "epoch": 19.6,
       "learning_rate": 0.0004426,
-      "loss": 0.3491,
+      "loss": 0.3149,
       "step": 1078
     },
     {
       "epoch": 19.62,
       "learning_rate": 0.0004425,
-      "loss": 0.333,
+      "loss": 0.3087,
       "step": 1079
     },
     {
       "epoch": 19.64,
       "learning_rate": 0.0004424,
-      "loss": 0.3539,
+      "loss": 0.3565,
       "step": 1080
     },
     {
       "epoch": 19.65,
       "learning_rate": 0.0004423,
-      "loss": 0.4239,
+      "loss": 0.3911,
       "step": 1081
     },
     {
       "epoch": 19.67,
       "learning_rate": 0.0004422,
-      "loss": 0.4012,
+      "loss": 0.3796,
       "step": 1082
     },
     {
       "epoch": 19.69,
       "learning_rate": 0.0004421,
-      "loss": 0.4621,
+      "loss": 0.4081,
       "step": 1083
     },
     {
       "epoch": 19.71,
       "learning_rate": 0.000442,
-      "loss": 0.3732,
+      "loss": 0.3487,
       "step": 1084
     },
     {
       "epoch": 19.73,
       "learning_rate": 0.0004419,
-      "loss": 0.3684,
+      "loss": 0.3529,
       "step": 1085
     },
     {
       "epoch": 19.75,
       "learning_rate": 0.00044180000000000006,
-      "loss": 0.4106,
+      "loss": 0.3163,
       "step": 1086
     },
     {
       "epoch": 19.76,
       "learning_rate": 0.0004417,
-      "loss": 0.3175,
+      "loss": 0.3193,
       "step": 1087
     },
     {
       "epoch": 19.78,
       "learning_rate": 0.0004416,
-      "loss": 0.3409,
+      "loss": 0.3008,
       "step": 1088
     },
     {
       "epoch": 19.8,
       "learning_rate": 0.0004415,
-      "loss": 0.3221,
+      "loss": 0.3012,
       "step": 1089
     },
     {
       "epoch": 19.82,
       "learning_rate": 0.00044140000000000005,
-      "loss": 0.3391,
+      "loss": 0.3619,
       "step": 1090
     },
     {
       "epoch": 19.84,
       "learning_rate": 0.00044130000000000005,
-      "loss": 0.3465,
+      "loss": 0.3361,
       "step": 1091
     },
     {
       "epoch": 19.85,
       "learning_rate": 0.0004412,
-      "loss": 0.3586,
+      "loss": 0.3348,
       "step": 1092
     },
     {
       "epoch": 19.87,
       "learning_rate": 0.0004411,
-      "loss": 0.409,
+      "loss": 0.3801,
       "step": 1093
     },
     {
       "epoch": 19.89,
       "learning_rate": 0.000441,
-      "loss": 0.3214,
+      "loss": 0.3199,
       "step": 1094
     },
     {
       "epoch": 19.91,
       "learning_rate": 0.00044090000000000004,
-      "loss": 0.3319,
+      "loss": 0.3479,
       "step": 1095
     },
     {
       "epoch": 19.93,
       "learning_rate": 0.00044080000000000004,
-      "loss": 0.3683,
+      "loss": 0.3614,
       "step": 1096
     },
     {
       "epoch": 19.95,
       "learning_rate": 0.0004407,
-      "loss": 0.3013,
+      "loss": 0.2979,
       "step": 1097
     },
     {
       "epoch": 19.96,
       "learning_rate": 0.0004406,
-      "loss": 0.3791,
+      "loss": 0.3634,
       "step": 1098
     },
     {
       "epoch": 19.98,
       "learning_rate": 0.00044050000000000003,
-      "loss": 0.3381,
+      "loss": 0.3181,
       "step": 1099
     },
     {
       "epoch": 20.0,
       "learning_rate": 0.0004404,
-      "loss": 0.3039,
+      "loss": 0.3027,
       "step": 1100
     },
     {
       "epoch": 20.02,
       "learning_rate": 0.0004403,
-      "loss": 0.3542,
+      "loss": 0.3487,
       "step": 1101
     },
     {
       "epoch": 20.04,
       "learning_rate": 0.00044019999999999997,
-      "loss": 0.3218,
+      "loss": 0.284,
       "step": 1102
     },
     {
       "epoch": 20.05,
       "learning_rate": 0.0004401,
-      "loss": 0.3137,
+      "loss": 0.2848,
       "step": 1103
     },
     {
       "epoch": 20.07,
       "learning_rate": 0.00044,
-      "loss": 0.374,
+      "loss": 0.3373,
       "step": 1104
     },
     {
       "epoch": 20.09,
       "learning_rate": 0.0004399,
-      "loss": 0.3705,
+      "loss": 0.3571,
       "step": 1105
     },
     {
       "epoch": 20.11,
       "learning_rate": 0.0004398,
-      "loss": 0.3226,
+      "loss": 0.2967,
       "step": 1106
     },
     {
       "epoch": 20.13,
       "learning_rate": 0.0004397,
-      "loss": 0.3625,
+      "loss": 0.323,
       "step": 1107
     },
     {
       "epoch": 20.15,
       "learning_rate": 0.0004396,
-      "loss": 0.3771,
+      "loss": 0.3582,
       "step": 1108
     },
     {
       "epoch": 20.16,
       "learning_rate": 0.0004395,
-      "loss": 0.4005,
+      "loss": 0.3582,
       "step": 1109
     },
     {
       "epoch": 20.18,
       "learning_rate": 0.0004394,
-      "loss": 0.3445,
+      "loss": 0.3277,
       "step": 1110
     },
     {
       "epoch": 20.2,
       "learning_rate": 0.00043930000000000005,
-      "loss": 0.2934,
+      "loss": 0.3169,
       "step": 1111
     },
     {
       "epoch": 20.22,
       "learning_rate": 0.0004392,
-      "loss": 0.3087,
+      "loss": 0.3139,
       "step": 1112
     },
     {
       "epoch": 20.24,
       "learning_rate": 0.0004391,
-      "loss": 0.2939,
+      "loss": 0.2819,
       "step": 1113
     },
     {
       "epoch": 20.25,
       "learning_rate": 0.000439,
-      "loss": 0.3229,
+      "loss": 0.3167,
       "step": 1114
     },
     {
       "epoch": 20.27,
       "learning_rate": 0.00043890000000000004,
-      "loss": 0.3392,
+      "loss": 0.3033,
       "step": 1115
     },
     {
       "epoch": 20.29,
       "learning_rate": 0.00043880000000000004,
-      "loss": 0.3602,
+      "loss": 0.3955,
       "step": 1116
     },
     {
       "epoch": 20.31,
       "learning_rate": 0.0004387,
-      "loss": 0.3745,
+      "loss": 0.3714,
       "step": 1117
     },
     {
       "epoch": 20.33,
       "learning_rate": 0.0004386,
-      "loss": 0.3473,
+      "loss": 0.3259,
       "step": 1118
     },
     {
       "epoch": 20.35,
       "learning_rate": 0.00043850000000000003,
-      "loss": 0.3372,
+      "loss": 0.3393,
       "step": 1119
     },
     {
       "epoch": 20.36,
       "learning_rate": 0.00043840000000000003,
-      "loss": 0.3173,
+      "loss": 0.3207,
       "step": 1120
     },
     {
       "epoch": 20.38,
       "learning_rate": 0.00043830000000000003,
-      "loss": 0.3112,
+      "loss": 0.3392,
       "step": 1121
     },
     {
       "epoch": 20.4,
       "learning_rate": 0.00043819999999999997,
-      "loss": 0.3215,
+      "loss": 0.2908,
       "step": 1122
     },
     {
       "epoch": 20.42,
       "learning_rate": 0.0004381,
-      "loss": 0.3949,
+      "loss": 0.3796,
       "step": 1123
     },
     {
       "epoch": 20.44,
       "learning_rate": 0.000438,
-      "loss": 0.3615,
+      "loss": 0.3399,
       "step": 1124
     },
     {
       "epoch": 20.45,
       "learning_rate": 0.0004379,
-      "loss": 0.2959,
+      "loss": 0.2728,
       "step": 1125
     },
     {
       "epoch": 20.47,
       "learning_rate": 0.0004378,
-      "loss": 0.3673,
+      "loss": 0.3686,
       "step": 1126
     },
     {
       "epoch": 20.49,
       "learning_rate": 0.0004377,
-      "loss": 0.3289,
+      "loss": 0.2997,
       "step": 1127
     },
     {
       "epoch": 20.51,
       "learning_rate": 0.0004376,
-      "loss": 0.333,
+      "loss": 0.3159,
       "step": 1128
     },
     {
       "epoch": 20.53,
       "learning_rate": 0.0004375,
-      "loss": 0.2853,
+      "loss": 0.2857,
       "step": 1129
     },
     {
       "epoch": 20.55,
       "learning_rate": 0.0004374,
-      "loss": 0.2675,
+      "loss": 0.333,
       "step": 1130
     },
     {
       "epoch": 20.56,
       "learning_rate": 0.0004373,
-      "loss": 0.3429,
+      "loss": 0.3385,
       "step": 1131
     },
     {
       "epoch": 20.58,
       "learning_rate": 0.0004372,
-      "loss": 0.3689,
+      "loss": 0.3528,
       "step": 1132
     },
     {
       "epoch": 20.6,
       "learning_rate": 0.0004371,
-      "loss": 0.392,
+      "loss": 0.3806,
       "step": 1133
     },
     {
       "epoch": 20.62,
       "learning_rate": 0.000437,
-      "loss": 0.3738,
+      "loss": 0.3737,
       "step": 1134
     },
     {
       "epoch": 20.64,
       "learning_rate": 0.0004369,
-      "loss": 0.337,
+      "loss": 0.3299,
       "step": 1135
     },
     {
       "epoch": 20.65,
       "learning_rate": 0.00043680000000000005,
-      "loss": 0.43,
+      "loss": 0.3757,
       "step": 1136
     },
     {
       "epoch": 20.67,
       "learning_rate": 0.0004367,
-      "loss": 0.394,
+      "loss": 0.388,
       "step": 1137
     },
     {
       "epoch": 20.69,
       "learning_rate": 0.0004366,
-      "loss": 0.3708,
+      "loss": 0.4258,
       "step": 1138
     },
     {
       "epoch": 20.71,
       "learning_rate": 0.0004365,
-      "loss": 0.3716,
+      "loss": 0.3239,
       "step": 1139
     },
     {
       "epoch": 20.73,
       "learning_rate": 0.00043640000000000004,
-      "loss": 0.3508,
+      "loss": 0.3519,
       "step": 1140
     },
     {
       "epoch": 20.75,
       "learning_rate": 0.00043630000000000003,
-      "loss": 0.3325,
+      "loss": 0.3589,
       "step": 1141
     },
     {
       "epoch": 20.76,
       "learning_rate": 0.0004362,
-      "loss": 0.3485,
+      "loss": 0.3319,
       "step": 1142
     },
     {
       "epoch": 20.78,
       "learning_rate": 0.0004361,
-      "loss": 0.3441,
+      "loss": 0.3178,
       "step": 1143
     },
     {
       "epoch": 20.8,
       "learning_rate": 0.000436,
-      "loss": 0.3314,
+      "loss": 0.3638,
       "step": 1144
     },
     {
       "epoch": 20.82,
       "learning_rate": 0.0004359,
-      "loss": 0.3363,
+      "loss": 0.3152,
       "step": 1145
     },
     {
       "epoch": 20.84,
       "learning_rate": 0.0004358,
-      "loss": 0.3367,
+      "loss": 0.3205,
       "step": 1146
     },
     {
       "epoch": 20.85,
       "learning_rate": 0.00043569999999999997,
-      "loss": 0.3023,
+      "loss": 0.2846,
       "step": 1147
     },
     {
       "epoch": 20.87,
       "learning_rate": 0.0004356,
-      "loss": 0.3397,
+      "loss": 0.3123,
       "step": 1148
     },
     {
       "epoch": 20.89,
       "learning_rate": 0.0004355,
-      "loss": 0.4168,
+      "loss": 0.4003,
       "step": 1149
     },
     {
       "epoch": 20.91,
       "learning_rate": 0.0004354,
-      "loss": 0.3766,
+      "loss": 0.3332,
       "step": 1150
     },
     {
       "epoch": 20.93,
       "learning_rate": 0.0004353,
-      "loss": 0.3695,
+      "loss": 0.3602,
       "step": 1151
     },
     {
       "epoch": 20.95,
       "learning_rate": 0.0004352,
-      "loss": 0.3585,
+      "loss": 0.3497,
       "step": 1152
     },
     {
       "epoch": 20.96,
       "learning_rate": 0.0004351,
-      "loss": 0.3678,
+      "loss": 0.3351,
       "step": 1153
     },
     {
       "epoch": 20.98,
       "learning_rate": 0.000435,
-      "loss": 0.331,
+      "loss": 0.3173,
       "step": 1154
     },
     {
       "epoch": 21.0,
       "learning_rate": 0.0004349,
-      "loss": 0.3555,
+      "loss": 0.3205,
       "step": 1155
     },
     {
       "epoch": 21.02,
       "learning_rate": 0.00043480000000000005,
-      "loss": 0.3184,
+      "loss": 0.3488,
       "step": 1156
     },
     {
       "epoch": 21.04,
       "learning_rate": 0.0004347,
-      "loss": 0.3231,
+      "loss": 0.3521,
       "step": 1157
     },
     {
       "epoch": 21.05,
       "learning_rate": 0.0004346,
-      "loss": 0.3135,
+      "loss": 0.2992,
       "step": 1158
     },
     {
       "epoch": 21.07,
       "learning_rate": 0.0004345,
-      "loss": 0.2855,
+      "loss": 0.3125,
       "step": 1159
     },
     {
       "epoch": 21.09,
       "learning_rate": 0.00043440000000000004,
-      "loss": 0.3959,
+      "loss": 0.417,
       "step": 1160
     },
     {
       "epoch": 21.11,
       "learning_rate": 0.00043430000000000004,
-      "loss": 0.312,
+      "loss": 0.3104,
       "step": 1161
     },
     {
       "epoch": 21.13,
       "learning_rate": 0.0004342,
-      "loss": 0.3335,
+      "loss": 0.3135,
       "step": 1162
     },
     {
       "epoch": 21.15,
       "learning_rate": 0.0004341,
-      "loss": 0.3105,
+      "loss": 0.3101,
       "step": 1163
     },
     {
       "epoch": 21.16,
       "learning_rate": 0.00043400000000000003,
-      "loss": 0.2914,
+      "loss": 0.2919,
       "step": 1164
     },
     {
       "epoch": 21.18,
       "learning_rate": 0.00043390000000000003,
-      "loss": 0.3781,
+      "loss": 0.3998,
       "step": 1165
     },
     {
       "epoch": 21.2,
       "learning_rate": 0.00043380000000000003,
-      "loss": 0.3422,
+      "loss": 0.3306,
       "step": 1166
     },
     {
       "epoch": 21.22,
       "learning_rate": 0.00043369999999999997,
-      "loss": 0.3476,
+      "loss": 0.3447,
       "step": 1167
     },
     {
       "epoch": 21.24,
       "learning_rate": 0.00043359999999999997,
-      "loss": 0.3327,
+      "loss": 0.3113,
       "step": 1168
     },
     {
       "epoch": 21.25,
       "learning_rate": 0.0004335,
-      "loss": 0.3568,
+      "loss": 0.3562,
       "step": 1169
     },
     {
       "epoch": 21.27,
       "learning_rate": 0.0004334,
-      "loss": 0.3333,
+      "loss": 0.3628,
       "step": 1170
     },
     {
       "epoch": 21.29,
       "learning_rate": 0.0004333,
-      "loss": 0.3129,
+      "loss": 0.3225,
       "step": 1171
     },
     {
       "epoch": 21.31,
       "learning_rate": 0.00043319999999999996,
-      "loss": 0.3117,
+      "loss": 0.298,
       "step": 1172
     },
     {
       "epoch": 21.33,
       "learning_rate": 0.0004331,
-      "loss": 0.297,
+      "loss": 0.3198,
       "step": 1173
     },
     {
       "epoch": 21.35,
       "learning_rate": 0.000433,
-      "loss": 0.3855,
+      "loss": 0.4005,
       "step": 1174
     },
     {
       "epoch": 21.36,
       "learning_rate": 0.0004329,
-      "loss": 0.3485,
+      "loss": 0.3205,
       "step": 1175
     },
     {
       "epoch": 21.38,
       "learning_rate": 0.0004328,
-      "loss": 0.3383,
+      "loss": 0.3064,
       "step": 1176
     },
     {
       "epoch": 21.4,
       "learning_rate": 0.0004327,
-      "loss": 0.3591,
+      "loss": 0.3541,
       "step": 1177
     },
     {
       "epoch": 21.42,
       "learning_rate": 0.0004326,
-      "loss": 0.3264,
+      "loss": 0.3335,
       "step": 1178
     },
     {
       "epoch": 21.44,
       "learning_rate": 0.0004325,
-      "loss": 0.3254,
+      "loss": 0.3069,
       "step": 1179
     },
     {
       "epoch": 21.45,
       "learning_rate": 0.0004324,
-      "loss": 0.3354,
+      "loss": 0.3229,
       "step": 1180
     },
     {
       "epoch": 21.47,
       "learning_rate": 0.00043230000000000005,
-      "loss": 0.3155,
+      "loss": 0.3376,
       "step": 1181
     },
     {
       "epoch": 21.49,
       "learning_rate": 0.0004322,
-      "loss": 0.3225,
+      "loss": 0.3188,
       "step": 1182
     },
     {
       "epoch": 21.51,
       "learning_rate": 0.0004321,
-      "loss": 0.3385,
+      "loss": 0.3404,
       "step": 1183
     },
     {
       "epoch": 21.53,
       "learning_rate": 0.000432,
-      "loss": 0.361,
+      "loss": 0.3486,
       "step": 1184
     },
     {
       "epoch": 21.55,
       "learning_rate": 0.00043190000000000004,
-      "loss": 0.2877,
+      "loss": 0.2621,
       "step": 1185
     },
     {
       "epoch": 21.56,
       "learning_rate": 0.00043180000000000003,
-      "loss": 0.4098,
+      "loss": 0.3896,
       "step": 1186
     },
     {
       "epoch": 21.58,
       "learning_rate": 0.0004317,
-      "loss": 0.3315,
+      "loss": 0.3499,
       "step": 1187
     },
     {
       "epoch": 21.6,
       "learning_rate": 0.0004316,
-      "loss": 0.3001,
+      "loss": 0.3189,
       "step": 1188
     },
     {
       "epoch": 21.62,
       "learning_rate": 0.0004315,
-      "loss": 0.3625,
+      "loss": 0.3635,
       "step": 1189
     },
     {
       "epoch": 21.64,
       "learning_rate": 0.0004314,
-      "loss": 0.3068,
+      "loss": 0.3123,
       "step": 1190
     },
     {
       "epoch": 21.65,
       "learning_rate": 0.0004313,
-      "loss": 0.3184,
+      "loss": 0.3366,
       "step": 1191
     },
     {
       "epoch": 21.67,
       "learning_rate": 0.0004312,
-      "loss": 0.3407,
+      "loss": 0.3446,
       "step": 1192
     },
     {
       "epoch": 21.69,
       "learning_rate": 0.0004311,
-      "loss": 0.3485,
+      "loss": 0.3352,
       "step": 1193
     },
     {
       "epoch": 21.71,
       "learning_rate": 0.000431,
-      "loss": 0.3026,
+      "loss": 0.3084,
       "step": 1194
     },
     {
       "epoch": 21.73,
       "learning_rate": 0.0004309,
-      "loss": 0.2983,
+      "loss": 0.3113,
       "step": 1195
     },
     {
       "epoch": 21.75,
       "learning_rate": 0.0004308,
-      "loss": 0.3029,
+      "loss": 0.3039,
       "step": 1196
     },
     {
       "epoch": 21.76,
       "learning_rate": 0.00043070000000000006,
-      "loss": 0.369,
+      "loss": 0.3864,
       "step": 1197
     },
     {
       "epoch": 21.78,
       "learning_rate": 0.0004306,
-      "loss": 0.2752,
+      "loss": 0.3015,
       "step": 1198
     },
     {
       "epoch": 21.8,
       "learning_rate": 0.0004305,
-      "loss": 0.3244,
+      "loss": 0.3484,
       "step": 1199
     },
     {
       "epoch": 21.82,
       "learning_rate": 0.0004304,
-      "loss": 0.365,
+      "loss": 0.3758,
       "step": 1200
     },
     {
       "epoch": 21.84,
       "learning_rate": 0.00043030000000000005,
-      "loss": 0.2619,
+      "loss": 0.2515,
       "step": 1201
     },
     {
       "epoch": 21.85,
       "learning_rate": 0.00043020000000000005,
-      "loss": 0.3071,
+      "loss": 0.3368,
       "step": 1202
     },
     {
       "epoch": 21.87,
       "learning_rate": 0.0004301,
-      "loss": 0.3496,
+      "loss": 0.371,
       "step": 1203
     },
     {
       "epoch": 21.89,
       "learning_rate": 0.00043,
-      "loss": 0.3945,
+      "loss": 0.3802,
       "step": 1204
     },
     {
       "epoch": 21.91,
       "learning_rate": 0.0004299,
-      "loss": 0.2969,
+      "loss": 0.353,
       "step": 1205
     },
     {
       "epoch": 21.93,
       "learning_rate": 0.00042980000000000004,
-      "loss": 0.3987,
+      "loss": 0.3982,
       "step": 1206
     },
     {
       "epoch": 21.95,
       "learning_rate": 0.00042970000000000004,
-      "loss": 0.3485,
+      "loss": 0.3598,
       "step": 1207
     },
     {
       "epoch": 21.96,
       "learning_rate": 0.0004296,
-      "loss": 0.2979,
+      "loss": 0.2958,
       "step": 1208
     },
     {
       "epoch": 21.98,
       "learning_rate": 0.0004295,
-      "loss": 0.3502,
+      "loss": 0.3639,
       "step": 1209
     },
     {
       "epoch": 22.0,
       "learning_rate": 0.00042940000000000003,
-      "loss": 0.3499,
+      "loss": 0.3679,
       "step": 1210
     },
     {
       "epoch": 22.02,
       "learning_rate": 0.0004293,
-      "loss": 0.3377,
+      "loss": 0.3751,
       "step": 1211
     },
     {
       "epoch": 22.04,
       "learning_rate": 0.0004292,
-      "loss": 0.2766,
+      "loss": 0.2787,
       "step": 1212
     },
     {
       "epoch": 22.05,
       "learning_rate": 0.00042909999999999997,
-      "loss": 0.3308,
+      "loss": 0.3078,
       "step": 1213
     },
     {
       "epoch": 22.07,
       "learning_rate": 0.000429,
-      "loss": 0.3066,
+      "loss": 0.286,
       "step": 1214
     },
     {
       "epoch": 22.09,
       "learning_rate": 0.0004289,
-      "loss": 0.3074,
+      "loss": 0.3562,
       "step": 1215
     },
     {
       "epoch": 22.11,
       "learning_rate": 0.0004288,
-      "loss": 0.3311,
+      "loss": 0.3346,
       "step": 1216
     },
     {
       "epoch": 22.13,
       "learning_rate": 0.0004287,
-      "loss": 0.2503,
+      "loss": 0.2584,
       "step": 1217
     },
     {
       "epoch": 22.15,
       "learning_rate": 0.0004286,
-      "loss": 0.2989,
+      "loss": 0.3668,
       "step": 1218
     },
     {
       "epoch": 22.16,
       "learning_rate": 0.0004285,
-      "loss": 0.3422,
+      "loss": 0.3774,
       "step": 1219
     },
     {
       "epoch": 22.18,
       "learning_rate": 0.0004284,
-      "loss": 0.3226,
+      "loss": 0.3279,
       "step": 1220
     },
     {
       "epoch": 22.2,
       "learning_rate": 0.0004283,
-      "loss": 0.3328,
+      "loss": 0.3872,
       "step": 1221
     },
     {
       "epoch": 22.22,
       "learning_rate": 0.00042820000000000005,
-      "loss": 0.3786,
+      "loss": 0.3461,
       "step": 1222
     },
     {
       "epoch": 22.24,
       "learning_rate": 0.0004281,
-      "loss": 0.2952,
+      "loss": 0.3196,
       "step": 1223
     },
     {
       "epoch": 22.25,
       "learning_rate": 0.000428,
-      "loss": 0.2944,
+      "loss": 0.3133,
       "step": 1224
     },
     {
       "epoch": 22.27,
       "learning_rate": 0.0004279,
-      "loss": 0.3299,
+      "loss": 0.3054,
       "step": 1225
     },
     {
       "epoch": 22.29,
       "learning_rate": 0.00042780000000000004,
-      "loss": 0.339,
+      "loss": 0.3388,
       "step": 1226
     },
     {
       "epoch": 22.31,
       "learning_rate": 0.00042770000000000004,
-      "loss": 0.3231,
+      "loss": 0.3814,
       "step": 1227
     },
     {
       "epoch": 22.33,
       "learning_rate": 0.0004276,
-      "loss": 0.3153,
+      "loss": 0.3266,
       "step": 1228
     },
     {
       "epoch": 22.35,
       "learning_rate": 0.0004275,
-      "loss": 0.3399,
+      "loss": 0.3579,
       "step": 1229
     },
     {
       "epoch": 22.36,
       "learning_rate": 0.00042740000000000003,
-      "loss": 0.2985,
+      "loss": 0.3148,
       "step": 1230
     },
     {
       "epoch": 22.38,
       "learning_rate": 0.00042730000000000003,
-      "loss": 0.3215,
+      "loss": 0.3436,
       "step": 1231
     },
     {
       "epoch": 22.4,
       "learning_rate": 0.00042720000000000003,
-      "loss": 0.3261,
+      "loss": 0.3268,
       "step": 1232
     },
     {
       "epoch": 22.42,
       "learning_rate": 0.0004271,
-      "loss": 0.3268,
+      "loss": 0.3638,
       "step": 1233
     },
     {
       "epoch": 22.44,
       "learning_rate": 0.000427,
-      "loss": 0.3002,
+      "loss": 0.3628,
       "step": 1234
     },
     {
       "epoch": 22.45,
       "learning_rate": 0.0004269,
-      "loss": 0.3056,
+      "loss": 0.2675,
       "step": 1235
     },
     {
       "epoch": 22.47,
       "learning_rate": 0.0004268,
-      "loss": 0.363,
+      "loss": 0.3539,
       "step": 1236
     },
     {
       "epoch": 22.49,
       "learning_rate": 0.0004267,
-      "loss": 0.3199,
+      "loss": 0.2967,
       "step": 1237
     },
     {
       "epoch": 22.51,
       "learning_rate": 0.00042659999999999996,
-      "loss": 0.3654,
+      "loss": 0.3653,
       "step": 1238
     },
     {
       "epoch": 22.53,
       "learning_rate": 0.0004265,
-      "loss": 0.3082,
+      "loss": 0.2861,
       "step": 1239
     },
     {
       "epoch": 22.55,
       "learning_rate": 0.0004264,
-      "loss": 0.3288,
+      "loss": 0.3145,
       "step": 1240
     },
     {
       "epoch": 22.56,
       "learning_rate": 0.0004263,
-      "loss": 0.3706,
+      "loss": 0.4007,
       "step": 1241
     },
     {
       "epoch": 22.58,
       "learning_rate": 0.0004262,
-      "loss": 0.3445,
+      "loss": 0.351,
       "step": 1242
     },
     {
       "epoch": 22.6,
       "learning_rate": 0.0004261,
-      "loss": 0.3291,
+      "loss": 0.2838,
       "step": 1243
     },
     {
       "epoch": 22.62,
       "learning_rate": 0.000426,
-      "loss": 0.3534,
+      "loss": 0.3596,
       "step": 1244
     },
     {
       "epoch": 22.64,
       "learning_rate": 0.0004259,
-      "loss": 0.3395,
+      "loss": 0.3041,
       "step": 1245
     },
     {
       "epoch": 22.65,
       "learning_rate": 0.0004258,
-      "loss": 0.3186,
+      "loss": 0.3254,
       "step": 1246
     },
     {
       "epoch": 22.67,
       "learning_rate": 0.00042570000000000005,
-      "loss": 0.3406,
+      "loss": 0.321,
       "step": 1247
     },
     {
       "epoch": 22.69,
       "learning_rate": 0.0004256,
-      "loss": 0.3143,
+      "loss": 0.3411,
       "step": 1248
     },
     {
       "epoch": 22.71,
       "learning_rate": 0.0004255,
-      "loss": 0.3099,
+      "loss": 0.3375,
       "step": 1249
     },
     {
       "epoch": 22.73,
       "learning_rate": 0.0004254,
-      "loss": 0.3032,
+      "loss": 0.3175,
       "step": 1250
     },
     {
       "epoch": 22.75,
       "learning_rate": 0.00042530000000000004,
-      "loss": 0.2782,
+      "loss": 0.2996,
       "step": 1251
     },
     {
       "epoch": 22.76,
       "learning_rate": 0.00042520000000000004,
-      "loss": 0.3324,
+      "loss": 0.3516,
       "step": 1252
     },
     {
       "epoch": 22.78,
       "learning_rate": 0.0004251,
-      "loss": 0.3116,
+      "loss": 0.3031,
       "step": 1253
     },
     {
       "epoch": 22.8,
       "learning_rate": 0.000425,
-      "loss": 0.2712,
+      "loss": 0.2806,
       "step": 1254
     },
     {
       "epoch": 22.82,
       "learning_rate": 0.00042490000000000003,
-      "loss": 0.3362,
+      "loss": 0.3361,
       "step": 1255
     },
     {
       "epoch": 22.84,
       "learning_rate": 0.0004248,
-      "loss": 0.3146,
+      "loss": 0.3125,
       "step": 1256
     },
     {
       "epoch": 22.85,
       "learning_rate": 0.0004247,
-      "loss": 0.3779,
+      "loss": 0.4131,
       "step": 1257
     },
     {
       "epoch": 22.87,
       "learning_rate": 0.00042459999999999997,
-      "loss": 0.2531,
+      "loss": 0.2709,
       "step": 1258
     },
     {
       "epoch": 22.89,
       "learning_rate": 0.0004245,
-      "loss": 0.2798,
+      "loss": 0.2886,
       "step": 1259
     },
     {
       "epoch": 22.91,
       "learning_rate": 0.0004244,
-      "loss": 0.2708,
+      "loss": 0.3089,
       "step": 1260
     },
     {
       "epoch": 22.93,
       "learning_rate": 0.0004243,
-      "loss": 0.3059,
+      "loss": 0.3098,
       "step": 1261
     },
     {
       "epoch": 22.95,
       "learning_rate": 0.0004242,
-      "loss": 0.3352,
+      "loss": 0.343,
       "step": 1262
     },
     {
       "epoch": 22.96,
       "learning_rate": 0.0004241,
-      "loss": 0.3348,
+      "loss": 0.3121,
       "step": 1263
     },
     {
       "epoch": 22.98,
       "learning_rate": 0.000424,
-      "loss": 0.3215,
+      "loss": 0.369,
       "step": 1264
     },
     {
       "epoch": 23.0,
       "learning_rate": 0.0004239,
-      "loss": 0.375,
+      "loss": 0.3609,
       "step": 1265
     },
     {
       "epoch": 23.02,
       "learning_rate": 0.0004238,
-      "loss": 0.3162,
+      "loss": 0.3149,
       "step": 1266
     },
     {
       "epoch": 23.04,
       "learning_rate": 0.00042370000000000005,
-      "loss": 0.3585,
+      "loss": 0.3547,
       "step": 1267
     },
     {
       "epoch": 23.05,
       "learning_rate": 0.0004236,
-      "loss": 0.3113,
+      "loss": 0.346,
       "step": 1268
     },
     {
       "epoch": 23.07,
       "learning_rate": 0.0004235,
-      "loss": 0.3289,
+      "loss": 0.3403,
       "step": 1269
     },
     {
       "epoch": 23.09,
       "learning_rate": 0.0004234,
-      "loss": 0.2673,
+      "loss": 0.3111,
       "step": 1270
     },
     {
       "epoch": 23.11,
       "learning_rate": 0.00042330000000000004,
-      "loss": 0.2653,
+      "loss": 0.287,
       "step": 1271
     },
     {
       "epoch": 23.13,
       "learning_rate": 0.00042320000000000004,
-      "loss": 0.2735,
+      "loss": 0.2829,
       "step": 1272
     },
     {
       "epoch": 23.15,
       "learning_rate": 0.0004231,
-      "loss": 0.3108,
+      "loss": 0.3485,
       "step": 1273
     },
     {
       "epoch": 23.16,
       "learning_rate": 0.000423,
-      "loss": 0.3263,
+      "loss": 0.3326,
       "step": 1274
     },
     {
       "epoch": 23.18,
       "learning_rate": 0.00042290000000000003,
-      "loss": 0.3057,
+      "loss": 0.3133,
       "step": 1275
     },
     {
       "epoch": 23.2,
       "learning_rate": 0.00042280000000000003,
-      "loss": 0.334,
+      "loss": 0.335,
       "step": 1276
     },
     {
       "epoch": 23.22,
       "learning_rate": 0.00042270000000000003,
-      "loss": 0.35,
+      "loss": 0.337,
       "step": 1277
     },
     {
       "epoch": 23.24,
       "learning_rate": 0.00042259999999999997,
-      "loss": 0.2612,
+      "loss": 0.2765,
       "step": 1278
     },
     {
       "epoch": 23.25,
       "learning_rate": 0.00042249999999999997,
-      "loss": 0.2972,
+      "loss": 0.301,
       "step": 1279
     },
     {
       "epoch": 23.27,
       "learning_rate": 0.0004224,
-      "loss": 0.3303,
+      "loss": 0.3398,
       "step": 1280
     },
     {
       "epoch": 23.29,
       "learning_rate": 0.0004223,
-      "loss": 0.2958,
+      "loss": 0.2941,
       "step": 1281
     },
     {
       "epoch": 23.31,
       "learning_rate": 0.0004222,
-      "loss": 0.2802,
+      "loss": 0.2913,
       "step": 1282
     },
     {
       "epoch": 23.33,
       "learning_rate": 0.00042209999999999996,
-      "loss": 0.2715,
+      "loss": 0.3078,
       "step": 1283
     },
     {
       "epoch": 23.35,
       "learning_rate": 0.000422,
-      "loss": 0.3491,
+      "loss": 0.3169,
       "step": 1284
     },
     {
       "epoch": 23.36,
       "learning_rate": 0.0004219,
-      "loss": 0.3401,
+      "loss": 0.3485,
       "step": 1285
     },
     {
       "epoch": 23.38,
       "learning_rate": 0.0004218,
-      "loss": 0.2667,
+      "loss": 0.2792,
       "step": 1286
     },
     {
       "epoch": 23.4,
       "learning_rate": 0.0004217,
-      "loss": 0.2973,
+      "loss": 0.3224,
       "step": 1287
     },
     {
       "epoch": 23.42,
       "learning_rate": 0.0004216,
-      "loss": 0.3052,
+      "loss": 0.2947,
       "step": 1288
     },
     {
       "epoch": 23.44,
       "learning_rate": 0.0004215,
-      "loss": 0.2941,
+      "loss": 0.3182,
       "step": 1289
     },
     {
       "epoch": 23.45,
       "learning_rate": 0.0004214,
-      "loss": 0.2973,
+      "loss": 0.2633,
       "step": 1290
     },
     {
       "epoch": 23.47,
       "learning_rate": 0.0004213,
-      "loss": 0.325,
+      "loss": 0.3377,
       "step": 1291
     },
     {
       "epoch": 23.49,
       "learning_rate": 0.00042120000000000005,
-      "loss": 0.3307,
+      "loss": 0.3558,
       "step": 1292
     },
     {
       "epoch": 23.51,
       "learning_rate": 0.0004211,
-      "loss": 0.3391,
+      "loss": 0.3094,
       "step": 1293
     },
     {
       "epoch": 23.53,
       "learning_rate": 0.000421,
-      "loss": 0.3,
+      "loss": 0.3117,
       "step": 1294
     },
     {
       "epoch": 23.55,
       "learning_rate": 0.0004209,
-      "loss": 0.3118,
+      "loss": 0.3153,
       "step": 1295
     },
     {
       "epoch": 23.56,
       "learning_rate": 0.00042080000000000004,
-      "loss": 0.3138,
+      "loss": 0.2782,
       "step": 1296
     },
     {
       "epoch": 23.58,
       "learning_rate": 0.00042070000000000003,
-      "loss": 0.2804,
+      "loss": 0.2921,
       "step": 1297
     },
     {
       "epoch": 23.6,
       "learning_rate": 0.0004206,
-      "loss": 0.29,
+      "loss": 0.2856,
       "step": 1298
     },
     {
       "epoch": 23.62,
       "learning_rate": 0.0004205,
-      "loss": 0.3379,
+      "loss": 0.3304,
       "step": 1299
     },
     {
       "epoch": 23.64,
       "learning_rate": 0.00042040000000000003,
-      "loss": 0.3337,
+      "loss": 0.3514,
       "step": 1300
     },
     {
       "epoch": 23.65,
       "learning_rate": 0.0004203,
-      "loss": 0.3144,
+      "loss": 0.2963,
       "step": 1301
     },
     {
       "epoch": 23.67,
       "learning_rate": 0.0004202,
-      "loss": 0.3943,
+      "loss": 0.3359,
       "step": 1302
     },
     {
       "epoch": 23.69,
       "learning_rate": 0.00042009999999999997,
-      "loss": 0.3107,
+      "loss": 0.3079,
       "step": 1303
     },
     {
       "epoch": 23.71,
       "learning_rate": 0.00042,
-      "loss": 0.3282,
+      "loss": 0.2833,
       "step": 1304
     },
     {
       "epoch": 23.73,
       "learning_rate": 0.0004199,
-      "loss": 0.331,
+      "loss": 0.3094,
       "step": 1305
     },
     {
       "epoch": 23.75,
       "learning_rate": 0.0004198,
-      "loss": 0.297,
+      "loss": 0.2909,
       "step": 1306
     },
     {
       "epoch": 23.76,
       "learning_rate": 0.0004197,
-      "loss": 0.3514,
+      "loss": 0.3297,
       "step": 1307
     },
     {
       "epoch": 23.78,
       "learning_rate": 0.0004196,
-      "loss": 0.3094,
+      "loss": 0.2878,
       "step": 1308
     },
     {
       "epoch": 23.8,
       "learning_rate": 0.0004195,
-      "loss": 0.3151,
+      "loss": 0.2707,
       "step": 1309
     },
     {
       "epoch": 23.82,
       "learning_rate": 0.0004194,
-      "loss": 0.3352,
+      "loss": 0.34,
       "step": 1310
     },
     {
       "epoch": 23.84,
       "learning_rate": 0.0004193,
-      "loss": 0.3136,
+      "loss": 0.296,
       "step": 1311
     },
     {
       "epoch": 23.85,
       "learning_rate": 0.00041920000000000005,
-      "loss": 0.2915,
+      "loss": 0.2911,
       "step": 1312
     },
     {
       "epoch": 23.87,
       "learning_rate": 0.0004191,
-      "loss": 0.2637,
+      "loss": 0.2518,
       "step": 1313
     },
     {
       "epoch": 23.89,
       "learning_rate": 0.000419,
-      "loss": 0.3347,
+      "loss": 0.3426,
       "step": 1314
     },
     {
       "epoch": 23.91,
       "learning_rate": 0.0004189,
-      "loss": 0.2917,
+      "loss": 0.2668,
       "step": 1315
     },
     {
       "epoch": 23.93,
       "learning_rate": 0.0004188,
-      "loss": 0.3642,
+      "loss": 0.2945,
       "step": 1316
     },
     {
       "epoch": 23.95,
       "learning_rate": 0.00041870000000000004,
-      "loss": 0.3046,
+      "loss": 0.3062,
       "step": 1317
     },
     {
       "epoch": 23.96,
       "learning_rate": 0.00041860000000000004,
-      "loss": 0.3094,
+      "loss": 0.3049,
       "step": 1318
     },
     {
       "epoch": 23.98,
       "learning_rate": 0.0004185,
-      "loss": 0.3391,
+      "loss": 0.2933,
       "step": 1319
     },
     {
       "epoch": 24.0,
       "learning_rate": 0.0004184,
-      "loss": 0.2268,
+      "loss": 0.2173,
       "step": 1320
     },
     {
       "epoch": 24.02,
       "learning_rate": 0.00041830000000000003,
-      "loss": 0.3322,
+      "loss": 0.3141,
       "step": 1321
     },
     {
       "epoch": 24.04,
       "learning_rate": 0.00041820000000000003,
-      "loss": 0.316,
+      "loss": 0.3327,
       "step": 1322
     },
     {
       "epoch": 24.05,
       "learning_rate": 0.0004181,
-      "loss": 0.291,
+      "loss": 0.2945,
       "step": 1323
     },
     {
       "epoch": 24.07,
       "learning_rate": 0.00041799999999999997,
-      "loss": 0.329,
+      "loss": 0.2612,
       "step": 1324
     },
     {
       "epoch": 24.09,
       "learning_rate": 0.0004179,
-      "loss": 0.3547,
+      "loss": 0.3348,
       "step": 1325
     },
     {
       "epoch": 24.11,
       "learning_rate": 0.0004178,
-      "loss": 0.3304,
+      "loss": 0.3164,
       "step": 1326
     },
     {
       "epoch": 24.13,
       "learning_rate": 0.0004177,
-      "loss": 0.3348,
+      "loss": 0.3685,
       "step": 1327
     },
     {
       "epoch": 24.15,
       "learning_rate": 0.0004176,
-      "loss": 0.3156,
+      "loss": 0.2791,
       "step": 1328
     },
     {
       "epoch": 24.16,
       "learning_rate": 0.0004175,
-      "loss": 0.3048,
+      "loss": 0.2915,
       "step": 1329
     },
     {
       "epoch": 24.18,
       "learning_rate": 0.0004174,
-      "loss": 0.2972,
+      "loss": 0.3097,
       "step": 1330
     },
     {
       "epoch": 24.2,
       "learning_rate": 0.0004173,
-      "loss": 0.3089,
+      "loss": 0.2441,
       "step": 1331
     },
     {
       "epoch": 24.22,
       "learning_rate": 0.0004172,
-      "loss": 0.2649,
+      "loss": 0.3241,
       "step": 1332
     },
     {
       "epoch": 24.24,
       "learning_rate": 0.00041710000000000006,
-      "loss": 0.242,
+      "loss": 0.2623,
       "step": 1333
     },
     {
       "epoch": 24.25,
       "learning_rate": 0.000417,
-      "loss": 0.3369,
+      "loss": 0.3325,
       "step": 1334
     },
     {
       "epoch": 24.27,
       "learning_rate": 0.0004169,
-      "loss": 0.2826,
+      "loss": 0.2749,
       "step": 1335
     },
     {
       "epoch": 24.29,
       "learning_rate": 0.0004168,
-      "loss": 0.3018,
+      "loss": 0.2502,
       "step": 1336
     },
     {
       "epoch": 24.31,
       "learning_rate": 0.00041670000000000005,
-      "loss": 0.2592,
+      "loss": 0.2566,
       "step": 1337
     },
     {
       "epoch": 24.33,
       "learning_rate": 0.00041660000000000004,
-      "loss": 0.2988,
+      "loss": 0.3215,
       "step": 1338
     },
     {
       "epoch": 24.35,
       "learning_rate": 0.0004165,
-      "loss": 0.2946,
+      "loss": 0.26,
       "step": 1339
     },
     {
       "epoch": 24.36,
       "learning_rate": 0.0004164,
-      "loss": 0.2881,
+      "loss": 0.2663,
       "step": 1340
     },
     {
       "epoch": 24.38,
       "learning_rate": 0.00041630000000000004,
-      "loss": 0.2918,
+      "loss": 0.2897,
       "step": 1341
     },
     {
       "epoch": 24.4,
       "learning_rate": 0.00041620000000000003,
-      "loss": 0.3425,
+      "loss": 0.3401,
       "step": 1342
     },
     {
       "epoch": 24.42,
       "learning_rate": 0.00041610000000000003,
-      "loss": 0.3171,
+      "loss": 0.2869,
       "step": 1343
     },
     {
       "epoch": 24.44,
       "learning_rate": 0.000416,
-      "loss": 0.3095,
+      "loss": 0.2957,
       "step": 1344
     },
     {
       "epoch": 24.45,
       "learning_rate": 0.0004159,
-      "loss": 0.3539,
+      "loss": 0.3552,
       "step": 1345
     },
     {
       "epoch": 24.47,
       "learning_rate": 0.0004158,
-      "loss": 0.3687,
+      "loss": 0.3618,
       "step": 1346
     },
     {
       "epoch": 24.49,
       "learning_rate": 0.0004157,
-      "loss": 0.3651,
+      "loss": 0.3407,
       "step": 1347
     },
     {
       "epoch": 24.51,
       "learning_rate": 0.0004156,
-      "loss": 0.3161,
+      "loss": 0.3322,
       "step": 1348
     },
     {
       "epoch": 24.53,
       "learning_rate": 0.00041549999999999996,
-      "loss": 0.3148,
+      "loss": 0.3117,
       "step": 1349
     },
     {
       "epoch": 24.55,
       "learning_rate": 0.0004154,
-      "loss": 0.3142,
+      "loss": 0.3174,
       "step": 1350
     },
     {
       "epoch": 24.56,
       "learning_rate": 0.0004153,
-      "loss": 0.2906,
+      "loss": 0.2844,
       "step": 1351
     },
     {
       "epoch": 24.58,
       "learning_rate": 0.0004152,
-      "loss": 0.3358,
+      "loss": 0.3577,
       "step": 1352
     },
     {
       "epoch": 24.6,
       "learning_rate": 0.0004151,
-      "loss": 0.3016,
+      "loss": 0.2877,
       "step": 1353
     },
     {
       "epoch": 24.62,
       "learning_rate": 0.000415,
-      "loss": 0.3202,
+      "loss": 0.2791,
       "step": 1354
     },
     {
       "epoch": 24.64,
       "learning_rate": 0.0004149,
-      "loss": 0.363,
+      "loss": 0.3377,
       "step": 1355
     },
     {
       "epoch": 24.65,
       "learning_rate": 0.0004148,
-      "loss": 0.3417,
+      "loss": 0.3318,
       "step": 1356
     },
     {
       "epoch": 24.67,
       "learning_rate": 0.0004147,
-      "loss": 0.2842,
+      "loss": 0.2862,
       "step": 1357
     },
     {
       "epoch": 24.69,
       "learning_rate": 0.00041460000000000005,
-      "loss": 0.338,
+      "loss": 0.2994,
       "step": 1358
     },
     {
       "epoch": 24.71,
       "learning_rate": 0.0004145,
-      "loss": 0.3374,
+      "loss": 0.3217,
       "step": 1359
     },
     {
       "epoch": 24.73,
       "learning_rate": 0.0004144,
-      "loss": 0.2996,
+      "loss": 0.281,
       "step": 1360
     },
     {
       "epoch": 24.75,
       "learning_rate": 0.0004143,
-      "loss": 0.3728,
+      "loss": 0.3568,
       "step": 1361
     },
     {
       "epoch": 24.76,
       "learning_rate": 0.00041420000000000004,
-      "loss": 0.2509,
+      "loss": 0.1998,
       "step": 1362
     },
     {
       "epoch": 24.78,
       "learning_rate": 0.00041410000000000004,
-      "loss": 0.3519,
+      "loss": 0.3449,
       "step": 1363
     },
     {
       "epoch": 24.8,
       "learning_rate": 0.000414,
-      "loss": 0.2996,
+      "loss": 0.2674,
       "step": 1364
     },
     {
       "epoch": 24.82,
       "learning_rate": 0.0004139,
-      "loss": 0.3258,
+      "loss": 0.3279,
       "step": 1365
     },
     {
       "epoch": 24.84,
       "learning_rate": 0.00041380000000000003,
-      "loss": 0.2714,
+      "loss": 0.2924,
       "step": 1366
     },
     {
       "epoch": 24.85,
       "learning_rate": 0.0004137,
-      "loss": 0.2973,
+      "loss": 0.3154,
       "step": 1367
     },
     {
       "epoch": 24.87,
       "learning_rate": 0.0004136,
-      "loss": 0.3199,
+      "loss": 0.3177,
       "step": 1368
     },
     {
       "epoch": 24.89,
       "learning_rate": 0.00041349999999999997,
-      "loss": 0.3286,
+      "loss": 0.3084,
       "step": 1369
     },
     {
       "epoch": 24.91,
       "learning_rate": 0.0004134,
-      "loss": 0.3168,
+      "loss": 0.2803,
       "step": 1370
     },
     {
       "epoch": 24.93,
       "learning_rate": 0.0004133,
-      "loss": 0.3168,
+      "loss": 0.2949,
       "step": 1371
     },
     {
       "epoch": 24.95,
       "learning_rate": 0.0004132,
-      "loss": 0.3208,
+      "loss": 0.3065,
       "step": 1372
     },
     {
       "epoch": 24.96,
       "learning_rate": 0.0004131,
-      "loss": 0.2743,
+      "loss": 0.2865,
       "step": 1373
     },
     {
       "epoch": 24.98,
       "learning_rate": 0.000413,
-      "loss": 0.3069,
+      "loss": 0.3344,
       "step": 1374
     },
     {
       "epoch": 25.0,
       "learning_rate": 0.0004129,
-      "loss": 0.2925,
+      "loss": 0.2553,
       "step": 1375
     },
     {
       "epoch": 25.02,
       "learning_rate": 0.0004128,
-      "loss": 0.2983,
+      "loss": 0.2922,
       "step": 1376
     },
     {
       "epoch": 25.04,
       "learning_rate": 0.0004127,
-      "loss": 0.3071,
+      "loss": 0.2836,
       "step": 1377
     },
     {
       "epoch": 25.05,
       "learning_rate": 0.00041260000000000005,
-      "loss": 0.296,
+      "loss": 0.2749,
       "step": 1378
     },
     {
       "epoch": 25.07,
       "learning_rate": 0.0004125,
-      "loss": 0.2877,
+      "loss": 0.3183,
       "step": 1379
     },
     {
       "epoch": 25.09,
       "learning_rate": 0.0004124,
-      "loss": 0.3446,
+      "loss": 0.2882,
       "step": 1380
     },
     {
       "epoch": 25.11,
       "learning_rate": 0.0004123,
-      "loss": 0.2842,
+      "loss": 0.2966,
       "step": 1381
     },
     {
       "epoch": 25.13,
       "learning_rate": 0.00041220000000000004,
-      "loss": 0.241,
+      "loss": 0.2481,
       "step": 1382
     },
     {
       "epoch": 25.15,
       "learning_rate": 0.00041210000000000004,
-      "loss": 0.2699,
+      "loss": 0.2664,
       "step": 1383
     },
     {
       "epoch": 25.16,
       "learning_rate": 0.000412,
-      "loss": 0.3219,
+      "loss": 0.3288,
       "step": 1384
     },
     {
       "epoch": 25.18,
       "learning_rate": 0.0004119,
-      "loss": 0.305,
+      "loss": 0.2796,
       "step": 1385
     },
     {
       "epoch": 25.2,
       "learning_rate": 0.00041180000000000003,
-      "loss": 0.3098,
+      "loss": 0.3012,
       "step": 1386
     },
     {
       "epoch": 25.22,
       "learning_rate": 0.00041170000000000003,
-      "loss": 0.2635,
+      "loss": 0.2878,
       "step": 1387
     },
     {
       "epoch": 25.24,
       "learning_rate": 0.00041160000000000003,
-      "loss": 0.2823,
+      "loss": 0.2947,
       "step": 1388
     },
     {
       "epoch": 25.25,
       "learning_rate": 0.0004115,
-      "loss": 0.3343,
+      "loss": 0.3202,
       "step": 1389
     },
     {
       "epoch": 25.27,
       "learning_rate": 0.00041139999999999997,
-      "loss": 0.287,
+      "loss": 0.2611,
       "step": 1390
     },
     {
       "epoch": 25.29,
       "learning_rate": 0.0004113,
-      "loss": 0.3036,
+      "loss": 0.265,
       "step": 1391
     },
     {
       "epoch": 25.31,
       "learning_rate": 0.0004112,
-      "loss": 0.3255,
+      "loss": 0.3155,
       "step": 1392
     },
     {
       "epoch": 25.33,
       "learning_rate": 0.0004111,
-      "loss": 0.3636,
+      "loss": 0.3154,
       "step": 1393
     },
     {
       "epoch": 25.35,
       "learning_rate": 0.00041099999999999996,
-      "loss": 0.3305,
+      "loss": 0.3237,
       "step": 1394
     },
     {
       "epoch": 25.36,
       "learning_rate": 0.0004109,
-      "loss": 0.253,
+      "loss": 0.2553,
       "step": 1395
     },
     {
       "epoch": 25.38,
       "learning_rate": 0.0004108,
-      "loss": 0.2763,
+      "loss": 0.2597,
       "step": 1396
     },
     {
       "epoch": 25.4,
       "learning_rate": 0.0004107,
-      "loss": 0.3164,
+      "loss": 0.2861,
       "step": 1397
     },
     {
       "epoch": 25.42,
       "learning_rate": 0.0004106,
-      "loss": 0.2734,
+      "loss": 0.263,
       "step": 1398
     },
     {
       "epoch": 25.44,
       "learning_rate": 0.0004105,
-      "loss": 0.259,
+      "loss": 0.2828,
       "step": 1399
     },
     {
       "epoch": 25.45,
       "learning_rate": 0.0004104,
-      "loss": 0.2499,
+      "loss": 0.214,
       "step": 1400
     },
     {
       "epoch": 25.47,
       "learning_rate": 0.0004103,
-      "loss": 0.3023,
+      "loss": 0.2807,
       "step": 1401
     },
     {
       "epoch": 25.49,
       "learning_rate": 0.0004102,
-      "loss": 0.2904,
+      "loss": 0.2625,
       "step": 1402
     },
     {
       "epoch": 25.51,
       "learning_rate": 0.00041010000000000005,
-      "loss": 0.3036,
+      "loss": 0.2893,
       "step": 1403
     },
     {
       "epoch": 25.53,
       "learning_rate": 0.00041,
-      "loss": 0.3014,
+      "loss": 0.2655,
       "step": 1404
     },
     {
       "epoch": 25.55,
       "learning_rate": 0.0004099,
-      "loss": 0.288,
+      "loss": 0.2613,
       "step": 1405
     },
     {
       "epoch": 25.56,
       "learning_rate": 0.0004098,
-      "loss": 0.337,
+      "loss": 0.3026,
       "step": 1406
     },
     {
       "epoch": 25.58,
       "learning_rate": 0.00040970000000000004,
-      "loss": 0.2934,
+      "loss": 0.3094,
       "step": 1407
     },
     {
       "epoch": 25.6,
       "learning_rate": 0.00040960000000000004,
-      "loss": 0.31,
+      "loss": 0.3023,
       "step": 1408
     },
     {
       "epoch": 25.62,
       "learning_rate": 0.0004095,
-      "loss": 0.2578,
+      "loss": 0.252,
       "step": 1409
     },
     {
       "epoch": 25.64,
       "learning_rate": 0.0004094,
-      "loss": 0.2393,
+      "loss": 0.2719,
       "step": 1410
     },
     {
       "epoch": 25.65,
       "learning_rate": 0.00040930000000000003,
-      "loss": 0.3252,
+      "loss": 0.3064,
       "step": 1411
     },
     {
       "epoch": 25.67,
       "learning_rate": 0.0004092,
-      "loss": 0.3377,
+      "loss": 0.294,
       "step": 1412
     },
     {
       "epoch": 25.69,
       "learning_rate": 0.0004091,
-      "loss": 0.2966,
+      "loss": 0.2923,
       "step": 1413
     },
     {
       "epoch": 25.71,
       "learning_rate": 0.00040899999999999997,
-      "loss": 0.2989,
+      "loss": 0.2983,
       "step": 1414
     },
     {
       "epoch": 25.73,
       "learning_rate": 0.0004089,
-      "loss": 0.2612,
+      "loss": 0.2983,
       "step": 1415
     },
     {
       "epoch": 25.75,
       "learning_rate": 0.0004088,
-      "loss": 0.269,
+      "loss": 0.3085,
       "step": 1416
     },
     {
       "epoch": 25.76,
       "learning_rate": 0.0004087,
-      "loss": 0.3236,
+      "loss": 0.2998,
       "step": 1417
     },
     {
       "epoch": 25.78,
       "learning_rate": 0.0004086,
-      "loss": 0.2761,
+      "loss": 0.2936,
       "step": 1418
     },
     {
       "epoch": 25.8,
       "learning_rate": 0.0004085,
-      "loss": 0.2894,
+      "loss": 0.3297,
       "step": 1419
     },
     {
       "epoch": 25.82,
       "learning_rate": 0.0004084,
-      "loss": 0.3416,
+      "loss": 0.3302,
       "step": 1420
     },
     {
       "epoch": 25.84,
       "learning_rate": 0.0004083,
-      "loss": 0.2724,
+      "loss": 0.2734,
       "step": 1421
     },
     {
       "epoch": 25.85,
       "learning_rate": 0.0004082,
-      "loss": 0.3074,
+      "loss": 0.3201,
       "step": 1422
     },
     {
       "epoch": 25.87,
       "learning_rate": 0.00040810000000000005,
-      "loss": 0.2969,
+      "loss": 0.3077,
       "step": 1423
     },
     {
       "epoch": 25.89,
       "learning_rate": 0.000408,
-      "loss": 0.3357,
+      "loss": 0.309,
       "step": 1424
     },
     {
       "epoch": 25.91,
       "learning_rate": 0.0004079,
-      "loss": 0.3439,
+      "loss": 0.3562,
       "step": 1425
     },
     {
       "epoch": 25.93,
       "learning_rate": 0.0004078,
-      "loss": 0.3047,
+      "loss": 0.2958,
       "step": 1426
     },
     {
       "epoch": 25.95,
       "learning_rate": 0.0004077,
-      "loss": 0.3043,
+      "loss": 0.2676,
       "step": 1427
     },
     {
       "epoch": 25.96,
       "learning_rate": 0.00040760000000000004,
-      "loss": 0.2791,
+      "loss": 0.2907,
       "step": 1428
     },
     {
       "epoch": 25.98,
       "learning_rate": 0.0004075,
-      "loss": 0.2949,
+      "loss": 0.2725,
       "step": 1429
     },
     {
       "epoch": 26.0,
       "learning_rate": 0.0004074,
-      "loss": 0.2335,
+      "loss": 0.2759,
       "step": 1430
     },
     {
       "epoch": 26.02,
       "learning_rate": 0.0004073,
-      "loss": 0.3207,
+      "loss": 0.3353,
       "step": 1431
     },
     {
       "epoch": 26.04,
       "learning_rate": 0.00040720000000000003,
-      "loss": 0.2842,
+      "loss": 0.3089,
       "step": 1432
     },
     {
       "epoch": 26.05,
       "learning_rate": 0.00040710000000000003,
-      "loss": 0.2422,
+      "loss": 0.2733,
       "step": 1433
     },
     {
       "epoch": 26.07,
       "learning_rate": 0.00040699999999999997,
-      "loss": 0.2978,
+      "loss": 0.3349,
       "step": 1434
     },
     {
       "epoch": 26.09,
       "learning_rate": 0.00040689999999999997,
-      "loss": 0.2849,
+      "loss": 0.3104,
       "step": 1435
     },
     {
       "epoch": 26.11,
       "learning_rate": 0.0004068,
-      "loss": 0.2515,
+      "loss": 0.2734,
       "step": 1436
     },
     {
@@ -8652,3705 +8652,3705 @@
     {
       "epoch": 26.15,
       "learning_rate": 0.0004066,
-      "loss": 0.2261,
+      "loss": 0.2614,
       "step": 1438
     },
     {
       "epoch": 26.16,
       "learning_rate": 0.00040649999999999996,
-      "loss": 0.1991,
+      "loss": 0.1849,
       "step": 1439
     },
     {
       "epoch": 26.18,
       "learning_rate": 0.0004064,
-      "loss": 0.3097,
+      "loss": 0.3178,
       "step": 1440
     },
     {
       "epoch": 26.2,
       "learning_rate": 0.0004063,
-      "loss": 0.2641,
+      "loss": 0.2699,
       "step": 1441
     },
     {
       "epoch": 26.22,
       "learning_rate": 0.0004062,
-      "loss": 0.304,
+      "loss": 0.3376,
       "step": 1442
     },
     {
       "epoch": 26.24,
       "learning_rate": 0.0004061,
-      "loss": 0.2186,
+      "loss": 0.2325,
       "step": 1443
     },
     {
       "epoch": 26.25,
       "learning_rate": 0.00040600000000000006,
-      "loss": 0.3161,
+      "loss": 0.2787,
       "step": 1444
     },
     {
       "epoch": 26.27,
       "learning_rate": 0.0004059,
-      "loss": 0.2479,
+      "loss": 0.2445,
       "step": 1445
     },
     {
       "epoch": 26.29,
       "learning_rate": 0.0004058,
-      "loss": 0.2622,
+      "loss": 0.2708,
       "step": 1446
     },
     {
       "epoch": 26.31,
       "learning_rate": 0.0004057,
-      "loss": 0.291,
+      "loss": 0.2739,
       "step": 1447
     },
     {
       "epoch": 26.33,
       "learning_rate": 0.00040560000000000005,
-      "loss": 0.2815,
+      "loss": 0.3308,
       "step": 1448
     },
     {
       "epoch": 26.35,
       "learning_rate": 0.00040550000000000004,
-      "loss": 0.2528,
+      "loss": 0.2392,
       "step": 1449
     },
     {
       "epoch": 26.36,
       "learning_rate": 0.0004054,
-      "loss": 0.3211,
+      "loss": 0.3315,
       "step": 1450
     },
     {
       "epoch": 26.38,
       "learning_rate": 0.0004053,
-      "loss": 0.3311,
+      "loss": 0.2937,
       "step": 1451
     },
     {
       "epoch": 26.4,
       "learning_rate": 0.00040520000000000004,
-      "loss": 0.2863,
+      "loss": 0.2268,
       "step": 1452
     },
     {
       "epoch": 26.42,
       "learning_rate": 0.00040510000000000003,
-      "loss": 0.3292,
+      "loss": 0.3229,
       "step": 1453
     },
     {
       "epoch": 26.44,
       "learning_rate": 0.00040500000000000003,
-      "loss": 0.2925,
+      "loss": 0.2752,
       "step": 1454
     },
     {
       "epoch": 26.45,
       "learning_rate": 0.0004049,
-      "loss": 0.2745,
+      "loss": 0.2464,
       "step": 1455
     },
     {
       "epoch": 26.47,
       "learning_rate": 0.00040480000000000003,
-      "loss": 0.2357,
+      "loss": 0.2273,
       "step": 1456
     },
     {
       "epoch": 26.49,
       "learning_rate": 0.0004047,
-      "loss": 0.2887,
+      "loss": 0.2701,
       "step": 1457
     },
     {
       "epoch": 26.51,
       "learning_rate": 0.0004046,
-      "loss": 0.265,
+      "loss": 0.2573,
       "step": 1458
     },
     {
       "epoch": 26.53,
       "learning_rate": 0.0004045,
-      "loss": 0.2889,
+      "loss": 0.2986,
       "step": 1459
     },
     {
       "epoch": 26.55,
       "learning_rate": 0.00040439999999999996,
-      "loss": 0.3505,
+      "loss": 0.2917,
       "step": 1460
     },
     {
       "epoch": 26.56,
       "learning_rate": 0.0004043,
-      "loss": 0.3564,
+      "loss": 0.2606,
       "step": 1461
     },
     {
       "epoch": 26.58,
       "learning_rate": 0.0004042,
-      "loss": 0.3198,
+      "loss": 0.3466,
       "step": 1462
     },
     {
       "epoch": 26.6,
       "learning_rate": 0.0004041,
-      "loss": 0.2668,
+      "loss": 0.2833,
       "step": 1463
     },
     {
       "epoch": 26.62,
       "learning_rate": 0.000404,
-      "loss": 0.309,
+      "loss": 0.3247,
       "step": 1464
     },
     {
       "epoch": 26.64,
       "learning_rate": 0.0004039,
-      "loss": 0.3068,
+      "loss": 0.2993,
       "step": 1465
     },
     {
       "epoch": 26.65,
       "learning_rate": 0.0004038,
-      "loss": 0.315,
+      "loss": 0.3072,
       "step": 1466
     },
     {
       "epoch": 26.67,
       "learning_rate": 0.0004037,
-      "loss": 0.2783,
+      "loss": 0.3096,
       "step": 1467
     },
     {
       "epoch": 26.69,
       "learning_rate": 0.0004036,
-      "loss": 0.3204,
+      "loss": 0.3283,
       "step": 1468
     },
     {
       "epoch": 26.71,
       "learning_rate": 0.00040350000000000005,
-      "loss": 0.2656,
+      "loss": 0.2502,
       "step": 1469
     },
     {
       "epoch": 26.73,
       "learning_rate": 0.0004034,
-      "loss": 0.2994,
+      "loss": 0.2853,
       "step": 1470
     },
     {
       "epoch": 26.75,
       "learning_rate": 0.0004033,
-      "loss": 0.2926,
+      "loss": 0.2814,
       "step": 1471
     },
     {
       "epoch": 26.76,
       "learning_rate": 0.0004032,
-      "loss": 0.2591,
+      "loss": 0.2328,
       "step": 1472
     },
     {
       "epoch": 26.78,
       "learning_rate": 0.00040310000000000004,
-      "loss": 0.2778,
+      "loss": 0.2739,
       "step": 1473
     },
     {
       "epoch": 26.8,
       "learning_rate": 0.00040300000000000004,
-      "loss": 0.2574,
+      "loss": 0.2307,
       "step": 1474
     },
     {
       "epoch": 26.82,
       "learning_rate": 0.0004029,
-      "loss": 0.3183,
+      "loss": 0.2823,
       "step": 1475
     },
     {
       "epoch": 26.84,
       "learning_rate": 0.0004028,
-      "loss": 0.3125,
+      "loss": 0.2597,
       "step": 1476
     },
     {
       "epoch": 26.85,
       "learning_rate": 0.00040270000000000003,
-      "loss": 0.2382,
+      "loss": 0.1849,
       "step": 1477
     },
     {
       "epoch": 26.87,
       "learning_rate": 0.00040260000000000003,
-      "loss": 0.24,
+      "loss": 0.2098,
       "step": 1478
     },
     {
       "epoch": 26.89,
       "learning_rate": 0.0004025,
-      "loss": 0.28,
+      "loss": 0.2329,
       "step": 1479
     },
     {
       "epoch": 26.91,
       "learning_rate": 0.00040239999999999997,
-      "loss": 0.269,
+      "loss": 0.2857,
       "step": 1480
     },
     {
       "epoch": 26.93,
       "learning_rate": 0.0004023,
-      "loss": 0.3353,
+      "loss": 0.3393,
       "step": 1481
     },
     {
       "epoch": 26.95,
       "learning_rate": 0.0004022,
-      "loss": 0.2545,
+      "loss": 0.254,
       "step": 1482
     },
     {
       "epoch": 26.96,
       "learning_rate": 0.0004021,
-      "loss": 0.2846,
+      "loss": 0.2834,
       "step": 1483
     },
     {
       "epoch": 26.98,
       "learning_rate": 0.000402,
-      "loss": 0.3488,
+      "loss": 0.3296,
       "step": 1484
     },
     {
       "epoch": 27.0,
       "learning_rate": 0.0004019,
-      "loss": 0.3,
+      "loss": 0.2071,
       "step": 1485
     },
     {
       "epoch": 27.02,
       "learning_rate": 0.0004018,
-      "loss": 0.2981,
+      "loss": 0.3035,
       "step": 1486
     },
     {
       "epoch": 27.04,
       "learning_rate": 0.0004017,
-      "loss": 0.2916,
+      "loss": 0.2477,
       "step": 1487
     },
     {
       "epoch": 27.05,
       "learning_rate": 0.0004016,
-      "loss": 0.2856,
+      "loss": 0.2243,
       "step": 1488
     },
     {
       "epoch": 27.07,
       "learning_rate": 0.00040150000000000006,
-      "loss": 0.2966,
+      "loss": 0.2545,
       "step": 1489
     },
     {
       "epoch": 27.09,
       "learning_rate": 0.0004014,
-      "loss": 0.2603,
+      "loss": 0.26,
       "step": 1490
     },
     {
       "epoch": 27.11,
       "learning_rate": 0.0004013,
-      "loss": 0.3073,
+      "loss": 0.2977,
       "step": 1491
     },
     {
       "epoch": 27.13,
       "learning_rate": 0.0004012,
-      "loss": 0.2512,
+      "loss": 0.258,
       "step": 1492
     },
     {
       "epoch": 27.15,
       "learning_rate": 0.00040110000000000005,
-      "loss": 0.307,
+      "loss": 0.2534,
       "step": 1493
     },
     {
       "epoch": 27.16,
       "learning_rate": 0.00040100000000000004,
-      "loss": 0.2331,
+      "loss": 0.2276,
       "step": 1494
     },
     {
       "epoch": 27.18,
       "learning_rate": 0.0004009,
-      "loss": 0.3299,
+      "loss": 0.3333,
       "step": 1495
     },
     {
       "epoch": 27.2,
       "learning_rate": 0.0004008,
-      "loss": 0.2699,
+      "loss": 0.2379,
       "step": 1496
     },
     {
       "epoch": 27.22,
       "learning_rate": 0.0004007,
-      "loss": 0.29,
+      "loss": 0.2537,
       "step": 1497
     },
     {
       "epoch": 27.24,
       "learning_rate": 0.00040060000000000003,
-      "loss": 0.2309,
+      "loss": 0.2293,
       "step": 1498
     },
     {
       "epoch": 27.25,
       "learning_rate": 0.00040050000000000003,
-      "loss": 0.3034,
+      "loss": 0.2877,
       "step": 1499
     },
     {
       "epoch": 27.27,
       "learning_rate": 0.0004004,
-      "loss": 0.344,
+      "loss": 0.2733,
       "step": 1500
     },
     {
       "epoch": 27.27,
-      "eval_cer": 0.0961783534985532,
-      "eval_loss": 0.40665939450263977,
-      "eval_runtime": 13.9324,
-      "eval_samples_per_second": 118.214,
-      "eval_steps_per_second": 1.866,
-      "eval_wer": 0.3850474925952405,
+      "eval_cer": 0.09649320079761316,
+      "eval_loss": 0.3965006172657013,
+      "eval_runtime": 14.1738,
+      "eval_samples_per_second": 116.201,
+      "eval_steps_per_second": 1.834,
+      "eval_wer": 0.38709018486365027,
       "step": 1500
     },
     {
       "epoch": 27.29,
       "learning_rate": 0.00040029999999999997,
-      "loss": 0.2813,
+      "loss": 0.2852,
       "step": 1501
     },
     {
       "epoch": 27.31,
       "learning_rate": 0.0004002,
-      "loss": 0.2969,
+      "loss": 0.2717,
       "step": 1502
     },
     {
       "epoch": 27.33,
       "learning_rate": 0.0004001,
-      "loss": 0.3117,
+      "loss": 0.2458,
       "step": 1503
     },
     {
       "epoch": 27.35,
       "learning_rate": 0.0004,
-      "loss": 0.3001,
+      "loss": 0.2713,
       "step": 1504
     },
     {
       "epoch": 27.36,
       "learning_rate": 0.00039989999999999996,
-      "loss": 0.2882,
+      "loss": 0.2829,
       "step": 1505
     },
     {
       "epoch": 27.38,
       "learning_rate": 0.0003998,
-      "loss": 0.2922,
+      "loss": 0.2794,
       "step": 1506
     },
     {
       "epoch": 27.4,
       "learning_rate": 0.0003997,
-      "loss": 0.3295,
+      "loss": 0.2893,
       "step": 1507
     },
     {
       "epoch": 27.42,
       "learning_rate": 0.0003996,
-      "loss": 0.3616,
+      "loss": 0.3011,
       "step": 1508
     },
     {
       "epoch": 27.44,
       "learning_rate": 0.0003995,
-      "loss": 0.3471,
+      "loss": 0.305,
       "step": 1509
     },
     {
       "epoch": 27.45,
       "learning_rate": 0.0003994,
-      "loss": 0.2818,
+      "loss": 0.2741,
       "step": 1510
     },
     {
       "epoch": 27.47,
       "learning_rate": 0.0003993,
-      "loss": 0.329,
+      "loss": 0.2971,
       "step": 1511
     },
     {
       "epoch": 27.49,
       "learning_rate": 0.0003992,
-      "loss": 0.3054,
+      "loss": 0.2908,
       "step": 1512
     },
     {
       "epoch": 27.51,
       "learning_rate": 0.0003991,
-      "loss": 0.2634,
+      "loss": 0.2502,
       "step": 1513
     },
     {
       "epoch": 27.53,
       "learning_rate": 0.00039900000000000005,
-      "loss": 0.3351,
+      "loss": 0.3076,
       "step": 1514
     },
     {
       "epoch": 27.55,
       "learning_rate": 0.0003989,
-      "loss": 0.3091,
+      "loss": 0.2935,
       "step": 1515
     },
     {
       "epoch": 27.56,
       "learning_rate": 0.0003988,
-      "loss": 0.2574,
+      "loss": 0.2575,
       "step": 1516
     },
     {
       "epoch": 27.58,
       "learning_rate": 0.0003987,
-      "loss": 0.3276,
+      "loss": 0.2955,
       "step": 1517
     },
     {
       "epoch": 27.6,
       "learning_rate": 0.00039860000000000004,
-      "loss": 0.2573,
+      "loss": 0.2293,
       "step": 1518
     },
     {
       "epoch": 27.62,
       "learning_rate": 0.00039850000000000004,
-      "loss": 0.2628,
+      "loss": 0.2802,
       "step": 1519
     },
     {
       "epoch": 27.64,
       "learning_rate": 0.0003984,
-      "loss": 0.312,
+      "loss": 0.3052,
       "step": 1520
     },
     {
       "epoch": 27.65,
       "learning_rate": 0.0003983,
-      "loss": 0.2444,
+      "loss": 0.2292,
       "step": 1521
     },
     {
       "epoch": 27.67,
       "learning_rate": 0.00039820000000000003,
-      "loss": 0.3148,
+      "loss": 0.2781,
       "step": 1522
     },
     {
       "epoch": 27.69,
       "learning_rate": 0.00039810000000000003,
-      "loss": 0.3083,
+      "loss": 0.2587,
       "step": 1523
     },
     {
       "epoch": 27.71,
       "learning_rate": 0.000398,
-      "loss": 0.2119,
+      "loss": 0.2268,
       "step": 1524
     },
     {
       "epoch": 27.73,
       "learning_rate": 0.00039789999999999997,
-      "loss": 0.2699,
+      "loss": 0.2606,
       "step": 1525
     },
     {
       "epoch": 27.75,
       "learning_rate": 0.0003978,
-      "loss": 0.2917,
+      "loss": 0.3081,
       "step": 1526
     },
     {
       "epoch": 27.76,
       "learning_rate": 0.0003977,
-      "loss": 0.2884,
+      "loss": 0.2618,
       "step": 1527
     },
     {
       "epoch": 27.78,
       "learning_rate": 0.0003976,
-      "loss": 0.2498,
+      "loss": 0.2419,
       "step": 1528
     },
     {
       "epoch": 27.8,
       "learning_rate": 0.0003975,
-      "loss": 0.3233,
+      "loss": 0.3141,
       "step": 1529
     },
     {
       "epoch": 27.82,
       "learning_rate": 0.0003974,
-      "loss": 0.2945,
+      "loss": 0.2377,
       "step": 1530
     },
     {
       "epoch": 27.84,
       "learning_rate": 0.0003973,
-      "loss": 0.2665,
+      "loss": 0.2492,
       "step": 1531
     },
     {
       "epoch": 27.85,
       "learning_rate": 0.0003972,
-      "loss": 0.3948,
+      "loss": 0.3541,
       "step": 1532
     },
     {
       "epoch": 27.87,
       "learning_rate": 0.0003971,
-      "loss": 0.2768,
+      "loss": 0.2726,
       "step": 1533
     },
     {
       "epoch": 27.89,
       "learning_rate": 0.00039700000000000005,
-      "loss": 0.3281,
+      "loss": 0.3302,
       "step": 1534
     },
     {
       "epoch": 27.91,
       "learning_rate": 0.0003969,
-      "loss": 0.3024,
+      "loss": 0.3251,
       "step": 1535
     },
     {
       "epoch": 27.93,
       "learning_rate": 0.0003968,
-      "loss": 0.3337,
+      "loss": 0.3257,
       "step": 1536
     },
     {
       "epoch": 27.95,
       "learning_rate": 0.0003967,
-      "loss": 0.2816,
+      "loss": 0.2646,
       "step": 1537
     },
     {
       "epoch": 27.96,
       "learning_rate": 0.0003966,
-      "loss": 0.2918,
+      "loss": 0.3045,
       "step": 1538
     },
     {
       "epoch": 27.98,
       "learning_rate": 0.00039650000000000004,
-      "loss": 0.2782,
+      "loss": 0.264,
       "step": 1539
     },
     {
       "epoch": 28.0,
       "learning_rate": 0.0003964,
-      "loss": 0.2722,
+      "loss": 0.2373,
       "step": 1540
     },
     {
       "epoch": 28.02,
       "learning_rate": 0.0003963,
-      "loss": 0.2832,
+      "loss": 0.2766,
       "step": 1541
     },
     {
       "epoch": 28.04,
       "learning_rate": 0.0003962,
-      "loss": 0.3126,
+      "loss": 0.2988,
       "step": 1542
     },
     {
       "epoch": 28.05,
       "learning_rate": 0.00039610000000000003,
-      "loss": 0.3507,
+      "loss": 0.307,
       "step": 1543
     },
     {
       "epoch": 28.07,
       "learning_rate": 0.00039600000000000003,
-      "loss": 0.2867,
+      "loss": 0.2753,
       "step": 1544
     },
     {
       "epoch": 28.09,
       "learning_rate": 0.0003959,
-      "loss": 0.2416,
+      "loss": 0.2612,
       "step": 1545
     },
     {
       "epoch": 28.11,
       "learning_rate": 0.00039579999999999997,
-      "loss": 0.2236,
+      "loss": 0.2334,
       "step": 1546
     },
     {
       "epoch": 28.13,
       "learning_rate": 0.0003957,
-      "loss": 0.2828,
+      "loss": 0.3025,
       "step": 1547
     },
     {
       "epoch": 28.15,
       "learning_rate": 0.0003956,
-      "loss": 0.2883,
+      "loss": 0.3335,
       "step": 1548
     },
     {
       "epoch": 28.16,
       "learning_rate": 0.0003955,
-      "loss": 0.318,
+      "loss": 0.3152,
       "step": 1549
     },
     {
       "epoch": 28.18,
       "learning_rate": 0.00039539999999999996,
-      "loss": 0.3166,
+      "loss": 0.3069,
       "step": 1550
     },
     {
       "epoch": 28.2,
       "learning_rate": 0.0003953,
-      "loss": 0.2457,
+      "loss": 0.2318,
       "step": 1551
     },
     {
       "epoch": 28.22,
       "learning_rate": 0.0003952,
-      "loss": 0.2697,
+      "loss": 0.2547,
       "step": 1552
     },
     {
       "epoch": 28.24,
       "learning_rate": 0.0003951,
-      "loss": 0.313,
+      "loss": 0.2608,
       "step": 1553
     },
     {
       "epoch": 28.25,
       "learning_rate": 0.000395,
-      "loss": 0.3132,
+      "loss": 0.3017,
       "step": 1554
     },
     {
       "epoch": 28.27,
       "learning_rate": 0.0003949,
-      "loss": 0.2762,
+      "loss": 0.2831,
       "step": 1555
     },
     {
       "epoch": 28.29,
       "learning_rate": 0.0003948,
-      "loss": 0.2712,
+      "loss": 0.2349,
       "step": 1556
     },
     {
       "epoch": 28.31,
       "learning_rate": 0.0003947,
-      "loss": 0.354,
+      "loss": 0.331,
       "step": 1557
     },
     {
       "epoch": 28.33,
       "learning_rate": 0.0003946,
-      "loss": 0.3128,
+      "loss": 0.2818,
       "step": 1558
     },
     {
       "epoch": 28.35,
       "learning_rate": 0.00039450000000000005,
-      "loss": 0.2924,
+      "loss": 0.2739,
       "step": 1559
     },
     {
       "epoch": 28.36,
       "learning_rate": 0.0003944,
-      "loss": 0.3157,
+      "loss": 0.3178,
       "step": 1560
     },
     {
       "epoch": 28.38,
       "learning_rate": 0.0003943,
-      "loss": 0.2702,
+      "loss": 0.2561,
       "step": 1561
     },
     {
       "epoch": 28.4,
       "learning_rate": 0.0003942,
-      "loss": 0.277,
+      "loss": 0.2929,
       "step": 1562
     },
     {
       "epoch": 28.42,
       "learning_rate": 0.00039410000000000004,
-      "loss": 0.2841,
+      "loss": 0.2421,
       "step": 1563
     },
     {
       "epoch": 28.44,
       "learning_rate": 0.00039400000000000004,
-      "loss": 0.3362,
+      "loss": 0.3062,
       "step": 1564
     },
     {
       "epoch": 28.45,
       "learning_rate": 0.0003939,
-      "loss": 0.2811,
+      "loss": 0.2543,
       "step": 1565
     },
     {
       "epoch": 28.47,
       "learning_rate": 0.0003938,
-      "loss": 0.2801,
+      "loss": 0.2814,
       "step": 1566
     },
     {
       "epoch": 28.49,
       "learning_rate": 0.00039370000000000003,
-      "loss": 0.2646,
+      "loss": 0.2807,
       "step": 1567
     },
     {
       "epoch": 28.51,
       "learning_rate": 0.0003936,
-      "loss": 0.3121,
+      "loss": 0.294,
       "step": 1568
     },
     {
       "epoch": 28.53,
       "learning_rate": 0.0003935,
-      "loss": 0.283,
+      "loss": 0.2639,
       "step": 1569
     },
     {
       "epoch": 28.55,
       "learning_rate": 0.0003934,
-      "loss": 0.2243,
+      "loss": 0.2366,
       "step": 1570
     },
     {
       "epoch": 28.56,
       "learning_rate": 0.00039329999999999996,
-      "loss": 0.2805,
+      "loss": 0.2677,
       "step": 1571
     },
     {
       "epoch": 28.58,
       "learning_rate": 0.0003932,
-      "loss": 0.2876,
+      "loss": 0.2626,
       "step": 1572
     },
     {
       "epoch": 28.6,
       "learning_rate": 0.0003931,
-      "loss": 0.3026,
+      "loss": 0.2917,
       "step": 1573
     },
     {
       "epoch": 28.62,
       "learning_rate": 0.000393,
-      "loss": 0.3215,
+      "loss": 0.29,
       "step": 1574
     },
     {
       "epoch": 28.64,
       "learning_rate": 0.0003929,
-      "loss": 0.3208,
+      "loss": 0.3074,
       "step": 1575
     },
     {
       "epoch": 28.65,
       "learning_rate": 0.0003928,
-      "loss": 0.3301,
+      "loss": 0.332,
       "step": 1576
     },
     {
       "epoch": 28.67,
       "learning_rate": 0.0003927,
-      "loss": 0.2689,
+      "loss": 0.2702,
       "step": 1577
     },
     {
       "epoch": 28.69,
       "learning_rate": 0.0003926,
-      "loss": 0.2851,
+      "loss": 0.2511,
       "step": 1578
     },
     {
       "epoch": 28.71,
       "learning_rate": 0.0003925,
-      "loss": 0.2814,
+      "loss": 0.2427,
       "step": 1579
     },
     {
       "epoch": 28.73,
       "learning_rate": 0.00039240000000000005,
-      "loss": 0.2894,
+      "loss": 0.2408,
       "step": 1580
     },
     {
       "epoch": 28.75,
       "learning_rate": 0.0003923,
-      "loss": 0.2913,
+      "loss": 0.2403,
       "step": 1581
     },
     {
       "epoch": 28.76,
       "learning_rate": 0.0003922,
-      "loss": 0.2763,
+      "loss": 0.2448,
       "step": 1582
     },
     {
       "epoch": 28.78,
       "learning_rate": 0.0003921,
-      "loss": 0.2672,
+      "loss": 0.2559,
       "step": 1583
     },
     {
       "epoch": 28.8,
       "learning_rate": 0.00039200000000000004,
-      "loss": 0.3255,
+      "loss": 0.2861,
       "step": 1584
     },
     {
       "epoch": 28.82,
       "learning_rate": 0.00039190000000000004,
-      "loss": 0.3611,
+      "loss": 0.2922,
       "step": 1585
     },
     {
       "epoch": 28.84,
       "learning_rate": 0.0003918,
-      "loss": 0.2507,
+      "loss": 0.2456,
       "step": 1586
     },
     {
       "epoch": 28.85,
       "learning_rate": 0.0003917,
-      "loss": 0.3005,
+      "loss": 0.2606,
       "step": 1587
     },
     {
       "epoch": 28.87,
       "learning_rate": 0.00039160000000000003,
-      "loss": 0.3,
+      "loss": 0.2493,
       "step": 1588
     },
     {
       "epoch": 28.89,
       "learning_rate": 0.00039150000000000003,
-      "loss": 0.2853,
+      "loss": 0.284,
       "step": 1589
     },
     {
       "epoch": 28.91,
       "learning_rate": 0.0003914,
-      "loss": 0.3148,
+      "loss": 0.2732,
       "step": 1590
     },
     {
       "epoch": 28.93,
       "learning_rate": 0.00039129999999999997,
-      "loss": 0.3599,
+      "loss": 0.3195,
       "step": 1591
     },
     {
       "epoch": 28.95,
       "learning_rate": 0.0003912,
-      "loss": 0.3209,
+      "loss": 0.2926,
       "step": 1592
     },
     {
       "epoch": 28.96,
       "learning_rate": 0.0003911,
-      "loss": 0.3587,
+      "loss": 0.2939,
       "step": 1593
     },
     {
       "epoch": 28.98,
       "learning_rate": 0.000391,
-      "loss": 0.3004,
+      "loss": 0.2431,
       "step": 1594
     },
     {
       "epoch": 29.0,
       "learning_rate": 0.0003909,
-      "loss": 0.254,
+      "loss": 0.2718,
       "step": 1595
     },
     {
       "epoch": 29.02,
       "learning_rate": 0.0003908,
-      "loss": 0.3033,
+      "loss": 0.2669,
       "step": 1596
     },
     {
       "epoch": 29.04,
       "learning_rate": 0.0003907,
-      "loss": 0.3306,
+      "loss": 0.2415,
       "step": 1597
     },
     {
       "epoch": 29.05,
       "learning_rate": 0.0003906,
-      "loss": 0.2489,
+      "loss": 0.2347,
       "step": 1598
     },
     {
       "epoch": 29.07,
       "learning_rate": 0.0003905,
-      "loss": 0.2802,
+      "loss": 0.2656,
       "step": 1599
     },
     {
       "epoch": 29.09,
       "learning_rate": 0.00039040000000000006,
-      "loss": 0.2795,
+      "loss": 0.2307,
       "step": 1600
     },
     {
       "epoch": 29.11,
       "learning_rate": 0.0003903,
-      "loss": 0.2726,
+      "loss": 0.2714,
       "step": 1601
     },
     {
       "epoch": 29.13,
       "learning_rate": 0.0003902,
-      "loss": 0.3016,
+      "loss": 0.2577,
       "step": 1602
     },
     {
       "epoch": 29.15,
       "learning_rate": 0.0003901,
-      "loss": 0.2935,
+      "loss": 0.2693,
       "step": 1603
     },
     {
       "epoch": 29.16,
       "learning_rate": 0.00039000000000000005,
-      "loss": 0.2617,
+      "loss": 0.2531,
       "step": 1604
     },
     {
       "epoch": 29.18,
       "learning_rate": 0.00038990000000000004,
-      "loss": 0.2782,
+      "loss": 0.2357,
       "step": 1605
     },
     {
       "epoch": 29.2,
       "learning_rate": 0.0003898,
-      "loss": 0.3108,
+      "loss": 0.2832,
       "step": 1606
     },
     {
       "epoch": 29.22,
       "learning_rate": 0.0003897,
-      "loss": 0.2988,
+      "loss": 0.2566,
       "step": 1607
     },
     {
       "epoch": 29.24,
       "learning_rate": 0.0003896,
-      "loss": 0.2965,
+      "loss": 0.2924,
       "step": 1608
     },
     {
       "epoch": 29.25,
       "learning_rate": 0.00038950000000000003,
-      "loss": 0.2901,
+      "loss": 0.2669,
       "step": 1609
     },
     {
       "epoch": 29.27,
       "learning_rate": 0.00038940000000000003,
-      "loss": 0.2717,
+      "loss": 0.2705,
       "step": 1610
     },
     {
       "epoch": 29.29,
       "learning_rate": 0.0003893,
-      "loss": 0.2185,
+      "loss": 0.2009,
       "step": 1611
     },
     {
       "epoch": 29.31,
       "learning_rate": 0.0003892,
-      "loss": 0.2834,
+      "loss": 0.2432,
       "step": 1612
     },
     {
       "epoch": 29.33,
       "learning_rate": 0.0003891,
-      "loss": 0.3045,
+      "loss": 0.2656,
       "step": 1613
     },
     {
       "epoch": 29.35,
       "learning_rate": 0.000389,
-      "loss": 0.3147,
+      "loss": 0.2851,
       "step": 1614
     },
     {
       "epoch": 29.36,
       "learning_rate": 0.0003889,
-      "loss": 0.3224,
+      "loss": 0.2948,
       "step": 1615
     },
     {
       "epoch": 29.38,
       "learning_rate": 0.00038879999999999996,
-      "loss": 0.3063,
+      "loss": 0.2895,
       "step": 1616
     },
     {
       "epoch": 29.4,
       "learning_rate": 0.0003887,
-      "loss": 0.2873,
+      "loss": 0.2647,
       "step": 1617
     },
     {
       "epoch": 29.42,
       "learning_rate": 0.0003886,
-      "loss": 0.2492,
+      "loss": 0.2159,
       "step": 1618
     },
     {
       "epoch": 29.44,
       "learning_rate": 0.0003885,
-      "loss": 0.254,
+      "loss": 0.2366,
       "step": 1619
     },
     {
       "epoch": 29.45,
       "learning_rate": 0.0003884,
-      "loss": 0.2136,
+      "loss": 0.1774,
       "step": 1620
     },
     {
       "epoch": 29.47,
       "learning_rate": 0.0003883,
-      "loss": 0.3145,
+      "loss": 0.2677,
       "step": 1621
     },
     {
       "epoch": 29.49,
       "learning_rate": 0.0003882,
-      "loss": 0.2378,
+      "loss": 0.2309,
       "step": 1622
     },
     {
       "epoch": 29.51,
       "learning_rate": 0.0003881,
-      "loss": 0.2579,
+      "loss": 0.2366,
       "step": 1623
     },
     {
       "epoch": 29.53,
       "learning_rate": 0.000388,
-      "loss": 0.299,
+      "loss": 0.2574,
       "step": 1624
     },
     {
       "epoch": 29.55,
       "learning_rate": 0.00038790000000000005,
-      "loss": 0.3432,
+      "loss": 0.3053,
       "step": 1625
     },
     {
       "epoch": 29.56,
       "learning_rate": 0.0003878,
-      "loss": 0.291,
+      "loss": 0.2426,
       "step": 1626
     },
     {
       "epoch": 29.58,
       "learning_rate": 0.0003877,
-      "loss": 0.3157,
+      "loss": 0.265,
       "step": 1627
     },
     {
       "epoch": 29.6,
       "learning_rate": 0.0003876,
-      "loss": 0.2837,
+      "loss": 0.2703,
       "step": 1628
     },
     {
       "epoch": 29.62,
       "learning_rate": 0.00038750000000000004,
-      "loss": 0.2963,
+      "loss": 0.2432,
       "step": 1629
     },
     {
       "epoch": 29.64,
       "learning_rate": 0.00038740000000000004,
-      "loss": 0.2852,
+      "loss": 0.2727,
       "step": 1630
     },
     {
       "epoch": 29.65,
       "learning_rate": 0.0003873,
-      "loss": 0.2602,
+      "loss": 0.2285,
       "step": 1631
     },
     {
       "epoch": 29.67,
       "learning_rate": 0.0003872,
-      "loss": 0.2916,
+      "loss": 0.2755,
       "step": 1632
     },
     {
       "epoch": 29.69,
       "learning_rate": 0.00038710000000000003,
-      "loss": 0.2771,
+      "loss": 0.2074,
       "step": 1633
     },
     {
       "epoch": 29.71,
       "learning_rate": 0.00038700000000000003,
-      "loss": 0.2819,
+      "loss": 0.2498,
       "step": 1634
     },
     {
       "epoch": 29.73,
       "learning_rate": 0.0003869,
-      "loss": 0.2166,
+      "loss": 0.1812,
       "step": 1635
     },
     {
       "epoch": 29.75,
       "learning_rate": 0.00038679999999999997,
-      "loss": 0.2923,
+      "loss": 0.3075,
       "step": 1636
     },
     {
       "epoch": 29.76,
       "learning_rate": 0.0003867,
-      "loss": 0.2622,
+      "loss": 0.2492,
       "step": 1637
     },
     {
       "epoch": 29.78,
       "learning_rate": 0.0003866,
-      "loss": 0.2836,
+      "loss": 0.3038,
       "step": 1638
     },
     {
       "epoch": 29.8,
       "learning_rate": 0.0003865,
-      "loss": 0.2674,
+      "loss": 0.2611,
       "step": 1639
     },
     {
       "epoch": 29.82,
       "learning_rate": 0.0003864,
-      "loss": 0.307,
+      "loss": 0.274,
       "step": 1640
     },
     {
       "epoch": 29.84,
       "learning_rate": 0.0003863,
-      "loss": 0.3385,
+      "loss": 0.2814,
       "step": 1641
     },
     {
       "epoch": 29.85,
       "learning_rate": 0.0003862,
-      "loss": 0.2919,
+      "loss": 0.267,
       "step": 1642
     },
     {
       "epoch": 29.87,
       "learning_rate": 0.0003861,
-      "loss": 0.2965,
+      "loss": 0.2522,
       "step": 1643
     },
     {
       "epoch": 29.89,
       "learning_rate": 0.000386,
-      "loss": 0.2744,
+      "loss": 0.2417,
       "step": 1644
     },
     {
       "epoch": 29.91,
       "learning_rate": 0.00038590000000000006,
-      "loss": 0.3535,
+      "loss": 0.2594,
       "step": 1645
     },
     {
       "epoch": 29.93,
       "learning_rate": 0.0003858,
-      "loss": 0.2883,
+      "loss": 0.2219,
       "step": 1646
     },
     {
       "epoch": 29.95,
       "learning_rate": 0.0003857,
-      "loss": 0.3224,
+      "loss": 0.2956,
       "step": 1647
     },
     {
       "epoch": 29.96,
       "learning_rate": 0.0003856,
-      "loss": 0.3053,
+      "loss": 0.2737,
       "step": 1648
     },
     {
       "epoch": 29.98,
       "learning_rate": 0.0003855,
-      "loss": 0.2972,
+      "loss": 0.2563,
       "step": 1649
     },
     {
       "epoch": 30.0,
       "learning_rate": 0.00038540000000000004,
-      "loss": 0.2185,
+      "loss": 0.2051,
       "step": 1650
     },
     {
       "epoch": 30.02,
       "learning_rate": 0.0003853,
-      "loss": 0.2627,
+      "loss": 0.221,
       "step": 1651
     },
     {
       "epoch": 30.04,
       "learning_rate": 0.0003852,
-      "loss": 0.2564,
+      "loss": 0.2733,
       "step": 1652
     },
     {
       "epoch": 30.05,
       "learning_rate": 0.0003851,
-      "loss": 0.28,
+      "loss": 0.2795,
       "step": 1653
     },
     {
       "epoch": 30.07,
       "learning_rate": 0.00038500000000000003,
-      "loss": 0.3616,
+      "loss": 0.3263,
       "step": 1654
     },
     {
       "epoch": 30.09,
       "learning_rate": 0.00038490000000000003,
-      "loss": 0.3098,
+      "loss": 0.2953,
       "step": 1655
     },
     {
       "epoch": 30.11,
       "learning_rate": 0.0003848,
-      "loss": 0.2889,
+      "loss": 0.256,
       "step": 1656
     },
     {
       "epoch": 30.13,
       "learning_rate": 0.00038469999999999997,
-      "loss": 0.2819,
+      "loss": 0.2443,
       "step": 1657
     },
     {
       "epoch": 30.15,
       "learning_rate": 0.0003846,
-      "loss": 0.2395,
+      "loss": 0.2416,
       "step": 1658
     },
     {
       "epoch": 30.16,
       "learning_rate": 0.0003845,
-      "loss": 0.3207,
+      "loss": 0.2792,
       "step": 1659
     },
     {
       "epoch": 30.18,
       "learning_rate": 0.0003844,
-      "loss": 0.2996,
+      "loss": 0.326,
       "step": 1660
     },
     {
       "epoch": 30.2,
       "learning_rate": 0.00038429999999999996,
-      "loss": 0.2717,
+      "loss": 0.2302,
       "step": 1661
     },
     {
       "epoch": 30.22,
       "learning_rate": 0.0003842,
-      "loss": 0.3189,
+      "loss": 0.2868,
       "step": 1662
     },
     {
       "epoch": 30.24,
       "learning_rate": 0.0003841,
-      "loss": 0.3162,
+      "loss": 0.2698,
       "step": 1663
     },
     {
       "epoch": 30.25,
       "learning_rate": 0.000384,
-      "loss": 0.2886,
+      "loss": 0.3063,
       "step": 1664
     },
     {
       "epoch": 30.27,
       "learning_rate": 0.0003839,
-      "loss": 0.2573,
+      "loss": 0.28,
       "step": 1665
     },
     {
       "epoch": 30.29,
       "learning_rate": 0.0003838,
-      "loss": 0.2915,
+      "loss": 0.2453,
       "step": 1666
     },
     {
       "epoch": 30.31,
       "learning_rate": 0.0003837,
-      "loss": 0.2498,
+      "loss": 0.2252,
       "step": 1667
     },
     {
       "epoch": 30.33,
       "learning_rate": 0.0003836,
-      "loss": 0.2785,
+      "loss": 0.2506,
       "step": 1668
     },
     {
       "epoch": 30.35,
       "learning_rate": 0.0003835,
-      "loss": 0.3502,
+      "loss": 0.2788,
       "step": 1669
     },
     {
       "epoch": 30.36,
       "learning_rate": 0.00038340000000000005,
-      "loss": 0.2755,
+      "loss": 0.2504,
       "step": 1670
     },
     {
       "epoch": 30.38,
       "learning_rate": 0.0003833,
-      "loss": 0.2553,
+      "loss": 0.218,
       "step": 1671
     },
     {
       "epoch": 30.4,
       "learning_rate": 0.0003832,
-      "loss": 0.282,
+      "loss": 0.3024,
       "step": 1672
     },
     {
       "epoch": 30.42,
       "learning_rate": 0.0003831,
-      "loss": 0.2488,
+      "loss": 0.2424,
       "step": 1673
     },
     {
       "epoch": 30.44,
       "learning_rate": 0.00038300000000000004,
-      "loss": 0.2674,
+      "loss": 0.2367,
       "step": 1674
     },
     {
       "epoch": 30.45,
       "learning_rate": 0.00038290000000000004,
-      "loss": 0.2196,
+      "loss": 0.2011,
       "step": 1675
     },
     {
       "epoch": 30.47,
       "learning_rate": 0.0003828,
-      "loss": 0.2543,
+      "loss": 0.2326,
       "step": 1676
     },
     {
       "epoch": 30.49,
       "learning_rate": 0.0003827,
-      "loss": 0.2555,
+      "loss": 0.2048,
       "step": 1677
     },
     {
       "epoch": 30.51,
       "learning_rate": 0.00038260000000000003,
-      "loss": 0.2827,
+      "loss": 0.2453,
       "step": 1678
     },
     {
       "epoch": 30.53,
       "learning_rate": 0.00038250000000000003,
-      "loss": 0.252,
+      "loss": 0.2243,
       "step": 1679
     },
     {
       "epoch": 30.55,
       "learning_rate": 0.0003824,
-      "loss": 0.273,
+      "loss": 0.211,
       "step": 1680
     },
     {
       "epoch": 30.56,
       "learning_rate": 0.00038229999999999997,
-      "loss": 0.2785,
+      "loss": 0.2728,
       "step": 1681
     },
     {
       "epoch": 30.58,
       "learning_rate": 0.00038219999999999997,
-      "loss": 0.3233,
+      "loss": 0.2776,
       "step": 1682
     },
     {
       "epoch": 30.6,
       "learning_rate": 0.0003821,
-      "loss": 0.282,
+      "loss": 0.2422,
       "step": 1683
     },
     {
       "epoch": 30.62,
       "learning_rate": 0.000382,
-      "loss": 0.2821,
+      "loss": 0.2321,
       "step": 1684
     },
     {
       "epoch": 30.64,
       "learning_rate": 0.0003819,
-      "loss": 0.27,
+      "loss": 0.2508,
       "step": 1685
     },
     {
       "epoch": 30.65,
       "learning_rate": 0.00038179999999999996,
-      "loss": 0.2826,
+      "loss": 0.2274,
       "step": 1686
     },
     {
       "epoch": 30.67,
       "learning_rate": 0.0003817,
-      "loss": 0.2754,
+      "loss": 0.2538,
       "step": 1687
     },
     {
       "epoch": 30.69,
       "learning_rate": 0.0003816,
-      "loss": 0.3035,
+      "loss": 0.299,
       "step": 1688
     },
     {
       "epoch": 30.71,
       "learning_rate": 0.0003815,
-      "loss": 0.3013,
+      "loss": 0.2509,
       "step": 1689
     },
     {
       "epoch": 30.73,
       "learning_rate": 0.0003814,
-      "loss": 0.2378,
+      "loss": 0.19,
       "step": 1690
     },
     {
       "epoch": 30.75,
       "learning_rate": 0.0003813,
-      "loss": 0.2585,
+      "loss": 0.2414,
       "step": 1691
     },
     {
       "epoch": 30.76,
       "learning_rate": 0.0003812,
-      "loss": 0.3175,
+      "loss": 0.2819,
       "step": 1692
     },
     {
       "epoch": 30.78,
       "learning_rate": 0.0003811,
-      "loss": 0.2827,
+      "loss": 0.2148,
       "step": 1693
     },
     {
       "epoch": 30.8,
       "learning_rate": 0.000381,
-      "loss": 0.2963,
+      "loss": 0.2399,
       "step": 1694
     },
     {
       "epoch": 30.82,
       "learning_rate": 0.00038090000000000004,
-      "loss": 0.2836,
+      "loss": 0.2557,
       "step": 1695
     },
     {
       "epoch": 30.84,
       "learning_rate": 0.00038080000000000004,
-      "loss": 0.3157,
+      "loss": 0.2981,
       "step": 1696
     },
     {
       "epoch": 30.85,
       "learning_rate": 0.0003807,
-      "loss": 0.2524,
+      "loss": 0.2453,
       "step": 1697
     },
     {
       "epoch": 30.87,
       "learning_rate": 0.0003806,
-      "loss": 0.2966,
+      "loss": 0.2582,
       "step": 1698
     },
     {
       "epoch": 30.89,
       "learning_rate": 0.00038050000000000003,
-      "loss": 0.3207,
+      "loss": 0.2936,
       "step": 1699
     },
     {
       "epoch": 30.91,
       "learning_rate": 0.00038040000000000003,
-      "loss": 0.2721,
+      "loss": 0.2797,
       "step": 1700
     },
     {
       "epoch": 30.93,
       "learning_rate": 0.00038030000000000003,
-      "loss": 0.2545,
+      "loss": 0.2295,
       "step": 1701
     },
     {
       "epoch": 30.95,
       "learning_rate": 0.00038019999999999997,
-      "loss": 0.239,
+      "loss": 0.2411,
       "step": 1702
     },
     {
       "epoch": 30.96,
       "learning_rate": 0.0003801,
-      "loss": 0.3009,
+      "loss": 0.298,
       "step": 1703
     },
     {
       "epoch": 30.98,
       "learning_rate": 0.00038,
-      "loss": 0.3165,
+      "loss": 0.3074,
       "step": 1704
     },
     {
       "epoch": 31.0,
       "learning_rate": 0.0003799,
-      "loss": 0.4182,
+      "loss": 0.3655,
       "step": 1705
     },
     {
       "epoch": 31.02,
       "learning_rate": 0.0003798,
-      "loss": 0.2333,
+      "loss": 0.2272,
       "step": 1706
     },
     {
       "epoch": 31.04,
       "learning_rate": 0.0003797,
-      "loss": 0.2627,
+      "loss": 0.2504,
       "step": 1707
     },
     {
       "epoch": 31.05,
       "learning_rate": 0.0003796,
-      "loss": 0.2459,
+      "loss": 0.2321,
       "step": 1708
     },
     {
       "epoch": 31.07,
       "learning_rate": 0.0003795,
-      "loss": 0.2581,
+      "loss": 0.2974,
       "step": 1709
     },
     {
       "epoch": 31.09,
       "learning_rate": 0.0003794,
-      "loss": 0.2697,
+      "loss": 0.268,
       "step": 1710
     },
     {
       "epoch": 31.11,
       "learning_rate": 0.00037930000000000006,
-      "loss": 0.2541,
+      "loss": 0.2197,
       "step": 1711
     },
     {
       "epoch": 31.13,
       "learning_rate": 0.0003792,
-      "loss": 0.269,
+      "loss": 0.2552,
       "step": 1712
     },
     {
       "epoch": 31.15,
       "learning_rate": 0.0003791,
-      "loss": 0.2575,
+      "loss": 0.2513,
       "step": 1713
     },
     {
       "epoch": 31.16,
       "learning_rate": 0.000379,
-      "loss": 0.3125,
+      "loss": 0.255,
       "step": 1714
     },
     {
       "epoch": 31.18,
       "learning_rate": 0.00037890000000000005,
-      "loss": 0.2632,
+      "loss": 0.2308,
       "step": 1715
     },
     {
       "epoch": 31.2,
       "learning_rate": 0.00037880000000000005,
-      "loss": 0.2821,
+      "loss": 0.2874,
       "step": 1716
     },
     {
       "epoch": 31.22,
       "learning_rate": 0.0003787,
-      "loss": 0.3559,
+      "loss": 0.3417,
       "step": 1717
     },
     {
       "epoch": 31.24,
       "learning_rate": 0.0003786,
-      "loss": 0.2684,
+      "loss": 0.2497,
       "step": 1718
     },
     {
       "epoch": 31.25,
       "learning_rate": 0.0003785,
-      "loss": 0.2862,
+      "loss": 0.2635,
       "step": 1719
     },
     {
       "epoch": 31.27,
       "learning_rate": 0.00037840000000000004,
-      "loss": 0.2674,
+      "loss": 0.2743,
       "step": 1720
     },
     {
       "epoch": 31.29,
       "learning_rate": 0.00037830000000000003,
-      "loss": 0.2892,
+      "loss": 0.2609,
       "step": 1721
     },
     {
       "epoch": 31.31,
       "learning_rate": 0.0003782,
-      "loss": 0.273,
+      "loss": 0.2275,
       "step": 1722
     },
     {
       "epoch": 31.33,
       "learning_rate": 0.0003781,
-      "loss": 0.2616,
+      "loss": 0.3091,
       "step": 1723
     },
     {
       "epoch": 31.35,
       "learning_rate": 0.000378,
-      "loss": 0.2849,
+      "loss": 0.2773,
       "step": 1724
     },
     {
       "epoch": 31.36,
       "learning_rate": 0.0003779,
-      "loss": 0.3186,
+      "loss": 0.3141,
       "step": 1725
     },
     {
       "epoch": 31.38,
       "learning_rate": 0.0003778,
-      "loss": 0.2511,
+      "loss": 0.2479,
       "step": 1726
     },
     {
       "epoch": 31.4,
       "learning_rate": 0.00037769999999999997,
-      "loss": 0.2358,
+      "loss": 0.2432,
       "step": 1727
     },
     {
       "epoch": 31.42,
       "learning_rate": 0.0003776,
-      "loss": 0.2813,
+      "loss": 0.2733,
       "step": 1728
     },
     {
       "epoch": 31.44,
       "learning_rate": 0.0003775,
-      "loss": 0.2829,
+      "loss": 0.2734,
       "step": 1729
     },
     {
       "epoch": 31.45,
       "learning_rate": 0.0003774,
-      "loss": 0.2421,
+      "loss": 0.1991,
       "step": 1730
     },
     {
       "epoch": 31.47,
       "learning_rate": 0.0003773,
-      "loss": 0.275,
+      "loss": 0.2427,
       "step": 1731
     },
     {
       "epoch": 31.49,
       "learning_rate": 0.0003772,
-      "loss": 0.2323,
+      "loss": 0.2417,
       "step": 1732
     },
     {
       "epoch": 31.51,
       "learning_rate": 0.0003771,
-      "loss": 0.2632,
+      "loss": 0.289,
       "step": 1733
     },
     {
       "epoch": 31.53,
       "learning_rate": 0.000377,
-      "loss": 0.3095,
+      "loss": 0.2822,
       "step": 1734
     },
     {
       "epoch": 31.55,
       "learning_rate": 0.0003769,
-      "loss": 0.236,
+      "loss": 0.2341,
       "step": 1735
     },
     {
       "epoch": 31.56,
       "learning_rate": 0.00037680000000000005,
-      "loss": 0.2979,
+      "loss": 0.2693,
       "step": 1736
     },
     {
       "epoch": 31.58,
       "learning_rate": 0.0003767,
-      "loss": 0.2167,
+      "loss": 0.2042,
       "step": 1737
     },
     {
       "epoch": 31.6,
       "learning_rate": 0.0003766,
-      "loss": 0.2835,
+      "loss": 0.2626,
       "step": 1738
     },
     {
       "epoch": 31.62,
       "learning_rate": 0.0003765,
-      "loss": 0.2358,
+      "loss": 0.2461,
       "step": 1739
     },
     {
       "epoch": 31.64,
       "learning_rate": 0.00037640000000000004,
-      "loss": 0.3161,
+      "loss": 0.2772,
       "step": 1740
     },
     {
       "epoch": 31.65,
       "learning_rate": 0.00037630000000000004,
-      "loss": 0.2947,
+      "loss": 0.2813,
       "step": 1741
     },
     {
       "epoch": 31.67,
       "learning_rate": 0.0003762,
-      "loss": 0.2855,
+      "loss": 0.2752,
       "step": 1742
     },
     {
       "epoch": 31.69,
       "learning_rate": 0.0003761,
-      "loss": 0.2826,
+      "loss": 0.2712,
       "step": 1743
     },
     {
       "epoch": 31.71,
       "learning_rate": 0.00037600000000000003,
-      "loss": 0.2746,
+      "loss": 0.2763,
       "step": 1744
     },
     {
       "epoch": 31.73,
       "learning_rate": 0.00037590000000000003,
-      "loss": 0.2356,
+      "loss": 0.2062,
       "step": 1745
     },
     {
       "epoch": 31.75,
       "learning_rate": 0.00037580000000000003,
-      "loss": 0.2798,
+      "loss": 0.2755,
       "step": 1746
     },
     {
       "epoch": 31.76,
       "learning_rate": 0.00037569999999999997,
-      "loss": 0.3097,
+      "loss": 0.2873,
       "step": 1747
     },
     {
       "epoch": 31.78,
       "learning_rate": 0.0003756,
-      "loss": 0.1708,
+      "loss": 0.1901,
       "step": 1748
     },
     {
       "epoch": 31.8,
       "learning_rate": 0.0003755,
-      "loss": 0.2665,
+      "loss": 0.2375,
       "step": 1749
     },
     {
       "epoch": 31.82,
       "learning_rate": 0.0003754,
-      "loss": 0.3132,
+      "loss": 0.3068,
       "step": 1750
     },
     {
       "epoch": 31.84,
       "learning_rate": 0.0003753,
-      "loss": 0.3156,
+      "loss": 0.2802,
       "step": 1751
     },
     {
       "epoch": 31.85,
       "learning_rate": 0.0003752,
-      "loss": 0.252,
+      "loss": 0.2115,
       "step": 1752
     },
     {
       "epoch": 31.87,
       "learning_rate": 0.0003751,
-      "loss": 0.2784,
+      "loss": 0.2623,
       "step": 1753
     },
     {
       "epoch": 31.89,
       "learning_rate": 0.000375,
-      "loss": 0.3233,
+      "loss": 0.2869,
       "step": 1754
     },
     {
       "epoch": 31.91,
       "learning_rate": 0.0003749,
-      "loss": 0.2337,
+      "loss": 0.2108,
       "step": 1755
     },
     {
       "epoch": 31.93,
       "learning_rate": 0.0003748,
-      "loss": 0.2716,
+      "loss": 0.235,
       "step": 1756
     },
     {
       "epoch": 31.95,
       "learning_rate": 0.0003747,
-      "loss": 0.2963,
+      "loss": 0.2439,
       "step": 1757
     },
     {
       "epoch": 31.96,
       "learning_rate": 0.0003746,
-      "loss": 0.3429,
+      "loss": 0.2597,
       "step": 1758
     },
     {
       "epoch": 31.98,
       "learning_rate": 0.0003745,
-      "loss": 0.3292,
+      "loss": 0.2743,
       "step": 1759
     },
     {
       "epoch": 32.0,
       "learning_rate": 0.0003744,
-      "loss": 0.1541,
+      "loss": 0.1828,
       "step": 1760
     },
     {
       "epoch": 32.02,
       "learning_rate": 0.00037430000000000005,
-      "loss": 0.2774,
+      "loss": 0.2075,
       "step": 1761
     },
     {
       "epoch": 32.04,
       "learning_rate": 0.0003742,
-      "loss": 0.2789,
+      "loss": 0.227,
       "step": 1762
     },
     {
       "epoch": 32.05,
       "learning_rate": 0.0003741,
-      "loss": 0.3285,
+      "loss": 0.2674,
       "step": 1763
     },
     {
       "epoch": 32.07,
       "learning_rate": 0.000374,
-      "loss": 0.2495,
+      "loss": 0.2563,
       "step": 1764
     },
     {
       "epoch": 32.09,
       "learning_rate": 0.00037390000000000004,
-      "loss": 0.253,
+      "loss": 0.2249,
       "step": 1765
     },
     {
       "epoch": 32.11,
       "learning_rate": 0.00037380000000000003,
-      "loss": 0.2627,
+      "loss": 0.2207,
       "step": 1766
     },
     {
       "epoch": 32.13,
       "learning_rate": 0.0003737,
-      "loss": 0.2821,
+      "loss": 0.2763,
       "step": 1767
     },
     {
       "epoch": 32.15,
       "learning_rate": 0.0003736,
-      "loss": 0.3065,
+      "loss": 0.2788,
       "step": 1768
     },
     {
       "epoch": 32.16,
       "learning_rate": 0.0003735,
-      "loss": 0.2813,
+      "loss": 0.2249,
       "step": 1769
     },
     {
       "epoch": 32.18,
       "learning_rate": 0.0003734,
-      "loss": 0.2602,
+      "loss": 0.2254,
       "step": 1770
     },
     {
       "epoch": 32.2,
       "learning_rate": 0.0003733,
-      "loss": 0.2144,
+      "loss": 0.2186,
       "step": 1771
     },
     {
       "epoch": 32.22,
       "learning_rate": 0.00037319999999999996,
-      "loss": 0.29,
+      "loss": 0.2637,
       "step": 1772
     },
     {
       "epoch": 32.24,
       "learning_rate": 0.0003731,
-      "loss": 0.2976,
+      "loss": 0.2387,
       "step": 1773
     },
     {
       "epoch": 32.25,
       "learning_rate": 0.000373,
-      "loss": 0.2741,
+      "loss": 0.2374,
       "step": 1774
     },
     {
       "epoch": 32.27,
       "learning_rate": 0.0003729,
-      "loss": 0.2582,
+      "loss": 0.2349,
       "step": 1775
     },
     {
       "epoch": 32.29,
       "learning_rate": 0.0003728,
-      "loss": 0.2772,
+      "loss": 0.2722,
       "step": 1776
     },
     {
       "epoch": 32.31,
       "learning_rate": 0.0003727,
-      "loss": 0.271,
+      "loss": 0.2318,
       "step": 1777
     },
     {
       "epoch": 32.33,
       "learning_rate": 0.0003726,
-      "loss": 0.3346,
+      "loss": 0.2855,
       "step": 1778
     },
     {
       "epoch": 32.35,
       "learning_rate": 0.0003725,
-      "loss": 0.2684,
+      "loss": 0.236,
       "step": 1779
     },
     {
       "epoch": 32.36,
       "learning_rate": 0.0003724,
-      "loss": 0.2639,
+      "loss": 0.224,
       "step": 1780
     },
     {
       "epoch": 32.38,
       "learning_rate": 0.00037230000000000005,
-      "loss": 0.2683,
+      "loss": 0.269,
       "step": 1781
     },
     {
       "epoch": 32.4,
       "learning_rate": 0.0003722,
-      "loss": 0.3281,
+      "loss": 0.3163,
       "step": 1782
     },
     {
       "epoch": 32.42,
       "learning_rate": 0.0003721,
-      "loss": 0.2244,
+      "loss": 0.1938,
       "step": 1783
     },
     {
       "epoch": 32.44,
       "learning_rate": 0.000372,
-      "loss": 0.2826,
+      "loss": 0.2635,
       "step": 1784
     },
     {
       "epoch": 32.45,
       "learning_rate": 0.00037190000000000004,
-      "loss": 0.2381,
+      "loss": 0.2221,
       "step": 1785
     },
     {
       "epoch": 32.47,
       "learning_rate": 0.00037180000000000004,
-      "loss": 0.2746,
+      "loss": 0.2674,
       "step": 1786
     },
     {
       "epoch": 32.49,
       "learning_rate": 0.0003717,
-      "loss": 0.224,
+      "loss": 0.2312,
       "step": 1787
     },
     {
       "epoch": 32.51,
       "learning_rate": 0.0003716,
-      "loss": 0.3058,
+      "loss": 0.2727,
       "step": 1788
     },
     {
       "epoch": 32.53,
       "learning_rate": 0.00037150000000000003,
-      "loss": 0.2665,
+      "loss": 0.2263,
       "step": 1789
     },
     {
       "epoch": 32.55,
       "learning_rate": 0.00037140000000000003,
-      "loss": 0.2722,
+      "loss": 0.2134,
       "step": 1790
     },
     {
       "epoch": 32.56,
       "learning_rate": 0.0003713,
-      "loss": 0.2819,
+      "loss": 0.273,
       "step": 1791
     },
     {
       "epoch": 32.58,
       "learning_rate": 0.00037119999999999997,
-      "loss": 0.2284,
+      "loss": 0.2563,
       "step": 1792
     },
     {
       "epoch": 32.6,
       "learning_rate": 0.00037109999999999997,
-      "loss": 0.2516,
+      "loss": 0.2284,
       "step": 1793
     },
     {
       "epoch": 32.62,
       "learning_rate": 0.000371,
-      "loss": 0.275,
+      "loss": 0.2455,
       "step": 1794
     },
     {
       "epoch": 32.64,
       "learning_rate": 0.0003709,
-      "loss": 0.3325,
+      "loss": 0.2927,
       "step": 1795
     },
     {
       "epoch": 32.65,
       "learning_rate": 0.0003708,
-      "loss": 0.256,
+      "loss": 0.2422,
       "step": 1796
     },
     {
       "epoch": 32.67,
       "learning_rate": 0.00037069999999999996,
-      "loss": 0.2444,
+      "loss": 0.2162,
       "step": 1797
     },
     {
       "epoch": 32.69,
       "learning_rate": 0.0003706,
-      "loss": 0.2296,
+      "loss": 0.1962,
       "step": 1798
     },
     {
       "epoch": 32.71,
       "learning_rate": 0.0003705,
-      "loss": 0.2899,
+      "loss": 0.2978,
       "step": 1799
     },
     {
       "epoch": 32.73,
       "learning_rate": 0.0003704,
-      "loss": 0.2608,
+      "loss": 0.271,
       "step": 1800
     },
     {
       "epoch": 32.75,
       "learning_rate": 0.0003703,
-      "loss": 0.2205,
+      "loss": 0.2327,
       "step": 1801
     },
     {
       "epoch": 32.76,
       "learning_rate": 0.0003702,
-      "loss": 0.2809,
+      "loss": 0.2375,
       "step": 1802
     },
     {
       "epoch": 32.78,
       "learning_rate": 0.0003701,
-      "loss": 0.25,
+      "loss": 0.1611,
       "step": 1803
     },
     {
       "epoch": 32.8,
       "learning_rate": 0.00037,
-      "loss": 0.2902,
+      "loss": 0.3003,
       "step": 1804
     },
     {
       "epoch": 32.82,
       "learning_rate": 0.0003699,
-      "loss": 0.2692,
+      "loss": 0.2898,
       "step": 1805
     },
     {
       "epoch": 32.84,
       "learning_rate": 0.00036980000000000004,
-      "loss": 0.2494,
+      "loss": 0.2363,
       "step": 1806
     },
     {
       "epoch": 32.85,
       "learning_rate": 0.0003697,
-      "loss": 0.2818,
+      "loss": 0.2898,
       "step": 1807
     },
     {
       "epoch": 32.87,
       "learning_rate": 0.0003696,
-      "loss": 0.2477,
+      "loss": 0.2422,
       "step": 1808
     },
     {
       "epoch": 32.89,
       "learning_rate": 0.0003695,
-      "loss": 0.2716,
+      "loss": 0.3303,
       "step": 1809
     },
     {
       "epoch": 32.91,
       "learning_rate": 0.00036940000000000003,
-      "loss": 0.2574,
+      "loss": 0.2823,
       "step": 1810
     },
     {
       "epoch": 32.93,
       "learning_rate": 0.00036930000000000003,
-      "loss": 0.3099,
+      "loss": 0.2792,
       "step": 1811
     },
     {
       "epoch": 32.95,
       "learning_rate": 0.0003692,
-      "loss": 0.3139,
+      "loss": 0.2792,
       "step": 1812
     },
     {
       "epoch": 32.96,
       "learning_rate": 0.00036909999999999997,
-      "loss": 0.2508,
+      "loss": 0.219,
       "step": 1813
     },
     {
       "epoch": 32.98,
       "learning_rate": 0.000369,
-      "loss": 0.2946,
+      "loss": 0.2464,
       "step": 1814
     },
     {
       "epoch": 33.0,
       "learning_rate": 0.0003689,
-      "loss": 0.2177,
+      "loss": 0.2676,
       "step": 1815
     },
     {
       "epoch": 33.02,
       "learning_rate": 0.0003688,
-      "loss": 0.2338,
+      "loss": 0.2103,
       "step": 1816
     },
     {
       "epoch": 33.04,
       "learning_rate": 0.0003687,
-      "loss": 0.2837,
+      "loss": 0.2831,
       "step": 1817
     },
     {
       "epoch": 33.05,
       "learning_rate": 0.0003686,
-      "loss": 0.1927,
+      "loss": 0.1783,
       "step": 1818
     },
     {
       "epoch": 33.07,
       "learning_rate": 0.0003685,
-      "loss": 0.2229,
+      "loss": 0.243,
       "step": 1819
     },
     {
       "epoch": 33.09,
       "learning_rate": 0.0003684,
-      "loss": 0.2974,
+      "loss": 0.2533,
       "step": 1820
     },
     {
       "epoch": 33.11,
       "learning_rate": 0.0003683,
-      "loss": 0.2388,
+      "loss": 0.2208,
       "step": 1821
     },
     {
       "epoch": 33.13,
       "learning_rate": 0.00036820000000000006,
-      "loss": 0.2451,
+      "loss": 0.2409,
       "step": 1822
     },
     {
       "epoch": 33.15,
       "learning_rate": 0.0003681,
-      "loss": 0.2532,
+      "loss": 0.2241,
       "step": 1823
     },
     {
       "epoch": 33.16,
       "learning_rate": 0.000368,
-      "loss": 0.2193,
+      "loss": 0.2004,
       "step": 1824
     },
     {
       "epoch": 33.18,
       "learning_rate": 0.0003679,
-      "loss": 0.2602,
+      "loss": 0.2816,
       "step": 1825
     },
     {
       "epoch": 33.2,
       "learning_rate": 0.00036780000000000005,
-      "loss": 0.336,
+      "loss": 0.2997,
       "step": 1826
     },
     {
       "epoch": 33.22,
       "learning_rate": 0.00036770000000000005,
-      "loss": 0.2503,
+      "loss": 0.2121,
       "step": 1827
     },
     {
       "epoch": 33.24,
       "learning_rate": 0.0003676,
-      "loss": 0.1789,
+      "loss": 0.1806,
       "step": 1828
     },
     {
       "epoch": 33.25,
       "learning_rate": 0.0003675,
-      "loss": 0.2537,
+      "loss": 0.2367,
       "step": 1829
     },
     {
       "epoch": 33.27,
       "learning_rate": 0.0003674,
-      "loss": 0.2369,
+      "loss": 0.2149,
       "step": 1830
     },
     {
       "epoch": 33.29,
       "learning_rate": 0.00036730000000000004,
-      "loss": 0.2668,
+      "loss": 0.256,
       "step": 1831
     },
     {
       "epoch": 33.31,
       "learning_rate": 0.00036720000000000004,
-      "loss": 0.2623,
+      "loss": 0.2358,
       "step": 1832
     },
     {
       "epoch": 33.33,
       "learning_rate": 0.0003671,
-      "loss": 0.2455,
+      "loss": 0.2016,
       "step": 1833
     },
     {
       "epoch": 33.35,
       "learning_rate": 0.000367,
-      "loss": 0.2496,
+      "loss": 0.2358,
       "step": 1834
     },
     {
       "epoch": 33.36,
       "learning_rate": 0.00036690000000000003,
-      "loss": 0.2317,
+      "loss": 0.2289,
       "step": 1835
     },
     {
       "epoch": 33.38,
       "learning_rate": 0.0003668,
-      "loss": 0.2964,
+      "loss": 0.2632,
       "step": 1836
     },
     {
       "epoch": 33.4,
       "learning_rate": 0.0003667,
-      "loss": 0.2772,
+      "loss": 0.2711,
       "step": 1837
     },
     {
       "epoch": 33.42,
       "learning_rate": 0.00036659999999999997,
-      "loss": 0.1834,
+      "loss": 0.1862,
       "step": 1838
     },
     {
       "epoch": 33.44,
       "learning_rate": 0.0003665,
-      "loss": 0.2538,
+      "loss": 0.2466,
       "step": 1839
     },
     {
       "epoch": 33.45,
       "learning_rate": 0.0003664,
-      "loss": 0.2244,
+      "loss": 0.2153,
       "step": 1840
     },
     {
       "epoch": 33.47,
       "learning_rate": 0.0003663,
-      "loss": 0.223,
+      "loss": 0.227,
       "step": 1841
     },
     {
       "epoch": 33.49,
       "learning_rate": 0.0003662,
-      "loss": 0.2173,
+      "loss": 0.2585,
       "step": 1842
     },
     {
       "epoch": 33.51,
       "learning_rate": 0.0003661,
-      "loss": 0.223,
+      "loss": 0.249,
       "step": 1843
     },
     {
       "epoch": 33.53,
       "learning_rate": 0.000366,
-      "loss": 0.2329,
+      "loss": 0.2467,
       "step": 1844
     },
     {
       "epoch": 33.55,
       "learning_rate": 0.0003659,
-      "loss": 0.2501,
+      "loss": 0.2439,
       "step": 1845
     },
     {
       "epoch": 33.56,
       "learning_rate": 0.0003658,
-      "loss": 0.2359,
+      "loss": 0.2634,
       "step": 1846
     },
     {
       "epoch": 33.58,
       "learning_rate": 0.00036570000000000005,
-      "loss": 0.2183,
+      "loss": 0.2356,
       "step": 1847
     },
     {
       "epoch": 33.6,
       "learning_rate": 0.0003656,
-      "loss": 0.2794,
+      "loss": 0.2833,
       "step": 1848
     },
     {
       "epoch": 33.62,
       "learning_rate": 0.0003655,
-      "loss": 0.279,
+      "loss": 0.2695,
       "step": 1849
     },
     {
       "epoch": 33.64,
       "learning_rate": 0.0003654,
-      "loss": 0.2465,
+      "loss": 0.2349,
       "step": 1850
     },
     {
       "epoch": 33.65,
       "learning_rate": 0.00036530000000000004,
-      "loss": 0.2488,
+      "loss": 0.2174,
       "step": 1851
     },
     {
       "epoch": 33.67,
       "learning_rate": 0.00036520000000000004,
-      "loss": 0.2146,
+      "loss": 0.2061,
       "step": 1852
     },
     {
       "epoch": 33.69,
       "learning_rate": 0.0003651,
-      "loss": 0.2389,
+      "loss": 0.2264,
       "step": 1853
     },
     {
       "epoch": 33.71,
       "learning_rate": 0.000365,
-      "loss": 0.2545,
+      "loss": 0.2428,
       "step": 1854
     },
     {
       "epoch": 33.73,
       "learning_rate": 0.00036490000000000003,
-      "loss": 0.2859,
+      "loss": 0.3006,
       "step": 1855
     },
     {
       "epoch": 33.75,
       "learning_rate": 0.00036480000000000003,
-      "loss": 0.2159,
+      "loss": 0.2507,
       "step": 1856
     },
     {
       "epoch": 33.76,
       "learning_rate": 0.00036470000000000003,
-      "loss": 0.2593,
+      "loss": 0.2289,
       "step": 1857
     },
     {
       "epoch": 33.78,
       "learning_rate": 0.00036459999999999997,
-      "loss": 0.2535,
+      "loss": 0.2345,
       "step": 1858
     },
     {
       "epoch": 33.8,
       "learning_rate": 0.0003645,
-      "loss": 0.2379,
+      "loss": 0.2122,
       "step": 1859
     },
     {
       "epoch": 33.82,
       "learning_rate": 0.0003644,
-      "loss": 0.2929,
+      "loss": 0.2971,
       "step": 1860
     },
     {
       "epoch": 33.84,
       "learning_rate": 0.0003643,
-      "loss": 0.2108,
+      "loss": 0.2229,
       "step": 1861
     },
     {
       "epoch": 33.85,
       "learning_rate": 0.0003642,
-      "loss": 0.2506,
+      "loss": 0.2486,
       "step": 1862
     },
     {
       "epoch": 33.87,
       "learning_rate": 0.00036409999999999996,
-      "loss": 0.2873,
+      "loss": 0.2771,
       "step": 1863
     },
     {
       "epoch": 33.89,
       "learning_rate": 0.000364,
-      "loss": 0.2658,
+      "loss": 0.2441,
       "step": 1864
     },
     {
       "epoch": 33.91,
       "learning_rate": 0.0003639,
-      "loss": 0.3134,
+      "loss": 0.2501,
       "step": 1865
     },
     {
       "epoch": 33.93,
       "learning_rate": 0.0003638,
-      "loss": 0.2457,
+      "loss": 0.2549,
       "step": 1866
     },
     {
       "epoch": 33.95,
       "learning_rate": 0.0003637,
-      "loss": 0.2435,
+      "loss": 0.2391,
       "step": 1867
     },
     {
       "epoch": 33.96,
       "learning_rate": 0.0003636,
-      "loss": 0.2379,
+      "loss": 0.256,
       "step": 1868
     },
     {
       "epoch": 33.98,
       "learning_rate": 0.0003635,
-      "loss": 0.2338,
+      "loss": 0.2887,
       "step": 1869
     },
     {
       "epoch": 34.0,
       "learning_rate": 0.0003634,
-      "loss": 0.1937,
+      "loss": 0.2379,
       "step": 1870
     },
     {
       "epoch": 34.02,
       "learning_rate": 0.0003633,
-      "loss": 0.2506,
+      "loss": 0.2597,
       "step": 1871
     },
     {
       "epoch": 34.04,
       "learning_rate": 0.00036320000000000005,
-      "loss": 0.2211,
+      "loss": 0.2325,
       "step": 1872
     },
     {
       "epoch": 34.05,
       "learning_rate": 0.0003631,
-      "loss": 0.2734,
+      "loss": 0.2388,
       "step": 1873
     },
     {
       "epoch": 34.07,
       "learning_rate": 0.000363,
-      "loss": 0.2262,
+      "loss": 0.2526,
       "step": 1874
     },
     {
       "epoch": 34.09,
       "learning_rate": 0.0003629,
-      "loss": 0.2609,
+      "loss": 0.2642,
       "step": 1875
     },
     {
       "epoch": 34.11,
       "learning_rate": 0.00036280000000000004,
-      "loss": 0.2965,
+      "loss": 0.2419,
       "step": 1876
     },
     {
       "epoch": 34.13,
       "learning_rate": 0.00036270000000000003,
-      "loss": 0.282,
+      "loss": 0.2982,
       "step": 1877
     },
     {
       "epoch": 34.15,
       "learning_rate": 0.0003626,
-      "loss": 0.2052,
+      "loss": 0.2096,
       "step": 1878
     },
     {
       "epoch": 34.16,
       "learning_rate": 0.0003625,
-      "loss": 0.226,
+      "loss": 0.2239,
       "step": 1879
     },
     {
       "epoch": 34.18,
       "learning_rate": 0.0003624,
-      "loss": 0.2285,
+      "loss": 0.2428,
       "step": 1880
     },
     {
       "epoch": 34.2,
       "learning_rate": 0.0003623,
-      "loss": 0.2129,
+      "loss": 0.2637,
       "step": 1881
     },
     {
       "epoch": 34.22,
       "learning_rate": 0.0003622,
-      "loss": 0.3141,
+      "loss": 0.266,
       "step": 1882
     },
     {
       "epoch": 34.24,
       "learning_rate": 0.00036209999999999997,
-      "loss": 0.2895,
+      "loss": 0.2309,
       "step": 1883
     },
     {
       "epoch": 34.25,
       "learning_rate": 0.000362,
-      "loss": 0.2559,
+      "loss": 0.2261,
       "step": 1884
     },
     {
       "epoch": 34.27,
       "learning_rate": 0.0003619,
-      "loss": 0.2925,
+      "loss": 0.2456,
       "step": 1885
     },
     {
       "epoch": 34.29,
       "learning_rate": 0.0003618,
-      "loss": 0.2764,
+      "loss": 0.2259,
       "step": 1886
     },
     {
       "epoch": 34.31,
       "learning_rate": 0.0003617,
-      "loss": 0.2733,
+      "loss": 0.2251,
       "step": 1887
     },
     {
       "epoch": 34.33,
       "learning_rate": 0.0003616,
-      "loss": 0.2603,
+      "loss": 0.2177,
       "step": 1888
     },
     {
       "epoch": 34.35,
       "learning_rate": 0.0003615,
-      "loss": 0.2786,
+      "loss": 0.2227,
       "step": 1889
     },
     {
       "epoch": 34.36,
       "learning_rate": 0.0003614,
-      "loss": 0.2644,
+      "loss": 0.2497,
       "step": 1890
     },
     {
       "epoch": 34.38,
       "learning_rate": 0.0003613,
-      "loss": 0.2499,
+      "loss": 0.2221,
       "step": 1891
     },
     {
       "epoch": 34.4,
       "learning_rate": 0.00036120000000000005,
-      "loss": 0.2687,
+      "loss": 0.2352,
       "step": 1892
     },
     {
       "epoch": 34.42,
       "learning_rate": 0.0003611,
-      "loss": 0.2299,
+      "loss": 0.2023,
       "step": 1893
     },
     {
       "epoch": 34.44,
       "learning_rate": 0.000361,
-      "loss": 0.2693,
+      "loss": 0.2515,
       "step": 1894
     },
     {
       "epoch": 34.45,
       "learning_rate": 0.0003609,
-      "loss": 0.2201,
+      "loss": 0.1821,
       "step": 1895
     },
     {
       "epoch": 34.47,
       "learning_rate": 0.00036080000000000004,
-      "loss": 0.2693,
+      "loss": 0.2473,
       "step": 1896
     },
     {
       "epoch": 34.49,
       "learning_rate": 0.00036070000000000004,
-      "loss": 0.1997,
+      "loss": 0.1575,
       "step": 1897
     },
     {
       "epoch": 34.51,
       "learning_rate": 0.0003606,
-      "loss": 0.2226,
+      "loss": 0.2485,
       "step": 1898
     },
     {
       "epoch": 34.53,
       "learning_rate": 0.0003605,
-      "loss": 0.2628,
+      "loss": 0.2417,
       "step": 1899
     },
     {
       "epoch": 34.55,
       "learning_rate": 0.00036040000000000003,
-      "loss": 0.2773,
+      "loss": 0.3127,
       "step": 1900
     },
     {
       "epoch": 34.56,
       "learning_rate": 0.00036030000000000003,
-      "loss": 0.265,
+      "loss": 0.2398,
       "step": 1901
     },
     {
       "epoch": 34.58,
       "learning_rate": 0.00036020000000000003,
-      "loss": 0.2453,
+      "loss": 0.2432,
       "step": 1902
     },
     {
       "epoch": 34.6,
       "learning_rate": 0.00036009999999999997,
-      "loss": 0.2614,
+      "loss": 0.2742,
       "step": 1903
     },
     {
       "epoch": 34.62,
       "learning_rate": 0.00035999999999999997,
-      "loss": 0.2528,
+      "loss": 0.233,
       "step": 1904
     },
     {
       "epoch": 34.64,
       "learning_rate": 0.0003599,
-      "loss": 0.2741,
+      "loss": 0.2441,
       "step": 1905
     },
     {
       "epoch": 34.65,
       "learning_rate": 0.0003598,
-      "loss": 0.2462,
+      "loss": 0.2528,
       "step": 1906
     },
     {
       "epoch": 34.67,
       "learning_rate": 0.0003597,
-      "loss": 0.2792,
+      "loss": 0.273,
       "step": 1907
     },
     {
       "epoch": 34.69,
       "learning_rate": 0.00035959999999999996,
-      "loss": 0.2357,
+      "loss": 0.284,
       "step": 1908
     },
     {
       "epoch": 34.71,
       "learning_rate": 0.0003595,
-      "loss": 0.2523,
+      "loss": 0.2455,
       "step": 1909
     },
     {
       "epoch": 34.73,
       "learning_rate": 0.0003594,
-      "loss": 0.1992,
+      "loss": 0.1946,
       "step": 1910
     },
     {
       "epoch": 34.75,
       "learning_rate": 0.0003593,
-      "loss": 0.2165,
+      "loss": 0.1874,
       "step": 1911
     },
     {
       "epoch": 34.76,
       "learning_rate": 0.0003592,
-      "loss": 0.2471,
+      "loss": 0.2184,
       "step": 1912
     },
     {
       "epoch": 34.78,
       "learning_rate": 0.0003591,
-      "loss": 0.2451,
+      "loss": 0.2429,
       "step": 1913
     },
     {
       "epoch": 34.8,
       "learning_rate": 0.000359,
-      "loss": 0.2522,
+      "loss": 0.2453,
       "step": 1914
     },
     {
       "epoch": 34.82,
       "learning_rate": 0.0003589,
-      "loss": 0.2085,
+      "loss": 0.2361,
       "step": 1915
     },
     {
       "epoch": 34.84,
       "learning_rate": 0.0003588,
-      "loss": 0.2803,
+      "loss": 0.2658,
       "step": 1916
     },
     {
       "epoch": 34.85,
       "learning_rate": 0.00035870000000000005,
-      "loss": 0.2444,
+      "loss": 0.2137,
       "step": 1917
     },
     {
       "epoch": 34.87,
       "learning_rate": 0.0003586,
-      "loss": 0.2125,
+      "loss": 0.2006,
       "step": 1918
     },
     {
       "epoch": 34.89,
       "learning_rate": 0.0003585,
-      "loss": 0.254,
+      "loss": 0.2482,
       "step": 1919
     },
     {
       "epoch": 34.91,
       "learning_rate": 0.0003584,
-      "loss": 0.2953,
+      "loss": 0.2768,
       "step": 1920
     },
     {
       "epoch": 34.93,
       "learning_rate": 0.00035830000000000004,
-      "loss": 0.2328,
+      "loss": 0.2345,
       "step": 1921
     },
     {
       "epoch": 34.95,
       "learning_rate": 0.00035820000000000003,
-      "loss": 0.2029,
+      "loss": 0.2159,
       "step": 1922
     },
     {
       "epoch": 34.96,
       "learning_rate": 0.0003581,
-      "loss": 0.3525,
+      "loss": 0.3589,
       "step": 1923
     },
     {
       "epoch": 34.98,
       "learning_rate": 0.000358,
-      "loss": 0.2584,
+      "loss": 0.2722,
       "step": 1924
     },
     {
       "epoch": 35.0,
       "learning_rate": 0.0003579,
-      "loss": 0.2637,
+      "loss": 0.2028,
       "step": 1925
     },
     {
       "epoch": 35.02,
       "learning_rate": 0.0003578,
-      "loss": 0.3088,
+      "loss": 0.3101,
       "step": 1926
     },
     {
       "epoch": 35.04,
       "learning_rate": 0.0003577,
-      "loss": 0.2986,
+      "loss": 0.2728,
       "step": 1927
     },
     {
       "epoch": 35.05,
       "learning_rate": 0.00035759999999999996,
-      "loss": 0.2161,
+      "loss": 0.2117,
       "step": 1928
     },
     {
       "epoch": 35.07,
       "learning_rate": 0.0003575,
-      "loss": 0.2309,
+      "loss": 0.2559,
       "step": 1929
     },
     {
       "epoch": 35.09,
       "learning_rate": 0.0003574,
-      "loss": 0.2293,
+      "loss": 0.219,
       "step": 1930
     },
     {
       "epoch": 35.11,
       "learning_rate": 0.0003573,
-      "loss": 0.2489,
+      "loss": 0.2537,
       "step": 1931
     },
     {
       "epoch": 35.13,
       "learning_rate": 0.0003572,
-      "loss": 0.2561,
+      "loss": 0.2401,
       "step": 1932
     },
     {
       "epoch": 35.15,
       "learning_rate": 0.0003571,
-      "loss": 0.2701,
+      "loss": 0.2737,
       "step": 1933
     },
     {
       "epoch": 35.16,
       "learning_rate": 0.000357,
-      "loss": 0.1897,
+      "loss": 0.1754,
       "step": 1934
     },
     {
       "epoch": 35.18,
       "learning_rate": 0.0003569,
-      "loss": 0.3118,
+      "loss": 0.2877,
       "step": 1935
     },
     {
       "epoch": 35.2,
       "learning_rate": 0.0003568,
-      "loss": 0.2615,
+      "loss": 0.2514,
       "step": 1936
     },
     {
       "epoch": 35.22,
       "learning_rate": 0.00035670000000000005,
-      "loss": 0.2286,
+      "loss": 0.2335,
       "step": 1937
     },
     {
       "epoch": 35.24,
       "learning_rate": 0.0003566,
-      "loss": 0.2106,
+      "loss": 0.1881,
       "step": 1938
     },
     {
       "epoch": 35.25,
       "learning_rate": 0.0003565,
-      "loss": 0.2416,
+      "loss": 0.2471,
       "step": 1939
     },
     {
       "epoch": 35.27,
       "learning_rate": 0.0003564,
-      "loss": 0.2218,
+      "loss": 0.226,
       "step": 1940
     },
     {
       "epoch": 35.29,
       "learning_rate": 0.0003563,
-      "loss": 0.244,
+      "loss": 0.2378,
       "step": 1941
     },
     {
       "epoch": 35.31,
       "learning_rate": 0.00035620000000000004,
-      "loss": 0.2932,
+      "loss": 0.2658,
       "step": 1942
     },
     {
       "epoch": 35.33,
       "learning_rate": 0.00035610000000000004,
-      "loss": 0.2736,
+      "loss": 0.2537,
       "step": 1943
     },
     {
       "epoch": 35.35,
       "learning_rate": 0.000356,
-      "loss": 0.256,
+      "loss": 0.2296,
       "step": 1944
     },
     {
       "epoch": 35.36,
       "learning_rate": 0.0003559,
-      "loss": 0.2747,
+      "loss": 0.2787,
       "step": 1945
     },
     {
       "epoch": 35.38,
       "learning_rate": 0.00035580000000000003,
-      "loss": 0.2787,
+      "loss": 0.2607,
       "step": 1946
     },
     {
       "epoch": 35.4,
       "learning_rate": 0.0003557,
-      "loss": 0.2279,
+      "loss": 0.2416,
       "step": 1947
     },
     {
       "epoch": 35.42,
       "learning_rate": 0.0003556,
-      "loss": 0.2101,
+      "loss": 0.1982,
       "step": 1948
     },
     {
       "epoch": 35.44,
       "learning_rate": 0.00035549999999999997,
-      "loss": 0.2688,
+      "loss": 0.2264,
       "step": 1949
     },
     {
       "epoch": 35.45,
       "learning_rate": 0.0003554,
-      "loss": 0.2316,
+      "loss": 0.1914,
       "step": 1950
     },
     {
       "epoch": 35.47,
       "learning_rate": 0.0003553,
-      "loss": 0.285,
+      "loss": 0.2587,
       "step": 1951
     },
     {
       "epoch": 35.49,
       "learning_rate": 0.0003552,
-      "loss": 0.2554,
+      "loss": 0.2066,
       "step": 1952
     },
     {
       "epoch": 35.51,
       "learning_rate": 0.0003551,
-      "loss": 0.2737,
+      "loss": 0.2336,
       "step": 1953
     },
     {
       "epoch": 35.53,
       "learning_rate": 0.000355,
-      "loss": 0.2893,
+      "loss": 0.2668,
       "step": 1954
     },
     {
       "epoch": 35.55,
       "learning_rate": 0.0003549,
-      "loss": 0.2757,
+      "loss": 0.2958,
       "step": 1955
     },
     {
       "epoch": 35.56,
       "learning_rate": 0.0003548,
-      "loss": 0.2307,
+      "loss": 0.233,
       "step": 1956
     },
     {
       "epoch": 35.58,
       "learning_rate": 0.0003547,
-      "loss": 0.2761,
+      "loss": 0.2672,
       "step": 1957
     },
     {
       "epoch": 35.6,
       "learning_rate": 0.00035460000000000005,
-      "loss": 0.2825,
+      "loss": 0.2512,
       "step": 1958
     },
     {
       "epoch": 35.62,
       "learning_rate": 0.0003545,
-      "loss": 0.2345,
+      "loss": 0.2253,
       "step": 1959
     },
     {
       "epoch": 35.64,
       "learning_rate": 0.0003544,
-      "loss": 0.2933,
+      "loss": 0.2897,
       "step": 1960
     },
     {
       "epoch": 35.65,
       "learning_rate": 0.0003543,
-      "loss": 0.2397,
+      "loss": 0.2175,
       "step": 1961
     },
     {
       "epoch": 35.67,
       "learning_rate": 0.00035420000000000004,
-      "loss": 0.2211,
+      "loss": 0.2271,
       "step": 1962
     },
     {
       "epoch": 35.69,
       "learning_rate": 0.00035410000000000004,
-      "loss": 0.2709,
+      "loss": 0.2392,
       "step": 1963
     },
     {
       "epoch": 35.71,
       "learning_rate": 0.000354,
-      "loss": 0.2703,
+      "loss": 0.2462,
       "step": 1964
     },
     {
       "epoch": 35.73,
       "learning_rate": 0.0003539,
-      "loss": 0.2335,
+      "loss": 0.2273,
       "step": 1965
     },
     {
       "epoch": 35.75,
       "learning_rate": 0.00035380000000000003,
-      "loss": 0.2593,
+      "loss": 0.2223,
       "step": 1966
     },
     {
       "epoch": 35.76,
       "learning_rate": 0.00035370000000000003,
-      "loss": 0.2137,
+      "loss": 0.1772,
       "step": 1967
     },
     {
       "epoch": 35.78,
       "learning_rate": 0.00035360000000000003,
-      "loss": 0.2403,
+      "loss": 0.2409,
       "step": 1968
     },
     {
       "epoch": 35.8,
       "learning_rate": 0.0003535,
-      "loss": 0.2707,
+      "loss": 0.25,
       "step": 1969
     },
     {
       "epoch": 35.82,
       "learning_rate": 0.0003534,
-      "loss": 0.3057,
+      "loss": 0.2589,
       "step": 1970
     },
     {
       "epoch": 35.84,
       "learning_rate": 0.0003533,
-      "loss": 0.2024,
+      "loss": 0.2139,
       "step": 1971
     },
     {
       "epoch": 35.85,
       "learning_rate": 0.0003532,
-      "loss": 0.2624,
+      "loss": 0.2755,
       "step": 1972
     },
     {
       "epoch": 35.87,
       "learning_rate": 0.0003531,
-      "loss": 0.2588,
+      "loss": 0.2344,
       "step": 1973
     },
     {
       "epoch": 35.89,
       "learning_rate": 0.00035299999999999996,
-      "loss": 0.196,
+      "loss": 0.2072,
       "step": 1974
     },
     {
       "epoch": 35.91,
       "learning_rate": 0.0003529,
-      "loss": 0.2424,
+      "loss": 0.2012,
       "step": 1975
     },
     {
       "epoch": 35.93,
       "learning_rate": 0.0003528,
-      "loss": 0.2331,
+      "loss": 0.2014,
       "step": 1976
     },
     {
       "epoch": 35.95,
       "learning_rate": 0.0003527,
-      "loss": 0.2024,
+      "loss": 0.1902,
       "step": 1977
     },
     {
       "epoch": 35.96,
       "learning_rate": 0.0003526,
-      "loss": 0.2599,
+      "loss": 0.2998,
       "step": 1978
     },
     {
       "epoch": 35.98,
       "learning_rate": 0.0003525,
-      "loss": 0.25,
+      "loss": 0.2494,
       "step": 1979
     },
     {
       "epoch": 36.0,
       "learning_rate": 0.0003524,
-      "loss": 0.2672,
+      "loss": 0.249,
       "step": 1980
     },
     {
       "epoch": 36.02,
       "learning_rate": 0.0003523,
-      "loss": 0.2459,
+      "loss": 0.2328,
       "step": 1981
     },
     {
       "epoch": 36.04,
       "learning_rate": 0.0003522,
-      "loss": 0.2112,
+      "loss": 0.2281,
       "step": 1982
     },
     {
       "epoch": 36.05,
       "learning_rate": 0.00035210000000000005,
-      "loss": 0.2294,
+      "loss": 0.2232,
       "step": 1983
     },
     {
       "epoch": 36.07,
       "learning_rate": 0.000352,
-      "loss": 0.2509,
+      "loss": 0.2537,
       "step": 1984
     },
     {
       "epoch": 36.09,
       "learning_rate": 0.0003519,
-      "loss": 0.2858,
+      "loss": 0.278,
       "step": 1985
     },
     {
       "epoch": 36.11,
       "learning_rate": 0.0003518,
-      "loss": 0.1867,
+      "loss": 0.1832,
       "step": 1986
     },
     {
       "epoch": 36.13,
       "learning_rate": 0.00035170000000000004,
-      "loss": 0.2232,
+      "loss": 0.1977,
       "step": 1987
     },
     {
       "epoch": 36.15,
       "learning_rate": 0.00035160000000000004,
-      "loss": 0.2018,
+      "loss": 0.2338,
       "step": 1988
     },
     {
       "epoch": 36.16,
       "learning_rate": 0.0003515,
-      "loss": 0.2034,
+      "loss": 0.1888,
       "step": 1989
     },
     {
       "epoch": 36.18,
       "learning_rate": 0.0003514,
-      "loss": 0.2447,
+      "loss": 0.2399,
       "step": 1990
     },
     {
       "epoch": 36.2,
       "learning_rate": 0.00035130000000000003,
-      "loss": 0.2385,
+      "loss": 0.2239,
       "step": 1991
     },
     {
       "epoch": 36.22,
       "learning_rate": 0.0003512,
-      "loss": 0.2089,
+      "loss": 0.2169,
       "step": 1992
     },
     {
       "epoch": 36.24,
       "learning_rate": 0.0003511,
-      "loss": 0.2142,
+      "loss": 0.1969,
       "step": 1993
     },
     {
       "epoch": 36.25,
       "learning_rate": 0.00035099999999999997,
-      "loss": 0.2805,
+      "loss": 0.2596,
       "step": 1994
     },
     {
       "epoch": 36.27,
       "learning_rate": 0.0003509,
-      "loss": 0.2552,
+      "loss": 0.2097,
       "step": 1995
     },
     {
       "epoch": 36.29,
       "learning_rate": 0.0003508,
-      "loss": 0.2471,
+      "loss": 0.2702,
       "step": 1996
     },
     {
       "epoch": 36.31,
       "learning_rate": 0.0003507,
-      "loss": 0.2928,
+      "loss": 0.2537,
       "step": 1997
     },
     {
       "epoch": 36.33,
       "learning_rate": 0.0003506,
-      "loss": 0.2559,
+      "loss": 0.2044,
       "step": 1998
     },
     {
       "epoch": 36.35,
       "learning_rate": 0.0003505,
-      "loss": 0.2113,
+      "loss": 0.1826,
       "step": 1999
     },
     {
       "epoch": 36.36,
       "learning_rate": 0.0003504,
-      "loss": 0.2853,
+      "loss": 0.2649,
       "step": 2000
     },
     {
       "epoch": 36.36,
-      "eval_cer": 0.09586350619949324,
-      "eval_loss": 0.417502224445343,
-      "eval_runtime": 13.8056,
-      "eval_samples_per_second": 119.3,
-      "eval_steps_per_second": 1.883,
-      "eval_wer": 0.38065570421815953,
+      "eval_cer": 0.09398941513366017,
+      "eval_loss": 0.43297263979911804,
+      "eval_runtime": 14.2255,
+      "eval_samples_per_second": 115.778,
+      "eval_steps_per_second": 1.828,
+      "eval_wer": 0.38290266571341025,
       "step": 2000
     },
     {
       "epoch": 36.38,
       "learning_rate": 0.0003503,
-      "loss": 0.2583,
+      "loss": 0.2187,
       "step": 2001
     },
     {
       "epoch": 36.4,
       "learning_rate": 0.0003502,
-      "loss": 0.2327,
+      "loss": 0.2117,
       "step": 2002
     },
     {
       "epoch": 36.42,
       "learning_rate": 0.00035010000000000005,
-      "loss": 0.2564,
+      "loss": 0.2196,
       "step": 2003
     },
     {
       "epoch": 36.44,
       "learning_rate": 0.00035,
-      "loss": 0.2391,
+      "loss": 0.2201,
       "step": 2004
     },
     {
       "epoch": 36.45,
       "learning_rate": 0.0003499,
-      "loss": 0.2701,
+      "loss": 0.2051,
       "step": 2005
     },
     {
       "epoch": 36.47,
       "learning_rate": 0.0003498,
-      "loss": 0.2245,
+      "loss": 0.2055,
       "step": 2006
     },
     {
       "epoch": 36.49,
       "learning_rate": 0.00034970000000000004,
-      "loss": 0.2206,
+      "loss": 0.2131,
       "step": 2007
     },
     {
       "epoch": 36.51,
       "learning_rate": 0.00034960000000000004,
-      "loss": 0.2292,
+      "loss": 0.2244,
       "step": 2008
     },
     {
       "epoch": 36.53,
       "learning_rate": 0.0003495,
-      "loss": 0.248,
+      "loss": 0.2379,
       "step": 2009
     },
     {
       "epoch": 36.55,
       "learning_rate": 0.0003494,
-      "loss": 0.2514,
+      "loss": 0.2251,
       "step": 2010
     },
     {
       "epoch": 36.56,
       "learning_rate": 0.00034930000000000003,
-      "loss": 0.2412,
+      "loss": 0.2449,
       "step": 2011
     },
     {
       "epoch": 36.58,
       "learning_rate": 0.00034920000000000003,
-      "loss": 0.2276,
+      "loss": 0.172,
       "step": 2012
     },
     {
       "epoch": 36.6,
       "learning_rate": 0.00034910000000000003,
-      "loss": 0.2299,
+      "loss": 0.211,
       "step": 2013
     },
     {
       "epoch": 36.62,
       "learning_rate": 0.00034899999999999997,
-      "loss": 0.2005,
+      "loss": 0.2086,
       "step": 2014
     },
     {
       "epoch": 36.64,
       "learning_rate": 0.00034889999999999997,
-      "loss": 0.2624,
+      "loss": 0.2516,
       "step": 2015
     },
     {
       "epoch": 36.65,
       "learning_rate": 0.0003488,
-      "loss": 0.2081,
+      "loss": 0.1943,
       "step": 2016
     },
     {
       "epoch": 36.67,
       "learning_rate": 0.0003487,
-      "loss": 0.2298,
+      "loss": 0.1858,
       "step": 2017
     },
     {
       "epoch": 36.69,
       "learning_rate": 0.0003486,
-      "loss": 0.2384,
+      "loss": 0.2057,
       "step": 2018
     },
     {
       "epoch": 36.71,
       "learning_rate": 0.00034849999999999996,
-      "loss": 0.27,
+      "loss": 0.2614,
       "step": 2019
     },
     {
       "epoch": 36.73,
       "learning_rate": 0.0003484,
-      "loss": 0.22,
+      "loss": 0.2146,
       "step": 2020
     },
     {
       "epoch": 36.75,
       "learning_rate": 0.0003483,
-      "loss": 0.2564,
+      "loss": 0.2203,
       "step": 2021
     },
     {
       "epoch": 36.76,
       "learning_rate": 0.0003482,
-      "loss": 0.2918,
+      "loss": 0.2383,
       "step": 2022
     },
     {
       "epoch": 36.78,
       "learning_rate": 0.0003481,
-      "loss": 0.2577,
+      "loss": 0.2111,
       "step": 2023
     },
     {
       "epoch": 36.8,
       "learning_rate": 0.000348,
-      "loss": 0.2246,
+      "loss": 0.2157,
       "step": 2024
     },
     {
       "epoch": 36.82,
       "learning_rate": 0.0003479,
-      "loss": 0.2789,
+      "loss": 0.2637,
       "step": 2025
     },
     {
       "epoch": 36.84,
       "learning_rate": 0.0003478,
-      "loss": 0.2669,
+      "loss": 0.2654,
       "step": 2026
     },
     {
       "epoch": 36.85,
       "learning_rate": 0.0003477,
-      "loss": 0.2835,
+      "loss": 0.2673,
       "step": 2027
     },
     {
       "epoch": 36.87,
       "learning_rate": 0.00034760000000000005,
-      "loss": 0.2679,
+      "loss": 0.2634,
       "step": 2028
     },
     {
       "epoch": 36.89,
       "learning_rate": 0.0003475,
-      "loss": 0.242,
+      "loss": 0.2596,
       "step": 2029
     },
     {
       "epoch": 36.91,
       "learning_rate": 0.0003474,
-      "loss": 0.2325,
+      "loss": 0.2152,
       "step": 2030
     },
     {
       "epoch": 36.93,
       "learning_rate": 0.0003473,
-      "loss": 0.2638,
+      "loss": 0.2797,
       "step": 2031
     },
     {
       "epoch": 36.95,
       "learning_rate": 0.00034720000000000004,
-      "loss": 0.299,
+      "loss": 0.3167,
       "step": 2032
     },
     {
       "epoch": 36.96,
       "learning_rate": 0.00034710000000000003,
-      "loss": 0.2199,
+      "loss": 0.1698,
       "step": 2033
     },
     {
       "epoch": 36.98,
       "learning_rate": 0.000347,
-      "loss": 0.2376,
+      "loss": 0.234,
       "step": 2034
     },
     {
       "epoch": 37.0,
       "learning_rate": 0.0003469,
-      "loss": 0.1587,
+      "loss": 0.1368,
       "step": 2035
     },
     {
       "epoch": 37.02,
       "learning_rate": 0.0003468,
-      "loss": 0.2693,
+      "loss": 0.2514,
       "step": 2036
     },
     {
       "epoch": 37.04,
       "learning_rate": 0.0003467,
-      "loss": 0.2439,
+      "loss": 0.2341,
       "step": 2037
     },
     {
       "epoch": 37.05,
       "learning_rate": 0.0003466,
-      "loss": 0.2882,
+      "loss": 0.2742,
       "step": 2038
     },
     {
       "epoch": 37.07,
       "learning_rate": 0.00034649999999999997,
-      "loss": 0.2789,
+      "loss": 0.2381,
       "step": 2039
     },
     {
       "epoch": 37.09,
       "learning_rate": 0.0003464,
-      "loss": 0.3062,
+      "loss": 0.2379,
       "step": 2040
     },
     {
       "epoch": 37.11,
       "learning_rate": 0.0003463,
-      "loss": 0.2057,
+      "loss": 0.2237,
       "step": 2041
     },
     {
       "epoch": 37.13,
       "learning_rate": 0.0003462,
-      "loss": 0.2924,
+      "loss": 0.2629,
       "step": 2042
     },
     {
       "epoch": 37.15,
       "learning_rate": 0.0003461,
-      "loss": 0.2398,
+      "loss": 0.2064,
       "step": 2043
     },
     {
       "epoch": 37.16,
       "learning_rate": 0.000346,
-      "loss": 0.257,
+      "loss": 0.2255,
       "step": 2044
     },
     {
       "epoch": 37.18,
       "learning_rate": 0.0003459,
-      "loss": 0.2283,
+      "loss": 0.2194,
       "step": 2045
     },
     {
       "epoch": 37.2,
       "learning_rate": 0.0003458,
-      "loss": 0.2423,
+      "loss": 0.2203,
       "step": 2046
     },
     {
       "epoch": 37.22,
       "learning_rate": 0.0003457,
-      "loss": 0.2721,
+      "loss": 0.2739,
       "step": 2047
     },
     {
       "epoch": 37.24,
       "learning_rate": 0.00034560000000000005,
-      "loss": 0.2134,
+      "loss": 0.197,
       "step": 2048
     },
     {
       "epoch": 37.25,
       "learning_rate": 0.0003455,
-      "loss": 0.2748,
+      "loss": 0.2424,
       "step": 2049
     },
     {
       "epoch": 37.27,
       "learning_rate": 0.0003454,
-      "loss": 0.2721,
+      "loss": 0.2313,
       "step": 2050
     },
     {
       "epoch": 37.29,
       "learning_rate": 0.0003453,
-      "loss": 0.2281,
+      "loss": 0.2405,
       "step": 2051
     },
     {
@@ -12362,16743 +12362,16743 @@
     {
       "epoch": 37.33,
       "learning_rate": 0.00034510000000000004,
-      "loss": 0.2222,
+      "loss": 0.1999,
       "step": 2053
     },
     {
       "epoch": 37.35,
       "learning_rate": 0.000345,
-      "loss": 0.2677,
+      "loss": 0.2277,
       "step": 2054
     },
     {
       "epoch": 37.36,
       "learning_rate": 0.0003449,
-      "loss": 0.2352,
+      "loss": 0.2134,
       "step": 2055
     },
     {
       "epoch": 37.38,
       "learning_rate": 0.0003448,
-      "loss": 0.269,
+      "loss": 0.2344,
       "step": 2056
     },
     {
       "epoch": 37.4,
       "learning_rate": 0.00034470000000000003,
-      "loss": 0.1859,
+      "loss": 0.1947,
       "step": 2057
     },
     {
       "epoch": 37.42,
       "learning_rate": 0.00034460000000000003,
-      "loss": 0.2711,
+      "loss": 0.2584,
       "step": 2058
     },
     {
       "epoch": 37.44,
       "learning_rate": 0.00034449999999999997,
-      "loss": 0.2292,
+      "loss": 0.2299,
       "step": 2059
     },
     {
       "epoch": 37.45,
       "learning_rate": 0.00034439999999999997,
-      "loss": 0.2235,
+      "loss": 0.2252,
       "step": 2060
     },
     {
       "epoch": 37.47,
       "learning_rate": 0.0003443,
-      "loss": 0.2747,
+      "loss": 0.2223,
       "step": 2061
     },
     {
       "epoch": 37.49,
       "learning_rate": 0.0003442,
-      "loss": 0.2476,
+      "loss": 0.2497,
       "step": 2062
     },
     {
       "epoch": 37.51,
       "learning_rate": 0.0003441,
-      "loss": 0.1799,
+      "loss": 0.1709,
       "step": 2063
     },
     {
       "epoch": 37.53,
       "learning_rate": 0.00034399999999999996,
-      "loss": 0.2022,
+      "loss": 0.1916,
       "step": 2064
     },
     {
       "epoch": 37.55,
       "learning_rate": 0.0003439,
-      "loss": 0.222,
+      "loss": 0.232,
       "step": 2065
     },
     {
       "epoch": 37.56,
       "learning_rate": 0.0003438,
-      "loss": 0.222,
+      "loss": 0.2349,
       "step": 2066
     },
     {
       "epoch": 37.58,
       "learning_rate": 0.0003437,
-      "loss": 0.2261,
+      "loss": 0.2044,
       "step": 2067
     },
     {
       "epoch": 37.6,
       "learning_rate": 0.0003436,
-      "loss": 0.2592,
+      "loss": 0.2458,
       "step": 2068
     },
     {
       "epoch": 37.62,
       "learning_rate": 0.00034350000000000006,
-      "loss": 0.236,
+      "loss": 0.2065,
       "step": 2069
     },
     {
       "epoch": 37.64,
       "learning_rate": 0.0003434,
-      "loss": 0.2402,
+      "loss": 0.2132,
       "step": 2070
     },
     {
       "epoch": 37.65,
       "learning_rate": 0.0003433,
-      "loss": 0.2058,
+      "loss": 0.2505,
       "step": 2071
     },
     {
       "epoch": 37.67,
       "learning_rate": 0.0003432,
-      "loss": 0.3142,
+      "loss": 0.312,
       "step": 2072
     },
     {
       "epoch": 37.69,
       "learning_rate": 0.00034310000000000005,
-      "loss": 0.236,
+      "loss": 0.2084,
       "step": 2073
     },
     {
       "epoch": 37.71,
       "learning_rate": 0.00034300000000000004,
-      "loss": 0.2292,
+      "loss": 0.2412,
       "step": 2074
     },
     {
       "epoch": 37.73,
       "learning_rate": 0.0003429,
-      "loss": 0.2639,
+      "loss": 0.2496,
       "step": 2075
     },
     {
       "epoch": 37.75,
       "learning_rate": 0.0003428,
-      "loss": 0.2487,
+      "loss": 0.2192,
       "step": 2076
     },
     {
       "epoch": 37.76,
       "learning_rate": 0.00034270000000000004,
-      "loss": 0.2727,
+      "loss": 0.2304,
       "step": 2077
     },
     {
       "epoch": 37.78,
       "learning_rate": 0.00034260000000000003,
-      "loss": 0.2834,
+      "loss": 0.2699,
       "step": 2078
     },
     {
       "epoch": 37.8,
       "learning_rate": 0.00034250000000000003,
-      "loss": 0.238,
+      "loss": 0.229,
       "step": 2079
     },
     {
       "epoch": 37.82,
       "learning_rate": 0.0003424,
-      "loss": 0.2698,
+      "loss": 0.2386,
       "step": 2080
     },
     {
       "epoch": 37.84,
       "learning_rate": 0.0003423,
-      "loss": 0.2329,
+      "loss": 0.1849,
       "step": 2081
     },
     {
       "epoch": 37.85,
       "learning_rate": 0.0003422,
-      "loss": 0.2492,
+      "loss": 0.2363,
       "step": 2082
     },
     {
       "epoch": 37.87,
       "learning_rate": 0.0003421,
-      "loss": 0.2793,
+      "loss": 0.2334,
       "step": 2083
     },
     {
       "epoch": 37.89,
       "learning_rate": 0.000342,
-      "loss": 0.2243,
+      "loss": 0.2016,
       "step": 2084
     },
     {
       "epoch": 37.91,
       "learning_rate": 0.00034189999999999996,
-      "loss": 0.2536,
+      "loss": 0.2205,
       "step": 2085
     },
     {
       "epoch": 37.93,
       "learning_rate": 0.0003418,
-      "loss": 0.2688,
+      "loss": 0.2405,
       "step": 2086
     },
     {
       "epoch": 37.95,
       "learning_rate": 0.0003417,
-      "loss": 0.2194,
+      "loss": 0.2157,
       "step": 2087
     },
     {
       "epoch": 37.96,
       "learning_rate": 0.0003416,
-      "loss": 0.2517,
+      "loss": 0.2147,
       "step": 2088
     },
     {
       "epoch": 37.98,
       "learning_rate": 0.0003415,
-      "loss": 0.2207,
+      "loss": 0.2129,
       "step": 2089
     },
     {
       "epoch": 38.0,
       "learning_rate": 0.0003414,
-      "loss": 0.2032,
+      "loss": 0.1358,
       "step": 2090
     },
     {
       "epoch": 38.02,
       "learning_rate": 0.0003413,
-      "loss": 0.2704,
+      "loss": 0.2507,
       "step": 2091
     },
     {
       "epoch": 38.04,
       "learning_rate": 0.0003412,
-      "loss": 0.2505,
+      "loss": 0.2191,
       "step": 2092
     },
     {
       "epoch": 38.05,
       "learning_rate": 0.0003411,
-      "loss": 0.2368,
+      "loss": 0.2138,
       "step": 2093
     },
     {
       "epoch": 38.07,
       "learning_rate": 0.00034100000000000005,
-      "loss": 0.2734,
+      "loss": 0.2544,
       "step": 2094
     },
     {
       "epoch": 38.09,
       "learning_rate": 0.0003409,
-      "loss": 0.2052,
+      "loss": 0.2159,
       "step": 2095
     },
     {
       "epoch": 38.11,
       "learning_rate": 0.0003408,
-      "loss": 0.2178,
+      "loss": 0.232,
       "step": 2096
     },
     {
       "epoch": 38.13,
       "learning_rate": 0.0003407,
-      "loss": 0.2338,
+      "loss": 0.2011,
       "step": 2097
     },
     {
       "epoch": 38.15,
       "learning_rate": 0.00034060000000000004,
-      "loss": 0.2514,
+      "loss": 0.2451,
       "step": 2098
     },
     {
       "epoch": 38.16,
       "learning_rate": 0.00034050000000000004,
-      "loss": 0.2045,
+      "loss": 0.2126,
       "step": 2099
     },
     {
       "epoch": 38.18,
       "learning_rate": 0.0003404,
-      "loss": 0.3074,
+      "loss": 0.3006,
       "step": 2100
     },
     {
       "epoch": 38.2,
       "learning_rate": 0.0003403,
-      "loss": 0.2531,
+      "loss": 0.2334,
       "step": 2101
     },
     {
       "epoch": 38.22,
       "learning_rate": 0.00034020000000000003,
-      "loss": 0.194,
+      "loss": 0.1732,
       "step": 2102
     },
     {
       "epoch": 38.24,
       "learning_rate": 0.0003401,
-      "loss": 0.1666,
+      "loss": 0.1778,
       "step": 2103
     },
     {
       "epoch": 38.25,
       "learning_rate": 0.00034,
-      "loss": 0.2615,
+      "loss": 0.2458,
       "step": 2104
     },
     {
       "epoch": 38.27,
       "learning_rate": 0.00033989999999999997,
-      "loss": 0.185,
+      "loss": 0.178,
       "step": 2105
     },
     {
       "epoch": 38.29,
       "learning_rate": 0.0003398,
-      "loss": 0.2467,
+      "loss": 0.2076,
       "step": 2106
     },
     {
       "epoch": 38.31,
       "learning_rate": 0.0003397,
-      "loss": 0.2488,
+      "loss": 0.2466,
       "step": 2107
     },
     {
       "epoch": 38.33,
       "learning_rate": 0.0003396,
-      "loss": 0.2779,
+      "loss": 0.2124,
       "step": 2108
     },
     {
       "epoch": 38.35,
       "learning_rate": 0.0003395,
-      "loss": 0.2413,
+      "loss": 0.239,
       "step": 2109
     },
     {
       "epoch": 38.36,
       "learning_rate": 0.0003394,
-      "loss": 0.2224,
+      "loss": 0.2176,
       "step": 2110
     },
     {
       "epoch": 38.38,
       "learning_rate": 0.0003393,
-      "loss": 0.2022,
+      "loss": 0.241,
       "step": 2111
     },
     {
       "epoch": 38.4,
       "learning_rate": 0.0003392,
-      "loss": 0.2497,
+      "loss": 0.2708,
       "step": 2112
     },
     {
       "epoch": 38.42,
       "learning_rate": 0.0003391,
-      "loss": 0.2158,
+      "loss": 0.232,
       "step": 2113
     },
     {
       "epoch": 38.44,
       "learning_rate": 0.00033900000000000005,
-      "loss": 0.2228,
+      "loss": 0.2175,
       "step": 2114
     },
     {
       "epoch": 38.45,
       "learning_rate": 0.0003389,
-      "loss": 0.191,
+      "loss": 0.1669,
       "step": 2115
     },
     {
       "epoch": 38.47,
       "learning_rate": 0.0003388,
-      "loss": 0.2397,
+      "loss": 0.2108,
       "step": 2116
     },
     {
       "epoch": 38.49,
       "learning_rate": 0.0003387,
-      "loss": 0.2486,
+      "loss": 0.2163,
       "step": 2117
     },
     {
       "epoch": 38.51,
       "learning_rate": 0.00033860000000000004,
-      "loss": 0.2263,
+      "loss": 0.2068,
       "step": 2118
     },
     {
       "epoch": 38.53,
       "learning_rate": 0.00033850000000000004,
-      "loss": 0.2098,
+      "loss": 0.1718,
       "step": 2119
     },
     {
       "epoch": 38.55,
       "learning_rate": 0.0003384,
-      "loss": 0.2264,
+      "loss": 0.2074,
       "step": 2120
     },
     {
       "epoch": 38.56,
       "learning_rate": 0.0003383,
-      "loss": 0.2287,
+      "loss": 0.2259,
       "step": 2121
     },
     {
       "epoch": 38.58,
       "learning_rate": 0.0003382,
-      "loss": 0.3024,
+      "loss": 0.2629,
       "step": 2122
     },
     {
       "epoch": 38.6,
       "learning_rate": 0.00033810000000000003,
-      "loss": 0.2228,
+      "loss": 0.2024,
       "step": 2123
     },
     {
       "epoch": 38.62,
       "learning_rate": 0.00033800000000000003,
-      "loss": 0.2426,
+      "loss": 0.2044,
       "step": 2124
     },
     {
       "epoch": 38.64,
       "learning_rate": 0.0003379,
-      "loss": 0.2106,
+      "loss": 0.1962,
       "step": 2125
     },
     {
       "epoch": 38.65,
       "learning_rate": 0.00033779999999999997,
-      "loss": 0.2827,
+      "loss": 0.2806,
       "step": 2126
     },
     {
       "epoch": 38.67,
       "learning_rate": 0.0003377,
-      "loss": 0.2324,
+      "loss": 0.212,
       "step": 2127
     },
     {
       "epoch": 38.69,
       "learning_rate": 0.0003376,
-      "loss": 0.2409,
+      "loss": 0.2563,
       "step": 2128
     },
     {
       "epoch": 38.71,
       "learning_rate": 0.0003375,
-      "loss": 0.2514,
+      "loss": 0.2297,
       "step": 2129
     },
     {
       "epoch": 38.73,
       "learning_rate": 0.00033739999999999996,
-      "loss": 0.2424,
+      "loss": 0.2307,
       "step": 2130
     },
     {
       "epoch": 38.75,
       "learning_rate": 0.0003373,
-      "loss": 0.2801,
+      "loss": 0.2647,
       "step": 2131
     },
     {
       "epoch": 38.76,
       "learning_rate": 0.0003372,
-      "loss": 0.2284,
+      "loss": 0.184,
       "step": 2132
     },
     {
       "epoch": 38.78,
       "learning_rate": 0.0003371,
-      "loss": 0.2562,
+      "loss": 0.1981,
       "step": 2133
     },
     {
       "epoch": 38.8,
       "learning_rate": 0.000337,
-      "loss": 0.2333,
+      "loss": 0.2281,
       "step": 2134
     },
     {
       "epoch": 38.82,
       "learning_rate": 0.0003369,
-      "loss": 0.2252,
+      "loss": 0.2236,
       "step": 2135
     },
     {
       "epoch": 38.84,
       "learning_rate": 0.0003368,
-      "loss": 0.2357,
+      "loss": 0.2082,
       "step": 2136
     },
     {
       "epoch": 38.85,
       "learning_rate": 0.0003367,
-      "loss": 0.2684,
+      "loss": 0.2143,
       "step": 2137
     },
     {
       "epoch": 38.87,
       "learning_rate": 0.0003366,
-      "loss": 0.3182,
+      "loss": 0.2487,
       "step": 2138
     },
     {
       "epoch": 38.89,
       "learning_rate": 0.00033650000000000005,
-      "loss": 0.2874,
+      "loss": 0.2612,
       "step": 2139
     },
     {
       "epoch": 38.91,
       "learning_rate": 0.0003364,
-      "loss": 0.2299,
+      "loss": 0.1709,
       "step": 2140
     },
     {
       "epoch": 38.93,
       "learning_rate": 0.0003363,
-      "loss": 0.2769,
+      "loss": 0.241,
       "step": 2141
     },
     {
       "epoch": 38.95,
       "learning_rate": 0.0003362,
-      "loss": 0.3082,
+      "loss": 0.3012,
       "step": 2142
     },
     {
       "epoch": 38.96,
       "learning_rate": 0.00033610000000000004,
-      "loss": 0.2636,
+      "loss": 0.2571,
       "step": 2143
     },
     {
       "epoch": 38.98,
       "learning_rate": 0.00033600000000000004,
-      "loss": 0.2272,
+      "loss": 0.1928,
       "step": 2144
     },
     {
       "epoch": 39.0,
       "learning_rate": 0.0003359,
-      "loss": 0.2648,
+      "loss": 0.2507,
       "step": 2145
     },
     {
       "epoch": 39.02,
       "learning_rate": 0.0003358,
-      "loss": 0.2482,
+      "loss": 0.2247,
       "step": 2146
     },
     {
       "epoch": 39.04,
       "learning_rate": 0.00033570000000000003,
-      "loss": 0.2332,
+      "loss": 0.217,
       "step": 2147
     },
     {
       "epoch": 39.05,
       "learning_rate": 0.0003356,
-      "loss": 0.2745,
+      "loss": 0.2087,
       "step": 2148
     },
     {
       "epoch": 39.07,
       "learning_rate": 0.0003355,
-      "loss": 0.2299,
+      "loss": 0.2047,
       "step": 2149
     },
     {
       "epoch": 39.09,
       "learning_rate": 0.00033539999999999997,
-      "loss": 0.3026,
+      "loss": 0.2712,
       "step": 2150
     },
     {
       "epoch": 39.11,
       "learning_rate": 0.0003353,
-      "loss": 0.2125,
+      "loss": 0.1759,
       "step": 2151
     },
     {
       "epoch": 39.13,
       "learning_rate": 0.0003352,
-      "loss": 0.2668,
+      "loss": 0.2464,
       "step": 2152
     },
     {
       "epoch": 39.15,
       "learning_rate": 0.0003351,
-      "loss": 0.2229,
+      "loss": 0.2313,
       "step": 2153
     },
     {
       "epoch": 39.16,
       "learning_rate": 0.000335,
-      "loss": 0.22,
+      "loss": 0.198,
       "step": 2154
     },
     {
       "epoch": 39.18,
       "learning_rate": 0.0003349,
-      "loss": 0.2783,
+      "loss": 0.2269,
       "step": 2155
     },
     {
       "epoch": 39.2,
       "learning_rate": 0.0003348,
-      "loss": 0.2893,
+      "loss": 0.2567,
       "step": 2156
     },
     {
       "epoch": 39.22,
       "learning_rate": 0.0003347,
-      "loss": 0.2527,
+      "loss": 0.1964,
       "step": 2157
     },
     {
       "epoch": 39.24,
       "learning_rate": 0.0003346,
-      "loss": 0.1943,
+      "loss": 0.1519,
       "step": 2158
     },
     {
       "epoch": 39.25,
       "learning_rate": 0.00033450000000000005,
-      "loss": 0.2491,
+      "loss": 0.2049,
       "step": 2159
     },
     {
       "epoch": 39.27,
       "learning_rate": 0.0003344,
-      "loss": 0.2411,
+      "loss": 0.1939,
       "step": 2160
     },
     {
       "epoch": 39.29,
       "learning_rate": 0.0003343,
-      "loss": 0.2978,
+      "loss": 0.2357,
       "step": 2161
     },
     {
       "epoch": 39.31,
       "learning_rate": 0.0003342,
-      "loss": 0.2919,
+      "loss": 0.2386,
       "step": 2162
     },
     {
       "epoch": 39.33,
       "learning_rate": 0.0003341,
-      "loss": 0.2599,
+      "loss": 0.2063,
       "step": 2163
     },
     {
       "epoch": 39.35,
       "learning_rate": 0.00033400000000000004,
-      "loss": 0.2272,
+      "loss": 0.1988,
       "step": 2164
     },
     {
       "epoch": 39.36,
       "learning_rate": 0.0003339,
-      "loss": 0.2803,
+      "loss": 0.2453,
       "step": 2165
     },
     {
       "epoch": 39.38,
       "learning_rate": 0.0003338,
-      "loss": 0.2515,
+      "loss": 0.2215,
       "step": 2166
     },
     {
       "epoch": 39.4,
       "learning_rate": 0.0003337,
-      "loss": 0.2462,
+      "loss": 0.1972,
       "step": 2167
     },
     {
       "epoch": 39.42,
       "learning_rate": 0.00033360000000000003,
-      "loss": 0.2841,
+      "loss": 0.2255,
       "step": 2168
     },
     {
       "epoch": 39.44,
       "learning_rate": 0.00033350000000000003,
-      "loss": 0.294,
+      "loss": 0.2615,
       "step": 2169
     },
     {
       "epoch": 39.45,
       "learning_rate": 0.00033339999999999997,
-      "loss": 0.1579,
+      "loss": 0.1722,
       "step": 2170
     },
     {
       "epoch": 39.47,
       "learning_rate": 0.00033329999999999997,
-      "loss": 0.2148,
+      "loss": 0.1869,
       "step": 2171
     },
     {
       "epoch": 39.49,
       "learning_rate": 0.0003332,
-      "loss": 0.2525,
+      "loss": 0.2049,
       "step": 2172
     },
     {
       "epoch": 39.51,
       "learning_rate": 0.0003331,
-      "loss": 0.1948,
+      "loss": 0.2002,
       "step": 2173
     },
     {
       "epoch": 39.53,
       "learning_rate": 0.000333,
-      "loss": 0.2042,
+      "loss": 0.1936,
       "step": 2174
     },
     {
       "epoch": 39.55,
       "learning_rate": 0.00033289999999999996,
-      "loss": 0.2607,
+      "loss": 0.229,
       "step": 2175
     },
     {
       "epoch": 39.56,
       "learning_rate": 0.0003328,
-      "loss": 0.246,
+      "loss": 0.2059,
       "step": 2176
     },
     {
       "epoch": 39.58,
       "learning_rate": 0.0003327,
-      "loss": 0.254,
+      "loss": 0.232,
       "step": 2177
     },
     {
       "epoch": 39.6,
       "learning_rate": 0.0003326,
-      "loss": 0.2274,
+      "loss": 0.1894,
       "step": 2178
     },
     {
       "epoch": 39.62,
       "learning_rate": 0.0003325,
-      "loss": 0.2578,
+      "loss": 0.2274,
       "step": 2179
     },
     {
       "epoch": 39.64,
       "learning_rate": 0.0003324,
-      "loss": 0.2105,
+      "loss": 0.1981,
       "step": 2180
     },
     {
       "epoch": 39.65,
       "learning_rate": 0.0003323,
-      "loss": 0.2102,
+      "loss": 0.2093,
       "step": 2181
     },
     {
       "epoch": 39.67,
       "learning_rate": 0.0003322,
-      "loss": 0.2376,
+      "loss": 0.2201,
       "step": 2182
     },
     {
       "epoch": 39.69,
       "learning_rate": 0.0003321,
-      "loss": 0.1924,
+      "loss": 0.1846,
       "step": 2183
     },
     {
       "epoch": 39.71,
       "learning_rate": 0.00033200000000000005,
-      "loss": 0.2663,
+      "loss": 0.2377,
       "step": 2184
     },
     {
       "epoch": 39.73,
       "learning_rate": 0.0003319,
-      "loss": 0.2154,
+      "loss": 0.244,
       "step": 2185
     },
     {
       "epoch": 39.75,
       "learning_rate": 0.0003318,
-      "loss": 0.2379,
+      "loss": 0.2156,
       "step": 2186
     },
     {
       "epoch": 39.76,
       "learning_rate": 0.0003317,
-      "loss": 0.2534,
+      "loss": 0.2556,
       "step": 2187
     },
     {
       "epoch": 39.78,
       "learning_rate": 0.00033160000000000004,
-      "loss": 0.2632,
+      "loss": 0.2112,
       "step": 2188
     },
     {
       "epoch": 39.8,
       "learning_rate": 0.00033150000000000003,
-      "loss": 0.2184,
+      "loss": 0.1863,
       "step": 2189
     },
     {
       "epoch": 39.82,
       "learning_rate": 0.0003314,
-      "loss": 0.2348,
+      "loss": 0.2016,
       "step": 2190
     },
     {
       "epoch": 39.84,
       "learning_rate": 0.0003313,
-      "loss": 0.2666,
+      "loss": 0.241,
       "step": 2191
     },
     {
       "epoch": 39.85,
       "learning_rate": 0.00033120000000000003,
-      "loss": 0.2284,
+      "loss": 0.2094,
       "step": 2192
     },
     {
       "epoch": 39.87,
       "learning_rate": 0.0003311,
-      "loss": 0.2288,
+      "loss": 0.2201,
       "step": 2193
     },
     {
       "epoch": 39.89,
       "learning_rate": 0.000331,
-      "loss": 0.2318,
+      "loss": 0.2457,
       "step": 2194
     },
     {
       "epoch": 39.91,
       "learning_rate": 0.0003309,
-      "loss": 0.1853,
+      "loss": 0.1718,
       "step": 2195
     },
     {
       "epoch": 39.93,
       "learning_rate": 0.00033079999999999996,
-      "loss": 0.2515,
+      "loss": 0.2301,
       "step": 2196
     },
     {
       "epoch": 39.95,
       "learning_rate": 0.0003307,
-      "loss": 0.2234,
+      "loss": 0.2112,
       "step": 2197
     },
     {
       "epoch": 39.96,
       "learning_rate": 0.0003306,
-      "loss": 0.2697,
+      "loss": 0.2642,
       "step": 2198
     },
     {
       "epoch": 39.98,
       "learning_rate": 0.0003305,
-      "loss": 0.2204,
+      "loss": 0.2039,
       "step": 2199
     },
     {
       "epoch": 40.0,
       "learning_rate": 0.0003304,
-      "loss": 0.2277,
+      "loss": 0.1772,
       "step": 2200
     },
     {
       "epoch": 40.02,
       "learning_rate": 0.0003303,
-      "loss": 0.2303,
+      "loss": 0.2115,
       "step": 2201
     },
     {
       "epoch": 40.04,
       "learning_rate": 0.0003302,
-      "loss": 0.2454,
+      "loss": 0.2419,
       "step": 2202
     },
     {
       "epoch": 40.05,
       "learning_rate": 0.0003301,
-      "loss": 0.255,
+      "loss": 0.2319,
       "step": 2203
     },
     {
       "epoch": 40.07,
       "learning_rate": 0.00033,
-      "loss": 0.2504,
+      "loss": 0.2256,
       "step": 2204
     },
     {
       "epoch": 40.09,
       "learning_rate": 0.00032990000000000005,
-      "loss": 0.2844,
+      "loss": 0.2116,
       "step": 2205
     },
     {
       "epoch": 40.11,
       "learning_rate": 0.0003298,
-      "loss": 0.1866,
+      "loss": 0.173,
       "step": 2206
     },
     {
       "epoch": 40.13,
       "learning_rate": 0.0003297,
-      "loss": 0.2103,
+      "loss": 0.2201,
       "step": 2207
     },
     {
       "epoch": 40.15,
       "learning_rate": 0.0003296,
-      "loss": 0.277,
+      "loss": 0.2255,
       "step": 2208
     },
     {
       "epoch": 40.16,
       "learning_rate": 0.00032950000000000004,
-      "loss": 0.2248,
+      "loss": 0.2286,
       "step": 2209
     },
     {
       "epoch": 40.18,
       "learning_rate": 0.00032940000000000004,
-      "loss": 0.238,
+      "loss": 0.2284,
       "step": 2210
     },
     {
       "epoch": 40.2,
       "learning_rate": 0.0003293,
-      "loss": 0.2435,
+      "loss": 0.2249,
       "step": 2211
     },
     {
       "epoch": 40.22,
       "learning_rate": 0.0003292,
-      "loss": 0.194,
+      "loss": 0.1793,
       "step": 2212
     },
     {
       "epoch": 40.24,
       "learning_rate": 0.00032910000000000003,
-      "loss": 0.2475,
+      "loss": 0.2219,
       "step": 2213
     },
     {
       "epoch": 40.25,
       "learning_rate": 0.00032900000000000003,
-      "loss": 0.2192,
+      "loss": 0.1839,
       "step": 2214
     },
     {
       "epoch": 40.27,
       "learning_rate": 0.0003289,
-      "loss": 0.2494,
+      "loss": 0.211,
       "step": 2215
     },
     {
       "epoch": 40.29,
       "learning_rate": 0.00032879999999999997,
-      "loss": 0.2468,
+      "loss": 0.2214,
       "step": 2216
     },
     {
       "epoch": 40.31,
       "learning_rate": 0.0003287,
-      "loss": 0.2213,
+      "loss": 0.1983,
       "step": 2217
     },
     {
       "epoch": 40.33,
       "learning_rate": 0.0003286,
-      "loss": 0.2663,
+      "loss": 0.2187,
       "step": 2218
     },
     {
       "epoch": 40.35,
       "learning_rate": 0.0003285,
-      "loss": 0.2721,
+      "loss": 0.2245,
       "step": 2219
     },
     {
       "epoch": 40.36,
       "learning_rate": 0.0003284,
-      "loss": 0.2611,
+      "loss": 0.2663,
       "step": 2220
     },
     {
       "epoch": 40.38,
       "learning_rate": 0.0003283,
-      "loss": 0.2061,
+      "loss": 0.2008,
       "step": 2221
     },
     {
       "epoch": 40.4,
       "learning_rate": 0.0003282,
-      "loss": 0.2019,
+      "loss": 0.2119,
       "step": 2222
     },
     {
       "epoch": 40.42,
       "learning_rate": 0.0003281,
-      "loss": 0.248,
+      "loss": 0.2228,
       "step": 2223
     },
     {
       "epoch": 40.44,
       "learning_rate": 0.000328,
-      "loss": 0.2176,
+      "loss": 0.2136,
       "step": 2224
     },
     {
       "epoch": 40.45,
       "learning_rate": 0.00032790000000000006,
-      "loss": 0.2759,
+      "loss": 0.2786,
       "step": 2225
     },
     {
       "epoch": 40.47,
       "learning_rate": 0.0003278,
-      "loss": 0.2548,
+      "loss": 0.2451,
       "step": 2226
     },
     {
       "epoch": 40.49,
       "learning_rate": 0.0003277,
-      "loss": 0.2483,
+      "loss": 0.2061,
       "step": 2227
     },
     {
       "epoch": 40.51,
       "learning_rate": 0.0003276,
-      "loss": 0.21,
+      "loss": 0.1814,
       "step": 2228
     },
     {
       "epoch": 40.53,
       "learning_rate": 0.00032750000000000005,
-      "loss": 0.2659,
+      "loss": 0.2342,
       "step": 2229
     },
     {
       "epoch": 40.55,
       "learning_rate": 0.00032740000000000004,
-      "loss": 0.1922,
+      "loss": 0.1783,
       "step": 2230
     },
     {
       "epoch": 40.56,
       "learning_rate": 0.0003273,
-      "loss": 0.251,
+      "loss": 0.257,
       "step": 2231
     },
     {
       "epoch": 40.58,
       "learning_rate": 0.0003272,
-      "loss": 0.2231,
+      "loss": 0.1955,
       "step": 2232
     },
     {
       "epoch": 40.6,
       "learning_rate": 0.0003271,
-      "loss": 0.3055,
+      "loss": 0.2762,
       "step": 2233
     },
     {
       "epoch": 40.62,
       "learning_rate": 0.00032700000000000003,
-      "loss": 0.2086,
+      "loss": 0.2135,
       "step": 2234
     },
     {
       "epoch": 40.64,
       "learning_rate": 0.00032690000000000003,
-      "loss": 0.2284,
+      "loss": 0.2123,
       "step": 2235
     },
     {
       "epoch": 40.65,
       "learning_rate": 0.0003268,
-      "loss": 0.2816,
+      "loss": 0.2358,
       "step": 2236
     },
     {
       "epoch": 40.67,
       "learning_rate": 0.00032669999999999997,
-      "loss": 0.2934,
+      "loss": 0.2667,
       "step": 2237
     },
     {
       "epoch": 40.69,
       "learning_rate": 0.0003266,
-      "loss": 0.2431,
+      "loss": 0.2368,
       "step": 2238
     },
     {
       "epoch": 40.71,
       "learning_rate": 0.0003265,
-      "loss": 0.2456,
+      "loss": 0.1807,
       "step": 2239
     },
     {
       "epoch": 40.73,
       "learning_rate": 0.0003264,
-      "loss": 0.2429,
+      "loss": 0.2248,
       "step": 2240
     },
     {
       "epoch": 40.75,
       "learning_rate": 0.00032629999999999996,
-      "loss": 0.2682,
+      "loss": 0.222,
       "step": 2241
     },
     {
       "epoch": 40.76,
       "learning_rate": 0.0003262,
-      "loss": 0.1852,
+      "loss": 0.1924,
       "step": 2242
     },
     {
       "epoch": 40.78,
       "learning_rate": 0.0003261,
-      "loss": 0.2152,
+      "loss": 0.1833,
       "step": 2243
     },
     {
       "epoch": 40.8,
       "learning_rate": 0.000326,
-      "loss": 0.2222,
+      "loss": 0.2062,
       "step": 2244
     },
     {
       "epoch": 40.82,
       "learning_rate": 0.0003259,
-      "loss": 0.249,
+      "loss": 0.2463,
       "step": 2245
     },
     {
       "epoch": 40.84,
       "learning_rate": 0.0003258,
-      "loss": 0.2442,
+      "loss": 0.2449,
       "step": 2246
     },
     {
       "epoch": 40.85,
       "learning_rate": 0.0003257,
-      "loss": 0.2401,
+      "loss": 0.2044,
       "step": 2247
     },
     {
       "epoch": 40.87,
       "learning_rate": 0.0003256,
-      "loss": 0.2763,
+      "loss": 0.2241,
       "step": 2248
     },
     {
       "epoch": 40.89,
       "learning_rate": 0.0003255,
-      "loss": 0.2594,
+      "loss": 0.1984,
       "step": 2249
     },
     {
       "epoch": 40.91,
       "learning_rate": 0.00032540000000000005,
-      "loss": 0.2377,
+      "loss": 0.2016,
       "step": 2250
     },
     {
       "epoch": 40.93,
       "learning_rate": 0.0003253,
-      "loss": 0.2403,
+      "loss": 0.2308,
       "step": 2251
     },
     {
       "epoch": 40.95,
       "learning_rate": 0.0003252,
-      "loss": 0.2792,
+      "loss": 0.2615,
       "step": 2252
     },
     {
       "epoch": 40.96,
       "learning_rate": 0.0003251,
-      "loss": 0.2783,
+      "loss": 0.2488,
       "step": 2253
     },
     {
       "epoch": 40.98,
       "learning_rate": 0.00032500000000000004,
-      "loss": 0.2389,
+      "loss": 0.2211,
       "step": 2254
     },
     {
       "epoch": 41.0,
-      "learning_rate": 0.00032500000000000004,
-      "loss": 0.2023,
+      "learning_rate": 0.00032490000000000004,
+      "loss": 0.1707,
       "step": 2255
     },
     {
       "epoch": 41.02,
-      "learning_rate": 0.00032490000000000004,
-      "loss": 0.2624,
+      "learning_rate": 0.0003248,
+      "loss": 0.2578,
       "step": 2256
     },
     {
       "epoch": 41.04,
-      "learning_rate": 0.0003248,
-      "loss": 0.2233,
+      "learning_rate": 0.0003247,
+      "loss": 0.1971,
       "step": 2257
     },
     {
       "epoch": 41.05,
-      "learning_rate": 0.0003247,
-      "loss": 0.2547,
+      "learning_rate": 0.00032460000000000003,
+      "loss": 0.2329,
       "step": 2258
     },
     {
       "epoch": 41.07,
-      "learning_rate": 0.00032460000000000003,
-      "loss": 0.2317,
+      "learning_rate": 0.00032450000000000003,
+      "loss": 0.2061,
       "step": 2259
     },
     {
       "epoch": 41.09,
-      "learning_rate": 0.00032450000000000003,
-      "loss": 0.2098,
+      "learning_rate": 0.0003244,
+      "loss": 0.1923,
       "step": 2260
     },
     {
       "epoch": 41.11,
-      "learning_rate": 0.0003244,
-      "loss": 0.1982,
+      "learning_rate": 0.00032429999999999997,
+      "loss": 0.1471,
       "step": 2261
     },
     {
       "epoch": 41.13,
-      "learning_rate": 0.00032429999999999997,
-      "loss": 0.1928,
+      "learning_rate": 0.0003242,
+      "loss": 0.2322,
       "step": 2262
     },
     {
       "epoch": 41.15,
-      "learning_rate": 0.0003242,
-      "loss": 0.296,
+      "learning_rate": 0.0003241,
+      "loss": 0.2457,
       "step": 2263
     },
     {
       "epoch": 41.16,
-      "learning_rate": 0.0003241,
-      "loss": 0.1598,
+      "learning_rate": 0.000324,
+      "loss": 0.1711,
       "step": 2264
     },
     {
       "epoch": 41.18,
-      "learning_rate": 0.000324,
-      "loss": 0.2257,
+      "learning_rate": 0.0003239,
+      "loss": 0.2086,
       "step": 2265
     },
     {
       "epoch": 41.2,
-      "learning_rate": 0.0003239,
-      "loss": 0.2265,
+      "learning_rate": 0.0003238,
+      "loss": 0.1998,
       "step": 2266
     },
     {
       "epoch": 41.22,
-      "learning_rate": 0.0003238,
-      "loss": 0.2136,
+      "learning_rate": 0.0003237,
+      "loss": 0.2053,
       "step": 2267
     },
     {
       "epoch": 41.24,
-      "learning_rate": 0.0003237,
-      "loss": 0.235,
+      "learning_rate": 0.0003236,
+      "loss": 0.2044,
       "step": 2268
     },
     {
       "epoch": 41.25,
-      "learning_rate": 0.0003236,
-      "loss": 0.2277,
+      "learning_rate": 0.0003235,
+      "loss": 0.2147,
       "step": 2269
     },
     {
       "epoch": 41.27,
-      "learning_rate": 0.0003235,
-      "loss": 0.2161,
+      "learning_rate": 0.00032340000000000005,
+      "loss": 0.1871,
       "step": 2270
     },
     {
       "epoch": 41.29,
-      "learning_rate": 0.00032340000000000005,
-      "loss": 0.2092,
+      "learning_rate": 0.0003233,
+      "loss": 0.1777,
       "step": 2271
     },
     {
       "epoch": 41.31,
-      "learning_rate": 0.0003233,
-      "loss": 0.2084,
+      "learning_rate": 0.0003232,
+      "loss": 0.214,
       "step": 2272
     },
     {
       "epoch": 41.33,
-      "learning_rate": 0.0003232,
-      "loss": 0.1633,
+      "learning_rate": 0.0003231,
+      "loss": 0.1622,
       "step": 2273
     },
     {
       "epoch": 41.35,
-      "learning_rate": 0.0003231,
-      "loss": 0.2018,
+      "learning_rate": 0.000323,
+      "loss": 0.1679,
       "step": 2274
     },
     {
       "epoch": 41.36,
-      "learning_rate": 0.000323,
-      "loss": 0.2302,
+      "learning_rate": 0.00032290000000000004,
+      "loss": 0.1972,
       "step": 2275
     },
     {
       "epoch": 41.38,
-      "learning_rate": 0.00032290000000000004,
-      "loss": 0.2701,
+      "learning_rate": 0.0003228,
+      "loss": 0.2402,
       "step": 2276
     },
     {
       "epoch": 41.4,
-      "learning_rate": 0.0003228,
-      "loss": 0.2104,
+      "learning_rate": 0.0003227,
+      "loss": 0.1906,
       "step": 2277
     },
     {
       "epoch": 41.42,
-      "learning_rate": 0.0003227,
-      "loss": 0.1866,
+      "learning_rate": 0.0003226,
+      "loss": 0.176,
       "step": 2278
     },
     {
       "epoch": 41.44,
-      "learning_rate": 0.0003226,
-      "loss": 0.225,
+      "learning_rate": 0.00032250000000000003,
+      "loss": 0.2029,
       "step": 2279
     },
     {
       "epoch": 41.45,
-      "learning_rate": 0.00032250000000000003,
-      "loss": 0.2115,
+      "learning_rate": 0.00032240000000000003,
+      "loss": 0.1885,
       "step": 2280
     },
     {
       "epoch": 41.47,
-      "learning_rate": 0.00032240000000000003,
-      "loss": 0.2382,
+      "learning_rate": 0.0003223,
+      "loss": 0.2455,
       "step": 2281
     },
     {
       "epoch": 41.49,
-      "learning_rate": 0.0003223,
-      "loss": 0.222,
+      "learning_rate": 0.00032219999999999997,
+      "loss": 0.2196,
       "step": 2282
     },
     {
       "epoch": 41.51,
-      "learning_rate": 0.00032219999999999997,
-      "loss": 0.2068,
+      "learning_rate": 0.0003221,
+      "loss": 0.198,
       "step": 2283
     },
     {
       "epoch": 41.53,
-      "learning_rate": 0.0003221,
-      "loss": 0.2492,
+      "learning_rate": 0.000322,
+      "loss": 0.23,
       "step": 2284
     },
     {
       "epoch": 41.55,
-      "learning_rate": 0.000322,
-      "loss": 0.2133,
+      "learning_rate": 0.0003219,
+      "loss": 0.1826,
       "step": 2285
     },
     {
       "epoch": 41.56,
-      "learning_rate": 0.0003219,
-      "loss": 0.227,
+      "learning_rate": 0.00032179999999999996,
+      "loss": 0.186,
       "step": 2286
     },
     {
       "epoch": 41.58,
-      "learning_rate": 0.00032179999999999996,
-      "loss": 0.2085,
+      "learning_rate": 0.0003217,
+      "loss": 0.1923,
       "step": 2287
     },
     {
       "epoch": 41.6,
-      "learning_rate": 0.0003217,
-      "loss": 0.2293,
+      "learning_rate": 0.0003216,
+      "loss": 0.1991,
       "step": 2288
     },
     {
       "epoch": 41.62,
-      "learning_rate": 0.0003216,
-      "loss": 0.2781,
+      "learning_rate": 0.0003215,
+      "loss": 0.2518,
       "step": 2289
     },
     {
       "epoch": 41.64,
-      "learning_rate": 0.0003215,
-      "loss": 0.294,
+      "learning_rate": 0.0003214,
+      "loss": 0.2412,
       "step": 2290
     },
     {
       "epoch": 41.65,
-      "learning_rate": 0.0003214,
-      "loss": 0.185,
+      "learning_rate": 0.0003213,
+      "loss": 0.1971,
       "step": 2291
     },
     {
       "epoch": 41.67,
-      "learning_rate": 0.0003213,
-      "loss": 0.2503,
+      "learning_rate": 0.0003212,
+      "loss": 0.1941,
       "step": 2292
     },
     {
       "epoch": 41.69,
-      "learning_rate": 0.0003212,
-      "loss": 0.2443,
+      "learning_rate": 0.0003211,
+      "loss": 0.1962,
       "step": 2293
     },
     {
       "epoch": 41.71,
-      "learning_rate": 0.0003211,
-      "loss": 0.2296,
+      "learning_rate": 0.000321,
+      "loss": 0.1894,
       "step": 2294
     },
     {
       "epoch": 41.73,
-      "learning_rate": 0.000321,
-      "loss": 0.2643,
+      "learning_rate": 0.00032090000000000005,
+      "loss": 0.2635,
       "step": 2295
     },
     {
       "epoch": 41.75,
-      "learning_rate": 0.00032090000000000005,
-      "loss": 0.2346,
+      "learning_rate": 0.0003208,
+      "loss": 0.2094,
       "step": 2296
     },
     {
       "epoch": 41.76,
-      "learning_rate": 0.0003208,
-      "loss": 0.2134,
+      "learning_rate": 0.0003207,
+      "loss": 0.2053,
       "step": 2297
     },
     {
       "epoch": 41.78,
-      "learning_rate": 0.0003207,
-      "loss": 0.2444,
+      "learning_rate": 0.0003206,
+      "loss": 0.1954,
       "step": 2298
     },
     {
       "epoch": 41.8,
-      "learning_rate": 0.0003206,
-      "loss": 0.2171,
+      "learning_rate": 0.00032050000000000004,
+      "loss": 0.2235,
       "step": 2299
     },
     {
       "epoch": 41.82,
-      "learning_rate": 0.00032050000000000004,
-      "loss": 0.1917,
+      "learning_rate": 0.00032040000000000004,
+      "loss": 0.1825,
       "step": 2300
     },
     {
       "epoch": 41.84,
-      "learning_rate": 0.00032040000000000004,
-      "loss": 0.2103,
+      "learning_rate": 0.0003203,
+      "loss": 0.1921,
       "step": 2301
     },
     {
       "epoch": 41.85,
-      "learning_rate": 0.0003203,
-      "loss": 0.1823,
+      "learning_rate": 0.0003202,
+      "loss": 0.1835,
       "step": 2302
     },
     {
       "epoch": 41.87,
-      "learning_rate": 0.0003202,
-      "loss": 0.2377,
+      "learning_rate": 0.00032010000000000003,
+      "loss": 0.2418,
       "step": 2303
     },
     {
       "epoch": 41.89,
-      "learning_rate": 0.00032010000000000003,
-      "loss": 0.2374,
+      "learning_rate": 0.00032,
+      "loss": 0.2246,
       "step": 2304
     },
     {
       "epoch": 41.91,
-      "learning_rate": 0.00032,
-      "loss": 0.1886,
+      "learning_rate": 0.0003199,
+      "loss": 0.1917,
       "step": 2305
     },
     {
       "epoch": 41.93,
-      "learning_rate": 0.0003199,
-      "loss": 0.2479,
+      "learning_rate": 0.00031979999999999997,
+      "loss": 0.2196,
       "step": 2306
     },
     {
       "epoch": 41.95,
-      "learning_rate": 0.00031979999999999997,
-      "loss": 0.2348,
+      "learning_rate": 0.00031969999999999996,
+      "loss": 0.1773,
       "step": 2307
     },
     {
       "epoch": 41.96,
-      "learning_rate": 0.00031969999999999996,
-      "loss": 0.2788,
+      "learning_rate": 0.0003196,
+      "loss": 0.2533,
       "step": 2308
     },
     {
       "epoch": 41.98,
-      "learning_rate": 0.0003196,
-      "loss": 0.2108,
+      "learning_rate": 0.0003195,
+      "loss": 0.1681,
       "step": 2309
     },
     {
       "epoch": 42.0,
-      "learning_rate": 0.0003195,
-      "loss": 0.0997,
+      "learning_rate": 0.0003194,
+      "loss": 0.1127,
       "step": 2310
     },
     {
       "epoch": 42.02,
-      "learning_rate": 0.0003194,
-      "loss": 0.243,
+      "learning_rate": 0.00031929999999999995,
+      "loss": 0.1968,
       "step": 2311
     },
     {
       "epoch": 42.04,
-      "learning_rate": 0.00031929999999999995,
-      "loss": 0.1987,
+      "learning_rate": 0.0003192,
+      "loss": 0.1493,
       "step": 2312
     },
     {
       "epoch": 42.05,
-      "learning_rate": 0.0003192,
-      "loss": 0.2213,
+      "learning_rate": 0.0003191,
+      "loss": 0.1855,
       "step": 2313
     },
     {
       "epoch": 42.07,
-      "learning_rate": 0.0003191,
-      "loss": 0.2069,
+      "learning_rate": 0.000319,
+      "loss": 0.1592,
       "step": 2314
     },
     {
       "epoch": 42.09,
-      "learning_rate": 0.000319,
-      "loss": 0.2296,
+      "learning_rate": 0.0003189,
+      "loss": 0.2071,
       "step": 2315
     },
     {
       "epoch": 42.11,
-      "learning_rate": 0.0003189,
-      "loss": 0.2325,
+      "learning_rate": 0.0003188,
+      "loss": 0.2147,
       "step": 2316
     },
     {
       "epoch": 42.13,
-      "learning_rate": 0.0003188,
-      "loss": 0.2617,
+      "learning_rate": 0.0003187,
+      "loss": 0.23,
       "step": 2317
     },
     {
       "epoch": 42.15,
-      "learning_rate": 0.0003187,
-      "loss": 0.2222,
+      "learning_rate": 0.0003186,
+      "loss": 0.2005,
       "step": 2318
     },
     {
       "epoch": 42.16,
-      "learning_rate": 0.0003186,
-      "loss": 0.1927,
+      "learning_rate": 0.0003185,
+      "loss": 0.1629,
       "step": 2319
     },
     {
       "epoch": 42.18,
-      "learning_rate": 0.0003185,
-      "loss": 0.2085,
+      "learning_rate": 0.00031840000000000004,
+      "loss": 0.1831,
       "step": 2320
     },
     {
       "epoch": 42.2,
-      "learning_rate": 0.00031840000000000004,
-      "loss": 0.2194,
+      "learning_rate": 0.00031830000000000004,
+      "loss": 0.1858,
       "step": 2321
     },
     {
       "epoch": 42.22,
-      "learning_rate": 0.00031830000000000004,
-      "loss": 0.2131,
+      "learning_rate": 0.0003182,
+      "loss": 0.192,
       "step": 2322
     },
     {
       "epoch": 42.24,
-      "learning_rate": 0.0003182,
-      "loss": 0.2323,
+      "learning_rate": 0.0003181,
+      "loss": 0.191,
       "step": 2323
     },
     {
       "epoch": 42.25,
-      "learning_rate": 0.0003181,
-      "loss": 0.2957,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 0.2563,
       "step": 2324
     },
     {
       "epoch": 42.27,
-      "learning_rate": 0.00031800000000000003,
-      "loss": 0.2397,
+      "learning_rate": 0.00031790000000000003,
+      "loss": 0.1946,
       "step": 2325
     },
     {
       "epoch": 42.29,
-      "learning_rate": 0.00031790000000000003,
-      "loss": 0.2756,
+      "learning_rate": 0.0003178,
+      "loss": 0.271,
       "step": 2326
     },
     {
       "epoch": 42.31,
-      "learning_rate": 0.0003178,
-      "loss": 0.2159,
+      "learning_rate": 0.00031769999999999997,
+      "loss": 0.2053,
       "step": 2327
     },
     {
       "epoch": 42.33,
-      "learning_rate": 0.00031769999999999997,
-      "loss": 0.2665,
+      "learning_rate": 0.0003176,
+      "loss": 0.2154,
       "step": 2328
     },
     {
       "epoch": 42.35,
-      "learning_rate": 0.0003176,
-      "loss": 0.2285,
+      "learning_rate": 0.0003175,
+      "loss": 0.19,
       "step": 2329
     },
     {
       "epoch": 42.36,
-      "learning_rate": 0.0003175,
-      "loss": 0.2548,
+      "learning_rate": 0.0003174,
+      "loss": 0.2519,
       "step": 2330
     },
     {
       "epoch": 42.38,
-      "learning_rate": 0.0003174,
-      "loss": 0.2,
+      "learning_rate": 0.0003173,
+      "loss": 0.1933,
       "step": 2331
     },
     {
       "epoch": 42.4,
-      "learning_rate": 0.0003173,
-      "loss": 0.2447,
+      "learning_rate": 0.0003172,
+      "loss": 0.2157,
       "step": 2332
     },
     {
       "epoch": 42.42,
-      "learning_rate": 0.0003172,
-      "loss": 0.2449,
+      "learning_rate": 0.0003171,
+      "loss": 0.2363,
       "step": 2333
     },
     {
       "epoch": 42.44,
-      "learning_rate": 0.0003171,
-      "loss": 0.2349,
+      "learning_rate": 0.000317,
+      "loss": 0.1977,
       "step": 2334
     },
     {
       "epoch": 42.45,
-      "learning_rate": 0.000317,
-      "loss": 0.1958,
+      "learning_rate": 0.0003169,
+      "loss": 0.2198,
       "step": 2335
     },
     {
       "epoch": 42.47,
-      "learning_rate": 0.0003169,
-      "loss": 0.2517,
+      "learning_rate": 0.00031680000000000006,
+      "loss": 0.2458,
       "step": 2336
     },
     {
       "epoch": 42.49,
-      "learning_rate": 0.00031680000000000006,
-      "loss": 0.2018,
+      "learning_rate": 0.0003167,
+      "loss": 0.1826,
       "step": 2337
     },
     {
       "epoch": 42.51,
-      "learning_rate": 0.0003167,
-      "loss": 0.2257,
+      "learning_rate": 0.0003166,
+      "loss": 0.184,
       "step": 2338
     },
     {
       "epoch": 42.53,
-      "learning_rate": 0.0003166,
-      "loss": 0.2308,
+      "learning_rate": 0.0003165,
+      "loss": 0.2176,
       "step": 2339
     },
     {
       "epoch": 42.55,
-      "learning_rate": 0.0003165,
-      "loss": 0.1883,
+      "learning_rate": 0.00031640000000000005,
+      "loss": 0.1818,
       "step": 2340
     },
     {
       "epoch": 42.56,
-      "learning_rate": 0.00031640000000000005,
-      "loss": 0.2122,
+      "learning_rate": 0.00031630000000000004,
+      "loss": 0.2207,
       "step": 2341
     },
     {
       "epoch": 42.58,
-      "learning_rate": 0.00031630000000000004,
-      "loss": 0.2318,
+      "learning_rate": 0.0003162,
+      "loss": 0.2074,
       "step": 2342
     },
     {
       "epoch": 42.6,
-      "learning_rate": 0.0003162,
-      "loss": 0.2631,
+      "learning_rate": 0.0003161,
+      "loss": 0.2384,
       "step": 2343
     },
     {
       "epoch": 42.62,
-      "learning_rate": 0.0003161,
-      "loss": 0.2042,
+      "learning_rate": 0.000316,
+      "loss": 0.2052,
       "step": 2344
     },
     {
       "epoch": 42.64,
-      "learning_rate": 0.000316,
-      "loss": 0.2016,
+      "learning_rate": 0.00031590000000000003,
+      "loss": 0.1844,
       "step": 2345
     },
     {
       "epoch": 42.65,
-      "learning_rate": 0.00031590000000000003,
-      "loss": 0.1915,
+      "learning_rate": 0.00031580000000000003,
+      "loss": 0.1865,
       "step": 2346
     },
     {
       "epoch": 42.67,
-      "learning_rate": 0.00031580000000000003,
-      "loss": 0.2651,
+      "learning_rate": 0.0003157,
+      "loss": 0.2196,
       "step": 2347
     },
     {
       "epoch": 42.69,
-      "learning_rate": 0.0003157,
-      "loss": 0.215,
+      "learning_rate": 0.0003156,
+      "loss": 0.169,
       "step": 2348
     },
     {
       "epoch": 42.71,
-      "learning_rate": 0.0003156,
-      "loss": 0.2061,
+      "learning_rate": 0.0003155,
+      "loss": 0.2213,
       "step": 2349
     },
     {
       "epoch": 42.73,
-      "learning_rate": 0.0003155,
-      "loss": 0.1628,
+      "learning_rate": 0.0003154,
+      "loss": 0.1413,
       "step": 2350
     },
     {
       "epoch": 42.75,
-      "learning_rate": 0.0003154,
-      "loss": 0.3041,
+      "learning_rate": 0.0003153,
+      "loss": 0.2735,
       "step": 2351
     },
     {
       "epoch": 42.76,
-      "learning_rate": 0.0003153,
-      "loss": 0.2215,
+      "learning_rate": 0.00031519999999999996,
+      "loss": 0.2051,
       "step": 2352
     },
     {
       "epoch": 42.78,
-      "learning_rate": 0.00031519999999999996,
-      "loss": 0.1916,
+      "learning_rate": 0.0003151,
+      "loss": 0.2185,
       "step": 2353
     },
     {
       "epoch": 42.8,
-      "learning_rate": 0.0003151,
-      "loss": 0.1683,
+      "learning_rate": 0.000315,
+      "loss": 0.1729,
       "step": 2354
     },
     {
       "epoch": 42.82,
-      "learning_rate": 0.000315,
-      "loss": 0.2313,
+      "learning_rate": 0.0003149,
+      "loss": 0.208,
       "step": 2355
     },
     {
       "epoch": 42.84,
-      "learning_rate": 0.0003149,
-      "loss": 0.2159,
+      "learning_rate": 0.0003148,
+      "loss": 0.2023,
       "step": 2356
     },
     {
       "epoch": 42.85,
-      "learning_rate": 0.0003148,
-      "loss": 0.2738,
+      "learning_rate": 0.0003147,
+      "loss": 0.2613,
       "step": 2357
     },
     {
       "epoch": 42.87,
-      "learning_rate": 0.0003147,
-      "loss": 0.2082,
+      "learning_rate": 0.0003146,
+      "loss": 0.1893,
       "step": 2358
     },
     {
       "epoch": 42.89,
-      "learning_rate": 0.0003146,
-      "loss": 0.213,
+      "learning_rate": 0.0003145,
+      "loss": 0.1943,
       "step": 2359
     },
     {
       "epoch": 42.91,
-      "learning_rate": 0.0003145,
-      "loss": 0.1518,
+      "learning_rate": 0.0003144,
+      "loss": 0.1263,
       "step": 2360
     },
     {
       "epoch": 42.93,
-      "learning_rate": 0.0003144,
-      "loss": 0.19,
+      "learning_rate": 0.00031430000000000005,
+      "loss": 0.1665,
       "step": 2361
     },
     {
       "epoch": 42.95,
-      "learning_rate": 0.00031430000000000005,
-      "loss": 0.188,
+      "learning_rate": 0.0003142,
+      "loss": 0.1682,
       "step": 2362
     },
     {
       "epoch": 42.96,
-      "learning_rate": 0.0003142,
-      "loss": 0.2329,
+      "learning_rate": 0.0003141,
+      "loss": 0.1722,
       "step": 2363
     },
     {
       "epoch": 42.98,
-      "learning_rate": 0.0003141,
-      "loss": 0.2297,
+      "learning_rate": 0.000314,
+      "loss": 0.2005,
       "step": 2364
     },
     {
       "epoch": 43.0,
-      "learning_rate": 0.000314,
-      "loss": 0.2218,
+      "learning_rate": 0.00031390000000000004,
+      "loss": 0.181,
       "step": 2365
     },
     {
       "epoch": 43.02,
-      "learning_rate": 0.00031390000000000004,
-      "loss": 0.2274,
+      "learning_rate": 0.00031380000000000004,
+      "loss": 0.2201,
       "step": 2366
     },
     {
       "epoch": 43.04,
-      "learning_rate": 0.00031380000000000004,
-      "loss": 0.2498,
+      "learning_rate": 0.0003137,
+      "loss": 0.2216,
       "step": 2367
     },
     {
       "epoch": 43.05,
-      "learning_rate": 0.0003137,
-      "loss": 0.2128,
+      "learning_rate": 0.0003136,
+      "loss": 0.1799,
       "step": 2368
     },
     {
       "epoch": 43.07,
-      "learning_rate": 0.0003136,
-      "loss": 0.1942,
+      "learning_rate": 0.00031350000000000003,
+      "loss": 0.1831,
       "step": 2369
     },
     {
       "epoch": 43.09,
-      "learning_rate": 0.00031350000000000003,
-      "loss": 0.2041,
+      "learning_rate": 0.00031340000000000003,
+      "loss": 0.1987,
       "step": 2370
     },
     {
       "epoch": 43.11,
-      "learning_rate": 0.00031340000000000003,
-      "loss": 0.1991,
+      "learning_rate": 0.0003133,
+      "loss": 0.1611,
       "step": 2371
     },
     {
       "epoch": 43.13,
-      "learning_rate": 0.0003133,
-      "loss": 0.231,
+      "learning_rate": 0.00031319999999999997,
+      "loss": 0.2186,
       "step": 2372
     },
     {
       "epoch": 43.15,
-      "learning_rate": 0.00031319999999999997,
-      "loss": 0.2161,
+      "learning_rate": 0.0003131,
+      "loss": 0.2059,
       "step": 2373
     },
     {
       "epoch": 43.16,
-      "learning_rate": 0.0003131,
-      "loss": 0.1806,
+      "learning_rate": 0.000313,
+      "loss": 0.149,
       "step": 2374
     },
     {
       "epoch": 43.18,
-      "learning_rate": 0.000313,
-      "loss": 0.2171,
+      "learning_rate": 0.0003129,
+      "loss": 0.2405,
       "step": 2375
     },
     {
       "epoch": 43.2,
-      "learning_rate": 0.0003129,
-      "loss": 0.193,
+      "learning_rate": 0.0003128,
+      "loss": 0.1794,
       "step": 2376
     },
     {
       "epoch": 43.22,
-      "learning_rate": 0.0003128,
-      "loss": 0.2224,
+      "learning_rate": 0.0003127,
+      "loss": 0.233,
       "step": 2377
     },
     {
       "epoch": 43.24,
-      "learning_rate": 0.0003127,
-      "loss": 0.1806,
+      "learning_rate": 0.0003126,
+      "loss": 0.1577,
       "step": 2378
     },
     {
       "epoch": 43.25,
-      "learning_rate": 0.0003126,
-      "loss": 0.1854,
+      "learning_rate": 0.0003125,
+      "loss": 0.185,
       "step": 2379
     },
     {
       "epoch": 43.27,
-      "learning_rate": 0.0003125,
-      "loss": 0.2483,
+      "learning_rate": 0.0003124,
+      "loss": 0.2373,
       "step": 2380
     },
     {
       "epoch": 43.29,
-      "learning_rate": 0.0003124,
-      "loss": 0.2114,
+      "learning_rate": 0.0003123,
+      "loss": 0.1742,
       "step": 2381
     },
     {
       "epoch": 43.31,
-      "learning_rate": 0.0003123,
-      "loss": 0.1942,
+      "learning_rate": 0.0003122,
+      "loss": 0.2067,
       "step": 2382
     },
     {
       "epoch": 43.33,
-      "learning_rate": 0.0003122,
-      "loss": 0.2269,
+      "learning_rate": 0.0003121,
+      "loss": 0.2031,
       "step": 2383
     },
     {
       "epoch": 43.35,
-      "learning_rate": 0.0003121,
-      "loss": 0.2151,
+      "learning_rate": 0.000312,
+      "loss": 0.1812,
       "step": 2384
     },
     {
       "epoch": 43.36,
-      "learning_rate": 0.000312,
-      "loss": 0.2487,
+      "learning_rate": 0.0003119,
+      "loss": 0.2716,
       "step": 2385
     },
     {
       "epoch": 43.38,
-      "learning_rate": 0.0003119,
-      "loss": 0.221,
+      "learning_rate": 0.00031180000000000004,
+      "loss": 0.2371,
       "step": 2386
     },
     {
       "epoch": 43.4,
-      "learning_rate": 0.00031180000000000004,
-      "loss": 0.2286,
+      "learning_rate": 0.0003117,
+      "loss": 0.1831,
       "step": 2387
     },
     {
       "epoch": 43.42,
-      "learning_rate": 0.0003117,
-      "loss": 0.1992,
+      "learning_rate": 0.0003116,
+      "loss": 0.195,
       "step": 2388
     },
     {
       "epoch": 43.44,
-      "learning_rate": 0.0003116,
-      "loss": 0.2253,
+      "learning_rate": 0.0003115,
+      "loss": 0.2332,
       "step": 2389
     },
     {
       "epoch": 43.45,
-      "learning_rate": 0.0003115,
-      "loss": 0.1722,
+      "learning_rate": 0.00031140000000000003,
+      "loss": 0.1903,
       "step": 2390
     },
     {
       "epoch": 43.47,
-      "learning_rate": 0.00031140000000000003,
-      "loss": 0.1996,
+      "learning_rate": 0.00031130000000000003,
+      "loss": 0.1811,
       "step": 2391
     },
     {
       "epoch": 43.49,
-      "learning_rate": 0.00031130000000000003,
-      "loss": 0.225,
+      "learning_rate": 0.0003112,
+      "loss": 0.1959,
       "step": 2392
     },
     {
       "epoch": 43.51,
-      "learning_rate": 0.0003112,
-      "loss": 0.1984,
+      "learning_rate": 0.00031109999999999997,
+      "loss": 0.2043,
       "step": 2393
     },
     {
       "epoch": 43.53,
-      "learning_rate": 0.00031109999999999997,
-      "loss": 0.2314,
+      "learning_rate": 0.000311,
+      "loss": 0.2232,
       "step": 2394
     },
     {
       "epoch": 43.55,
-      "learning_rate": 0.000311,
-      "loss": 0.2532,
+      "learning_rate": 0.0003109,
+      "loss": 0.2042,
       "step": 2395
     },
     {
       "epoch": 43.56,
-      "learning_rate": 0.0003109,
-      "loss": 0.2403,
+      "learning_rate": 0.0003108,
+      "loss": 0.2214,
       "step": 2396
     },
     {
       "epoch": 43.58,
-      "learning_rate": 0.0003108,
-      "loss": 0.207,
+      "learning_rate": 0.00031069999999999996,
+      "loss": 0.2046,
       "step": 2397
     },
     {
       "epoch": 43.6,
-      "learning_rate": 0.00031069999999999996,
-      "loss": 0.2082,
+      "learning_rate": 0.0003106,
+      "loss": 0.2,
       "step": 2398
     },
     {
       "epoch": 43.62,
-      "learning_rate": 0.0003106,
-      "loss": 0.224,
+      "learning_rate": 0.0003105,
+      "loss": 0.2196,
       "step": 2399
     },
     {
       "epoch": 43.64,
-      "learning_rate": 0.0003105,
-      "loss": 0.1903,
+      "learning_rate": 0.0003104,
+      "loss": 0.1884,
       "step": 2400
     },
     {
       "epoch": 43.65,
-      "learning_rate": 0.0003104,
-      "loss": 0.2913,
+      "learning_rate": 0.0003103,
+      "loss": 0.2479,
       "step": 2401
     },
     {
       "epoch": 43.67,
-      "learning_rate": 0.0003103,
-      "loss": 0.1403,
+      "learning_rate": 0.0003102,
+      "loss": 0.1163,
       "step": 2402
     },
     {
       "epoch": 43.69,
-      "learning_rate": 0.0003102,
-      "loss": 0.2446,
+      "learning_rate": 0.0003101,
+      "loss": 0.2312,
       "step": 2403
     },
     {
       "epoch": 43.71,
-      "learning_rate": 0.0003101,
-      "loss": 0.2219,
+      "learning_rate": 0.00031,
+      "loss": 0.2213,
       "step": 2404
     },
     {
       "epoch": 43.73,
-      "learning_rate": 0.00031,
-      "loss": 0.2478,
+      "learning_rate": 0.0003099,
+      "loss": 0.2258,
       "step": 2405
     },
     {
       "epoch": 43.75,
-      "learning_rate": 0.0003099,
-      "loss": 0.2123,
+      "learning_rate": 0.00030980000000000005,
+      "loss": 0.1813,
       "step": 2406
     },
     {
       "epoch": 43.76,
-      "learning_rate": 0.00030980000000000005,
-      "loss": 0.2107,
+      "learning_rate": 0.0003097,
+      "loss": 0.1853,
       "step": 2407
     },
     {
       "epoch": 43.78,
-      "learning_rate": 0.0003097,
-      "loss": 0.1652,
+      "learning_rate": 0.0003096,
+      "loss": 0.1853,
       "step": 2408
     },
     {
       "epoch": 43.8,
-      "learning_rate": 0.0003096,
-      "loss": 0.1747,
+      "learning_rate": 0.0003095,
+      "loss": 0.1631,
       "step": 2409
     },
     {
       "epoch": 43.82,
-      "learning_rate": 0.0003095,
-      "loss": 0.2218,
+      "learning_rate": 0.00030940000000000004,
+      "loss": 0.2179,
       "step": 2410
     },
     {
       "epoch": 43.84,
-      "learning_rate": 0.00030940000000000004,
-      "loss": 0.2099,
+      "learning_rate": 0.00030930000000000004,
+      "loss": 0.2397,
       "step": 2411
     },
     {
       "epoch": 43.85,
-      "learning_rate": 0.00030930000000000004,
-      "loss": 0.1526,
+      "learning_rate": 0.0003092,
+      "loss": 0.1958,
       "step": 2412
     },
     {
       "epoch": 43.87,
-      "learning_rate": 0.0003092,
-      "loss": 0.2497,
+      "learning_rate": 0.0003091,
+      "loss": 0.231,
       "step": 2413
     },
     {
       "epoch": 43.89,
-      "learning_rate": 0.0003091,
-      "loss": 0.2185,
+      "learning_rate": 0.00030900000000000003,
+      "loss": 0.1684,
       "step": 2414
     },
     {
       "epoch": 43.91,
-      "learning_rate": 0.00030900000000000003,
-      "loss": 0.1299,
+      "learning_rate": 0.00030890000000000003,
+      "loss": 0.1642,
       "step": 2415
     },
     {
       "epoch": 43.93,
-      "learning_rate": 0.00030890000000000003,
-      "loss": 0.2216,
+      "learning_rate": 0.0003088,
+      "loss": 0.2277,
       "step": 2416
     },
     {
       "epoch": 43.95,
-      "learning_rate": 0.0003088,
-      "loss": 0.2219,
+      "learning_rate": 0.00030869999999999997,
+      "loss": 0.2116,
       "step": 2417
     },
     {
       "epoch": 43.96,
-      "learning_rate": 0.00030869999999999997,
-      "loss": 0.1841,
+      "learning_rate": 0.00030859999999999997,
+      "loss": 0.1783,
       "step": 2418
     },
     {
       "epoch": 43.98,
-      "learning_rate": 0.00030859999999999997,
-      "loss": 0.1886,
+      "learning_rate": 0.0003085,
+      "loss": 0.1769,
       "step": 2419
     },
     {
       "epoch": 44.0,
-      "learning_rate": 0.0003085,
-      "loss": 0.1995,
+      "learning_rate": 0.0003084,
+      "loss": 0.1749,
       "step": 2420
     },
     {
       "epoch": 44.02,
-      "learning_rate": 0.0003084,
-      "loss": 0.2279,
+      "learning_rate": 0.0003083,
+      "loss": 0.2603,
       "step": 2421
     },
     {
       "epoch": 44.04,
-      "learning_rate": 0.0003083,
-      "loss": 0.2087,
+      "learning_rate": 0.00030819999999999996,
+      "loss": 0.2069,
       "step": 2422
     },
     {
       "epoch": 44.05,
-      "learning_rate": 0.00030819999999999996,
-      "loss": 0.1871,
+      "learning_rate": 0.0003081,
+      "loss": 0.1787,
       "step": 2423
     },
     {
       "epoch": 44.07,
-      "learning_rate": 0.0003081,
-      "loss": 0.2098,
+      "learning_rate": 0.000308,
+      "loss": 0.1644,
       "step": 2424
     },
     {
       "epoch": 44.09,
-      "learning_rate": 0.000308,
-      "loss": 0.1933,
+      "learning_rate": 0.0003079,
+      "loss": 0.2092,
       "step": 2425
     },
     {
       "epoch": 44.11,
-      "learning_rate": 0.0003079,
-      "loss": 0.2163,
-      "step": 2426
+      "learning_rate": 0.0003078,
+      "loss": 0.221,
+      "step": 2426
     },
     {
       "epoch": 44.13,
-      "learning_rate": 0.0003078,
-      "loss": 0.2239,
+      "learning_rate": 0.0003077,
+      "loss": 0.2115,
       "step": 2427
     },
     {
       "epoch": 44.15,
-      "learning_rate": 0.0003077,
-      "loss": 0.2393,
+      "learning_rate": 0.0003076,
+      "loss": 0.1942,
       "step": 2428
     },
     {
       "epoch": 44.16,
-      "learning_rate": 0.0003076,
-      "loss": 0.176,
+      "learning_rate": 0.0003075,
+      "loss": 0.1671,
       "step": 2429
     },
     {
       "epoch": 44.18,
-      "learning_rate": 0.0003075,
-      "loss": 0.2258,
+      "learning_rate": 0.0003074,
+      "loss": 0.226,
       "step": 2430
     },
     {
       "epoch": 44.2,
-      "learning_rate": 0.0003074,
-      "loss": 0.2036,
+      "learning_rate": 0.00030730000000000004,
+      "loss": 0.2012,
       "step": 2431
     },
     {
       "epoch": 44.22,
-      "learning_rate": 0.00030730000000000004,
-      "loss": 0.1768,
+      "learning_rate": 0.0003072,
+      "loss": 0.1406,
       "step": 2432
     },
     {
       "epoch": 44.24,
-      "learning_rate": 0.0003072,
-      "loss": 0.2138,
+      "learning_rate": 0.0003071,
+      "loss": 0.1811,
       "step": 2433
     },
     {
       "epoch": 44.25,
-      "learning_rate": 0.0003071,
-      "loss": 0.2094,
+      "learning_rate": 0.000307,
+      "loss": 0.2106,
       "step": 2434
     },
     {
       "epoch": 44.27,
-      "learning_rate": 0.000307,
-      "loss": 0.2065,
+      "learning_rate": 0.00030690000000000003,
+      "loss": 0.2581,
       "step": 2435
     },
     {
       "epoch": 44.29,
-      "learning_rate": 0.00030690000000000003,
-      "loss": 0.2226,
+      "learning_rate": 0.00030680000000000003,
+      "loss": 0.1726,
       "step": 2436
     },
     {
       "epoch": 44.31,
-      "learning_rate": 0.00030680000000000003,
-      "loss": 0.1491,
+      "learning_rate": 0.0003067,
+      "loss": 0.1348,
       "step": 2437
     },
     {
       "epoch": 44.33,
-      "learning_rate": 0.0003067,
-      "loss": 0.1929,
+      "learning_rate": 0.00030659999999999997,
+      "loss": 0.1893,
       "step": 2438
     },
     {
       "epoch": 44.35,
-      "learning_rate": 0.00030659999999999997,
-      "loss": 0.1836,
+      "learning_rate": 0.0003065,
+      "loss": 0.1476,
       "step": 2439
     },
     {
       "epoch": 44.36,
-      "learning_rate": 0.0003065,
-      "loss": 0.2286,
+      "learning_rate": 0.0003064,
+      "loss": 0.201,
       "step": 2440
     },
     {
       "epoch": 44.38,
-      "learning_rate": 0.0003064,
-      "loss": 0.2282,
+      "learning_rate": 0.0003063,
+      "loss": 0.2111,
       "step": 2441
     },
     {
       "epoch": 44.4,
-      "learning_rate": 0.0003063,
-      "loss": 0.2068,
+      "learning_rate": 0.0003062,
+      "loss": 0.1874,
       "step": 2442
     },
     {
       "epoch": 44.42,
-      "learning_rate": 0.0003062,
-      "loss": 0.2256,
+      "learning_rate": 0.0003061,
+      "loss": 0.2054,
       "step": 2443
     },
     {
       "epoch": 44.44,
-      "learning_rate": 0.0003061,
-      "loss": 0.1988,
+      "learning_rate": 0.000306,
+      "loss": 0.1536,
       "step": 2444
     },
     {
       "epoch": 44.45,
-      "learning_rate": 0.000306,
-      "loss": 0.1694,
+      "learning_rate": 0.0003059,
+      "loss": 0.1353,
       "step": 2445
     },
     {
       "epoch": 44.47,
-      "learning_rate": 0.0003059,
-      "loss": 0.2135,
+      "learning_rate": 0.0003058,
+      "loss": 0.1744,
       "step": 2446
     },
     {
       "epoch": 44.49,
-      "learning_rate": 0.0003058,
-      "loss": 0.2412,
+      "learning_rate": 0.00030570000000000006,
+      "loss": 0.249,
       "step": 2447
     },
     {
       "epoch": 44.51,
-      "learning_rate": 0.00030570000000000006,
-      "loss": 0.1668,
+      "learning_rate": 0.0003056,
+      "loss": 0.1712,
       "step": 2448
     },
     {
       "epoch": 44.53,
-      "learning_rate": 0.0003056,
-      "loss": 0.2408,
+      "learning_rate": 0.0003055,
+      "loss": 0.1993,
       "step": 2449
     },
     {
       "epoch": 44.55,
-      "learning_rate": 0.0003055,
-      "loss": 0.1828,
+      "learning_rate": 0.0003054,
+      "loss": 0.1848,
       "step": 2450
     },
     {
       "epoch": 44.56,
-      "learning_rate": 0.0003054,
-      "loss": 0.2096,
+      "learning_rate": 0.00030530000000000005,
+      "loss": 0.1736,
       "step": 2451
     },
     {
       "epoch": 44.58,
-      "learning_rate": 0.00030530000000000005,
-      "loss": 0.2106,
+      "learning_rate": 0.00030520000000000005,
+      "loss": 0.2077,
       "step": 2452
     },
     {
       "epoch": 44.6,
-      "learning_rate": 0.00030520000000000005,
-      "loss": 0.2266,
+      "learning_rate": 0.0003051,
+      "loss": 0.2112,
       "step": 2453
     },
     {
       "epoch": 44.62,
-      "learning_rate": 0.0003051,
-      "loss": 0.1978,
+      "learning_rate": 0.000305,
+      "loss": 0.1939,
       "step": 2454
     },
     {
       "epoch": 44.64,
-      "learning_rate": 0.000305,
-      "loss": 0.1748,
+      "learning_rate": 0.0003049,
+      "loss": 0.1816,
       "step": 2455
     },
     {
       "epoch": 44.65,
-      "learning_rate": 0.0003049,
-      "loss": 0.1777,
+      "learning_rate": 0.00030480000000000004,
+      "loss": 0.1655,
       "step": 2456
     },
     {
       "epoch": 44.67,
-      "learning_rate": 0.00030480000000000004,
-      "loss": 0.1807,
+      "learning_rate": 0.00030470000000000003,
+      "loss": 0.1799,
       "step": 2457
     },
     {
       "epoch": 44.69,
-      "learning_rate": 0.00030470000000000003,
-      "loss": 0.206,
+      "learning_rate": 0.0003046,
+      "loss": 0.1736,
       "step": 2458
     },
     {
       "epoch": 44.71,
-      "learning_rate": 0.0003046,
-      "loss": 0.2605,
+      "learning_rate": 0.0003045,
+      "loss": 0.2177,
       "step": 2459
     },
     {
       "epoch": 44.73,
-      "learning_rate": 0.0003045,
-      "loss": 0.2108,
+      "learning_rate": 0.0003044,
+      "loss": 0.1697,
       "step": 2460
     },
     {
       "epoch": 44.75,
-      "learning_rate": 0.0003044,
-      "loss": 0.2223,
+      "learning_rate": 0.0003043,
+      "loss": 0.1782,
       "step": 2461
     },
     {
       "epoch": 44.76,
-      "learning_rate": 0.0003043,
-      "loss": 0.1599,
+      "learning_rate": 0.0003042,
+      "loss": 0.1199,
       "step": 2462
     },
     {
       "epoch": 44.78,
-      "learning_rate": 0.0003042,
-      "loss": 0.2068,
+      "learning_rate": 0.00030409999999999996,
+      "loss": 0.16,
       "step": 2463
     },
     {
       "epoch": 44.8,
-      "learning_rate": 0.00030409999999999996,
-      "loss": 0.1939,
+      "learning_rate": 0.000304,
+      "loss": 0.1607,
       "step": 2464
     },
     {
       "epoch": 44.82,
-      "learning_rate": 0.000304,
-      "loss": 0.2262,
+      "learning_rate": 0.0003039,
+      "loss": 0.204,
       "step": 2465
     },
     {
       "epoch": 44.84,
-      "learning_rate": 0.0003039,
-      "loss": 0.2332,
+      "learning_rate": 0.0003038,
+      "loss": 0.2011,
       "step": 2466
     },
     {
       "epoch": 44.85,
-      "learning_rate": 0.0003038,
-      "loss": 0.2204,
+      "learning_rate": 0.0003037,
+      "loss": 0.1659,
       "step": 2467
     },
     {
       "epoch": 44.87,
-      "learning_rate": 0.0003037,
-      "loss": 0.2148,
+      "learning_rate": 0.0003036,
+      "loss": 0.1612,
       "step": 2468
     },
     {
       "epoch": 44.89,
-      "learning_rate": 0.0003036,
-      "loss": 0.2166,
+      "learning_rate": 0.0003035,
+      "loss": 0.1827,
       "step": 2469
     },
     {
       "epoch": 44.91,
-      "learning_rate": 0.0003035,
-      "loss": 0.1673,
+      "learning_rate": 0.0003034,
+      "loss": 0.165,
       "step": 2470
     },
     {
       "epoch": 44.93,
-      "learning_rate": 0.0003034,
-      "loss": 0.2387,
+      "learning_rate": 0.0003033,
+      "loss": 0.2217,
       "step": 2471
     },
     {
       "epoch": 44.95,
-      "learning_rate": 0.0003033,
-      "loss": 0.2071,
+      "learning_rate": 0.00030320000000000005,
+      "loss": 0.1783,
       "step": 2472
     },
     {
       "epoch": 44.96,
-      "learning_rate": 0.00030320000000000005,
-      "loss": 0.2017,
+      "learning_rate": 0.0003031,
+      "loss": 0.1973,
       "step": 2473
     },
     {
       "epoch": 44.98,
-      "learning_rate": 0.0003031,
-      "loss": 0.1953,
+      "learning_rate": 0.000303,
+      "loss": 0.1583,
       "step": 2474
     },
     {
       "epoch": 45.0,
-      "learning_rate": 0.000303,
-      "loss": 0.1719,
+      "learning_rate": 0.0003029,
+      "loss": 0.1625,
       "step": 2475
     },
     {
       "epoch": 45.02,
-      "learning_rate": 0.0003029,
-      "loss": 0.2227,
+      "learning_rate": 0.00030280000000000004,
+      "loss": 0.2128,
       "step": 2476
     },
     {
       "epoch": 45.04,
-      "learning_rate": 0.00030280000000000004,
-      "loss": 0.1806,
+      "learning_rate": 0.00030270000000000004,
+      "loss": 0.1694,
       "step": 2477
     },
     {
       "epoch": 45.05,
-      "learning_rate": 0.00030270000000000004,
-      "loss": 0.2338,
+      "learning_rate": 0.0003026,
+      "loss": 0.2227,
       "step": 2478
     },
     {
       "epoch": 45.07,
-      "learning_rate": 0.0003026,
-      "loss": 0.2071,
+      "learning_rate": 0.0003025,
+      "loss": 0.1924,
       "step": 2479
     },
     {
       "epoch": 45.09,
-      "learning_rate": 0.0003025,
-      "loss": 0.2117,
+      "learning_rate": 0.00030240000000000003,
+      "loss": 0.1928,
       "step": 2480
     },
     {
       "epoch": 45.11,
-      "learning_rate": 0.00030240000000000003,
-      "loss": 0.2113,
+      "learning_rate": 0.00030230000000000003,
+      "loss": 0.1836,
       "step": 2481
     },
     {
       "epoch": 45.13,
-      "learning_rate": 0.00030230000000000003,
-      "loss": 0.1985,
+      "learning_rate": 0.00030220000000000003,
+      "loss": 0.2056,
       "step": 2482
     },
     {
       "epoch": 45.15,
-      "learning_rate": 0.00030220000000000003,
-      "loss": 0.2562,
+      "learning_rate": 0.00030209999999999997,
+      "loss": 0.2081,
       "step": 2483
     },
     {
       "epoch": 45.16,
-      "learning_rate": 0.00030209999999999997,
-      "loss": 0.2242,
+      "learning_rate": 0.000302,
+      "loss": 0.2343,
       "step": 2484
     },
     {
       "epoch": 45.18,
-      "learning_rate": 0.000302,
-      "loss": 0.176,
+      "learning_rate": 0.0003019,
+      "loss": 0.1387,
       "step": 2485
     },
     {
       "epoch": 45.2,
-      "learning_rate": 0.0003019,
-      "loss": 0.1972,
+      "learning_rate": 0.0003018,
+      "loss": 0.1722,
       "step": 2486
     },
     {
       "epoch": 45.22,
-      "learning_rate": 0.0003018,
-      "loss": 0.2052,
+      "learning_rate": 0.0003017,
+      "loss": 0.1706,
       "step": 2487
     },
     {
       "epoch": 45.24,
-      "learning_rate": 0.0003017,
-      "loss": 0.2254,
+      "learning_rate": 0.00030159999999999996,
+      "loss": 0.2183,
       "step": 2488
     },
     {
       "epoch": 45.25,
-      "learning_rate": 0.00030159999999999996,
-      "loss": 0.2146,
+      "learning_rate": 0.0003015,
+      "loss": 0.2101,
       "step": 2489
     },
     {
       "epoch": 45.27,
-      "learning_rate": 0.0003015,
-      "loss": 0.1642,
+      "learning_rate": 0.0003014,
+      "loss": 0.1605,
       "step": 2490
     },
     {
       "epoch": 45.29,
-      "learning_rate": 0.0003014,
-      "loss": 0.1753,
+      "learning_rate": 0.0003013,
+      "loss": 0.1714,
       "step": 2491
     },
     {
       "epoch": 45.31,
-      "learning_rate": 0.0003013,
-      "loss": 0.2031,
+      "learning_rate": 0.0003012,
+      "loss": 0.1776,
       "step": 2492
     },
     {
       "epoch": 45.33,
-      "learning_rate": 0.0003012,
-      "loss": 0.2049,
+      "learning_rate": 0.0003011,
+      "loss": 0.1946,
       "step": 2493
     },
     {
       "epoch": 45.35,
-      "learning_rate": 0.0003011,
-      "loss": 0.2191,
+      "learning_rate": 0.000301,
+      "loss": 0.226,
       "step": 2494
     },
     {
       "epoch": 45.36,
-      "learning_rate": 0.000301,
-      "loss": 0.1727,
+      "learning_rate": 0.0003009,
+      "loss": 0.1732,
       "step": 2495
     },
     {
       "epoch": 45.38,
-      "learning_rate": 0.0003009,
-      "loss": 0.2135,
+      "learning_rate": 0.0003008,
+      "loss": 0.2209,
       "step": 2496
     },
     {
       "epoch": 45.4,
-      "learning_rate": 0.0003008,
-      "loss": 0.234,
+      "learning_rate": 0.00030070000000000004,
+      "loss": 0.2021,
       "step": 2497
     },
     {
       "epoch": 45.42,
-      "learning_rate": 0.00030070000000000004,
-      "loss": 0.1881,
+      "learning_rate": 0.0003006,
+      "loss": 0.1633,
       "step": 2498
     },
     {
       "epoch": 45.44,
-      "learning_rate": 0.0003006,
-      "loss": 0.2075,
+      "learning_rate": 0.0003005,
+      "loss": 0.1515,
       "step": 2499
     },
     {
       "epoch": 45.45,
-      "learning_rate": 0.0003005,
-      "loss": 0.1718,
+      "learning_rate": 0.0003004,
+      "loss": 0.1546,
       "step": 2500
     },
     {
       "epoch": 45.45,
-      "eval_cer": 0.09199538223961379,
-      "eval_loss": 0.4219995439052582,
-      "eval_runtime": 13.9794,
-      "eval_samples_per_second": 117.816,
-      "eval_steps_per_second": 1.86,
-      "eval_wer": 0.3611479930548463,
+      "eval_cer": 0.09042114574431401,
+      "eval_loss": 0.4126885235309601,
+      "eval_runtime": 14.1276,
+      "eval_samples_per_second": 116.58,
+      "eval_steps_per_second": 1.84,
+      "eval_wer": 0.3602287815340619,
       "step": 2500
     },
     {
       "epoch": 45.47,
-      "learning_rate": 0.0003004,
-      "loss": 0.2299,
+      "learning_rate": 0.00030030000000000004,
+      "loss": 0.2105,
       "step": 2501
     },
     {
       "epoch": 45.49,
-      "learning_rate": 0.00030030000000000004,
-      "loss": 0.2225,
+      "learning_rate": 0.00030020000000000003,
+      "loss": 0.2316,
       "step": 2502
     },
     {
       "epoch": 45.51,
-      "learning_rate": 0.00030020000000000003,
-      "loss": 0.1819,
+      "learning_rate": 0.0003001,
+      "loss": 0.1583,
       "step": 2503
     },
     {
       "epoch": 45.53,
-      "learning_rate": 0.0003001,
-      "loss": 0.207,
+      "learning_rate": 0.0003,
+      "loss": 0.1951,
       "step": 2504
     },
     {
       "epoch": 45.55,
-      "learning_rate": 0.0003,
-      "loss": 0.2104,
+      "learning_rate": 0.0002999,
+      "loss": 0.2058,
       "step": 2505
     },
     {
       "epoch": 45.56,
-      "learning_rate": 0.0002999,
-      "loss": 0.2134,
+      "learning_rate": 0.0002998,
+      "loss": 0.1974,
       "step": 2506
     },
     {
       "epoch": 45.58,
-      "learning_rate": 0.0002998,
-      "loss": 0.2157,
+      "learning_rate": 0.0002997,
+      "loss": 0.1803,
       "step": 2507
     },
     {
       "epoch": 45.6,
-      "learning_rate": 0.0002997,
-      "loss": 0.2129,
+      "learning_rate": 0.00029959999999999996,
+      "loss": 0.2012,
       "step": 2508
     },
     {
       "epoch": 45.62,
-      "learning_rate": 0.00029959999999999996,
-      "loss": 0.2274,
+      "learning_rate": 0.0002995,
+      "loss": 0.2081,
       "step": 2509
     },
     {
       "epoch": 45.64,
-      "learning_rate": 0.0002995,
-      "loss": 0.2728,
+      "learning_rate": 0.0002994,
+      "loss": 0.2216,
       "step": 2510
     },
     {
       "epoch": 45.65,
-      "learning_rate": 0.0002994,
-      "loss": 0.1827,
+      "learning_rate": 0.0002993,
+      "loss": 0.1778,
       "step": 2511
     },
     {
       "epoch": 45.67,
-      "learning_rate": 0.0002993,
-      "loss": 0.2088,
+      "learning_rate": 0.0002992,
+      "loss": 0.2094,
       "step": 2512
     },
     {
       "epoch": 45.69,
-      "learning_rate": 0.0002992,
-      "loss": 0.1735,
+      "learning_rate": 0.0002991,
+      "loss": 0.1502,
       "step": 2513
     },
     {
       "epoch": 45.71,
-      "learning_rate": 0.0002991,
-      "loss": 0.1769,
+      "learning_rate": 0.000299,
+      "loss": 0.2022,
       "step": 2514
     },
     {
       "epoch": 45.73,
-      "learning_rate": 0.000299,
-      "loss": 0.2024,
+      "learning_rate": 0.0002989,
+      "loss": 0.1581,
       "step": 2515
     },
     {
       "epoch": 45.75,
-      "learning_rate": 0.0002989,
-      "loss": 0.1918,
+      "learning_rate": 0.0002988,
+      "loss": 0.1788,
       "step": 2516
     },
     {
       "epoch": 45.76,
-      "learning_rate": 0.0002988,
-      "loss": 0.2167,
+      "learning_rate": 0.00029870000000000005,
+      "loss": 0.2054,
       "step": 2517
     },
     {
       "epoch": 45.78,
-      "learning_rate": 0.00029870000000000005,
-      "loss": 0.1653,
+      "learning_rate": 0.0002986,
+      "loss": 0.1605,
       "step": 2518
     },
     {
       "epoch": 45.8,
-      "learning_rate": 0.0002986,
-      "loss": 0.238,
+      "learning_rate": 0.0002985,
+      "loss": 0.2244,
       "step": 2519
     },
     {
       "epoch": 45.82,
-      "learning_rate": 0.0002985,
-      "loss": 0.1717,
+      "learning_rate": 0.0002984,
+      "loss": 0.135,
       "step": 2520
     },
     {
       "epoch": 45.84,
-      "learning_rate": 0.0002984,
-      "loss": 0.211,
+      "learning_rate": 0.00029830000000000004,
+      "loss": 0.1977,
       "step": 2521
     },
     {
       "epoch": 45.85,
-      "learning_rate": 0.00029830000000000004,
-      "loss": 0.1916,
+      "learning_rate": 0.00029820000000000004,
+      "loss": 0.2097,
       "step": 2522
     },
     {
       "epoch": 45.87,
-      "learning_rate": 0.00029820000000000004,
-      "loss": 0.2156,
+      "learning_rate": 0.0002981,
+      "loss": 0.1817,
       "step": 2523
     },
     {
       "epoch": 45.89,
-      "learning_rate": 0.0002981,
-      "loss": 0.1763,
+      "learning_rate": 0.000298,
+      "loss": 0.1637,
       "step": 2524
     },
     {
       "epoch": 45.91,
-      "learning_rate": 0.000298,
-      "loss": 0.1558,
+      "learning_rate": 0.00029790000000000003,
+      "loss": 0.1501,
       "step": 2525
     },
     {
       "epoch": 45.93,
-      "learning_rate": 0.00029790000000000003,
-      "loss": 0.246,
+      "learning_rate": 0.00029780000000000003,
+      "loss": 0.2338,
       "step": 2526
     },
     {
       "epoch": 45.95,
-      "learning_rate": 0.00029780000000000003,
-      "loss": 0.1656,
+      "learning_rate": 0.0002977,
+      "loss": 0.1885,
       "step": 2527
     },
     {
       "epoch": 45.96,
-      "learning_rate": 0.0002977,
-      "loss": 0.1827,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.1542,
       "step": 2528
     },
     {
       "epoch": 45.98,
-      "learning_rate": 0.00029759999999999997,
-      "loss": 0.2882,
+      "learning_rate": 0.00029749999999999997,
+      "loss": 0.2372,
       "step": 2529
     },
     {
       "epoch": 46.0,
-      "learning_rate": 0.00029749999999999997,
-      "loss": 0.1663,
+      "learning_rate": 0.0002974,
+      "loss": 0.1637,
       "step": 2530
     },
     {
       "epoch": 46.02,
-      "learning_rate": 0.0002974,
-      "loss": 0.1946,
+      "learning_rate": 0.0002973,
+      "loss": 0.1931,
       "step": 2531
     },
     {
       "epoch": 46.04,
-      "learning_rate": 0.0002973,
-      "loss": 0.2183,
+      "learning_rate": 0.0002972,
+      "loss": 0.2282,
       "step": 2532
     },
     {
       "epoch": 46.05,
-      "learning_rate": 0.0002972,
-      "loss": 0.1979,
+      "learning_rate": 0.00029709999999999996,
+      "loss": 0.185,
       "step": 2533
     },
     {
       "epoch": 46.07,
-      "learning_rate": 0.00029709999999999996,
-      "loss": 0.1846,
+      "learning_rate": 0.000297,
+      "loss": 0.1902,
       "step": 2534
     },
     {
       "epoch": 46.09,
-      "learning_rate": 0.000297,
-      "loss": 0.2062,
+      "learning_rate": 0.0002969,
+      "loss": 0.2026,
       "step": 2535
     },
     {
       "epoch": 46.11,
-      "learning_rate": 0.0002969,
-      "loss": 0.1811,
+      "learning_rate": 0.0002968,
+      "loss": 0.1545,
       "step": 2536
     },
     {
       "epoch": 46.13,
-      "learning_rate": 0.0002968,
-      "loss": 0.1936,
+      "learning_rate": 0.0002967,
+      "loss": 0.1898,
       "step": 2537
     },
     {
       "epoch": 46.15,
-      "learning_rate": 0.0002967,
-      "loss": 0.2268,
+      "learning_rate": 0.0002966,
+      "loss": 0.2193,
       "step": 2538
     },
     {
       "epoch": 46.16,
-      "learning_rate": 0.0002966,
-      "loss": 0.207,
+      "learning_rate": 0.0002965,
+      "loss": 0.1966,
       "step": 2539
     },
     {
       "epoch": 46.18,
-      "learning_rate": 0.0002965,
-      "loss": 0.2419,
+      "learning_rate": 0.0002964,
+      "loss": 0.2077,
       "step": 2540
     },
     {
       "epoch": 46.2,
-      "learning_rate": 0.0002964,
-      "loss": 0.1939,
+      "learning_rate": 0.0002963,
+      "loss": 0.189,
       "step": 2541
     },
     {
       "epoch": 46.22,
-      "learning_rate": 0.0002963,
-      "loss": 0.2035,
+      "learning_rate": 0.00029620000000000004,
+      "loss": 0.1519,
       "step": 2542
     },
     {
       "epoch": 46.24,
-      "learning_rate": 0.00029620000000000004,
-      "loss": 0.1859,
+      "learning_rate": 0.0002961,
+      "loss": 0.1667,
       "step": 2543
     },
     {
       "epoch": 46.25,
-      "learning_rate": 0.0002961,
-      "loss": 0.1603,
+      "learning_rate": 0.000296,
+      "loss": 0.1495,
       "step": 2544
     },
     {
       "epoch": 46.27,
-      "learning_rate": 0.000296,
-      "loss": 0.2161,
+      "learning_rate": 0.0002959,
+      "loss": 0.2113,
       "step": 2545
     },
     {
       "epoch": 46.29,
-      "learning_rate": 0.0002959,
-      "loss": 0.2148,
+      "learning_rate": 0.00029580000000000003,
+      "loss": 0.2072,
       "step": 2546
     },
     {
       "epoch": 46.31,
-      "learning_rate": 0.00029580000000000003,
-      "loss": 0.2025,
+      "learning_rate": 0.00029570000000000003,
+      "loss": 0.1952,
       "step": 2547
     },
     {
       "epoch": 46.33,
-      "learning_rate": 0.00029570000000000003,
-      "loss": 0.1473,
+      "learning_rate": 0.0002956,
+      "loss": 0.2115,
       "step": 2548
     },
     {
       "epoch": 46.35,
-      "learning_rate": 0.0002956,
-      "loss": 0.1759,
+      "learning_rate": 0.00029549999999999997,
+      "loss": 0.1517,
       "step": 2549
     },
     {
       "epoch": 46.36,
-      "learning_rate": 0.00029549999999999997,
-      "loss": 0.1945,
+      "learning_rate": 0.0002954,
+      "loss": 0.1987,
       "step": 2550
     },
     {
       "epoch": 46.38,
-      "learning_rate": 0.0002954,
-      "loss": 0.1876,
+      "learning_rate": 0.0002953,
+      "loss": 0.1671,
       "step": 2551
     },
     {
       "epoch": 46.4,
-      "learning_rate": 0.0002953,
-      "loss": 0.1844,
+      "learning_rate": 0.0002952,
+      "loss": 0.1872,
       "step": 2552
     },
     {
       "epoch": 46.42,
-      "learning_rate": 0.0002952,
-      "loss": 0.1633,
+      "learning_rate": 0.00029509999999999996,
+      "loss": 0.2037,
       "step": 2553
     },
     {
       "epoch": 46.44,
-      "learning_rate": 0.00029509999999999996,
-      "loss": 0.1793,
+      "learning_rate": 0.000295,
+      "loss": 0.1844,
       "step": 2554
     },
     {
       "epoch": 46.45,
-      "learning_rate": 0.000295,
-      "loss": 0.1962,
+      "learning_rate": 0.0002949,
+      "loss": 0.1584,
       "step": 2555
     },
     {
       "epoch": 46.47,
-      "learning_rate": 0.0002949,
-      "loss": 0.1735,
+      "learning_rate": 0.0002948,
+      "loss": 0.1737,
       "step": 2556
     },
     {
       "epoch": 46.49,
-      "learning_rate": 0.0002948,
-      "loss": 0.1796,
+      "learning_rate": 0.0002947,
+      "loss": 0.1612,
       "step": 2557
     },
     {
       "epoch": 46.51,
-      "learning_rate": 0.0002947,
-      "loss": 0.1783,
+      "learning_rate": 0.0002946,
+      "loss": 0.1876,
       "step": 2558
     },
     {
       "epoch": 46.53,
-      "learning_rate": 0.0002946,
-      "loss": 0.1971,
+      "learning_rate": 0.0002945,
+      "loss": 0.1965,
       "step": 2559
     },
     {
       "epoch": 46.55,
-      "learning_rate": 0.0002945,
-      "loss": 0.2347,
+      "learning_rate": 0.0002944,
+      "loss": 0.2066,
       "step": 2560
     },
     {
       "epoch": 46.56,
-      "learning_rate": 0.0002944,
-      "loss": 0.1421,
+      "learning_rate": 0.0002943,
+      "loss": 0.1263,
       "step": 2561
     },
     {
       "epoch": 46.58,
-      "learning_rate": 0.0002943,
-      "loss": 0.1733,
+      "learning_rate": 0.00029420000000000005,
+      "loss": 0.1693,
       "step": 2562
     },
     {
       "epoch": 46.6,
-      "learning_rate": 0.00029420000000000005,
-      "loss": 0.2076,
+      "learning_rate": 0.0002941,
+      "loss": 0.1874,
       "step": 2563
     },
     {
       "epoch": 46.62,
-      "learning_rate": 0.0002941,
-      "loss": 0.2398,
+      "learning_rate": 0.000294,
+      "loss": 0.1869,
       "step": 2564
     },
     {
       "epoch": 46.64,
-      "learning_rate": 0.000294,
-      "loss": 0.2108,
+      "learning_rate": 0.0002939,
+      "loss": 0.203,
       "step": 2565
     },
     {
       "epoch": 46.65,
-      "learning_rate": 0.0002939,
-      "loss": 0.2356,
+      "learning_rate": 0.0002938,
+      "loss": 0.2209,
       "step": 2566
     },
     {
       "epoch": 46.67,
-      "learning_rate": 0.0002938,
-      "loss": 0.1671,
+      "learning_rate": 0.00029370000000000004,
+      "loss": 0.1276,
       "step": 2567
     },
     {
       "epoch": 46.69,
-      "learning_rate": 0.00029370000000000004,
-      "loss": 0.199,
+      "learning_rate": 0.00029360000000000003,
+      "loss": 0.2109,
       "step": 2568
     },
     {
       "epoch": 46.71,
-      "learning_rate": 0.00029360000000000003,
-      "loss": 0.1918,
+      "learning_rate": 0.0002935,
+      "loss": 0.1647,
       "step": 2569
     },
     {
       "epoch": 46.73,
-      "learning_rate": 0.0002935,
-      "loss": 0.2138,
+      "learning_rate": 0.0002934,
+      "loss": 0.199,
       "step": 2570
     },
     {
       "epoch": 46.75,
-      "learning_rate": 0.0002934,
-      "loss": 0.1778,
+      "learning_rate": 0.00029330000000000003,
+      "loss": 0.1726,
       "step": 2571
     },
     {
       "epoch": 46.76,
-      "learning_rate": 0.00029330000000000003,
-      "loss": 0.1854,
+      "learning_rate": 0.0002932,
+      "loss": 0.1591,
       "step": 2572
     },
     {
       "epoch": 46.78,
-      "learning_rate": 0.0002932,
-      "loss": 0.1823,
+      "learning_rate": 0.0002931,
+      "loss": 0.1498,
       "step": 2573
     },
     {
       "epoch": 46.8,
-      "learning_rate": 0.0002931,
-      "loss": 0.2181,
+      "learning_rate": 0.00029299999999999997,
+      "loss": 0.1689,
       "step": 2574
     },
     {
       "epoch": 46.82,
-      "learning_rate": 0.00029299999999999997,
-      "loss": 0.2346,
+      "learning_rate": 0.0002929,
+      "loss": 0.2171,
       "step": 2575
     },
     {
       "epoch": 46.84,
-      "learning_rate": 0.0002929,
-      "loss": 0.1876,
+      "learning_rate": 0.0002928,
+      "loss": 0.2038,
       "step": 2576
     },
     {
       "epoch": 46.85,
-      "learning_rate": 0.0002928,
-      "loss": 0.1876,
+      "learning_rate": 0.0002927,
+      "loss": 0.1706,
       "step": 2577
     },
     {
       "epoch": 46.87,
-      "learning_rate": 0.0002927,
-      "loss": 0.2015,
+      "learning_rate": 0.0002926,
+      "loss": 0.1642,
       "step": 2578
     },
     {
       "epoch": 46.89,
-      "learning_rate": 0.0002926,
-      "loss": 0.2077,
+      "learning_rate": 0.0002925,
+      "loss": 0.2146,
       "step": 2579
     },
     {
       "epoch": 46.91,
-      "learning_rate": 0.0002925,
-      "loss": 0.1699,
+      "learning_rate": 0.0002924,
+      "loss": 0.1475,
       "step": 2580
     },
     {
       "epoch": 46.93,
-      "learning_rate": 0.0002924,
-      "loss": 0.2003,
+      "learning_rate": 0.0002923,
+      "loss": 0.2034,
       "step": 2581
     },
     {
       "epoch": 46.95,
-      "learning_rate": 0.0002923,
-      "loss": 0.1808,
+      "learning_rate": 0.0002922,
+      "loss": 0.1786,
       "step": 2582
     },
     {
       "epoch": 46.96,
-      "learning_rate": 0.0002922,
-      "loss": 0.1927,
+      "learning_rate": 0.00029210000000000005,
+      "loss": 0.1748,
       "step": 2583
     },
     {
       "epoch": 46.98,
-      "learning_rate": 0.00029210000000000005,
-      "loss": 0.1661,
+      "learning_rate": 0.000292,
+      "loss": 0.1849,
       "step": 2584
     },
     {
       "epoch": 47.0,
-      "learning_rate": 0.000292,
-      "loss": 0.1595,
+      "learning_rate": 0.0002919,
+      "loss": 0.1887,
       "step": 2585
     },
     {
       "epoch": 47.02,
-      "learning_rate": 0.0002919,
-      "loss": 0.233,
+      "learning_rate": 0.0002918,
+      "loss": 0.2191,
       "step": 2586
     },
     {
       "epoch": 47.04,
-      "learning_rate": 0.0002918,
-      "loss": 0.1832,
+      "learning_rate": 0.00029170000000000004,
+      "loss": 0.166,
       "step": 2587
     },
     {
       "epoch": 47.05,
-      "learning_rate": 0.00029170000000000004,
-      "loss": 0.1827,
+      "learning_rate": 0.00029160000000000004,
+      "loss": 0.1749,
       "step": 2588
     },
     {
       "epoch": 47.07,
-      "learning_rate": 0.00029160000000000004,
-      "loss": 0.2097,
+      "learning_rate": 0.0002915,
+      "loss": 0.1837,
       "step": 2589
     },
     {
       "epoch": 47.09,
-      "learning_rate": 0.0002915,
-      "loss": 0.1868,
+      "learning_rate": 0.0002914,
+      "loss": 0.2203,
       "step": 2590
     },
     {
       "epoch": 47.11,
-      "learning_rate": 0.0002914,
-      "loss": 0.2284,
+      "learning_rate": 0.00029130000000000003,
+      "loss": 0.1719,
       "step": 2591
     },
     {
       "epoch": 47.13,
-      "learning_rate": 0.00029130000000000003,
-      "loss": 0.2308,
+      "learning_rate": 0.00029120000000000003,
+      "loss": 0.1852,
       "step": 2592
     },
     {
       "epoch": 47.15,
-      "learning_rate": 0.00029120000000000003,
-      "loss": 0.207,
+      "learning_rate": 0.00029110000000000003,
+      "loss": 0.1871,
       "step": 2593
     },
     {
       "epoch": 47.16,
-      "learning_rate": 0.00029110000000000003,
-      "loss": 0.2232,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 0.2095,
       "step": 2594
     },
     {
       "epoch": 47.18,
-      "learning_rate": 0.00029099999999999997,
-      "loss": 0.2039,
+      "learning_rate": 0.0002909,
+      "loss": 0.1871,
       "step": 2595
     },
     {
       "epoch": 47.2,
-      "learning_rate": 0.0002909,
-      "loss": 0.205,
+      "learning_rate": 0.0002908,
+      "loss": 0.1746,
       "step": 2596
     },
     {
       "epoch": 47.22,
-      "learning_rate": 0.0002908,
-      "loss": 0.2456,
+      "learning_rate": 0.0002907,
+      "loss": 0.2386,
       "step": 2597
     },
     {
       "epoch": 47.24,
-      "learning_rate": 0.0002907,
-      "loss": 0.123,
+      "learning_rate": 0.0002906,
+      "loss": 0.1333,
       "step": 2598
     },
     {
       "epoch": 47.25,
-      "learning_rate": 0.0002906,
-      "loss": 0.1912,
+      "learning_rate": 0.00029049999999999996,
+      "loss": 0.1805,
       "step": 2599
     },
     {
       "epoch": 47.27,
-      "learning_rate": 0.00029049999999999996,
-      "loss": 0.2139,
+      "learning_rate": 0.0002904,
+      "loss": 0.1911,
       "step": 2600
     },
     {
       "epoch": 47.29,
-      "learning_rate": 0.0002904,
-      "loss": 0.1857,
+      "learning_rate": 0.0002903,
+      "loss": 0.1563,
       "step": 2601
     },
     {
       "epoch": 47.31,
-      "learning_rate": 0.0002903,
-      "loss": 0.2086,
+      "learning_rate": 0.0002902,
+      "loss": 0.2049,
       "step": 2602
     },
     {
       "epoch": 47.33,
-      "learning_rate": 0.0002902,
-      "loss": 0.1732,
+      "learning_rate": 0.0002901,
+      "loss": 0.1648,
       "step": 2603
     },
     {
       "epoch": 47.35,
-      "learning_rate": 0.0002901,
-      "loss": 0.175,
+      "learning_rate": 0.00029,
+      "loss": 0.158,
       "step": 2604
     },
     {
       "epoch": 47.36,
-      "learning_rate": 0.00029,
-      "loss": 0.2449,
+      "learning_rate": 0.0002899,
+      "loss": 0.2185,
       "step": 2605
     },
     {
       "epoch": 47.38,
-      "learning_rate": 0.0002899,
-      "loss": 0.1699,
+      "learning_rate": 0.0002898,
+      "loss": 0.1888,
       "step": 2606
     },
     {
       "epoch": 47.4,
-      "learning_rate": 0.0002898,
-      "loss": 0.1605,
+      "learning_rate": 0.0002897,
+      "loss": 0.1545,
       "step": 2607
     },
     {
       "epoch": 47.42,
-      "learning_rate": 0.0002897,
-      "loss": 0.1942,
+      "learning_rate": 0.00028960000000000005,
+      "loss": 0.2102,
       "step": 2608
     },
     {
       "epoch": 47.44,
-      "learning_rate": 0.00028960000000000005,
-      "loss": 0.2204,
+      "learning_rate": 0.0002895,
+      "loss": 0.2249,
       "step": 2609
     },
     {
       "epoch": 47.45,
-      "learning_rate": 0.0002895,
-      "loss": 0.1909,
+      "learning_rate": 0.0002894,
+      "loss": 0.1975,
       "step": 2610
     },
     {
       "epoch": 47.47,
-      "learning_rate": 0.0002894,
-      "loss": 0.2111,
+      "learning_rate": 0.0002893,
+      "loss": 0.2175,
       "step": 2611
     },
     {
       "epoch": 47.49,
-      "learning_rate": 0.0002893,
-      "loss": 0.2331,
+      "learning_rate": 0.00028920000000000004,
+      "loss": 0.2088,
       "step": 2612
     },
     {
       "epoch": 47.51,
-      "learning_rate": 0.00028920000000000004,
-      "loss": 0.2427,
+      "learning_rate": 0.00028910000000000003,
+      "loss": 0.2221,
       "step": 2613
     },
     {
       "epoch": 47.53,
-      "learning_rate": 0.00028910000000000003,
-      "loss": 0.2072,
+      "learning_rate": 0.000289,
+      "loss": 0.1574,
       "step": 2614
     },
     {
       "epoch": 47.55,
-      "learning_rate": 0.000289,
-      "loss": 0.176,
+      "learning_rate": 0.0002889,
+      "loss": 0.1616,
       "step": 2615
     },
     {
       "epoch": 47.56,
-      "learning_rate": 0.0002889,
-      "loss": 0.2041,
+      "learning_rate": 0.0002888,
+      "loss": 0.2467,
       "step": 2616
     },
     {
       "epoch": 47.58,
-      "learning_rate": 0.0002888,
-      "loss": 0.1971,
+      "learning_rate": 0.0002887,
+      "loss": 0.1872,
       "step": 2617
     },
     {
       "epoch": 47.6,
-      "learning_rate": 0.0002887,
-      "loss": 0.2051,
+      "learning_rate": 0.0002886,
+      "loss": 0.1951,
       "step": 2618
     },
     {
       "epoch": 47.62,
-      "learning_rate": 0.0002886,
-      "loss": 0.1738,
+      "learning_rate": 0.00028849999999999997,
+      "loss": 0.1668,
       "step": 2619
     },
     {
       "epoch": 47.64,
-      "learning_rate": 0.00028849999999999997,
-      "loss": 0.1918,
+      "learning_rate": 0.0002884,
+      "loss": 0.1706,
       "step": 2620
     },
     {
       "epoch": 47.65,
-      "learning_rate": 0.0002884,
-      "loss": 0.2157,
+      "learning_rate": 0.0002883,
+      "loss": 0.1999,
       "step": 2621
     },
     {
       "epoch": 47.67,
-      "learning_rate": 0.0002883,
-      "loss": 0.1938,
+      "learning_rate": 0.0002882,
+      "loss": 0.1664,
       "step": 2622
     },
     {
       "epoch": 47.69,
-      "learning_rate": 0.0002882,
-      "loss": 0.2802,
+      "learning_rate": 0.0002881,
+      "loss": 0.2241,
       "step": 2623
     },
     {
       "epoch": 47.71,
-      "learning_rate": 0.0002881,
-      "loss": 0.2276,
+      "learning_rate": 0.000288,
+      "loss": 0.1912,
       "step": 2624
     },
     {
       "epoch": 47.73,
-      "learning_rate": 0.000288,
-      "loss": 0.2444,
+      "learning_rate": 0.0002879,
+      "loss": 0.2309,
       "step": 2625
     },
     {
       "epoch": 47.75,
-      "learning_rate": 0.0002879,
-      "loss": 0.1628,
+      "learning_rate": 0.0002878,
+      "loss": 0.1841,
       "step": 2626
     },
     {
       "epoch": 47.76,
-      "learning_rate": 0.0002878,
-      "loss": 0.2129,
+      "learning_rate": 0.0002877,
+      "loss": 0.1954,
       "step": 2627
     },
     {
       "epoch": 47.78,
-      "learning_rate": 0.0002877,
-      "loss": 0.1773,
+      "learning_rate": 0.00028760000000000005,
+      "loss": 0.1493,
       "step": 2628
     },
     {
       "epoch": 47.8,
-      "learning_rate": 0.00028760000000000005,
-      "loss": 0.1793,
+      "learning_rate": 0.0002875,
+      "loss": 0.1748,
       "step": 2629
     },
     {
       "epoch": 47.82,
-      "learning_rate": 0.0002875,
-      "loss": 0.2154,
+      "learning_rate": 0.0002874,
+      "loss": 0.195,
       "step": 2630
     },
     {
       "epoch": 47.84,
-      "learning_rate": 0.0002874,
-      "loss": 0.1801,
+      "learning_rate": 0.0002873,
+      "loss": 0.1711,
       "step": 2631
     },
     {
       "epoch": 47.85,
-      "learning_rate": 0.0002873,
-      "loss": 0.2116,
+      "learning_rate": 0.00028720000000000004,
+      "loss": 0.1952,
       "step": 2632
     },
     {
       "epoch": 47.87,
-      "learning_rate": 0.00028720000000000004,
-      "loss": 0.2283,
+      "learning_rate": 0.00028710000000000004,
+      "loss": 0.1998,
       "step": 2633
     },
     {
       "epoch": 47.89,
-      "learning_rate": 0.00028710000000000004,
-      "loss": 0.1821,
+      "learning_rate": 0.000287,
+      "loss": 0.1567,
       "step": 2634
     },
     {
       "epoch": 47.91,
-      "learning_rate": 0.000287,
-      "loss": 0.2025,
+      "learning_rate": 0.0002869,
+      "loss": 0.1478,
       "step": 2635
     },
     {
       "epoch": 47.93,
-      "learning_rate": 0.0002869,
-      "loss": 0.212,
+      "learning_rate": 0.00028680000000000003,
+      "loss": 0.1909,
       "step": 2636
     },
     {
       "epoch": 47.95,
-      "learning_rate": 0.00028680000000000003,
-      "loss": 0.2165,
+      "learning_rate": 0.00028670000000000003,
+      "loss": 0.198,
       "step": 2637
     },
     {
       "epoch": 47.96,
-      "learning_rate": 0.00028670000000000003,
-      "loss": 0.198,
+      "learning_rate": 0.00028660000000000003,
+      "loss": 0.1701,
       "step": 2638
     },
     {
       "epoch": 47.98,
-      "learning_rate": 0.00028660000000000003,
-      "loss": 0.2389,
+      "learning_rate": 0.00028649999999999997,
+      "loss": 0.2044,
       "step": 2639
     },
     {
       "epoch": 48.0,
-      "learning_rate": 0.00028649999999999997,
-      "loss": 0.1521,
+      "learning_rate": 0.00028639999999999997,
+      "loss": 0.1507,
       "step": 2640
     },
     {
       "epoch": 48.02,
-      "learning_rate": 0.00028639999999999997,
-      "loss": 0.2499,
+      "learning_rate": 0.0002863,
+      "loss": 0.1961,
       "step": 2641
     },
     {
       "epoch": 48.04,
-      "learning_rate": 0.0002863,
-      "loss": 0.1845,
+      "learning_rate": 0.0002862,
+      "loss": 0.1775,
       "step": 2642
     },
     {
       "epoch": 48.05,
-      "learning_rate": 0.0002862,
-      "loss": 0.1991,
+      "learning_rate": 0.0002861,
+      "loss": 0.1713,
       "step": 2643
     },
     {
       "epoch": 48.07,
-      "learning_rate": 0.0002861,
-      "loss": 0.2017,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 0.1936,
       "step": 2644
     },
     {
       "epoch": 48.09,
-      "learning_rate": 0.00028599999999999996,
-      "loss": 0.1836,
+      "learning_rate": 0.0002859,
+      "loss": 0.1912,
       "step": 2645
     },
     {
       "epoch": 48.11,
-      "learning_rate": 0.0002859,
-      "loss": 0.1378,
+      "learning_rate": 0.0002858,
+      "loss": 0.1425,
       "step": 2646
     },
     {
       "epoch": 48.13,
-      "learning_rate": 0.0002858,
-      "loss": 0.2082,
+      "learning_rate": 0.0002857,
+      "loss": 0.1966,
       "step": 2647
     },
     {
       "epoch": 48.15,
-      "learning_rate": 0.0002857,
-      "loss": 0.2499,
+      "learning_rate": 0.0002856,
+      "loss": 0.2081,
       "step": 2648
     },
     {
       "epoch": 48.16,
-      "learning_rate": 0.0002856,
-      "loss": 0.2091,
+      "learning_rate": 0.0002855,
+      "loss": 0.1787,
       "step": 2649
     },
     {
       "epoch": 48.18,
-      "learning_rate": 0.0002855,
-      "loss": 0.1903,
+      "learning_rate": 0.0002854,
+      "loss": 0.1921,
       "step": 2650
     },
     {
       "epoch": 48.2,
-      "learning_rate": 0.0002854,
-      "loss": 0.1934,
+      "learning_rate": 0.0002853,
+      "loss": 0.1514,
       "step": 2651
     },
     {
       "epoch": 48.22,
-      "learning_rate": 0.0002853,
-      "loss": 0.1914,
+      "learning_rate": 0.0002852,
+      "loss": 0.1684,
       "step": 2652
     },
     {
       "epoch": 48.24,
-      "learning_rate": 0.0002852,
-      "loss": 0.1712,
+      "learning_rate": 0.00028510000000000005,
+      "loss": 0.1637,
       "step": 2653
     },
     {
       "epoch": 48.25,
-      "learning_rate": 0.00028510000000000005,
-      "loss": 0.1682,
+      "learning_rate": 0.000285,
+      "loss": 0.1587,
       "step": 2654
     },
     {
       "epoch": 48.27,
-      "learning_rate": 0.000285,
-      "loss": 0.1911,
+      "learning_rate": 0.0002849,
+      "loss": 0.1481,
       "step": 2655
     },
     {
       "epoch": 48.29,
-      "learning_rate": 0.0002849,
-      "loss": 0.1752,
+      "learning_rate": 0.0002848,
+      "loss": 0.1742,
       "step": 2656
     },
     {
       "epoch": 48.31,
-      "learning_rate": 0.0002848,
-      "loss": 0.1953,
+      "learning_rate": 0.00028470000000000004,
+      "loss": 0.1743,
       "step": 2657
     },
     {
       "epoch": 48.33,
-      "learning_rate": 0.00028470000000000004,
-      "loss": 0.2195,
+      "learning_rate": 0.00028460000000000003,
+      "loss": 0.1845,
       "step": 2658
     },
     {
       "epoch": 48.35,
-      "learning_rate": 0.00028460000000000003,
-      "loss": 0.2534,
+      "learning_rate": 0.0002845,
+      "loss": 0.2031,
       "step": 2659
     },
     {
       "epoch": 48.36,
-      "learning_rate": 0.0002845,
-      "loss": 0.2496,
+      "learning_rate": 0.0002844,
+      "loss": 0.199,
       "step": 2660
     },
     {
       "epoch": 48.38,
-      "learning_rate": 0.0002844,
-      "loss": 0.1749,
+      "learning_rate": 0.0002843,
+      "loss": 0.1624,
       "step": 2661
     },
     {
       "epoch": 48.4,
-      "learning_rate": 0.0002843,
-      "loss": 0.1836,
+      "learning_rate": 0.0002842,
+      "loss": 0.1982,
       "step": 2662
     },
     {
       "epoch": 48.42,
-      "learning_rate": 0.0002842,
-      "loss": 0.2114,
+      "learning_rate": 0.0002841,
+      "loss": 0.1812,
       "step": 2663
     },
     {
       "epoch": 48.44,
-      "learning_rate": 0.0002841,
-      "loss": 0.1753,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 0.1593,
       "step": 2664
     },
     {
       "epoch": 48.45,
-      "learning_rate": 0.00028399999999999996,
-      "loss": 0.2052,
+      "learning_rate": 0.0002839,
+      "loss": 0.1516,
       "step": 2665
     },
     {
       "epoch": 48.47,
-      "learning_rate": 0.0002839,
-      "loss": 0.1985,
+      "learning_rate": 0.0002838,
+      "loss": 0.2029,
       "step": 2666
     },
     {
       "epoch": 48.49,
-      "learning_rate": 0.0002838,
-      "loss": 0.2102,
+      "learning_rate": 0.0002837,
+      "loss": 0.2042,
       "step": 2667
     },
     {
       "epoch": 48.51,
-      "learning_rate": 0.0002837,
-      "loss": 0.1761,
+      "learning_rate": 0.0002836,
+      "loss": 0.1611,
       "step": 2668
     },
     {
       "epoch": 48.53,
-      "learning_rate": 0.0002836,
-      "loss": 0.1555,
+      "learning_rate": 0.0002835,
+      "loss": 0.1463,
       "step": 2669
     },
     {
       "epoch": 48.55,
-      "learning_rate": 0.0002835,
-      "loss": 0.1753,
+      "learning_rate": 0.0002834,
+      "loss": 0.1876,
       "step": 2670
     },
     {
       "epoch": 48.56,
-      "learning_rate": 0.0002834,
-      "loss": 0.1332,
+      "learning_rate": 0.0002833,
+      "loss": 0.1261,
       "step": 2671
     },
     {
       "epoch": 48.58,
-      "learning_rate": 0.0002833,
-      "loss": 0.2106,
+      "learning_rate": 0.0002832,
+      "loss": 0.2022,
       "step": 2672
     },
     {
       "epoch": 48.6,
-      "learning_rate": 0.0002832,
-      "loss": 0.1798,
+      "learning_rate": 0.00028310000000000005,
+      "loss": 0.1429,
       "step": 2673
     },
     {
       "epoch": 48.62,
-      "learning_rate": 0.00028310000000000005,
-      "loss": 0.2107,
+      "learning_rate": 0.000283,
+      "loss": 0.191,
       "step": 2674
     },
     {
       "epoch": 48.64,
-      "learning_rate": 0.000283,
-      "loss": 0.2092,
+      "learning_rate": 0.0002829,
+      "loss": 0.2045,
       "step": 2675
     },
     {
       "epoch": 48.65,
-      "learning_rate": 0.0002829,
-      "loss": 0.1742,
+      "learning_rate": 0.0002828,
+      "loss": 0.1713,
       "step": 2676
     },
     {
       "epoch": 48.67,
-      "learning_rate": 0.0002828,
-      "loss": 0.1849,
+      "learning_rate": 0.0002827,
+      "loss": 0.1314,
       "step": 2677
     },
     {
       "epoch": 48.69,
-      "learning_rate": 0.0002827,
-      "loss": 0.1817,
+      "learning_rate": 0.00028260000000000004,
+      "loss": 0.1642,
       "step": 2678
     },
     {
       "epoch": 48.71,
-      "learning_rate": 0.00028260000000000004,
-      "loss": 0.2091,
+      "learning_rate": 0.0002825,
+      "loss": 0.2042,
       "step": 2679
     },
     {
       "epoch": 48.73,
-      "learning_rate": 0.0002825,
-      "loss": 0.2092,
+      "learning_rate": 0.0002824,
+      "loss": 0.1875,
       "step": 2680
     },
     {
       "epoch": 48.75,
-      "learning_rate": 0.0002824,
-      "loss": 0.2028,
+      "learning_rate": 0.0002823,
+      "loss": 0.1944,
       "step": 2681
     },
     {
       "epoch": 48.76,
-      "learning_rate": 0.0002823,
-      "loss": 0.1583,
+      "learning_rate": 0.00028220000000000003,
+      "loss": 0.1194,
       "step": 2682
     },
     {
       "epoch": 48.78,
-      "learning_rate": 0.00028220000000000003,
-      "loss": 0.1852,
+      "learning_rate": 0.0002821,
+      "loss": 0.1821,
       "step": 2683
     },
     {
       "epoch": 48.8,
-      "learning_rate": 0.0002821,
-      "loss": 0.2054,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.2399,
       "step": 2684
     },
     {
       "epoch": 48.82,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 0.1803,
+      "learning_rate": 0.00028189999999999997,
+      "loss": 0.1883,
       "step": 2685
     },
     {
       "epoch": 48.84,
-      "learning_rate": 0.00028189999999999997,
-      "loss": 0.1565,
+      "learning_rate": 0.0002818,
+      "loss": 0.1666,
       "step": 2686
     },
     {
       "epoch": 48.85,
-      "learning_rate": 0.0002818,
-      "loss": 0.1982,
+      "learning_rate": 0.0002817,
+      "loss": 0.1802,
       "step": 2687
     },
     {
       "epoch": 48.87,
-      "learning_rate": 0.0002817,
-      "loss": 0.1288,
+      "learning_rate": 0.0002816,
+      "loss": 0.1259,
       "step": 2688
     },
     {
       "epoch": 48.89,
-      "learning_rate": 0.0002816,
-      "loss": 0.1784,
+      "learning_rate": 0.00028149999999999996,
+      "loss": 0.2332,
       "step": 2689
     },
     {
       "epoch": 48.91,
-      "learning_rate": 0.00028149999999999996,
-      "loss": 0.2098,
+      "learning_rate": 0.0002814,
+      "loss": 0.1452,
       "step": 2690
     },
     {
       "epoch": 48.93,
-      "learning_rate": 0.0002814,
-      "loss": 0.21,
+      "learning_rate": 0.0002813,
+      "loss": 0.2312,
       "step": 2691
     },
     {
       "epoch": 48.95,
-      "learning_rate": 0.0002813,
-      "loss": 0.2063,
+      "learning_rate": 0.0002812,
+      "loss": 0.1881,
       "step": 2692
     },
     {
       "epoch": 48.96,
-      "learning_rate": 0.0002812,
-      "loss": 0.1507,
+      "learning_rate": 0.0002811,
+      "loss": 0.1582,
       "step": 2693
     },
     {
       "epoch": 48.98,
-      "learning_rate": 0.0002811,
-      "loss": 0.1766,
+      "learning_rate": 0.00028100000000000005,
+      "loss": 0.1762,
       "step": 2694
     },
     {
       "epoch": 49.0,
-      "learning_rate": 0.00028100000000000005,
-      "loss": 0.1388,
+      "learning_rate": 0.0002809,
+      "loss": 0.1691,
       "step": 2695
     },
     {
       "epoch": 49.02,
-      "learning_rate": 0.0002809,
-      "loss": 0.1745,
+      "learning_rate": 0.0002808,
+      "loss": 0.1778,
       "step": 2696
     },
     {
       "epoch": 49.04,
-      "learning_rate": 0.0002808,
-      "loss": 0.2425,
+      "learning_rate": 0.0002807,
+      "loss": 0.239,
       "step": 2697
     },
     {
       "epoch": 49.05,
-      "learning_rate": 0.0002807,
-      "loss": 0.1841,
+      "learning_rate": 0.00028060000000000004,
+      "loss": 0.1772,
       "step": 2698
     },
     {
       "epoch": 49.07,
-      "learning_rate": 0.00028060000000000004,
-      "loss": 0.2203,
+      "learning_rate": 0.00028050000000000004,
+      "loss": 0.198,
       "step": 2699
     },
     {
       "epoch": 49.09,
-      "learning_rate": 0.00028050000000000004,
-      "loss": 0.1533,
+      "learning_rate": 0.0002804,
+      "loss": 0.1578,
       "step": 2700
     },
     {
       "epoch": 49.11,
-      "learning_rate": 0.0002804,
-      "loss": 0.191,
+      "learning_rate": 0.0002803,
+      "loss": 0.1675,
       "step": 2701
     },
     {
       "epoch": 49.13,
-      "learning_rate": 0.0002803,
-      "loss": 0.221,
+      "learning_rate": 0.00028020000000000003,
+      "loss": 0.1847,
       "step": 2702
     },
     {
       "epoch": 49.15,
-      "learning_rate": 0.00028020000000000003,
-      "loss": 0.2226,
+      "learning_rate": 0.00028010000000000003,
+      "loss": 0.2134,
       "step": 2703
     },
     {
       "epoch": 49.16,
-      "learning_rate": 0.00028010000000000003,
-      "loss": 0.1886,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 0.1722,
       "step": 2704
     },
     {
       "epoch": 49.18,
-      "learning_rate": 0.00028000000000000003,
-      "loss": 0.1513,
+      "learning_rate": 0.0002799,
+      "loss": 0.1365,
       "step": 2705
     },
     {
       "epoch": 49.2,
-      "learning_rate": 0.0002799,
-      "loss": 0.1888,
+      "learning_rate": 0.0002798,
+      "loss": 0.1818,
       "step": 2706
     },
     {
       "epoch": 49.22,
-      "learning_rate": 0.0002798,
-      "loss": 0.1889,
+      "learning_rate": 0.0002797,
+      "loss": 0.19,
       "step": 2707
     },
     {
       "epoch": 49.24,
-      "learning_rate": 0.0002797,
-      "loss": 0.1708,
+      "learning_rate": 0.0002796,
+      "loss": 0.1734,
       "step": 2708
     },
     {
       "epoch": 49.25,
-      "learning_rate": 0.0002796,
-      "loss": 0.1648,
+      "learning_rate": 0.0002795,
+      "loss": 0.1609,
       "step": 2709
     },
     {
       "epoch": 49.27,
-      "learning_rate": 0.0002795,
-      "loss": 0.1858,
+      "learning_rate": 0.00027939999999999996,
+      "loss": 0.1839,
       "step": 2710
     },
     {
       "epoch": 49.29,
-      "learning_rate": 0.00027939999999999996,
-      "loss": 0.1924,
+      "learning_rate": 0.0002793,
+      "loss": 0.1714,
       "step": 2711
     },
     {
       "epoch": 49.31,
-      "learning_rate": 0.0002793,
-      "loss": 0.2039,
+      "learning_rate": 0.0002792,
+      "loss": 0.2063,
       "step": 2712
     },
     {
       "epoch": 49.33,
-      "learning_rate": 0.0002792,
-      "loss": 0.2049,
+      "learning_rate": 0.0002791,
+      "loss": 0.2247,
       "step": 2713
     },
     {
       "epoch": 49.35,
-      "learning_rate": 0.0002791,
-      "loss": 0.2071,
+      "learning_rate": 0.000279,
+      "loss": 0.2189,
       "step": 2714
     },
     {
       "epoch": 49.36,
-      "learning_rate": 0.000279,
-      "loss": 0.1989,
+      "learning_rate": 0.0002789,
+      "loss": 0.2245,
       "step": 2715
     },
     {
       "epoch": 49.38,
-      "learning_rate": 0.0002789,
-      "loss": 0.1592,
+      "learning_rate": 0.0002788,
+      "loss": 0.1448,
       "step": 2716
     },
     {
       "epoch": 49.4,
-      "learning_rate": 0.0002788,
-      "loss": 0.2194,
+      "learning_rate": 0.0002787,
+      "loss": 0.2084,
       "step": 2717
     },
     {
       "epoch": 49.42,
-      "learning_rate": 0.0002787,
-      "loss": 0.1739,
+      "learning_rate": 0.0002786,
+      "loss": 0.1877,
       "step": 2718
     },
     {
       "epoch": 49.44,
-      "learning_rate": 0.0002786,
-      "loss": 0.153,
+      "learning_rate": 0.00027850000000000005,
+      "loss": 0.1432,
       "step": 2719
     },
     {
       "epoch": 49.45,
-      "learning_rate": 0.00027850000000000005,
-      "loss": 0.1671,
+      "learning_rate": 0.0002784,
+      "loss": 0.156,
       "step": 2720
     },
     {
       "epoch": 49.47,
-      "learning_rate": 0.0002784,
-      "loss": 0.1897,
+      "learning_rate": 0.0002783,
+      "loss": 0.1725,
       "step": 2721
     },
     {
       "epoch": 49.49,
-      "learning_rate": 0.0002783,
-      "loss": 0.2162,
+      "learning_rate": 0.0002782,
+      "loss": 0.1973,
       "step": 2722
     },
     {
       "epoch": 49.51,
-      "learning_rate": 0.0002782,
-      "loss": 0.1934,
+      "learning_rate": 0.00027810000000000004,
+      "loss": 0.1859,
       "step": 2723
     },
     {
       "epoch": 49.53,
-      "learning_rate": 0.00027810000000000004,
-      "loss": 0.1571,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 0.164,
       "step": 2724
     },
     {
       "epoch": 49.55,
-      "learning_rate": 0.00027800000000000004,
-      "loss": 0.2193,
+      "learning_rate": 0.0002779,
+      "loss": 0.2119,
       "step": 2725
     },
     {
       "epoch": 49.56,
-      "learning_rate": 0.0002779,
-      "loss": 0.1611,
+      "learning_rate": 0.0002778,
+      "loss": 0.1699,
       "step": 2726
     },
     {
       "epoch": 49.58,
-      "learning_rate": 0.0002778,
-      "loss": 0.2231,
+      "learning_rate": 0.00027770000000000003,
+      "loss": 0.2002,
       "step": 2727
     },
     {
       "epoch": 49.6,
-      "learning_rate": 0.00027770000000000003,
-      "loss": 0.1966,
+      "learning_rate": 0.0002776,
+      "loss": 0.1596,
       "step": 2728
     },
     {
       "epoch": 49.62,
-      "learning_rate": 0.0002776,
-      "loss": 0.2177,
+      "learning_rate": 0.0002775,
+      "loss": 0.2192,
       "step": 2729
     },
     {
       "epoch": 49.64,
-      "learning_rate": 0.0002775,
-      "loss": 0.2234,
+      "learning_rate": 0.00027739999999999997,
+      "loss": 0.1886,
       "step": 2730
     },
     {
       "epoch": 49.65,
-      "learning_rate": 0.00027739999999999997,
-      "loss": 0.1877,
+      "learning_rate": 0.0002773,
+      "loss": 0.1761,
       "step": 2731
     },
     {
       "epoch": 49.67,
-      "learning_rate": 0.0002773,
-      "loss": 0.1923,
+      "learning_rate": 0.0002772,
+      "loss": 0.168,
       "step": 2732
     },
     {
       "epoch": 49.69,
-      "learning_rate": 0.0002772,
-      "loss": 0.1891,
+      "learning_rate": 0.0002771,
+      "loss": 0.1932,
       "step": 2733
     },
     {
       "epoch": 49.71,
-      "learning_rate": 0.0002771,
-      "loss": 0.1737,
+      "learning_rate": 0.000277,
+      "loss": 0.1791,
       "step": 2734
     },
     {
       "epoch": 49.73,
-      "learning_rate": 0.000277,
-      "loss": 0.2455,
+      "learning_rate": 0.0002769,
+      "loss": 0.2422,
       "step": 2735
     },
     {
       "epoch": 49.75,
-      "learning_rate": 0.0002769,
-      "loss": 0.2131,
+      "learning_rate": 0.0002768,
+      "loss": 0.1615,
       "step": 2736
     },
     {
       "epoch": 49.76,
-      "learning_rate": 0.0002768,
-      "loss": 0.2125,
+      "learning_rate": 0.0002767,
+      "loss": 0.1706,
       "step": 2737
     },
     {
       "epoch": 49.78,
-      "learning_rate": 0.0002767,
-      "loss": 0.1807,
+      "learning_rate": 0.0002766,
+      "loss": 0.1474,
       "step": 2738
     },
     {
       "epoch": 49.8,
-      "learning_rate": 0.0002766,
-      "loss": 0.1943,
+      "learning_rate": 0.00027650000000000005,
+      "loss": 0.1991,
       "step": 2739
     },
     {
       "epoch": 49.82,
-      "learning_rate": 0.00027650000000000005,
-      "loss": 0.2333,
+      "learning_rate": 0.0002764,
+      "loss": 0.2304,
       "step": 2740
     },
     {
       "epoch": 49.84,
-      "learning_rate": 0.0002764,
-      "loss": 0.1938,
+      "learning_rate": 0.0002763,
+      "loss": 0.1817,
       "step": 2741
     },
     {
       "epoch": 49.85,
-      "learning_rate": 0.0002763,
-      "loss": 0.2229,
+      "learning_rate": 0.0002762,
+      "loss": 0.178,
       "step": 2742
     },
     {
       "epoch": 49.87,
-      "learning_rate": 0.0002762,
-      "loss": 0.1773,
+      "learning_rate": 0.00027610000000000004,
+      "loss": 0.1625,
       "step": 2743
     },
     {
       "epoch": 49.89,
-      "learning_rate": 0.00027610000000000004,
-      "loss": 0.2043,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 0.2022,
       "step": 2744
     },
     {
       "epoch": 49.91,
-      "learning_rate": 0.00027600000000000004,
-      "loss": 0.1791,
+      "learning_rate": 0.0002759,
+      "loss": 0.1769,
       "step": 2745
     },
     {
       "epoch": 49.93,
-      "learning_rate": 0.0002759,
-      "loss": 0.2273,
+      "learning_rate": 0.0002758,
+      "loss": 0.2156,
       "step": 2746
     },
     {
       "epoch": 49.95,
-      "learning_rate": 0.0002758,
-      "loss": 0.2387,
+      "learning_rate": 0.0002757,
+      "loss": 0.2215,
       "step": 2747
     },
     {
       "epoch": 49.96,
-      "learning_rate": 0.0002757,
-      "loss": 0.1761,
+      "learning_rate": 0.00027560000000000003,
+      "loss": 0.1931,
       "step": 2748
     },
     {
       "epoch": 49.98,
-      "learning_rate": 0.00027560000000000003,
-      "loss": 0.217,
+      "learning_rate": 0.00027550000000000003,
+      "loss": 0.1846,
       "step": 2749
     },
     {
       "epoch": 50.0,
-      "learning_rate": 0.00027550000000000003,
-      "loss": 0.1558,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 0.124,
       "step": 2750
     },
     {
       "epoch": 50.02,
-      "learning_rate": 0.00027539999999999997,
-      "loss": 0.1986,
+      "learning_rate": 0.00027529999999999997,
+      "loss": 0.1686,
       "step": 2751
     },
     {
       "epoch": 50.04,
-      "learning_rate": 0.00027529999999999997,
-      "loss": 0.1888,
+      "learning_rate": 0.0002752,
+      "loss": 0.1712,
       "step": 2752
     },
     {
       "epoch": 50.05,
-      "learning_rate": 0.0002752,
-      "loss": 0.1512,
+      "learning_rate": 0.0002751,
+      "loss": 0.1367,
       "step": 2753
     },
     {
       "epoch": 50.07,
-      "learning_rate": 0.0002751,
-      "loss": 0.1945,
+      "learning_rate": 0.000275,
+      "loss": 0.177,
       "step": 2754
     },
     {
       "epoch": 50.09,
-      "learning_rate": 0.000275,
-      "loss": 0.1741,
+      "learning_rate": 0.00027489999999999996,
+      "loss": 0.1899,
       "step": 2755
     },
     {
       "epoch": 50.11,
-      "learning_rate": 0.00027489999999999996,
-      "loss": 0.1985,
+      "learning_rate": 0.0002748,
+      "loss": 0.1693,
       "step": 2756
     },
     {
       "epoch": 50.13,
-      "learning_rate": 0.0002748,
-      "loss": 0.1832,
+      "learning_rate": 0.0002747,
+      "loss": 0.1726,
       "step": 2757
     },
     {
       "epoch": 50.15,
-      "learning_rate": 0.0002747,
-      "loss": 0.2221,
+      "learning_rate": 0.0002746,
+      "loss": 0.2017,
       "step": 2758
     },
     {
       "epoch": 50.16,
-      "learning_rate": 0.0002746,
-      "loss": 0.1564,
+      "learning_rate": 0.0002745,
+      "loss": 0.1458,
       "step": 2759
     },
     {
       "epoch": 50.18,
-      "learning_rate": 0.0002745,
-      "loss": 0.1996,
+      "learning_rate": 0.0002744,
+      "loss": 0.179,
       "step": 2760
     },
     {
       "epoch": 50.2,
-      "learning_rate": 0.0002744,
-      "loss": 0.2189,
+      "learning_rate": 0.0002743,
+      "loss": 0.1663,
       "step": 2761
     },
     {
       "epoch": 50.22,
-      "learning_rate": 0.0002743,
-      "loss": 0.194,
+      "learning_rate": 0.0002742,
+      "loss": 0.1932,
       "step": 2762
     },
     {
       "epoch": 50.24,
-      "learning_rate": 0.0002742,
-      "loss": 0.1992,
+      "learning_rate": 0.0002741,
+      "loss": 0.1778,
       "step": 2763
     },
     {
       "epoch": 50.25,
-      "learning_rate": 0.0002741,
-      "loss": 0.1767,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 0.1561,
       "step": 2764
     },
     {
       "epoch": 50.27,
-      "learning_rate": 0.00027400000000000005,
-      "loss": 0.2113,
+      "learning_rate": 0.0002739,
+      "loss": 0.1988,
       "step": 2765
     },
     {
       "epoch": 50.29,
-      "learning_rate": 0.0002739,
-      "loss": 0.1528,
+      "learning_rate": 0.0002738,
+      "loss": 0.1575,
       "step": 2766
     },
     {
       "epoch": 50.31,
-      "learning_rate": 0.0002738,
-      "loss": 0.159,
+      "learning_rate": 0.0002737,
+      "loss": 0.1631,
       "step": 2767
     },
     {
       "epoch": 50.33,
-      "learning_rate": 0.0002737,
-      "loss": 0.2048,
+      "learning_rate": 0.00027360000000000004,
+      "loss": 0.1733,
       "step": 2768
     },
     {
       "epoch": 50.35,
-      "learning_rate": 0.00027360000000000004,
-      "loss": 0.2166,
+      "learning_rate": 0.00027350000000000003,
+      "loss": 0.2027,
       "step": 2769
     },
     {
       "epoch": 50.36,
-      "learning_rate": 0.00027350000000000003,
-      "loss": 0.2012,
+      "learning_rate": 0.0002734,
+      "loss": 0.201,
       "step": 2770
     },
     {
       "epoch": 50.38,
-      "learning_rate": 0.0002734,
-      "loss": 0.1768,
+      "learning_rate": 0.0002733,
+      "loss": 0.1547,
       "step": 2771
     },
     {
       "epoch": 50.4,
-      "learning_rate": 0.0002733,
-      "loss": 0.246,
+      "learning_rate": 0.0002732,
+      "loss": 0.206,
       "step": 2772
     },
     {
       "epoch": 50.42,
-      "learning_rate": 0.0002732,
-      "loss": 0.1926,
+      "learning_rate": 0.0002731,
+      "loss": 0.148,
       "step": 2773
     },
     {
       "epoch": 50.44,
-      "learning_rate": 0.0002731,
-      "loss": 0.178,
+      "learning_rate": 0.000273,
+      "loss": 0.1975,
       "step": 2774
     },
     {
       "epoch": 50.45,
-      "learning_rate": 0.000273,
-      "loss": 0.1411,
+      "learning_rate": 0.00027289999999999997,
+      "loss": 0.1025,
       "step": 2775
     },
     {
       "epoch": 50.47,
-      "learning_rate": 0.00027289999999999997,
-      "loss": 0.2104,
+      "learning_rate": 0.0002728,
+      "loss": 0.1854,
       "step": 2776
     },
     {
       "epoch": 50.49,
-      "learning_rate": 0.0002728,
-      "loss": 0.1994,
+      "learning_rate": 0.0002727,
+      "loss": 0.1797,
       "step": 2777
     },
     {
       "epoch": 50.51,
-      "learning_rate": 0.0002727,
-      "loss": 0.2102,
+      "learning_rate": 0.0002726,
+      "loss": 0.213,
       "step": 2778
     },
     {
       "epoch": 50.53,
-      "learning_rate": 0.0002726,
-      "loss": 0.1906,
+      "learning_rate": 0.0002725,
+      "loss": 0.1621,
       "step": 2779
     },
     {
       "epoch": 50.55,
-      "learning_rate": 0.0002725,
-      "loss": 0.1623,
+      "learning_rate": 0.0002724,
+      "loss": 0.1724,
       "step": 2780
     },
     {
       "epoch": 50.56,
-      "learning_rate": 0.0002724,
-      "loss": 0.198,
+      "learning_rate": 0.0002723,
+      "loss": 0.1373,
       "step": 2781
     },
     {
       "epoch": 50.58,
-      "learning_rate": 0.0002723,
-      "loss": 0.2052,
+      "learning_rate": 0.0002722,
+      "loss": 0.1826,
       "step": 2782
     },
     {
       "epoch": 50.6,
-      "learning_rate": 0.0002722,
-      "loss": 0.2061,
+      "learning_rate": 0.0002721,
+      "loss": 0.2082,
       "step": 2783
     },
     {
       "epoch": 50.62,
-      "learning_rate": 0.0002721,
-      "loss": 0.2015,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 0.1764,
       "step": 2784
     },
     {
       "epoch": 50.64,
-      "learning_rate": 0.00027200000000000005,
-      "loss": 0.155,
+      "learning_rate": 0.0002719,
+      "loss": 0.1232,
       "step": 2785
     },
     {
       "epoch": 50.65,
-      "learning_rate": 0.0002719,
-      "loss": 0.1859,
+      "learning_rate": 0.0002718,
+      "loss": 0.1443,
       "step": 2786
     },
     {
       "epoch": 50.67,
-      "learning_rate": 0.0002718,
-      "loss": 0.1613,
+      "learning_rate": 0.0002717,
+      "loss": 0.1306,
       "step": 2787
     },
     {
       "epoch": 50.69,
-      "learning_rate": 0.0002717,
-      "loss": 0.1618,
+      "learning_rate": 0.0002716,
+      "loss": 0.1564,
       "step": 2788
     },
     {
       "epoch": 50.71,
-      "learning_rate": 0.0002716,
-      "loss": 0.1982,
+      "learning_rate": 0.00027150000000000004,
+      "loss": 0.173,
       "step": 2789
     },
     {
       "epoch": 50.73,
-      "learning_rate": 0.00027150000000000004,
-      "loss": 0.2118,
+      "learning_rate": 0.0002714,
+      "loss": 0.18,
       "step": 2790
     },
     {
       "epoch": 50.75,
-      "learning_rate": 0.0002714,
-      "loss": 0.1649,
+      "learning_rate": 0.0002713,
+      "loss": 0.1631,
       "step": 2791
     },
     {
       "epoch": 50.76,
-      "learning_rate": 0.0002713,
-      "loss": 0.1898,
+      "learning_rate": 0.0002712,
+      "loss": 0.1691,
       "step": 2792
     },
     {
       "epoch": 50.78,
-      "learning_rate": 0.0002712,
-      "loss": 0.1974,
+      "learning_rate": 0.00027110000000000003,
+      "loss": 0.1532,
       "step": 2793
     },
     {
       "epoch": 50.8,
-      "learning_rate": 0.00027110000000000003,
-      "loss": 0.1178,
+      "learning_rate": 0.00027100000000000003,
+      "loss": 0.0934,
       "step": 2794
     },
     {
       "epoch": 50.82,
-      "learning_rate": 0.00027100000000000003,
-      "loss": 0.2126,
+      "learning_rate": 0.00027089999999999997,
+      "loss": 0.2109,
       "step": 2795
     },
     {
       "epoch": 50.84,
-      "learning_rate": 0.00027089999999999997,
-      "loss": 0.1808,
+      "learning_rate": 0.00027079999999999997,
+      "loss": 0.1399,
       "step": 2796
     },
     {
       "epoch": 50.85,
-      "learning_rate": 0.00027079999999999997,
-      "loss": 0.1753,
+      "learning_rate": 0.0002707,
+      "loss": 0.1733,
       "step": 2797
     },
     {
       "epoch": 50.87,
-      "learning_rate": 0.0002707,
-      "loss": 0.2312,
+      "learning_rate": 0.0002706,
+      "loss": 0.2056,
       "step": 2798
     },
     {
       "epoch": 50.89,
-      "learning_rate": 0.0002706,
-      "loss": 0.1744,
+      "learning_rate": 0.0002705,
+      "loss": 0.1531,
       "step": 2799
     },
     {
       "epoch": 50.91,
-      "learning_rate": 0.0002705,
-      "loss": 0.1826,
+      "learning_rate": 0.00027039999999999996,
+      "loss": 0.1565,
       "step": 2800
     },
     {
       "epoch": 50.93,
-      "learning_rate": 0.00027039999999999996,
-      "loss": 0.1997,
+      "learning_rate": 0.0002703,
+      "loss": 0.1705,
       "step": 2801
     },
     {
       "epoch": 50.95,
-      "learning_rate": 0.0002703,
-      "loss": 0.188,
+      "learning_rate": 0.0002702,
+      "loss": 0.1812,
       "step": 2802
     },
     {
       "epoch": 50.96,
-      "learning_rate": 0.0002702,
-      "loss": 0.2143,
+      "learning_rate": 0.0002701,
+      "loss": 0.16,
       "step": 2803
     },
     {
       "epoch": 50.98,
-      "learning_rate": 0.0002701,
-      "loss": 0.1703,
+      "learning_rate": 0.00027,
+      "loss": 0.1568,
       "step": 2804
     },
     {
       "epoch": 51.0,
-      "learning_rate": 0.00027,
-      "loss": 0.1202,
+      "learning_rate": 0.0002699,
+      "loss": 0.1211,
       "step": 2805
     },
     {
       "epoch": 51.02,
-      "learning_rate": 0.0002699,
-      "loss": 0.1537,
+      "learning_rate": 0.0002698,
+      "loss": 0.1495,
       "step": 2806
     },
     {
       "epoch": 51.04,
-      "learning_rate": 0.0002698,
-      "loss": 0.1447,
+      "learning_rate": 0.0002697,
+      "loss": 0.1493,
       "step": 2807
     },
     {
       "epoch": 51.05,
-      "learning_rate": 0.0002697,
-      "loss": 0.1417,
+      "learning_rate": 0.0002696,
+      "loss": 0.1552,
       "step": 2808
     },
     {
       "epoch": 51.07,
-      "learning_rate": 0.0002696,
-      "loss": 0.1664,
+      "learning_rate": 0.00026950000000000005,
+      "loss": 0.1519,
       "step": 2809
     },
     {
       "epoch": 51.09,
-      "learning_rate": 0.00026950000000000005,
-      "loss": 0.2352,
+      "learning_rate": 0.0002694,
+      "loss": 0.1995,
       "step": 2810
     },
     {
       "epoch": 51.11,
-      "learning_rate": 0.0002694,
-      "loss": 0.1927,
+      "learning_rate": 0.0002693,
+      "loss": 0.1832,
       "step": 2811
     },
     {
       "epoch": 51.13,
-      "learning_rate": 0.0002693,
-      "loss": 0.1627,
+      "learning_rate": 0.0002692,
+      "loss": 0.1528,
       "step": 2812
     },
     {
       "epoch": 51.15,
-      "learning_rate": 0.0002692,
-      "loss": 0.2076,
+      "learning_rate": 0.00026910000000000004,
+      "loss": 0.1894,
       "step": 2813
     },
     {
       "epoch": 51.16,
-      "learning_rate": 0.00026910000000000004,
-      "loss": 0.1671,
+      "learning_rate": 0.00026900000000000003,
+      "loss": 0.1502,
       "step": 2814
     },
     {
       "epoch": 51.18,
-      "learning_rate": 0.00026900000000000003,
-      "loss": 0.1987,
+      "learning_rate": 0.0002689,
+      "loss": 0.1704,
       "step": 2815
     },
     {
       "epoch": 51.2,
-      "learning_rate": 0.0002689,
-      "loss": 0.1478,
+      "learning_rate": 0.0002688,
+      "loss": 0.1581,
       "step": 2816
     },
     {
       "epoch": 51.22,
-      "learning_rate": 0.0002688,
-      "loss": 0.1821,
+      "learning_rate": 0.0002687,
+      "loss": 0.1543,
       "step": 2817
     },
     {
       "epoch": 51.24,
-      "learning_rate": 0.0002687,
-      "loss": 0.1714,
+      "learning_rate": 0.0002686,
+      "loss": 0.1562,
       "step": 2818
     },
     {
       "epoch": 51.25,
-      "learning_rate": 0.0002686,
-      "loss": 0.1514,
+      "learning_rate": 0.0002685,
+      "loss": 0.1513,
       "step": 2819
     },
     {
       "epoch": 51.27,
-      "learning_rate": 0.0002685,
-      "loss": 0.1539,
+      "learning_rate": 0.0002684,
+      "loss": 0.1745,
       "step": 2820
     },
     {
       "epoch": 51.29,
-      "learning_rate": 0.0002684,
-      "loss": 0.17,
+      "learning_rate": 0.00026829999999999996,
+      "loss": 0.1892,
       "step": 2821
     },
     {
       "epoch": 51.31,
-      "learning_rate": 0.00026829999999999996,
-      "loss": 0.1562,
+      "learning_rate": 0.0002682,
+      "loss": 0.1613,
       "step": 2822
     },
     {
       "epoch": 51.33,
-      "learning_rate": 0.0002682,
-      "loss": 0.2135,
+      "learning_rate": 0.0002681,
+      "loss": 0.1799,
       "step": 2823
     },
     {
       "epoch": 51.35,
-      "learning_rate": 0.0002681,
-      "loss": 0.202,
+      "learning_rate": 0.000268,
+      "loss": 0.1631,
       "step": 2824
     },
     {
       "epoch": 51.36,
-      "learning_rate": 0.000268,
-      "loss": 0.1835,
+      "learning_rate": 0.0002679,
+      "loss": 0.1641,
       "step": 2825
     },
     {
       "epoch": 51.38,
-      "learning_rate": 0.0002679,
-      "loss": 0.124,
+      "learning_rate": 0.0002678,
+      "loss": 0.114,
       "step": 2826
     },
     {
       "epoch": 51.4,
-      "learning_rate": 0.0002678,
-      "loss": 0.1599,
+      "learning_rate": 0.0002677,
+      "loss": 0.1352,
       "step": 2827
     },
     {
       "epoch": 51.42,
-      "learning_rate": 0.0002677,
-      "loss": 0.1963,
+      "learning_rate": 0.0002676,
+      "loss": 0.2001,
       "step": 2828
     },
     {
       "epoch": 51.44,
-      "learning_rate": 0.0002676,
-      "loss": 0.2131,
+      "learning_rate": 0.0002675,
+      "loss": 0.2174,
       "step": 2829
     },
     {
       "epoch": 51.45,
-      "learning_rate": 0.0002675,
-      "loss": 0.1503,
+      "learning_rate": 0.00026740000000000005,
+      "loss": 0.1188,
       "step": 2830
     },
     {
       "epoch": 51.47,
-      "learning_rate": 0.00026740000000000005,
-      "loss": 0.1563,
+      "learning_rate": 0.0002673,
+      "loss": 0.1514,
       "step": 2831
     },
     {
       "epoch": 51.49,
-      "learning_rate": 0.0002673,
-      "loss": 0.1495,
+      "learning_rate": 0.0002672,
+      "loss": 0.1614,
       "step": 2832
     },
     {
       "epoch": 51.51,
-      "learning_rate": 0.0002672,
-      "loss": 0.1892,
+      "learning_rate": 0.0002671,
+      "loss": 0.1693,
       "step": 2833
     },
     {
       "epoch": 51.53,
-      "learning_rate": 0.0002671,
-      "loss": 0.1592,
+      "learning_rate": 0.00026700000000000004,
+      "loss": 0.1278,
       "step": 2834
     },
     {
       "epoch": 51.55,
-      "learning_rate": 0.00026700000000000004,
-      "loss": 0.192,
+      "learning_rate": 0.00026690000000000004,
+      "loss": 0.1696,
       "step": 2835
     },
     {
       "epoch": 51.56,
-      "learning_rate": 0.00026690000000000004,
-      "loss": 0.2008,
+      "learning_rate": 0.0002668,
+      "loss": 0.1729,
       "step": 2836
     },
     {
       "epoch": 51.58,
-      "learning_rate": 0.0002668,
-      "loss": 0.2115,
+      "learning_rate": 0.0002667,
+      "loss": 0.1623,
       "step": 2837
     },
     {
       "epoch": 51.6,
-      "learning_rate": 0.0002667,
-      "loss": 0.1918,
+      "learning_rate": 0.00026660000000000003,
+      "loss": 0.1794,
       "step": 2838
     },
     {
       "epoch": 51.62,
-      "learning_rate": 0.00026660000000000003,
-      "loss": 0.1615,
+      "learning_rate": 0.0002665,
+      "loss": 0.1598,
       "step": 2839
     },
     {
       "epoch": 51.64,
-      "learning_rate": 0.0002665,
-      "loss": 0.209,
+      "learning_rate": 0.0002664,
+      "loss": 0.1715,
       "step": 2840
     },
     {
       "epoch": 51.65,
-      "learning_rate": 0.0002664,
-      "loss": 0.1602,
+      "learning_rate": 0.00026629999999999997,
+      "loss": 0.1175,
       "step": 2841
     },
     {
       "epoch": 51.67,
-      "learning_rate": 0.00026629999999999997,
-      "loss": 0.1678,
+      "learning_rate": 0.0002662,
+      "loss": 0.1835,
       "step": 2842
     },
     {
       "epoch": 51.69,
-      "learning_rate": 0.0002662,
-      "loss": 0.1858,
+      "learning_rate": 0.0002661,
+      "loss": 0.1764,
       "step": 2843
     },
     {
       "epoch": 51.71,
-      "learning_rate": 0.0002661,
-      "loss": 0.1973,
+      "learning_rate": 0.000266,
+      "loss": 0.189,
       "step": 2844
     },
     {
       "epoch": 51.73,
-      "learning_rate": 0.000266,
-      "loss": 0.1966,
+      "learning_rate": 0.0002659,
+      "loss": 0.1816,
       "step": 2845
     },
     {
       "epoch": 51.75,
-      "learning_rate": 0.0002659,
-      "loss": 0.1649,
+      "learning_rate": 0.0002658,
+      "loss": 0.1686,
       "step": 2846
     },
     {
       "epoch": 51.76,
-      "learning_rate": 0.0002658,
-      "loss": 0.2215,
+      "learning_rate": 0.0002657,
+      "loss": 0.2155,
       "step": 2847
     },
     {
       "epoch": 51.78,
-      "learning_rate": 0.0002657,
-      "loss": 0.1927,
+      "learning_rate": 0.0002656,
+      "loss": 0.1722,
       "step": 2848
     },
     {
       "epoch": 51.8,
-      "learning_rate": 0.0002656,
-      "loss": 0.1531,
+      "learning_rate": 0.0002655,
+      "loss": 0.1392,
       "step": 2849
     },
     {
       "epoch": 51.82,
-      "learning_rate": 0.0002655,
-      "loss": 0.1976,
+      "learning_rate": 0.00026540000000000005,
+      "loss": 0.1542,
       "step": 2850
     },
     {
       "epoch": 51.84,
-      "learning_rate": 0.00026540000000000005,
-      "loss": 0.1897,
+      "learning_rate": 0.0002653,
+      "loss": 0.1939,
       "step": 2851
     },
     {
       "epoch": 51.85,
-      "learning_rate": 0.0002653,
-      "loss": 0.1754,
+      "learning_rate": 0.0002652,
+      "loss": 0.171,
       "step": 2852
     },
     {
       "epoch": 51.87,
-      "learning_rate": 0.0002652,
-      "loss": 0.1561,
+      "learning_rate": 0.0002651,
+      "loss": 0.1288,
       "step": 2853
     },
     {
       "epoch": 51.89,
-      "learning_rate": 0.0002651,
-      "loss": 0.1865,
+      "learning_rate": 0.00026500000000000004,
+      "loss": 0.2024,
       "step": 2854
     },
     {
       "epoch": 51.91,
-      "learning_rate": 0.00026500000000000004,
-      "loss": 0.1869,
+      "learning_rate": 0.00026490000000000004,
+      "loss": 0.1632,
       "step": 2855
     },
     {
       "epoch": 51.93,
-      "learning_rate": 0.00026490000000000004,
-      "loss": 0.2051,
+      "learning_rate": 0.0002648,
+      "loss": 0.1923,
       "step": 2856
     },
     {
       "epoch": 51.95,
-      "learning_rate": 0.0002648,
-      "loss": 0.1658,
+      "learning_rate": 0.0002647,
+      "loss": 0.1468,
       "step": 2857
     },
     {
       "epoch": 51.96,
-      "learning_rate": 0.0002647,
-      "loss": 0.2112,
+      "learning_rate": 0.0002646,
+      "loss": 0.2033,
       "step": 2858
     },
     {
       "epoch": 51.98,
-      "learning_rate": 0.0002646,
-      "loss": 0.1618,
+      "learning_rate": 0.00026450000000000003,
+      "loss": 0.1535,
       "step": 2859
     },
     {
       "epoch": 52.0,
-      "learning_rate": 0.00026450000000000003,
-      "loss": 0.1394,
+      "learning_rate": 0.00026440000000000003,
+      "loss": 0.2268,
       "step": 2860
     },
     {
       "epoch": 52.02,
-      "learning_rate": 0.00026440000000000003,
-      "loss": 0.166,
+      "learning_rate": 0.0002643,
+      "loss": 0.1728,
       "step": 2861
     },
     {
       "epoch": 52.04,
-      "learning_rate": 0.0002643,
-      "loss": 0.126,
+      "learning_rate": 0.00026419999999999997,
+      "loss": 0.1626,
       "step": 2862
     },
     {
       "epoch": 52.05,
-      "learning_rate": 0.00026419999999999997,
-      "loss": 0.2381,
+      "learning_rate": 0.0002641,
+      "loss": 0.1959,
       "step": 2863
     },
     {
       "epoch": 52.07,
-      "learning_rate": 0.0002641,
-      "loss": 0.1964,
+      "learning_rate": 0.000264,
+      "loss": 0.197,
       "step": 2864
     },
     {
       "epoch": 52.09,
-      "learning_rate": 0.000264,
-      "loss": 0.1772,
+      "learning_rate": 0.0002639,
+      "loss": 0.1848,
       "step": 2865
     },
     {
       "epoch": 52.11,
-      "learning_rate": 0.0002639,
-      "loss": 0.173,
+      "learning_rate": 0.00026379999999999996,
+      "loss": 0.1777,
       "step": 2866
     },
     {
       "epoch": 52.13,
-      "learning_rate": 0.00026379999999999996,
-      "loss": 0.1859,
+      "learning_rate": 0.0002637,
+      "loss": 0.1627,
       "step": 2867
     },
     {
       "epoch": 52.15,
-      "learning_rate": 0.0002637,
-      "loss": 0.1949,
+      "learning_rate": 0.0002636,
+      "loss": 0.1698,
       "step": 2868
     },
     {
       "epoch": 52.16,
-      "learning_rate": 0.0002636,
-      "loss": 0.1892,
+      "learning_rate": 0.0002635,
+      "loss": 0.1394,
       "step": 2869
     },
     {
       "epoch": 52.18,
-      "learning_rate": 0.0002635,
-      "loss": 0.1887,
+      "learning_rate": 0.0002634,
+      "loss": 0.2012,
       "step": 2870
     },
     {
       "epoch": 52.2,
-      "learning_rate": 0.0002634,
-      "loss": 0.1445,
+      "learning_rate": 0.0002633,
+      "loss": 0.1539,
       "step": 2871
     },
     {
       "epoch": 52.22,
-      "learning_rate": 0.0002633,
-      "loss": 0.1671,
+      "learning_rate": 0.0002632,
+      "loss": 0.1818,
       "step": 2872
     },
     {
       "epoch": 52.24,
-      "learning_rate": 0.0002632,
-      "loss": 0.1565,
+      "learning_rate": 0.0002631,
+      "loss": 0.1431,
       "step": 2873
     },
     {
       "epoch": 52.25,
-      "learning_rate": 0.0002631,
-      "loss": 0.2129,
+      "learning_rate": 0.000263,
+      "loss": 0.2193,
       "step": 2874
     },
     {
       "epoch": 52.27,
-      "learning_rate": 0.000263,
-      "loss": 0.2102,
+      "learning_rate": 0.00026290000000000005,
+      "loss": 0.2141,
       "step": 2875
     },
     {
       "epoch": 52.29,
-      "learning_rate": 0.00026290000000000005,
-      "loss": 0.2357,
+      "learning_rate": 0.0002628,
+      "loss": 0.2115,
       "step": 2876
     },
     {
       "epoch": 52.31,
-      "learning_rate": 0.0002628,
-      "loss": 0.1949,
+      "learning_rate": 0.0002627,
+      "loss": 0.182,
       "step": 2877
     },
     {
       "epoch": 52.33,
-      "learning_rate": 0.0002627,
-      "loss": 0.1743,
+      "learning_rate": 0.0002626,
+      "loss": 0.1608,
       "step": 2878
     },
     {
       "epoch": 52.35,
-      "learning_rate": 0.0002626,
-      "loss": 0.1671,
+      "learning_rate": 0.00026250000000000004,
+      "loss": 0.1567,
       "step": 2879
     },
     {
       "epoch": 52.36,
-      "learning_rate": 0.00026250000000000004,
-      "loss": 0.2162,
+      "learning_rate": 0.00026240000000000004,
+      "loss": 0.2167,
       "step": 2880
     },
     {
       "epoch": 52.38,
-      "learning_rate": 0.00026240000000000004,
-      "loss": 0.1422,
+      "learning_rate": 0.0002623,
+      "loss": 0.128,
       "step": 2881
     },
     {
       "epoch": 52.4,
-      "learning_rate": 0.0002623,
-      "loss": 0.199,
+      "learning_rate": 0.0002622,
+      "loss": 0.1836,
       "step": 2882
     },
     {
       "epoch": 52.42,
-      "learning_rate": 0.0002622,
-      "loss": 0.1564,
+      "learning_rate": 0.00026210000000000003,
+      "loss": 0.1465,
       "step": 2883
     },
     {
       "epoch": 52.44,
-      "learning_rate": 0.00026210000000000003,
-      "loss": 0.167,
+      "learning_rate": 0.000262,
+      "loss": 0.1666,
       "step": 2884
     },
     {
       "epoch": 52.45,
-      "learning_rate": 0.000262,
-      "loss": 0.2053,
+      "learning_rate": 0.0002619,
+      "loss": 0.1864,
       "step": 2885
     },
     {
       "epoch": 52.47,
-      "learning_rate": 0.0002619,
-      "loss": 0.215,
+      "learning_rate": 0.00026179999999999997,
+      "loss": 0.1933,
       "step": 2886
     },
     {
       "epoch": 52.49,
-      "learning_rate": 0.00026179999999999997,
-      "loss": 0.2028,
+      "learning_rate": 0.0002617,
+      "loss": 0.1629,
       "step": 2887
     },
     {
       "epoch": 52.51,
-      "learning_rate": 0.0002617,
-      "loss": 0.1966,
+      "learning_rate": 0.0002616,
+      "loss": 0.1675,
       "step": 2888
     },
     {
       "epoch": 52.53,
-      "learning_rate": 0.0002616,
-      "loss": 0.1629,
+      "learning_rate": 0.0002615,
+      "loss": 0.1417,
       "step": 2889
     },
     {
       "epoch": 52.55,
-      "learning_rate": 0.0002615,
-      "loss": 0.2043,
+      "learning_rate": 0.0002614,
+      "loss": 0.192,
       "step": 2890
     },
     {
       "epoch": 52.56,
-      "learning_rate": 0.0002614,
-      "loss": 0.2005,
+      "learning_rate": 0.0002613,
+      "loss": 0.1661,
       "step": 2891
     },
     {
       "epoch": 52.58,
-      "learning_rate": 0.0002613,
-      "loss": 0.1916,
+      "learning_rate": 0.0002612,
+      "loss": 0.1642,
       "step": 2892
     },
     {
       "epoch": 52.6,
-      "learning_rate": 0.0002612,
-      "loss": 0.1766,
+      "learning_rate": 0.0002611,
+      "loss": 0.1858,
       "step": 2893
     },
     {
       "epoch": 52.62,
-      "learning_rate": 0.0002611,
-      "loss": 0.2084,
+      "learning_rate": 0.000261,
+      "loss": 0.1716,
       "step": 2894
     },
     {
       "epoch": 52.64,
-      "learning_rate": 0.000261,
-      "loss": 0.1962,
+      "learning_rate": 0.00026090000000000005,
+      "loss": 0.1861,
       "step": 2895
     },
     {
       "epoch": 52.65,
-      "learning_rate": 0.00026090000000000005,
-      "loss": 0.1681,
+      "learning_rate": 0.0002608,
+      "loss": 0.1501,
       "step": 2896
     },
     {
       "epoch": 52.67,
-      "learning_rate": 0.0002608,
-      "loss": 0.1909,
+      "learning_rate": 0.0002607,
+      "loss": 0.1667,
       "step": 2897
     },
     {
       "epoch": 52.69,
-      "learning_rate": 0.0002607,
-      "loss": 0.1835,
+      "learning_rate": 0.0002606,
+      "loss": 0.1637,
       "step": 2898
     },
     {
       "epoch": 52.71,
-      "learning_rate": 0.0002606,
-      "loss": 0.2006,
+      "learning_rate": 0.0002605,
+      "loss": 0.1677,
       "step": 2899
     },
     {
       "epoch": 52.73,
-      "learning_rate": 0.0002605,
-      "loss": 0.163,
+      "learning_rate": 0.00026040000000000004,
+      "loss": 0.1743,
       "step": 2900
     },
     {
       "epoch": 52.75,
-      "learning_rate": 0.00026040000000000004,
-      "loss": 0.1825,
+      "learning_rate": 0.0002603,
+      "loss": 0.1669,
       "step": 2901
     },
     {
       "epoch": 52.76,
-      "learning_rate": 0.0002603,
-      "loss": 0.1896,
+      "learning_rate": 0.0002602,
+      "loss": 0.1759,
       "step": 2902
     },
     {
       "epoch": 52.78,
-      "learning_rate": 0.0002602,
-      "loss": 0.1659,
+      "learning_rate": 0.0002601,
+      "loss": 0.165,
       "step": 2903
     },
     {
       "epoch": 52.8,
-      "learning_rate": 0.0002601,
-      "loss": 0.1862,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 0.1424,
       "step": 2904
     },
     {
       "epoch": 52.82,
-      "learning_rate": 0.00026000000000000003,
-      "loss": 0.193,
+      "learning_rate": 0.00025990000000000003,
+      "loss": 0.1841,
       "step": 2905
     },
     {
       "epoch": 52.84,
-      "learning_rate": 0.00025990000000000003,
-      "loss": 0.1609,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.1474,
       "step": 2906
     },
     {
       "epoch": 52.85,
-      "learning_rate": 0.00025979999999999997,
-      "loss": 0.2062,
+      "learning_rate": 0.00025969999999999997,
+      "loss": 0.1784,
       "step": 2907
     },
     {
       "epoch": 52.87,
-      "learning_rate": 0.00025969999999999997,
-      "loss": 0.164,
+      "learning_rate": 0.0002596,
+      "loss": 0.1661,
       "step": 2908
     },
     {
       "epoch": 52.89,
-      "learning_rate": 0.0002596,
-      "loss": 0.1949,
+      "learning_rate": 0.0002595,
+      "loss": 0.171,
       "step": 2909
     },
     {
       "epoch": 52.91,
-      "learning_rate": 0.0002595,
-      "loss": 0.1972,
+      "learning_rate": 0.0002594,
+      "loss": 0.1911,
       "step": 2910
     },
     {
       "epoch": 52.93,
-      "learning_rate": 0.0002594,
-      "loss": 0.1983,
+      "learning_rate": 0.00025929999999999996,
+      "loss": 0.1681,
       "step": 2911
     },
     {
       "epoch": 52.95,
-      "learning_rate": 0.00025929999999999996,
-      "loss": 0.1942,
+      "learning_rate": 0.0002592,
+      "loss": 0.1706,
       "step": 2912
     },
     {
       "epoch": 52.96,
-      "learning_rate": 0.0002592,
-      "loss": 0.1963,
+      "learning_rate": 0.0002591,
+      "loss": 0.2091,
       "step": 2913
     },
     {
       "epoch": 52.98,
-      "learning_rate": 0.0002591,
-      "loss": 0.1733,
+      "learning_rate": 0.000259,
+      "loss": 0.1523,
       "step": 2914
     },
     {
       "epoch": 53.0,
-      "learning_rate": 0.000259,
-      "loss": 0.1198,
+      "learning_rate": 0.0002589,
+      "loss": 0.1279,
       "step": 2915
     },
     {
       "epoch": 53.02,
-      "learning_rate": 0.0002589,
-      "loss": 0.2175,
+      "learning_rate": 0.0002588,
+      "loss": 0.1777,
       "step": 2916
     },
     {
       "epoch": 53.04,
-      "learning_rate": 0.0002588,
-      "loss": 0.2022,
+      "learning_rate": 0.0002587,
+      "loss": 0.1861,
       "step": 2917
     },
     {
       "epoch": 53.05,
-      "learning_rate": 0.0002587,
-      "loss": 0.1788,
+      "learning_rate": 0.0002586,
+      "loss": 0.147,
       "step": 2918
     },
     {
       "epoch": 53.07,
-      "learning_rate": 0.0002586,
-      "loss": 0.254,
+      "learning_rate": 0.0002585,
+      "loss": 0.2279,
       "step": 2919
     },
     {
       "epoch": 53.09,
-      "learning_rate": 0.0002585,
-      "loss": 0.1947,
+      "learning_rate": 0.00025840000000000005,
+      "loss": 0.1622,
       "step": 2920
     },
     {
       "epoch": 53.11,
-      "learning_rate": 0.00025840000000000005,
-      "loss": 0.1309,
+      "learning_rate": 0.0002583,
+      "loss": 0.1389,
       "step": 2921
     },
     {
       "epoch": 53.13,
-      "learning_rate": 0.0002583,
-      "loss": 0.1848,
+      "learning_rate": 0.0002582,
+      "loss": 0.184,
       "step": 2922
     },
     {
       "epoch": 53.15,
-      "learning_rate": 0.0002582,
-      "loss": 0.1604,
+      "learning_rate": 0.0002581,
+      "loss": 0.1598,
       "step": 2923
     },
     {
       "epoch": 53.16,
-      "learning_rate": 0.0002581,
-      "loss": 0.1723,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 0.151,
       "step": 2924
     },
     {
       "epoch": 53.18,
-      "learning_rate": 0.00025800000000000004,
-      "loss": 0.1534,
+      "learning_rate": 0.00025790000000000003,
+      "loss": 0.1425,
       "step": 2925
     },
     {
       "epoch": 53.2,
-      "learning_rate": 0.00025790000000000003,
-      "loss": 0.1378,
+      "learning_rate": 0.0002578,
+      "loss": 0.1266,
       "step": 2926
     },
     {
       "epoch": 53.22,
-      "learning_rate": 0.0002578,
-      "loss": 0.162,
+      "learning_rate": 0.0002577,
+      "loss": 0.1652,
       "step": 2927
     },
     {
       "epoch": 53.24,
-      "learning_rate": 0.0002577,
-      "loss": 0.1463,
+      "learning_rate": 0.00025760000000000003,
+      "loss": 0.1362,
       "step": 2928
     },
     {
       "epoch": 53.25,
-      "learning_rate": 0.00025760000000000003,
-      "loss": 0.1764,
+      "learning_rate": 0.0002575,
+      "loss": 0.1445,
       "step": 2929
     },
     {
       "epoch": 53.27,
-      "learning_rate": 0.0002575,
-      "loss": 0.1954,
+      "learning_rate": 0.0002574,
+      "loss": 0.1825,
       "step": 2930
     },
     {
       "epoch": 53.29,
-      "learning_rate": 0.0002574,
-      "loss": 0.1851,
+      "learning_rate": 0.00025729999999999997,
+      "loss": 0.1752,
       "step": 2931
     },
     {
       "epoch": 53.31,
-      "learning_rate": 0.00025729999999999997,
-      "loss": 0.1521,
+      "learning_rate": 0.00025719999999999996,
+      "loss": 0.1649,
       "step": 2932
     },
     {
       "epoch": 53.33,
-      "learning_rate": 0.00025719999999999996,
-      "loss": 0.1735,
+      "learning_rate": 0.0002571,
+      "loss": 0.1789,
       "step": 2933
     },
     {
       "epoch": 53.35,
-      "learning_rate": 0.0002571,
-      "loss": 0.1129,
+      "learning_rate": 0.000257,
+      "loss": 0.1229,
       "step": 2934
     },
     {
       "epoch": 53.36,
-      "learning_rate": 0.000257,
-      "loss": 0.1627,
+      "learning_rate": 0.0002569,
+      "loss": 0.1457,
       "step": 2935
     },
     {
       "epoch": 53.38,
-      "learning_rate": 0.0002569,
-      "loss": 0.2237,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.2047,
       "step": 2936
     },
     {
       "epoch": 53.4,
-      "learning_rate": 0.00025679999999999995,
-      "loss": 0.1974,
+      "learning_rate": 0.0002567,
+      "loss": 0.1827,
       "step": 2937
     },
     {
       "epoch": 53.42,
-      "learning_rate": 0.0002567,
-      "loss": 0.1491,
+      "learning_rate": 0.0002566,
+      "loss": 0.1385,
       "step": 2938
     },
     {
       "epoch": 53.44,
-      "learning_rate": 0.0002566,
-      "loss": 0.2328,
+      "learning_rate": 0.0002565,
+      "loss": 0.1789,
       "step": 2939
     },
     {
       "epoch": 53.45,
-      "learning_rate": 0.0002565,
-      "loss": 0.1695,
+      "learning_rate": 0.0002564,
+      "loss": 0.1605,
       "step": 2940
     },
     {
       "epoch": 53.47,
-      "learning_rate": 0.0002564,
-      "loss": 0.2211,
+      "learning_rate": 0.0002563,
+      "loss": 0.2231,
       "step": 2941
     },
     {
       "epoch": 53.49,
-      "learning_rate": 0.0002563,
-      "loss": 0.2112,
+      "learning_rate": 0.0002562,
+      "loss": 0.2234,
       "step": 2942
     },
     {
       "epoch": 53.51,
-      "learning_rate": 0.0002562,
-      "loss": 0.1662,
+      "learning_rate": 0.0002561,
+      "loss": 0.1553,
       "step": 2943
     },
     {
       "epoch": 53.53,
-      "learning_rate": 0.0002561,
-      "loss": 0.1874,
+      "learning_rate": 0.000256,
+      "loss": 0.1719,
       "step": 2944
     },
     {
       "epoch": 53.55,
-      "learning_rate": 0.000256,
-      "loss": 0.2083,
+      "learning_rate": 0.00025590000000000004,
+      "loss": 0.1941,
       "step": 2945
     },
     {
       "epoch": 53.56,
-      "learning_rate": 0.00025590000000000004,
-      "loss": 0.2341,
+      "learning_rate": 0.00025580000000000004,
+      "loss": 0.2297,
       "step": 2946
     },
     {
       "epoch": 53.58,
-      "learning_rate": 0.00025580000000000004,
-      "loss": 0.2234,
+      "learning_rate": 0.0002557,
+      "loss": 0.1776,
       "step": 2947
     },
     {
       "epoch": 53.6,
-      "learning_rate": 0.0002557,
-      "loss": 0.201,
+      "learning_rate": 0.0002556,
+      "loss": 0.1861,
       "step": 2948
     },
     {
       "epoch": 53.62,
-      "learning_rate": 0.0002556,
-      "loss": 0.2116,
+      "learning_rate": 0.00025550000000000003,
+      "loss": 0.2459,
       "step": 2949
     },
     {
       "epoch": 53.64,
-      "learning_rate": 0.00025550000000000003,
-      "loss": 0.2128,
+      "learning_rate": 0.00025540000000000003,
+      "loss": 0.1791,
       "step": 2950
     },
     {
       "epoch": 53.65,
-      "learning_rate": 0.00025540000000000003,
-      "loss": 0.2004,
+      "learning_rate": 0.0002553,
+      "loss": 0.1933,
       "step": 2951
     },
     {
       "epoch": 53.67,
-      "learning_rate": 0.0002553,
-      "loss": 0.1486,
+      "learning_rate": 0.00025519999999999997,
+      "loss": 0.1417,
       "step": 2952
     },
     {
       "epoch": 53.69,
-      "learning_rate": 0.00025519999999999997,
-      "loss": 0.1484,
+      "learning_rate": 0.0002551,
+      "loss": 0.1679,
       "step": 2953
     },
     {
       "epoch": 53.71,
-      "learning_rate": 0.0002551,
-      "loss": 0.2004,
+      "learning_rate": 0.000255,
+      "loss": 0.2116,
       "step": 2954
     },
     {
       "epoch": 53.73,
-      "learning_rate": 0.000255,
-      "loss": 0.205,
+      "learning_rate": 0.0002549,
+      "loss": 0.1636,
       "step": 2955
     },
     {
       "epoch": 53.75,
-      "learning_rate": 0.0002549,
-      "loss": 0.1774,
+      "learning_rate": 0.0002548,
+      "loss": 0.1829,
       "step": 2956
     },
     {
       "epoch": 53.76,
-      "learning_rate": 0.0002548,
-      "loss": 0.1941,
+      "learning_rate": 0.0002547,
+      "loss": 0.1689,
       "step": 2957
     },
     {
       "epoch": 53.78,
-      "learning_rate": 0.0002547,
-      "loss": 0.1459,
+      "learning_rate": 0.0002546,
+      "loss": 0.1708,
       "step": 2958
     },
     {
       "epoch": 53.8,
-      "learning_rate": 0.0002546,
-      "loss": 0.1739,
+      "learning_rate": 0.0002545,
+      "loss": 0.1684,
       "step": 2959
     },
     {
       "epoch": 53.82,
-      "learning_rate": 0.0002545,
-      "loss": 0.175,
+      "learning_rate": 0.0002544,
+      "loss": 0.1863,
       "step": 2960
     },
     {
       "epoch": 53.84,
-      "learning_rate": 0.0002544,
-      "loss": 0.1465,
+      "learning_rate": 0.00025430000000000006,
+      "loss": 0.1487,
       "step": 2961
     },
     {
       "epoch": 53.85,
-      "learning_rate": 0.00025430000000000006,
-      "loss": 0.1199,
+      "learning_rate": 0.0002542,
+      "loss": 0.1196,
       "step": 2962
     },
     {
       "epoch": 53.87,
-      "learning_rate": 0.0002542,
-      "loss": 0.1924,
+      "learning_rate": 0.0002541,
+      "loss": 0.1759,
       "step": 2963
     },
     {
       "epoch": 53.89,
-      "learning_rate": 0.0002541,
-      "loss": 0.1761,
+      "learning_rate": 0.000254,
+      "loss": 0.1882,
       "step": 2964
     },
     {
       "epoch": 53.91,
-      "learning_rate": 0.000254,
-      "loss": 0.18,
+      "learning_rate": 0.00025390000000000005,
+      "loss": 0.1578,
       "step": 2965
     },
     {
       "epoch": 53.93,
-      "learning_rate": 0.00025390000000000005,
-      "loss": 0.1901,
+      "learning_rate": 0.00025380000000000004,
+      "loss": 0.2155,
       "step": 2966
     },
     {
       "epoch": 53.95,
-      "learning_rate": 0.00025380000000000004,
-      "loss": 0.1746,
+      "learning_rate": 0.0002537,
+      "loss": 0.1731,
       "step": 2967
     },
     {
       "epoch": 53.96,
-      "learning_rate": 0.0002537,
-      "loss": 0.1568,
+      "learning_rate": 0.0002536,
+      "loss": 0.1483,
       "step": 2968
     },
     {
       "epoch": 53.98,
-      "learning_rate": 0.0002536,
-      "loss": 0.1691,
+      "learning_rate": 0.0002535,
+      "loss": 0.1581,
       "step": 2969
     },
     {
       "epoch": 54.0,
-      "learning_rate": 0.0002535,
-      "loss": 0.1948,
+      "learning_rate": 0.00025340000000000003,
+      "loss": 0.1982,
       "step": 2970
     },
     {
       "epoch": 54.02,
-      "learning_rate": 0.00025340000000000003,
-      "loss": 0.2317,
+      "learning_rate": 0.00025330000000000003,
+      "loss": 0.1923,
       "step": 2971
     },
     {
       "epoch": 54.04,
-      "learning_rate": 0.00025330000000000003,
-      "loss": 0.2309,
+      "learning_rate": 0.0002532,
+      "loss": 0.2029,
       "step": 2972
     },
     {
       "epoch": 54.05,
-      "learning_rate": 0.0002532,
-      "loss": 0.1643,
+      "learning_rate": 0.00025309999999999997,
+      "loss": 0.1555,
       "step": 2973
     },
     {
       "epoch": 54.07,
-      "learning_rate": 0.00025309999999999997,
-      "loss": 0.1874,
+      "learning_rate": 0.000253,
+      "loss": 0.1888,
       "step": 2974
     },
     {
       "epoch": 54.09,
-      "learning_rate": 0.000253,
-      "loss": 0.2026,
+      "learning_rate": 0.0002529,
+      "loss": 0.1895,
       "step": 2975
     },
     {
       "epoch": 54.11,
-      "learning_rate": 0.0002529,
-      "loss": 0.1143,
+      "learning_rate": 0.0002528,
+      "loss": 0.0945,
       "step": 2976
     },
     {
       "epoch": 54.13,
-      "learning_rate": 0.0002528,
-      "loss": 0.1857,
+      "learning_rate": 0.00025269999999999996,
+      "loss": 0.1912,
       "step": 2977
     },
     {
       "epoch": 54.15,
-      "learning_rate": 0.00025269999999999996,
-      "loss": 0.1402,
+      "learning_rate": 0.0002526,
+      "loss": 0.1369,
       "step": 2978
     },
     {
       "epoch": 54.16,
-      "learning_rate": 0.0002526,
-      "loss": 0.1467,
+      "learning_rate": 0.0002525,
+      "loss": 0.1572,
       "step": 2979
     },
     {
       "epoch": 54.18,
-      "learning_rate": 0.0002525,
-      "loss": 0.2057,
+      "learning_rate": 0.0002524,
+      "loss": 0.1679,
       "step": 2980
     },
     {
       "epoch": 54.2,
-      "learning_rate": 0.0002524,
-      "loss": 0.1563,
+      "learning_rate": 0.0002523,
+      "loss": 0.1496,
       "step": 2981
     },
     {
       "epoch": 54.22,
-      "learning_rate": 0.0002523,
-      "loss": 0.1775,
+      "learning_rate": 0.0002522,
+      "loss": 0.1575,
       "step": 2982
     },
     {
       "epoch": 54.24,
-      "learning_rate": 0.0002522,
-      "loss": 0.1908,
+      "learning_rate": 0.0002521,
+      "loss": 0.166,
       "step": 2983
     },
     {
       "epoch": 54.25,
-      "learning_rate": 0.0002521,
-      "loss": 0.1706,
+      "learning_rate": 0.000252,
+      "loss": 0.1444,
       "step": 2984
     },
     {
       "epoch": 54.27,
-      "learning_rate": 0.000252,
-      "loss": 0.138,
+      "learning_rate": 0.0002519,
+      "loss": 0.1433,
       "step": 2985
     },
     {
       "epoch": 54.29,
-      "learning_rate": 0.0002519,
-      "loss": 0.1659,
+      "learning_rate": 0.00025180000000000005,
+      "loss": 0.1624,
       "step": 2986
     },
     {
       "epoch": 54.31,
-      "learning_rate": 0.00025180000000000005,
-      "loss": 0.2005,
+      "learning_rate": 0.0002517,
+      "loss": 0.1766,
       "step": 2987
     },
     {
       "epoch": 54.33,
-      "learning_rate": 0.0002517,
-      "loss": 0.1755,
+      "learning_rate": 0.0002516,
+      "loss": 0.1647,
       "step": 2988
     },
     {
       "epoch": 54.35,
-      "learning_rate": 0.0002516,
-      "loss": 0.1711,
+      "learning_rate": 0.0002515,
+      "loss": 0.1793,
       "step": 2989
     },
     {
       "epoch": 54.36,
-      "learning_rate": 0.0002515,
-      "loss": 0.1987,
+      "learning_rate": 0.00025140000000000004,
+      "loss": 0.1925,
       "step": 2990
     },
     {
       "epoch": 54.38,
-      "learning_rate": 0.00025140000000000004,
-      "loss": 0.1714,
+      "learning_rate": 0.00025130000000000004,
+      "loss": 0.1591,
       "step": 2991
     },
     {
       "epoch": 54.4,
-      "learning_rate": 0.00025130000000000004,
-      "loss": 0.211,
+      "learning_rate": 0.0002512,
+      "loss": 0.1836,
       "step": 2992
     },
     {
       "epoch": 54.42,
-      "learning_rate": 0.0002512,
-      "loss": 0.2388,
+      "learning_rate": 0.0002511,
+      "loss": 0.2093,
       "step": 2993
     },
     {
       "epoch": 54.44,
-      "learning_rate": 0.0002511,
-      "loss": 0.2047,
+      "learning_rate": 0.00025100000000000003,
+      "loss": 0.1877,
       "step": 2994
     },
     {
       "epoch": 54.45,
-      "learning_rate": 0.00025100000000000003,
-      "loss": 0.1337,
+      "learning_rate": 0.0002509,
+      "loss": 0.1181,
       "step": 2995
     },
     {
       "epoch": 54.47,
-      "learning_rate": 0.0002509,
-      "loss": 0.1744,
+      "learning_rate": 0.0002508,
+      "loss": 0.1646,
       "step": 2996
     },
     {
       "epoch": 54.49,
-      "learning_rate": 0.0002508,
-      "loss": 0.1427,
+      "learning_rate": 0.00025069999999999997,
+      "loss": 0.1361,
       "step": 2997
     },
     {
       "epoch": 54.51,
-      "learning_rate": 0.00025069999999999997,
-      "loss": 0.1693,
-      "step": 2998
+      "learning_rate": 0.0002506,
+      "loss": 0.1643,
+      "step": 2998
     },
     {
       "epoch": 54.53,
-      "learning_rate": 0.0002506,
-      "loss": 0.2113,
+      "learning_rate": 0.0002505,
+      "loss": 0.2093,
       "step": 2999
     },
     {
       "epoch": 54.55,
-      "learning_rate": 0.0002505,
-      "loss": 0.1733,
+      "learning_rate": 0.0002504,
+      "loss": 0.151,
       "step": 3000
     },
     {
       "epoch": 54.55,
-      "eval_cer": 0.09289494595121366,
-      "eval_loss": 0.41024285554885864,
-      "eval_runtime": 13.9284,
-      "eval_samples_per_second": 118.248,
-      "eval_steps_per_second": 1.867,
-      "eval_wer": 0.3712593197834746,
+      "eval_cer": 0.09031619664462735,
+      "eval_loss": 0.4243168532848358,
+      "eval_runtime": 14.1625,
+      "eval_samples_per_second": 116.293,
+      "eval_steps_per_second": 1.836,
+      "eval_wer": 0.35777755081197016,
       "step": 3000
     },
     {
       "epoch": 54.56,
-      "learning_rate": 0.0002504,
-      "loss": 0.1989,
+      "learning_rate": 0.0002503,
+      "loss": 0.191,
       "step": 3001
     },
     {
       "epoch": 54.58,
-      "learning_rate": 0.0002503,
-      "loss": 0.2131,
+      "learning_rate": 0.0002502,
+      "loss": 0.1925,
       "step": 3002
     },
     {
       "epoch": 54.6,
-      "learning_rate": 0.0002502,
-      "loss": 0.2049,
+      "learning_rate": 0.0002501,
+      "loss": 0.2005,
       "step": 3003
     },
     {
       "epoch": 54.62,
-      "learning_rate": 0.0002501,
-      "loss": 0.1368,
+      "learning_rate": 0.00025,
+      "loss": 0.1507,
       "step": 3004
     },
     {
       "epoch": 54.64,
-      "learning_rate": 0.00025,
-      "loss": 0.2063,
+      "learning_rate": 0.0002499,
+      "loss": 0.1792,
       "step": 3005
     },
     {
       "epoch": 54.65,
-      "learning_rate": 0.0002499,
-      "loss": 0.1756,
+      "learning_rate": 0.0002498,
+      "loss": 0.186,
       "step": 3006
     },
     {
       "epoch": 54.67,
-      "learning_rate": 0.0002498,
-      "loss": 0.155,
+      "learning_rate": 0.0002497,
+      "loss": 0.1554,
       "step": 3007
     },
     {
       "epoch": 54.69,
-      "learning_rate": 0.0002497,
-      "loss": 0.1621,
+      "learning_rate": 0.0002496,
+      "loss": 0.1509,
       "step": 3008
     },
     {
       "epoch": 54.71,
-      "learning_rate": 0.0002496,
-      "loss": 0.1982,
+      "learning_rate": 0.0002495,
+      "loss": 0.1832,
       "step": 3009
     },
     {
       "epoch": 54.73,
-      "learning_rate": 0.0002495,
-      "loss": 0.1882,
+      "learning_rate": 0.0002494,
+      "loss": 0.1874,
       "step": 3010
     },
     {
       "epoch": 54.75,
-      "learning_rate": 0.0002494,
-      "loss": 0.1879,
+      "learning_rate": 0.0002493,
+      "loss": 0.1802,
       "step": 3011
     },
     {
       "epoch": 54.76,
-      "learning_rate": 0.0002493,
-      "loss": 0.1483,
+      "learning_rate": 0.0002492,
+      "loss": 0.1463,
       "step": 3012
     },
     {
       "epoch": 54.78,
-      "learning_rate": 0.0002492,
-      "loss": 0.1733,
+      "learning_rate": 0.0002491,
+      "loss": 0.1695,
       "step": 3013
     },
     {
       "epoch": 54.8,
-      "learning_rate": 0.0002491,
-      "loss": 0.1675,
+      "learning_rate": 0.000249,
+      "loss": 0.1603,
       "step": 3014
     },
     {
       "epoch": 54.82,
-      "learning_rate": 0.000249,
-      "loss": 0.1935,
+      "learning_rate": 0.00024890000000000003,
+      "loss": 0.1683,
       "step": 3015
     },
     {
       "epoch": 54.84,
-      "learning_rate": 0.00024890000000000003,
-      "loss": 0.1921,
+      "learning_rate": 0.0002488,
+      "loss": 0.1244,
       "step": 3016
     },
     {
       "epoch": 54.85,
-      "learning_rate": 0.0002488,
-      "loss": 0.1796,
+      "learning_rate": 0.00024870000000000003,
+      "loss": 0.169,
       "step": 3017
     },
     {
       "epoch": 54.87,
-      "learning_rate": 0.00024870000000000003,
-      "loss": 0.1706,
+      "learning_rate": 0.00024859999999999997,
+      "loss": 0.152,
       "step": 3018
     },
     {
       "epoch": 54.89,
-      "learning_rate": 0.00024859999999999997,
-      "loss": 0.1596,
+      "learning_rate": 0.0002485,
+      "loss": 0.1428,
       "step": 3019
     },
     {
       "epoch": 54.91,
-      "learning_rate": 0.0002485,
-      "loss": 0.1244,
+      "learning_rate": 0.0002484,
+      "loss": 0.124,
       "step": 3020
     },
     {
       "epoch": 54.93,
-      "learning_rate": 0.0002484,
-      "loss": 0.1488,
+      "learning_rate": 0.0002483,
+      "loss": 0.1697,
       "step": 3021
     },
     {
       "epoch": 54.95,
-      "learning_rate": 0.0002483,
-      "loss": 0.1404,
+      "learning_rate": 0.0002482,
+      "loss": 0.1627,
       "step": 3022
     },
     {
       "epoch": 54.96,
-      "learning_rate": 0.0002482,
-      "loss": 0.1626,
+      "learning_rate": 0.0002481,
+      "loss": 0.1424,
       "step": 3023
     },
     {
       "epoch": 54.98,
-      "learning_rate": 0.0002481,
-      "loss": 0.1463,
+      "learning_rate": 0.000248,
+      "loss": 0.1423,
       "step": 3024
     },
     {
       "epoch": 55.0,
-      "learning_rate": 0.000248,
-      "loss": 0.2114,
+      "learning_rate": 0.0002479,
+      "loss": 0.1758,
       "step": 3025
     },
     {
       "epoch": 55.02,
-      "learning_rate": 0.0002479,
-      "loss": 0.1686,
+      "learning_rate": 0.0002478,
+      "loss": 0.1532,
       "step": 3026
     },
     {
       "epoch": 55.04,
-      "learning_rate": 0.0002478,
-      "loss": 0.2188,
+      "learning_rate": 0.0002477,
+      "loss": 0.1685,
       "step": 3027
     },
     {
       "epoch": 55.05,
-      "learning_rate": 0.0002477,
-      "loss": 0.1299,
+      "learning_rate": 0.0002476,
+      "loss": 0.1276,
       "step": 3028
     },
     {
       "epoch": 55.07,
-      "learning_rate": 0.0002476,
-      "loss": 0.2026,
+      "learning_rate": 0.0002475,
+      "loss": 0.1768,
       "step": 3029
     },
     {
       "epoch": 55.09,
-      "learning_rate": 0.0002475,
-      "loss": 0.1638,
+      "learning_rate": 0.0002474,
+      "loss": 0.157,
       "step": 3030
     },
     {
       "epoch": 55.11,
-      "learning_rate": 0.0002474,
-      "loss": 0.1902,
+      "learning_rate": 0.0002473,
+      "loss": 0.1438,
       "step": 3031
     },
     {
       "epoch": 55.13,
-      "learning_rate": 0.0002473,
-      "loss": 0.2485,
+      "learning_rate": 0.0002472,
+      "loss": 0.2165,
       "step": 3032
     },
     {
       "epoch": 55.15,
-      "learning_rate": 0.0002472,
-      "loss": 0.183,
+      "learning_rate": 0.0002471,
+      "loss": 0.1635,
       "step": 3033
     },
     {
       "epoch": 55.16,
-      "learning_rate": 0.0002471,
-      "loss": 0.2012,
+      "learning_rate": 0.000247,
+      "loss": 0.1591,
       "step": 3034
     },
     {
       "epoch": 55.18,
-      "learning_rate": 0.000247,
-      "loss": 0.1692,
+      "learning_rate": 0.00024690000000000004,
+      "loss": 0.1516,
       "step": 3035
     },
     {
       "epoch": 55.2,
-      "learning_rate": 0.00024690000000000004,
-      "loss": 0.1542,
+      "learning_rate": 0.0002468,
+      "loss": 0.135,
       "step": 3036
     },
     {
       "epoch": 55.22,
-      "learning_rate": 0.0002468,
-      "loss": 0.1667,
+      "learning_rate": 0.00024670000000000003,
+      "loss": 0.1614,
       "step": 3037
     },
     {
       "epoch": 55.24,
-      "learning_rate": 0.00024670000000000003,
-      "loss": 0.2365,
+      "learning_rate": 0.00024660000000000003,
+      "loss": 0.1955,
       "step": 3038
     },
     {
       "epoch": 55.25,
-      "learning_rate": 0.00024660000000000003,
-      "loss": 0.2281,
+      "learning_rate": 0.00024650000000000003,
+      "loss": 0.207,
       "step": 3039
     },
     {
       "epoch": 55.27,
-      "learning_rate": 0.00024650000000000003,
-      "loss": 0.1427,
+      "learning_rate": 0.0002464,
+      "loss": 0.1308,
       "step": 3040
     },
     {
       "epoch": 55.29,
-      "learning_rate": 0.0002464,
-      "loss": 0.1828,
+      "learning_rate": 0.00024629999999999997,
+      "loss": 0.1761,
       "step": 3041
     },
     {
       "epoch": 55.31,
-      "learning_rate": 0.00024629999999999997,
-      "loss": 0.2036,
+      "learning_rate": 0.0002462,
+      "loss": 0.2233,
       "step": 3042
     },
     {
       "epoch": 55.33,
-      "learning_rate": 0.0002462,
-      "loss": 0.215,
+      "learning_rate": 0.0002461,
+      "loss": 0.2229,
       "step": 3043
     },
     {
       "epoch": 55.35,
-      "learning_rate": 0.0002461,
-      "loss": 0.2252,
+      "learning_rate": 0.000246,
+      "loss": 0.1722,
       "step": 3044
     },
     {
       "epoch": 55.36,
-      "learning_rate": 0.000246,
-      "loss": 0.1851,
+      "learning_rate": 0.0002459,
+      "loss": 0.1541,
       "step": 3045
     },
     {
       "epoch": 55.38,
-      "learning_rate": 0.0002459,
-      "loss": 0.2246,
+      "learning_rate": 0.0002458,
+      "loss": 0.2522,
       "step": 3046
     },
     {
       "epoch": 55.4,
-      "learning_rate": 0.0002458,
-      "loss": 0.1566,
+      "learning_rate": 0.0002457,
+      "loss": 0.1294,
       "step": 3047
     },
     {
       "epoch": 55.42,
-      "learning_rate": 0.0002457,
-      "loss": 0.1612,
+      "learning_rate": 0.0002456,
+      "loss": 0.1624,
       "step": 3048
     },
     {
       "epoch": 55.44,
-      "learning_rate": 0.0002456,
-      "loss": 0.1585,
+      "learning_rate": 0.0002455,
+      "loss": 0.1582,
       "step": 3049
     },
     {
       "epoch": 55.45,
-      "learning_rate": 0.0002455,
-      "loss": 0.1325,
+      "learning_rate": 0.0002454,
+      "loss": 0.167,
       "step": 3050
     },
     {
       "epoch": 55.47,
-      "learning_rate": 0.0002454,
-      "loss": 0.2072,
+      "learning_rate": 0.0002453,
+      "loss": 0.1977,
       "step": 3051
     },
     {
       "epoch": 55.49,
-      "learning_rate": 0.0002453,
-      "loss": 0.1877,
+      "learning_rate": 0.0002452,
+      "loss": 0.1512,
       "step": 3052
     },
     {
       "epoch": 55.51,
-      "learning_rate": 0.0002452,
-      "loss": 0.1736,
+      "learning_rate": 0.0002451,
+      "loss": 0.16,
       "step": 3053
     },
     {
       "epoch": 55.53,
-      "learning_rate": 0.0002451,
-      "loss": 0.1791,
+      "learning_rate": 0.000245,
+      "loss": 0.2206,
       "step": 3054
     },
     {
       "epoch": 55.55,
-      "learning_rate": 0.000245,
-      "loss": 0.1689,
+      "learning_rate": 0.0002449,
+      "loss": 0.1673,
       "step": 3055
     },
     {
       "epoch": 55.56,
-      "learning_rate": 0.0002449,
-      "loss": 0.1384,
+      "learning_rate": 0.0002448,
+      "loss": 0.1576,
       "step": 3056
     },
     {
       "epoch": 55.58,
-      "learning_rate": 0.0002448,
-      "loss": 0.1724,
+      "learning_rate": 0.0002447,
+      "loss": 0.1735,
       "step": 3057
     },
     {
       "epoch": 55.6,
-      "learning_rate": 0.0002447,
-      "loss": 0.1722,
+      "learning_rate": 0.00024460000000000004,
+      "loss": 0.171,
       "step": 3058
     },
     {
       "epoch": 55.62,
-      "learning_rate": 0.00024460000000000004,
-      "loss": 0.1464,
+      "learning_rate": 0.0002445,
+      "loss": 0.1418,
       "step": 3059
     },
     {
       "epoch": 55.64,
-      "learning_rate": 0.0002445,
-      "loss": 0.1833,
+      "learning_rate": 0.00024440000000000003,
+      "loss": 0.1808,
       "step": 3060
     },
     {
       "epoch": 55.65,
-      "learning_rate": 0.00024440000000000003,
-      "loss": 0.1836,
+      "learning_rate": 0.0002443,
+      "loss": 0.1657,
       "step": 3061
     },
     {
       "epoch": 55.67,
-      "learning_rate": 0.0002443,
-      "loss": 0.1648,
+      "learning_rate": 0.0002442,
+      "loss": 0.1872,
       "step": 3062
     },
     {
       "epoch": 55.69,
-      "learning_rate": 0.0002442,
-      "loss": 0.1651,
+      "learning_rate": 0.00024410000000000002,
+      "loss": 0.132,
       "step": 3063
     },
     {
       "epoch": 55.71,
-      "learning_rate": 0.00024410000000000002,
-      "loss": 0.2326,
+      "learning_rate": 0.000244,
+      "loss": 0.2246,
       "step": 3064
     },
     {
       "epoch": 55.73,
-      "learning_rate": 0.000244,
-      "loss": 0.1892,
+      "learning_rate": 0.00024390000000000002,
+      "loss": 0.1931,
       "step": 3065
     },
     {
       "epoch": 55.75,
-      "learning_rate": 0.00024390000000000002,
-      "loss": 0.1763,
+      "learning_rate": 0.0002438,
+      "loss": 0.1726,
       "step": 3066
     },
     {
       "epoch": 55.76,
-      "learning_rate": 0.0002438,
-      "loss": 0.1449,
+      "learning_rate": 0.00024370000000000001,
+      "loss": 0.1324,
       "step": 3067
     },
     {
       "epoch": 55.78,
-      "learning_rate": 0.00024370000000000001,
-      "loss": 0.2074,
+      "learning_rate": 0.0002436,
+      "loss": 0.1874,
       "step": 3068
     },
     {
       "epoch": 55.8,
-      "learning_rate": 0.0002436,
-      "loss": 0.2013,
+      "learning_rate": 0.0002435,
+      "loss": 0.1605,
       "step": 3069
     },
     {
       "epoch": 55.82,
-      "learning_rate": 0.0002435,
-      "loss": 0.2282,
+      "learning_rate": 0.0002434,
+      "loss": 0.2226,
       "step": 3070
     },
     {
       "epoch": 55.84,
-      "learning_rate": 0.0002434,
-      "loss": 0.1716,
+      "learning_rate": 0.0002433,
+      "loss": 0.1491,
       "step": 3071
     },
     {
       "epoch": 55.85,
-      "learning_rate": 0.0002433,
-      "loss": 0.1268,
+      "learning_rate": 0.0002432,
+      "loss": 0.1492,
       "step": 3072
     },
     {
       "epoch": 55.87,
-      "learning_rate": 0.0002432,
-      "loss": 0.139,
+      "learning_rate": 0.00024310000000000003,
+      "loss": 0.1666,
       "step": 3073
     },
     {
       "epoch": 55.89,
-      "learning_rate": 0.00024310000000000003,
-      "loss": 0.1578,
+      "learning_rate": 0.000243,
+      "loss": 0.1493,
       "step": 3074
     },
     {
       "epoch": 55.91,
-      "learning_rate": 0.000243,
-      "loss": 0.1768,
+      "learning_rate": 0.00024290000000000002,
+      "loss": 0.1242,
       "step": 3075
     },
     {
       "epoch": 55.93,
-      "learning_rate": 0.00024290000000000002,
-      "loss": 0.2057,
+      "learning_rate": 0.0002428,
+      "loss": 0.1791,
       "step": 3076
     },
     {
       "epoch": 55.95,
-      "learning_rate": 0.0002428,
-      "loss": 0.1936,
+      "learning_rate": 0.00024270000000000002,
+      "loss": 0.181,
       "step": 3077
     },
     {
       "epoch": 55.96,
-      "learning_rate": 0.00024270000000000002,
-      "loss": 0.1716,
+      "learning_rate": 0.00024260000000000002,
+      "loss": 0.1565,
       "step": 3078
     },
     {
       "epoch": 55.98,
-      "learning_rate": 0.00024260000000000002,
-      "loss": 0.2028,
+      "learning_rate": 0.00024249999999999999,
+      "loss": 0.1714,
       "step": 3079
     },
     {
       "epoch": 56.0,
-      "learning_rate": 0.00024249999999999999,
-      "loss": 0.1149,
+      "learning_rate": 0.0002424,
+      "loss": 0.1165,
       "step": 3080
     },
     {
       "epoch": 56.02,
-      "learning_rate": 0.0002424,
-      "loss": 0.1648,
+      "learning_rate": 0.00024229999999999998,
+      "loss": 0.1484,
       "step": 3081
     },
     {
       "epoch": 56.04,
-      "learning_rate": 0.00024229999999999998,
-      "loss": 0.1431,
+      "learning_rate": 0.0002422,
+      "loss": 0.1428,
       "step": 3082
     },
     {
       "epoch": 56.05,
-      "learning_rate": 0.0002422,
-      "loss": 0.1579,
+      "learning_rate": 0.0002421,
+      "loss": 0.1521,
       "step": 3083
     },
     {
       "epoch": 56.07,
-      "learning_rate": 0.0002421,
-      "loss": 0.1771,
+      "learning_rate": 0.000242,
+      "loss": 0.2,
       "step": 3084
     },
     {
       "epoch": 56.09,
-      "learning_rate": 0.000242,
-      "loss": 0.1701,
+      "learning_rate": 0.0002419,
+      "loss": 0.1615,
       "step": 3085
     },
     {
       "epoch": 56.11,
-      "learning_rate": 0.0002419,
-      "loss": 0.1616,
+      "learning_rate": 0.0002418,
+      "loss": 0.1623,
       "step": 3086
     },
     {
       "epoch": 56.13,
-      "learning_rate": 0.0002418,
-      "loss": 0.2383,
+      "learning_rate": 0.0002417,
+      "loss": 0.2285,
       "step": 3087
     },
     {
       "epoch": 56.15,
-      "learning_rate": 0.0002417,
-      "loss": 0.1948,
+      "learning_rate": 0.00024160000000000002,
+      "loss": 0.1452,
       "step": 3088
     },
     {
       "epoch": 56.16,
-      "learning_rate": 0.00024160000000000002,
-      "loss": 0.1743,
+      "learning_rate": 0.0002415,
+      "loss": 0.1752,
       "step": 3089
     },
     {
       "epoch": 56.18,
-      "learning_rate": 0.0002415,
-      "loss": 0.2435,
+      "learning_rate": 0.0002414,
+      "loss": 0.1966,
       "step": 3090
     },
     {
       "epoch": 56.2,
-      "learning_rate": 0.0002414,
-      "loss": 0.1904,
+      "learning_rate": 0.00024129999999999998,
+      "loss": 0.1575,
       "step": 3091
     },
     {
       "epoch": 56.22,
-      "learning_rate": 0.00024129999999999998,
-      "loss": 0.1509,
+      "learning_rate": 0.0002412,
+      "loss": 0.1363,
       "step": 3092
     },
     {
       "epoch": 56.24,
-      "learning_rate": 0.0002412,
-      "loss": 0.2042,
+      "learning_rate": 0.0002411,
+      "loss": 0.1746,
       "step": 3093
     },
     {
       "epoch": 56.25,
-      "learning_rate": 0.0002411,
-      "loss": 0.1742,
+      "learning_rate": 0.000241,
+      "loss": 0.1533,
       "step": 3094
     },
     {
       "epoch": 56.27,
-      "learning_rate": 0.000241,
-      "loss": 0.1618,
+      "learning_rate": 0.0002409,
+      "loss": 0.1323,
       "step": 3095
     },
     {
       "epoch": 56.29,
-      "learning_rate": 0.0002409,
-      "loss": 0.1771,
+      "learning_rate": 0.0002408,
+      "loss": 0.1689,
       "step": 3096
     },
     {
       "epoch": 56.31,
-      "learning_rate": 0.0002408,
-      "loss": 0.1646,
+      "learning_rate": 0.0002407,
+      "loss": 0.1713,
       "step": 3097
     },
     {
       "epoch": 56.33,
-      "learning_rate": 0.0002407,
-      "loss": 0.1749,
+      "learning_rate": 0.00024060000000000002,
+      "loss": 0.1669,
       "step": 3098
     },
     {
       "epoch": 56.35,
-      "learning_rate": 0.00024060000000000002,
-      "loss": 0.1553,
+      "learning_rate": 0.0002405,
+      "loss": 0.1291,
       "step": 3099
     },
     {
       "epoch": 56.36,
-      "learning_rate": 0.0002405,
-      "loss": 0.1567,
+      "learning_rate": 0.00024040000000000002,
+      "loss": 0.1361,
       "step": 3100
     },
     {
       "epoch": 56.38,
-      "learning_rate": 0.00024040000000000002,
-      "loss": 0.2124,
+      "learning_rate": 0.0002403,
+      "loss": 0.1831,
       "step": 3101
     },
     {
       "epoch": 56.4,
-      "learning_rate": 0.0002403,
-      "loss": 0.1456,
+      "learning_rate": 0.0002402,
+      "loss": 0.1452,
       "step": 3102
     },
     {
       "epoch": 56.42,
-      "learning_rate": 0.0002402,
-      "loss": 0.1999,
+      "learning_rate": 0.0002401,
+      "loss": 0.173,
       "step": 3103
     },
     {
       "epoch": 56.44,
-      "learning_rate": 0.0002401,
-      "loss": 0.2026,
+      "learning_rate": 0.00024,
+      "loss": 0.1679,
       "step": 3104
     },
     {
       "epoch": 56.45,
-      "learning_rate": 0.00024,
-      "loss": 0.1379,
+      "learning_rate": 0.0002399,
+      "loss": 0.1177,
       "step": 3105
     },
     {
       "epoch": 56.47,
-      "learning_rate": 0.0002399,
-      "loss": 0.187,
+      "learning_rate": 0.00023980000000000003,
+      "loss": 0.1606,
       "step": 3106
     },
     {
       "epoch": 56.49,
-      "learning_rate": 0.00023980000000000003,
-      "loss": 0.1721,
+      "learning_rate": 0.0002397,
+      "loss": 0.1468,
       "step": 3107
     },
     {
       "epoch": 56.51,
-      "learning_rate": 0.0002397,
-      "loss": 0.1568,
+      "learning_rate": 0.00023960000000000002,
+      "loss": 0.1558,
       "step": 3108
     },
     {
       "epoch": 56.53,
-      "learning_rate": 0.00023960000000000002,
-      "loss": 0.163,
+      "learning_rate": 0.0002395,
+      "loss": 0.1336,
       "step": 3109
     },
     {
       "epoch": 56.55,
-      "learning_rate": 0.0002395,
-      "loss": 0.1071,
+      "learning_rate": 0.00023940000000000002,
+      "loss": 0.1004,
       "step": 3110
     },
     {
       "epoch": 56.56,
-      "learning_rate": 0.00023940000000000002,
-      "loss": 0.1012,
+      "learning_rate": 0.00023930000000000002,
+      "loss": 0.1069,
       "step": 3111
     },
     {
       "epoch": 56.58,
-      "learning_rate": 0.00023930000000000002,
-      "loss": 0.1761,
+      "learning_rate": 0.00023920000000000001,
+      "loss": 0.1427,
       "step": 3112
     },
     {
       "epoch": 56.6,
-      "learning_rate": 0.00023920000000000001,
-      "loss": 0.159,
+      "learning_rate": 0.0002391,
+      "loss": 0.1377,
       "step": 3113
     },
     {
       "epoch": 56.62,
-      "learning_rate": 0.0002391,
-      "loss": 0.1429,
+      "learning_rate": 0.00023899999999999998,
+      "loss": 0.1405,
       "step": 3114
     },
     {
       "epoch": 56.64,
-      "learning_rate": 0.00023899999999999998,
-      "loss": 0.1388,
+      "learning_rate": 0.0002389,
+      "loss": 0.1334,
       "step": 3115
     },
     {
       "epoch": 56.65,
-      "learning_rate": 0.0002389,
-      "loss": 0.1751,
+      "learning_rate": 0.0002388,
+      "loss": 0.1658,
       "step": 3116
     },
     {
       "epoch": 56.67,
-      "learning_rate": 0.0002388,
-      "loss": 0.1412,
+      "learning_rate": 0.0002387,
+      "loss": 0.1428,
       "step": 3117
     },
     {
       "epoch": 56.69,
-      "learning_rate": 0.0002387,
-      "loss": 0.1464,
+      "learning_rate": 0.0002386,
+      "loss": 0.147,
       "step": 3118
     },
     {
       "epoch": 56.71,
-      "learning_rate": 0.0002386,
-      "loss": 0.1378,
+      "learning_rate": 0.0002385,
+      "loss": 0.1217,
       "step": 3119
     },
     {
       "epoch": 56.73,
-      "learning_rate": 0.0002385,
-      "loss": 0.1948,
+      "learning_rate": 0.0002384,
+      "loss": 0.1869,
       "step": 3120
     },
     {
       "epoch": 56.75,
-      "learning_rate": 0.0002384,
-      "loss": 0.1667,
+      "learning_rate": 0.00023830000000000002,
+      "loss": 0.1603,
       "step": 3121
     },
     {
       "epoch": 56.76,
-      "learning_rate": 0.00023830000000000002,
-      "loss": 0.1737,
+      "learning_rate": 0.0002382,
+      "loss": 0.1521,
       "step": 3122
     },
     {
       "epoch": 56.78,
-      "learning_rate": 0.0002382,
-      "loss": 0.1213,
+      "learning_rate": 0.00023810000000000001,
+      "loss": 0.1304,
       "step": 3123
     },
     {
       "epoch": 56.8,
-      "learning_rate": 0.00023810000000000001,
-      "loss": 0.1299,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 0.1522,
       "step": 3124
     },
     {
       "epoch": 56.82,
-      "learning_rate": 0.00023799999999999998,
-      "loss": 0.2017,
+      "learning_rate": 0.0002379,
+      "loss": 0.172,
       "step": 3125
     },
     {
       "epoch": 56.84,
-      "learning_rate": 0.0002379,
-      "loss": 0.1568,
+      "learning_rate": 0.0002378,
+      "loss": 0.1542,
       "step": 3126
     },
     {
       "epoch": 56.85,
-      "learning_rate": 0.0002378,
-      "loss": 0.1274,
+      "learning_rate": 0.0002377,
+      "loss": 0.114,
       "step": 3127
     },
     {
       "epoch": 56.87,
-      "learning_rate": 0.0002377,
-      "loss": 0.1202,
+      "learning_rate": 0.0002376,
+      "loss": 0.1397,
       "step": 3128
     },
     {
       "epoch": 56.89,
-      "learning_rate": 0.0002376,
-      "loss": 0.1625,
+      "learning_rate": 0.0002375,
+      "loss": 0.1593,
       "step": 3129
     },
     {
       "epoch": 56.91,
-      "learning_rate": 0.0002375,
-      "loss": 0.1136,
+      "learning_rate": 0.0002374,
+      "loss": 0.095,
       "step": 3130
     },
     {
       "epoch": 56.93,
-      "learning_rate": 0.0002374,
-      "loss": 0.19,
+      "learning_rate": 0.00023730000000000002,
+      "loss": 0.1732,
       "step": 3131
     },
     {
       "epoch": 56.95,
-      "learning_rate": 0.00023730000000000002,
-      "loss": 0.2211,
+      "learning_rate": 0.0002372,
+      "loss": 0.1796,
       "step": 3132
     },
     {
       "epoch": 56.96,
-      "learning_rate": 0.0002372,
-      "loss": 0.1828,
+      "learning_rate": 0.00023710000000000002,
+      "loss": 0.1808,
       "step": 3133
     },
     {
       "epoch": 56.98,
-      "learning_rate": 0.00023710000000000002,
-      "loss": 0.135,
+      "learning_rate": 0.000237,
+      "loss": 0.1528,
       "step": 3134
     },
     {
       "epoch": 57.0,
-      "learning_rate": 0.000237,
-      "loss": 0.1503,
+      "learning_rate": 0.0002369,
+      "loss": 0.0719,
       "step": 3135
     },
     {
       "epoch": 57.02,
-      "learning_rate": 0.0002369,
-      "loss": 0.1752,
+      "learning_rate": 0.0002368,
+      "loss": 0.1747,
       "step": 3136
     },
     {
       "epoch": 57.04,
-      "learning_rate": 0.0002368,
-      "loss": 0.1769,
+      "learning_rate": 0.0002367,
+      "loss": 0.1612,
       "step": 3137
     },
     {
       "epoch": 57.05,
-      "learning_rate": 0.0002367,
-      "loss": 0.1858,
+      "learning_rate": 0.0002366,
+      "loss": 0.184,
       "step": 3138
     },
     {
       "epoch": 57.07,
-      "learning_rate": 0.0002366,
-      "loss": 0.1415,
+      "learning_rate": 0.0002365,
+      "loss": 0.1297,
       "step": 3139
     },
     {
       "epoch": 57.09,
-      "learning_rate": 0.0002365,
-      "loss": 0.1703,
-      "step": 3140
+      "learning_rate": 0.0002364,
+      "loss": 0.1385,
+      "step": 3140
     },
     {
       "epoch": 57.11,
-      "learning_rate": 0.0002364,
-      "loss": 0.1285,
+      "learning_rate": 0.00023630000000000002,
+      "loss": 0.1,
       "step": 3141
     },
     {
       "epoch": 57.13,
-      "learning_rate": 0.00023630000000000002,
-      "loss": 0.1777,
+      "learning_rate": 0.0002362,
+      "loss": 0.1459,
       "step": 3142
     },
     {
       "epoch": 57.15,
-      "learning_rate": 0.0002362,
-      "loss": 0.1379,
+      "learning_rate": 0.00023610000000000002,
+      "loss": 0.1293,
       "step": 3143
     },
     {
       "epoch": 57.16,
-      "learning_rate": 0.00023610000000000002,
-      "loss": 0.1516,
+      "learning_rate": 0.000236,
+      "loss": 0.148,
       "step": 3144
     },
     {
       "epoch": 57.18,
-      "learning_rate": 0.000236,
-      "loss": 0.1766,
+      "learning_rate": 0.00023590000000000001,
+      "loss": 0.1913,
       "step": 3145
     },
     {
       "epoch": 57.2,
-      "learning_rate": 0.00023590000000000001,
-      "loss": 0.1772,
+      "learning_rate": 0.0002358,
+      "loss": 0.1477,
       "step": 3146
     },
     {
       "epoch": 57.22,
-      "learning_rate": 0.0002358,
-      "loss": 0.1097,
+      "learning_rate": 0.0002357,
+      "loss": 0.1307,
       "step": 3147
     },
     {
       "epoch": 57.24,
-      "learning_rate": 0.0002357,
-      "loss": 0.1645,
+      "learning_rate": 0.0002356,
+      "loss": 0.1435,
       "step": 3148
     },
     {
       "epoch": 57.25,
-      "learning_rate": 0.0002356,
-      "loss": 0.1742,
+      "learning_rate": 0.0002355,
+      "loss": 0.1608,
       "step": 3149
     },
     {
       "epoch": 57.27,
-      "learning_rate": 0.0002355,
-      "loss": 0.206,
+      "learning_rate": 0.0002354,
+      "loss": 0.1868,
       "step": 3150
     },
     {
       "epoch": 57.29,
-      "learning_rate": 0.0002354,
-      "loss": 0.1331,
+      "learning_rate": 0.00023530000000000003,
+      "loss": 0.1044,
       "step": 3151
     },
     {
       "epoch": 57.31,
-      "learning_rate": 0.00023530000000000003,
-      "loss": 0.1717,
+      "learning_rate": 0.0002352,
+      "loss": 0.1606,
       "step": 3152
     },
     {
       "epoch": 57.33,
-      "learning_rate": 0.0002352,
-      "loss": 0.1625,
+      "learning_rate": 0.0002351,
+      "loss": 0.1489,
       "step": 3153
     },
     {
       "epoch": 57.35,
-      "learning_rate": 0.0002351,
-      "loss": 0.1233,
+      "learning_rate": 0.000235,
+      "loss": 0.148,
       "step": 3154
     },
     {
       "epoch": 57.36,
-      "learning_rate": 0.000235,
-      "loss": 0.1464,
+      "learning_rate": 0.0002349,
+      "loss": 0.1398,
       "step": 3155
     },
     {
       "epoch": 57.38,
-      "learning_rate": 0.0002349,
-      "loss": 0.171,
+      "learning_rate": 0.00023480000000000002,
+      "loss": 0.1463,
       "step": 3156
     },
     {
       "epoch": 57.4,
-      "learning_rate": 0.00023480000000000002,
-      "loss": 0.162,
+      "learning_rate": 0.00023469999999999999,
+      "loss": 0.195,
       "step": 3157
     },
     {
       "epoch": 57.42,
-      "learning_rate": 0.00023469999999999999,
-      "loss": 0.2083,
+      "learning_rate": 0.0002346,
+      "loss": 0.2026,
       "step": 3158
     },
     {
       "epoch": 57.44,
-      "learning_rate": 0.0002346,
-      "loss": 0.1396,
+      "learning_rate": 0.00023449999999999998,
+      "loss": 0.1107,
       "step": 3159
     },
     {
       "epoch": 57.45,
-      "learning_rate": 0.00023449999999999998,
-      "loss": 0.135,
+      "learning_rate": 0.0002344,
+      "loss": 0.1397,
       "step": 3160
     },
     {
       "epoch": 57.47,
-      "learning_rate": 0.0002344,
-      "loss": 0.1837,
+      "learning_rate": 0.0002343,
+      "loss": 0.181,
       "step": 3161
     },
     {
       "epoch": 57.49,
-      "learning_rate": 0.0002343,
-      "loss": 0.1449,
+      "learning_rate": 0.0002342,
+      "loss": 0.1585,
       "step": 3162
     },
     {
       "epoch": 57.51,
-      "learning_rate": 0.0002342,
-      "loss": 0.1374,
+      "learning_rate": 0.0002341,
+      "loss": 0.1437,
       "step": 3163
     },
     {
       "epoch": 57.53,
-      "learning_rate": 0.0002341,
-      "loss": 0.1607,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 0.1521,
       "step": 3164
     },
     {
       "epoch": 57.55,
-      "learning_rate": 0.00023400000000000002,
-      "loss": 0.1668,
+      "learning_rate": 0.0002339,
+      "loss": 0.1658,
       "step": 3165
     },
     {
       "epoch": 57.56,
-      "learning_rate": 0.0002339,
-      "loss": 0.1412,
+      "learning_rate": 0.00023380000000000002,
+      "loss": 0.1078,
       "step": 3166
     },
     {
       "epoch": 57.58,
-      "learning_rate": 0.00023380000000000002,
-      "loss": 0.1909,
+      "learning_rate": 0.0002337,
+      "loss": 0.1636,
       "step": 3167
     },
     {
       "epoch": 57.6,
-      "learning_rate": 0.0002337,
-      "loss": 0.1612,
+      "learning_rate": 0.0002336,
+      "loss": 0.1434,
       "step": 3168
     },
     {
       "epoch": 57.62,
-      "learning_rate": 0.0002336,
-      "loss": 0.1417,
+      "learning_rate": 0.0002335,
+      "loss": 0.1471,
       "step": 3169
     },
     {
       "epoch": 57.64,
-      "learning_rate": 0.0002335,
-      "loss": 0.1581,
+      "learning_rate": 0.0002334,
+      "loss": 0.1461,
       "step": 3170
     },
     {
       "epoch": 57.65,
-      "learning_rate": 0.0002334,
-      "loss": 0.21,
+      "learning_rate": 0.0002333,
+      "loss": 0.1749,
       "step": 3171
     },
     {
       "epoch": 57.67,
-      "learning_rate": 0.0002333,
-      "loss": 0.1438,
+      "learning_rate": 0.0002332,
+      "loss": 0.1698,
       "step": 3172
     },
     {
       "epoch": 57.69,
-      "learning_rate": 0.0002332,
-      "loss": 0.1345,
+      "learning_rate": 0.0002331,
+      "loss": 0.1357,
       "step": 3173
     },
     {
       "epoch": 57.71,
-      "learning_rate": 0.0002331,
-      "loss": 0.1987,
+      "learning_rate": 0.00023300000000000003,
+      "loss": 0.2255,
       "step": 3174
     },
     {
       "epoch": 57.73,
-      "learning_rate": 0.00023300000000000003,
-      "loss": 0.1681,
+      "learning_rate": 0.0002329,
+      "loss": 0.1677,
       "step": 3175
     },
     {
       "epoch": 57.75,
-      "learning_rate": 0.0002329,
-      "loss": 0.1677,
+      "learning_rate": 0.00023280000000000002,
+      "loss": 0.1578,
       "step": 3176
     },
     {
       "epoch": 57.76,
-      "learning_rate": 0.00023280000000000002,
-      "loss": 0.1808,
+      "learning_rate": 0.0002327,
+      "loss": 0.1792,
       "step": 3177
     },
     {
       "epoch": 57.78,
-      "learning_rate": 0.0002327,
-      "loss": 0.1533,
+      "learning_rate": 0.00023260000000000002,
+      "loss": 0.1519,
       "step": 3178
     },
     {
       "epoch": 57.8,
-      "learning_rate": 0.00023260000000000002,
-      "loss": 0.1604,
+      "learning_rate": 0.0002325,
+      "loss": 0.1404,
       "step": 3179
     },
     {
       "epoch": 57.82,
-      "learning_rate": 0.0002325,
-      "loss": 0.1446,
+      "learning_rate": 0.0002324,
+      "loss": 0.1426,
       "step": 3180
     },
     {
       "epoch": 57.84,
-      "learning_rate": 0.0002324,
-      "loss": 0.1603,
+      "learning_rate": 0.0002323,
+      "loss": 0.1468,
       "step": 3181
     },
     {
       "epoch": 57.85,
-      "learning_rate": 0.0002323,
-      "loss": 0.1612,
+      "learning_rate": 0.0002322,
+      "loss": 0.1505,
       "step": 3182
     },
     {
       "epoch": 57.87,
-      "learning_rate": 0.0002322,
-      "loss": 0.1174,
+      "learning_rate": 0.0002321,
+      "loss": 0.1154,
       "step": 3183
     },
     {
       "epoch": 57.89,
-      "learning_rate": 0.0002321,
-      "loss": 0.1621,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 0.1364,
       "step": 3184
     },
     {
       "epoch": 57.91,
-      "learning_rate": 0.00023200000000000003,
-      "loss": 0.22,
+      "learning_rate": 0.0002319,
+      "loss": 0.2053,
       "step": 3185
     },
     {
       "epoch": 57.93,
-      "learning_rate": 0.0002319,
-      "loss": 0.1955,
+      "learning_rate": 0.00023180000000000002,
+      "loss": 0.1956,
       "step": 3186
     },
     {
       "epoch": 57.95,
-      "learning_rate": 0.00023180000000000002,
-      "loss": 0.2012,
+      "learning_rate": 0.0002317,
+      "loss": 0.1904,
       "step": 3187
     },
     {
       "epoch": 57.96,
-      "learning_rate": 0.0002317,
-      "loss": 0.1603,
+      "learning_rate": 0.0002316,
+      "loss": 0.1525,
       "step": 3188
     },
     {
       "epoch": 57.98,
-      "learning_rate": 0.0002316,
-      "loss": 0.1715,
+      "learning_rate": 0.00023150000000000002,
+      "loss": 0.1771,
       "step": 3189
     },
     {
       "epoch": 58.0,
-      "learning_rate": 0.00023150000000000002,
-      "loss": 0.2343,
+      "learning_rate": 0.0002314,
+      "loss": 0.242,
       "step": 3190
     },
     {
       "epoch": 58.02,
-      "learning_rate": 0.0002314,
-      "loss": 0.1583,
+      "learning_rate": 0.0002313,
+      "loss": 0.1582,
       "step": 3191
     },
     {
       "epoch": 58.04,
-      "learning_rate": 0.0002313,
-      "loss": 0.1837,
+      "learning_rate": 0.00023119999999999998,
+      "loss": 0.2339,
       "step": 3192
     },
     {
       "epoch": 58.05,
-      "learning_rate": 0.00023119999999999998,
-      "loss": 0.1804,
+      "learning_rate": 0.0002311,
+      "loss": 0.171,
       "step": 3193
     },
     {
       "epoch": 58.07,
-      "learning_rate": 0.0002311,
-      "loss": 0.1637,
+      "learning_rate": 0.000231,
+      "loss": 0.1251,
       "step": 3194
     },
     {
       "epoch": 58.09,
-      "learning_rate": 0.000231,
-      "loss": 0.1502,
+      "learning_rate": 0.0002309,
+      "loss": 0.1417,
       "step": 3195
     },
     {
       "epoch": 58.11,
-      "learning_rate": 0.0002309,
-      "loss": 0.1598,
+      "learning_rate": 0.0002308,
+      "loss": 0.1503,
       "step": 3196
     },
     {
       "epoch": 58.13,
-      "learning_rate": 0.0002308,
-      "loss": 0.1609,
+      "learning_rate": 0.0002307,
+      "loss": 0.1509,
       "step": 3197
     },
     {
       "epoch": 58.15,
-      "learning_rate": 0.0002307,
-      "loss": 0.1694,
+      "learning_rate": 0.0002306,
+      "loss": 0.1506,
       "step": 3198
     },
     {
       "epoch": 58.16,
-      "learning_rate": 0.0002306,
-      "loss": 0.1457,
+      "learning_rate": 0.00023050000000000002,
+      "loss": 0.1414,
       "step": 3199
     },
     {
       "epoch": 58.18,
-      "learning_rate": 0.00023050000000000002,
-      "loss": 0.1402,
+      "learning_rate": 0.0002304,
+      "loss": 0.1555,
       "step": 3200
     },
     {
       "epoch": 58.2,
-      "learning_rate": 0.0002304,
-      "loss": 0.141,
+      "learning_rate": 0.00023030000000000001,
+      "loss": 0.1369,
       "step": 3201
     },
     {
       "epoch": 58.22,
-      "learning_rate": 0.00023030000000000001,
-      "loss": 0.1069,
+      "learning_rate": 0.00023019999999999998,
+      "loss": 0.108,
       "step": 3202
     },
     {
       "epoch": 58.24,
-      "learning_rate": 0.00023019999999999998,
-      "loss": 0.1746,
+      "learning_rate": 0.0002301,
+      "loss": 0.1419,
       "step": 3203
     },
     {
       "epoch": 58.25,
-      "learning_rate": 0.0002301,
-      "loss": 0.148,
+      "learning_rate": 0.00023,
+      "loss": 0.1576,
       "step": 3204
     },
     {
       "epoch": 58.27,
-      "learning_rate": 0.00023,
-      "loss": 0.172,
+      "learning_rate": 0.0002299,
+      "loss": 0.1506,
       "step": 3205
     },
     {
       "epoch": 58.29,
-      "learning_rate": 0.0002299,
-      "loss": 0.1907,
+      "learning_rate": 0.0002298,
+      "loss": 0.1806,
       "step": 3206
     },
     {
       "epoch": 58.31,
-      "learning_rate": 0.0002298,
-      "loss": 0.1791,
+      "learning_rate": 0.0002297,
+      "loss": 0.1966,
       "step": 3207
     },
     {
       "epoch": 58.33,
-      "learning_rate": 0.0002297,
-      "loss": 0.1983,
+      "learning_rate": 0.0002296,
+      "loss": 0.185,
       "step": 3208
     },
     {
       "epoch": 58.35,
-      "learning_rate": 0.0002296,
-      "loss": 0.1405,
+      "learning_rate": 0.00022950000000000002,
+      "loss": 0.1316,
       "step": 3209
     },
     {
       "epoch": 58.36,
-      "learning_rate": 0.00022950000000000002,
-      "loss": 0.1855,
+      "learning_rate": 0.0002294,
+      "loss": 0.1763,
       "step": 3210
     },
     {
       "epoch": 58.38,
-      "learning_rate": 0.0002294,
-      "loss": 0.1644,
+      "learning_rate": 0.00022930000000000002,
+      "loss": 0.1679,
       "step": 3211
     },
     {
       "epoch": 58.4,
-      "learning_rate": 0.00022930000000000002,
-      "loss": 0.1735,
+      "learning_rate": 0.0002292,
+      "loss": 0.156,
       "step": 3212
     },
     {
       "epoch": 58.42,
-      "learning_rate": 0.0002292,
-      "loss": 0.1477,
+      "learning_rate": 0.0002291,
+      "loss": 0.1275,
       "step": 3213
     },
     {
       "epoch": 58.44,
-      "learning_rate": 0.0002291,
-      "loss": 0.1574,
+      "learning_rate": 0.000229,
+      "loss": 0.1582,
       "step": 3214
     },
     {
       "epoch": 58.45,
-      "learning_rate": 0.000229,
-      "loss": 0.1525,
+      "learning_rate": 0.0002289,
+      "loss": 0.1681,
       "step": 3215
     },
     {
       "epoch": 58.47,
-      "learning_rate": 0.0002289,
-      "loss": 0.1681,
+      "learning_rate": 0.0002288,
+      "loss": 0.1605,
       "step": 3216
     },
     {
       "epoch": 58.49,
-      "learning_rate": 0.0002288,
-      "loss": 0.1306,
+      "learning_rate": 0.0002287,
+      "loss": 0.1339,
       "step": 3217
     },
     {
       "epoch": 58.51,
-      "learning_rate": 0.0002287,
-      "loss": 0.1464,
+      "learning_rate": 0.0002286,
+      "loss": 0.1679,
       "step": 3218
     },
     {
       "epoch": 58.53,
-      "learning_rate": 0.0002286,
-      "loss": 0.1668,
+      "learning_rate": 0.00022850000000000002,
+      "loss": 0.162,
       "step": 3219
     },
     {
       "epoch": 58.55,
-      "learning_rate": 0.00022850000000000002,
-      "loss": 0.2135,
+      "learning_rate": 0.0002284,
+      "loss": 0.2092,
       "step": 3220
     },
     {
       "epoch": 58.56,
-      "learning_rate": 0.0002284,
-      "loss": 0.2004,
+      "learning_rate": 0.00022830000000000002,
+      "loss": 0.1468,
       "step": 3221
     },
     {
       "epoch": 58.58,
-      "learning_rate": 0.00022830000000000002,
-      "loss": 0.1413,
+      "learning_rate": 0.0002282,
+      "loss": 0.1582,
       "step": 3222
     },
     {
       "epoch": 58.6,
-      "learning_rate": 0.0002282,
-      "loss": 0.1487,
+      "learning_rate": 0.00022810000000000001,
+      "loss": 0.138,
       "step": 3223
     },
     {
       "epoch": 58.62,
-      "learning_rate": 0.00022810000000000001,
-      "loss": 0.171,
+      "learning_rate": 0.000228,
+      "loss": 0.1567,
       "step": 3224
     },
     {
       "epoch": 58.64,
-      "learning_rate": 0.000228,
-      "loss": 0.1502,
+      "learning_rate": 0.00022789999999999998,
+      "loss": 0.1115,
       "step": 3225
     },
     {
       "epoch": 58.65,
-      "learning_rate": 0.00022789999999999998,
-      "loss": 0.1604,
+      "learning_rate": 0.0002278,
+      "loss": 0.1969,
       "step": 3226
     },
     {
       "epoch": 58.67,
-      "learning_rate": 0.0002278,
-      "loss": 0.1366,
+      "learning_rate": 0.0002277,
+      "loss": 0.1564,
       "step": 3227
     },
     {
       "epoch": 58.69,
-      "learning_rate": 0.0002277,
-      "loss": 0.1252,
+      "learning_rate": 0.0002276,
+      "loss": 0.1698,
       "step": 3228
     },
     {
       "epoch": 58.71,
-      "learning_rate": 0.0002276,
-      "loss": 0.1526,
+      "learning_rate": 0.0002275,
+      "loss": 0.1483,
       "step": 3229
     },
     {
       "epoch": 58.73,
-      "learning_rate": 0.0002275,
-      "loss": 0.1673,
+      "learning_rate": 0.0002274,
+      "loss": 0.1762,
       "step": 3230
     },
     {
       "epoch": 58.75,
-      "learning_rate": 0.0002274,
-      "loss": 0.1422,
+      "learning_rate": 0.0002273,
+      "loss": 0.1397,
       "step": 3231
     },
     {
       "epoch": 58.76,
-      "learning_rate": 0.0002273,
-      "loss": 0.1838,
+      "learning_rate": 0.00022720000000000002,
+      "loss": 0.1977,
       "step": 3232
     },
     {
       "epoch": 58.78,
-      "learning_rate": 0.00022720000000000002,
-      "loss": 0.1604,
+      "learning_rate": 0.0002271,
+      "loss": 0.1391,
       "step": 3233
     },
     {
       "epoch": 58.8,
-      "learning_rate": 0.0002271,
-      "loss": 0.1229,
+      "learning_rate": 0.00022700000000000002,
+      "loss": 0.1024,
       "step": 3234
     },
     {
       "epoch": 58.82,
-      "learning_rate": 0.00022700000000000002,
-      "loss": 0.1873,
+      "learning_rate": 0.00022689999999999999,
+      "loss": 0.1827,
       "step": 3235
     },
     {
       "epoch": 58.84,
-      "learning_rate": 0.00022689999999999999,
-      "loss": 0.1594,
+      "learning_rate": 0.0002268,
+      "loss": 0.1455,
       "step": 3236
     },
     {
       "epoch": 58.85,
-      "learning_rate": 0.0002268,
-      "loss": 0.1641,
+      "learning_rate": 0.0002267,
+      "loss": 0.1505,
       "step": 3237
     },
     {
       "epoch": 58.87,
-      "learning_rate": 0.0002267,
-      "loss": 0.1952,
+      "learning_rate": 0.0002266,
+      "loss": 0.1748,
       "step": 3238
     },
     {
       "epoch": 58.89,
-      "learning_rate": 0.0002266,
-      "loss": 0.154,
+      "learning_rate": 0.0002265,
+      "loss": 0.129,
       "step": 3239
     },
     {
       "epoch": 58.91,
-      "learning_rate": 0.0002265,
-      "loss": 0.176,
+      "learning_rate": 0.0002264,
+      "loss": 0.1367,
       "step": 3240
     },
     {
       "epoch": 58.93,
-      "learning_rate": 0.0002264,
-      "loss": 0.1607,
+      "learning_rate": 0.0002263,
+      "loss": 0.1546,
       "step": 3241
     },
     {
       "epoch": 58.95,
-      "learning_rate": 0.0002263,
-      "loss": 0.2018,
+      "learning_rate": 0.00022620000000000002,
+      "loss": 0.1571,
       "step": 3242
     },
     {
       "epoch": 58.96,
-      "learning_rate": 0.00022620000000000002,
-      "loss": 0.1853,
+      "learning_rate": 0.0002261,
+      "loss": 0.1589,
       "step": 3243
     },
     {
       "epoch": 58.98,
-      "learning_rate": 0.0002261,
-      "loss": 0.1255,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 0.1264,
       "step": 3244
     },
     {
       "epoch": 59.0,
-      "learning_rate": 0.00022600000000000002,
-      "loss": 0.1745,
+      "learning_rate": 0.0002259,
+      "loss": 0.1364,
       "step": 3245
     },
     {
       "epoch": 59.02,
-      "learning_rate": 0.0002259,
-      "loss": 0.1553,
+      "learning_rate": 0.0002258,
+      "loss": 0.1339,
       "step": 3246
     },
     {
       "epoch": 59.04,
-      "learning_rate": 0.0002258,
-      "loss": 0.1731,
+      "learning_rate": 0.0002257,
+      "loss": 0.1561,
       "step": 3247
     },
     {
       "epoch": 59.05,
-      "learning_rate": 0.0002257,
-      "loss": 0.143,
+      "learning_rate": 0.0002256,
+      "loss": 0.1486,
       "step": 3248
     },
     {
       "epoch": 59.07,
-      "learning_rate": 0.0002256,
-      "loss": 0.1345,
+      "learning_rate": 0.0002255,
+      "loss": 0.1311,
       "step": 3249
     },
     {
       "epoch": 59.09,
-      "learning_rate": 0.0002255,
-      "loss": 0.1529,
+      "learning_rate": 0.0002254,
+      "loss": 0.142,
       "step": 3250
     },
     {
       "epoch": 59.11,
-      "learning_rate": 0.0002254,
-      "loss": 0.1273,
+      "learning_rate": 0.0002253,
+      "loss": 0.1256,
       "step": 3251
     },
     {
       "epoch": 59.13,
-      "learning_rate": 0.0002253,
-      "loss": 0.15,
+      "learning_rate": 0.00022520000000000003,
+      "loss": 0.1179,
       "step": 3252
     },
     {
       "epoch": 59.15,
-      "learning_rate": 0.00022520000000000003,
-      "loss": 0.1606,
+      "learning_rate": 0.0002251,
+      "loss": 0.1523,
       "step": 3253
     },
     {
       "epoch": 59.16,
-      "learning_rate": 0.0002251,
-      "loss": 0.1687,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 0.1454,
       "step": 3254
     },
     {
       "epoch": 59.18,
-      "learning_rate": 0.00022500000000000002,
-      "loss": 0.1194,
+      "learning_rate": 0.0002249,
+      "loss": 0.1325,
       "step": 3255
     },
     {
       "epoch": 59.2,
-      "learning_rate": 0.0002249,
-      "loss": 0.2026,
+      "learning_rate": 0.00022480000000000002,
+      "loss": 0.1868,
       "step": 3256
     },
     {
       "epoch": 59.22,
-      "learning_rate": 0.00022480000000000002,
-      "loss": 0.1563,
+      "learning_rate": 0.00022470000000000001,
+      "loss": 0.1209,
       "step": 3257
     },
     {
       "epoch": 59.24,
-      "learning_rate": 0.00022470000000000001,
-      "loss": 0.1565,
+      "learning_rate": 0.0002246,
+      "loss": 0.1576,
       "step": 3258
     },
     {
       "epoch": 59.25,
-      "learning_rate": 0.0002246,
-      "loss": 0.1515,
+      "learning_rate": 0.0002245,
+      "loss": 0.1202,
       "step": 3259
     },
     {
       "epoch": 59.27,
-      "learning_rate": 0.0002245,
-      "loss": 0.1774,
+      "learning_rate": 0.0002244,
+      "loss": 0.1412,
       "step": 3260
     },
     {
       "epoch": 59.29,
-      "learning_rate": 0.0002244,
-      "loss": 0.1886,
+      "learning_rate": 0.0002243,
+      "loss": 0.1485,
       "step": 3261
     },
     {
       "epoch": 59.31,
-      "learning_rate": 0.0002243,
-      "loss": 0.2104,
+      "learning_rate": 0.00022420000000000003,
+      "loss": 0.1838,
       "step": 3262
     },
     {
       "epoch": 59.33,
-      "learning_rate": 0.00022420000000000003,
-      "loss": 0.1743,
+      "learning_rate": 0.0002241,
+      "loss": 0.1378,
       "step": 3263
     },
     {
       "epoch": 59.35,
-      "learning_rate": 0.0002241,
-      "loss": 0.1727,
+      "learning_rate": 0.000224,
+      "loss": 0.1881,
       "step": 3264
     },
     {
       "epoch": 59.36,
-      "learning_rate": 0.000224,
-      "loss": 0.1507,
+      "learning_rate": 0.0002239,
+      "loss": 0.1388,
       "step": 3265
     },
     {
       "epoch": 59.38,
-      "learning_rate": 0.0002239,
-      "loss": 0.14,
+      "learning_rate": 0.0002238,
+      "loss": 0.127,
       "step": 3266
     },
     {
       "epoch": 59.4,
-      "learning_rate": 0.0002238,
-      "loss": 0.1858,
+      "learning_rate": 0.00022370000000000002,
+      "loss": 0.159,
       "step": 3267
     },
     {
       "epoch": 59.42,
-      "learning_rate": 0.00022370000000000002,
-      "loss": 0.2025,
+      "learning_rate": 0.0002236,
+      "loss": 0.1956,
       "step": 3268
     },
     {
       "epoch": 59.44,
-      "learning_rate": 0.0002236,
-      "loss": 0.2122,
+      "learning_rate": 0.0002235,
+      "loss": 0.1555,
       "step": 3269
     },
     {
       "epoch": 59.45,
-      "learning_rate": 0.0002235,
-      "loss": 0.1368,
+      "learning_rate": 0.00022339999999999998,
+      "loss": 0.103,
       "step": 3270
     },
     {
       "epoch": 59.47,
-      "learning_rate": 0.00022339999999999998,
-      "loss": 0.2011,
+      "learning_rate": 0.0002233,
+      "loss": 0.1749,
       "step": 3271
     },
     {
       "epoch": 59.49,
-      "learning_rate": 0.0002233,
-      "loss": 0.1455,
+      "learning_rate": 0.0002232,
+      "loss": 0.1256,
       "step": 3272
     },
     {
       "epoch": 59.51,
-      "learning_rate": 0.0002232,
-      "loss": 0.154,
+      "learning_rate": 0.0002231,
+      "loss": 0.1867,
       "step": 3273
     },
     {
       "epoch": 59.53,
-      "learning_rate": 0.0002231,
-      "loss": 0.1747,
+      "learning_rate": 0.000223,
+      "loss": 0.1638,
       "step": 3274
     },
     {
       "epoch": 59.55,
-      "learning_rate": 0.000223,
-      "loss": 0.1607,
+      "learning_rate": 0.0002229,
+      "loss": 0.1844,
       "step": 3275
     },
     {
       "epoch": 59.56,
-      "learning_rate": 0.0002229,
-      "loss": 0.1744,
+      "learning_rate": 0.0002228,
+      "loss": 0.1464,
       "step": 3276
     },
     {
       "epoch": 59.58,
-      "learning_rate": 0.0002228,
-      "loss": 0.1702,
+      "learning_rate": 0.00022270000000000002,
+      "loss": 0.1605,
       "step": 3277
     },
     {
       "epoch": 59.6,
-      "learning_rate": 0.00022270000000000002,
-      "loss": 0.1831,
+      "learning_rate": 0.0002226,
+      "loss": 0.1574,
       "step": 3278
     },
     {
       "epoch": 59.62,
-      "learning_rate": 0.0002226,
-      "loss": 0.1431,
-      "step": 3279
+      "learning_rate": 0.00022250000000000001,
+      "loss": 0.1421,
+      "step": 3279
     },
     {
       "epoch": 59.64,
-      "learning_rate": 0.00022250000000000001,
-      "loss": 0.2102,
+      "learning_rate": 0.00022239999999999998,
+      "loss": 0.1734,
       "step": 3280
     },
     {
       "epoch": 59.65,
-      "learning_rate": 0.00022239999999999998,
-      "loss": 0.1376,
+      "learning_rate": 0.0002223,
+      "loss": 0.1358,
       "step": 3281
     },
     {
       "epoch": 59.67,
-      "learning_rate": 0.0002223,
-      "loss": 0.1654,
+      "learning_rate": 0.0002222,
+      "loss": 0.1558,
       "step": 3282
     },
     {
       "epoch": 59.69,
-      "learning_rate": 0.0002222,
-      "loss": 0.1281,
+      "learning_rate": 0.0002221,
+      "loss": 0.1498,
       "step": 3283
     },
     {
       "epoch": 59.71,
-      "learning_rate": 0.0002221,
-      "loss": 0.1625,
+      "learning_rate": 0.000222,
+      "loss": 0.1508,
       "step": 3284
     },
     {
       "epoch": 59.73,
-      "learning_rate": 0.000222,
-      "loss": 0.2001,
+      "learning_rate": 0.0002219,
+      "loss": 0.1714,
       "step": 3285
     },
     {
       "epoch": 59.75,
-      "learning_rate": 0.0002219,
-      "loss": 0.1684,
+      "learning_rate": 0.0002218,
+      "loss": 0.1879,
       "step": 3286
     },
     {
       "epoch": 59.76,
-      "learning_rate": 0.0002218,
-      "loss": 0.1791,
+      "learning_rate": 0.00022170000000000002,
+      "loss": 0.1284,
       "step": 3287
     },
     {
       "epoch": 59.78,
-      "learning_rate": 0.00022170000000000002,
-      "loss": 0.1523,
+      "learning_rate": 0.0002216,
+      "loss": 0.1518,
       "step": 3288
     },
     {
       "epoch": 59.8,
-      "learning_rate": 0.0002216,
-      "loss": 0.1617,
+      "learning_rate": 0.00022150000000000002,
+      "loss": 0.1376,
       "step": 3289
     },
     {
       "epoch": 59.82,
-      "learning_rate": 0.00022150000000000002,
-      "loss": 0.1444,
+      "learning_rate": 0.00022140000000000001,
+      "loss": 0.1306,
       "step": 3290
     },
     {
       "epoch": 59.84,
-      "learning_rate": 0.00022140000000000001,
-      "loss": 0.1893,
+      "learning_rate": 0.0002213,
+      "loss": 0.1796,
       "step": 3291
     },
     {
       "epoch": 59.85,
-      "learning_rate": 0.0002213,
-      "loss": 0.127,
+      "learning_rate": 0.0002212,
+      "loss": 0.1453,
       "step": 3292
     },
     {
       "epoch": 59.87,
-      "learning_rate": 0.0002212,
-      "loss": 0.1252,
+      "learning_rate": 0.0002211,
+      "loss": 0.1108,
       "step": 3293
     },
     {
       "epoch": 59.89,
-      "learning_rate": 0.0002211,
-      "loss": 0.1359,
+      "learning_rate": 0.000221,
+      "loss": 0.1526,
       "step": 3294
     },
     {
       "epoch": 59.91,
-      "learning_rate": 0.000221,
-      "loss": 0.1005,
+      "learning_rate": 0.00022090000000000003,
+      "loss": 0.1033,
       "step": 3295
     },
     {
       "epoch": 59.93,
-      "learning_rate": 0.00022090000000000003,
-      "loss": 0.1718,
+      "learning_rate": 0.0002208,
+      "loss": 0.1696,
       "step": 3296
     },
     {
       "epoch": 59.95,
-      "learning_rate": 0.0002208,
-      "loss": 0.1533,
+      "learning_rate": 0.00022070000000000002,
+      "loss": 0.1486,
       "step": 3297
     },
     {
       "epoch": 59.96,
-      "learning_rate": 0.00022070000000000002,
-      "loss": 0.1649,
+      "learning_rate": 0.0002206,
+      "loss": 0.1214,
       "step": 3298
     },
     {
       "epoch": 59.98,
-      "learning_rate": 0.0002206,
-      "loss": 0.1533,
+      "learning_rate": 0.0002205,
+      "loss": 0.1235,
       "step": 3299
     },
     {
       "epoch": 60.0,
-      "learning_rate": 0.0002205,
-      "loss": 0.0978,
+      "learning_rate": 0.00022040000000000002,
+      "loss": 0.0944,
       "step": 3300
     },
     {
       "epoch": 60.02,
-      "learning_rate": 0.00022040000000000002,
-      "loss": 0.2027,
+      "learning_rate": 0.0002203,
+      "loss": 0.1748,
       "step": 3301
     },
     {
       "epoch": 60.04,
-      "learning_rate": 0.0002203,
-      "loss": 0.1342,
+      "learning_rate": 0.0002202,
+      "loss": 0.1168,
       "step": 3302
     },
     {
       "epoch": 60.05,
-      "learning_rate": 0.0002202,
-      "loss": 0.1453,
+      "learning_rate": 0.00022009999999999998,
+      "loss": 0.1292,
       "step": 3303
     },
     {
       "epoch": 60.07,
-      "learning_rate": 0.00022009999999999998,
-      "loss": 0.1557,
+      "learning_rate": 0.00022,
+      "loss": 0.1421,
       "step": 3304
     },
     {
       "epoch": 60.09,
-      "learning_rate": 0.00022,
-      "loss": 0.1706,
+      "learning_rate": 0.0002199,
+      "loss": 0.142,
       "step": 3305
     },
     {
       "epoch": 60.11,
-      "learning_rate": 0.0002199,
-      "loss": 0.1186,
+      "learning_rate": 0.0002198,
+      "loss": 0.1369,
       "step": 3306
     },
     {
       "epoch": 60.13,
-      "learning_rate": 0.0002198,
-      "loss": 0.145,
+      "learning_rate": 0.0002197,
+      "loss": 0.1367,
       "step": 3307
     },
     {
       "epoch": 60.15,
-      "learning_rate": 0.0002197,
-      "loss": 0.1761,
+      "learning_rate": 0.0002196,
+      "loss": 0.1681,
       "step": 3308
     },
     {
       "epoch": 60.16,
-      "learning_rate": 0.0002196,
-      "loss": 0.1562,
+      "learning_rate": 0.0002195,
+      "loss": 0.1412,
       "step": 3309
     },
     {
       "epoch": 60.18,
-      "learning_rate": 0.0002195,
-      "loss": 0.1503,
+      "learning_rate": 0.00021940000000000002,
+      "loss": 0.1095,
       "step": 3310
     },
     {
       "epoch": 60.2,
-      "learning_rate": 0.00021940000000000002,
-      "loss": 0.1289,
+      "learning_rate": 0.0002193,
+      "loss": 0.1063,
       "step": 3311
     },
     {
       "epoch": 60.22,
-      "learning_rate": 0.0002193,
-      "loss": 0.1447,
+      "learning_rate": 0.00021920000000000002,
+      "loss": 0.1179,
       "step": 3312
     },
     {
       "epoch": 60.24,
-      "learning_rate": 0.00021920000000000002,
-      "loss": 0.1314,
+      "learning_rate": 0.00021909999999999999,
+      "loss": 0.138,
       "step": 3313
     },
     {
       "epoch": 60.25,
-      "learning_rate": 0.00021909999999999999,
-      "loss": 0.1597,
+      "learning_rate": 0.000219,
+      "loss": 0.1608,
       "step": 3314
     },
     {
       "epoch": 60.27,
-      "learning_rate": 0.000219,
-      "loss": 0.1663,
+      "learning_rate": 0.0002189,
+      "loss": 0.1726,
       "step": 3315
     },
     {
       "epoch": 60.29,
-      "learning_rate": 0.0002189,
-      "loss": 0.149,
+      "learning_rate": 0.0002188,
+      "loss": 0.1398,
       "step": 3316
     },
     {
       "epoch": 60.31,
-      "learning_rate": 0.0002188,
-      "loss": 0.1666,
+      "learning_rate": 0.0002187,
+      "loss": 0.1601,
       "step": 3317
     },
     {
       "epoch": 60.33,
-      "learning_rate": 0.0002187,
-      "loss": 0.1576,
+      "learning_rate": 0.0002186,
+      "loss": 0.1695,
       "step": 3318
     },
     {
       "epoch": 60.35,
-      "learning_rate": 0.0002186,
-      "loss": 0.1394,
+      "learning_rate": 0.0002185,
+      "loss": 0.165,
       "step": 3319
     },
     {
       "epoch": 60.36,
-      "learning_rate": 0.0002185,
-      "loss": 0.2112,
+      "learning_rate": 0.00021840000000000002,
+      "loss": 0.2056,
       "step": 3320
     },
     {
       "epoch": 60.38,
-      "learning_rate": 0.00021840000000000002,
-      "loss": 0.1593,
+      "learning_rate": 0.0002183,
+      "loss": 0.1616,
       "step": 3321
     },
     {
       "epoch": 60.4,
-      "learning_rate": 0.0002183,
-      "loss": 0.1634,
+      "learning_rate": 0.00021820000000000002,
+      "loss": 0.1519,
       "step": 3322
     },
     {
       "epoch": 60.42,
-      "learning_rate": 0.00021820000000000002,
-      "loss": 0.1319,
+      "learning_rate": 0.0002181,
+      "loss": 0.1566,
       "step": 3323
     },
     {
       "epoch": 60.44,
-      "learning_rate": 0.0002181,
-      "loss": 0.1619,
+      "learning_rate": 0.000218,
+      "loss": 0.1265,
       "step": 3324
     },
     {
       "epoch": 60.45,
-      "learning_rate": 0.000218,
-      "loss": 0.0935,
+      "learning_rate": 0.0002179,
+      "loss": 0.0937,
       "step": 3325
     },
     {
       "epoch": 60.47,
-      "learning_rate": 0.0002179,
-      "loss": 0.2333,
+      "learning_rate": 0.0002178,
+      "loss": 0.2069,
       "step": 3326
     },
     {
       "epoch": 60.49,
-      "learning_rate": 0.0002178,
-      "loss": 0.1898,
+      "learning_rate": 0.0002177,
+      "loss": 0.1561,
       "step": 3327
     },
     {
       "epoch": 60.51,
-      "learning_rate": 0.0002177,
-      "loss": 0.1576,
+      "learning_rate": 0.0002176,
+      "loss": 0.1493,
       "step": 3328
     },
     {
       "epoch": 60.53,
-      "learning_rate": 0.0002176,
-      "loss": 0.1642,
+      "learning_rate": 0.0002175,
+      "loss": 0.1432,
       "step": 3329
     },
     {
       "epoch": 60.55,
-      "learning_rate": 0.0002175,
-      "loss": 0.1276,
+      "learning_rate": 0.00021740000000000003,
+      "loss": 0.1138,
       "step": 3330
     },
     {
       "epoch": 60.56,
-      "learning_rate": 0.00021740000000000003,
-      "loss": 0.109,
+      "learning_rate": 0.0002173,
+      "loss": 0.1127,
       "step": 3331
     },
     {
       "epoch": 60.58,
-      "learning_rate": 0.0002173,
-      "loss": 0.1432,
+      "learning_rate": 0.00021720000000000002,
+      "loss": 0.1535,
       "step": 3332
     },
     {
       "epoch": 60.6,
-      "learning_rate": 0.00021720000000000002,
-      "loss": 0.204,
+      "learning_rate": 0.0002171,
+      "loss": 0.1806,
       "step": 3333
     },
     {
       "epoch": 60.62,
-      "learning_rate": 0.0002171,
-      "loss": 0.1409,
+      "learning_rate": 0.00021700000000000002,
+      "loss": 0.1181,
       "step": 3334
     },
     {
       "epoch": 60.64,
-      "learning_rate": 0.00021700000000000002,
-      "loss": 0.184,
+      "learning_rate": 0.00021690000000000001,
+      "loss": 0.189,
       "step": 3335
     },
     {
       "epoch": 60.65,
-      "learning_rate": 0.00021690000000000001,
-      "loss": 0.1825,
+      "learning_rate": 0.00021679999999999998,
+      "loss": 0.1652,
       "step": 3336
     },
     {
       "epoch": 60.67,
-      "learning_rate": 0.00021679999999999998,
-      "loss": 0.1448,
+      "learning_rate": 0.0002167,
+      "loss": 0.1155,
       "step": 3337
     },
     {
       "epoch": 60.69,
-      "learning_rate": 0.0002167,
-      "loss": 0.138,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 0.1119,
       "step": 3338
     },
     {
       "epoch": 60.71,
-      "learning_rate": 0.00021659999999999998,
-      "loss": 0.1387,
+      "learning_rate": 0.0002165,
+      "loss": 0.1454,
       "step": 3339
     },
     {
       "epoch": 60.73,
-      "learning_rate": 0.0002165,
-      "loss": 0.1586,
+      "learning_rate": 0.0002164,
+      "loss": 0.1411,
       "step": 3340
     },
     {
       "epoch": 60.75,
-      "learning_rate": 0.0002164,
-      "loss": 0.1749,
+      "learning_rate": 0.0002163,
+      "loss": 0.1693,
       "step": 3341
     },
     {
       "epoch": 60.76,
-      "learning_rate": 0.0002163,
-      "loss": 0.1584,
+      "learning_rate": 0.0002162,
+      "loss": 0.1673,
       "step": 3342
     },
     {
       "epoch": 60.78,
-      "learning_rate": 0.0002162,
-      "loss": 0.1767,
+      "learning_rate": 0.0002161,
+      "loss": 0.1172,
       "step": 3343
     },
     {
       "epoch": 60.8,
-      "learning_rate": 0.0002161,
-      "loss": 0.1321,
+      "learning_rate": 0.000216,
+      "loss": 0.1381,
       "step": 3344
     },
     {
       "epoch": 60.82,
-      "learning_rate": 0.000216,
-      "loss": 0.1692,
+      "learning_rate": 0.00021590000000000002,
+      "loss": 0.1841,
       "step": 3345
     },
     {
       "epoch": 60.84,
-      "learning_rate": 0.00021590000000000002,
-      "loss": 0.1265,
+      "learning_rate": 0.0002158,
+      "loss": 0.1467,
       "step": 3346
     },
     {
       "epoch": 60.85,
-      "learning_rate": 0.0002158,
-      "loss": 0.1685,
+      "learning_rate": 0.0002157,
+      "loss": 0.119,
       "step": 3347
     },
     {
       "epoch": 60.87,
-      "learning_rate": 0.0002157,
-      "loss": 0.2136,
+      "learning_rate": 0.0002156,
+      "loss": 0.1985,
       "step": 3348
     },
     {
       "epoch": 60.89,
-      "learning_rate": 0.0002156,
-      "loss": 0.1678,
+      "learning_rate": 0.0002155,
+      "loss": 0.1515,
       "step": 3349
     },
     {
       "epoch": 60.91,
-      "learning_rate": 0.0002155,
-      "loss": 0.1415,
+      "learning_rate": 0.0002154,
+      "loss": 0.094,
       "step": 3350
     },
     {
       "epoch": 60.93,
-      "learning_rate": 0.0002154,
-      "loss": 0.1607,
+      "learning_rate": 0.0002153,
+      "loss": 0.1554,
       "step": 3351
     },
     {
       "epoch": 60.95,
-      "learning_rate": 0.0002153,
-      "loss": 0.1707,
+      "learning_rate": 0.0002152,
+      "loss": 0.1357,
       "step": 3352
     },
     {
       "epoch": 60.96,
-      "learning_rate": 0.0002152,
-      "loss": 0.1686,
+      "learning_rate": 0.00021510000000000002,
+      "loss": 0.1527,
       "step": 3353
     },
     {
       "epoch": 60.98,
-      "learning_rate": 0.00021510000000000002,
-      "loss": 0.1239,
+      "learning_rate": 0.000215,
+      "loss": 0.1035,
       "step": 3354
     },
     {
       "epoch": 61.0,
-      "learning_rate": 0.000215,
-      "loss": 0.1357,
+      "learning_rate": 0.00021490000000000002,
+      "loss": 0.1366,
       "step": 3355
     },
     {
       "epoch": 61.02,
-      "learning_rate": 0.00021490000000000002,
-      "loss": 0.1563,
+      "learning_rate": 0.0002148,
+      "loss": 0.1465,
       "step": 3356
     },
     {
       "epoch": 61.04,
-      "learning_rate": 0.0002148,
-      "loss": 0.1716,
+      "learning_rate": 0.00021470000000000001,
+      "loss": 0.1632,
       "step": 3357
     },
     {
       "epoch": 61.05,
-      "learning_rate": 0.00021470000000000001,
-      "loss": 0.2118,
+      "learning_rate": 0.0002146,
+      "loss": 0.194,
       "step": 3358
     },
     {
       "epoch": 61.07,
-      "learning_rate": 0.0002146,
-      "loss": 0.1633,
+      "learning_rate": 0.0002145,
+      "loss": 0.1317,
       "step": 3359
     },
     {
       "epoch": 61.09,
-      "learning_rate": 0.0002145,
-      "loss": 0.1277,
+      "learning_rate": 0.0002144,
+      "loss": 0.1369,
       "step": 3360
     },
     {
       "epoch": 61.11,
-      "learning_rate": 0.0002144,
-      "loss": 0.1752,
+      "learning_rate": 0.0002143,
+      "loss": 0.1763,
       "step": 3361
     },
     {
       "epoch": 61.13,
-      "learning_rate": 0.0002143,
-      "loss": 0.1623,
+      "learning_rate": 0.0002142,
+      "loss": 0.1606,
       "step": 3362
     },
     {
       "epoch": 61.15,
-      "learning_rate": 0.0002142,
-      "loss": 0.1681,
+      "learning_rate": 0.00021410000000000003,
+      "loss": 0.173,
       "step": 3363
     },
     {
       "epoch": 61.16,
-      "learning_rate": 0.00021410000000000003,
-      "loss": 0.1581,
+      "learning_rate": 0.000214,
+      "loss": 0.1562,
       "step": 3364
     },
     {
       "epoch": 61.18,
-      "learning_rate": 0.000214,
-      "loss": 0.1472,
+      "learning_rate": 0.00021390000000000002,
+      "loss": 0.146,
       "step": 3365
     },
     {
       "epoch": 61.2,
-      "learning_rate": 0.00021390000000000002,
-      "loss": 0.1458,
+      "learning_rate": 0.0002138,
+      "loss": 0.1625,
       "step": 3366
     },
     {
       "epoch": 61.22,
-      "learning_rate": 0.0002138,
-      "loss": 0.1349,
+      "learning_rate": 0.00021370000000000002,
+      "loss": 0.1397,
       "step": 3367
     },
     {
       "epoch": 61.24,
-      "learning_rate": 0.00021370000000000002,
-      "loss": 0.1458,
+      "learning_rate": 0.00021360000000000001,
+      "loss": 0.1579,
       "step": 3368
     },
     {
       "epoch": 61.25,
-      "learning_rate": 0.00021360000000000001,
-      "loss": 0.1636,
+      "learning_rate": 0.0002135,
+      "loss": 0.1364,
       "step": 3369
     },
     {
       "epoch": 61.27,
-      "learning_rate": 0.0002135,
-      "loss": 0.1201,
+      "learning_rate": 0.0002134,
+      "loss": 0.1098,
       "step": 3370
     },
     {
       "epoch": 61.29,
-      "learning_rate": 0.0002134,
-      "loss": 0.1549,
+      "learning_rate": 0.00021329999999999998,
+      "loss": 0.156,
       "step": 3371
     },
     {
       "epoch": 61.31,
-      "learning_rate": 0.00021329999999999998,
-      "loss": 0.1487,
+      "learning_rate": 0.0002132,
+      "loss": 0.137,
       "step": 3372
     },
     {
       "epoch": 61.33,
-      "learning_rate": 0.0002132,
-      "loss": 0.1134,
+      "learning_rate": 0.0002131,
+      "loss": 0.0935,
       "step": 3373
     },
     {
       "epoch": 61.35,
-      "learning_rate": 0.0002131,
-      "loss": 0.1265,
+      "learning_rate": 0.000213,
+      "loss": 0.1438,
       "step": 3374
     },
     {
       "epoch": 61.36,
-      "learning_rate": 0.000213,
-      "loss": 0.148,
+      "learning_rate": 0.0002129,
+      "loss": 0.1208,
       "step": 3375
     },
     {
       "epoch": 61.38,
-      "learning_rate": 0.0002129,
-      "loss": 0.1657,
+      "learning_rate": 0.0002128,
+      "loss": 0.1587,
       "step": 3376
     },
     {
       "epoch": 61.4,
-      "learning_rate": 0.0002128,
-      "loss": 0.2187,
+      "learning_rate": 0.0002127,
+      "loss": 0.2102,
       "step": 3377
     },
     {
       "epoch": 61.42,
-      "learning_rate": 0.0002127,
-      "loss": 0.1351,
+      "learning_rate": 0.00021260000000000002,
+      "loss": 0.1506,
       "step": 3378
     },
     {
       "epoch": 61.44,
-      "learning_rate": 0.00021260000000000002,
-      "loss": 0.1622,
+      "learning_rate": 0.0002125,
+      "loss": 0.1433,
       "step": 3379
     },
     {
       "epoch": 61.45,
-      "learning_rate": 0.0002125,
-      "loss": 0.1595,
+      "learning_rate": 0.0002124,
+      "loss": 0.1475,
       "step": 3380
     },
     {
       "epoch": 61.47,
-      "learning_rate": 0.0002124,
-      "loss": 0.1395,
+      "learning_rate": 0.00021229999999999998,
+      "loss": 0.1377,
       "step": 3381
     },
     {
       "epoch": 61.49,
-      "learning_rate": 0.00021229999999999998,
-      "loss": 0.1583,
+      "learning_rate": 0.0002122,
+      "loss": 0.13,
       "step": 3382
     },
     {
       "epoch": 61.51,
-      "learning_rate": 0.0002122,
-      "loss": 0.1386,
+      "learning_rate": 0.0002121,
+      "loss": 0.1134,
       "step": 3383
     },
     {
       "epoch": 61.53,
-      "learning_rate": 0.0002121,
-      "loss": 0.1567,
+      "learning_rate": 0.000212,
+      "loss": 0.1355,
       "step": 3384
     },
     {
       "epoch": 61.55,
-      "learning_rate": 0.000212,
-      "loss": 0.1403,
+      "learning_rate": 0.0002119,
+      "loss": 0.1233,
       "step": 3385
     },
     {
       "epoch": 61.56,
-      "learning_rate": 0.0002119,
-      "loss": 0.1744,
+      "learning_rate": 0.0002118,
+      "loss": 0.1765,
       "step": 3386
     },
     {
       "epoch": 61.58,
-      "learning_rate": 0.0002118,
-      "loss": 0.1342,
+      "learning_rate": 0.0002117,
+      "loss": 0.1408,
       "step": 3387
     },
     {
       "epoch": 61.6,
-      "learning_rate": 0.0002117,
-      "loss": 0.215,
+      "learning_rate": 0.00021160000000000002,
+      "loss": 0.1745,
       "step": 3388
     },
     {
       "epoch": 61.62,
-      "learning_rate": 0.00021160000000000002,
-      "loss": 0.1473,
+      "learning_rate": 0.0002115,
+      "loss": 0.1326,
       "step": 3389
     },
     {
       "epoch": 61.64,
-      "learning_rate": 0.0002115,
-      "loss": 0.1493,
+      "learning_rate": 0.00021140000000000002,
+      "loss": 0.1599,
       "step": 3390
     },
     {
       "epoch": 61.65,
-      "learning_rate": 0.00021140000000000002,
-      "loss": 0.1996,
+      "learning_rate": 0.00021129999999999999,
+      "loss": 0.1663,
       "step": 3391
     },
     {
       "epoch": 61.67,
-      "learning_rate": 0.00021129999999999999,
-      "loss": 0.1685,
+      "learning_rate": 0.0002112,
+      "loss": 0.1284,
       "step": 3392
     },
     {
       "epoch": 61.69,
-      "learning_rate": 0.0002112,
-      "loss": 0.148,
+      "learning_rate": 0.0002111,
+      "loss": 0.1847,
       "step": 3393
     },
     {
       "epoch": 61.71,
-      "learning_rate": 0.0002111,
-      "loss": 0.1896,
+      "learning_rate": 0.000211,
+      "loss": 0.1702,
       "step": 3394
     },
     {
       "epoch": 61.73,
-      "learning_rate": 0.000211,
-      "loss": 0.1626,
+      "learning_rate": 0.0002109,
+      "loss": 0.1636,
       "step": 3395
     },
     {
       "epoch": 61.75,
-      "learning_rate": 0.0002109,
-      "loss": 0.1656,
+      "learning_rate": 0.0002108,
+      "loss": 0.144,
       "step": 3396
     },
     {
       "epoch": 61.76,
-      "learning_rate": 0.0002108,
-      "loss": 0.1194,
+      "learning_rate": 0.0002107,
+      "loss": 0.1363,
       "step": 3397
     },
     {
       "epoch": 61.78,
-      "learning_rate": 0.0002107,
-      "loss": 0.1424,
+      "learning_rate": 0.00021060000000000002,
+      "loss": 0.1263,
       "step": 3398
     },
     {
       "epoch": 61.8,
-      "learning_rate": 0.00021060000000000002,
-      "loss": 0.1509,
+      "learning_rate": 0.0002105,
+      "loss": 0.1488,
       "step": 3399
     },
     {
       "epoch": 61.82,
-      "learning_rate": 0.0002105,
-      "loss": 0.144,
+      "learning_rate": 0.00021040000000000002,
+      "loss": 0.1447,
       "step": 3400
     },
     {
       "epoch": 61.84,
-      "learning_rate": 0.00021040000000000002,
-      "loss": 0.1552,
+      "learning_rate": 0.0002103,
+      "loss": 0.1407,
       "step": 3401
     },
     {
       "epoch": 61.85,
-      "learning_rate": 0.0002103,
-      "loss": 0.1595,
+      "learning_rate": 0.00021020000000000001,
+      "loss": 0.15,
       "step": 3402
     },
     {
       "epoch": 61.87,
-      "learning_rate": 0.00021020000000000001,
-      "loss": 0.1608,
+      "learning_rate": 0.0002101,
+      "loss": 0.1568,
       "step": 3403
     },
     {
       "epoch": 61.89,
-      "learning_rate": 0.0002101,
-      "loss": 0.1481,
+      "learning_rate": 0.00021,
+      "loss": 0.1596,
       "step": 3404
     },
     {
       "epoch": 61.91,
-      "learning_rate": 0.00021,
-      "loss": 0.1086,
+      "learning_rate": 0.0002099,
+      "loss": 0.1056,
       "step": 3405
     },
     {
       "epoch": 61.93,
-      "learning_rate": 0.0002099,
-      "loss": 0.2072,
+      "learning_rate": 0.0002098,
+      "loss": 0.1945,
       "step": 3406
     },
     {
       "epoch": 61.95,
-      "learning_rate": 0.0002098,
-      "loss": 0.1934,
+      "learning_rate": 0.0002097,
+      "loss": 0.1735,
       "step": 3407
     },
     {
       "epoch": 61.96,
-      "learning_rate": 0.0002097,
-      "loss": 0.1581,
+      "learning_rate": 0.00020960000000000003,
+      "loss": 0.1588,
       "step": 3408
     },
     {
       "epoch": 61.98,
-      "learning_rate": 0.00020960000000000003,
-      "loss": 0.11,
+      "learning_rate": 0.0002095,
+      "loss": 0.1128,
       "step": 3409
     },
     {
       "epoch": 62.0,
-      "learning_rate": 0.0002095,
-      "loss": 0.1255,
+      "learning_rate": 0.0002094,
+      "loss": 0.1313,
       "step": 3410
     },
     {
       "epoch": 62.02,
-      "learning_rate": 0.0002094,
-      "loss": 0.1704,
+      "learning_rate": 0.00020930000000000002,
+      "loss": 0.1862,
       "step": 3411
     },
     {
       "epoch": 62.04,
-      "learning_rate": 0.00020930000000000002,
-      "loss": 0.1587,
+      "learning_rate": 0.0002092,
+      "loss": 0.1604,
       "step": 3412
     },
     {
       "epoch": 62.05,
-      "learning_rate": 0.0002092,
-      "loss": 0.151,
+      "learning_rate": 0.00020910000000000001,
+      "loss": 0.1608,
       "step": 3413
     },
     {
       "epoch": 62.07,
-      "learning_rate": 0.00020910000000000001,
-      "loss": 0.1685,
+      "learning_rate": 0.00020899999999999998,
+      "loss": 0.1325,
       "step": 3414
     },
     {
       "epoch": 62.09,
-      "learning_rate": 0.00020899999999999998,
-      "loss": 0.1387,
+      "learning_rate": 0.0002089,
+      "loss": 0.1426,
       "step": 3415
     },
     {
       "epoch": 62.11,
-      "learning_rate": 0.0002089,
-      "loss": 0.1293,
+      "learning_rate": 0.0002088,
+      "loss": 0.1433,
       "step": 3416
     },
     {
       "epoch": 62.13,
-      "learning_rate": 0.0002088,
-      "loss": 0.1521,
+      "learning_rate": 0.0002087,
+      "loss": 0.1688,
       "step": 3417
     },
     {
       "epoch": 62.15,
-      "learning_rate": 0.0002087,
-      "loss": 0.133,
+      "learning_rate": 0.0002086,
+      "loss": 0.1377,
       "step": 3418
     },
     {
       "epoch": 62.16,
-      "learning_rate": 0.0002086,
-      "loss": 0.1636,
+      "learning_rate": 0.0002085,
+      "loss": 0.1551,
       "step": 3419
     },
     {
       "epoch": 62.18,
-      "learning_rate": 0.0002085,
-      "loss": 0.1956,
+      "learning_rate": 0.0002084,
+      "loss": 0.1833,
       "step": 3420
     },
     {
       "epoch": 62.2,
-      "learning_rate": 0.0002084,
-      "loss": 0.1695,
+      "learning_rate": 0.00020830000000000002,
+      "loss": 0.1891,
       "step": 3421
     },
     {
       "epoch": 62.22,
-      "learning_rate": 0.00020830000000000002,
-      "loss": 0.1425,
+      "learning_rate": 0.0002082,
+      "loss": 0.1252,
       "step": 3422
     },
     {
       "epoch": 62.24,
-      "learning_rate": 0.0002082,
-      "loss": 0.1481,
+      "learning_rate": 0.00020810000000000002,
+      "loss": 0.1243,
       "step": 3423
     },
     {
       "epoch": 62.25,
-      "learning_rate": 0.00020810000000000002,
-      "loss": 0.1567,
+      "learning_rate": 0.000208,
+      "loss": 0.148,
       "step": 3424
     },
     {
       "epoch": 62.27,
-      "learning_rate": 0.000208,
-      "loss": 0.1724,
+      "learning_rate": 0.0002079,
+      "loss": 0.155,
       "step": 3425
     },
     {
       "epoch": 62.29,
-      "learning_rate": 0.0002079,
-      "loss": 0.1145,
+      "learning_rate": 0.0002078,
+      "loss": 0.1127,
       "step": 3426
     },
     {
       "epoch": 62.31,
-      "learning_rate": 0.0002078,
-      "loss": 0.1106,
+      "learning_rate": 0.0002077,
+      "loss": 0.0927,
       "step": 3427
     },
     {
       "epoch": 62.33,
-      "learning_rate": 0.0002077,
-      "loss": 0.1519,
+      "learning_rate": 0.0002076,
+      "loss": 0.1488,
       "step": 3428
     },
     {
       "epoch": 62.35,
-      "learning_rate": 0.0002076,
-      "loss": 0.1932,
+      "learning_rate": 0.0002075,
+      "loss": 0.1754,
       "step": 3429
     },
     {
       "epoch": 62.36,
-      "learning_rate": 0.0002075,
-      "loss": 0.1579,
+      "learning_rate": 0.0002074,
+      "loss": 0.143,
       "step": 3430
     },
     {
       "epoch": 62.38,
-      "learning_rate": 0.0002074,
-      "loss": 0.1154,
+      "learning_rate": 0.00020730000000000002,
+      "loss": 0.1115,
       "step": 3431
     },
     {
       "epoch": 62.4,
-      "learning_rate": 0.00020730000000000002,
-      "loss": 0.1667,
+      "learning_rate": 0.0002072,
+      "loss": 0.1512,
       "step": 3432
     },
     {
       "epoch": 62.42,
-      "learning_rate": 0.0002072,
-      "loss": 0.1686,
+      "learning_rate": 0.00020710000000000002,
+      "loss": 0.1525,
       "step": 3433
     },
     {
       "epoch": 62.44,
-      "learning_rate": 0.00020710000000000002,
-      "loss": 0.1189,
+      "learning_rate": 0.000207,
+      "loss": 0.1543,
       "step": 3434
     },
     {
       "epoch": 62.45,
-      "learning_rate": 0.000207,
-      "loss": 0.1908,
+      "learning_rate": 0.00020690000000000001,
+      "loss": 0.1628,
       "step": 3435
     },
     {
       "epoch": 62.47,
-      "learning_rate": 0.00020690000000000001,
-      "loss": 0.1472,
+      "learning_rate": 0.0002068,
+      "loss": 0.1379,
       "step": 3436
     },
     {
       "epoch": 62.49,
-      "learning_rate": 0.0002068,
-      "loss": 0.1641,
+      "learning_rate": 0.0002067,
+      "loss": 0.1333,
       "step": 3437
     },
     {
       "epoch": 62.51,
-      "learning_rate": 0.0002067,
-      "loss": 0.1328,
+      "learning_rate": 0.0002066,
+      "loss": 0.1208,
       "step": 3438
     },
     {
       "epoch": 62.53,
-      "learning_rate": 0.0002066,
-      "loss": 0.1403,
+      "learning_rate": 0.0002065,
+      "loss": 0.1327,
       "step": 3439
     },
     {
       "epoch": 62.55,
-      "learning_rate": 0.0002065,
-      "loss": 0.1404,
+      "learning_rate": 0.0002064,
+      "loss": 0.1553,
       "step": 3440
     },
     {
       "epoch": 62.56,
-      "learning_rate": 0.0002064,
-      "loss": 0.1649,
+      "learning_rate": 0.00020630000000000003,
+      "loss": 0.1401,
       "step": 3441
     },
     {
       "epoch": 62.58,
-      "learning_rate": 0.00020630000000000003,
-      "loss": 0.1614,
+      "learning_rate": 0.0002062,
+      "loss": 0.1341,
       "step": 3442
     },
     {
       "epoch": 62.6,
-      "learning_rate": 0.0002062,
-      "loss": 0.1887,
+      "learning_rate": 0.00020610000000000002,
+      "loss": 0.1518,
       "step": 3443
     },
     {
       "epoch": 62.62,
-      "learning_rate": 0.00020610000000000002,
-      "loss": 0.1151,
+      "learning_rate": 0.000206,
+      "loss": 0.1127,
       "step": 3444
     },
     {
       "epoch": 62.64,
-      "learning_rate": 0.000206,
-      "loss": 0.1476,
+      "learning_rate": 0.00020590000000000002,
+      "loss": 0.14,
       "step": 3445
     },
     {
       "epoch": 62.65,
-      "learning_rate": 0.00020590000000000002,
-      "loss": 0.0978,
+      "learning_rate": 0.00020580000000000002,
+      "loss": 0.1063,
       "step": 3446
     },
     {
       "epoch": 62.67,
-      "learning_rate": 0.00020580000000000002,
-      "loss": 0.1181,
+      "learning_rate": 0.00020569999999999999,
+      "loss": 0.1313,
       "step": 3447
     },
     {
       "epoch": 62.69,
-      "learning_rate": 0.00020569999999999999,
-      "loss": 0.1421,
+      "learning_rate": 0.0002056,
+      "loss": 0.1142,
       "step": 3448
     },
     {
       "epoch": 62.71,
-      "learning_rate": 0.0002056,
-      "loss": 0.1488,
+      "learning_rate": 0.00020549999999999998,
+      "loss": 0.1292,
       "step": 3449
     },
     {
       "epoch": 62.73,
-      "learning_rate": 0.00020549999999999998,
-      "loss": 0.1283,
+      "learning_rate": 0.0002054,
+      "loss": 0.0979,
       "step": 3450
     },
     {
       "epoch": 62.75,
-      "learning_rate": 0.0002054,
-      "loss": 0.159,
+      "learning_rate": 0.0002053,
+      "loss": 0.1492,
       "step": 3451
     },
     {
       "epoch": 62.76,
-      "learning_rate": 0.0002053,
-      "loss": 0.1544,
+      "learning_rate": 0.0002052,
+      "loss": 0.1283,
       "step": 3452
     },
     {
       "epoch": 62.78,
-      "learning_rate": 0.0002052,
-      "loss": 0.1134,
+      "learning_rate": 0.0002051,
+      "loss": 0.1022,
       "step": 3453
     },
     {
       "epoch": 62.8,
-      "learning_rate": 0.0002051,
-      "loss": 0.1056,
+      "learning_rate": 0.000205,
+      "loss": 0.1133,
       "step": 3454
     },
     {
       "epoch": 62.82,
-      "learning_rate": 0.000205,
-      "loss": 0.1684,
+      "learning_rate": 0.0002049,
+      "loss": 0.1576,
       "step": 3455
     },
     {
       "epoch": 62.84,
-      "learning_rate": 0.0002049,
-      "loss": 0.1337,
+      "learning_rate": 0.00020480000000000002,
+      "loss": 0.1117,
       "step": 3456
     },
     {
       "epoch": 62.85,
-      "learning_rate": 0.00020480000000000002,
-      "loss": 0.1426,
+      "learning_rate": 0.0002047,
+      "loss": 0.1337,
       "step": 3457
     },
     {
       "epoch": 62.87,
-      "learning_rate": 0.0002047,
-      "loss": 0.1223,
+      "learning_rate": 0.0002046,
+      "loss": 0.0966,
       "step": 3458
     },
     {
       "epoch": 62.89,
-      "learning_rate": 0.0002046,
-      "loss": 0.1775,
+      "learning_rate": 0.00020449999999999998,
+      "loss": 0.1248,
       "step": 3459
     },
     {
       "epoch": 62.91,
-      "learning_rate": 0.00020449999999999998,
-      "loss": 0.0958,
+      "learning_rate": 0.0002044,
+      "loss": 0.0954,
       "step": 3460
     },
     {
       "epoch": 62.93,
-      "learning_rate": 0.0002044,
-      "loss": 0.1909,
+      "learning_rate": 0.0002043,
+      "loss": 0.131,
       "step": 3461
     },
     {
       "epoch": 62.95,
-      "learning_rate": 0.0002043,
-      "loss": 0.134,
+      "learning_rate": 0.0002042,
+      "loss": 0.138,
       "step": 3462
     },
     {
       "epoch": 62.96,
-      "learning_rate": 0.0002042,
-      "loss": 0.1835,
+      "learning_rate": 0.0002041,
+      "loss": 0.1552,
       "step": 3463
     },
     {
       "epoch": 62.98,
-      "learning_rate": 0.0002041,
-      "loss": 0.1457,
+      "learning_rate": 0.000204,
+      "loss": 0.134,
       "step": 3464
     },
     {
       "epoch": 63.0,
-      "learning_rate": 0.000204,
-      "loss": 0.1634,
+      "learning_rate": 0.0002039,
+      "loss": 0.1014,
       "step": 3465
     },
     {
       "epoch": 63.02,
-      "learning_rate": 0.0002039,
-      "loss": 0.1522,
+      "learning_rate": 0.00020380000000000002,
+      "loss": 0.1265,
       "step": 3466
     },
     {
       "epoch": 63.04,
-      "learning_rate": 0.00020380000000000002,
-      "loss": 0.1216,
+      "learning_rate": 0.0002037,
+      "loss": 0.1126,
       "step": 3467
     },
     {
       "epoch": 63.05,
-      "learning_rate": 0.0002037,
-      "loss": 0.1256,
+      "learning_rate": 0.00020360000000000002,
+      "loss": 0.109,
       "step": 3468
     },
     {
       "epoch": 63.07,
-      "learning_rate": 0.00020360000000000002,
-      "loss": 0.1654,
+      "learning_rate": 0.00020349999999999999,
+      "loss": 0.1262,
       "step": 3469
     },
     {
       "epoch": 63.09,
-      "learning_rate": 0.00020349999999999999,
-      "loss": 0.1577,
+      "learning_rate": 0.0002034,
+      "loss": 0.154,
       "step": 3470
     },
     {
       "epoch": 63.11,
-      "learning_rate": 0.0002034,
-      "loss": 0.1074,
+      "learning_rate": 0.0002033,
+      "loss": 0.0784,
       "step": 3471
     },
     {
       "epoch": 63.13,
-      "learning_rate": 0.0002033,
-      "loss": 0.1548,
+      "learning_rate": 0.0002032,
+      "loss": 0.1152,
       "step": 3472
     },
     {
       "epoch": 63.15,
-      "learning_rate": 0.0002032,
-      "loss": 0.1497,
+      "learning_rate": 0.0002031,
+      "loss": 0.108,
       "step": 3473
     },
     {
       "epoch": 63.16,
-      "learning_rate": 0.0002031,
-      "loss": 0.1171,
+      "learning_rate": 0.00020300000000000003,
+      "loss": 0.0971,
       "step": 3474
     },
     {
       "epoch": 63.18,
-      "learning_rate": 0.00020300000000000003,
-      "loss": 0.1717,
+      "learning_rate": 0.0002029,
+      "loss": 0.1426,
       "step": 3475
     },
     {
       "epoch": 63.2,
-      "learning_rate": 0.0002029,
-      "loss": 0.1368,
+      "learning_rate": 0.00020280000000000002,
+      "loss": 0.1253,
       "step": 3476
     },
     {
       "epoch": 63.22,
-      "learning_rate": 0.00020280000000000002,
-      "loss": 0.1239,
+      "learning_rate": 0.0002027,
+      "loss": 0.1048,
       "step": 3477
     },
     {
       "epoch": 63.24,
-      "learning_rate": 0.0002027,
-      "loss": 0.1654,
+      "learning_rate": 0.00020260000000000002,
+      "loss": 0.1341,
       "step": 3478
     },
     {
       "epoch": 63.25,
-      "learning_rate": 0.00020260000000000002,
-      "loss": 0.1797,
+      "learning_rate": 0.00020250000000000002,
+      "loss": 0.1835,
       "step": 3479
     },
     {
       "epoch": 63.27,
-      "learning_rate": 0.00020250000000000002,
-      "loss": 0.1325,
+      "learning_rate": 0.00020240000000000001,
+      "loss": 0.1101,
       "step": 3480
     },
     {
       "epoch": 63.29,
-      "learning_rate": 0.00020240000000000001,
-      "loss": 0.1659,
+      "learning_rate": 0.0002023,
+      "loss": 0.169,
       "step": 3481
     },
     {
       "epoch": 63.31,
-      "learning_rate": 0.0002023,
-      "loss": 0.1419,
+      "learning_rate": 0.00020219999999999998,
+      "loss": 0.1198,
       "step": 3482
     },
     {
       "epoch": 63.33,
-      "learning_rate": 0.00020219999999999998,
-      "loss": 0.1526,
+      "learning_rate": 0.0002021,
+      "loss": 0.1584,
       "step": 3483
     },
     {
       "epoch": 63.35,
-      "learning_rate": 0.0002021,
-      "loss": 0.086,
+      "learning_rate": 0.000202,
+      "loss": 0.0818,
       "step": 3484
     },
     {
       "epoch": 63.36,
-      "learning_rate": 0.000202,
-      "loss": 0.18,
+      "learning_rate": 0.0002019,
+      "loss": 0.1475,
       "step": 3485
     },
     {
       "epoch": 63.38,
-      "learning_rate": 0.0002019,
-      "loss": 0.1635,
+      "learning_rate": 0.0002018,
+      "loss": 0.1544,
       "step": 3486
     },
     {
       "epoch": 63.4,
-      "learning_rate": 0.0002018,
-      "loss": 0.1393,
+      "learning_rate": 0.0002017,
+      "loss": 0.1175,
       "step": 3487
     },
     {
       "epoch": 63.42,
-      "learning_rate": 0.0002017,
-      "loss": 0.1412,
+      "learning_rate": 0.0002016,
+      "loss": 0.1068,
       "step": 3488
     },
     {
       "epoch": 63.44,
-      "learning_rate": 0.0002016,
-      "loss": 0.1636,
+      "learning_rate": 0.00020150000000000002,
+      "loss": 0.1544,
       "step": 3489
     },
     {
       "epoch": 63.45,
-      "learning_rate": 0.00020150000000000002,
-      "loss": 0.0824,
+      "learning_rate": 0.0002014,
+      "loss": 0.0961,
       "step": 3490
     },
     {
       "epoch": 63.47,
-      "learning_rate": 0.0002014,
-      "loss": 0.1556,
+      "learning_rate": 0.00020130000000000001,
+      "loss": 0.161,
       "step": 3491
     },
     {
       "epoch": 63.49,
-      "learning_rate": 0.00020130000000000001,
-      "loss": 0.1529,
+      "learning_rate": 0.00020119999999999998,
+      "loss": 0.1711,
       "step": 3492
     },
     {
       "epoch": 63.51,
-      "learning_rate": 0.00020119999999999998,
-      "loss": 0.1572,
+      "learning_rate": 0.0002011,
+      "loss": 0.1171,
       "step": 3493
     },
     {
       "epoch": 63.53,
-      "learning_rate": 0.0002011,
-      "loss": 0.1083,
+      "learning_rate": 0.000201,
+      "loss": 0.1071,
       "step": 3494
     },
     {
       "epoch": 63.55,
-      "learning_rate": 0.000201,
-      "loss": 0.094,
+      "learning_rate": 0.0002009,
+      "loss": 0.1122,
       "step": 3495
     },
     {
       "epoch": 63.56,
-      "learning_rate": 0.0002009,
-      "loss": 0.0936,
+      "learning_rate": 0.0002008,
+      "loss": 0.1431,
       "step": 3496
     },
     {
       "epoch": 63.58,
-      "learning_rate": 0.0002008,
-      "loss": 0.1482,
+      "learning_rate": 0.0002007,
+      "loss": 0.166,
       "step": 3497
     },
     {
       "epoch": 63.6,
-      "learning_rate": 0.0002007,
-      "loss": 0.1902,
+      "learning_rate": 0.0002006,
+      "loss": 0.1558,
       "step": 3498
     },
     {
       "epoch": 63.62,
-      "learning_rate": 0.0002006,
-      "loss": 0.1699,
+      "learning_rate": 0.00020050000000000002,
+      "loss": 0.1402,
       "step": 3499
     },
     {
       "epoch": 63.64,
-      "learning_rate": 0.00020050000000000002,
-      "loss": 0.2094,
+      "learning_rate": 0.0002004,
+      "loss": 0.1898,
       "step": 3500
     },
     {
       "epoch": 63.64,
-      "eval_cer": 0.0864030944991679,
-      "eval_loss": 0.41124406456947327,
-      "eval_runtime": 13.976,
-      "eval_samples_per_second": 117.845,
-      "eval_steps_per_second": 1.86,
-      "eval_wer": 0.33837197426207744,
+      "eval_cer": 0.08301473785214171,
+      "eval_loss": 0.4083641469478607,
+      "eval_runtime": 14.1402,
+      "eval_samples_per_second": 116.476,
+      "eval_steps_per_second": 1.839,
+      "eval_wer": 0.33622714738024717,
       "step": 3500
     },
     {
       "epoch": 63.65,
-      "learning_rate": 0.0002004,
-      "loss": 0.1381,
+      "learning_rate": 0.00020030000000000002,
+      "loss": 0.1295,
       "step": 3501
     },
     {
       "epoch": 63.67,
-      "learning_rate": 0.00020030000000000002,
-      "loss": 0.1148,
+      "learning_rate": 0.0002002,
+      "loss": 0.0975,
       "step": 3502
     },
     {
       "epoch": 63.69,
-      "learning_rate": 0.0002002,
-      "loss": 0.1513,
+      "learning_rate": 0.0002001,
+      "loss": 0.1454,
       "step": 3503
     },
     {
       "epoch": 63.71,
-      "learning_rate": 0.0002001,
-      "loss": 0.1322,
+      "learning_rate": 0.0002,
+      "loss": 0.1185,
       "step": 3504
     },
     {
       "epoch": 63.73,
-      "learning_rate": 0.0002,
-      "loss": 0.1732,
+      "learning_rate": 0.0001999,
+      "loss": 0.1493,
       "step": 3505
     },
     {
       "epoch": 63.75,
-      "learning_rate": 0.0001999,
-      "loss": 0.151,
+      "learning_rate": 0.0001998,
+      "loss": 0.1194,
       "step": 3506
     },
     {
       "epoch": 63.76,
-      "learning_rate": 0.0001998,
-      "loss": 0.1477,
+      "learning_rate": 0.0001997,
+      "loss": 0.1308,
       "step": 3507
     },
     {
       "epoch": 63.78,
-      "learning_rate": 0.0001997,
-      "loss": 0.1254,
+      "learning_rate": 0.0001996,
+      "loss": 0.1129,
       "step": 3508
     },
     {
       "epoch": 63.8,
-      "learning_rate": 0.0001996,
-      "loss": 0.1439,
+      "learning_rate": 0.00019950000000000002,
+      "loss": 0.138,
       "step": 3509
     },
     {
       "epoch": 63.82,
-      "learning_rate": 0.00019950000000000002,
-      "loss": 0.1456,
+      "learning_rate": 0.0001994,
+      "loss": 0.1429,
       "step": 3510
     },
     {
       "epoch": 63.84,
-      "learning_rate": 0.0001994,
-      "loss": 0.1291,
+      "learning_rate": 0.00019930000000000002,
+      "loss": 0.1347,
       "step": 3511
     },
     {
       "epoch": 63.85,
-      "learning_rate": 0.00019930000000000002,
-      "loss": 0.1434,
+      "learning_rate": 0.0001992,
+      "loss": 0.1212,
       "step": 3512
     },
     {
       "epoch": 63.87,
-      "learning_rate": 0.0001992,
-      "loss": 0.1343,
+      "learning_rate": 0.00019910000000000001,
+      "loss": 0.0919,
       "step": 3513
     },
     {
       "epoch": 63.89,
-      "learning_rate": 0.00019910000000000001,
-      "loss": 0.1411,
+      "learning_rate": 0.000199,
+      "loss": 0.1445,
       "step": 3514
     },
     {
       "epoch": 63.91,
-      "learning_rate": 0.000199,
-      "loss": 0.1068,
+      "learning_rate": 0.0001989,
+      "loss": 0.116,
       "step": 3515
     },
     {
       "epoch": 63.93,
-      "learning_rate": 0.0001989,
-      "loss": 0.1518,
+      "learning_rate": 0.0001988,
+      "loss": 0.1613,
       "step": 3516
     },
     {
       "epoch": 63.95,
-      "learning_rate": 0.0001988,
-      "loss": 0.1547,
+      "learning_rate": 0.0001987,
+      "loss": 0.1485,
       "step": 3517
     },
     {
       "epoch": 63.96,
-      "learning_rate": 0.0001987,
-      "loss": 0.1666,
+      "learning_rate": 0.0001986,
+      "loss": 0.1427,
       "step": 3518
     },
     {
       "epoch": 63.98,
-      "learning_rate": 0.0001986,
-      "loss": 0.1308,
+      "learning_rate": 0.00019850000000000003,
+      "loss": 0.1337,
       "step": 3519
     },
     {
       "epoch": 64.0,
-      "learning_rate": 0.00019850000000000003,
-      "loss": 0.0736,
+      "learning_rate": 0.0001984,
+      "loss": 0.0619,
       "step": 3520
     },
     {
       "epoch": 64.02,
-      "learning_rate": 0.0001984,
-      "loss": 0.1244,
+      "learning_rate": 0.0001983,
+      "loss": 0.1189,
       "step": 3521
     },
     {
       "epoch": 64.04,
-      "learning_rate": 0.0001983,
-      "loss": 0.1604,
+      "learning_rate": 0.0001982,
+      "loss": 0.1282,
       "step": 3522
     },
     {
       "epoch": 64.05,
-      "learning_rate": 0.0001982,
-      "loss": 0.1481,
+      "learning_rate": 0.0001981,
+      "loss": 0.1529,
       "step": 3523
     },
     {
       "epoch": 64.07,
-      "learning_rate": 0.0001981,
-      "loss": 0.1187,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 0.1144,
       "step": 3524
     },
     {
       "epoch": 64.09,
-      "learning_rate": 0.00019800000000000002,
-      "loss": 0.1339,
+      "learning_rate": 0.00019789999999999999,
+      "loss": 0.1299,
       "step": 3525
     },
     {
       "epoch": 64.11,
-      "learning_rate": 0.00019789999999999999,
-      "loss": 0.1395,
+      "learning_rate": 0.0001978,
+      "loss": 0.1095,
       "step": 3526
     },
     {
       "epoch": 64.13,
-      "learning_rate": 0.0001978,
-      "loss": 0.1429,
+      "learning_rate": 0.00019769999999999998,
+      "loss": 0.1579,
       "step": 3527
     },
     {
       "epoch": 64.15,
-      "learning_rate": 0.00019769999999999998,
-      "loss": 0.1173,
+      "learning_rate": 0.0001976,
+      "loss": 0.1081,
       "step": 3528
     },
     {
       "epoch": 64.16,
-      "learning_rate": 0.0001976,
-      "loss": 0.1188,
+      "learning_rate": 0.0001975,
+      "loss": 0.1254,
       "step": 3529
     },
     {
       "epoch": 64.18,
-      "learning_rate": 0.0001975,
-      "loss": 0.1116,
+      "learning_rate": 0.0001974,
+      "loss": 0.1191,
       "step": 3530
     },
     {
       "epoch": 64.2,
-      "learning_rate": 0.0001974,
-      "loss": 0.1466,
+      "learning_rate": 0.0001973,
+      "loss": 0.1378,
       "step": 3531
     },
     {
       "epoch": 64.22,
-      "learning_rate": 0.0001973,
-      "loss": 0.1099,
+      "learning_rate": 0.0001972,
+      "loss": 0.1047,
       "step": 3532
     },
     {
       "epoch": 64.24,
-      "learning_rate": 0.0001972,
-      "loss": 0.1377,
+      "learning_rate": 0.0001971,
+      "loss": 0.1171,
       "step": 3533
     },
     {
       "epoch": 64.25,
-      "learning_rate": 0.0001971,
-      "loss": 0.1356,
+      "learning_rate": 0.00019700000000000002,
+      "loss": 0.1147,
       "step": 3534
     },
     {
       "epoch": 64.27,
-      "learning_rate": 0.00019700000000000002,
-      "loss": 0.118,
+      "learning_rate": 0.0001969,
+      "loss": 0.1243,
       "step": 3535
     },
     {
       "epoch": 64.29,
-      "learning_rate": 0.0001969,
-      "loss": 0.15,
+      "learning_rate": 0.0001968,
+      "loss": 0.1081,
       "step": 3536
     },
     {
       "epoch": 64.31,
-      "learning_rate": 0.0001968,
-      "loss": 0.1365,
+      "learning_rate": 0.0001967,
+      "loss": 0.1129,
       "step": 3537
     },
     {
       "epoch": 64.33,
-      "learning_rate": 0.0001967,
-      "loss": 0.1138,
+      "learning_rate": 0.0001966,
+      "loss": 0.093,
       "step": 3538
     },
     {
       "epoch": 64.35,
-      "learning_rate": 0.0001966,
-      "loss": 0.1372,
+      "learning_rate": 0.0001965,
+      "loss": 0.1031,
       "step": 3539
     },
     {
       "epoch": 64.36,
-      "learning_rate": 0.0001965,
-      "loss": 0.1253,
+      "learning_rate": 0.0001964,
+      "loss": 0.1224,
       "step": 3540
     },
     {
       "epoch": 64.38,
-      "learning_rate": 0.0001964,
-      "loss": 0.1455,
+      "learning_rate": 0.0001963,
+      "loss": 0.1212,
       "step": 3541
     },
     {
       "epoch": 64.4,
-      "learning_rate": 0.0001963,
-      "loss": 0.131,
+      "learning_rate": 0.00019620000000000003,
+      "loss": 0.134,
       "step": 3542
     },
     {
       "epoch": 64.42,
-      "learning_rate": 0.00019620000000000003,
-      "loss": 0.1599,
+      "learning_rate": 0.0001961,
+      "loss": 0.1364,
       "step": 3543
     },
     {
       "epoch": 64.44,
-      "learning_rate": 0.0001961,
-      "loss": 0.099,
+      "learning_rate": 0.00019600000000000002,
+      "loss": 0.0926,
       "step": 3544
     },
     {
       "epoch": 64.45,
-      "learning_rate": 0.00019600000000000002,
-      "loss": 0.1072,
+      "learning_rate": 0.0001959,
+      "loss": 0.1013,
       "step": 3545
     },
     {
       "epoch": 64.47,
-      "learning_rate": 0.0001959,
-      "loss": 0.1952,
+      "learning_rate": 0.00019580000000000002,
+      "loss": 0.1746,
       "step": 3546
     },
     {
       "epoch": 64.49,
-      "learning_rate": 0.00019580000000000002,
-      "loss": 0.1405,
+      "learning_rate": 0.0001957,
+      "loss": 0.1163,
       "step": 3547
     },
     {
       "epoch": 64.51,
-      "learning_rate": 0.0001957,
-      "loss": 0.1382,
+      "learning_rate": 0.0001956,
+      "loss": 0.1302,
       "step": 3548
     },
     {
       "epoch": 64.53,
-      "learning_rate": 0.0001956,
-      "loss": 0.1258,
+      "learning_rate": 0.0001955,
+      "loss": 0.108,
       "step": 3549
     },
     {
       "epoch": 64.55,
-      "learning_rate": 0.0001955,
-      "loss": 0.1387,
+      "learning_rate": 0.0001954,
+      "loss": 0.1009,
       "step": 3550
     },
     {
       "epoch": 64.56,
-      "learning_rate": 0.0001954,
-      "loss": 0.122,
+      "learning_rate": 0.0001953,
+      "loss": 0.1064,
       "step": 3551
     },
     {
       "epoch": 64.58,
-      "learning_rate": 0.0001953,
-      "loss": 0.1414,
+      "learning_rate": 0.00019520000000000003,
+      "loss": 0.1429,
       "step": 3552
     },
     {
       "epoch": 64.6,
-      "learning_rate": 0.00019520000000000003,
-      "loss": 0.1361,
+      "learning_rate": 0.0001951,
+      "loss": 0.1257,
       "step": 3553
     },
     {
       "epoch": 64.62,
-      "learning_rate": 0.0001951,
-      "loss": 0.1165,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 0.1092,
       "step": 3554
     },
     {
       "epoch": 64.64,
-      "learning_rate": 0.00019500000000000002,
-      "loss": 0.1316,
+      "learning_rate": 0.0001949,
+      "loss": 0.1269,
       "step": 3555
     },
     {
       "epoch": 64.65,
-      "learning_rate": 0.0001949,
-      "loss": 0.1302,
+      "learning_rate": 0.0001948,
+      "loss": 0.1242,
       "step": 3556
     },
     {
       "epoch": 64.67,
-      "learning_rate": 0.0001948,
-      "loss": 0.1162,
+      "learning_rate": 0.00019470000000000002,
+      "loss": 0.0964,
       "step": 3557
     },
     {
       "epoch": 64.69,
-      "learning_rate": 0.00019470000000000002,
-      "loss": 0.1635,
+      "learning_rate": 0.0001946,
+      "loss": 0.1334,
       "step": 3558
     },
     {
       "epoch": 64.71,
-      "learning_rate": 0.0001946,
-      "loss": 0.1139,
+      "learning_rate": 0.0001945,
+      "loss": 0.1081,
       "step": 3559
     },
     {
       "epoch": 64.73,
-      "learning_rate": 0.0001945,
-      "loss": 0.1585,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 0.1268,
       "step": 3560
     },
     {
       "epoch": 64.75,
-      "learning_rate": 0.00019439999999999998,
-      "loss": 0.1143,
+      "learning_rate": 0.0001943,
+      "loss": 0.1374,
       "step": 3561
     },
     {
       "epoch": 64.76,
-      "learning_rate": 0.0001943,
-      "loss": 0.1351,
+      "learning_rate": 0.0001942,
+      "loss": 0.1172,
       "step": 3562
     },
     {
       "epoch": 64.78,
-      "learning_rate": 0.0001942,
-      "loss": 0.1036,
+      "learning_rate": 0.0001941,
+      "loss": 0.0892,
       "step": 3563
     },
     {
       "epoch": 64.8,
-      "learning_rate": 0.0001941,
-      "loss": 0.1276,
+      "learning_rate": 0.000194,
+      "loss": 0.1156,
       "step": 3564
     },
     {
       "epoch": 64.82,
-      "learning_rate": 0.000194,
-      "loss": 0.1406,
+      "learning_rate": 0.0001939,
+      "loss": 0.1402,
       "step": 3565
     },
     {
       "epoch": 64.84,
-      "learning_rate": 0.0001939,
-      "loss": 0.1522,
+      "learning_rate": 0.0001938,
+      "loss": 0.1288,
       "step": 3566
     },
     {
       "epoch": 64.85,
-      "learning_rate": 0.0001938,
-      "loss": 0.1558,
+      "learning_rate": 0.00019370000000000002,
+      "loss": 0.137,
       "step": 3567
     },
     {
       "epoch": 64.87,
-      "learning_rate": 0.00019370000000000002,
-      "loss": 0.0933,
+      "learning_rate": 0.0001936,
+      "loss": 0.0668,
       "step": 3568
     },
     {
       "epoch": 64.89,
-      "learning_rate": 0.0001936,
-      "loss": 0.1238,
+      "learning_rate": 0.00019350000000000001,
+      "loss": 0.1043,
       "step": 3569
     },
     {
       "epoch": 64.91,
-      "learning_rate": 0.00019350000000000001,
-      "loss": 0.1517,
+      "learning_rate": 0.00019339999999999998,
+      "loss": 0.1167,
       "step": 3570
     },
     {
       "epoch": 64.93,
-      "learning_rate": 0.00019339999999999998,
-      "loss": 0.1692,
+      "learning_rate": 0.0001933,
+      "loss": 0.1434,
       "step": 3571
     },
     {
       "epoch": 64.95,
-      "learning_rate": 0.0001933,
-      "loss": 0.1288,
+      "learning_rate": 0.0001932,
+      "loss": 0.1194,
       "step": 3572
     },
     {
       "epoch": 64.96,
-      "learning_rate": 0.0001932,
-      "loss": 0.1429,
+      "learning_rate": 0.0001931,
+      "loss": 0.1159,
       "step": 3573
     },
     {
       "epoch": 64.98,
-      "learning_rate": 0.0001931,
-      "loss": 0.1764,
+      "learning_rate": 0.000193,
+      "loss": 0.1605,
       "step": 3574
     },
     {
       "epoch": 65.0,
-      "learning_rate": 0.000193,
-      "loss": 0.1153,
+      "learning_rate": 0.0001929,
+      "loss": 0.0605,
       "step": 3575
     },
     {
       "epoch": 65.02,
-      "learning_rate": 0.0001929,
-      "loss": 0.1504,
+      "learning_rate": 0.0001928,
+      "loss": 0.1687,
       "step": 3576
     },
     {
       "epoch": 65.04,
-      "learning_rate": 0.0001928,
-      "loss": 0.1137,
+      "learning_rate": 0.00019270000000000002,
+      "loss": 0.1133,
       "step": 3577
     },
     {
       "epoch": 65.05,
-      "learning_rate": 0.00019270000000000002,
-      "loss": 0.1542,
+      "learning_rate": 0.0001926,
+      "loss": 0.1387,
       "step": 3578
     },
     {
       "epoch": 65.07,
-      "learning_rate": 0.0001926,
-      "loss": 0.1764,
+      "learning_rate": 0.00019250000000000002,
+      "loss": 0.1658,
       "step": 3579
     },
     {
       "epoch": 65.09,
-      "learning_rate": 0.00019250000000000002,
-      "loss": 0.1507,
+      "learning_rate": 0.0001924,
+      "loss": 0.1542,
       "step": 3580
     },
     {
       "epoch": 65.11,
-      "learning_rate": 0.0001924,
-      "loss": 0.1434,
+      "learning_rate": 0.0001923,
+      "loss": 0.1096,
       "step": 3581
     },
     {
       "epoch": 65.13,
-      "learning_rate": 0.0001923,
-      "loss": 0.1327,
+      "learning_rate": 0.0001922,
+      "loss": 0.1185,
       "step": 3582
     },
     {
       "epoch": 65.15,
-      "learning_rate": 0.0001922,
-      "loss": 0.1211,
+      "learning_rate": 0.0001921,
+      "loss": 0.1255,
       "step": 3583
     },
     {
       "epoch": 65.16,
-      "learning_rate": 0.0001921,
-      "loss": 0.1363,
+      "learning_rate": 0.000192,
+      "loss": 0.118,
       "step": 3584
     },
     {
       "epoch": 65.18,
-      "learning_rate": 0.000192,
-      "loss": 0.1556,
+      "learning_rate": 0.0001919,
+      "loss": 0.1284,
       "step": 3585
     },
     {
       "epoch": 65.2,
-      "learning_rate": 0.0001919,
-      "loss": 0.1534,
+      "learning_rate": 0.0001918,
+      "loss": 0.1457,
       "step": 3586
     },
     {
       "epoch": 65.22,
-      "learning_rate": 0.0001918,
-      "loss": 0.1174,
+      "learning_rate": 0.00019170000000000002,
+      "loss": 0.0813,
       "step": 3587
     },
     {
       "epoch": 65.24,
-      "learning_rate": 0.00019170000000000002,
-      "loss": 0.1288,
+      "learning_rate": 0.0001916,
+      "loss": 0.1481,
       "step": 3588
     },
     {
       "epoch": 65.25,
-      "learning_rate": 0.0001916,
-      "loss": 0.1345,
+      "learning_rate": 0.00019150000000000002,
+      "loss": 0.124,
       "step": 3589
     },
     {
       "epoch": 65.27,
-      "learning_rate": 0.00019150000000000002,
-      "loss": 0.1596,
+      "learning_rate": 0.0001914,
+      "loss": 0.1368,
       "step": 3590
     },
     {
       "epoch": 65.29,
-      "learning_rate": 0.0001914,
-      "loss": 0.1294,
+      "learning_rate": 0.00019130000000000001,
+      "loss": 0.117,
       "step": 3591
     },
     {
       "epoch": 65.31,
-      "learning_rate": 0.00019130000000000001,
-      "loss": 0.1516,
+      "learning_rate": 0.0001912,
+      "loss": 0.1242,
       "step": 3592
     },
     {
       "epoch": 65.33,
-      "learning_rate": 0.0001912,
-      "loss": 0.1329,
+      "learning_rate": 0.00019109999999999998,
+      "loss": 0.1422,
       "step": 3593
     },
     {
       "epoch": 65.35,
-      "learning_rate": 0.00019109999999999998,
-      "loss": 0.1476,
+      "learning_rate": 0.000191,
+      "loss": 0.1072,
       "step": 3594
     },
     {
       "epoch": 65.36,
-      "learning_rate": 0.000191,
-      "loss": 0.1587,
+      "learning_rate": 0.00019089999999999998,
+      "loss": 0.1636,
       "step": 3595
     },
     {
       "epoch": 65.38,
-      "learning_rate": 0.00019089999999999998,
-      "loss": 0.1444,
+      "learning_rate": 0.0001908,
+      "loss": 0.1222,
       "step": 3596
     },
     {
       "epoch": 65.4,
-      "learning_rate": 0.0001908,
-      "loss": 0.1379,
+      "learning_rate": 0.0001907,
+      "loss": 0.1232,
       "step": 3597
     },
     {
       "epoch": 65.42,
-      "learning_rate": 0.0001907,
-      "loss": 0.1491,
+      "learning_rate": 0.0001906,
+      "loss": 0.1025,
       "step": 3598
     },
     {
       "epoch": 65.44,
-      "learning_rate": 0.0001906,
-      "loss": 0.148,
+      "learning_rate": 0.0001905,
+      "loss": 0.1371,
       "step": 3599
     },
     {
       "epoch": 65.45,
-      "learning_rate": 0.0001905,
-      "loss": 0.1169,
+      "learning_rate": 0.00019040000000000002,
+      "loss": 0.1058,
       "step": 3600
     },
     {
       "epoch": 65.47,
-      "learning_rate": 0.00019040000000000002,
-      "loss": 0.1674,
+      "learning_rate": 0.0001903,
+      "loss": 0.1223,
       "step": 3601
     },
     {
       "epoch": 65.49,
-      "learning_rate": 0.0001903,
-      "loss": 0.1333,
+      "learning_rate": 0.00019020000000000002,
+      "loss": 0.1117,
       "step": 3602
     },
     {
       "epoch": 65.51,
-      "learning_rate": 0.00019020000000000002,
-      "loss": 0.1524,
+      "learning_rate": 0.00019009999999999999,
+      "loss": 0.1179,
       "step": 3603
     },
     {
       "epoch": 65.53,
-      "learning_rate": 0.00019009999999999999,
-      "loss": 0.1139,
+      "learning_rate": 0.00019,
+      "loss": 0.1163,
       "step": 3604
     },
     {
       "epoch": 65.55,
-      "learning_rate": 0.00019,
-      "loss": 0.1239,
+      "learning_rate": 0.0001899,
+      "loss": 0.1451,
       "step": 3605
     },
     {
       "epoch": 65.56,
-      "learning_rate": 0.0001899,
-      "loss": 0.095,
+      "learning_rate": 0.0001898,
+      "loss": 0.0633,
       "step": 3606
     },
     {
       "epoch": 65.58,
-      "learning_rate": 0.0001898,
-      "loss": 0.136,
+      "learning_rate": 0.0001897,
+      "loss": 0.1328,
       "step": 3607
     },
     {
       "epoch": 65.6,
-      "learning_rate": 0.0001897,
-      "loss": 0.1325,
+      "learning_rate": 0.0001896,
+      "loss": 0.1306,
       "step": 3608
     },
     {
       "epoch": 65.62,
-      "learning_rate": 0.0001896,
-      "loss": 0.1138,
+      "learning_rate": 0.0001895,
+      "loss": 0.0923,
       "step": 3609
     },
     {
       "epoch": 65.64,
-      "learning_rate": 0.0001895,
-      "loss": 0.1422,
+      "learning_rate": 0.00018940000000000002,
+      "loss": 0.1332,
       "step": 3610
     },
     {
       "epoch": 65.65,
-      "learning_rate": 0.00018940000000000002,
-      "loss": 0.1707,
+      "learning_rate": 0.0001893,
+      "loss": 0.1553,
       "step": 3611
     },
     {
       "epoch": 65.67,
-      "learning_rate": 0.0001893,
-      "loss": 0.1063,
+      "learning_rate": 0.00018920000000000002,
+      "loss": 0.1012,
       "step": 3612
     },
     {
       "epoch": 65.69,
-      "learning_rate": 0.00018920000000000002,
-      "loss": 0.2016,
+      "learning_rate": 0.0001891,
+      "loss": 0.1861,
       "step": 3613
     },
     {
       "epoch": 65.71,
-      "learning_rate": 0.0001891,
-      "loss": 0.1823,
+      "learning_rate": 0.000189,
+      "loss": 0.1754,
       "step": 3614
     },
     {
       "epoch": 65.73,
-      "learning_rate": 0.000189,
-      "loss": 0.1307,
+      "learning_rate": 0.0001889,
+      "loss": 0.1161,
       "step": 3615
     },
     {
       "epoch": 65.75,
-      "learning_rate": 0.0001889,
-      "loss": 0.1283,
+      "learning_rate": 0.0001888,
+      "loss": 0.1077,
       "step": 3616
     },
     {
       "epoch": 65.76,
-      "learning_rate": 0.0001888,
-      "loss": 0.1407,
+      "learning_rate": 0.0001887,
+      "loss": 0.1437,
       "step": 3617
     },
     {
       "epoch": 65.78,
-      "learning_rate": 0.0001887,
-      "loss": 0.1505,
+      "learning_rate": 0.0001886,
+      "loss": 0.1504,
       "step": 3618
     },
     {
       "epoch": 65.8,
-      "learning_rate": 0.0001886,
-      "loss": 0.1261,
+      "learning_rate": 0.0001885,
+      "loss": 0.1328,
       "step": 3619
     },
     {
       "epoch": 65.82,
-      "learning_rate": 0.0001885,
-      "loss": 0.1886,
+      "learning_rate": 0.00018840000000000003,
+      "loss": 0.1568,
       "step": 3620
     },
     {
       "epoch": 65.84,
-      "learning_rate": 0.00018840000000000003,
-      "loss": 0.147,
+      "learning_rate": 0.0001883,
+      "loss": 0.1387,
       "step": 3621
     },
     {
       "epoch": 65.85,
-      "learning_rate": 0.0001883,
-      "loss": 0.1613,
+      "learning_rate": 0.00018820000000000002,
+      "loss": 0.1216,
       "step": 3622
     },
     {
       "epoch": 65.87,
-      "learning_rate": 0.00018820000000000002,
-      "loss": 0.1312,
+      "learning_rate": 0.0001881,
+      "loss": 0.1093,
       "step": 3623
     },
     {
       "epoch": 65.89,
-      "learning_rate": 0.0001881,
-      "loss": 0.1265,
+      "learning_rate": 0.00018800000000000002,
+      "loss": 0.1127,
       "step": 3624
     },
     {
       "epoch": 65.91,
-      "learning_rate": 0.00018800000000000002,
-      "loss": 0.1019,
+      "learning_rate": 0.00018790000000000001,
+      "loss": 0.1074,
       "step": 3625
     },
     {
       "epoch": 65.93,
-      "learning_rate": 0.00018790000000000001,
-      "loss": 0.1362,
+      "learning_rate": 0.0001878,
+      "loss": 0.1088,
       "step": 3626
     },
     {
       "epoch": 65.95,
-      "learning_rate": 0.0001878,
-      "loss": 0.1465,
+      "learning_rate": 0.0001877,
+      "loss": 0.1144,
       "step": 3627
     },
     {
       "epoch": 65.96,
-      "learning_rate": 0.0001877,
-      "loss": 0.1301,
+      "learning_rate": 0.0001876,
+      "loss": 0.1081,
       "step": 3628
     },
     {
       "epoch": 65.98,
-      "learning_rate": 0.0001876,
-      "loss": 0.1252,
+      "learning_rate": 0.0001875,
+      "loss": 0.1397,
       "step": 3629
     },
     {
       "epoch": 66.0,
-      "learning_rate": 0.0001875,
-      "loss": 0.0704,
+      "learning_rate": 0.0001874,
+      "loss": 0.0443,
       "step": 3630
     },
     {
       "epoch": 66.02,
-      "learning_rate": 0.0001874,
-      "loss": 0.1595,
+      "learning_rate": 0.0001873,
+      "loss": 0.1623,
       "step": 3631
     },
     {
       "epoch": 66.04,
-      "learning_rate": 0.0001873,
-      "loss": 0.1288,
+      "learning_rate": 0.0001872,
+      "loss": 0.1089,
       "step": 3632
     },
     {
       "epoch": 66.05,
-      "learning_rate": 0.0001872,
-      "loss": 0.1714,
+      "learning_rate": 0.0001871,
+      "loss": 0.1878,
       "step": 3633
     },
     {
       "epoch": 66.07,
-      "learning_rate": 0.0001871,
-      "loss": 0.1047,
+      "learning_rate": 0.000187,
+      "loss": 0.1063,
       "step": 3634
     },
     {
       "epoch": 66.09,
-      "learning_rate": 0.000187,
-      "loss": 0.1362,
+      "learning_rate": 0.00018690000000000002,
+      "loss": 0.1262,
       "step": 3635
     },
     {
       "epoch": 66.11,
-      "learning_rate": 0.00018690000000000002,
-      "loss": 0.1336,
+      "learning_rate": 0.0001868,
+      "loss": 0.1316,
       "step": 3636
     },
     {
       "epoch": 66.13,
-      "learning_rate": 0.0001868,
-      "loss": 0.1935,
+      "learning_rate": 0.0001867,
+      "loss": 0.161,
       "step": 3637
     },
     {
       "epoch": 66.15,
-      "learning_rate": 0.0001867,
-      "loss": 0.1592,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 0.1596,
       "step": 3638
     },
     {
       "epoch": 66.16,
-      "learning_rate": 0.00018659999999999998,
-      "loss": 0.1616,
+      "learning_rate": 0.0001865,
+      "loss": 0.1446,
       "step": 3639
     },
     {
       "epoch": 66.18,
-      "learning_rate": 0.0001865,
-      "loss": 0.163,
+      "learning_rate": 0.0001864,
+      "loss": 0.1288,
       "step": 3640
     },
     {
       "epoch": 66.2,
-      "learning_rate": 0.0001864,
-      "loss": 0.1469,
+      "learning_rate": 0.0001863,
+      "loss": 0.1241,
       "step": 3641
     },
     {
       "epoch": 66.22,
-      "learning_rate": 0.0001863,
-      "loss": 0.1119,
+      "learning_rate": 0.0001862,
+      "loss": 0.1054,
       "step": 3642
     },
     {
       "epoch": 66.24,
-      "learning_rate": 0.0001862,
-      "loss": 0.1433,
+      "learning_rate": 0.0001861,
+      "loss": 0.123,
       "step": 3643
     },
     {
       "epoch": 66.25,
-      "learning_rate": 0.0001861,
-      "loss": 0.1096,
+      "learning_rate": 0.000186,
+      "loss": 0.1114,
       "step": 3644
     },
     {
       "epoch": 66.27,
-      "learning_rate": 0.000186,
-      "loss": 0.1283,
+      "learning_rate": 0.00018590000000000002,
+      "loss": 0.1449,
       "step": 3645
     },
     {
       "epoch": 66.29,
-      "learning_rate": 0.00018590000000000002,
-      "loss": 0.1426,
+      "learning_rate": 0.0001858,
+      "loss": 0.1353,
       "step": 3646
     },
     {
       "epoch": 66.31,
-      "learning_rate": 0.0001858,
-      "loss": 0.1716,
+      "learning_rate": 0.00018570000000000001,
+      "loss": 0.1393,
       "step": 3647
     },
     {
       "epoch": 66.33,
-      "learning_rate": 0.00018570000000000001,
-      "loss": 0.1406,
+      "learning_rate": 0.00018559999999999998,
+      "loss": 0.0963,
       "step": 3648
     },
     {
       "epoch": 66.35,
-      "learning_rate": 0.00018559999999999998,
-      "loss": 0.1229,
+      "learning_rate": 0.0001855,
+      "loss": 0.1085,
       "step": 3649
     },
     {
       "epoch": 66.36,
-      "learning_rate": 0.0001855,
-      "loss": 0.1236,
+      "learning_rate": 0.0001854,
+      "loss": 0.1363,
       "step": 3650
     },
     {
       "epoch": 66.38,
-      "learning_rate": 0.0001854,
-      "loss": 0.1404,
+      "learning_rate": 0.0001853,
+      "loss": 0.178,
       "step": 3651
     },
     {
       "epoch": 66.4,
-      "learning_rate": 0.0001853,
-      "loss": 0.1529,
+      "learning_rate": 0.0001852,
+      "loss": 0.1452,
       "step": 3652
     },
     {
       "epoch": 66.42,
-      "learning_rate": 0.0001852,
-      "loss": 0.1242,
+      "learning_rate": 0.0001851,
+      "loss": 0.1288,
       "step": 3653
     },
     {
       "epoch": 66.44,
-      "learning_rate": 0.0001851,
-      "loss": 0.124,
+      "learning_rate": 0.000185,
+      "loss": 0.1031,
       "step": 3654
     },
     {
       "epoch": 66.45,
-      "learning_rate": 0.000185,
-      "loss": 0.1169,
+      "learning_rate": 0.00018490000000000002,
+      "loss": 0.0801,
       "step": 3655
     },
     {
       "epoch": 66.47,
-      "learning_rate": 0.00018490000000000002,
-      "loss": 0.1508,
+      "learning_rate": 0.0001848,
+      "loss": 0.1162,
       "step": 3656
     },
     {
       "epoch": 66.49,
-      "learning_rate": 0.0001848,
-      "loss": 0.1467,
+      "learning_rate": 0.00018470000000000002,
+      "loss": 0.1301,
       "step": 3657
     },
     {
       "epoch": 66.51,
-      "learning_rate": 0.00018470000000000002,
-      "loss": 0.1608,
+      "learning_rate": 0.0001846,
+      "loss": 0.1759,
       "step": 3658
     },
     {
       "epoch": 66.53,
-      "learning_rate": 0.0001846,
-      "loss": 0.1427,
+      "learning_rate": 0.0001845,
+      "loss": 0.154,
       "step": 3659
     },
     {
       "epoch": 66.55,
-      "learning_rate": 0.0001845,
-      "loss": 0.1164,
+      "learning_rate": 0.0001844,
+      "loss": 0.1219,
       "step": 3660
     },
     {
       "epoch": 66.56,
-      "learning_rate": 0.0001844,
-      "loss": 0.1864,
+      "learning_rate": 0.0001843,
+      "loss": 0.1407,
       "step": 3661
     },
     {
       "epoch": 66.58,
-      "learning_rate": 0.0001843,
-      "loss": 0.149,
+      "learning_rate": 0.0001842,
+      "loss": 0.1343,
       "step": 3662
     },
     {
       "epoch": 66.6,
-      "learning_rate": 0.0001842,
-      "loss": 0.1541,
+      "learning_rate": 0.00018410000000000003,
+      "loss": 0.1393,
       "step": 3663
     },
     {
       "epoch": 66.62,
-      "learning_rate": 0.00018410000000000003,
-      "loss": 0.1022,
+      "learning_rate": 0.000184,
+      "loss": 0.1122,
       "step": 3664
     },
     {
       "epoch": 66.64,
-      "learning_rate": 0.000184,
-      "loss": 0.1619,
+      "learning_rate": 0.00018390000000000002,
+      "loss": 0.1283,
       "step": 3665
     },
     {
       "epoch": 66.65,
-      "learning_rate": 0.00018390000000000002,
-      "loss": 0.1298,
+      "learning_rate": 0.0001838,
+      "loss": 0.1247,
       "step": 3666
     },
     {
       "epoch": 66.67,
-      "learning_rate": 0.0001838,
-      "loss": 0.1219,
+      "learning_rate": 0.0001837,
+      "loss": 0.1161,
       "step": 3667
     },
     {
       "epoch": 66.69,
-      "learning_rate": 0.0001837,
-      "loss": 0.1425,
+      "learning_rate": 0.00018360000000000002,
+      "loss": 0.1307,
       "step": 3668
     },
     {
       "epoch": 66.71,
-      "learning_rate": 0.00018360000000000002,
-      "loss": 0.1387,
+      "learning_rate": 0.0001835,
+      "loss": 0.129,
       "step": 3669
     },
     {
       "epoch": 66.73,
-      "learning_rate": 0.0001835,
-      "loss": 0.1466,
+      "learning_rate": 0.0001834,
+      "loss": 0.1375,
       "step": 3670
     },
     {
       "epoch": 66.75,
-      "learning_rate": 0.0001834,
-      "loss": 0.0918,
+      "learning_rate": 0.00018329999999999998,
+      "loss": 0.0899,
       "step": 3671
     },
     {
       "epoch": 66.76,
-      "learning_rate": 0.00018329999999999998,
-      "loss": 0.1192,
+      "learning_rate": 0.0001832,
+      "loss": 0.0971,
       "step": 3672
     },
     {
       "epoch": 66.78,
-      "learning_rate": 0.0001832,
-      "loss": 0.1283,
+      "learning_rate": 0.0001831,
+      "loss": 0.1049,
       "step": 3673
     },
     {
       "epoch": 66.8,
-      "learning_rate": 0.0001831,
-      "loss": 0.1058,
+      "learning_rate": 0.000183,
+      "loss": 0.1085,
       "step": 3674
     },
     {
       "epoch": 66.82,
-      "learning_rate": 0.000183,
-      "loss": 0.1619,
+      "learning_rate": 0.0001829,
+      "loss": 0.1672,
       "step": 3675
     },
     {
       "epoch": 66.84,
-      "learning_rate": 0.0001829,
-      "loss": 0.1131,
+      "learning_rate": 0.0001828,
+      "loss": 0.1221,
       "step": 3676
     },
     {
       "epoch": 66.85,
-      "learning_rate": 0.0001828,
-      "loss": 0.145,
+      "learning_rate": 0.0001827,
+      "loss": 0.1359,
       "step": 3677
     },
     {
       "epoch": 66.87,
-      "learning_rate": 0.0001827,
-      "loss": 0.1416,
+      "learning_rate": 0.00018260000000000002,
+      "loss": 0.1286,
       "step": 3678
     },
     {
       "epoch": 66.89,
-      "learning_rate": 0.00018260000000000002,
-      "loss": 0.1257,
+      "learning_rate": 0.0001825,
+      "loss": 0.1046,
       "step": 3679
     },
     {
       "epoch": 66.91,
-      "learning_rate": 0.0001825,
-      "loss": 0.0997,
+      "learning_rate": 0.00018240000000000002,
+      "loss": 0.0887,
       "step": 3680
     },
     {
       "epoch": 66.93,
-      "learning_rate": 0.00018240000000000002,
-      "loss": 0.1323,
+      "learning_rate": 0.00018229999999999999,
+      "loss": 0.1124,
       "step": 3681
     },
     {
       "epoch": 66.95,
-      "learning_rate": 0.00018229999999999999,
-      "loss": 0.1328,
+      "learning_rate": 0.0001822,
+      "loss": 0.1245,
       "step": 3682
     },
     {
       "epoch": 66.96,
-      "learning_rate": 0.0001822,
-      "loss": 0.1485,
+      "learning_rate": 0.0001821,
+      "loss": 0.1261,
       "step": 3683
     },
     {
       "epoch": 66.98,
-      "learning_rate": 0.0001821,
-      "loss": 0.0923,
+      "learning_rate": 0.000182,
+      "loss": 0.1037,
       "step": 3684
     },
     {
       "epoch": 67.0,
-      "learning_rate": 0.000182,
-      "loss": 0.0437,
+      "learning_rate": 0.0001819,
+      "loss": 0.0813,
       "step": 3685
     },
     {
       "epoch": 67.02,
-      "learning_rate": 0.0001819,
-      "loss": 0.1354,
+      "learning_rate": 0.0001818,
+      "loss": 0.135,
       "step": 3686
     },
     {
       "epoch": 67.04,
-      "learning_rate": 0.0001818,
-      "loss": 0.1443,
+      "learning_rate": 0.0001817,
+      "loss": 0.1166,
       "step": 3687
     },
     {
       "epoch": 67.05,
-      "learning_rate": 0.0001817,
-      "loss": 0.1195,
+      "learning_rate": 0.00018160000000000002,
+      "loss": 0.0982,
       "step": 3688
     },
     {
       "epoch": 67.07,
-      "learning_rate": 0.00018160000000000002,
-      "loss": 0.1062,
+      "learning_rate": 0.0001815,
+      "loss": 0.1303,
       "step": 3689
     },
     {
       "epoch": 67.09,
-      "learning_rate": 0.0001815,
-      "loss": 0.1075,
+      "learning_rate": 0.00018140000000000002,
+      "loss": 0.1188,
       "step": 3690
     },
     {
       "epoch": 67.11,
-      "learning_rate": 0.00018140000000000002,
-      "loss": 0.1326,
+      "learning_rate": 0.0001813,
+      "loss": 0.1047,
       "step": 3691
     },
     {
       "epoch": 67.13,
-      "learning_rate": 0.0001813,
-      "loss": 0.1349,
+      "learning_rate": 0.0001812,
+      "loss": 0.1084,
       "step": 3692
     },
     {
       "epoch": 67.15,
-      "learning_rate": 0.0001812,
-      "loss": 0.139,
+      "learning_rate": 0.0001811,
+      "loss": 0.1388,
       "step": 3693
     },
     {
       "epoch": 67.16,
-      "learning_rate": 0.0001811,
-      "loss": 0.1202,
+      "learning_rate": 0.000181,
+      "loss": 0.1234,
       "step": 3694
     },
     {
       "epoch": 67.18,
-      "learning_rate": 0.000181,
-      "loss": 0.1434,
+      "learning_rate": 0.0001809,
+      "loss": 0.1647,
       "step": 3695
     },
     {
       "epoch": 67.2,
-      "learning_rate": 0.0001809,
-      "loss": 0.1447,
+      "learning_rate": 0.0001808,
+      "loss": 0.1105,
       "step": 3696
     },
     {
       "epoch": 67.22,
-      "learning_rate": 0.0001808,
-      "loss": 0.131,
+      "learning_rate": 0.0001807,
+      "loss": 0.1223,
       "step": 3697
     },
     {
       "epoch": 67.24,
-      "learning_rate": 0.0001807,
-      "loss": 0.1331,
+      "learning_rate": 0.00018060000000000003,
+      "loss": 0.1141,
       "step": 3698
     },
     {
       "epoch": 67.25,
-      "learning_rate": 0.00018060000000000003,
-      "loss": 0.183,
+      "learning_rate": 0.0001805,
+      "loss": 0.1489,
       "step": 3699
     },
     {
       "epoch": 67.27,
-      "learning_rate": 0.0001805,
-      "loss": 0.1243,
+      "learning_rate": 0.00018040000000000002,
+      "loss": 0.1073,
       "step": 3700
     },
     {
       "epoch": 67.29,
-      "learning_rate": 0.00018040000000000002,
-      "loss": 0.0951,
+      "learning_rate": 0.0001803,
+      "loss": 0.1043,
       "step": 3701
     },
     {
       "epoch": 67.31,
-      "learning_rate": 0.0001803,
-      "loss": 0.2057,
+      "learning_rate": 0.00018020000000000002,
+      "loss": 0.1652,
       "step": 3702
     },
     {
       "epoch": 67.33,
-      "learning_rate": 0.00018020000000000002,
-      "loss": 0.1497,
+      "learning_rate": 0.00018010000000000001,
+      "loss": 0.1373,
       "step": 3703
     },
     {
       "epoch": 67.35,
-      "learning_rate": 0.00018010000000000001,
-      "loss": 0.1084,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.0968,
       "step": 3704
     },
     {
       "epoch": 67.36,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 0.1208,
+      "learning_rate": 0.0001799,
+      "loss": 0.1018,
       "step": 3705
     },
     {
       "epoch": 67.38,
-      "learning_rate": 0.0001799,
-      "loss": 0.119,
+      "learning_rate": 0.00017979999999999998,
+      "loss": 0.1182,
       "step": 3706
     },
     {
       "epoch": 67.4,
-      "learning_rate": 0.00017979999999999998,
-      "loss": 0.1059,
+      "learning_rate": 0.0001797,
+      "loss": 0.1125,
       "step": 3707
     },
     {
       "epoch": 67.42,
-      "learning_rate": 0.0001797,
-      "loss": 0.1238,
+      "learning_rate": 0.0001796,
+      "loss": 0.1168,
       "step": 3708
     },
     {
       "epoch": 67.44,
-      "learning_rate": 0.0001796,
-      "loss": 0.1054,
+      "learning_rate": 0.0001795,
+      "loss": 0.1359,
       "step": 3709
     },
     {
       "epoch": 67.45,
-      "learning_rate": 0.0001795,
-      "loss": 0.1346,
+      "learning_rate": 0.0001794,
+      "loss": 0.1178,
       "step": 3710
     },
     {
       "epoch": 67.47,
-      "learning_rate": 0.0001794,
-      "loss": 0.162,
+      "learning_rate": 0.0001793,
+      "loss": 0.1442,
       "step": 3711
     },
     {
       "epoch": 67.49,
-      "learning_rate": 0.0001793,
-      "loss": 0.1623,
+      "learning_rate": 0.0001792,
+      "loss": 0.1315,
       "step": 3712
     },
     {
       "epoch": 67.51,
-      "learning_rate": 0.0001792,
-      "loss": 0.1054,
+      "learning_rate": 0.00017910000000000002,
+      "loss": 0.1051,
       "step": 3713
     },
     {
       "epoch": 67.53,
-      "learning_rate": 0.00017910000000000002,
-      "loss": 0.1448,
+      "learning_rate": 0.000179,
+      "loss": 0.1392,
       "step": 3714
     },
     {
       "epoch": 67.55,
-      "learning_rate": 0.000179,
-      "loss": 0.133,
+      "learning_rate": 0.0001789,
+      "loss": 0.1362,
       "step": 3715
     },
     {
       "epoch": 67.56,
-      "learning_rate": 0.0001789,
-      "loss": 0.1046,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 0.0911,
       "step": 3716
     },
     {
       "epoch": 67.58,
-      "learning_rate": 0.00017879999999999998,
-      "loss": 0.1878,
+      "learning_rate": 0.0001787,
+      "loss": 0.143,
       "step": 3717
     },
     {
       "epoch": 67.6,
-      "learning_rate": 0.0001787,
-      "loss": 0.146,
+      "learning_rate": 0.0001786,
+      "loss": 0.1295,
       "step": 3718
     },
     {
       "epoch": 67.62,
-      "learning_rate": 0.0001786,
-      "loss": 0.1044,
+      "learning_rate": 0.0001785,
+      "loss": 0.1069,
       "step": 3719
     },
     {
       "epoch": 67.64,
-      "learning_rate": 0.0001785,
-      "loss": 0.1284,
+      "learning_rate": 0.0001784,
+      "loss": 0.1115,
       "step": 3720
     },
     {
       "epoch": 67.65,
-      "learning_rate": 0.0001784,
-      "loss": 0.1119,
+      "learning_rate": 0.0001783,
+      "loss": 0.0867,
       "step": 3721
     },
     {
       "epoch": 67.67,
-      "learning_rate": 0.0001783,
-      "loss": 0.1176,
+      "learning_rate": 0.0001782,
+      "loss": 0.0857,
       "step": 3722
     },
     {
       "epoch": 67.69,
-      "learning_rate": 0.0001782,
-      "loss": 0.1445,
+      "learning_rate": 0.00017810000000000002,
+      "loss": 0.1231,
       "step": 3723
     },
     {
       "epoch": 67.71,
-      "learning_rate": 0.00017810000000000002,
-      "loss": 0.1618,
+      "learning_rate": 0.000178,
+      "loss": 0.143,
       "step": 3724
     },
     {
       "epoch": 67.73,
-      "learning_rate": 0.000178,
-      "loss": 0.143,
+      "learning_rate": 0.00017790000000000001,
+      "loss": 0.1177,
       "step": 3725
     },
     {
       "epoch": 67.75,
-      "learning_rate": 0.00017790000000000001,
-      "loss": 0.1352,
+      "learning_rate": 0.0001778,
+      "loss": 0.1235,
       "step": 3726
     },
     {
       "epoch": 67.76,
-      "learning_rate": 0.0001778,
-      "loss": 0.1698,
+      "learning_rate": 0.0001777,
+      "loss": 0.1192,
       "step": 3727
     },
     {
       "epoch": 67.78,
-      "learning_rate": 0.0001777,
-      "loss": 0.1307,
+      "learning_rate": 0.0001776,
+      "loss": 0.13,
       "step": 3728
     },
     {
       "epoch": 67.8,
-      "learning_rate": 0.0001776,
-      "loss": 0.0905,
+      "learning_rate": 0.0001775,
+      "loss": 0.0885,
       "step": 3729
     },
     {
       "epoch": 67.82,
-      "learning_rate": 0.0001775,
-      "loss": 0.1379,
+      "learning_rate": 0.0001774,
+      "loss": 0.1354,
       "step": 3730
     },
     {
       "epoch": 67.84,
-      "learning_rate": 0.0001774,
-      "loss": 0.1763,
+      "learning_rate": 0.00017730000000000003,
+      "loss": 0.1722,
       "step": 3731
     },
     {
       "epoch": 67.85,
-      "learning_rate": 0.00017730000000000003,
-      "loss": 0.1455,
+      "learning_rate": 0.0001772,
+      "loss": 0.1369,
       "step": 3732
     },
     {
       "epoch": 67.87,
-      "learning_rate": 0.0001772,
-      "loss": 0.1268,
+      "learning_rate": 0.00017710000000000002,
+      "loss": 0.1296,
       "step": 3733
     },
     {
       "epoch": 67.89,
-      "learning_rate": 0.00017710000000000002,
-      "loss": 0.1125,
+      "learning_rate": 0.000177,
+      "loss": 0.0849,
       "step": 3734
     },
     {
       "epoch": 67.91,
-      "learning_rate": 0.000177,
-      "loss": 0.1262,
+      "learning_rate": 0.00017690000000000002,
+      "loss": 0.1156,
       "step": 3735
     },
     {
       "epoch": 67.93,
-      "learning_rate": 0.00017690000000000002,
-      "loss": 0.1389,
+      "learning_rate": 0.00017680000000000001,
+      "loss": 0.1275,
       "step": 3736
     },
     {
       "epoch": 67.95,
-      "learning_rate": 0.00017680000000000001,
-      "loss": 0.1249,
+      "learning_rate": 0.0001767,
+      "loss": 0.1022,
       "step": 3737
     },
     {
       "epoch": 67.96,
-      "learning_rate": 0.0001767,
-      "loss": 0.1484,
+      "learning_rate": 0.0001766,
+      "loss": 0.1226,
       "step": 3738
     },
     {
       "epoch": 67.98,
-      "learning_rate": 0.0001766,
-      "loss": 0.1588,
+      "learning_rate": 0.00017649999999999998,
+      "loss": 0.1264,
       "step": 3739
     },
     {
       "epoch": 68.0,
-      "learning_rate": 0.00017649999999999998,
-      "loss": 0.1317,
+      "learning_rate": 0.0001764,
+      "loss": 0.1657,
       "step": 3740
     },
     {
       "epoch": 68.02,
-      "learning_rate": 0.0001764,
-      "loss": 0.1365,
+      "learning_rate": 0.0001763,
+      "loss": 0.1428,
       "step": 3741
     },
     {
       "epoch": 68.04,
-      "learning_rate": 0.0001763,
-      "loss": 0.1652,
+      "learning_rate": 0.0001762,
+      "loss": 0.133,
       "step": 3742
     },
     {
       "epoch": 68.05,
-      "learning_rate": 0.0001762,
-      "loss": 0.1581,
+      "learning_rate": 0.0001761,
+      "loss": 0.1145,
       "step": 3743
     },
     {
       "epoch": 68.07,
-      "learning_rate": 0.0001761,
-      "loss": 0.1223,
+      "learning_rate": 0.000176,
+      "loss": 0.1177,
       "step": 3744
     },
     {
       "epoch": 68.09,
-      "learning_rate": 0.000176,
-      "loss": 0.101,
+      "learning_rate": 0.0001759,
+      "loss": 0.0996,
       "step": 3745
     },
     {
       "epoch": 68.11,
-      "learning_rate": 0.0001759,
-      "loss": 0.1496,
+      "learning_rate": 0.00017580000000000002,
+      "loss": 0.1143,
       "step": 3746
     },
     {
       "epoch": 68.13,
-      "learning_rate": 0.00017580000000000002,
-      "loss": 0.194,
+      "learning_rate": 0.0001757,
+      "loss": 0.2006,
       "step": 3747
     },
     {
       "epoch": 68.15,
-      "learning_rate": 0.0001757,
-      "loss": 0.1439,
+      "learning_rate": 0.0001756,
+      "loss": 0.1178,
       "step": 3748
     },
     {
       "epoch": 68.16,
-      "learning_rate": 0.0001756,
-      "loss": 0.1342,
+      "learning_rate": 0.00017549999999999998,
+      "loss": 0.1323,
       "step": 3749
     },
     {
       "epoch": 68.18,
-      "learning_rate": 0.00017549999999999998,
-      "loss": 0.1444,
+      "learning_rate": 0.0001754,
+      "loss": 0.12,
       "step": 3750
     },
     {
       "epoch": 68.2,
-      "learning_rate": 0.0001754,
-      "loss": 0.1299,
+      "learning_rate": 0.0001753,
+      "loss": 0.1049,
       "step": 3751
     },
     {
       "epoch": 68.22,
-      "learning_rate": 0.0001753,
-      "loss": 0.1214,
+      "learning_rate": 0.0001752,
+      "loss": 0.1009,
       "step": 3752
     },
     {
       "epoch": 68.24,
-      "learning_rate": 0.0001752,
-      "loss": 0.1289,
+      "learning_rate": 0.0001751,
+      "loss": 0.1029,
       "step": 3753
     },
     {
       "epoch": 68.25,
-      "learning_rate": 0.0001751,
-      "loss": 0.1404,
+      "learning_rate": 0.000175,
+      "loss": 0.1418,
       "step": 3754
     },
     {
       "epoch": 68.27,
-      "learning_rate": 0.000175,
-      "loss": 0.1407,
+      "learning_rate": 0.0001749,
+      "loss": 0.1363,
       "step": 3755
     },
     {
       "epoch": 68.29,
-      "learning_rate": 0.0001749,
-      "loss": 0.1256,
+      "learning_rate": 0.00017480000000000002,
+      "loss": 0.103,
       "step": 3756
     },
     {
       "epoch": 68.31,
-      "learning_rate": 0.00017480000000000002,
-      "loss": 0.0858,
+      "learning_rate": 0.0001747,
+      "loss": 0.0775,
       "step": 3757
     },
     {
       "epoch": 68.33,
-      "learning_rate": 0.0001747,
-      "loss": 0.1665,
+      "learning_rate": 0.00017460000000000002,
+      "loss": 0.1221,
       "step": 3758
     },
     {
       "epoch": 68.35,
-      "learning_rate": 0.00017460000000000002,
-      "loss": 0.104,
+      "learning_rate": 0.00017449999999999999,
+      "loss": 0.0895,
       "step": 3759
     },
     {
       "epoch": 68.36,
-      "learning_rate": 0.00017449999999999999,
-      "loss": 0.144,
+      "learning_rate": 0.0001744,
+      "loss": 0.1568,
       "step": 3760
     },
     {
       "epoch": 68.38,
-      "learning_rate": 0.0001744,
-      "loss": 0.1417,
+      "learning_rate": 0.0001743,
+      "loss": 0.1193,
       "step": 3761
     },
     {
       "epoch": 68.4,
-      "learning_rate": 0.0001743,
-      "loss": 0.097,
+      "learning_rate": 0.0001742,
+      "loss": 0.0669,
       "step": 3762
     },
     {
       "epoch": 68.42,
-      "learning_rate": 0.0001742,
-      "loss": 0.1205,
+      "learning_rate": 0.0001741,
+      "loss": 0.1024,
       "step": 3763
     },
     {
       "epoch": 68.44,
-      "learning_rate": 0.0001741,
-      "loss": 0.1235,
+      "learning_rate": 0.000174,
+      "loss": 0.1282,
       "step": 3764
     },
     {
       "epoch": 68.45,
-      "learning_rate": 0.000174,
-      "loss": 0.1106,
+      "learning_rate": 0.0001739,
+      "loss": 0.0952,
       "step": 3765
     },
     {
       "epoch": 68.47,
-      "learning_rate": 0.0001739,
-      "loss": 0.1681,
+      "learning_rate": 0.00017380000000000002,
+      "loss": 0.1527,
       "step": 3766
     },
     {
       "epoch": 68.49,
-      "learning_rate": 0.00017380000000000002,
-      "loss": 0.1459,
+      "learning_rate": 0.0001737,
+      "loss": 0.1301,
       "step": 3767
     },
     {
       "epoch": 68.51,
-      "learning_rate": 0.0001737,
-      "loss": 0.13,
+      "learning_rate": 0.00017360000000000002,
+      "loss": 0.1187,
       "step": 3768
     },
     {
       "epoch": 68.53,
-      "learning_rate": 0.00017360000000000002,
-      "loss": 0.1394,
+      "learning_rate": 0.0001735,
+      "loss": 0.1129,
       "step": 3769
     },
     {
       "epoch": 68.55,
-      "learning_rate": 0.0001735,
-      "loss": 0.0904,
+      "learning_rate": 0.0001734,
+      "loss": 0.0884,
       "step": 3770
     },
     {
       "epoch": 68.56,
-      "learning_rate": 0.0001734,
-      "loss": 0.1314,
+      "learning_rate": 0.0001733,
+      "loss": 0.1402,
       "step": 3771
     },
     {
       "epoch": 68.58,
-      "learning_rate": 0.0001733,
-      "loss": 0.1688,
+      "learning_rate": 0.0001732,
+      "loss": 0.1479,
       "step": 3772
     },
     {
       "epoch": 68.6,
-      "learning_rate": 0.0001732,
-      "loss": 0.1417,
+      "learning_rate": 0.0001731,
+      "loss": 0.1492,
       "step": 3773
     },
     {
       "epoch": 68.62,
-      "learning_rate": 0.0001731,
-      "loss": 0.1256,
+      "learning_rate": 0.000173,
+      "loss": 0.1134,
       "step": 3774
     },
     {
       "epoch": 68.64,
-      "learning_rate": 0.000173,
-      "loss": 0.1554,
+      "learning_rate": 0.0001729,
+      "loss": 0.1674,
       "step": 3775
     },
     {
       "epoch": 68.65,
-      "learning_rate": 0.0001729,
-      "loss": 0.0933,
+      "learning_rate": 0.00017280000000000003,
+      "loss": 0.1067,
       "step": 3776
     },
     {
       "epoch": 68.67,
-      "learning_rate": 0.00017280000000000003,
-      "loss": 0.1417,
+      "learning_rate": 0.0001727,
+      "loss": 0.1156,
       "step": 3777
     },
     {
       "epoch": 68.69,
-      "learning_rate": 0.0001727,
-      "loss": 0.134,
+      "learning_rate": 0.0001726,
+      "loss": 0.087,
       "step": 3778
     },
     {
       "epoch": 68.71,
-      "learning_rate": 0.0001726,
-      "loss": 0.0973,
+      "learning_rate": 0.0001725,
+      "loss": 0.0993,
       "step": 3779
     },
     {
       "epoch": 68.73,
-      "learning_rate": 0.0001725,
-      "loss": 0.195,
+      "learning_rate": 0.0001724,
+      "loss": 0.1611,
       "step": 3780
     },
     {
       "epoch": 68.75,
-      "learning_rate": 0.0001724,
-      "loss": 0.1441,
+      "learning_rate": 0.00017230000000000001,
+      "loss": 0.1068,
       "step": 3781
     },
     {
       "epoch": 68.76,
-      "learning_rate": 0.00017230000000000001,
-      "loss": 0.1469,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 0.1694,
       "step": 3782
     },
     {
       "epoch": 68.78,
-      "learning_rate": 0.00017219999999999998,
-      "loss": 0.1097,
+      "learning_rate": 0.0001721,
+      "loss": 0.0947,
       "step": 3783
     },
     {
       "epoch": 68.8,
-      "learning_rate": 0.0001721,
-      "loss": 0.0931,
+      "learning_rate": 0.00017199999999999998,
+      "loss": 0.0859,
       "step": 3784
     },
     {
       "epoch": 68.82,
-      "learning_rate": 0.00017199999999999998,
-      "loss": 0.1485,
+      "learning_rate": 0.0001719,
+      "loss": 0.1413,
       "step": 3785
     },
     {
       "epoch": 68.84,
-      "learning_rate": 0.0001719,
-      "loss": 0.1401,
+      "learning_rate": 0.0001718,
+      "loss": 0.1368,
       "step": 3786
     },
     {
       "epoch": 68.85,
-      "learning_rate": 0.0001718,
-      "loss": 0.1415,
+      "learning_rate": 0.0001717,
+      "loss": 0.1136,
       "step": 3787
     },
     {
       "epoch": 68.87,
-      "learning_rate": 0.0001717,
-      "loss": 0.1486,
+      "learning_rate": 0.0001716,
+      "loss": 0.1361,
       "step": 3788
     },
     {
       "epoch": 68.89,
-      "learning_rate": 0.0001716,
-      "loss": 0.1184,
+      "learning_rate": 0.00017150000000000002,
+      "loss": 0.1307,
       "step": 3789
     },
     {
       "epoch": 68.91,
-      "learning_rate": 0.00017150000000000002,
-      "loss": 0.0836,
+      "learning_rate": 0.0001714,
+      "loss": 0.0554,
       "step": 3790
     },
     {
       "epoch": 68.93,
-      "learning_rate": 0.0001714,
-      "loss": 0.1448,
+      "learning_rate": 0.00017130000000000002,
+      "loss": 0.1354,
       "step": 3791
     },
     {
       "epoch": 68.95,
-      "learning_rate": 0.00017130000000000002,
-      "loss": 0.1175,
+      "learning_rate": 0.0001712,
+      "loss": 0.1189,
       "step": 3792
     },
     {
       "epoch": 68.96,
-      "learning_rate": 0.0001712,
-      "loss": 0.1177,
+      "learning_rate": 0.0001711,
+      "loss": 0.0998,
       "step": 3793
     },
     {
       "epoch": 68.98,
-      "learning_rate": 0.0001711,
-      "loss": 0.1565,
+      "learning_rate": 0.000171,
+      "loss": 0.1313,
       "step": 3794
     },
     {
       "epoch": 69.0,
-      "learning_rate": 0.000171,
-      "loss": 0.1122,
+      "learning_rate": 0.0001709,
+      "loss": 0.092,
       "step": 3795
     },
     {
       "epoch": 69.02,
-      "learning_rate": 0.0001709,
-      "loss": 0.1449,
+      "learning_rate": 0.0001708,
+      "loss": 0.1459,
       "step": 3796
     },
     {
       "epoch": 69.04,
-      "learning_rate": 0.0001708,
-      "loss": 0.1208,
+      "learning_rate": 0.0001707,
+      "loss": 0.0956,
       "step": 3797
     },
     {
       "epoch": 69.05,
-      "learning_rate": 0.0001707,
-      "loss": 0.1236,
+      "learning_rate": 0.0001706,
+      "loss": 0.1248,
       "step": 3798
     },
     {
       "epoch": 69.07,
-      "learning_rate": 0.0001706,
-      "loss": 0.1395,
+      "learning_rate": 0.00017050000000000002,
+      "loss": 0.1374,
       "step": 3799
     },
     {
       "epoch": 69.09,
-      "learning_rate": 0.00017050000000000002,
-      "loss": 0.1266,
+      "learning_rate": 0.0001704,
+      "loss": 0.1298,
       "step": 3800
     },
     {
       "epoch": 69.11,
-      "learning_rate": 0.0001704,
-      "loss": 0.1392,
+      "learning_rate": 0.00017030000000000002,
+      "loss": 0.1326,
       "step": 3801
     },
     {
       "epoch": 69.13,
-      "learning_rate": 0.00017030000000000002,
-      "loss": 0.1457,
+      "learning_rate": 0.0001702,
+      "loss": 0.1478,
       "step": 3802
     },
     {
       "epoch": 69.15,
-      "learning_rate": 0.0001702,
-      "loss": 0.1294,
+      "learning_rate": 0.00017010000000000001,
+      "loss": 0.1302,
       "step": 3803
     },
     {
       "epoch": 69.16,
-      "learning_rate": 0.00017010000000000001,
-      "loss": 0.1353,
+      "learning_rate": 0.00017,
+      "loss": 0.1561,
       "step": 3804
     },
     {
       "epoch": 69.18,
-      "learning_rate": 0.00017,
-      "loss": 0.1276,
+      "learning_rate": 0.0001699,
+      "loss": 0.1118,
       "step": 3805
     },
     {
       "epoch": 69.2,
-      "learning_rate": 0.0001699,
-      "loss": 0.1253,
+      "learning_rate": 0.0001698,
+      "loss": 0.1325,
       "step": 3806
     },
     {
       "epoch": 69.22,
-      "learning_rate": 0.0001698,
-      "loss": 0.0698,
+      "learning_rate": 0.0001697,
+      "loss": 0.092,
       "step": 3807
     },
     {
       "epoch": 69.24,
-      "learning_rate": 0.0001697,
-      "loss": 0.1291,
+      "learning_rate": 0.0001696,
+      "loss": 0.1309,
       "step": 3808
     },
     {
       "epoch": 69.25,
-      "learning_rate": 0.0001696,
-      "loss": 0.1372,
+      "learning_rate": 0.00016950000000000003,
+      "loss": 0.1013,
       "step": 3809
     },
     {
       "epoch": 69.27,
-      "learning_rate": 0.00016950000000000003,
-      "loss": 0.1407,
+      "learning_rate": 0.0001694,
+      "loss": 0.1266,
       "step": 3810
     },
     {
       "epoch": 69.29,
-      "learning_rate": 0.0001694,
-      "loss": 0.0922,
+      "learning_rate": 0.00016930000000000002,
+      "loss": 0.0713,
       "step": 3811
     },
     {
       "epoch": 69.31,
-      "learning_rate": 0.00016930000000000002,
-      "loss": 0.1066,
+      "learning_rate": 0.0001692,
+      "loss": 0.1008,
       "step": 3812
     },
     {
       "epoch": 69.33,
-      "learning_rate": 0.0001692,
-      "loss": 0.1064,
+      "learning_rate": 0.0001691,
+      "loss": 0.0658,
       "step": 3813
     },
     {
       "epoch": 69.35,
-      "learning_rate": 0.0001691,
-      "loss": 0.1207,
+      "learning_rate": 0.00016900000000000002,
+      "loss": 0.1176,
       "step": 3814
     },
     {
       "epoch": 69.36,
-      "learning_rate": 0.00016900000000000002,
-      "loss": 0.1719,
+      "learning_rate": 0.00016889999999999999,
+      "loss": 0.161,
       "step": 3815
     },
     {
       "epoch": 69.38,
-      "learning_rate": 0.00016889999999999999,
-      "loss": 0.1217,
+      "learning_rate": 0.0001688,
+      "loss": 0.1319,
       "step": 3816
     },
     {
       "epoch": 69.4,
-      "learning_rate": 0.0001688,
-      "loss": 0.1553,
+      "learning_rate": 0.00016869999999999998,
+      "loss": 0.1597,
       "step": 3817
     },
     {
       "epoch": 69.42,
-      "learning_rate": 0.00016869999999999998,
-      "loss": 0.1458,
+      "learning_rate": 0.0001686,
+      "loss": 0.16,
       "step": 3818
     },
     {
       "epoch": 69.44,
-      "learning_rate": 0.0001686,
-      "loss": 0.1052,
+      "learning_rate": 0.0001685,
+      "loss": 0.1118,
       "step": 3819
     },
     {
       "epoch": 69.45,
-      "learning_rate": 0.0001685,
-      "loss": 0.105,
+      "learning_rate": 0.0001684,
+      "loss": 0.1199,
       "step": 3820
     },
     {
       "epoch": 69.47,
-      "learning_rate": 0.0001684,
-      "loss": 0.1367,
+      "learning_rate": 0.0001683,
+      "loss": 0.12,
       "step": 3821
     },
     {
       "epoch": 69.49,
-      "learning_rate": 0.0001683,
-      "loss": 0.1519,
+      "learning_rate": 0.0001682,
+      "loss": 0.1325,
       "step": 3822
     },
     {
       "epoch": 69.51,
-      "learning_rate": 0.0001682,
-      "loss": 0.1604,
+      "learning_rate": 0.0001681,
+      "loss": 0.1396,
       "step": 3823
     },
     {
       "epoch": 69.53,
-      "learning_rate": 0.0001681,
-      "loss": 0.127,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 0.0992,
       "step": 3824
     },
     {
       "epoch": 69.55,
-      "learning_rate": 0.00016800000000000002,
-      "loss": 0.1066,
+      "learning_rate": 0.0001679,
+      "loss": 0.1342,
       "step": 3825
     },
     {
       "epoch": 69.56,
-      "learning_rate": 0.0001679,
-      "loss": 0.1389,
+      "learning_rate": 0.0001678,
+      "loss": 0.1166,
       "step": 3826
     },
     {
       "epoch": 69.58,
-      "learning_rate": 0.0001678,
-      "loss": 0.1303,
+      "learning_rate": 0.00016769999999999998,
+      "loss": 0.1353,
       "step": 3827
     },
     {
       "epoch": 69.6,
-      "learning_rate": 0.00016769999999999998,
-      "loss": 0.13,
+      "learning_rate": 0.0001676,
+      "loss": 0.1047,
       "step": 3828
     },
     {
       "epoch": 69.62,
-      "learning_rate": 0.0001676,
-      "loss": 0.1286,
+      "learning_rate": 0.0001675,
+      "loss": 0.128,
       "step": 3829
     },
     {
       "epoch": 69.64,
-      "learning_rate": 0.0001675,
-      "loss": 0.1163,
+      "learning_rate": 0.0001674,
+      "loss": 0.1122,
       "step": 3830
     },
     {
       "epoch": 69.65,
-      "learning_rate": 0.0001674,
-      "loss": 0.1422,
+      "learning_rate": 0.0001673,
+      "loss": 0.1301,
       "step": 3831
     },
     {
       "epoch": 69.67,
-      "learning_rate": 0.0001673,
-      "loss": 0.0932,
+      "learning_rate": 0.0001672,
+      "loss": 0.0838,
       "step": 3832
     },
     {
       "epoch": 69.69,
-      "learning_rate": 0.0001672,
-      "loss": 0.0933,
+      "learning_rate": 0.0001671,
+      "loss": 0.1054,
       "step": 3833
     },
     {
       "epoch": 69.71,
-      "learning_rate": 0.0001671,
-      "loss": 0.1503,
+      "learning_rate": 0.00016700000000000002,
+      "loss": 0.1419,
       "step": 3834
     },
     {
       "epoch": 69.73,
-      "learning_rate": 0.00016700000000000002,
-      "loss": 0.1255,
+      "learning_rate": 0.0001669,
+      "loss": 0.1133,
       "step": 3835
     },
     {
       "epoch": 69.75,
-      "learning_rate": 0.0001669,
-      "loss": 0.1286,
+      "learning_rate": 0.00016680000000000002,
+      "loss": 0.1166,
       "step": 3836
     },
     {
       "epoch": 69.76,
-      "learning_rate": 0.00016680000000000002,
-      "loss": 0.1535,
+      "learning_rate": 0.00016669999999999999,
+      "loss": 0.1381,
       "step": 3837
     },
     {
       "epoch": 69.78,
-      "learning_rate": 0.00016669999999999999,
-      "loss": 0.1046,
+      "learning_rate": 0.0001666,
+      "loss": 0.0983,
       "step": 3838
     },
     {
       "epoch": 69.8,
-      "learning_rate": 0.0001666,
-      "loss": 0.1338,
+      "learning_rate": 0.0001665,
+      "loss": 0.1098,
       "step": 3839
     },
     {
       "epoch": 69.82,
-      "learning_rate": 0.0001665,
-      "loss": 0.1749,
+      "learning_rate": 0.0001664,
+      "loss": 0.1702,
       "step": 3840
     },
     {
       "epoch": 69.84,
-      "learning_rate": 0.0001664,
-      "loss": 0.1126,
+      "learning_rate": 0.0001663,
+      "loss": 0.0935,
       "step": 3841
     },
     {
       "epoch": 69.85,
-      "learning_rate": 0.0001663,
-      "loss": 0.1219,
+      "learning_rate": 0.0001662,
+      "loss": 0.1205,
       "step": 3842
     },
     {
       "epoch": 69.87,
-      "learning_rate": 0.0001662,
-      "loss": 0.0948,
+      "learning_rate": 0.0001661,
+      "loss": 0.1052,
       "step": 3843
     },
     {
       "epoch": 69.89,
-      "learning_rate": 0.0001661,
-      "loss": 0.0971,
+      "learning_rate": 0.00016600000000000002,
+      "loss": 0.1011,
       "step": 3844
     },
     {
       "epoch": 69.91,
-      "learning_rate": 0.00016600000000000002,
-      "loss": 0.0667,
+      "learning_rate": 0.0001659,
+      "loss": 0.0829,
       "step": 3845
     },
     {
       "epoch": 69.93,
-      "learning_rate": 0.0001659,
-      "loss": 0.0884,
+      "learning_rate": 0.00016580000000000002,
+      "loss": 0.1039,
       "step": 3846
     },
     {
       "epoch": 69.95,
-      "learning_rate": 0.00016580000000000002,
-      "loss": 0.1028,
+      "learning_rate": 0.0001657,
+      "loss": 0.0993,
       "step": 3847
     },
     {
       "epoch": 69.96,
-      "learning_rate": 0.0001657,
-      "loss": 0.1376,
+      "learning_rate": 0.00016560000000000001,
+      "loss": 0.1461,
       "step": 3848
     },
     {
       "epoch": 69.98,
-      "learning_rate": 0.00016560000000000001,
-      "loss": 0.0809,
+      "learning_rate": 0.0001655,
+      "loss": 0.0864,
       "step": 3849
     },
     {
       "epoch": 70.0,
-      "learning_rate": 0.0001655,
-      "loss": 0.0698,
+      "learning_rate": 0.00016539999999999998,
+      "loss": 0.1143,
       "step": 3850
     },
     {
       "epoch": 70.02,
-      "learning_rate": 0.00016539999999999998,
-      "loss": 0.1469,
+      "learning_rate": 0.0001653,
+      "loss": 0.1572,
       "step": 3851
     },
     {
       "epoch": 70.04,
-      "learning_rate": 0.0001653,
+      "learning_rate": 0.0001652,
       "loss": 0.1141,
       "step": 3852
     },
     {
       "epoch": 70.05,
-      "learning_rate": 0.0001652,
-      "loss": 0.14,
+      "learning_rate": 0.0001651,
+      "loss": 0.1426,
       "step": 3853
     },
     {
       "epoch": 70.07,
-      "learning_rate": 0.0001651,
-      "loss": 0.0947,
+      "learning_rate": 0.000165,
+      "loss": 0.0978,
       "step": 3854
     },
     {
       "epoch": 70.09,
-      "learning_rate": 0.000165,
-      "loss": 0.1235,
+      "learning_rate": 0.0001649,
+      "loss": 0.1078,
       "step": 3855
     },
     {
       "epoch": 70.11,
-      "learning_rate": 0.0001649,
-      "loss": 0.1794,
+      "learning_rate": 0.0001648,
+      "loss": 0.1335,
       "step": 3856
     },
     {
       "epoch": 70.13,
-      "learning_rate": 0.0001648,
-      "loss": 0.1051,
+      "learning_rate": 0.00016470000000000002,
+      "loss": 0.0873,
       "step": 3857
     },
     {
       "epoch": 70.15,
-      "learning_rate": 0.00016470000000000002,
-      "loss": 0.1232,
+      "learning_rate": 0.0001646,
+      "loss": 0.1199,
       "step": 3858
     },
     {
       "epoch": 70.16,
-      "learning_rate": 0.0001646,
-      "loss": 0.1342,
+      "learning_rate": 0.00016450000000000001,
+      "loss": 0.1257,
       "step": 3859
     },
     {
       "epoch": 70.18,
-      "learning_rate": 0.00016450000000000001,
-      "loss": 0.1096,
+      "learning_rate": 0.00016439999999999998,
+      "loss": 0.1174,
       "step": 3860
     },
     {
       "epoch": 70.2,
-      "learning_rate": 0.00016439999999999998,
-      "loss": 0.124,
+      "learning_rate": 0.0001643,
+      "loss": 0.1121,
       "step": 3861
     },
     {
       "epoch": 70.22,
-      "learning_rate": 0.0001643,
-      "loss": 0.0878,
+      "learning_rate": 0.0001642,
+      "loss": 0.1085,
       "step": 3862
     },
     {
       "epoch": 70.24,
-      "learning_rate": 0.0001642,
-      "loss": 0.1811,
+      "learning_rate": 0.0001641,
+      "loss": 0.147,
       "step": 3863
     },
     {
       "epoch": 70.25,
-      "learning_rate": 0.0001641,
-      "loss": 0.1687,
+      "learning_rate": 0.000164,
+      "loss": 0.1621,
       "step": 3864
     },
     {
       "epoch": 70.27,
-      "learning_rate": 0.000164,
-      "loss": 0.1531,
+      "learning_rate": 0.0001639,
+      "loss": 0.128,
       "step": 3865
     },
     {
       "epoch": 70.29,
-      "learning_rate": 0.0001639,
-      "loss": 0.1359,
+      "learning_rate": 0.0001638,
+      "loss": 0.1391,
       "step": 3866
     },
     {
       "epoch": 70.31,
-      "learning_rate": 0.0001638,
-      "loss": 0.1131,
+      "learning_rate": 0.00016370000000000002,
+      "loss": 0.1008,
       "step": 3867
     },
     {
       "epoch": 70.33,
-      "learning_rate": 0.00016370000000000002,
-      "loss": 0.1027,
+      "learning_rate": 0.0001636,
+      "loss": 0.1257,
       "step": 3868
     },
     {
       "epoch": 70.35,
-      "learning_rate": 0.0001636,
-      "loss": 0.1592,
+      "learning_rate": 0.00016350000000000002,
+      "loss": 0.142,
       "step": 3869
     },
     {
       "epoch": 70.36,
-      "learning_rate": 0.00016350000000000002,
-      "loss": 0.1515,
+      "learning_rate": 0.0001634,
+      "loss": 0.1265,
       "step": 3870
     },
     {
       "epoch": 70.38,
-      "learning_rate": 0.0001634,
-      "loss": 0.1026,
+      "learning_rate": 0.0001633,
+      "loss": 0.101,
       "step": 3871
     },
     {
       "epoch": 70.4,
-      "learning_rate": 0.0001633,
-      "loss": 0.1515,
+      "learning_rate": 0.0001632,
+      "loss": 0.1372,
       "step": 3872
     },
     {
       "epoch": 70.42,
-      "learning_rate": 0.0001632,
-      "loss": 0.1119,
+      "learning_rate": 0.0001631,
+      "loss": 0.1016,
       "step": 3873
     },
     {
       "epoch": 70.44,
-      "learning_rate": 0.0001631,
-      "loss": 0.1121,
+      "learning_rate": 0.000163,
+      "loss": 0.1062,
       "step": 3874
     },
     {
       "epoch": 70.45,
-      "learning_rate": 0.000163,
-      "loss": 0.0822,
+      "learning_rate": 0.0001629,
+      "loss": 0.0942,
       "step": 3875
     },
     {
       "epoch": 70.47,
-      "learning_rate": 0.0001629,
-      "loss": 0.1695,
+      "learning_rate": 0.0001628,
+      "loss": 0.1539,
       "step": 3876
     },
     {
       "epoch": 70.49,
-      "learning_rate": 0.0001628,
-      "loss": 0.1468,
+      "learning_rate": 0.00016270000000000002,
+      "loss": 0.1276,
       "step": 3877
     },
     {
       "epoch": 70.51,
-      "learning_rate": 0.00016270000000000002,
-      "loss": 0.1208,
+      "learning_rate": 0.0001626,
+      "loss": 0.1076,
       "step": 3878
     },
     {
       "epoch": 70.53,
-      "learning_rate": 0.0001626,
-      "loss": 0.0953,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 0.1135,
       "step": 3879
     },
     {
       "epoch": 70.55,
-      "learning_rate": 0.00016250000000000002,
-      "loss": 0.1113,
+      "learning_rate": 0.0001624,
+      "loss": 0.1069,
       "step": 3880
     },
     {
       "epoch": 70.56,
-      "learning_rate": 0.0001624,
-      "loss": 0.0774,
+      "learning_rate": 0.00016230000000000001,
+      "loss": 0.0797,
       "step": 3881
     },
     {
       "epoch": 70.58,
-      "learning_rate": 0.00016230000000000001,
-      "loss": 0.1268,
+      "learning_rate": 0.0001622,
+      "loss": 0.1303,
       "step": 3882
     },
     {
       "epoch": 70.6,
-      "learning_rate": 0.0001622,
-      "loss": 0.1555,
+      "learning_rate": 0.0001621,
+      "loss": 0.116,
       "step": 3883
     },
     {
       "epoch": 70.62,
-      "learning_rate": 0.0001621,
-      "loss": 0.1567,
+      "learning_rate": 0.000162,
+      "loss": 0.1334,
       "step": 3884
     },
     {
       "epoch": 70.64,
-      "learning_rate": 0.000162,
-      "loss": 0.0892,
+      "learning_rate": 0.0001619,
+      "loss": 0.0853,
       "step": 3885
     },
     {
       "epoch": 70.65,
-      "learning_rate": 0.0001619,
-      "loss": 0.0743,
+      "learning_rate": 0.0001618,
+      "loss": 0.0939,
       "step": 3886
     },
     {
       "epoch": 70.67,
-      "learning_rate": 0.0001618,
-      "loss": 0.0991,
+      "learning_rate": 0.00016170000000000003,
+      "loss": 0.0815,
       "step": 3887
     },
     {
       "epoch": 70.69,
-      "learning_rate": 0.00016170000000000003,
-      "loss": 0.096,
+      "learning_rate": 0.0001616,
+      "loss": 0.0882,
       "step": 3888
     },
     {
       "epoch": 70.71,
-      "learning_rate": 0.0001616,
-      "loss": 0.1251,
+      "learning_rate": 0.0001615,
+      "loss": 0.1077,
       "step": 3889
     },
     {
       "epoch": 70.73,
-      "learning_rate": 0.0001615,
-      "loss": 0.0855,
+      "learning_rate": 0.0001614,
+      "loss": 0.0672,
       "step": 3890
     },
     {
       "epoch": 70.75,
-      "learning_rate": 0.0001614,
-      "loss": 0.1172,
+      "learning_rate": 0.0001613,
+      "loss": 0.1129,
       "step": 3891
     },
     {
       "epoch": 70.76,
-      "learning_rate": 0.0001613,
-      "loss": 0.1642,
+      "learning_rate": 0.00016120000000000002,
+      "loss": 0.1604,
       "step": 3892
     },
     {
       "epoch": 70.78,
-      "learning_rate": 0.00016120000000000002,
-      "loss": 0.1108,
+      "learning_rate": 0.00016109999999999999,
+      "loss": 0.1275,
       "step": 3893
     },
     {
       "epoch": 70.8,
-      "learning_rate": 0.00016109999999999999,
-      "loss": 0.1118,
+      "learning_rate": 0.000161,
+      "loss": 0.1061,
       "step": 3894
     },
     {
       "epoch": 70.82,
-      "learning_rate": 0.000161,
-      "loss": 0.154,
+      "learning_rate": 0.00016089999999999998,
+      "loss": 0.1474,
       "step": 3895
     },
     {
       "epoch": 70.84,
-      "learning_rate": 0.00016089999999999998,
-      "loss": 0.1376,
+      "learning_rate": 0.0001608,
+      "loss": 0.1218,
       "step": 3896
     },
     {
       "epoch": 70.85,
-      "learning_rate": 0.0001608,
-      "loss": 0.09,
+      "learning_rate": 0.0001607,
+      "loss": 0.085,
       "step": 3897
     },
     {
       "epoch": 70.87,
-      "learning_rate": 0.0001607,
-      "loss": 0.1012,
+      "learning_rate": 0.0001606,
+      "loss": 0.0888,
       "step": 3898
     },
     {
       "epoch": 70.89,
-      "learning_rate": 0.0001606,
-      "loss": 0.1232,
+      "learning_rate": 0.0001605,
+      "loss": 0.0948,
       "step": 3899
     },
     {
       "epoch": 70.91,
-      "learning_rate": 0.0001605,
-      "loss": 0.0939,
+      "learning_rate": 0.0001604,
+      "loss": 0.114,
       "step": 3900
     },
     {
       "epoch": 70.93,
-      "learning_rate": 0.0001604,
-      "loss": 0.1273,
+      "learning_rate": 0.0001603,
+      "loss": 0.1313,
       "step": 3901
     },
     {
       "epoch": 70.95,
-      "learning_rate": 0.0001603,
-      "loss": 0.1562,
+      "learning_rate": 0.00016020000000000002,
+      "loss": 0.1345,
       "step": 3902
     },
     {
       "epoch": 70.96,
-      "learning_rate": 0.00016020000000000002,
-      "loss": 0.1377,
+      "learning_rate": 0.0001601,
+      "loss": 0.1016,
       "step": 3903
     },
     {
       "epoch": 70.98,
-      "learning_rate": 0.0001601,
-      "loss": 0.0564,
+      "learning_rate": 0.00016,
+      "loss": 0.0603,
       "step": 3904
     },
     {
       "epoch": 71.0,
-      "learning_rate": 0.00016,
+      "learning_rate": 0.00015989999999999998,
       "loss": 0.0924,
       "step": 3905
     },
     {
       "epoch": 71.02,
-      "learning_rate": 0.00015989999999999998,
-      "loss": 0.1751,
+      "learning_rate": 0.0001598,
+      "loss": 0.1552,
       "step": 3906
     },
     {
       "epoch": 71.04,
-      "learning_rate": 0.0001598,
-      "loss": 0.0955,
+      "learning_rate": 0.0001597,
+      "loss": 0.0927,
       "step": 3907
     },
     {
       "epoch": 71.05,
-      "learning_rate": 0.0001597,
-      "loss": 0.1609,
+      "learning_rate": 0.0001596,
+      "loss": 0.1309,
       "step": 3908
     },
     {
       "epoch": 71.07,
-      "learning_rate": 0.0001596,
-      "loss": 0.1325,
+      "learning_rate": 0.0001595,
+      "loss": 0.1036,
       "step": 3909
     },
     {
       "epoch": 71.09,
-      "learning_rate": 0.0001595,
-      "loss": 0.1555,
+      "learning_rate": 0.0001594,
+      "loss": 0.1453,
       "step": 3910
     },
     {
       "epoch": 71.11,
-      "learning_rate": 0.0001594,
-      "loss": 0.1395,
+      "learning_rate": 0.0001593,
+      "loss": 0.1602,
       "step": 3911
     },
     {
       "epoch": 71.13,
-      "learning_rate": 0.0001593,
-      "loss": 0.1649,
+      "learning_rate": 0.00015920000000000002,
+      "loss": 0.1499,
       "step": 3912
     },
     {
       "epoch": 71.15,
-      "learning_rate": 0.00015920000000000002,
-      "loss": 0.1401,
+      "learning_rate": 0.0001591,
+      "loss": 0.1254,
       "step": 3913
     },
     {
       "epoch": 71.16,
-      "learning_rate": 0.0001591,
-      "loss": 0.1368,
+      "learning_rate": 0.00015900000000000002,
+      "loss": 0.1254,
       "step": 3914
     },
     {
       "epoch": 71.18,
-      "learning_rate": 0.00015900000000000002,
-      "loss": 0.153,
+      "learning_rate": 0.0001589,
+      "loss": 0.137,
       "step": 3915
     },
     {
       "epoch": 71.2,
-      "learning_rate": 0.0001589,
-      "loss": 0.1192,
+      "learning_rate": 0.0001588,
+      "loss": 0.1085,
       "step": 3916
     },
     {
       "epoch": 71.22,
-      "learning_rate": 0.0001588,
-      "loss": 0.1056,
+      "learning_rate": 0.0001587,
+      "loss": 0.0961,
       "step": 3917
     },
     {
       "epoch": 71.24,
-      "learning_rate": 0.0001587,
-      "loss": 0.1347,
+      "learning_rate": 0.0001586,
+      "loss": 0.1154,
       "step": 3918
     },
     {
       "epoch": 71.25,
-      "learning_rate": 0.0001586,
-      "loss": 0.1272,
+      "learning_rate": 0.0001585,
+      "loss": 0.1515,
       "step": 3919
     },
     {
       "epoch": 71.27,
-      "learning_rate": 0.0001585,
-      "loss": 0.068,
+      "learning_rate": 0.00015840000000000003,
+      "loss": 0.0631,
       "step": 3920
     },
     {
       "epoch": 71.29,
-      "learning_rate": 0.00015840000000000003,
-      "loss": 0.1199,
+      "learning_rate": 0.0001583,
+      "loss": 0.1134,
       "step": 3921
     },
     {
       "epoch": 71.31,
-      "learning_rate": 0.0001583,
-      "loss": 0.1424,
+      "learning_rate": 0.00015820000000000002,
+      "loss": 0.1066,
       "step": 3922
     },
     {
       "epoch": 71.33,
-      "learning_rate": 0.00015820000000000002,
-      "loss": 0.133,
+      "learning_rate": 0.0001581,
+      "loss": 0.0956,
       "step": 3923
     },
     {
       "epoch": 71.35,
-      "learning_rate": 0.0001581,
-      "loss": 0.1137,
+      "learning_rate": 0.000158,
+      "loss": 0.1073,
       "step": 3924
     },
     {
       "epoch": 71.36,
-      "learning_rate": 0.000158,
-      "loss": 0.1062,
+      "learning_rate": 0.00015790000000000002,
+      "loss": 0.0996,
       "step": 3925
     },
     {
       "epoch": 71.38,
-      "learning_rate": 0.00015790000000000002,
-      "loss": 0.1415,
+      "learning_rate": 0.0001578,
+      "loss": 0.1159,
       "step": 3926
     },
     {
       "epoch": 71.4,
-      "learning_rate": 0.0001578,
-      "loss": 0.1375,
+      "learning_rate": 0.0001577,
+      "loss": 0.1021,
       "step": 3927
     },
     {
       "epoch": 71.42,
-      "learning_rate": 0.0001577,
-      "loss": 0.1209,
+      "learning_rate": 0.00015759999999999998,
+      "loss": 0.1243,
       "step": 3928
     },
     {
       "epoch": 71.44,
-      "learning_rate": 0.00015759999999999998,
-      "loss": 0.1264,
+      "learning_rate": 0.0001575,
+      "loss": 0.1095,
       "step": 3929
     },
     {
       "epoch": 71.45,
-      "learning_rate": 0.0001575,
-      "loss": 0.0849,
+      "learning_rate": 0.0001574,
+      "loss": 0.0666,
       "step": 3930
     },
     {
       "epoch": 71.47,
-      "learning_rate": 0.0001574,
-      "loss": 0.1685,
+      "learning_rate": 0.0001573,
+      "loss": 0.1508,
       "step": 3931
     },
     {
       "epoch": 71.49,
-      "learning_rate": 0.0001573,
-      "loss": 0.2053,
+      "learning_rate": 0.0001572,
+      "loss": 0.1618,
       "step": 3932
     },
     {
       "epoch": 71.51,
-      "learning_rate": 0.0001572,
-      "loss": 0.1347,
+      "learning_rate": 0.0001571,
+      "loss": 0.1476,
       "step": 3933
     },
     {
       "epoch": 71.53,
-      "learning_rate": 0.0001571,
-      "loss": 0.123,
+      "learning_rate": 0.000157,
+      "loss": 0.1104,
       "step": 3934
     },
     {
       "epoch": 71.55,
-      "learning_rate": 0.000157,
-      "loss": 0.1382,
+      "learning_rate": 0.00015690000000000002,
+      "loss": 0.124,
       "step": 3935
     },
     {
       "epoch": 71.56,
-      "learning_rate": 0.00015690000000000002,
-      "loss": 0.1084,
+      "learning_rate": 0.0001568,
+      "loss": 0.0826,
       "step": 3936
     },
     {
       "epoch": 71.58,
-      "learning_rate": 0.0001568,
-      "loss": 0.1271,
+      "learning_rate": 0.00015670000000000001,
+      "loss": 0.109,
       "step": 3937
     },
     {
       "epoch": 71.6,
-      "learning_rate": 0.00015670000000000001,
-      "loss": 0.1274,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 0.1049,
       "step": 3938
     },
     {
       "epoch": 71.62,
-      "learning_rate": 0.00015659999999999998,
-      "loss": 0.0934,
+      "learning_rate": 0.0001565,
+      "loss": 0.1005,
       "step": 3939
     },
     {
       "epoch": 71.64,
-      "learning_rate": 0.0001565,
-      "loss": 0.1216,
+      "learning_rate": 0.0001564,
+      "loss": 0.1128,
       "step": 3940
     },
     {
       "epoch": 71.65,
-      "learning_rate": 0.0001564,
-      "loss": 0.0886,
+      "learning_rate": 0.0001563,
+      "loss": 0.084,
       "step": 3941
     },
     {
       "epoch": 71.67,
-      "learning_rate": 0.0001563,
-      "loss": 0.1132,
+      "learning_rate": 0.0001562,
+      "loss": 0.0926,
       "step": 3942
     },
     {
       "epoch": 71.69,
-      "learning_rate": 0.0001562,
-      "loss": 0.1298,
+      "learning_rate": 0.0001561,
+      "loss": 0.131,
       "step": 3943
     },
     {
       "epoch": 71.71,
-      "learning_rate": 0.0001561,
-      "loss": 0.1227,
+      "learning_rate": 0.000156,
+      "loss": 0.11,
       "step": 3944
     },
     {
       "epoch": 71.73,
-      "learning_rate": 0.000156,
-      "loss": 0.1033,
+      "learning_rate": 0.00015590000000000002,
+      "loss": 0.1008,
       "step": 3945
     },
     {
       "epoch": 71.75,
-      "learning_rate": 0.00015590000000000002,
-      "loss": 0.0969,
+      "learning_rate": 0.0001558,
+      "loss": 0.0989,
       "step": 3946
     },
     {
       "epoch": 71.76,
-      "learning_rate": 0.0001558,
-      "loss": 0.1479,
+      "learning_rate": 0.00015570000000000002,
+      "loss": 0.1161,
       "step": 3947
     },
     {
       "epoch": 71.78,
-      "learning_rate": 0.00015570000000000002,
-      "loss": 0.1553,
+      "learning_rate": 0.0001556,
+      "loss": 0.1545,
       "step": 3948
     },
     {
       "epoch": 71.8,
-      "learning_rate": 0.0001556,
-      "loss": 0.0998,
+      "learning_rate": 0.0001555,
+      "loss": 0.1091,
       "step": 3949
     },
     {
       "epoch": 71.82,
-      "learning_rate": 0.0001555,
-      "loss": 0.1237,
+      "learning_rate": 0.0001554,
+      "loss": 0.1152,
       "step": 3950
     },
     {
       "epoch": 71.84,
-      "learning_rate": 0.0001554,
-      "loss": 0.1389,
+      "learning_rate": 0.0001553,
+      "loss": 0.1469,
       "step": 3951
     },
     {
       "epoch": 71.85,
-      "learning_rate": 0.0001553,
-      "loss": 0.1266,
+      "learning_rate": 0.0001552,
+      "loss": 0.118,
       "step": 3952
     },
     {
       "epoch": 71.87,
-      "learning_rate": 0.0001552,
-      "loss": 0.1134,
+      "learning_rate": 0.0001551,
+      "loss": 0.0966,
       "step": 3953
     },
     {
       "epoch": 71.89,
-      "learning_rate": 0.0001551,
-      "loss": 0.1173,
+      "learning_rate": 0.000155,
+      "loss": 0.0926,
       "step": 3954
     },
     {
       "epoch": 71.91,
-      "learning_rate": 0.000155,
-      "loss": 0.1253,
+      "learning_rate": 0.00015490000000000002,
+      "loss": 0.1559,
       "step": 3955
     },
     {
       "epoch": 71.93,
-      "learning_rate": 0.00015490000000000002,
-      "loss": 0.1233,
+      "learning_rate": 0.0001548,
+      "loss": 0.1346,
       "step": 3956
     },
     {
       "epoch": 71.95,
-      "learning_rate": 0.0001548,
-      "loss": 0.1214,
+      "learning_rate": 0.00015470000000000002,
+      "loss": 0.1111,
       "step": 3957
     },
     {
       "epoch": 71.96,
-      "learning_rate": 0.00015470000000000002,
-      "loss": 0.1367,
+      "learning_rate": 0.0001546,
+      "loss": 0.1391,
       "step": 3958
     },
     {
       "epoch": 71.98,
-      "learning_rate": 0.0001546,
-      "loss": 0.1005,
+      "learning_rate": 0.00015450000000000001,
+      "loss": 0.0944,
       "step": 3959
     },
     {
       "epoch": 72.0,
-      "learning_rate": 0.00015450000000000001,
-      "loss": 0.1375,
+      "learning_rate": 0.0001544,
+      "loss": 0.1456,
       "step": 3960
     },
     {
       "epoch": 72.02,
-      "learning_rate": 0.0001544,
-      "loss": 0.1558,
+      "learning_rate": 0.00015429999999999998,
+      "loss": 0.134,
       "step": 3961
     },
     {
       "epoch": 72.04,
-      "learning_rate": 0.00015429999999999998,
-      "loss": 0.156,
+      "learning_rate": 0.0001542,
+      "loss": 0.1181,
       "step": 3962
     },
     {
       "epoch": 72.05,
-      "learning_rate": 0.0001542,
-      "loss": 0.1437,
+      "learning_rate": 0.00015409999999999998,
+      "loss": 0.1117,
       "step": 3963
     },
     {
       "epoch": 72.07,
-      "learning_rate": 0.00015409999999999998,
-      "loss": 0.0873,
+      "learning_rate": 0.000154,
+      "loss": 0.0814,
       "step": 3964
     },
     {
       "epoch": 72.09,
-      "learning_rate": 0.000154,
-      "loss": 0.1199,
+      "learning_rate": 0.0001539,
+      "loss": 0.1162,
       "step": 3965
     },
     {
       "epoch": 72.11,
-      "learning_rate": 0.0001539,
-      "loss": 0.1161,
+      "learning_rate": 0.0001538,
+      "loss": 0.1139,
       "step": 3966
     },
     {
       "epoch": 72.13,
-      "learning_rate": 0.0001538,
-      "loss": 0.0993,
+      "learning_rate": 0.0001537,
+      "loss": 0.0912,
       "step": 3967
     },
     {
       "epoch": 72.15,
-      "learning_rate": 0.0001537,
-      "loss": 0.1175,
+      "learning_rate": 0.0001536,
+      "loss": 0.1053,
       "step": 3968
     },
     {
       "epoch": 72.16,
-      "learning_rate": 0.0001536,
-      "loss": 0.1115,
+      "learning_rate": 0.0001535,
+      "loss": 0.1038,
       "step": 3969
     },
     {
       "epoch": 72.18,
-      "learning_rate": 0.0001535,
-      "loss": 0.0713,
+      "learning_rate": 0.00015340000000000002,
+      "loss": 0.0684,
       "step": 3970
     },
     {
       "epoch": 72.2,
-      "learning_rate": 0.00015340000000000002,
-      "loss": 0.0939,
+      "learning_rate": 0.00015329999999999999,
+      "loss": 0.0789,
       "step": 3971
     },
     {
       "epoch": 72.22,
-      "learning_rate": 0.00015329999999999999,
-      "loss": 0.1137,
+      "learning_rate": 0.0001532,
+      "loss": 0.0881,
       "step": 3972
     },
     {
       "epoch": 72.24,
-      "learning_rate": 0.0001532,
-      "loss": 0.1494,
+      "learning_rate": 0.0001531,
+      "loss": 0.1176,
       "step": 3973
     },
     {
       "epoch": 72.25,
-      "learning_rate": 0.0001531,
-      "loss": 0.1669,
+      "learning_rate": 0.000153,
+      "loss": 0.1324,
       "step": 3974
     },
     {
       "epoch": 72.27,
-      "learning_rate": 0.000153,
-      "loss": 0.1316,
+      "learning_rate": 0.0001529,
+      "loss": 0.1357,
       "step": 3975
     },
     {
       "epoch": 72.29,
-      "learning_rate": 0.0001529,
-      "loss": 0.0935,
+      "learning_rate": 0.0001528,
+      "loss": 0.0904,
       "step": 3976
     },
     {
       "epoch": 72.31,
-      "learning_rate": 0.0001528,
-      "loss": 0.0872,
+      "learning_rate": 0.0001527,
+      "loss": 0.0606,
       "step": 3977
     },
     {
       "epoch": 72.33,
-      "learning_rate": 0.0001527,
-      "loss": 0.1203,
+      "learning_rate": 0.00015260000000000002,
+      "loss": 0.0843,
       "step": 3978
     },
     {
       "epoch": 72.35,
-      "learning_rate": 0.00015260000000000002,
-      "loss": 0.1025,
+      "learning_rate": 0.0001525,
+      "loss": 0.1017,
       "step": 3979
     },
     {
       "epoch": 72.36,
-      "learning_rate": 0.0001525,
-      "loss": 0.12,
+      "learning_rate": 0.00015240000000000002,
+      "loss": 0.1051,
       "step": 3980
     },
     {
       "epoch": 72.38,
-      "learning_rate": 0.00015240000000000002,
-      "loss": 0.1371,
+      "learning_rate": 0.0001523,
+      "loss": 0.12,
       "step": 3981
     },
     {
       "epoch": 72.4,
-      "learning_rate": 0.0001523,
-      "loss": 0.1149,
+      "learning_rate": 0.0001522,
+      "loss": 0.1324,
       "step": 3982
     },
     {
       "epoch": 72.42,
-      "learning_rate": 0.0001522,
-      "loss": 0.1313,
+      "learning_rate": 0.0001521,
+      "loss": 0.1092,
       "step": 3983
     },
     {
       "epoch": 72.44,
-      "learning_rate": 0.0001521,
-      "loss": 0.0979,
+      "learning_rate": 0.000152,
+      "loss": 0.1067,
       "step": 3984
     },
     {
       "epoch": 72.45,
-      "learning_rate": 0.000152,
-      "loss": 0.0985,
+      "learning_rate": 0.0001519,
+      "loss": 0.1028,
       "step": 3985
     },
     {
       "epoch": 72.47,
-      "learning_rate": 0.0001519,
-      "loss": 0.1201,
+      "learning_rate": 0.0001518,
+      "loss": 0.1233,
       "step": 3986
     },
     {
       "epoch": 72.49,
-      "learning_rate": 0.0001518,
-      "loss": 0.1109,
+      "learning_rate": 0.0001517,
+      "loss": 0.1114,
       "step": 3987
     },
     {
       "epoch": 72.51,
-      "learning_rate": 0.0001517,
-      "loss": 0.1452,
+      "learning_rate": 0.00015160000000000003,
+      "loss": 0.1154,
       "step": 3988
     },
     {
       "epoch": 72.53,
-      "learning_rate": 0.00015160000000000003,
-      "loss": 0.0983,
+      "learning_rate": 0.0001515,
+      "loss": 0.0775,
       "step": 3989
     },
     {
       "epoch": 72.55,
-      "learning_rate": 0.0001515,
-      "loss": 0.0787,
+      "learning_rate": 0.00015140000000000002,
+      "loss": 0.0791,
       "step": 3990
     },
     {
       "epoch": 72.56,
-      "learning_rate": 0.00015140000000000002,
-      "loss": 0.0947,
+      "learning_rate": 0.0001513,
+      "loss": 0.0857,
       "step": 3991
     },
     {
       "epoch": 72.58,
-      "learning_rate": 0.0001513,
-      "loss": 0.1165,
+      "learning_rate": 0.00015120000000000002,
+      "loss": 0.1024,
       "step": 3992
     },
     {
       "epoch": 72.6,
-      "learning_rate": 0.00015120000000000002,
-      "loss": 0.1249,
+      "learning_rate": 0.00015110000000000001,
+      "loss": 0.1304,
       "step": 3993
     },
     {
       "epoch": 72.62,
-      "learning_rate": 0.00015110000000000001,
-      "loss": 0.1174,
+      "learning_rate": 0.000151,
+      "loss": 0.0976,
       "step": 3994
     },
     {
       "epoch": 72.64,
-      "learning_rate": 0.000151,
-      "loss": 0.1248,
+      "learning_rate": 0.0001509,
+      "loss": 0.0968,
       "step": 3995
     },
     {
       "epoch": 72.65,
-      "learning_rate": 0.0001509,
-      "loss": 0.1176,
+      "learning_rate": 0.00015079999999999998,
+      "loss": 0.1247,
       "step": 3996
     },
     {
       "epoch": 72.67,
-      "learning_rate": 0.00015079999999999998,
-      "loss": 0.1273,
+      "learning_rate": 0.0001507,
+      "loss": 0.1034,
       "step": 3997
     },
     {
       "epoch": 72.69,
-      "learning_rate": 0.0001507,
-      "loss": 0.0723,
+      "learning_rate": 0.0001506,
+      "loss": 0.0806,
       "step": 3998
     },
     {
       "epoch": 72.71,
-      "learning_rate": 0.0001506,
-      "loss": 0.1166,
+      "learning_rate": 0.0001505,
+      "loss": 0.087,
       "step": 3999
     },
     {
       "epoch": 72.73,
-      "learning_rate": 0.0001505,
-      "loss": 0.1163,
+      "learning_rate": 0.0001504,
+      "loss": 0.1107,
       "step": 4000
     },
     {
       "epoch": 72.73,
-      "eval_cer": 0.08262492691044843,
-      "eval_loss": 0.4061971604824066,
-      "eval_runtime": 13.9927,
-      "eval_samples_per_second": 117.705,
-      "eval_steps_per_second": 1.858,
-      "eval_wer": 0.33326524359105303,
+      "eval_cer": 0.08099071950104199,
+      "eval_loss": 0.3907454311847687,
+      "eval_runtime": 14.2457,
+      "eval_samples_per_second": 115.613,
+      "eval_steps_per_second": 1.825,
+      "eval_wer": 0.3306097436421203,
       "step": 4000
     },
     {
       "epoch": 72.75,
-      "learning_rate": 0.0001504,
-      "loss": 0.1051,
+      "learning_rate": 0.0001503,
+      "loss": 0.098,
       "step": 4001
     },
     {
       "epoch": 72.76,
-      "learning_rate": 0.0001503,
-      "loss": 0.1383,
+      "learning_rate": 0.0001502,
+      "loss": 0.1334,
       "step": 4002
     },
     {
       "epoch": 72.78,
-      "learning_rate": 0.0001502,
-      "loss": 0.1199,
+      "learning_rate": 0.00015010000000000002,
+      "loss": 0.1132,
       "step": 4003
     },
     {
       "epoch": 72.8,
-      "learning_rate": 0.00015010000000000002,
-      "loss": 0.1062,
+      "learning_rate": 0.00015,
+      "loss": 0.117,
       "step": 4004
     },
     {
       "epoch": 72.82,
-      "learning_rate": 0.00015,
-      "loss": 0.1099,
+      "learning_rate": 0.0001499,
+      "loss": 0.1057,
       "step": 4005
     },
     {
       "epoch": 72.84,
-      "learning_rate": 0.0001499,
-      "loss": 0.1334,
+      "learning_rate": 0.00014979999999999998,
+      "loss": 0.1132,
       "step": 4006
     },
     {
       "epoch": 72.85,
-      "learning_rate": 0.00014979999999999998,
-      "loss": 0.1252,
+      "learning_rate": 0.0001497,
+      "loss": 0.1081,
       "step": 4007
     },
     {
       "epoch": 72.87,
-      "learning_rate": 0.0001497,
-      "loss": 0.1167,
+      "learning_rate": 0.0001496,
+      "loss": 0.1156,
       "step": 4008
     },
     {
       "epoch": 72.89,
-      "learning_rate": 0.0001496,
-      "loss": 0.127,
+      "learning_rate": 0.0001495,
+      "loss": 0.1147,
       "step": 4009
     },
     {
       "epoch": 72.91,
-      "learning_rate": 0.0001495,
-      "loss": 0.1306,
+      "learning_rate": 0.0001494,
+      "loss": 0.0857,
       "step": 4010
     },
     {
       "epoch": 72.93,
-      "learning_rate": 0.0001494,
-      "loss": 0.1424,
+      "learning_rate": 0.0001493,
+      "loss": 0.1058,
       "step": 4011
     },
     {
       "epoch": 72.95,
-      "learning_rate": 0.0001493,
-      "loss": 0.1451,
+      "learning_rate": 0.0001492,
+      "loss": 0.124,
       "step": 4012
     },
     {
       "epoch": 72.96,
-      "learning_rate": 0.0001492,
-      "loss": 0.1393,
+      "learning_rate": 0.00014910000000000002,
+      "loss": 0.1182,
       "step": 4013
     },
     {
       "epoch": 72.98,
-      "learning_rate": 0.00014910000000000002,
-      "loss": 0.1107,
+      "learning_rate": 0.000149,
+      "loss": 0.0919,
       "step": 4014
     },
     {
       "epoch": 73.0,
-      "learning_rate": 0.000149,
-      "loss": 0.2028,
+      "learning_rate": 0.00014890000000000001,
+      "loss": 0.1394,
       "step": 4015
     },
     {
       "epoch": 73.02,
-      "learning_rate": 0.00014890000000000001,
-      "loss": 0.1169,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 0.1039,
       "step": 4016
     },
     {
       "epoch": 73.04,
-      "learning_rate": 0.00014879999999999998,
-      "loss": 0.1264,
+      "learning_rate": 0.0001487,
+      "loss": 0.1043,
       "step": 4017
     },
     {
       "epoch": 73.05,
-      "learning_rate": 0.0001487,
-      "loss": 0.0941,
+      "learning_rate": 0.0001486,
+      "loss": 0.0855,
       "step": 4018
     },
     {
       "epoch": 73.07,
-      "learning_rate": 0.0001486,
-      "loss": 0.1362,
+      "learning_rate": 0.0001485,
+      "loss": 0.1462,
       "step": 4019
     },
     {
       "epoch": 73.09,
-      "learning_rate": 0.0001485,
-      "loss": 0.1337,
+      "learning_rate": 0.0001484,
+      "loss": 0.1154,
       "step": 4020
     },
     {
       "epoch": 73.11,
-      "learning_rate": 0.0001484,
-      "loss": 0.1182,
+      "learning_rate": 0.0001483,
+      "loss": 0.0942,
       "step": 4021
     },
     {
       "epoch": 73.13,
-      "learning_rate": 0.0001483,
-      "loss": 0.1681,
+      "learning_rate": 0.0001482,
+      "loss": 0.1471,
       "step": 4022
     },
     {
       "epoch": 73.15,
-      "learning_rate": 0.0001482,
-      "loss": 0.12,
+      "learning_rate": 0.00014810000000000002,
+      "loss": 0.1078,
       "step": 4023
     },
     {
       "epoch": 73.16,
-      "learning_rate": 0.00014810000000000002,
-      "loss": 0.1258,
+      "learning_rate": 0.000148,
+      "loss": 0.1287,
       "step": 4024
     },
     {
       "epoch": 73.18,
-      "learning_rate": 0.000148,
-      "loss": 0.132,
+      "learning_rate": 0.00014790000000000002,
+      "loss": 0.1502,
       "step": 4025
     },
     {
       "epoch": 73.2,
-      "learning_rate": 0.00014790000000000002,
-      "loss": 0.1172,
+      "learning_rate": 0.0001478,
+      "loss": 0.1171,
       "step": 4026
     },
     {
       "epoch": 73.22,
-      "learning_rate": 0.0001478,
-      "loss": 0.1186,
+      "learning_rate": 0.0001477,
+      "loss": 0.1079,
       "step": 4027
     },
     {
       "epoch": 73.24,
-      "learning_rate": 0.0001477,
-      "loss": 0.1064,
+      "learning_rate": 0.0001476,
+      "loss": 0.1103,
       "step": 4028
     },
     {
       "epoch": 73.25,
-      "learning_rate": 0.0001476,
-      "loss": 0.1163,
+      "learning_rate": 0.0001475,
+      "loss": 0.0878,
       "step": 4029
     },
     {
       "epoch": 73.27,
-      "learning_rate": 0.0001475,
-      "loss": 0.0948,
+      "learning_rate": 0.0001474,
+      "loss": 0.0916,
       "step": 4030
     },
     {
       "epoch": 73.29,
-      "learning_rate": 0.0001474,
-      "loss": 0.1101,
+      "learning_rate": 0.0001473,
+      "loss": 0.0997,
       "step": 4031
     },
     {
       "epoch": 73.31,
-      "learning_rate": 0.0001473,
-      "loss": 0.107,
+      "learning_rate": 0.0001472,
+      "loss": 0.1001,
       "step": 4032
     },
     {
       "epoch": 73.33,
-      "learning_rate": 0.0001472,
-      "loss": 0.1164,
+      "learning_rate": 0.00014710000000000002,
+      "loss": 0.0913,
       "step": 4033
     },
     {
       "epoch": 73.35,
-      "learning_rate": 0.00014710000000000002,
-      "loss": 0.1259,
+      "learning_rate": 0.000147,
+      "loss": 0.1025,
       "step": 4034
     },
     {
       "epoch": 73.36,
-      "learning_rate": 0.000147,
-      "loss": 0.1187,
+      "learning_rate": 0.0001469,
+      "loss": 0.1098,
       "step": 4035
     },
     {
       "epoch": 73.38,
-      "learning_rate": 0.0001469,
-      "loss": 0.1123,
+      "learning_rate": 0.00014680000000000002,
+      "loss": 0.1094,
       "step": 4036
     },
     {
       "epoch": 73.4,
-      "learning_rate": 0.00014680000000000002,
-      "loss": 0.1037,
+      "learning_rate": 0.0001467,
+      "loss": 0.0955,
       "step": 4037
     },
     {
       "epoch": 73.42,
-      "learning_rate": 0.0001467,
-      "loss": 0.1543,
+      "learning_rate": 0.0001466,
+      "loss": 0.1154,
       "step": 4038
     },
     {
       "epoch": 73.44,
-      "learning_rate": 0.0001466,
-      "loss": 0.0816,
+      "learning_rate": 0.00014649999999999998,
+      "loss": 0.0887,
       "step": 4039
     },
     {
       "epoch": 73.45,
-      "learning_rate": 0.00014649999999999998,
-      "loss": 0.0946,
+      "learning_rate": 0.0001464,
+      "loss": 0.1068,
       "step": 4040
     },
     {
       "epoch": 73.47,
-      "learning_rate": 0.0001464,
-      "loss": 0.1324,
+      "learning_rate": 0.0001463,
+      "loss": 0.1235,
       "step": 4041
     },
     {
       "epoch": 73.49,
-      "learning_rate": 0.0001463,
-      "loss": 0.1228,
+      "learning_rate": 0.0001462,
+      "loss": 0.1267,
       "step": 4042
     },
     {
       "epoch": 73.51,
-      "learning_rate": 0.0001462,
-      "loss": 0.1122,
+      "learning_rate": 0.0001461,
+      "loss": 0.1079,
       "step": 4043
     },
     {
       "epoch": 73.53,
-      "learning_rate": 0.0001461,
-      "loss": 0.1331,
+      "learning_rate": 0.000146,
+      "loss": 0.1176,
       "step": 4044
     },
     {
       "epoch": 73.55,
-      "learning_rate": 0.000146,
-      "loss": 0.1004,
+      "learning_rate": 0.0001459,
+      "loss": 0.1097,
       "step": 4045
     },
     {
       "epoch": 73.56,
-      "learning_rate": 0.0001459,
-      "loss": 0.1005,
+      "learning_rate": 0.00014580000000000002,
+      "loss": 0.0949,
       "step": 4046
     },
     {
       "epoch": 73.58,
-      "learning_rate": 0.00014580000000000002,
-      "loss": 0.1579,
+      "learning_rate": 0.0001457,
+      "loss": 0.1214,
       "step": 4047
     },
     {
       "epoch": 73.6,
-      "learning_rate": 0.0001457,
-      "loss": 0.1251,
+      "learning_rate": 0.00014560000000000002,
+      "loss": 0.1113,
       "step": 4048
     },
     {
       "epoch": 73.62,
-      "learning_rate": 0.00014560000000000002,
-      "loss": 0.1131,
+      "learning_rate": 0.00014549999999999999,
+      "loss": 0.1032,
       "step": 4049
     },
     {
       "epoch": 73.64,
-      "learning_rate": 0.00014549999999999999,
-      "loss": 0.1257,
+      "learning_rate": 0.0001454,
+      "loss": 0.1087,
       "step": 4050
     },
     {
       "epoch": 73.65,
-      "learning_rate": 0.0001454,
-      "loss": 0.1468,
+      "learning_rate": 0.0001453,
+      "loss": 0.1034,
       "step": 4051
     },
     {
       "epoch": 73.67,
-      "learning_rate": 0.0001453,
-      "loss": 0.0879,
+      "learning_rate": 0.0001452,
+      "loss": 0.0755,
       "step": 4052
     },
     {
       "epoch": 73.69,
-      "learning_rate": 0.0001452,
-      "loss": 0.1505,
+      "learning_rate": 0.0001451,
+      "loss": 0.1219,
       "step": 4053
     },
     {
       "epoch": 73.71,
-      "learning_rate": 0.0001451,
-      "loss": 0.0999,
+      "learning_rate": 0.000145,
+      "loss": 0.1272,
       "step": 4054
     },
     {
       "epoch": 73.73,
-      "learning_rate": 0.000145,
-      "loss": 0.1124,
+      "learning_rate": 0.0001449,
+      "loss": 0.1089,
       "step": 4055
     },
     {
       "epoch": 73.75,
-      "learning_rate": 0.0001449,
-      "loss": 0.1274,
+      "learning_rate": 0.00014480000000000002,
+      "loss": 0.1134,
       "step": 4056
     },
     {
       "epoch": 73.76,
-      "learning_rate": 0.00014480000000000002,
-      "loss": 0.0944,
+      "learning_rate": 0.0001447,
+      "loss": 0.077,
       "step": 4057
     },
     {
       "epoch": 73.78,
-      "learning_rate": 0.0001447,
-      "loss": 0.1295,
+      "learning_rate": 0.00014460000000000002,
+      "loss": 0.1106,
       "step": 4058
     },
     {
       "epoch": 73.8,
-      "learning_rate": 0.00014460000000000002,
-      "loss": 0.1188,
+      "learning_rate": 0.0001445,
+      "loss": 0.1067,
       "step": 4059
     },
     {
       "epoch": 73.82,
-      "learning_rate": 0.0001445,
-      "loss": 0.128,
+      "learning_rate": 0.0001444,
+      "loss": 0.1186,
       "step": 4060
     },
     {
       "epoch": 73.84,
-      "learning_rate": 0.0001444,
-      "loss": 0.1006,
+      "learning_rate": 0.0001443,
+      "loss": 0.0928,
       "step": 4061
     },
     {
       "epoch": 73.85,
-      "learning_rate": 0.0001443,
-      "loss": 0.1066,
+      "learning_rate": 0.0001442,
+      "loss": 0.0894,
       "step": 4062
     },
     {
       "epoch": 73.87,
-      "learning_rate": 0.0001442,
-      "loss": 0.1274,
+      "learning_rate": 0.0001441,
+      "loss": 0.1182,
       "step": 4063
     },
     {
       "epoch": 73.89,
-      "learning_rate": 0.0001441,
-      "loss": 0.1229,
+      "learning_rate": 0.000144,
+      "loss": 0.1063,
       "step": 4064
     },
     {
       "epoch": 73.91,
-      "learning_rate": 0.000144,
-      "loss": 0.0599,
+      "learning_rate": 0.0001439,
+      "loss": 0.0524,
       "step": 4065
     },
     {
       "epoch": 73.93,
-      "learning_rate": 0.0001439,
-      "loss": 0.1263,
+      "learning_rate": 0.00014380000000000003,
+      "loss": 0.0993,
       "step": 4066
     },
     {
       "epoch": 73.95,
-      "learning_rate": 0.00014380000000000003,
-      "loss": 0.1307,
+      "learning_rate": 0.0001437,
+      "loss": 0.1002,
       "step": 4067
     },
     {
       "epoch": 73.96,
-      "learning_rate": 0.0001437,
-      "loss": 0.1253,
+      "learning_rate": 0.00014360000000000002,
+      "loss": 0.1132,
       "step": 4068
     },
     {
       "epoch": 73.98,
-      "learning_rate": 0.00014360000000000002,
-      "loss": 0.1324,
+      "learning_rate": 0.0001435,
+      "loss": 0.1126,
       "step": 4069
     },
     {
       "epoch": 74.0,
-      "learning_rate": 0.0001435,
-      "loss": 0.0779,
+      "learning_rate": 0.00014340000000000002,
+      "loss": 0.0616,
       "step": 4070
     },
     {
       "epoch": 74.02,
-      "learning_rate": 0.00014340000000000002,
-      "loss": 0.1284,
+      "learning_rate": 0.00014330000000000001,
+      "loss": 0.114,
       "step": 4071
     },
     {
       "epoch": 74.04,
-      "learning_rate": 0.00014330000000000001,
-      "loss": 0.1293,
+      "learning_rate": 0.00014319999999999998,
+      "loss": 0.1258,
       "step": 4072
     },
     {
       "epoch": 74.05,
-      "learning_rate": 0.00014319999999999998,
-      "loss": 0.1318,
+      "learning_rate": 0.0001431,
+      "loss": 0.1162,
       "step": 4073
     },
     {
       "epoch": 74.07,
-      "learning_rate": 0.0001431,
-      "loss": 0.0911,
+      "learning_rate": 0.00014299999999999998,
+      "loss": 0.0949,
       "step": 4074
     },
     {
       "epoch": 74.09,
-      "learning_rate": 0.00014299999999999998,
-      "loss": 0.1148,
+      "learning_rate": 0.0001429,
+      "loss": 0.1163,
       "step": 4075
     },
     {
       "epoch": 74.11,
-      "learning_rate": 0.0001429,
-      "loss": 0.14,
+      "learning_rate": 0.0001428,
+      "loss": 0.1271,
       "step": 4076
     },
     {
       "epoch": 74.13,
-      "learning_rate": 0.0001428,
-      "loss": 0.123,
+      "learning_rate": 0.0001427,
+      "loss": 0.0927,
       "step": 4077
     },
     {
       "epoch": 74.15,
-      "learning_rate": 0.0001427,
-      "loss": 0.1121,
+      "learning_rate": 0.0001426,
+      "loss": 0.1359,
       "step": 4078
     },
     {
       "epoch": 74.16,
-      "learning_rate": 0.0001426,
-      "loss": 0.1137,
+      "learning_rate": 0.0001425,
+      "loss": 0.0997,
       "step": 4079
     },
     {
       "epoch": 74.18,
-      "learning_rate": 0.0001425,
-      "loss": 0.1156,
+      "learning_rate": 0.0001424,
+      "loss": 0.0854,
       "step": 4080
     },
     {
       "epoch": 74.2,
-      "learning_rate": 0.0001424,
-      "loss": 0.102,
+      "learning_rate": 0.00014230000000000002,
+      "loss": 0.0949,
       "step": 4081
     },
     {
       "epoch": 74.22,
-      "learning_rate": 0.00014230000000000002,
-      "loss": 0.1024,
+      "learning_rate": 0.0001422,
+      "loss": 0.0728,
       "step": 4082
     },
     {
       "epoch": 74.24,
-      "learning_rate": 0.0001422,
-      "loss": 0.088,
+      "learning_rate": 0.0001421,
+      "loss": 0.0864,
       "step": 4083
     },
     {
       "epoch": 74.25,
-      "learning_rate": 0.0001421,
-      "loss": 0.1187,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 0.1079,
       "step": 4084
     },
     {
       "epoch": 74.27,
-      "learning_rate": 0.00014199999999999998,
-      "loss": 0.1723,
+      "learning_rate": 0.0001419,
+      "loss": 0.1479,
       "step": 4085
     },
     {
       "epoch": 74.29,
-      "learning_rate": 0.0001419,
-      "loss": 0.0948,
+      "learning_rate": 0.0001418,
+      "loss": 0.0914,
       "step": 4086
     },
     {
       "epoch": 74.31,
-      "learning_rate": 0.0001418,
-      "loss": 0.1404,
+      "learning_rate": 0.0001417,
+      "loss": 0.1242,
       "step": 4087
     },
     {
       "epoch": 74.33,
-      "learning_rate": 0.0001417,
-      "loss": 0.1051,
+      "learning_rate": 0.0001416,
+      "loss": 0.0907,
       "step": 4088
     },
     {
       "epoch": 74.35,
-      "learning_rate": 0.0001416,
-      "loss": 0.1118,
+      "learning_rate": 0.0001415,
+      "loss": 0.0999,
       "step": 4089
     },
     {
       "epoch": 74.36,
-      "learning_rate": 0.0001415,
-      "loss": 0.1098,
+      "learning_rate": 0.0001414,
+      "loss": 0.099,
       "step": 4090
     },
     {
       "epoch": 74.38,
-      "learning_rate": 0.0001414,
-      "loss": 0.1069,
+      "learning_rate": 0.00014130000000000002,
+      "loss": 0.1033,
       "step": 4091
     },
     {
       "epoch": 74.4,
-      "learning_rate": 0.00014130000000000002,
-      "loss": 0.1295,
+      "learning_rate": 0.0001412,
+      "loss": 0.1111,
       "step": 4092
     },
     {
       "epoch": 74.42,
-      "learning_rate": 0.0001412,
-      "loss": 0.1299,
+      "learning_rate": 0.00014110000000000001,
+      "loss": 0.1266,
       "step": 4093
     },
     {
       "epoch": 74.44,
-      "learning_rate": 0.00014110000000000001,
-      "loss": 0.0983,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 0.1154,
       "step": 4094
     },
     {
       "epoch": 74.45,
-      "learning_rate": 0.00014099999999999998,
-      "loss": 0.1057,
+      "learning_rate": 0.0001409,
+      "loss": 0.0854,
       "step": 4095
     },
     {
       "epoch": 74.47,
-      "learning_rate": 0.0001409,
-      "loss": 0.1547,
+      "learning_rate": 0.0001408,
+      "loss": 0.1419,
       "step": 4096
     },
     {
       "epoch": 74.49,
-      "learning_rate": 0.0001408,
-      "loss": 0.1147,
+      "learning_rate": 0.0001407,
+      "loss": 0.1225,
       "step": 4097
     },
     {
       "epoch": 74.51,
-      "learning_rate": 0.0001407,
-      "loss": 0.1225,
+      "learning_rate": 0.0001406,
+      "loss": 0.1042,
       "step": 4098
     },
     {
       "epoch": 74.53,
-      "learning_rate": 0.0001406,
-      "loss": 0.0923,
+      "learning_rate": 0.00014050000000000003,
+      "loss": 0.0879,
       "step": 4099
     },
     {
       "epoch": 74.55,
-      "learning_rate": 0.00014050000000000003,
-      "loss": 0.125,
+      "learning_rate": 0.0001404,
+      "loss": 0.1063,
       "step": 4100
     },
     {
       "epoch": 74.56,
-      "learning_rate": 0.0001404,
-      "loss": 0.0775,
+      "learning_rate": 0.00014030000000000002,
+      "loss": 0.0795,
       "step": 4101
     },
     {
       "epoch": 74.58,
-      "learning_rate": 0.00014030000000000002,
-      "loss": 0.1047,
+      "learning_rate": 0.0001402,
+      "loss": 0.0999,
       "step": 4102
     },
     {
       "epoch": 74.6,
-      "learning_rate": 0.0001402,
-      "loss": 0.0914,
+      "learning_rate": 0.00014010000000000002,
+      "loss": 0.0795,
       "step": 4103
     },
     {
       "epoch": 74.62,
-      "learning_rate": 0.00014010000000000002,
-      "loss": 0.1357,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 0.0993,
       "step": 4104
     },
     {
       "epoch": 74.64,
-      "learning_rate": 0.00014000000000000001,
-      "loss": 0.0991,
+      "learning_rate": 0.0001399,
+      "loss": 0.1289,
       "step": 4105
     },
     {
       "epoch": 74.65,
-      "learning_rate": 0.0001399,
-      "loss": 0.0862,
+      "learning_rate": 0.0001398,
+      "loss": 0.1054,
       "step": 4106
     },
     {
       "epoch": 74.67,
-      "learning_rate": 0.0001398,
-      "loss": 0.09,
+      "learning_rate": 0.00013969999999999998,
+      "loss": 0.0886,
       "step": 4107
     },
     {
       "epoch": 74.69,
-      "learning_rate": 0.00013969999999999998,
-      "loss": 0.1232,
+      "learning_rate": 0.0001396,
+      "loss": 0.1055,
       "step": 4108
     },
     {
       "epoch": 74.71,
-      "learning_rate": 0.0001396,
-      "loss": 0.1081,
+      "learning_rate": 0.0001395,
+      "loss": 0.1275,
       "step": 4109
     },
     {
       "epoch": 74.73,
-      "learning_rate": 0.0001395,
-      "loss": 0.1255,
+      "learning_rate": 0.0001394,
+      "loss": 0.1137,
       "step": 4110
     },
     {
       "epoch": 74.75,
-      "learning_rate": 0.0001394,
-      "loss": 0.1387,
+      "learning_rate": 0.0001393,
+      "loss": 0.1101,
       "step": 4111
     },
     {
       "epoch": 74.76,
-      "learning_rate": 0.0001393,
-      "loss": 0.1231,
+      "learning_rate": 0.0001392,
+      "loss": 0.1339,
       "step": 4112
     },
     {
       "epoch": 74.78,
-      "learning_rate": 0.0001392,
-      "loss": 0.0883,
+      "learning_rate": 0.0001391,
+      "loss": 0.0788,
       "step": 4113
     },
     {
       "epoch": 74.8,
-      "learning_rate": 0.0001391,
-      "loss": 0.1176,
+      "learning_rate": 0.00013900000000000002,
+      "loss": 0.0841,
       "step": 4114
     },
     {
       "epoch": 74.82,
-      "learning_rate": 0.00013900000000000002,
-      "loss": 0.1481,
+      "learning_rate": 0.0001389,
+      "loss": 0.1362,
       "step": 4115
     },
     {
       "epoch": 74.84,
-      "learning_rate": 0.0001389,
-      "loss": 0.136,
+      "learning_rate": 0.0001388,
+      "loss": 0.128,
       "step": 4116
     },
     {
       "epoch": 74.85,
-      "learning_rate": 0.0001388,
-      "loss": 0.1487,
+      "learning_rate": 0.00013869999999999998,
+      "loss": 0.123,
       "step": 4117
     },
     {
       "epoch": 74.87,
-      "learning_rate": 0.00013869999999999998,
-      "loss": 0.124,
+      "learning_rate": 0.0001386,
+      "loss": 0.1445,
       "step": 4118
     },
     {
       "epoch": 74.89,
-      "learning_rate": 0.0001386,
-      "loss": 0.095,
+      "learning_rate": 0.0001385,
+      "loss": 0.0947,
       "step": 4119
     },
     {
       "epoch": 74.91,
-      "learning_rate": 0.0001385,
-      "loss": 0.098,
+      "learning_rate": 0.0001384,
+      "loss": 0.0743,
       "step": 4120
     },
     {
       "epoch": 74.93,
-      "learning_rate": 0.0001384,
-      "loss": 0.1357,
+      "learning_rate": 0.0001383,
+      "loss": 0.1078,
       "step": 4121
     },
     {
       "epoch": 74.95,
-      "learning_rate": 0.0001383,
-      "loss": 0.0947,
+      "learning_rate": 0.0001382,
+      "loss": 0.1172,
       "step": 4122
     },
     {
       "epoch": 74.96,
-      "learning_rate": 0.0001382,
-      "loss": 0.1308,
+      "learning_rate": 0.0001381,
+      "loss": 0.1054,
       "step": 4123
     },
     {
       "epoch": 74.98,
-      "learning_rate": 0.0001381,
-      "loss": 0.1527,
+      "learning_rate": 0.00013800000000000002,
+      "loss": 0.1454,
       "step": 4124
     },
     {
       "epoch": 75.0,
-      "learning_rate": 0.00013800000000000002,
-      "loss": 0.0912,
+      "learning_rate": 0.0001379,
+      "loss": 0.073,
       "step": 4125
     },
     {
       "epoch": 75.02,
-      "learning_rate": 0.0001379,
-      "loss": 0.1264,
+      "learning_rate": 0.00013780000000000002,
+      "loss": 0.1242,
       "step": 4126
     },
     {
       "epoch": 75.04,
-      "learning_rate": 0.00013780000000000002,
-      "loss": 0.1009,
+      "learning_rate": 0.00013769999999999999,
+      "loss": 0.0674,
       "step": 4127
     },
     {
       "epoch": 75.05,
-      "learning_rate": 0.00013769999999999999,
-      "loss": 0.1294,
+      "learning_rate": 0.0001376,
+      "loss": 0.1229,
       "step": 4128
     },
     {
       "epoch": 75.07,
-      "learning_rate": 0.0001376,
-      "loss": 0.1,
+      "learning_rate": 0.0001375,
+      "loss": 0.0987,
       "step": 4129
     },
     {
       "epoch": 75.09,
-      "learning_rate": 0.0001375,
-      "loss": 0.1548,
+      "learning_rate": 0.0001374,
+      "loss": 0.1313,
       "step": 4130
     },
     {
       "epoch": 75.11,
-      "learning_rate": 0.0001374,
-      "loss": 0.0783,
+      "learning_rate": 0.0001373,
+      "loss": 0.0782,
       "step": 4131
     },
     {
       "epoch": 75.13,
-      "learning_rate": 0.0001373,
-      "loss": 0.1496,
+      "learning_rate": 0.0001372,
+      "loss": 0.1291,
       "step": 4132
     },
     {
       "epoch": 75.15,
-      "learning_rate": 0.0001372,
-      "loss": 0.1118,
+      "learning_rate": 0.0001371,
+      "loss": 0.0994,
       "step": 4133
     },
     {
       "epoch": 75.16,
-      "learning_rate": 0.0001371,
-      "loss": 0.0859,
+      "learning_rate": 0.00013700000000000002,
+      "loss": 0.0695,
       "step": 4134
     },
     {
       "epoch": 75.18,
-      "learning_rate": 0.00013700000000000002,
-      "loss": 0.102,
+      "learning_rate": 0.0001369,
+      "loss": 0.0913,
       "step": 4135
     },
     {
       "epoch": 75.2,
-      "learning_rate": 0.0001369,
-      "loss": 0.1276,
+      "learning_rate": 0.00013680000000000002,
+      "loss": 0.0992,
       "step": 4136
     },
     {
       "epoch": 75.22,
-      "learning_rate": 0.00013680000000000002,
-      "loss": 0.0838,
+      "learning_rate": 0.0001367,
+      "loss": 0.0876,
       "step": 4137
     },
     {
       "epoch": 75.24,
-      "learning_rate": 0.0001367,
-      "loss": 0.1191,
+      "learning_rate": 0.0001366,
+      "loss": 0.1188,
       "step": 4138
     },
     {
       "epoch": 75.25,
-      "learning_rate": 0.0001366,
-      "loss": 0.1071,
+      "learning_rate": 0.0001365,
+      "loss": 0.1182,
       "step": 4139
     },
     {
       "epoch": 75.27,
-      "learning_rate": 0.0001365,
-      "loss": 0.115,
+      "learning_rate": 0.0001364,
+      "loss": 0.1049,
       "step": 4140
     },
     {
       "epoch": 75.29,
-      "learning_rate": 0.0001364,
-      "loss": 0.0943,
+      "learning_rate": 0.0001363,
+      "loss": 0.1131,
       "step": 4141
     },
     {
       "epoch": 75.31,
-      "learning_rate": 0.0001363,
-      "loss": 0.0876,
+      "learning_rate": 0.0001362,
+      "loss": 0.098,
       "step": 4142
     },
     {
       "epoch": 75.33,
-      "learning_rate": 0.0001362,
-      "loss": 0.1174,
+      "learning_rate": 0.0001361,
+      "loss": 0.1127,
       "step": 4143
     },
     {
       "epoch": 75.35,
-      "learning_rate": 0.0001361,
-      "loss": 0.1151,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 0.1199,
       "step": 4144
     },
     {
       "epoch": 75.36,
-      "learning_rate": 0.00013600000000000003,
-      "loss": 0.139,
+      "learning_rate": 0.0001359,
+      "loss": 0.1328,
       "step": 4145
     },
     {
       "epoch": 75.38,
-      "learning_rate": 0.0001359,
-      "loss": 0.1264,
+      "learning_rate": 0.0001358,
+      "loss": 0.1328,
       "step": 4146
     },
     {
       "epoch": 75.4,
-      "learning_rate": 0.0001358,
-      "loss": 0.1012,
+      "learning_rate": 0.0001357,
+      "loss": 0.0927,
       "step": 4147
     },
     {
       "epoch": 75.42,
-      "learning_rate": 0.0001357,
-      "loss": 0.1627,
+      "learning_rate": 0.0001356,
+      "loss": 0.1336,
       "step": 4148
     },
     {
       "epoch": 75.44,
-      "learning_rate": 0.0001356,
-      "loss": 0.102,
+      "learning_rate": 0.00013550000000000001,
+      "loss": 0.1077,
       "step": 4149
     },
     {
       "epoch": 75.45,
-      "learning_rate": 0.00013550000000000001,
-      "loss": 0.1107,
+      "learning_rate": 0.00013539999999999998,
+      "loss": 0.1145,
       "step": 4150
     },
     {
       "epoch": 75.47,
-      "learning_rate": 0.00013539999999999998,
-      "loss": 0.1451,
+      "learning_rate": 0.0001353,
+      "loss": 0.1306,
       "step": 4151
     },
     {
       "epoch": 75.49,
-      "learning_rate": 0.0001353,
-      "loss": 0.1377,
+      "learning_rate": 0.00013519999999999998,
+      "loss": 0.1262,
       "step": 4152
     },
     {
       "epoch": 75.51,
-      "learning_rate": 0.00013519999999999998,
-      "loss": 0.1121,
+      "learning_rate": 0.0001351,
+      "loss": 0.1182,
       "step": 4153
     },
     {
       "epoch": 75.53,
-      "learning_rate": 0.0001351,
-      "loss": 0.1016,
+      "learning_rate": 0.000135,
+      "loss": 0.1084,
       "step": 4154
     },
     {
       "epoch": 75.55,
-      "learning_rate": 0.000135,
-      "loss": 0.0748,
+      "learning_rate": 0.0001349,
+      "loss": 0.0618,
       "step": 4155
     },
     {
       "epoch": 75.56,
-      "learning_rate": 0.0001349,
-      "loss": 0.1047,
+      "learning_rate": 0.0001348,
+      "loss": 0.1009,
       "step": 4156
     },
     {
       "epoch": 75.58,
-      "learning_rate": 0.0001348,
-      "loss": 0.1458,
+      "learning_rate": 0.0001347,
+      "loss": 0.1337,
       "step": 4157
     },
     {
       "epoch": 75.6,
-      "learning_rate": 0.0001347,
-      "loss": 0.0942,
+      "learning_rate": 0.0001346,
+      "loss": 0.0915,
       "step": 4158
     },
     {
       "epoch": 75.62,
-      "learning_rate": 0.0001346,
-      "loss": 0.13,
+      "learning_rate": 0.00013450000000000002,
+      "loss": 0.1247,
       "step": 4159
     },
     {
       "epoch": 75.64,
-      "learning_rate": 0.00013450000000000002,
-      "loss": 0.1063,
+      "learning_rate": 0.0001344,
+      "loss": 0.1294,
       "step": 4160
     },
     {
       "epoch": 75.65,
-      "learning_rate": 0.0001344,
-      "loss": 0.1127,
+      "learning_rate": 0.0001343,
+      "loss": 0.089,
       "step": 4161
     },
     {
       "epoch": 75.67,
-      "learning_rate": 0.0001343,
-      "loss": 0.1296,
+      "learning_rate": 0.0001342,
+      "loss": 0.1023,
       "step": 4162
     },
     {
       "epoch": 75.69,
-      "learning_rate": 0.0001342,
-      "loss": 0.117,
+      "learning_rate": 0.0001341,
+      "loss": 0.1065,
       "step": 4163
     },
     {
       "epoch": 75.71,
-      "learning_rate": 0.0001341,
-      "loss": 0.1272,
+      "learning_rate": 0.000134,
+      "loss": 0.1171,
       "step": 4164
     },
     {
       "epoch": 75.73,
-      "learning_rate": 0.000134,
-      "loss": 0.1185,
+      "learning_rate": 0.0001339,
+      "loss": 0.0971,
       "step": 4165
     },
     {
       "epoch": 75.75,
-      "learning_rate": 0.0001339,
-      "loss": 0.1035,
+      "learning_rate": 0.0001338,
+      "loss": 0.0862,
       "step": 4166
     },
     {
       "epoch": 75.76,
-      "learning_rate": 0.0001338,
-      "loss": 0.1021,
+      "learning_rate": 0.00013370000000000002,
+      "loss": 0.082,
       "step": 4167
     },
     {
       "epoch": 75.78,
-      "learning_rate": 0.00013370000000000002,
-      "loss": 0.1002,
+      "learning_rate": 0.0001336,
+      "loss": 0.073,
       "step": 4168
     },
     {
       "epoch": 75.8,
-      "learning_rate": 0.0001336,
-      "loss": 0.0764,
+      "learning_rate": 0.00013350000000000002,
+      "loss": 0.0753,
       "step": 4169
     },
     {
       "epoch": 75.82,
-      "learning_rate": 0.00013350000000000002,
-      "loss": 0.1138,
+      "learning_rate": 0.0001334,
+      "loss": 0.0968,
       "step": 4170
     },
     {
       "epoch": 75.84,
-      "learning_rate": 0.0001334,
-      "loss": 0.1189,
+      "learning_rate": 0.00013330000000000001,
+      "loss": 0.101,
       "step": 4171
     },
     {
       "epoch": 75.85,
-      "learning_rate": 0.00013330000000000001,
-      "loss": 0.0829,
+      "learning_rate": 0.0001332,
+      "loss": 0.0911,
       "step": 4172
     },
     {
       "epoch": 75.87,
-      "learning_rate": 0.0001332,
-      "loss": 0.1169,
+      "learning_rate": 0.0001331,
+      "loss": 0.1043,
       "step": 4173
     },
     {
       "epoch": 75.89,
-      "learning_rate": 0.0001331,
-      "loss": 0.0803,
+      "learning_rate": 0.000133,
+      "loss": 0.1043,
       "step": 4174
     },
     {
       "epoch": 75.91,
-      "learning_rate": 0.000133,
-      "loss": 0.0752,
+      "learning_rate": 0.0001329,
+      "loss": 0.0602,
       "step": 4175
     },
     {
       "epoch": 75.93,
-      "learning_rate": 0.0001329,
-      "loss": 0.1283,
+      "learning_rate": 0.0001328,
+      "loss": 0.1201,
       "step": 4176
     },
     {
       "epoch": 75.95,
-      "learning_rate": 0.0001328,
-      "loss": 0.1185,
+      "learning_rate": 0.00013270000000000003,
+      "loss": 0.1011,
       "step": 4177
     },
     {
       "epoch": 75.96,
-      "learning_rate": 0.00013270000000000003,
-      "loss": 0.0831,
+      "learning_rate": 0.0001326,
+      "loss": 0.0749,
       "step": 4178
     },
     {
       "epoch": 75.98,
-      "learning_rate": 0.0001326,
-      "loss": 0.1135,
+      "learning_rate": 0.00013250000000000002,
+      "loss": 0.1133,
       "step": 4179
     },
     {
       "epoch": 76.0,
-      "learning_rate": 0.00013250000000000002,
-      "loss": 0.0819,
+      "learning_rate": 0.0001324,
+      "loss": 0.0436,
       "step": 4180
     },
     {
       "epoch": 76.02,
-      "learning_rate": 0.0001324,
-      "loss": 0.1121,
+      "learning_rate": 0.0001323,
+      "loss": 0.0978,
       "step": 4181
     },
     {
       "epoch": 76.04,
-      "learning_rate": 0.0001323,
-      "loss": 0.1648,
+      "learning_rate": 0.00013220000000000001,
+      "loss": 0.1822,
       "step": 4182
     },
     {
       "epoch": 76.05,
-      "learning_rate": 0.00013220000000000001,
-      "loss": 0.1028,
+      "learning_rate": 0.00013209999999999999,
+      "loss": 0.0868,
       "step": 4183
     },
     {
       "epoch": 76.07,
-      "learning_rate": 0.00013209999999999999,
-      "loss": 0.1287,
+      "learning_rate": 0.000132,
+      "loss": 0.1033,
       "step": 4184
     },
     {
       "epoch": 76.09,
-      "learning_rate": 0.000132,
-      "loss": 0.1093,
+      "learning_rate": 0.00013189999999999998,
+      "loss": 0.0734,
       "step": 4185
     },
     {
       "epoch": 76.11,
-      "learning_rate": 0.00013189999999999998,
-      "loss": 0.1426,
+      "learning_rate": 0.0001318,
+      "loss": 0.1312,
       "step": 4186
     },
     {
       "epoch": 76.13,
-      "learning_rate": 0.0001318,
-      "loss": 0.1289,
+      "learning_rate": 0.0001317,
+      "loss": 0.1236,
       "step": 4187
     },
     {
       "epoch": 76.15,
-      "learning_rate": 0.0001317,
-      "loss": 0.1155,
+      "learning_rate": 0.0001316,
+      "loss": 0.1072,
       "step": 4188
     },
     {
       "epoch": 76.16,
-      "learning_rate": 0.0001316,
-      "loss": 0.105,
+      "learning_rate": 0.0001315,
+      "loss": 0.0795,
       "step": 4189
     },
     {
       "epoch": 76.18,
-      "learning_rate": 0.0001315,
-      "loss": 0.0935,
+      "learning_rate": 0.0001314,
+      "loss": 0.0871,
       "step": 4190
     },
     {
       "epoch": 76.2,
-      "learning_rate": 0.0001314,
-      "loss": 0.0895,
+      "learning_rate": 0.0001313,
+      "loss": 0.0887,
       "step": 4191
     },
     {
       "epoch": 76.22,
-      "learning_rate": 0.0001313,
-      "loss": 0.1025,
+      "learning_rate": 0.00013120000000000002,
+      "loss": 0.089,
       "step": 4192
     },
     {
       "epoch": 76.24,
-      "learning_rate": 0.00013120000000000002,
-      "loss": 0.1079,
+      "learning_rate": 0.0001311,
+      "loss": 0.0945,
       "step": 4193
     },
     {
       "epoch": 76.25,
-      "learning_rate": 0.0001311,
-      "loss": 0.0959,
+      "learning_rate": 0.000131,
+      "loss": 0.0817,
       "step": 4194
     },
     {
       "epoch": 76.27,
-      "learning_rate": 0.000131,
-      "loss": 0.0952,
+      "learning_rate": 0.00013089999999999998,
+      "loss": 0.1068,
       "step": 4195
     },
     {
       "epoch": 76.29,
-      "learning_rate": 0.00013089999999999998,
-      "loss": 0.0862,
+      "learning_rate": 0.0001308,
+      "loss": 0.0842,
       "step": 4196
     },
     {
       "epoch": 76.31,
-      "learning_rate": 0.0001308,
-      "loss": 0.1202,
+      "learning_rate": 0.0001307,
+      "loss": 0.1164,
       "step": 4197
     },
     {
       "epoch": 76.33,
-      "learning_rate": 0.0001307,
-      "loss": 0.0833,
+      "learning_rate": 0.0001306,
+      "loss": 0.0875,
       "step": 4198
     },
     {
       "epoch": 76.35,
-      "learning_rate": 0.0001306,
-      "loss": 0.0751,
+      "learning_rate": 0.0001305,
+      "loss": 0.0932,
       "step": 4199
     },
     {
       "epoch": 76.36,
-      "learning_rate": 0.0001305,
-      "loss": 0.1229,
+      "learning_rate": 0.0001304,
+      "loss": 0.1157,
       "step": 4200
     },
     {
       "epoch": 76.38,
-      "learning_rate": 0.0001304,
-      "loss": 0.1246,
+      "learning_rate": 0.0001303,
+      "loss": 0.0997,
       "step": 4201
     },
     {
       "epoch": 76.4,
-      "learning_rate": 0.0001303,
-      "loss": 0.0916,
+      "learning_rate": 0.00013020000000000002,
+      "loss": 0.0998,
       "step": 4202
     },
     {
       "epoch": 76.42,
-      "learning_rate": 0.00013020000000000002,
-      "loss": 0.1207,
+      "learning_rate": 0.0001301,
+      "loss": 0.1086,
       "step": 4203
     },
     {
       "epoch": 76.44,
-      "learning_rate": 0.0001301,
-      "loss": 0.1028,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 0.1361,
       "step": 4204
     },
     {
       "epoch": 76.45,
-      "learning_rate": 0.00013000000000000002,
-      "loss": 0.072,
+      "learning_rate": 0.00012989999999999999,
+      "loss": 0.0681,
       "step": 4205
     },
     {
       "epoch": 76.47,
-      "learning_rate": 0.00012989999999999999,
-      "loss": 0.146,
+      "learning_rate": 0.0001298,
+      "loss": 0.125,
       "step": 4206
     },
     {
       "epoch": 76.49,
-      "learning_rate": 0.0001298,
-      "loss": 0.0925,
+      "learning_rate": 0.0001297,
+      "loss": 0.0905,
       "step": 4207
     },
     {
       "epoch": 76.51,
-      "learning_rate": 0.0001297,
-      "loss": 0.1388,
+      "learning_rate": 0.0001296,
+      "loss": 0.1062,
       "step": 4208
     },
     {
       "epoch": 76.53,
-      "learning_rate": 0.0001296,
-      "loss": 0.0874,
+      "learning_rate": 0.0001295,
+      "loss": 0.0769,
       "step": 4209
     },
     {
       "epoch": 76.55,
-      "learning_rate": 0.0001295,
-      "loss": 0.0922,
+      "learning_rate": 0.0001294,
+      "loss": 0.0873,
       "step": 4210
     },
     {
       "epoch": 76.56,
-      "learning_rate": 0.0001294,
-      "loss": 0.1078,
+      "learning_rate": 0.0001293,
+      "loss": 0.088,
       "step": 4211
     },
     {
       "epoch": 76.58,
-      "learning_rate": 0.0001293,
-      "loss": 0.0842,
+      "learning_rate": 0.00012920000000000002,
+      "loss": 0.0651,
       "step": 4212
     },
     {
       "epoch": 76.6,
-      "learning_rate": 0.00012920000000000002,
-      "loss": 0.0758,
+      "learning_rate": 0.0001291,
+      "loss": 0.0642,
       "step": 4213
     },
     {
       "epoch": 76.62,
-      "learning_rate": 0.0001291,
-      "loss": 0.1183,
+      "learning_rate": 0.00012900000000000002,
+      "loss": 0.0716,
       "step": 4214
     },
     {
       "epoch": 76.64,
-      "learning_rate": 0.00012900000000000002,
-      "loss": 0.0897,
+      "learning_rate": 0.0001289,
+      "loss": 0.0755,
       "step": 4215
     },
     {
       "epoch": 76.65,
-      "learning_rate": 0.0001289,
-      "loss": 0.0869,
+      "learning_rate": 0.00012880000000000001,
+      "loss": 0.0729,
       "step": 4216
     },
     {
       "epoch": 76.67,
-      "learning_rate": 0.00012880000000000001,
-      "loss": 0.106,
+      "learning_rate": 0.0001287,
+      "loss": 0.1109,
       "step": 4217
     },
     {
       "epoch": 76.69,
-      "learning_rate": 0.0001287,
-      "loss": 0.1072,
+      "learning_rate": 0.00012859999999999998,
+      "loss": 0.0955,
       "step": 4218
     },
     {
       "epoch": 76.71,
-      "learning_rate": 0.00012859999999999998,
-      "loss": 0.1123,
+      "learning_rate": 0.0001285,
+      "loss": 0.1167,
       "step": 4219
     },
     {
       "epoch": 76.73,
-      "learning_rate": 0.0001285,
+      "learning_rate": 0.00012839999999999998,
       "loss": 0.1194,
       "step": 4220
     },
     {
       "epoch": 76.75,
-      "learning_rate": 0.00012839999999999998,
-      "loss": 0.0986,
+      "learning_rate": 0.0001283,
+      "loss": 0.0953,
       "step": 4221
     },
     {
       "epoch": 76.76,
-      "learning_rate": 0.0001283,
-      "loss": 0.1085,
+      "learning_rate": 0.0001282,
+      "loss": 0.1049,
       "step": 4222
     },
     {
       "epoch": 76.78,
-      "learning_rate": 0.0001282,
-      "loss": 0.0928,
+      "learning_rate": 0.0001281,
+      "loss": 0.0993,
       "step": 4223
     },
     {
       "epoch": 76.8,
-      "learning_rate": 0.0001281,
-      "loss": 0.098,
+      "learning_rate": 0.000128,
+      "loss": 0.0984,
       "step": 4224
     },
     {
       "epoch": 76.82,
-      "learning_rate": 0.000128,
-      "loss": 0.102,
+      "learning_rate": 0.00012790000000000002,
+      "loss": 0.1074,
       "step": 4225
     },
     {
       "epoch": 76.84,
-      "learning_rate": 0.00012790000000000002,
-      "loss": 0.1385,
+      "learning_rate": 0.0001278,
+      "loss": 0.1253,
       "step": 4226
     },
     {
       "epoch": 76.85,
-      "learning_rate": 0.0001278,
-      "loss": 0.0864,
+      "learning_rate": 0.00012770000000000001,
+      "loss": 0.0716,
       "step": 4227
     },
     {
       "epoch": 76.87,
-      "learning_rate": 0.00012770000000000001,
-      "loss": 0.1306,
+      "learning_rate": 0.00012759999999999998,
+      "loss": 0.1198,
       "step": 4228
     },
     {
       "epoch": 76.89,
-      "learning_rate": 0.00012759999999999998,
-      "loss": 0.0826,
+      "learning_rate": 0.0001275,
+      "loss": 0.0769,
       "step": 4229
     },
     {
       "epoch": 76.91,
-      "learning_rate": 0.0001275,
-      "loss": 0.0945,
+      "learning_rate": 0.0001274,
+      "loss": 0.078,
       "step": 4230
     },
     {
       "epoch": 76.93,
-      "learning_rate": 0.0001274,
-      "loss": 0.0935,
+      "learning_rate": 0.0001273,
+      "loss": 0.0925,
       "step": 4231
     },
     {
       "epoch": 76.95,
-      "learning_rate": 0.0001273,
-      "loss": 0.1107,
+      "learning_rate": 0.0001272,
+      "loss": 0.0977,
       "step": 4232
     },
     {
       "epoch": 76.96,
-      "learning_rate": 0.0001272,
-      "loss": 0.1277,
+      "learning_rate": 0.0001271,
+      "loss": 0.1046,
       "step": 4233
     },
     {
       "epoch": 76.98,
-      "learning_rate": 0.0001271,
-      "loss": 0.094,
+      "learning_rate": 0.000127,
+      "loss": 0.0832,
       "step": 4234
     },
     {
       "epoch": 77.0,
-      "learning_rate": 0.000127,
-      "loss": 0.0338,
+      "learning_rate": 0.00012690000000000002,
+      "loss": 0.03,
       "step": 4235
     },
     {
       "epoch": 77.02,
-      "learning_rate": 0.00012690000000000002,
-      "loss": 0.1199,
+      "learning_rate": 0.0001268,
+      "loss": 0.1105,
       "step": 4236
     },
     {
       "epoch": 77.04,
-      "learning_rate": 0.0001268,
-      "loss": 0.0961,
+      "learning_rate": 0.00012670000000000002,
+      "loss": 0.0963,
       "step": 4237
     },
     {
       "epoch": 77.05,
-      "learning_rate": 0.00012670000000000002,
-      "loss": 0.1441,
+      "learning_rate": 0.0001266,
+      "loss": 0.1191,
       "step": 4238
     },
     {
       "epoch": 77.07,
-      "learning_rate": 0.0001266,
-      "loss": 0.1214,
+      "learning_rate": 0.0001265,
+      "loss": 0.1166,
       "step": 4239
     },
     {
       "epoch": 77.09,
-      "learning_rate": 0.0001265,
-      "loss": 0.0831,
+      "learning_rate": 0.0001264,
+      "loss": 0.0834,
       "step": 4240
     },
     {
       "epoch": 77.11,
-      "learning_rate": 0.0001264,
-      "loss": 0.1032,
+      "learning_rate": 0.0001263,
+      "loss": 0.1317,
       "step": 4241
     },
     {
       "epoch": 77.13,
-      "learning_rate": 0.0001263,
-      "loss": 0.1133,
+      "learning_rate": 0.0001262,
+      "loss": 0.114,
       "step": 4242
     },
     {
       "epoch": 77.15,
-      "learning_rate": 0.0001262,
-      "loss": 0.1086,
+      "learning_rate": 0.0001261,
+      "loss": 0.1044,
       "step": 4243
     },
     {
       "epoch": 77.16,
-      "learning_rate": 0.0001261,
-      "loss": 0.104,
+      "learning_rate": 0.000126,
+      "loss": 0.097,
       "step": 4244
     },
     {
       "epoch": 77.18,
-      "learning_rate": 0.000126,
-      "loss": 0.1033,
+      "learning_rate": 0.00012590000000000002,
+      "loss": 0.0889,
       "step": 4245
     },
     {
       "epoch": 77.2,
-      "learning_rate": 0.00012590000000000002,
-      "loss": 0.1128,
+      "learning_rate": 0.0001258,
+      "loss": 0.0844,
       "step": 4246
     },
     {
       "epoch": 77.22,
-      "learning_rate": 0.0001258,
-      "loss": 0.117,
+      "learning_rate": 0.00012570000000000002,
+      "loss": 0.0909,
       "step": 4247
     },
     {
       "epoch": 77.24,
-      "learning_rate": 0.00012570000000000002,
-      "loss": 0.0989,
+      "learning_rate": 0.0001256,
+      "loss": 0.1015,
       "step": 4248
     },
     {
       "epoch": 77.25,
-      "learning_rate": 0.0001256,
-      "loss": 0.1489,
+      "learning_rate": 0.00012550000000000001,
+      "loss": 0.1448,
       "step": 4249
     },
     {
       "epoch": 77.27,
-      "learning_rate": 0.00012550000000000001,
-      "loss": 0.1331,
+      "learning_rate": 0.0001254,
+      "loss": 0.1349,
       "step": 4250
     },
     {
       "epoch": 77.29,
-      "learning_rate": 0.0001254,
-      "loss": 0.098,
+      "learning_rate": 0.0001253,
+      "loss": 0.1097,
       "step": 4251
     },
     {
       "epoch": 77.31,
-      "learning_rate": 0.0001253,
-      "loss": 0.1035,
+      "learning_rate": 0.0001252,
+      "loss": 0.1094,
       "step": 4252
     },
     {
       "epoch": 77.33,
-      "learning_rate": 0.0001252,
-      "loss": 0.0982,
+      "learning_rate": 0.0001251,
+      "loss": 0.1036,
       "step": 4253
     },
     {
       "epoch": 77.35,
-      "learning_rate": 0.0001251,
-      "loss": 0.1078,
+      "learning_rate": 0.000125,
+      "loss": 0.0908,
       "step": 4254
     },
     {
       "epoch": 77.36,
-      "learning_rate": 0.000125,
-      "loss": 0.1083,
+      "learning_rate": 0.0001249,
+      "loss": 0.0896,
       "step": 4255
     },
     {
       "epoch": 77.38,
-      "learning_rate": 0.0001249,
-      "loss": 0.083,
+      "learning_rate": 0.0001248,
+      "loss": 0.0838,
       "step": 4256
     },
     {
       "epoch": 77.4,
-      "learning_rate": 0.0001248,
-      "loss": 0.1338,
+      "learning_rate": 0.0001247,
+      "loss": 0.1192,
       "step": 4257
     },
     {
       "epoch": 77.42,
-      "learning_rate": 0.0001247,
-      "loss": 0.1287,
+      "learning_rate": 0.0001246,
+      "loss": 0.0902,
       "step": 4258
     },
     {
       "epoch": 77.44,
-      "learning_rate": 0.0001246,
-      "loss": 0.1058,
+      "learning_rate": 0.0001245,
+      "loss": 0.1215,
       "step": 4259
     },
     {
       "epoch": 77.45,
-      "learning_rate": 0.0001245,
-      "loss": 0.081,
+      "learning_rate": 0.0001244,
+      "loss": 0.0951,
       "step": 4260
     },
     {
       "epoch": 77.47,
-      "learning_rate": 0.0001244,
-      "loss": 0.0772,
+      "learning_rate": 0.00012429999999999999,
+      "loss": 0.0802,
       "step": 4261
     },
     {
       "epoch": 77.49,
-      "learning_rate": 0.00012429999999999999,
-      "loss": 0.0712,
+      "learning_rate": 0.0001242,
+      "loss": 0.0592,
       "step": 4262
     },
     {
       "epoch": 77.51,
-      "learning_rate": 0.0001242,
-      "loss": 0.0819,
+      "learning_rate": 0.0001241,
+      "loss": 0.0656,
       "step": 4263
     },
     {
       "epoch": 77.53,
-      "learning_rate": 0.0001241,
-      "loss": 0.091,
+      "learning_rate": 0.000124,
+      "loss": 0.0817,
       "step": 4264
     },
     {
       "epoch": 77.55,
-      "learning_rate": 0.000124,
-      "loss": 0.1008,
+      "learning_rate": 0.0001239,
+      "loss": 0.0831,
       "step": 4265
     },
     {
       "epoch": 77.56,
-      "learning_rate": 0.0001239,
-      "loss": 0.0815,
+      "learning_rate": 0.0001238,
+      "loss": 0.077,
       "step": 4266
     },
     {
       "epoch": 77.58,
-      "learning_rate": 0.0001238,
-      "loss": 0.1053,
+      "learning_rate": 0.0001237,
+      "loss": 0.0906,
       "step": 4267
     },
     {
       "epoch": 77.6,
-      "learning_rate": 0.0001237,
-      "loss": 0.11,
+      "learning_rate": 0.0001236,
+      "loss": 0.096,
       "step": 4268
     },
     {
       "epoch": 77.62,
-      "learning_rate": 0.0001236,
-      "loss": 0.1179,
+      "learning_rate": 0.0001235,
+      "loss": 0.1017,
       "step": 4269
     },
     {
       "epoch": 77.64,
-      "learning_rate": 0.0001235,
-      "loss": 0.0938,
+      "learning_rate": 0.0001234,
+      "loss": 0.0809,
       "step": 4270
     },
     {
       "epoch": 77.65,
-      "learning_rate": 0.0001234,
-      "loss": 0.0879,
+      "learning_rate": 0.00012330000000000002,
+      "loss": 0.0598,
       "step": 4271
     },
     {
       "epoch": 77.67,
-      "learning_rate": 0.00012330000000000002,
-      "loss": 0.077,
+      "learning_rate": 0.0001232,
+      "loss": 0.0781,
       "step": 4272
     },
     {
       "epoch": 77.69,
-      "learning_rate": 0.0001232,
-      "loss": 0.0892,
+      "learning_rate": 0.0001231,
+      "loss": 0.0764,
       "step": 4273
     },
     {
       "epoch": 77.71,
-      "learning_rate": 0.0001231,
-      "loss": 0.0895,
+      "learning_rate": 0.000123,
+      "loss": 0.0762,
       "step": 4274
     },
     {
       "epoch": 77.73,
-      "learning_rate": 0.000123,
-      "loss": 0.1236,
+      "learning_rate": 0.0001229,
+      "loss": 0.0952,
       "step": 4275
     },
     {
       "epoch": 77.75,
-      "learning_rate": 0.0001229,
-      "loss": 0.0784,
+      "learning_rate": 0.0001228,
+      "loss": 0.0786,
       "step": 4276
     },
     {
       "epoch": 77.76,
-      "learning_rate": 0.0001228,
-      "loss": 0.1223,
+      "learning_rate": 0.0001227,
+      "loss": 0.0877,
       "step": 4277
     },
     {
       "epoch": 77.78,
-      "learning_rate": 0.0001227,
-      "loss": 0.1045,
+      "learning_rate": 0.0001226,
+      "loss": 0.1036,
       "step": 4278
     },
     {
       "epoch": 77.8,
-      "learning_rate": 0.0001226,
-      "loss": 0.0862,
+      "learning_rate": 0.0001225,
+      "loss": 0.0634,
       "step": 4279
     },
     {
       "epoch": 77.82,
-      "learning_rate": 0.0001225,
-      "loss": 0.1159,
+      "learning_rate": 0.0001224,
+      "loss": 0.1081,
       "step": 4280
     },
     {
       "epoch": 77.84,
-      "learning_rate": 0.0001224,
-      "loss": 0.1239,
+      "learning_rate": 0.00012230000000000002,
+      "loss": 0.0876,
       "step": 4281
     },
     {
       "epoch": 77.85,
-      "learning_rate": 0.00012230000000000002,
-      "loss": 0.1211,
+      "learning_rate": 0.00012220000000000002,
+      "loss": 0.0919,
       "step": 4282
     },
     {
       "epoch": 77.87,
       "learning_rate": 0.00012220000000000002,
-      "loss": 0.0949,
+      "loss": 0.0937,
       "step": 4283
     },
     {
       "epoch": 77.89,
       "learning_rate": 0.0001221,
-      "loss": 0.1441,
+      "loss": 0.1063,
       "step": 4284
     },
     {
       "epoch": 77.91,
       "learning_rate": 0.000122,
-      "loss": 0.1111,
+      "loss": 0.0802,
       "step": 4285
     },
     {
       "epoch": 77.93,
       "learning_rate": 0.0001219,
-      "loss": 0.1022,
+      "loss": 0.1047,
       "step": 4286
     },
     {
       "epoch": 77.95,
       "learning_rate": 0.0001218,
-      "loss": 0.1012,
+      "loss": 0.0942,
       "step": 4287
     },
     {
       "epoch": 77.96,
       "learning_rate": 0.0001217,
-      "loss": 0.1177,
+      "loss": 0.0916,
       "step": 4288
     },
     {
       "epoch": 77.98,
       "learning_rate": 0.0001216,
-      "loss": 0.0669,
+      "loss": 0.0684,
       "step": 4289
     },
     {
       "epoch": 78.0,
       "learning_rate": 0.0001215,
-      "loss": 0.0422,
+      "loss": 0.0286,
       "step": 4290
     },
     {
       "epoch": 78.02,
       "learning_rate": 0.0001214,
-      "loss": 0.1043,
+      "loss": 0.1031,
       "step": 4291
     },
     {
       "epoch": 78.04,
       "learning_rate": 0.00012130000000000001,
-      "loss": 0.0873,
+      "loss": 0.083,
       "step": 4292
     },
     {
       "epoch": 78.05,
       "learning_rate": 0.0001212,
-      "loss": 0.0879,
+      "loss": 0.0894,
       "step": 4293
     },
     {
       "epoch": 78.07,
       "learning_rate": 0.0001211,
-      "loss": 0.0715,
+      "loss": 0.0809,
       "step": 4294
     },
     {
       "epoch": 78.09,
       "learning_rate": 0.000121,
-      "loss": 0.0884,
+      "loss": 0.0873,
       "step": 4295
     },
     {
       "epoch": 78.11,
       "learning_rate": 0.0001209,
-      "loss": 0.0965,
+      "loss": 0.0872,
       "step": 4296
     },
     {
       "epoch": 78.13,
       "learning_rate": 0.00012080000000000001,
-      "loss": 0.1292,
+      "loss": 0.1139,
       "step": 4297
     },
     {
       "epoch": 78.15,
       "learning_rate": 0.0001207,
-      "loss": 0.1121,
+      "loss": 0.1008,
       "step": 4298
     },
     {
       "epoch": 78.16,
       "learning_rate": 0.0001206,
-      "loss": 0.1038,
+      "loss": 0.1132,
       "step": 4299
     },
     {
       "epoch": 78.18,
       "learning_rate": 0.0001205,
-      "loss": 0.0891,
+      "loss": 0.0948,
       "step": 4300
     },
     {
       "epoch": 78.2,
       "learning_rate": 0.0001204,
-      "loss": 0.1323,
+      "loss": 0.1155,
       "step": 4301
     },
     {
       "epoch": 78.22,
       "learning_rate": 0.00012030000000000001,
-      "loss": 0.1303,
+      "loss": 0.1019,
       "step": 4302
     },
     {
       "epoch": 78.24,
       "learning_rate": 0.00012020000000000001,
-      "loss": 0.074,
+      "loss": 0.0663,
       "step": 4303
     },
     {
       "epoch": 78.25,
       "learning_rate": 0.0001201,
-      "loss": 0.0917,
+      "loss": 0.0749,
       "step": 4304
     },
     {
       "epoch": 78.27,
       "learning_rate": 0.00012,
-      "loss": 0.1076,
+      "loss": 0.0841,
       "step": 4305
     },
     {
       "epoch": 78.29,
       "learning_rate": 0.00011990000000000001,
-      "loss": 0.1053,
+      "loss": 0.0953,
       "step": 4306
     },
     {
       "epoch": 78.31,
       "learning_rate": 0.00011980000000000001,
-      "loss": 0.0978,
+      "loss": 0.0989,
       "step": 4307
     },
     {
       "epoch": 78.33,
       "learning_rate": 0.00011970000000000001,
-      "loss": 0.0824,
+      "loss": 0.0986,
       "step": 4308
     },
     {
       "epoch": 78.35,
       "learning_rate": 0.00011960000000000001,
-      "loss": 0.0781,
+      "loss": 0.1017,
       "step": 4309
     },
     {
       "epoch": 78.36,
       "learning_rate": 0.00011949999999999999,
-      "loss": 0.1374,
+      "loss": 0.1253,
       "step": 4310
     },
     {
       "epoch": 78.38,
       "learning_rate": 0.0001194,
-      "loss": 0.0933,
+      "loss": 0.0925,
       "step": 4311
     },
     {
       "epoch": 78.4,
       "learning_rate": 0.0001193,
-      "loss": 0.0729,
+      "loss": 0.0638,
       "step": 4312
     },
     {
       "epoch": 78.42,
       "learning_rate": 0.0001192,
-      "loss": 0.1043,
+      "loss": 0.1001,
       "step": 4313
     },
     {
       "epoch": 78.44,
       "learning_rate": 0.0001191,
-      "loss": 0.067,
+      "loss": 0.0701,
       "step": 4314
     },
     {
       "epoch": 78.45,
       "learning_rate": 0.00011899999999999999,
-      "loss": 0.078,
+      "loss": 0.0562,
       "step": 4315
     },
     {
       "epoch": 78.47,
       "learning_rate": 0.0001189,
-      "loss": 0.1259,
+      "loss": 0.1178,
       "step": 4316
     },
     {
       "epoch": 78.49,
       "learning_rate": 0.0001188,
-      "loss": 0.1154,
+      "loss": 0.1093,
       "step": 4317
     },
     {
       "epoch": 78.51,
       "learning_rate": 0.0001187,
-      "loss": 0.0896,
+      "loss": 0.0737,
       "step": 4318
     },
     {
       "epoch": 78.53,
       "learning_rate": 0.0001186,
-      "loss": 0.0954,
+      "loss": 0.0764,
       "step": 4319
     },
     {
       "epoch": 78.55,
       "learning_rate": 0.0001185,
-      "loss": 0.0633,
+      "loss": 0.0484,
       "step": 4320
     },
     {
       "epoch": 78.56,
       "learning_rate": 0.0001184,
-      "loss": 0.0733,
+      "loss": 0.0715,
       "step": 4321
     },
     {
       "epoch": 78.58,
       "learning_rate": 0.0001183,
-      "loss": 0.1348,
+      "loss": 0.1445,
       "step": 4322
     },
     {
       "epoch": 78.6,
       "learning_rate": 0.0001182,
-      "loss": 0.1552,
+      "loss": 0.1117,
       "step": 4323
     },
     {
       "epoch": 78.62,
       "learning_rate": 0.0001181,
-      "loss": 0.0738,
+      "loss": 0.0561,
       "step": 4324
     },
     {
       "epoch": 78.64,
       "learning_rate": 0.000118,
-      "loss": 0.1115,
+      "loss": 0.1117,
       "step": 4325
     },
     {
       "epoch": 78.65,
       "learning_rate": 0.0001179,
-      "loss": 0.0912,
+      "loss": 0.0641,
       "step": 4326
     },
     {
       "epoch": 78.67,
       "learning_rate": 0.0001178,
-      "loss": 0.1367,
+      "loss": 0.0971,
       "step": 4327
     },
     {
       "epoch": 78.69,
       "learning_rate": 0.0001177,
-      "loss": 0.0875,
+      "loss": 0.0704,
       "step": 4328
     },
     {
       "epoch": 78.71,
       "learning_rate": 0.0001176,
-      "loss": 0.0958,
+      "loss": 0.0922,
       "step": 4329
     },
     {
       "epoch": 78.73,
       "learning_rate": 0.0001175,
-      "loss": 0.1092,
+      "loss": 0.0954,
       "step": 4330
     },
     {
       "epoch": 78.75,
       "learning_rate": 0.00011740000000000001,
-      "loss": 0.1154,
+      "loss": 0.0802,
       "step": 4331
     },
     {
       "epoch": 78.76,
       "learning_rate": 0.0001173,
-      "loss": 0.0963,
+      "loss": 0.0989,
       "step": 4332
     },
     {
       "epoch": 78.78,
       "learning_rate": 0.0001172,
-      "loss": 0.125,
+      "loss": 0.1061,
       "step": 4333
     },
     {
       "epoch": 78.8,
       "learning_rate": 0.0001171,
-      "loss": 0.0829,
+      "loss": 0.0729,
       "step": 4334
     },
     {
       "epoch": 78.82,
       "learning_rate": 0.00011700000000000001,
-      "loss": 0.1113,
+      "loss": 0.102,
       "step": 4335
     },
     {
       "epoch": 78.84,
       "learning_rate": 0.00011690000000000001,
-      "loss": 0.133,
+      "loss": 0.0845,
       "step": 4336
     },
     {
       "epoch": 78.85,
       "learning_rate": 0.0001168,
-      "loss": 0.1539,
+      "loss": 0.1469,
       "step": 4337
     },
     {
       "epoch": 78.87,
       "learning_rate": 0.0001167,
-      "loss": 0.1056,
+      "loss": 0.1055,
       "step": 4338
     },
     {
       "epoch": 78.89,
       "learning_rate": 0.0001166,
-      "loss": 0.1242,
+      "loss": 0.0986,
       "step": 4339
     },
     {
       "epoch": 78.91,
       "learning_rate": 0.00011650000000000001,
-      "loss": 0.0561,
+      "loss": 0.0641,
       "step": 4340
     },
     {
       "epoch": 78.93,
       "learning_rate": 0.00011640000000000001,
-      "loss": 0.098,
+      "loss": 0.1152,
       "step": 4341
     },
     {
       "epoch": 78.95,
       "learning_rate": 0.00011630000000000001,
-      "loss": 0.097,
+      "loss": 0.1054,
       "step": 4342
     },
     {
       "epoch": 78.96,
       "learning_rate": 0.0001162,
-      "loss": 0.111,
+      "loss": 0.0842,
       "step": 4343
     },
     {
       "epoch": 78.98,
       "learning_rate": 0.0001161,
-      "loss": 0.0773,
+      "loss": 0.0454,
       "step": 4344
     },
     {
       "epoch": 79.0,
       "learning_rate": 0.00011600000000000001,
-      "loss": 0.0957,
+      "loss": 0.0777,
       "step": 4345
     },
     {
       "epoch": 79.02,
       "learning_rate": 0.00011590000000000001,
-      "loss": 0.0947,
+      "loss": 0.0985,
       "step": 4346
     },
     {
       "epoch": 79.04,
       "learning_rate": 0.0001158,
-      "loss": 0.0897,
+      "loss": 0.0976,
       "step": 4347
     },
     {
       "epoch": 79.05,
       "learning_rate": 0.0001157,
-      "loss": 0.1066,
+      "loss": 0.1118,
       "step": 4348
     },
     {
       "epoch": 79.07,
       "learning_rate": 0.00011559999999999999,
-      "loss": 0.0842,
+      "loss": 0.0738,
       "step": 4349
     },
     {
       "epoch": 79.09,
       "learning_rate": 0.0001155,
-      "loss": 0.0933,
+      "loss": 0.1286,
       "step": 4350
     },
     {
       "epoch": 79.11,
       "learning_rate": 0.0001154,
-      "loss": 0.0515,
+      "loss": 0.0581,
       "step": 4351
     },
     {
       "epoch": 79.13,
       "learning_rate": 0.0001153,
-      "loss": 0.1315,
+      "loss": 0.0986,
       "step": 4352
     },
     {
       "epoch": 79.15,
       "learning_rate": 0.0001152,
-      "loss": 0.0938,
+      "loss": 0.0958,
       "step": 4353
     },
     {
       "epoch": 79.16,
       "learning_rate": 0.00011509999999999999,
-      "loss": 0.1278,
+      "loss": 0.1259,
       "step": 4354
     },
     {
       "epoch": 79.18,
       "learning_rate": 0.000115,
-      "loss": 0.1187,
+      "loss": 0.0915,
       "step": 4355
     },
     {
       "epoch": 79.2,
       "learning_rate": 0.0001149,
-      "loss": 0.1106,
+      "loss": 0.0792,
       "step": 4356
     },
     {
       "epoch": 79.22,
       "learning_rate": 0.0001148,
-      "loss": 0.1155,
+      "loss": 0.1049,
       "step": 4357
     },
     {
       "epoch": 79.24,
       "learning_rate": 0.0001147,
-      "loss": 0.0783,
+      "loss": 0.0836,
       "step": 4358
     },
     {
       "epoch": 79.25,
       "learning_rate": 0.0001146,
-      "loss": 0.1046,
+      "loss": 0.0982,
       "step": 4359
     },
     {
       "epoch": 79.27,
       "learning_rate": 0.0001145,
-      "loss": 0.082,
+      "loss": 0.0787,
       "step": 4360
     },
     {
       "epoch": 79.29,
       "learning_rate": 0.0001144,
-      "loss": 0.0873,
+      "loss": 0.0811,
       "step": 4361
     },
     {
       "epoch": 79.31,
       "learning_rate": 0.0001143,
-      "loss": 0.0712,
+      "loss": 0.0662,
       "step": 4362
     },
     {
       "epoch": 79.33,
       "learning_rate": 0.0001142,
-      "loss": 0.0663,
+      "loss": 0.0845,
       "step": 4363
     },
     {
       "epoch": 79.35,
       "learning_rate": 0.0001141,
-      "loss": 0.0712,
+      "loss": 0.0859,
       "step": 4364
     },
     {
       "epoch": 79.36,
       "learning_rate": 0.000114,
-      "loss": 0.1267,
+      "loss": 0.1105,
       "step": 4365
     },
     {
       "epoch": 79.38,
       "learning_rate": 0.0001139,
-      "loss": 0.1083,
+      "loss": 0.0922,
       "step": 4366
     },
     {
       "epoch": 79.4,
       "learning_rate": 0.0001138,
-      "loss": 0.1,
+      "loss": 0.0827,
       "step": 4367
     },
     {
       "epoch": 79.42,
       "learning_rate": 0.0001137,
-      "loss": 0.1273,
+      "loss": 0.1164,
       "step": 4368
     },
     {
       "epoch": 79.44,
       "learning_rate": 0.00011360000000000001,
-      "loss": 0.0881,
+      "loss": 0.0659,
       "step": 4369
     },
     {
       "epoch": 79.45,
       "learning_rate": 0.00011350000000000001,
-      "loss": 0.0784,
+      "loss": 0.0971,
       "step": 4370
     },
     {
       "epoch": 79.47,
       "learning_rate": 0.0001134,
-      "loss": 0.1343,
+      "loss": 0.1403,
       "step": 4371
     },
     {
       "epoch": 79.49,
       "learning_rate": 0.0001133,
-      "loss": 0.0763,
+      "loss": 0.0802,
       "step": 4372
     },
     {
       "epoch": 79.51,
       "learning_rate": 0.0001132,
-      "loss": 0.1147,
+      "loss": 0.112,
       "step": 4373
     },
     {
       "epoch": 79.53,
       "learning_rate": 0.00011310000000000001,
-      "loss": 0.0761,
+      "loss": 0.0827,
       "step": 4374
     },
     {
       "epoch": 79.55,
       "learning_rate": 0.00011300000000000001,
-      "loss": 0.0879,
+      "loss": 0.0683,
       "step": 4375
     },
     {
       "epoch": 79.56,
       "learning_rate": 0.0001129,
-      "loss": 0.0489,
+      "loss": 0.0644,
       "step": 4376
     },
     {
       "epoch": 79.58,
       "learning_rate": 0.0001128,
-      "loss": 0.1202,
+      "loss": 0.1267,
       "step": 4377
     },
     {
       "epoch": 79.6,
       "learning_rate": 0.0001127,
-      "loss": 0.0776,
+      "loss": 0.0684,
       "step": 4378
     },
     {
       "epoch": 79.62,
       "learning_rate": 0.00011260000000000001,
-      "loss": 0.1273,
+      "loss": 0.1163,
       "step": 4379
     },
     {
       "epoch": 79.64,
       "learning_rate": 0.00011250000000000001,
-      "loss": 0.1366,
+      "loss": 0.1131,
       "step": 4380
     },
     {
       "epoch": 79.65,
       "learning_rate": 0.00011240000000000001,
-      "loss": 0.1338,
+      "loss": 0.1094,
       "step": 4381
     },
     {
       "epoch": 79.67,
       "learning_rate": 0.0001123,
-      "loss": 0.1168,
+      "loss": 0.0988,
       "step": 4382
     },
     {
       "epoch": 79.69,
       "learning_rate": 0.0001122,
-      "loss": 0.0876,
+      "loss": 0.0725,
       "step": 4383
     },
     {
       "epoch": 79.71,
       "learning_rate": 0.00011210000000000001,
-      "loss": 0.1183,
+      "loss": 0.0996,
       "step": 4384
     },
     {
       "epoch": 79.73,
       "learning_rate": 0.000112,
-      "loss": 0.073,
+      "loss": 0.0732,
       "step": 4385
     },
     {
       "epoch": 79.75,
       "learning_rate": 0.0001119,
-      "loss": 0.1049,
+      "loss": 0.1008,
       "step": 4386
     },
     {
       "epoch": 79.76,
       "learning_rate": 0.0001118,
-      "loss": 0.0826,
+      "loss": 0.0915,
       "step": 4387
     },
     {
       "epoch": 79.78,
       "learning_rate": 0.00011169999999999999,
-      "loss": 0.1412,
+      "loss": 0.1282,
       "step": 4388
     },
     {
       "epoch": 79.8,
       "learning_rate": 0.0001116,
-      "loss": 0.0802,
+      "loss": 0.0837,
       "step": 4389
     },
     {
       "epoch": 79.82,
       "learning_rate": 0.0001115,
-      "loss": 0.1352,
+      "loss": 0.1193,
       "step": 4390
     },
     {
       "epoch": 79.84,
       "learning_rate": 0.0001114,
-      "loss": 0.0915,
+      "loss": 0.086,
       "step": 4391
     },
     {
       "epoch": 79.85,
       "learning_rate": 0.0001113,
-      "loss": 0.0872,
+      "loss": 0.0785,
       "step": 4392
     },
     {
       "epoch": 79.87,
       "learning_rate": 0.00011119999999999999,
-      "loss": 0.062,
+      "loss": 0.0726,
       "step": 4393
     },
     {
       "epoch": 79.89,
       "learning_rate": 0.0001111,
-      "loss": 0.0912,
+      "loss": 0.1069,
       "step": 4394
     },
     {
       "epoch": 79.91,
       "learning_rate": 0.000111,
-      "loss": 0.1381,
+      "loss": 0.0457,
       "step": 4395
     },
     {
       "epoch": 79.93,
       "learning_rate": 0.0001109,
-      "loss": 0.1316,
+      "loss": 0.1258,
       "step": 4396
     },
     {
       "epoch": 79.95,
       "learning_rate": 0.0001108,
-      "loss": 0.1064,
+      "loss": 0.0926,
       "step": 4397
     },
     {
       "epoch": 79.96,
       "learning_rate": 0.00011070000000000001,
-      "loss": 0.1025,
+      "loss": 0.1057,
       "step": 4398
     },
     {
       "epoch": 79.98,
       "learning_rate": 0.0001106,
-      "loss": 0.1149,
+      "loss": 0.101,
       "step": 4399
     },
     {
       "epoch": 80.0,
       "learning_rate": 0.0001105,
-      "loss": 0.0348,
+      "loss": 0.055,
       "step": 4400
     },
     {
       "epoch": 80.02,
       "learning_rate": 0.0001104,
-      "loss": 0.1112,
+      "loss": 0.0989,
       "step": 4401
     },
     {
       "epoch": 80.04,
       "learning_rate": 0.0001103,
-      "loss": 0.1085,
+      "loss": 0.0967,
       "step": 4402
     },
     {
       "epoch": 80.05,
       "learning_rate": 0.00011020000000000001,
-      "loss": 0.1093,
+      "loss": 0.1017,
       "step": 4403
     },
     {
       "epoch": 80.07,
       "learning_rate": 0.0001101,
-      "loss": 0.0883,
+      "loss": 0.1108,
       "step": 4404
     },
     {
       "epoch": 80.09,
       "learning_rate": 0.00011,
-      "loss": 0.1226,
+      "loss": 0.1274,
       "step": 4405
     },
     {
       "epoch": 80.11,
       "learning_rate": 0.0001099,
-      "loss": 0.1027,
+      "loss": 0.0746,
       "step": 4406
     },
     {
       "epoch": 80.13,
       "learning_rate": 0.0001098,
-      "loss": 0.0988,
+      "loss": 0.078,
       "step": 4407
     },
     {
       "epoch": 80.15,
       "learning_rate": 0.00010970000000000001,
-      "loss": 0.1076,
+      "loss": 0.082,
       "step": 4408
     },
     {
       "epoch": 80.16,
       "learning_rate": 0.00010960000000000001,
-      "loss": 0.1021,
+      "loss": 0.1107,
       "step": 4409
     },
     {
       "epoch": 80.18,
       "learning_rate": 0.0001095,
-      "loss": 0.1082,
+      "loss": 0.0796,
       "step": 4410
     },
     {
       "epoch": 80.2,
       "learning_rate": 0.0001094,
-      "loss": 0.097,
+      "loss": 0.0761,
       "step": 4411
     },
     {
       "epoch": 80.22,
       "learning_rate": 0.0001093,
-      "loss": 0.0954,
+      "loss": 0.1059,
       "step": 4412
     },
     {
       "epoch": 80.24,
       "learning_rate": 0.00010920000000000001,
-      "loss": 0.0807,
+      "loss": 0.0919,
       "step": 4413
     },
     {
       "epoch": 80.25,
       "learning_rate": 0.00010910000000000001,
-      "loss": 0.0992,
+      "loss": 0.0903,
       "step": 4414
     },
     {
       "epoch": 80.27,
       "learning_rate": 0.000109,
-      "loss": 0.0802,
+      "loss": 0.0877,
       "step": 4415
     },
     {
       "epoch": 80.29,
       "learning_rate": 0.0001089,
-      "loss": 0.0689,
+      "loss": 0.0859,
       "step": 4416
     },
     {
       "epoch": 80.31,
       "learning_rate": 0.0001088,
-      "loss": 0.0854,
+      "loss": 0.081,
       "step": 4417
     },
     {
       "epoch": 80.33,
       "learning_rate": 0.00010870000000000001,
-      "loss": 0.0912,
+      "loss": 0.0748,
       "step": 4418
     },
     {
       "epoch": 80.35,
       "learning_rate": 0.00010860000000000001,
-      "loss": 0.1151,
+      "loss": 0.0718,
       "step": 4419
     },
     {
       "epoch": 80.36,
       "learning_rate": 0.00010850000000000001,
-      "loss": 0.114,
+      "loss": 0.0775,
       "step": 4420
     },
     {
       "epoch": 80.38,
       "learning_rate": 0.00010839999999999999,
-      "loss": 0.08,
+      "loss": 0.0603,
       "step": 4421
     },
     {
       "epoch": 80.4,
       "learning_rate": 0.00010829999999999999,
-      "loss": 0.1038,
+      "loss": 0.1089,
       "step": 4422
     },
     {
       "epoch": 80.42,
       "learning_rate": 0.0001082,
-      "loss": 0.1218,
+      "loss": 0.109,
       "step": 4423
     },
     {
       "epoch": 80.44,
       "learning_rate": 0.0001081,
-      "loss": 0.083,
+      "loss": 0.1079,
       "step": 4424
     },
     {
       "epoch": 80.45,
       "learning_rate": 0.000108,
-      "loss": 0.0779,
+      "loss": 0.0756,
       "step": 4425
     },
     {
       "epoch": 80.47,
       "learning_rate": 0.0001079,
-      "loss": 0.1209,
+      "loss": 0.1138,
       "step": 4426
     },
     {
       "epoch": 80.49,
       "learning_rate": 0.0001078,
-      "loss": 0.1197,
+      "loss": 0.107,
       "step": 4427
     },
     {
       "epoch": 80.51,
       "learning_rate": 0.0001077,
-      "loss": 0.098,
+      "loss": 0.1287,
       "step": 4428
     },
     {
       "epoch": 80.53,
       "learning_rate": 0.0001076,
-      "loss": 0.0825,
+      "loss": 0.0492,
       "step": 4429
     },
     {
       "epoch": 80.55,
       "learning_rate": 0.0001075,
-      "loss": 0.1252,
+      "loss": 0.1155,
       "step": 4430
     },
     {
       "epoch": 80.56,
       "learning_rate": 0.0001074,
-      "loss": 0.0948,
+      "loss": 0.0801,
       "step": 4431
     },
     {
       "epoch": 80.58,
       "learning_rate": 0.0001073,
-      "loss": 0.0804,
+      "loss": 0.083,
       "step": 4432
     },
     {
       "epoch": 80.6,
       "learning_rate": 0.0001072,
-      "loss": 0.0733,
+      "loss": 0.0746,
       "step": 4433
     },
     {
       "epoch": 80.62,
       "learning_rate": 0.0001071,
-      "loss": 0.1318,
+      "loss": 0.1202,
       "step": 4434
     },
     {
       "epoch": 80.64,
       "learning_rate": 0.000107,
-      "loss": 0.0945,
+      "loss": 0.0984,
       "step": 4435
     },
     {
       "epoch": 80.65,
       "learning_rate": 0.0001069,
-      "loss": 0.0957,
+      "loss": 0.0738,
       "step": 4436
     },
     {
       "epoch": 80.67,
       "learning_rate": 0.00010680000000000001,
-      "loss": 0.1301,
+      "loss": 0.1002,
       "step": 4437
     },
     {
       "epoch": 80.69,
       "learning_rate": 0.0001067,
-      "loss": 0.0779,
+      "loss": 0.0663,
       "step": 4438
     },
     {
       "epoch": 80.71,
       "learning_rate": 0.0001066,
-      "loss": 0.1229,
+      "loss": 0.1123,
       "step": 4439
     },
     {
       "epoch": 80.73,
       "learning_rate": 0.0001065,
-      "loss": 0.1096,
+      "loss": 0.0936,
       "step": 4440
     },
     {
       "epoch": 80.75,
       "learning_rate": 0.0001064,
-      "loss": 0.0949,
+      "loss": 0.0733,
       "step": 4441
     },
     {
       "epoch": 80.76,
       "learning_rate": 0.00010630000000000001,
-      "loss": 0.088,
+      "loss": 0.068,
       "step": 4442
     },
     {
       "epoch": 80.78,
       "learning_rate": 0.0001062,
-      "loss": 0.0624,
+      "loss": 0.0574,
       "step": 4443
     },
     {
       "epoch": 80.8,
       "learning_rate": 0.0001061,
-      "loss": 0.0955,
+      "loss": 0.0722,
       "step": 4444
     },
     {
       "epoch": 80.82,
       "learning_rate": 0.000106,
-      "loss": 0.0837,
+      "loss": 0.0871,
       "step": 4445
     },
     {
       "epoch": 80.84,
       "learning_rate": 0.0001059,
-      "loss": 0.1078,
+      "loss": 0.0836,
       "step": 4446
     },
     {
       "epoch": 80.85,
       "learning_rate": 0.00010580000000000001,
-      "loss": 0.0859,
+      "loss": 0.0676,
       "step": 4447
     },
     {
       "epoch": 80.87,
       "learning_rate": 0.00010570000000000001,
-      "loss": 0.0877,
+      "loss": 0.0805,
       "step": 4448
     },
     {
       "epoch": 80.89,
       "learning_rate": 0.0001056,
-      "loss": 0.0945,
+      "loss": 0.0696,
       "step": 4449
     },
     {
       "epoch": 80.91,
       "learning_rate": 0.0001055,
-      "loss": 0.08,
+      "loss": 0.057,
       "step": 4450
     },
     {
       "epoch": 80.93,
       "learning_rate": 0.0001054,
-      "loss": 0.1307,
+      "loss": 0.0847,
       "step": 4451
     },
     {
       "epoch": 80.95,
       "learning_rate": 0.00010530000000000001,
-      "loss": 0.0937,
+      "loss": 0.0758,
       "step": 4452
     },
     {
       "epoch": 80.96,
       "learning_rate": 0.00010520000000000001,
-      "loss": 0.0826,
+      "loss": 0.0692,
       "step": 4453
     },
     {
       "epoch": 80.98,
       "learning_rate": 0.00010510000000000001,
-      "loss": 0.1126,
+      "loss": 0.1087,
       "step": 4454
     },
     {
       "epoch": 81.0,
       "learning_rate": 0.000105,
-      "loss": 0.079,
+      "loss": 0.0701,
       "step": 4455
     },
     {
       "epoch": 81.02,
       "learning_rate": 0.0001049,
-      "loss": 0.1104,
+      "loss": 0.1012,
       "step": 4456
     },
     {
       "epoch": 81.04,
       "learning_rate": 0.00010480000000000001,
-      "loss": 0.094,
+      "loss": 0.0984,
       "step": 4457
     },
     {
       "epoch": 81.05,
       "learning_rate": 0.0001047,
-      "loss": 0.1176,
+      "loss": 0.1036,
       "step": 4458
     },
     {
       "epoch": 81.07,
       "learning_rate": 0.0001046,
-      "loss": 0.0907,
+      "loss": 0.1004,
       "step": 4459
     },
     {
       "epoch": 81.09,
       "learning_rate": 0.00010449999999999999,
-      "loss": 0.1007,
+      "loss": 0.0992,
       "step": 4460
     },
     {
       "epoch": 81.11,
       "learning_rate": 0.0001044,
-      "loss": 0.0752,
+      "loss": 0.0636,
       "step": 4461
     },
     {
       "epoch": 81.13,
       "learning_rate": 0.0001043,
-      "loss": 0.0936,
+      "loss": 0.0775,
       "step": 4462
     },
     {
       "epoch": 81.15,
       "learning_rate": 0.0001042,
-      "loss": 0.1011,
+      "loss": 0.077,
       "step": 4463
     },
     {
       "epoch": 81.16,
       "learning_rate": 0.0001041,
-      "loss": 0.0873,
+      "loss": 0.0882,
       "step": 4464
     },
     {
       "epoch": 81.18,
       "learning_rate": 0.000104,
-      "loss": 0.1007,
+      "loss": 0.1135,
       "step": 4465
     },
     {
       "epoch": 81.2,
       "learning_rate": 0.0001039,
-      "loss": 0.0846,
+      "loss": 0.0787,
       "step": 4466
     },
     {
       "epoch": 81.22,
       "learning_rate": 0.0001038,
-      "loss": 0.0841,
+      "loss": 0.0561,
       "step": 4467
     },
     {
       "epoch": 81.24,
       "learning_rate": 0.0001037,
-      "loss": 0.0978,
+      "loss": 0.0864,
       "step": 4468
     },
     {
       "epoch": 81.25,
       "learning_rate": 0.0001036,
-      "loss": 0.0909,
+      "loss": 0.0793,
       "step": 4469
     },
     {
       "epoch": 81.27,
       "learning_rate": 0.0001035,
-      "loss": 0.1226,
+      "loss": 0.1082,
       "step": 4470
     },
     {
       "epoch": 81.29,
       "learning_rate": 0.0001034,
-      "loss": 0.0993,
+      "loss": 0.0779,
       "step": 4471
     },
     {
       "epoch": 81.31,
       "learning_rate": 0.0001033,
-      "loss": 0.1098,
+      "loss": 0.0927,
       "step": 4472
     },
     {
       "epoch": 81.33,
       "learning_rate": 0.0001032,
-      "loss": 0.0849,
+      "loss": 0.0788,
       "step": 4473
     },
     {
       "epoch": 81.35,
       "learning_rate": 0.0001031,
-      "loss": 0.1046,
+      "loss": 0.1044,
       "step": 4474
     },
     {
       "epoch": 81.36,
       "learning_rate": 0.000103,
-      "loss": 0.0998,
+      "loss": 0.0881,
       "step": 4475
     },
     {
       "epoch": 81.38,
       "learning_rate": 0.00010290000000000001,
-      "loss": 0.0936,
+      "loss": 0.0837,
       "step": 4476
     },
     {
       "epoch": 81.4,
       "learning_rate": 0.0001028,
-      "loss": 0.0834,
+      "loss": 0.0882,
       "step": 4477
     },
     {
       "epoch": 81.42,
       "learning_rate": 0.0001027,
-      "loss": 0.0994,
+      "loss": 0.075,
       "step": 4478
     },
     {
       "epoch": 81.44,
       "learning_rate": 0.0001026,
-      "loss": 0.1052,
+      "loss": 0.076,
       "step": 4479
     },
     {
       "epoch": 81.45,
       "learning_rate": 0.0001025,
-      "loss": 0.0559,
+      "loss": 0.0597,
       "step": 4480
     },
     {
       "epoch": 81.47,
       "learning_rate": 0.00010240000000000001,
-      "loss": 0.1484,
+      "loss": 0.1153,
       "step": 4481
     },
     {
       "epoch": 81.49,
       "learning_rate": 0.0001023,
-      "loss": 0.1131,
+      "loss": 0.1062,
       "step": 4482
     },
     {
       "epoch": 81.51,
       "learning_rate": 0.0001022,
-      "loss": 0.0862,
+      "loss": 0.0724,
       "step": 4483
     },
     {
       "epoch": 81.53,
       "learning_rate": 0.0001021,
-      "loss": 0.092,
+      "loss": 0.0963,
       "step": 4484
     },
     {
       "epoch": 81.55,
       "learning_rate": 0.000102,
-      "loss": 0.0892,
+      "loss": 0.0824,
       "step": 4485
     },
     {
       "epoch": 81.56,
       "learning_rate": 0.00010190000000000001,
-      "loss": 0.1033,
+      "loss": 0.0968,
       "step": 4486
     },
     {
       "epoch": 81.58,
       "learning_rate": 0.00010180000000000001,
-      "loss": 0.118,
+      "loss": 0.1224,
       "step": 4487
     },
     {
       "epoch": 81.6,
       "learning_rate": 0.0001017,
-      "loss": 0.0956,
+      "loss": 0.0737,
       "step": 4488
     },
     {
       "epoch": 81.62,
       "learning_rate": 0.0001016,
-      "loss": 0.1089,
+      "loss": 0.1009,
       "step": 4489
     },
     {
       "epoch": 81.64,
       "learning_rate": 0.00010150000000000001,
-      "loss": 0.0958,
+      "loss": 0.0805,
       "step": 4490
     },
     {
       "epoch": 81.65,
       "learning_rate": 0.00010140000000000001,
-      "loss": 0.079,
+      "loss": 0.0845,
       "step": 4491
     },
     {
       "epoch": 81.67,
       "learning_rate": 0.00010130000000000001,
-      "loss": 0.103,
+      "loss": 0.1161,
       "step": 4492
     },
     {
       "epoch": 81.69,
       "learning_rate": 0.00010120000000000001,
-      "loss": 0.0816,
+      "loss": 0.0613,
       "step": 4493
     },
     {
       "epoch": 81.71,
       "learning_rate": 0.00010109999999999999,
-      "loss": 0.1345,
+      "loss": 0.1091,
       "step": 4494
     },
     {
       "epoch": 81.73,
       "learning_rate": 0.000101,
-      "loss": 0.0992,
+      "loss": 0.0884,
       "step": 4495
     },
     {
       "epoch": 81.75,
       "learning_rate": 0.0001009,
-      "loss": 0.079,
+      "loss": 0.0862,
       "step": 4496
     },
     {
       "epoch": 81.76,
       "learning_rate": 0.0001008,
-      "loss": 0.0889,
+      "loss": 0.0833,
       "step": 4497
     },
     {
       "epoch": 81.78,
       "learning_rate": 0.0001007,
-      "loss": 0.116,
+      "loss": 0.1155,
       "step": 4498
     },
     {
       "epoch": 81.8,
       "learning_rate": 0.00010059999999999999,
-      "loss": 0.0824,
+      "loss": 0.0689,
       "step": 4499
     },
     {
       "epoch": 81.82,
       "learning_rate": 0.0001005,
-      "loss": 0.0966,
+      "loss": 0.0876,
       "step": 4500
     },
     {
       "epoch": 81.82,
-      "eval_cer": 0.07922157753489557,
-      "eval_loss": 0.40071213245391846,
-      "eval_runtime": 14.0214,
-      "eval_samples_per_second": 117.463,
-      "eval_steps_per_second": 1.854,
-      "eval_wer": 0.31641303237667245,
+      "eval_cer": 0.0781121156239224,
+      "eval_loss": 0.42646142840385437,
+      "eval_runtime": 14.3616,
+      "eval_samples_per_second": 114.681,
+      "eval_steps_per_second": 1.81,
+      "eval_wer": 0.31927280155244614,
       "step": 4500
     },
     {
       "epoch": 81.84,
       "learning_rate": 0.0001004,
-      "loss": 0.1176,
+      "loss": 0.1068,
       "step": 4501
     },
     {
       "epoch": 81.85,
       "learning_rate": 0.0001003,
-      "loss": 0.0992,
+      "loss": 0.0877,
       "step": 4502
     },
     {
       "epoch": 81.87,
       "learning_rate": 0.0001002,
-      "loss": 0.1074,
+      "loss": 0.1018,
       "step": 4503
     },
     {
       "epoch": 81.89,
       "learning_rate": 0.0001001,
-      "loss": 0.111,
+      "loss": 0.1011,
       "step": 4504
     },
     {
       "epoch": 81.91,
       "learning_rate": 0.0001,
-      "loss": 0.0882,
+      "loss": 0.0879,
       "step": 4505
     },
     {
       "epoch": 81.93,
       "learning_rate": 9.99e-05,
-      "loss": 0.104,
+      "loss": 0.1,
       "step": 4506
     },
     {
       "epoch": 81.95,
       "learning_rate": 9.98e-05,
-      "loss": 0.0894,
+      "loss": 0.1184,
       "step": 4507
     },
     {
       "epoch": 81.96,
       "learning_rate": 9.97e-05,
-      "loss": 0.1013,
+      "loss": 0.1082,
       "step": 4508
     },
     {
       "epoch": 81.98,
       "learning_rate": 9.96e-05,
-      "loss": 0.0982,
+      "loss": 0.0813,
       "step": 4509
     },
     {
       "epoch": 82.0,
       "learning_rate": 9.95e-05,
-      "loss": 0.0418,
+      "loss": 0.0421,
       "step": 4510
     },
     {
       "epoch": 82.02,
       "learning_rate": 9.94e-05,
-      "loss": 0.0914,
+      "loss": 0.1016,
       "step": 4511
     },
     {
       "epoch": 82.04,
       "learning_rate": 9.93e-05,
-      "loss": 0.0935,
+      "loss": 0.0971,
       "step": 4512
     },
     {
       "epoch": 82.05,
       "learning_rate": 9.92e-05,
-      "loss": 0.0919,
+      "loss": 0.0856,
       "step": 4513
     },
     {
       "epoch": 82.07,
       "learning_rate": 9.91e-05,
-      "loss": 0.1056,
+      "loss": 0.0985,
       "step": 4514
     },
     {
       "epoch": 82.09,
       "learning_rate": 9.900000000000001e-05,
-      "loss": 0.0547,
+      "loss": 0.0512,
       "step": 4515
     },
     {
       "epoch": 82.11,
       "learning_rate": 9.89e-05,
-      "loss": 0.0842,
+      "loss": 0.073,
       "step": 4516
     },
     {
       "epoch": 82.13,
       "learning_rate": 9.88e-05,
-      "loss": 0.0814,
+      "loss": 0.0827,
       "step": 4517
     },
     {
       "epoch": 82.15,
       "learning_rate": 9.87e-05,
-      "loss": 0.102,
+      "loss": 0.1157,
       "step": 4518
     },
     {
       "epoch": 82.16,
       "learning_rate": 9.86e-05,
-      "loss": 0.1061,
+      "loss": 0.0911,
       "step": 4519
     },
     {
       "epoch": 82.18,
       "learning_rate": 9.850000000000001e-05,
-      "loss": 0.0881,
+      "loss": 0.0768,
       "step": 4520
     },
     {
       "epoch": 82.2,
       "learning_rate": 9.84e-05,
-      "loss": 0.1055,
+      "loss": 0.0913,
       "step": 4521
     },
     {
       "epoch": 82.22,
       "learning_rate": 9.83e-05,
-      "loss": 0.1246,
+      "loss": 0.1082,
       "step": 4522
     },
     {
       "epoch": 82.24,
       "learning_rate": 9.82e-05,
-      "loss": 0.0753,
+      "loss": 0.0782,
       "step": 4523
     },
     {
       "epoch": 82.25,
       "learning_rate": 9.810000000000001e-05,
-      "loss": 0.0868,
+      "loss": 0.0936,
       "step": 4524
     },
     {
       "epoch": 82.27,
       "learning_rate": 9.800000000000001e-05,
-      "loss": 0.0761,
+      "loss": 0.0716,
       "step": 4525
     },
     {
       "epoch": 82.29,
       "learning_rate": 9.790000000000001e-05,
-      "loss": 0.0848,
+      "loss": 0.0661,
       "step": 4526
     },
     {
       "epoch": 82.31,
       "learning_rate": 9.78e-05,
-      "loss": 0.0803,
+      "loss": 0.0616,
       "step": 4527
     },
     {
       "epoch": 82.33,
       "learning_rate": 9.77e-05,
-      "loss": 0.0721,
+      "loss": 0.0723,
       "step": 4528
     },
     {
       "epoch": 82.35,
       "learning_rate": 9.760000000000001e-05,
-      "loss": 0.1059,
+      "loss": 0.0847,
       "step": 4529
     },
     {
       "epoch": 82.36,
       "learning_rate": 9.750000000000001e-05,
-      "loss": 0.1063,
+      "loss": 0.1,
       "step": 4530
     },
     {
       "epoch": 82.38,
       "learning_rate": 9.74e-05,
-      "loss": 0.1034,
+      "loss": 0.0907,
       "step": 4531
     },
     {
       "epoch": 82.4,
       "learning_rate": 9.73e-05,
-      "loss": 0.0915,
+      "loss": 0.1135,
       "step": 4532
     },
     {
       "epoch": 82.42,
       "learning_rate": 9.719999999999999e-05,
-      "loss": 0.0875,
+      "loss": 0.0765,
       "step": 4533
     },
     {
       "epoch": 82.44,
       "learning_rate": 9.71e-05,
-      "loss": 0.0607,
+      "loss": 0.0763,
       "step": 4534
     },
     {
       "epoch": 82.45,
       "learning_rate": 9.7e-05,
-      "loss": 0.0631,
+      "loss": 0.0314,
       "step": 4535
     },
     {
       "epoch": 82.47,
       "learning_rate": 9.69e-05,
-      "loss": 0.0829,
+      "loss": 0.1031,
       "step": 4536
     },
     {
       "epoch": 82.49,
       "learning_rate": 9.68e-05,
-      "loss": 0.1137,
+      "loss": 0.0916,
       "step": 4537
     },
     {
       "epoch": 82.51,
       "learning_rate": 9.669999999999999e-05,
-      "loss": 0.1158,
+      "loss": 0.1105,
       "step": 4538
     },
     {
       "epoch": 82.53,
       "learning_rate": 9.66e-05,
-      "loss": 0.1019,
+      "loss": 0.1037,
       "step": 4539
     },
     {
       "epoch": 82.55,
       "learning_rate": 9.65e-05,
-      "loss": 0.0843,
+      "loss": 0.076,
       "step": 4540
     },
     {
       "epoch": 82.56,
       "learning_rate": 9.64e-05,
-      "loss": 0.0686,
+      "loss": 0.0553,
       "step": 4541
     },
     {
       "epoch": 82.58,
       "learning_rate": 9.63e-05,
-      "loss": 0.1119,
+      "loss": 0.1021,
       "step": 4542
     },
     {
       "epoch": 82.6,
       "learning_rate": 9.62e-05,
-      "loss": 0.0833,
+      "loss": 0.0717,
       "step": 4543
     },
     {
       "epoch": 82.62,
       "learning_rate": 9.61e-05,
-      "loss": 0.122,
+      "loss": 0.1057,
       "step": 4544
     },
     {
       "epoch": 82.64,
       "learning_rate": 9.6e-05,
-      "loss": 0.0634,
+      "loss": 0.0684,
       "step": 4545
     },
     {
       "epoch": 82.65,
       "learning_rate": 9.59e-05,
-      "loss": 0.0858,
+      "loss": 0.0862,
       "step": 4546
     },
     {
       "epoch": 82.67,
       "learning_rate": 9.58e-05,
-      "loss": 0.0854,
+      "loss": 0.0543,
       "step": 4547
     },
     {
       "epoch": 82.69,
       "learning_rate": 9.57e-05,
-      "loss": 0.0858,
+      "loss": 0.0836,
       "step": 4548
     },
     {
       "epoch": 82.71,
       "learning_rate": 9.56e-05,
-      "loss": 0.0814,
+      "loss": 0.0778,
       "step": 4549
     },
     {
       "epoch": 82.73,
       "learning_rate": 9.55e-05,
-      "loss": 0.0914,
+      "loss": 0.1065,
       "step": 4550
     },
     {
       "epoch": 82.75,
       "learning_rate": 9.54e-05,
-      "loss": 0.0949,
+      "loss": 0.0999,
       "step": 4551
     },
     {
       "epoch": 82.76,
       "learning_rate": 9.53e-05,
-      "loss": 0.0812,
+      "loss": 0.0742,
       "step": 4552
     },
     {
       "epoch": 82.78,
       "learning_rate": 9.520000000000001e-05,
-      "loss": 0.1149,
+      "loss": 0.1021,
       "step": 4553
     },
     {
       "epoch": 82.8,
       "learning_rate": 9.510000000000001e-05,
-      "loss": 0.0957,
+      "loss": 0.0942,
       "step": 4554
     },
     {
       "epoch": 82.82,
       "learning_rate": 9.5e-05,
-      "loss": 0.0778,
+      "loss": 0.0718,
       "step": 4555
     },
     {
       "epoch": 82.84,
       "learning_rate": 9.49e-05,
-      "loss": 0.0969,
+      "loss": 0.0798,
       "step": 4556
     },
     {
       "epoch": 82.85,
       "learning_rate": 9.48e-05,
-      "loss": 0.0879,
+      "loss": 0.0994,
       "step": 4557
     },
     {
       "epoch": 82.87,
       "learning_rate": 9.470000000000001e-05,
-      "loss": 0.0796,
+      "loss": 0.0883,
       "step": 4558
     },
     {
       "epoch": 82.89,
       "learning_rate": 9.460000000000001e-05,
-      "loss": 0.0754,
+      "loss": 0.0892,
       "step": 4559
     },
     {
       "epoch": 82.91,
       "learning_rate": 9.45e-05,
-      "loss": 0.0468,
+      "loss": 0.0501,
       "step": 4560
     },
     {
       "epoch": 82.93,
       "learning_rate": 9.44e-05,
-      "loss": 0.0862,
+      "loss": 0.0935,
       "step": 4561
     },
     {
       "epoch": 82.95,
       "learning_rate": 9.43e-05,
-      "loss": 0.1067,
+      "loss": 0.1076,
       "step": 4562
     },
     {
       "epoch": 82.96,
       "learning_rate": 9.420000000000001e-05,
-      "loss": 0.0686,
+      "loss": 0.0654,
       "step": 4563
     },
     {
       "epoch": 82.98,
       "learning_rate": 9.410000000000001e-05,
-      "loss": 0.0777,
+      "loss": 0.0744,
       "step": 4564
     },
     {
       "epoch": 83.0,
       "learning_rate": 9.400000000000001e-05,
-      "loss": 0.0423,
+      "loss": 0.0351,
       "step": 4565
     },
     {
       "epoch": 83.02,
       "learning_rate": 9.39e-05,
-      "loss": 0.1369,
+      "loss": 0.1143,
       "step": 4566
     },
     {
       "epoch": 83.04,
       "learning_rate": 9.38e-05,
-      "loss": 0.1036,
+      "loss": 0.1158,
       "step": 4567
     },
     {
       "epoch": 83.05,
       "learning_rate": 9.37e-05,
-      "loss": 0.0789,
+      "loss": 0.0746,
       "step": 4568
     },
     {
       "epoch": 83.07,
       "learning_rate": 9.36e-05,
-      "loss": 0.0886,
+      "loss": 0.1146,
       "step": 4569
     },
     {
       "epoch": 83.09,
       "learning_rate": 9.35e-05,
-      "loss": 0.1032,
+      "loss": 0.0992,
       "step": 4570
     },
     {
       "epoch": 83.11,
       "learning_rate": 9.34e-05,
-      "loss": 0.0625,
+      "loss": 0.0523,
       "step": 4571
     },
     {
       "epoch": 83.13,
       "learning_rate": 9.329999999999999e-05,
-      "loss": 0.1003,
+      "loss": 0.1054,
       "step": 4572
     },
     {
       "epoch": 83.15,
       "learning_rate": 9.32e-05,
-      "loss": 0.1146,
+      "loss": 0.099,
       "step": 4573
     },
     {
       "epoch": 83.16,
       "learning_rate": 9.31e-05,
-      "loss": 0.1108,
+      "loss": 0.086,
       "step": 4574
     },
     {
       "epoch": 83.18,
       "learning_rate": 9.3e-05,
-      "loss": 0.0996,
+      "loss": 0.0973,
       "step": 4575
     },
     {
       "epoch": 83.2,
       "learning_rate": 9.29e-05,
-      "loss": 0.0771,
+      "loss": 0.0649,
       "step": 4576
     },
     {
       "epoch": 83.22,
       "learning_rate": 9.279999999999999e-05,
-      "loss": 0.0877,
+      "loss": 0.0867,
       "step": 4577
     },
     {
       "epoch": 83.24,
       "learning_rate": 9.27e-05,
-      "loss": 0.1019,
+      "loss": 0.1023,
       "step": 4578
     },
     {
       "epoch": 83.25,
       "learning_rate": 9.26e-05,
-      "loss": 0.0816,
+      "loss": 0.0783,
       "step": 4579
     },
     {
       "epoch": 83.27,
       "learning_rate": 9.25e-05,
-      "loss": 0.1176,
+      "loss": 0.0922,
       "step": 4580
     },
     {
       "epoch": 83.29,
       "learning_rate": 9.24e-05,
-      "loss": 0.0804,
+      "loss": 0.0953,
       "step": 4581
     },
     {
       "epoch": 83.31,
       "learning_rate": 9.23e-05,
-      "loss": 0.107,
+      "loss": 0.1248,
       "step": 4582
     },
     {
       "epoch": 83.33,
       "learning_rate": 9.22e-05,
-      "loss": 0.1105,
+      "loss": 0.0835,
       "step": 4583
     },
     {
       "epoch": 83.35,
       "learning_rate": 9.21e-05,
-      "loss": 0.0666,
+      "loss": 0.0732,
       "step": 4584
     },
     {
       "epoch": 83.36,
       "learning_rate": 9.2e-05,
-      "loss": 0.0683,
+      "loss": 0.0944,
       "step": 4585
     },
     {
       "epoch": 83.38,
       "learning_rate": 9.19e-05,
-      "loss": 0.1243,
+      "loss": 0.1142,
       "step": 4586
     },
     {
       "epoch": 83.4,
       "learning_rate": 9.180000000000001e-05,
-      "loss": 0.1194,
+      "loss": 0.1097,
       "step": 4587
     },
     {
       "epoch": 83.42,
       "learning_rate": 9.17e-05,
-      "loss": 0.0915,
+      "loss": 0.1023,
       "step": 4588
     },
     {
       "epoch": 83.44,
       "learning_rate": 9.16e-05,
-      "loss": 0.0964,
+      "loss": 0.0705,
       "step": 4589
     },
     {
       "epoch": 83.45,
       "learning_rate": 9.15e-05,
-      "loss": 0.1074,
+      "loss": 0.0802,
       "step": 4590
     },
     {
       "epoch": 83.47,
       "learning_rate": 9.14e-05,
-      "loss": 0.0974,
+      "loss": 0.0942,
       "step": 4591
     },
     {
       "epoch": 83.49,
       "learning_rate": 9.130000000000001e-05,
-      "loss": 0.0865,
+      "loss": 0.0866,
       "step": 4592
     },
     {
       "epoch": 83.51,
       "learning_rate": 9.120000000000001e-05,
-      "loss": 0.101,
+      "loss": 0.0794,
       "step": 4593
     },
     {
       "epoch": 83.53,
       "learning_rate": 9.11e-05,
-      "loss": 0.1202,
+      "loss": 0.0941,
       "step": 4594
     },
     {
       "epoch": 83.55,
       "learning_rate": 9.1e-05,
-      "loss": 0.1124,
+      "loss": 0.1138,
       "step": 4595
     },
     {
       "epoch": 83.56,
       "learning_rate": 9.09e-05,
-      "loss": 0.1194,
+      "loss": 0.0955,
       "step": 4596
     },
     {
       "epoch": 83.58,
       "learning_rate": 9.080000000000001e-05,
-      "loss": 0.1242,
+      "loss": 0.1216,
       "step": 4597
     },
     {
       "epoch": 83.6,
       "learning_rate": 9.070000000000001e-05,
-      "loss": 0.0939,
+      "loss": 0.0951,
       "step": 4598
     },
     {
       "epoch": 83.62,
       "learning_rate": 9.06e-05,
-      "loss": 0.0855,
+      "loss": 0.0766,
       "step": 4599
     },
     {
       "epoch": 83.64,
       "learning_rate": 9.05e-05,
-      "loss": 0.0675,
+      "loss": 0.0545,
       "step": 4600
     },
     {
       "epoch": 83.65,
       "learning_rate": 9.04e-05,
-      "loss": 0.1224,
+      "loss": 0.0896,
       "step": 4601
     },
     {
       "epoch": 83.67,
       "learning_rate": 9.030000000000001e-05,
-      "loss": 0.1069,
+      "loss": 0.1034,
       "step": 4602
     },
     {
       "epoch": 83.69,
       "learning_rate": 9.020000000000001e-05,
-      "loss": 0.0999,
+      "loss": 0.0971,
       "step": 4603
     },
     {
       "epoch": 83.71,
       "learning_rate": 9.010000000000001e-05,
-      "loss": 0.125,
+      "loss": 0.1331,
       "step": 4604
     },
     {
       "epoch": 83.73,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 0.0888,
+      "loss": 0.1087,
       "step": 4605
     },
     {
       "epoch": 83.75,
       "learning_rate": 8.989999999999999e-05,
-      "loss": 0.12,
+      "loss": 0.1143,
       "step": 4606
     },
     {
       "epoch": 83.76,
       "learning_rate": 8.98e-05,
-      "loss": 0.1266,
+      "loss": 0.0985,
       "step": 4607
     },
     {
       "epoch": 83.78,
       "learning_rate": 8.97e-05,
-      "loss": 0.122,
+      "loss": 0.0925,
       "step": 4608
     },
     {
       "epoch": 83.8,
       "learning_rate": 8.96e-05,
-      "loss": 0.0732,
+      "loss": 0.0592,
       "step": 4609
     },
     {
       "epoch": 83.82,
       "learning_rate": 8.95e-05,
-      "loss": 0.1127,
+      "loss": 0.0978,
       "step": 4610
     },
     {
       "epoch": 83.84,
       "learning_rate": 8.939999999999999e-05,
-      "loss": 0.0799,
+      "loss": 0.0726,
       "step": 4611
     },
     {
       "epoch": 83.85,
       "learning_rate": 8.93e-05,
-      "loss": 0.0989,
+      "loss": 0.0839,
       "step": 4612
     },
     {
       "epoch": 83.87,
       "learning_rate": 8.92e-05,
-      "loss": 0.1102,
+      "loss": 0.1023,
       "step": 4613
     },
     {
       "epoch": 83.89,
       "learning_rate": 8.91e-05,
-      "loss": 0.106,
+      "loss": 0.1012,
       "step": 4614
     },
     {
       "epoch": 83.91,
       "learning_rate": 8.9e-05,
-      "loss": 0.0623,
+      "loss": 0.0622,
       "step": 4615
     },
     {
       "epoch": 83.93,
       "learning_rate": 8.89e-05,
-      "loss": 0.1152,
+      "loss": 0.1033,
       "step": 4616
     },
     {
       "epoch": 83.95,
       "learning_rate": 8.88e-05,
-      "loss": 0.0677,
+      "loss": 0.0597,
       "step": 4617
     },
     {
       "epoch": 83.96,
       "learning_rate": 8.87e-05,
-      "loss": 0.0497,
+      "loss": 0.0517,
       "step": 4618
     },
     {
       "epoch": 83.98,
       "learning_rate": 8.86e-05,
-      "loss": 0.0735,
+      "loss": 0.06,
       "step": 4619
     },
     {
       "epoch": 84.0,
       "learning_rate": 8.85e-05,
-      "loss": 0.0504,
+      "loss": 0.1204,
       "step": 4620
     },
     {
       "epoch": 84.02,
       "learning_rate": 8.840000000000001e-05,
-      "loss": 0.1039,
+      "loss": 0.1065,
       "step": 4621
     },
     {
       "epoch": 84.04,
       "learning_rate": 8.83e-05,
-      "loss": 0.0756,
+      "loss": 0.0965,
       "step": 4622
     },
     {
       "epoch": 84.05,
       "learning_rate": 8.82e-05,
-      "loss": 0.092,
+      "loss": 0.091,
       "step": 4623
     },
     {
       "epoch": 84.07,
       "learning_rate": 8.81e-05,
-      "loss": 0.1118,
+      "loss": 0.0857,
       "step": 4624
     },
     {
       "epoch": 84.09,
       "learning_rate": 8.8e-05,
-      "loss": 0.1172,
+      "loss": 0.0949,
       "step": 4625
     },
     {
       "epoch": 84.11,
       "learning_rate": 8.790000000000001e-05,
-      "loss": 0.0623,
+      "loss": 0.0593,
       "step": 4626
     },
     {
       "epoch": 84.13,
       "learning_rate": 8.78e-05,
-      "loss": 0.1142,
+      "loss": 0.1098,
       "step": 4627
     },
     {
       "epoch": 84.15,
       "learning_rate": 8.77e-05,
-      "loss": 0.0821,
+      "loss": 0.0664,
       "step": 4628
     },
     {
       "epoch": 84.16,
       "learning_rate": 8.76e-05,
-      "loss": 0.0731,
+      "loss": 0.0665,
       "step": 4629
     },
     {
       "epoch": 84.18,
       "learning_rate": 8.75e-05,
-      "loss": 0.1319,
+      "loss": 0.117,
       "step": 4630
     },
     {
       "epoch": 84.2,
       "learning_rate": 8.740000000000001e-05,
-      "loss": 0.0962,
+      "loss": 0.0828,
       "step": 4631
     },
     {
       "epoch": 84.22,
       "learning_rate": 8.730000000000001e-05,
-      "loss": 0.0655,
+      "loss": 0.0519,
       "step": 4632
     },
     {
       "epoch": 84.24,
       "learning_rate": 8.72e-05,
-      "loss": 0.0661,
+      "loss": 0.0715,
       "step": 4633
     },
     {
       "epoch": 84.25,
       "learning_rate": 8.71e-05,
-      "loss": 0.1111,
+      "loss": 0.1194,
       "step": 4634
     },
     {
       "epoch": 84.27,
       "learning_rate": 8.7e-05,
-      "loss": 0.0651,
+      "loss": 0.0585,
       "step": 4635
     },
     {
       "epoch": 84.29,
       "learning_rate": 8.690000000000001e-05,
-      "loss": 0.0876,
+      "loss": 0.0705,
       "step": 4636
     },
     {
       "epoch": 84.31,
       "learning_rate": 8.680000000000001e-05,
-      "loss": 0.0848,
+      "loss": 0.0756,
       "step": 4637
     },
     {
       "epoch": 84.33,
       "learning_rate": 8.67e-05,
-      "loss": 0.0976,
+      "loss": 0.0765,
       "step": 4638
     },
     {
       "epoch": 84.35,
       "learning_rate": 8.66e-05,
-      "loss": 0.0755,
+      "loss": 0.0529,
       "step": 4639
     },
     {
       "epoch": 84.36,
       "learning_rate": 8.65e-05,
-      "loss": 0.1003,
+      "loss": 0.0748,
       "step": 4640
     },
     {
       "epoch": 84.38,
       "learning_rate": 8.640000000000001e-05,
-      "loss": 0.0886,
+      "loss": 0.081,
       "step": 4641
     },
     {
       "epoch": 84.4,
       "learning_rate": 8.63e-05,
-      "loss": 0.0972,
+      "loss": 0.0899,
       "step": 4642
     },
     {
       "epoch": 84.42,
       "learning_rate": 8.62e-05,
-      "loss": 0.0773,
+      "loss": 0.0699,
       "step": 4643
     },
     {
       "epoch": 84.44,
       "learning_rate": 8.609999999999999e-05,
-      "loss": 0.0758,
+      "loss": 0.0664,
       "step": 4644
     },
     {
       "epoch": 84.45,
       "learning_rate": 8.599999999999999e-05,
-      "loss": 0.0697,
+      "loss": 0.0889,
       "step": 4645
     },
     {
       "epoch": 84.47,
       "learning_rate": 8.59e-05,
-      "loss": 0.1131,
+      "loss": 0.0957,
       "step": 4646
     },
     {
       "epoch": 84.49,
       "learning_rate": 8.58e-05,
-      "loss": 0.0775,
+      "loss": 0.0619,
       "step": 4647
     },
     {
       "epoch": 84.51,
       "learning_rate": 8.57e-05,
-      "loss": 0.0783,
+      "loss": 0.0592,
       "step": 4648
     },
     {
       "epoch": 84.53,
       "learning_rate": 8.56e-05,
-      "loss": 0.1171,
+      "loss": 0.091,
       "step": 4649
     },
     {
       "epoch": 84.55,
       "learning_rate": 8.55e-05,
-      "loss": 0.0949,
+      "loss": 0.0846,
       "step": 4650
     },
     {
       "epoch": 84.56,
       "learning_rate": 8.54e-05,
-      "loss": 0.0625,
+      "loss": 0.0543,
       "step": 4651
     },
     {
       "epoch": 84.58,
       "learning_rate": 8.53e-05,
-      "loss": 0.0824,
+      "loss": 0.0676,
       "step": 4652
     },
     {
       "epoch": 84.6,
       "learning_rate": 8.52e-05,
-      "loss": 0.0631,
+      "loss": 0.0567,
       "step": 4653
     },
     {
       "epoch": 84.62,
       "learning_rate": 8.51e-05,
-      "loss": 0.0756,
+      "loss": 0.0792,
       "step": 4654
     },
     {
       "epoch": 84.64,
       "learning_rate": 8.5e-05,
-      "loss": 0.0617,
+      "loss": 0.0529,
       "step": 4655
     },
     {
       "epoch": 84.65,
       "learning_rate": 8.49e-05,
-      "loss": 0.096,
+      "loss": 0.0711,
       "step": 4656
     },
     {
       "epoch": 84.67,
       "learning_rate": 8.48e-05,
-      "loss": 0.0915,
+      "loss": 0.0643,
       "step": 4657
     },
     {
       "epoch": 84.69,
       "learning_rate": 8.47e-05,
-      "loss": 0.0778,
+      "loss": 0.0803,
       "step": 4658
     },
     {
       "epoch": 84.71,
       "learning_rate": 8.46e-05,
-      "loss": 0.108,
+      "loss": 0.092,
       "step": 4659
     },
     {
       "epoch": 84.73,
       "learning_rate": 8.450000000000001e-05,
-      "loss": 0.0876,
+      "loss": 0.087,
       "step": 4660
     },
     {
       "epoch": 84.75,
       "learning_rate": 8.44e-05,
-      "loss": 0.0932,
+      "loss": 0.0858,
       "step": 4661
     },
     {
       "epoch": 84.76,
       "learning_rate": 8.43e-05,
-      "loss": 0.0707,
+      "loss": 0.0694,
       "step": 4662
     },
     {
       "epoch": 84.78,
       "learning_rate": 8.42e-05,
-      "loss": 0.1253,
+      "loss": 0.0961,
       "step": 4663
     },
     {
       "epoch": 84.8,
       "learning_rate": 8.41e-05,
-      "loss": 0.0606,
+      "loss": 0.0637,
       "step": 4664
     },
     {
       "epoch": 84.82,
       "learning_rate": 8.400000000000001e-05,
-      "loss": 0.1104,
+      "loss": 0.1057,
       "step": 4665
     },
     {
       "epoch": 84.84,
       "learning_rate": 8.39e-05,
-      "loss": 0.0734,
+      "loss": 0.0616,
       "step": 4666
     },
     {
       "epoch": 84.85,
       "learning_rate": 8.38e-05,
-      "loss": 0.1203,
+      "loss": 0.106,
       "step": 4667
     },
     {
       "epoch": 84.87,
       "learning_rate": 8.37e-05,
-      "loss": 0.0614,
+      "loss": 0.0524,
       "step": 4668
     },
     {
       "epoch": 84.89,
       "learning_rate": 8.36e-05,
-      "loss": 0.0997,
+      "loss": 0.1049,
       "step": 4669
     },
     {
       "epoch": 84.91,
       "learning_rate": 8.350000000000001e-05,
-      "loss": 0.0565,
+      "loss": 0.0569,
       "step": 4670
     },
     {
       "epoch": 84.93,
       "learning_rate": 8.340000000000001e-05,
-      "loss": 0.0813,
+      "loss": 0.0795,
       "step": 4671
     },
     {
       "epoch": 84.95,
       "learning_rate": 8.33e-05,
-      "loss": 0.1054,
+      "loss": 0.0928,
       "step": 4672
     },
     {
       "epoch": 84.96,
       "learning_rate": 8.32e-05,
-      "loss": 0.1141,
+      "loss": 0.0985,
       "step": 4673
     },
     {
       "epoch": 84.98,
       "learning_rate": 8.31e-05,
-      "loss": 0.1009,
+      "loss": 0.0708,
       "step": 4674
     },
     {
       "epoch": 85.0,
       "learning_rate": 8.300000000000001e-05,
-      "loss": 0.0707,
+      "loss": 0.0461,
       "step": 4675
     },
     {
       "epoch": 85.02,
       "learning_rate": 8.290000000000001e-05,
-      "loss": 0.107,
+      "loss": 0.0993,
       "step": 4676
     },
     {
       "epoch": 85.04,
       "learning_rate": 8.280000000000001e-05,
-      "loss": 0.096,
+      "loss": 0.0699,
       "step": 4677
     },
     {
       "epoch": 85.05,
       "learning_rate": 8.269999999999999e-05,
-      "loss": 0.085,
+      "loss": 0.0962,
       "step": 4678
     },
     {
       "epoch": 85.07,
       "learning_rate": 8.26e-05,
-      "loss": 0.1076,
+      "loss": 0.0898,
       "step": 4679
     },
     {
       "epoch": 85.09,
       "learning_rate": 8.25e-05,
-      "loss": 0.0966,
+      "loss": 0.1117,
       "step": 4680
     },
     {
       "epoch": 85.11,
       "learning_rate": 8.24e-05,
-      "loss": 0.0737,
+      "loss": 0.0749,
       "step": 4681
     },
     {
       "epoch": 85.13,
       "learning_rate": 8.23e-05,
-      "loss": 0.0766,
+      "loss": 0.0765,
       "step": 4682
     },
     {
       "epoch": 85.15,
       "learning_rate": 8.219999999999999e-05,
-      "loss": 0.0969,
+      "loss": 0.0754,
       "step": 4683
     },
     {
       "epoch": 85.16,
       "learning_rate": 8.21e-05,
-      "loss": 0.1056,
+      "loss": 0.0646,
       "step": 4684
     },
     {
       "epoch": 85.18,
       "learning_rate": 8.2e-05,
-      "loss": 0.0636,
+      "loss": 0.0519,
       "step": 4685
     },
     {
       "epoch": 85.2,
       "learning_rate": 8.19e-05,
-      "loss": 0.0994,
+      "loss": 0.0724,
       "step": 4686
     },
     {
       "epoch": 85.22,
       "learning_rate": 8.18e-05,
-      "loss": 0.0933,
+      "loss": 0.087,
       "step": 4687
     },
     {
       "epoch": 85.24,
       "learning_rate": 8.17e-05,
-      "loss": 0.0728,
+      "loss": 0.0636,
       "step": 4688
     },
     {
       "epoch": 85.25,
       "learning_rate": 8.16e-05,
-      "loss": 0.1017,
+      "loss": 0.0893,
       "step": 4689
     },
     {
       "epoch": 85.27,
       "learning_rate": 8.15e-05,
-      "loss": 0.1003,
+      "loss": 0.0983,
       "step": 4690
     },
     {
       "epoch": 85.29,
       "learning_rate": 8.14e-05,
-      "loss": 0.1038,
+      "loss": 0.1197,
       "step": 4691
     },
     {
       "epoch": 85.31,
       "learning_rate": 8.13e-05,
-      "loss": 0.0775,
+      "loss": 0.0929,
       "step": 4692
     },
     {
       "epoch": 85.33,
       "learning_rate": 8.12e-05,
-      "loss": 0.0642,
+      "loss": 0.0729,
       "step": 4693
     },
     {
       "epoch": 85.35,
       "learning_rate": 8.11e-05,
-      "loss": 0.0686,
+      "loss": 0.0649,
       "step": 4694
     },
     {
       "epoch": 85.36,
       "learning_rate": 8.1e-05,
-      "loss": 0.1342,
+      "loss": 0.1143,
       "step": 4695
     },
     {
       "epoch": 85.38,
       "learning_rate": 8.09e-05,
-      "loss": 0.1066,
+      "loss": 0.1051,
       "step": 4696
     },
     {
       "epoch": 85.4,
       "learning_rate": 8.08e-05,
-      "loss": 0.1628,
+      "loss": 0.1273,
       "step": 4697
     },
     {
       "epoch": 85.42,
       "learning_rate": 8.07e-05,
-      "loss": 0.0712,
+      "loss": 0.0624,
       "step": 4698
     },
     {
       "epoch": 85.44,
       "learning_rate": 8.060000000000001e-05,
-      "loss": 0.0829,
+      "loss": 0.0997,
       "step": 4699
     },
     {
       "epoch": 85.45,
       "learning_rate": 8.05e-05,
-      "loss": 0.0756,
+      "loss": 0.069,
       "step": 4700
     },
     {
       "epoch": 85.47,
       "learning_rate": 8.04e-05,
-      "loss": 0.1223,
+      "loss": 0.1079,
       "step": 4701
     },
     {
       "epoch": 85.49,
       "learning_rate": 8.03e-05,
-      "loss": 0.0672,
+      "loss": 0.0603,
       "step": 4702
     },
     {
       "epoch": 85.51,
       "learning_rate": 8.02e-05,
-      "loss": 0.0723,
+      "loss": 0.0806,
       "step": 4703
     },
     {
       "epoch": 85.53,
       "learning_rate": 8.010000000000001e-05,
-      "loss": 0.0919,
+      "loss": 0.0935,
       "step": 4704
     },
     {
       "epoch": 85.55,
       "learning_rate": 8e-05,
-      "loss": 0.0747,
+      "loss": 0.0716,
       "step": 4705
     },
     {
       "epoch": 85.56,
       "learning_rate": 7.99e-05,
-      "loss": 0.1096,
+      "loss": 0.1151,
       "step": 4706
     },
     {
       "epoch": 85.58,
       "learning_rate": 7.98e-05,
-      "loss": 0.1495,
+      "loss": 0.1244,
       "step": 4707
     },
     {
       "epoch": 85.6,
       "learning_rate": 7.97e-05,
-      "loss": 0.055,
+      "loss": 0.0637,
       "step": 4708
     },
     {
       "epoch": 85.62,
       "learning_rate": 7.960000000000001e-05,
-      "loss": 0.1202,
+      "loss": 0.0989,
       "step": 4709
     },
     {
       "epoch": 85.64,
       "learning_rate": 7.950000000000001e-05,
-      "loss": 0.0748,
+      "loss": 0.0876,
       "step": 4710
     },
     {
       "epoch": 85.65,
       "learning_rate": 7.94e-05,
-      "loss": 0.0832,
+      "loss": 0.08,
       "step": 4711
     },
     {
       "epoch": 85.67,
       "learning_rate": 7.93e-05,
-      "loss": 0.0904,
+      "loss": 0.0653,
       "step": 4712
     },
     {
       "epoch": 85.69,
       "learning_rate": 7.920000000000001e-05,
-      "loss": 0.1235,
+      "loss": 0.1009,
       "step": 4713
     },
     {
       "epoch": 85.71,
       "learning_rate": 7.910000000000001e-05,
-      "loss": 0.1118,
+      "loss": 0.0986,
       "step": 4714
     },
     {
       "epoch": 85.73,
       "learning_rate": 7.9e-05,
-      "loss": 0.0836,
+      "loss": 0.0954,
       "step": 4715
     },
     {
       "epoch": 85.75,
       "learning_rate": 7.89e-05,
-      "loss": 0.0546,
+      "loss": 0.0828,
       "step": 4716
     },
     {
       "epoch": 85.76,
       "learning_rate": 7.879999999999999e-05,
-      "loss": 0.0782,
+      "loss": 0.0941,
       "step": 4717
     },
     {
       "epoch": 85.78,
       "learning_rate": 7.87e-05,
-      "loss": 0.0584,
+      "loss": 0.0574,
       "step": 4718
     },
     {
       "epoch": 85.8,
       "learning_rate": 7.86e-05,
-      "loss": 0.0989,
+      "loss": 0.0804,
       "step": 4719
     },
     {
       "epoch": 85.82,
       "learning_rate": 7.85e-05,
-      "loss": 0.1049,
+      "loss": 0.0999,
       "step": 4720
     },
     {
       "epoch": 85.84,
       "learning_rate": 7.84e-05,
-      "loss": 0.0853,
+      "loss": 0.0759,
       "step": 4721
     },
     {
       "epoch": 85.85,
       "learning_rate": 7.829999999999999e-05,
-      "loss": 0.0929,
+      "loss": 0.0878,
       "step": 4722
     },
     {
       "epoch": 85.87,
       "learning_rate": 7.82e-05,
-      "loss": 0.1007,
+      "loss": 0.0912,
       "step": 4723
     },
     {
       "epoch": 85.89,
       "learning_rate": 7.81e-05,
-      "loss": 0.0895,
+      "loss": 0.0945,
       "step": 4724
     },
     {
       "epoch": 85.91,
       "learning_rate": 7.8e-05,
-      "loss": 0.0535,
+      "loss": 0.0751,
       "step": 4725
     },
     {
       "epoch": 85.93,
       "learning_rate": 7.79e-05,
-      "loss": 0.1232,
+      "loss": 0.0934,
       "step": 4726
     },
     {
       "epoch": 85.95,
       "learning_rate": 7.78e-05,
-      "loss": 0.112,
+      "loss": 0.0897,
       "step": 4727
     },
     {
       "epoch": 85.96,
       "learning_rate": 7.77e-05,
-      "loss": 0.0945,
+      "loss": 0.0792,
       "step": 4728
     },
     {
       "epoch": 85.98,
       "learning_rate": 7.76e-05,
-      "loss": 0.0871,
+      "loss": 0.0878,
       "step": 4729
     },
     {
       "epoch": 86.0,
       "learning_rate": 7.75e-05,
-      "loss": 0.0879,
+      "loss": 0.0545,
       "step": 4730
     },
     {
       "epoch": 86.02,
       "learning_rate": 7.74e-05,
-      "loss": 0.1132,
+      "loss": 0.0956,
       "step": 4731
     },
     {
       "epoch": 86.04,
       "learning_rate": 7.73e-05,
-      "loss": 0.0926,
+      "loss": 0.077,
       "step": 4732
     },
     {
       "epoch": 86.05,
       "learning_rate": 7.72e-05,
-      "loss": 0.1077,
+      "loss": 0.1181,
       "step": 4733
     },
     {
       "epoch": 86.07,
       "learning_rate": 7.71e-05,
-      "loss": 0.1032,
+      "loss": 0.0803,
       "step": 4734
     },
     {
       "epoch": 86.09,
       "learning_rate": 7.7e-05,
-      "loss": 0.1127,
+      "loss": 0.1001,
       "step": 4735
     },
     {
       "epoch": 86.11,
       "learning_rate": 7.69e-05,
-      "loss": 0.0553,
+      "loss": 0.0371,
       "step": 4736
     },
     {
       "epoch": 86.13,
       "learning_rate": 7.68e-05,
-      "loss": 0.0908,
+      "loss": 0.0967,
       "step": 4737
     },
     {
       "epoch": 86.15,
       "learning_rate": 7.670000000000001e-05,
-      "loss": 0.0963,
+      "loss": 0.0974,
       "step": 4738
     },
     {
       "epoch": 86.16,
       "learning_rate": 7.66e-05,
-      "loss": 0.0765,
+      "loss": 0.075,
       "step": 4739
     },
     {
       "epoch": 86.18,
       "learning_rate": 7.65e-05,
-      "loss": 0.0949,
+      "loss": 0.0801,
       "step": 4740
     },
     {
       "epoch": 86.2,
       "learning_rate": 7.64e-05,
-      "loss": 0.0986,
+      "loss": 0.0819,
       "step": 4741
     },
     {
       "epoch": 86.22,
       "learning_rate": 7.630000000000001e-05,
-      "loss": 0.1081,
+      "loss": 0.0845,
       "step": 4742
     },
     {
       "epoch": 86.24,
       "learning_rate": 7.620000000000001e-05,
-      "loss": 0.0822,
+      "loss": 0.0846,
       "step": 4743
     },
     {
       "epoch": 86.25,
       "learning_rate": 7.61e-05,
-      "loss": 0.1106,
+      "loss": 0.0834,
       "step": 4744
     },
     {
       "epoch": 86.27,
       "learning_rate": 7.6e-05,
-      "loss": 0.1054,
+      "loss": 0.1114,
       "step": 4745
     },
     {
       "epoch": 86.29,
       "learning_rate": 7.59e-05,
-      "loss": 0.1003,
+      "loss": 0.0621,
       "step": 4746
     },
     {
       "epoch": 86.31,
       "learning_rate": 7.580000000000001e-05,
-      "loss": 0.065,
+      "loss": 0.0538,
       "step": 4747
     },
     {
       "epoch": 86.33,
       "learning_rate": 7.570000000000001e-05,
-      "loss": 0.0568,
+      "loss": 0.0691,
       "step": 4748
     },
     {
       "epoch": 86.35,
       "learning_rate": 7.560000000000001e-05,
-      "loss": 0.0772,
+      "loss": 0.0739,
       "step": 4749
     },
     {
       "epoch": 86.36,
       "learning_rate": 7.55e-05,
-      "loss": 0.0897,
+      "loss": 0.0954,
       "step": 4750
     },
     {
       "epoch": 86.38,
       "learning_rate": 7.539999999999999e-05,
-      "loss": 0.1171,
+      "loss": 0.0936,
       "step": 4751
     },
     {
       "epoch": 86.4,
       "learning_rate": 7.53e-05,
-      "loss": 0.0831,
+      "loss": 0.0771,
       "step": 4752
     },
     {
       "epoch": 86.42,
       "learning_rate": 7.52e-05,
-      "loss": 0.0623,
+      "loss": 0.0871,
       "step": 4753
     },
     {
       "epoch": 86.44,
       "learning_rate": 7.51e-05,
-      "loss": 0.1036,
+      "loss": 0.0621,
       "step": 4754
     },
     {
       "epoch": 86.45,
       "learning_rate": 7.5e-05,
-      "loss": 0.089,
+      "loss": 0.0724,
       "step": 4755
     },
     {
       "epoch": 86.47,
       "learning_rate": 7.489999999999999e-05,
-      "loss": 0.0871,
+      "loss": 0.074,
       "step": 4756
     },
     {
       "epoch": 86.49,
       "learning_rate": 7.48e-05,
-      "loss": 0.1193,
+      "loss": 0.1034,
       "step": 4757
     },
     {
       "epoch": 86.51,
       "learning_rate": 7.47e-05,
-      "loss": 0.0771,
+      "loss": 0.0588,
       "step": 4758
     },
     {
       "epoch": 86.53,
       "learning_rate": 7.46e-05,
-      "loss": 0.0946,
+      "loss": 0.085,
       "step": 4759
     },
     {
       "epoch": 86.55,
       "learning_rate": 7.45e-05,
-      "loss": 0.0829,
+      "loss": 0.0769,
       "step": 4760
     },
     {
       "epoch": 86.56,
       "learning_rate": 7.439999999999999e-05,
-      "loss": 0.0952,
+      "loss": 0.1143,
       "step": 4761
     },
     {
       "epoch": 86.58,
       "learning_rate": 7.43e-05,
-      "loss": 0.116,
+      "loss": 0.1012,
       "step": 4762
     },
     {
       "epoch": 86.6,
       "learning_rate": 7.42e-05,
-      "loss": 0.0795,
+      "loss": 0.0774,
       "step": 4763
     },
     {
       "epoch": 86.62,
       "learning_rate": 7.41e-05,
-      "loss": 0.0887,
+      "loss": 0.063,
       "step": 4764
     },
     {
       "epoch": 86.64,
       "learning_rate": 7.4e-05,
-      "loss": 0.082,
+      "loss": 0.0975,
       "step": 4765
     },
     {
       "epoch": 86.65,
       "learning_rate": 7.39e-05,
-      "loss": 0.0821,
+      "loss": 0.0576,
       "step": 4766
     },
     {
       "epoch": 86.67,
       "learning_rate": 7.38e-05,
-      "loss": 0.0713,
+      "loss": 0.0805,
       "step": 4767
     },
     {
       "epoch": 86.69,
       "learning_rate": 7.37e-05,
-      "loss": 0.0642,
+      "loss": 0.0725,
       "step": 4768
     },
     {
       "epoch": 86.71,
       "learning_rate": 7.36e-05,
-      "loss": 0.0965,
+      "loss": 0.0807,
       "step": 4769
     },
     {
       "epoch": 86.73,
       "learning_rate": 7.35e-05,
-      "loss": 0.0847,
+      "loss": 0.076,
       "step": 4770
     },
     {
       "epoch": 86.75,
       "learning_rate": 7.340000000000001e-05,
-      "loss": 0.1121,
+      "loss": 0.0849,
       "step": 4771
     },
     {
       "epoch": 86.76,
       "learning_rate": 7.33e-05,
-      "loss": 0.0831,
+      "loss": 0.0618,
       "step": 4772
     },
     {
       "epoch": 86.78,
       "learning_rate": 7.32e-05,
-      "loss": 0.0905,
+      "loss": 0.0756,
       "step": 4773
     },
     {
       "epoch": 86.8,
       "learning_rate": 7.31e-05,
-      "loss": 0.067,
+      "loss": 0.0643,
       "step": 4774
     },
     {
       "epoch": 86.82,
       "learning_rate": 7.3e-05,
-      "loss": 0.1031,
+      "loss": 0.0953,
       "step": 4775
     },
     {
       "epoch": 86.84,
       "learning_rate": 7.290000000000001e-05,
-      "loss": 0.0905,
+      "loss": 0.0753,
       "step": 4776
     },
     {
       "epoch": 86.85,
       "learning_rate": 7.280000000000001e-05,
-      "loss": 0.1046,
+      "loss": 0.0927,
       "step": 4777
     },
     {
       "epoch": 86.87,
       "learning_rate": 7.27e-05,
-      "loss": 0.0777,
+      "loss": 0.0692,
       "step": 4778
     },
     {
       "epoch": 86.89,
       "learning_rate": 7.26e-05,
-      "loss": 0.1231,
+      "loss": 0.0923,
       "step": 4779
     },
     {
       "epoch": 86.91,
       "learning_rate": 7.25e-05,
-      "loss": 0.0567,
+      "loss": 0.071,
       "step": 4780
     },
     {
       "epoch": 86.93,
       "learning_rate": 7.240000000000001e-05,
-      "loss": 0.0849,
+      "loss": 0.0675,
       "step": 4781
     },
     {
       "epoch": 86.95,
       "learning_rate": 7.230000000000001e-05,
-      "loss": 0.0775,
+      "loss": 0.0632,
       "step": 4782
     },
     {
       "epoch": 86.96,
       "learning_rate": 7.22e-05,
-      "loss": 0.0953,
+      "loss": 0.0719,
       "step": 4783
     },
     {
       "epoch": 86.98,
       "learning_rate": 7.21e-05,
-      "loss": 0.0631,
+      "loss": 0.0701,
       "step": 4784
     },
     {
       "epoch": 87.0,
       "learning_rate": 7.2e-05,
-      "loss": 0.0755,
+      "loss": 0.0951,
       "step": 4785
     },
     {
       "epoch": 87.02,
       "learning_rate": 7.190000000000001e-05,
-      "loss": 0.1211,
+      "loss": 0.0952,
       "step": 4786
     },
     {
       "epoch": 87.04,
       "learning_rate": 7.180000000000001e-05,
-      "loss": 0.0828,
+      "loss": 0.0622,
       "step": 4787
     },
     {
       "epoch": 87.05,
       "learning_rate": 7.170000000000001e-05,
-      "loss": 0.1272,
+      "loss": 0.1287,
       "step": 4788
     },
     {
       "epoch": 87.07,
       "learning_rate": 7.159999999999999e-05,
-      "loss": 0.126,
+      "loss": 0.1195,
       "step": 4789
     },
     {
       "epoch": 87.09,
       "learning_rate": 7.149999999999999e-05,
-      "loss": 0.0711,
+      "loss": 0.0715,
       "step": 4790
     },
     {
       "epoch": 87.11,
       "learning_rate": 7.14e-05,
-      "loss": 0.0651,
+      "loss": 0.0512,
       "step": 4791
     },
     {
       "epoch": 87.13,
       "learning_rate": 7.13e-05,
-      "loss": 0.1224,
+      "loss": 0.1065,
       "step": 4792
     },
     {
       "epoch": 87.15,
       "learning_rate": 7.12e-05,
-      "loss": 0.0783,
+      "loss": 0.0672,
       "step": 4793
     },
     {
       "epoch": 87.16,
       "learning_rate": 7.11e-05,
-      "loss": 0.0863,
+      "loss": 0.0726,
       "step": 4794
     },
     {
       "epoch": 87.18,
       "learning_rate": 7.099999999999999e-05,
-      "loss": 0.0864,
+      "loss": 0.0749,
       "step": 4795
     },
     {
       "epoch": 87.2,
       "learning_rate": 7.09e-05,
-      "loss": 0.0771,
+      "loss": 0.0829,
       "step": 4796
     },
     {
       "epoch": 87.22,
       "learning_rate": 7.08e-05,
-      "loss": 0.082,
+      "loss": 0.0797,
       "step": 4797
     },
     {
       "epoch": 87.24,
       "learning_rate": 7.07e-05,
-      "loss": 0.0902,
+      "loss": 0.0728,
       "step": 4798
     },
     {
       "epoch": 87.25,
       "learning_rate": 7.06e-05,
-      "loss": 0.0937,
+      "loss": 0.0884,
       "step": 4799
     },
     {
       "epoch": 87.27,
       "learning_rate": 7.049999999999999e-05,
-      "loss": 0.0828,
+      "loss": 0.0684,
       "step": 4800
     },
     {
       "epoch": 87.29,
       "learning_rate": 7.04e-05,
-      "loss": 0.0796,
+      "loss": 0.0678,
       "step": 4801
     },
     {
       "epoch": 87.31,
       "learning_rate": 7.03e-05,
-      "loss": 0.0913,
+      "loss": 0.0853,
       "step": 4802
     },
     {
       "epoch": 87.33,
       "learning_rate": 7.02e-05,
-      "loss": 0.0999,
+      "loss": 0.0711,
       "step": 4803
     },
     {
       "epoch": 87.35,
       "learning_rate": 7.01e-05,
-      "loss": 0.0508,
+      "loss": 0.0552,
       "step": 4804
     },
     {
       "epoch": 87.36,
       "learning_rate": 7.000000000000001e-05,
-      "loss": 0.0959,
+      "loss": 0.089,
       "step": 4805
     },
     {
       "epoch": 87.38,
       "learning_rate": 6.99e-05,
-      "loss": 0.0664,
+      "loss": 0.0687,
       "step": 4806
     },
     {
       "epoch": 87.4,
       "learning_rate": 6.98e-05,
-      "loss": 0.0947,
+      "loss": 0.0613,
       "step": 4807
     },
     {
       "epoch": 87.42,
       "learning_rate": 6.97e-05,
-      "loss": 0.0785,
+      "loss": 0.0876,
       "step": 4808
     },
     {
       "epoch": 87.44,
       "learning_rate": 6.96e-05,
-      "loss": 0.0673,
+      "loss": 0.0373,
       "step": 4809
     },
     {
       "epoch": 87.45,
       "learning_rate": 6.950000000000001e-05,
-      "loss": 0.0754,
+      "loss": 0.055,
       "step": 4810
     },
     {
       "epoch": 87.47,
       "learning_rate": 6.94e-05,
-      "loss": 0.1132,
+      "loss": 0.091,
       "step": 4811
     },
     {
       "epoch": 87.49,
       "learning_rate": 6.93e-05,
-      "loss": 0.0881,
+      "loss": 0.0572,
       "step": 4812
     },
     {
       "epoch": 87.51,
       "learning_rate": 6.92e-05,
-      "loss": 0.1001,
+      "loss": 0.0773,
       "step": 4813
     },
     {
       "epoch": 87.53,
       "learning_rate": 6.91e-05,
-      "loss": 0.099,
+      "loss": 0.0747,
       "step": 4814
     },
     {
       "epoch": 87.55,
       "learning_rate": 6.900000000000001e-05,
-      "loss": 0.0831,
+      "loss": 0.0662,
       "step": 4815
     },
     {
       "epoch": 87.56,
       "learning_rate": 6.890000000000001e-05,
-      "loss": 0.0571,
+      "loss": 0.0434,
       "step": 4816
     },
     {
       "epoch": 87.58,
       "learning_rate": 6.88e-05,
-      "loss": 0.0749,
+      "loss": 0.0822,
       "step": 4817
     },
     {
       "epoch": 87.6,
       "learning_rate": 6.87e-05,
-      "loss": 0.1042,
+      "loss": 0.0958,
       "step": 4818
     },
     {
       "epoch": 87.62,
       "learning_rate": 6.86e-05,
-      "loss": 0.1114,
+      "loss": 0.085,
       "step": 4819
     },
     {
       "epoch": 87.64,
       "learning_rate": 6.850000000000001e-05,
-      "loss": 0.0882,
+      "loss": 0.0716,
       "step": 4820
     },
     {
       "epoch": 87.65,
       "learning_rate": 6.840000000000001e-05,
-      "loss": 0.0797,
+      "loss": 0.0687,
       "step": 4821
     },
     {
       "epoch": 87.67,
       "learning_rate": 6.83e-05,
-      "loss": 0.0881,
+      "loss": 0.0848,
       "step": 4822
     },
     {
       "epoch": 87.69,
       "learning_rate": 6.82e-05,
-      "loss": 0.0713,
+      "loss": 0.0748,
       "step": 4823
     },
     {
       "epoch": 87.71,
       "learning_rate": 6.81e-05,
-      "loss": 0.0815,
+      "loss": 0.0942,
       "step": 4824
     },
     {
       "epoch": 87.73,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 0.0801,
+      "loss": 0.0797,
       "step": 4825
     },
     {
       "epoch": 87.75,
       "learning_rate": 6.79e-05,
-      "loss": 0.0817,
+      "loss": 0.0924,
       "step": 4826
     },
     {
       "epoch": 87.76,
       "learning_rate": 6.78e-05,
-      "loss": 0.0559,
+      "loss": 0.0455,
       "step": 4827
     },
     {
       "epoch": 87.78,
       "learning_rate": 6.769999999999999e-05,
-      "loss": 0.0918,
+      "loss": 0.0774,
       "step": 4828
     },
     {
       "epoch": 87.8,
       "learning_rate": 6.759999999999999e-05,
-      "loss": 0.0501,
+      "loss": 0.0797,
       "step": 4829
     },
     {
       "epoch": 87.82,
       "learning_rate": 6.75e-05,
-      "loss": 0.1115,
+      "loss": 0.0928,
       "step": 4830
     },
     {
       "epoch": 87.84,
       "learning_rate": 6.74e-05,
-      "loss": 0.1179,
+      "loss": 0.1014,
       "step": 4831
     },
     {
       "epoch": 87.85,
       "learning_rate": 6.73e-05,
-      "loss": 0.0799,
+      "loss": 0.1115,
       "step": 4832
     },
     {
       "epoch": 87.87,
       "learning_rate": 6.72e-05,
-      "loss": 0.0721,
+      "loss": 0.0768,
       "step": 4833
     },
     {
       "epoch": 87.89,
       "learning_rate": 6.71e-05,
-      "loss": 0.1036,
+      "loss": 0.0984,
       "step": 4834
     },
     {
@@ -29110,4021 +29110,4021 @@
     {
       "epoch": 87.93,
       "learning_rate": 6.69e-05,
-      "loss": 0.1164,
+      "loss": 0.0774,
       "step": 4836
     },
     {
       "epoch": 87.95,
       "learning_rate": 6.68e-05,
-      "loss": 0.0685,
+      "loss": 0.0674,
       "step": 4837
     },
     {
       "epoch": 87.96,
       "learning_rate": 6.67e-05,
-      "loss": 0.0663,
+      "loss": 0.0482,
       "step": 4838
     },
     {
       "epoch": 87.98,
       "learning_rate": 6.66e-05,
-      "loss": 0.12,
+      "loss": 0.089,
       "step": 4839
     },
     {
       "epoch": 88.0,
       "learning_rate": 6.65e-05,
-      "loss": 0.0378,
+      "loss": 0.0744,
       "step": 4840
     },
     {
       "epoch": 88.02,
       "learning_rate": 6.64e-05,
-      "loss": 0.0968,
+      "loss": 0.0963,
       "step": 4841
     },
     {
       "epoch": 88.04,
       "learning_rate": 6.63e-05,
-      "loss": 0.0896,
+      "loss": 0.0752,
       "step": 4842
     },
     {
       "epoch": 88.05,
       "learning_rate": 6.62e-05,
-      "loss": 0.0574,
+      "loss": 0.0479,
       "step": 4843
     },
     {
       "epoch": 88.07,
       "learning_rate": 6.610000000000001e-05,
-      "loss": 0.0791,
+      "loss": 0.0662,
       "step": 4844
     },
     {
       "epoch": 88.09,
       "learning_rate": 6.6e-05,
-      "loss": 0.0821,
+      "loss": 0.0782,
       "step": 4845
     },
     {
       "epoch": 88.11,
       "learning_rate": 6.59e-05,
-      "loss": 0.1091,
+      "loss": 0.0907,
       "step": 4846
     },
     {
       "epoch": 88.13,
       "learning_rate": 6.58e-05,
-      "loss": 0.0753,
+      "loss": 0.0624,
       "step": 4847
     },
     {
       "epoch": 88.15,
       "learning_rate": 6.57e-05,
-      "loss": 0.0994,
+      "loss": 0.0835,
       "step": 4848
     },
     {
       "epoch": 88.16,
       "learning_rate": 6.560000000000001e-05,
-      "loss": 0.0773,
+      "loss": 0.0609,
       "step": 4849
     },
     {
       "epoch": 88.18,
       "learning_rate": 6.55e-05,
-      "loss": 0.0901,
+      "loss": 0.0951,
       "step": 4850
     },
     {
       "epoch": 88.2,
       "learning_rate": 6.54e-05,
-      "loss": 0.0652,
+      "loss": 0.0726,
       "step": 4851
     },
     {
       "epoch": 88.22,
       "learning_rate": 6.53e-05,
-      "loss": 0.062,
+      "loss": 0.0478,
       "step": 4852
     },
     {
       "epoch": 88.24,
       "learning_rate": 6.52e-05,
-      "loss": 0.0826,
+      "loss": 0.0794,
       "step": 4853
     },
     {
       "epoch": 88.25,
       "learning_rate": 6.510000000000001e-05,
-      "loss": 0.111,
+      "loss": 0.0766,
       "step": 4854
     },
     {
       "epoch": 88.27,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 0.1096,
+      "loss": 0.0905,
       "step": 4855
     },
     {
       "epoch": 88.29,
       "learning_rate": 6.49e-05,
-      "loss": 0.0687,
+      "loss": 0.0786,
       "step": 4856
     },
     {
       "epoch": 88.31,
       "learning_rate": 6.48e-05,
-      "loss": 0.0899,
+      "loss": 0.0689,
       "step": 4857
     },
     {
       "epoch": 88.33,
       "learning_rate": 6.47e-05,
-      "loss": 0.094,
+      "loss": 0.0997,
       "step": 4858
     },
     {
       "epoch": 88.35,
       "learning_rate": 6.460000000000001e-05,
-      "loss": 0.0811,
+      "loss": 0.0551,
       "step": 4859
     },
     {
       "epoch": 88.36,
       "learning_rate": 6.450000000000001e-05,
-      "loss": 0.0739,
+      "loss": 0.0569,
       "step": 4860
     },
     {
       "epoch": 88.38,
       "learning_rate": 6.440000000000001e-05,
-      "loss": 0.1124,
+      "loss": 0.0969,
       "step": 4861
     },
     {
       "epoch": 88.4,
       "learning_rate": 6.429999999999999e-05,
-      "loss": 0.1004,
+      "loss": 0.0708,
       "step": 4862
     },
     {
       "epoch": 88.42,
       "learning_rate": 6.419999999999999e-05,
-      "loss": 0.1239,
+      "loss": 0.1088,
       "step": 4863
     },
     {
       "epoch": 88.44,
       "learning_rate": 6.41e-05,
-      "loss": 0.0878,
+      "loss": 0.0685,
       "step": 4864
     },
     {
       "epoch": 88.45,
       "learning_rate": 6.4e-05,
-      "loss": 0.0584,
+      "loss": 0.037,
       "step": 4865
     },
     {
       "epoch": 88.47,
       "learning_rate": 6.39e-05,
-      "loss": 0.1105,
+      "loss": 0.1161,
       "step": 4866
     },
     {
       "epoch": 88.49,
       "learning_rate": 6.379999999999999e-05,
-      "loss": 0.0817,
+      "loss": 0.0728,
       "step": 4867
     },
     {
       "epoch": 88.51,
       "learning_rate": 6.37e-05,
-      "loss": 0.0752,
+      "loss": 0.066,
       "step": 4868
     },
     {
       "epoch": 88.53,
       "learning_rate": 6.36e-05,
-      "loss": 0.0739,
+      "loss": 0.0509,
       "step": 4869
     },
     {
       "epoch": 88.55,
       "learning_rate": 6.35e-05,
-      "loss": 0.0844,
+      "loss": 0.0743,
       "step": 4870
     },
     {
       "epoch": 88.56,
       "learning_rate": 6.34e-05,
-      "loss": 0.0891,
+      "loss": 0.0596,
       "step": 4871
     },
     {
       "epoch": 88.58,
       "learning_rate": 6.33e-05,
-      "loss": 0.1144,
+      "loss": 0.1101,
       "step": 4872
     },
     {
       "epoch": 88.6,
       "learning_rate": 6.32e-05,
-      "loss": 0.0765,
+      "loss": 0.0759,
       "step": 4873
     },
     {
       "epoch": 88.62,
       "learning_rate": 6.31e-05,
-      "loss": 0.0706,
+      "loss": 0.0729,
       "step": 4874
     },
     {
       "epoch": 88.64,
       "learning_rate": 6.3e-05,
-      "loss": 0.072,
+      "loss": 0.0602,
       "step": 4875
     },
     {
       "epoch": 88.65,
       "learning_rate": 6.29e-05,
-      "loss": 0.0835,
+      "loss": 0.08,
       "step": 4876
     },
     {
       "epoch": 88.67,
       "learning_rate": 6.28e-05,
-      "loss": 0.0734,
+      "loss": 0.0659,
       "step": 4877
     },
     {
       "epoch": 88.69,
       "learning_rate": 6.27e-05,
-      "loss": 0.1083,
+      "loss": 0.0769,
       "step": 4878
     },
     {
       "epoch": 88.71,
       "learning_rate": 6.26e-05,
-      "loss": 0.1138,
+      "loss": 0.1172,
       "step": 4879
     },
     {
       "epoch": 88.73,
       "learning_rate": 6.25e-05,
-      "loss": 0.0797,
+      "loss": 0.0551,
       "step": 4880
     },
     {
       "epoch": 88.75,
       "learning_rate": 6.24e-05,
-      "loss": 0.0743,
+      "loss": 0.0708,
       "step": 4881
     },
     {
       "epoch": 88.76,
       "learning_rate": 6.23e-05,
-      "loss": 0.0736,
+      "loss": 0.0704,
       "step": 4882
     },
     {
       "epoch": 88.78,
       "learning_rate": 6.22e-05,
-      "loss": 0.0618,
+      "loss": 0.06,
       "step": 4883
     },
     {
       "epoch": 88.8,
       "learning_rate": 6.21e-05,
-      "loss": 0.1317,
+      "loss": 0.0996,
       "step": 4884
     },
     {
       "epoch": 88.82,
       "learning_rate": 6.2e-05,
-      "loss": 0.1018,
+      "loss": 0.0786,
       "step": 4885
     },
     {
       "epoch": 88.84,
       "learning_rate": 6.19e-05,
-      "loss": 0.1112,
+      "loss": 0.101,
       "step": 4886
     },
     {
       "epoch": 88.85,
       "learning_rate": 6.18e-05,
-      "loss": 0.061,
+      "loss": 0.0785,
       "step": 4887
     },
     {
       "epoch": 88.87,
       "learning_rate": 6.17e-05,
-      "loss": 0.087,
+      "loss": 0.0743,
       "step": 4888
     },
     {
       "epoch": 88.89,
       "learning_rate": 6.16e-05,
-      "loss": 0.0652,
+      "loss": 0.0475,
       "step": 4889
     },
     {
       "epoch": 88.91,
       "learning_rate": 6.15e-05,
-      "loss": 0.0906,
+      "loss": 0.073,
       "step": 4890
     },
     {
       "epoch": 88.93,
       "learning_rate": 6.14e-05,
-      "loss": 0.1041,
+      "loss": 0.084,
       "step": 4891
     },
     {
       "epoch": 88.95,
       "learning_rate": 6.13e-05,
-      "loss": 0.0573,
+      "loss": 0.0495,
       "step": 4892
     },
     {
       "epoch": 88.96,
       "learning_rate": 6.12e-05,
-      "loss": 0.0832,
+      "loss": 0.0765,
       "step": 4893
     },
     {
       "epoch": 88.98,
       "learning_rate": 6.110000000000001e-05,
-      "loss": 0.0809,
+      "loss": 0.0677,
       "step": 4894
     },
     {
       "epoch": 89.0,
       "learning_rate": 6.1e-05,
-      "loss": 0.0893,
+      "loss": 0.0808,
       "step": 4895
     },
     {
       "epoch": 89.02,
       "learning_rate": 6.09e-05,
-      "loss": 0.093,
+      "loss": 0.0565,
       "step": 4896
     },
     {
       "epoch": 89.04,
       "learning_rate": 6.08e-05,
-      "loss": 0.0708,
+      "loss": 0.0639,
       "step": 4897
     },
     {
       "epoch": 89.05,
       "learning_rate": 6.07e-05,
-      "loss": 0.1107,
+      "loss": 0.126,
       "step": 4898
     },
     {
       "epoch": 89.07,
       "learning_rate": 6.06e-05,
-      "loss": 0.0994,
+      "loss": 0.0816,
       "step": 4899
     },
     {
       "epoch": 89.09,
       "learning_rate": 6.05e-05,
-      "loss": 0.0989,
+      "loss": 0.0873,
       "step": 4900
     },
     {
       "epoch": 89.11,
       "learning_rate": 6.0400000000000004e-05,
-      "loss": 0.0999,
+      "loss": 0.0758,
       "step": 4901
     },
     {
       "epoch": 89.13,
       "learning_rate": 6.03e-05,
-      "loss": 0.087,
+      "loss": 0.1098,
       "step": 4902
     },
     {
       "epoch": 89.15,
       "learning_rate": 6.02e-05,
-      "loss": 0.0753,
+      "loss": 0.0909,
       "step": 4903
     },
     {
       "epoch": 89.16,
       "learning_rate": 6.0100000000000004e-05,
-      "loss": 0.1084,
+      "loss": 0.0913,
       "step": 4904
     },
     {
       "epoch": 89.18,
       "learning_rate": 6e-05,
-      "loss": 0.0548,
+      "loss": 0.0487,
       "step": 4905
     },
     {
       "epoch": 89.2,
       "learning_rate": 5.9900000000000006e-05,
-      "loss": 0.0931,
+      "loss": 0.0686,
       "step": 4906
     },
     {
       "epoch": 89.22,
       "learning_rate": 5.9800000000000003e-05,
-      "loss": 0.079,
+      "loss": 0.0825,
       "step": 4907
     },
     {
       "epoch": 89.24,
       "learning_rate": 5.97e-05,
-      "loss": 0.0796,
+      "loss": 0.0932,
       "step": 4908
     },
     {
       "epoch": 89.25,
       "learning_rate": 5.96e-05,
-      "loss": 0.077,
+      "loss": 0.0777,
       "step": 4909
     },
     {
       "epoch": 89.27,
       "learning_rate": 5.9499999999999996e-05,
-      "loss": 0.1076,
+      "loss": 0.1033,
       "step": 4910
     },
     {
       "epoch": 89.29,
       "learning_rate": 5.94e-05,
-      "loss": 0.0951,
+      "loss": 0.0695,
       "step": 4911
     },
     {
       "epoch": 89.31,
       "learning_rate": 5.93e-05,
-      "loss": 0.083,
+      "loss": 0.0701,
       "step": 4912
     },
     {
       "epoch": 89.33,
       "learning_rate": 5.92e-05,
-      "loss": 0.0795,
+      "loss": 0.0799,
       "step": 4913
     },
     {
       "epoch": 89.35,
       "learning_rate": 5.91e-05,
-      "loss": 0.0514,
+      "loss": 0.0668,
       "step": 4914
     },
     {
       "epoch": 89.36,
       "learning_rate": 5.9e-05,
-      "loss": 0.0966,
+      "loss": 0.0882,
       "step": 4915
     },
     {
       "epoch": 89.38,
       "learning_rate": 5.89e-05,
-      "loss": 0.0788,
+      "loss": 0.0829,
       "step": 4916
     },
     {
       "epoch": 89.4,
       "learning_rate": 5.88e-05,
-      "loss": 0.0991,
+      "loss": 0.1009,
       "step": 4917
     },
     {
       "epoch": 89.42,
       "learning_rate": 5.8700000000000004e-05,
-      "loss": 0.0653,
+      "loss": 0.061,
       "step": 4918
     },
     {
       "epoch": 89.44,
       "learning_rate": 5.86e-05,
-      "loss": 0.0727,
+      "loss": 0.0628,
       "step": 4919
     },
     {
       "epoch": 89.45,
       "learning_rate": 5.8500000000000006e-05,
-      "loss": 0.0515,
+      "loss": 0.0508,
       "step": 4920
     },
     {
       "epoch": 89.47,
       "learning_rate": 5.84e-05,
-      "loss": 0.1202,
+      "loss": 0.1084,
       "step": 4921
     },
     {
       "epoch": 89.49,
       "learning_rate": 5.83e-05,
-      "loss": 0.1002,
+      "loss": 0.0843,
       "step": 4922
     },
     {
       "epoch": 89.51,
       "learning_rate": 5.8200000000000005e-05,
-      "loss": 0.1013,
+      "loss": 0.0994,
       "step": 4923
     },
     {
       "epoch": 89.53,
       "learning_rate": 5.81e-05,
-      "loss": 0.0484,
+      "loss": 0.0358,
       "step": 4924
     },
     {
       "epoch": 89.55,
       "learning_rate": 5.800000000000001e-05,
-      "loss": 0.0629,
+      "loss": 0.0723,
       "step": 4925
     },
     {
       "epoch": 89.56,
       "learning_rate": 5.79e-05,
-      "loss": 0.0401,
+      "loss": 0.0306,
       "step": 4926
     },
     {
       "epoch": 89.58,
       "learning_rate": 5.7799999999999995e-05,
-      "loss": 0.0765,
+      "loss": 0.0757,
       "step": 4927
     },
     {
       "epoch": 89.6,
       "learning_rate": 5.77e-05,
-      "loss": 0.0777,
+      "loss": 0.0796,
       "step": 4928
     },
     {
       "epoch": 89.62,
       "learning_rate": 5.76e-05,
-      "loss": 0.0801,
+      "loss": 0.0658,
       "step": 4929
     },
     {
       "epoch": 89.64,
       "learning_rate": 5.75e-05,
-      "loss": 0.048,
+      "loss": 0.0377,
       "step": 4930
     },
     {
       "epoch": 89.65,
       "learning_rate": 5.74e-05,
-      "loss": 0.0826,
+      "loss": 0.0885,
       "step": 4931
     },
     {
       "epoch": 89.67,
       "learning_rate": 5.73e-05,
-      "loss": 0.0602,
+      "loss": 0.0622,
       "step": 4932
     },
     {
       "epoch": 89.69,
       "learning_rate": 5.72e-05,
-      "loss": 0.0728,
+      "loss": 0.0729,
       "step": 4933
     },
     {
       "epoch": 89.71,
       "learning_rate": 5.71e-05,
-      "loss": 0.0735,
+      "loss": 0.0702,
       "step": 4934
     },
     {
       "epoch": 89.73,
       "learning_rate": 5.7e-05,
-      "loss": 0.0943,
+      "loss": 0.0744,
       "step": 4935
     },
     {
       "epoch": 89.75,
       "learning_rate": 5.69e-05,
-      "loss": 0.0929,
+      "loss": 0.0714,
       "step": 4936
     },
     {
       "epoch": 89.76,
       "learning_rate": 5.6800000000000005e-05,
-      "loss": 0.0801,
+      "loss": 0.0655,
       "step": 4937
     },
     {
       "epoch": 89.78,
       "learning_rate": 5.67e-05,
-      "loss": 0.0808,
+      "loss": 0.0936,
       "step": 4938
     },
     {
       "epoch": 89.8,
       "learning_rate": 5.66e-05,
-      "loss": 0.0705,
+      "loss": 0.067,
       "step": 4939
     },
     {
       "epoch": 89.82,
       "learning_rate": 5.6500000000000005e-05,
-      "loss": 0.1048,
+      "loss": 0.0999,
       "step": 4940
     },
     {
       "epoch": 89.84,
       "learning_rate": 5.64e-05,
-      "loss": 0.0662,
+      "loss": 0.0661,
       "step": 4941
     },
     {
       "epoch": 89.85,
       "learning_rate": 5.6300000000000006e-05,
-      "loss": 0.0942,
+      "loss": 0.0829,
       "step": 4942
     },
     {
       "epoch": 89.87,
       "learning_rate": 5.6200000000000004e-05,
-      "loss": 0.0721,
+      "loss": 0.066,
       "step": 4943
     },
     {
       "epoch": 89.89,
       "learning_rate": 5.61e-05,
-      "loss": 0.0668,
+      "loss": 0.0662,
       "step": 4944
     },
     {
       "epoch": 89.91,
       "learning_rate": 5.6e-05,
-      "loss": 0.0777,
+      "loss": 0.0593,
       "step": 4945
     },
     {
       "epoch": 89.93,
       "learning_rate": 5.59e-05,
-      "loss": 0.0906,
+      "loss": 0.089,
       "step": 4946
     },
     {
       "epoch": 89.95,
       "learning_rate": 5.58e-05,
-      "loss": 0.0904,
+      "loss": 0.0769,
       "step": 4947
     },
     {
       "epoch": 89.96,
       "learning_rate": 5.57e-05,
-      "loss": 0.0933,
+      "loss": 0.0749,
       "step": 4948
     },
     {
       "epoch": 89.98,
       "learning_rate": 5.5599999999999996e-05,
-      "loss": 0.0796,
+      "loss": 0.0626,
       "step": 4949
     },
     {
       "epoch": 90.0,
       "learning_rate": 5.55e-05,
-      "loss": 0.0868,
+      "loss": 0.0598,
       "step": 4950
     },
     {
       "epoch": 90.02,
       "learning_rate": 5.54e-05,
-      "loss": 0.0925,
+      "loss": 0.0941,
       "step": 4951
     },
     {
       "epoch": 90.04,
       "learning_rate": 5.53e-05,
-      "loss": 0.0806,
+      "loss": 0.108,
       "step": 4952
     },
     {
       "epoch": 90.05,
       "learning_rate": 5.52e-05,
-      "loss": 0.0766,
+      "loss": 0.0649,
       "step": 4953
     },
     {
       "epoch": 90.07,
       "learning_rate": 5.5100000000000004e-05,
-      "loss": 0.1134,
+      "loss": 0.0991,
       "step": 4954
     },
     {
       "epoch": 90.09,
       "learning_rate": 5.5e-05,
-      "loss": 0.0891,
+      "loss": 0.0715,
       "step": 4955
     },
     {
       "epoch": 90.11,
       "learning_rate": 5.49e-05,
-      "loss": 0.0564,
+      "loss": 0.0506,
       "step": 4956
     },
     {
       "epoch": 90.13,
       "learning_rate": 5.4800000000000004e-05,
-      "loss": 0.0768,
+      "loss": 0.081,
       "step": 4957
     },
     {
       "epoch": 90.15,
       "learning_rate": 5.47e-05,
-      "loss": 0.1386,
+      "loss": 0.1142,
       "step": 4958
     },
     {
       "epoch": 90.16,
       "learning_rate": 5.4600000000000006e-05,
-      "loss": 0.0806,
+      "loss": 0.0662,
       "step": 4959
     },
     {
       "epoch": 90.18,
       "learning_rate": 5.45e-05,
-      "loss": 0.0723,
+      "loss": 0.0892,
       "step": 4960
     },
     {
       "epoch": 90.2,
       "learning_rate": 5.44e-05,
-      "loss": 0.0782,
+      "loss": 0.0838,
       "step": 4961
     },
     {
       "epoch": 90.22,
       "learning_rate": 5.4300000000000005e-05,
-      "loss": 0.0717,
+      "loss": 0.0836,
       "step": 4962
     },
     {
       "epoch": 90.24,
       "learning_rate": 5.4199999999999996e-05,
-      "loss": 0.0884,
+      "loss": 0.0953,
       "step": 4963
     },
     {
       "epoch": 90.25,
       "learning_rate": 5.41e-05,
-      "loss": 0.0689,
+      "loss": 0.0701,
       "step": 4964
     },
     {
       "epoch": 90.27,
       "learning_rate": 5.4e-05,
-      "loss": 0.1003,
+      "loss": 0.0837,
       "step": 4965
     },
     {
       "epoch": 90.29,
       "learning_rate": 5.39e-05,
-      "loss": 0.0944,
+      "loss": 0.0872,
       "step": 4966
     },
     {
       "epoch": 90.31,
       "learning_rate": 5.38e-05,
-      "loss": 0.0949,
+      "loss": 0.0959,
       "step": 4967
     },
     {
       "epoch": 90.33,
       "learning_rate": 5.37e-05,
-      "loss": 0.069,
+      "loss": 0.064,
       "step": 4968
     },
     {
       "epoch": 90.35,
       "learning_rate": 5.36e-05,
-      "loss": 0.0876,
+      "loss": 0.0545,
       "step": 4969
     },
     {
       "epoch": 90.36,
       "learning_rate": 5.35e-05,
-      "loss": 0.0898,
+      "loss": 0.0828,
       "step": 4970
     },
     {
       "epoch": 90.38,
       "learning_rate": 5.3400000000000004e-05,
-      "loss": 0.0715,
+      "loss": 0.0697,
       "step": 4971
     },
     {
       "epoch": 90.4,
       "learning_rate": 5.33e-05,
-      "loss": 0.0919,
+      "loss": 0.0857,
       "step": 4972
     },
     {
       "epoch": 90.42,
       "learning_rate": 5.32e-05,
-      "loss": 0.0824,
+      "loss": 0.0788,
       "step": 4973
     },
     {
       "epoch": 90.44,
       "learning_rate": 5.31e-05,
-      "loss": 0.0977,
+      "loss": 0.0739,
       "step": 4974
     },
     {
       "epoch": 90.45,
       "learning_rate": 5.3e-05,
-      "loss": 0.0667,
+      "loss": 0.0537,
       "step": 4975
     },
     {
       "epoch": 90.47,
       "learning_rate": 5.2900000000000005e-05,
-      "loss": 0.0878,
+      "loss": 0.0731,
       "step": 4976
     },
     {
       "epoch": 90.49,
       "learning_rate": 5.28e-05,
-      "loss": 0.0699,
+      "loss": 0.0759,
       "step": 4977
     },
     {
       "epoch": 90.51,
       "learning_rate": 5.27e-05,
-      "loss": 0.0895,
+      "loss": 0.0773,
       "step": 4978
     },
     {
       "epoch": 90.53,
       "learning_rate": 5.2600000000000005e-05,
-      "loss": 0.0545,
+      "loss": 0.0762,
       "step": 4979
     },
     {
       "epoch": 90.55,
       "learning_rate": 5.25e-05,
-      "loss": 0.0971,
+      "loss": 0.0786,
       "step": 4980
     },
     {
       "epoch": 90.56,
       "learning_rate": 5.2400000000000007e-05,
-      "loss": 0.0723,
+      "loss": 0.0387,
       "step": 4981
     },
     {
       "epoch": 90.58,
       "learning_rate": 5.23e-05,
-      "loss": 0.0763,
+      "loss": 0.0684,
       "step": 4982
     },
     {
       "epoch": 90.6,
       "learning_rate": 5.22e-05,
-      "loss": 0.0804,
+      "loss": 0.0664,
       "step": 4983
     },
     {
       "epoch": 90.62,
       "learning_rate": 5.21e-05,
-      "loss": 0.1074,
+      "loss": 0.0952,
       "step": 4984
     },
     {
       "epoch": 90.64,
       "learning_rate": 5.2e-05,
-      "loss": 0.097,
+      "loss": 0.1009,
       "step": 4985
     },
     {
       "epoch": 90.65,
       "learning_rate": 5.19e-05,
-      "loss": 0.0922,
+      "loss": 0.0822,
       "step": 4986
     },
     {
       "epoch": 90.67,
       "learning_rate": 5.18e-05,
-      "loss": 0.0554,
+      "loss": 0.0642,
       "step": 4987
     },
     {
       "epoch": 90.69,
       "learning_rate": 5.17e-05,
-      "loss": 0.1063,
+      "loss": 0.0722,
       "step": 4988
     },
     {
       "epoch": 90.71,
       "learning_rate": 5.16e-05,
-      "loss": 0.0904,
+      "loss": 0.0895,
       "step": 4989
     },
     {
       "epoch": 90.73,
       "learning_rate": 5.15e-05,
-      "loss": 0.07,
+      "loss": 0.0549,
       "step": 4990
     },
     {
       "epoch": 90.75,
       "learning_rate": 5.14e-05,
-      "loss": 0.0962,
+      "loss": 0.1059,
       "step": 4991
     },
     {
       "epoch": 90.76,
       "learning_rate": 5.13e-05,
-      "loss": 0.1132,
+      "loss": 0.0922,
       "step": 4992
     },
     {
       "epoch": 90.78,
       "learning_rate": 5.1200000000000004e-05,
-      "loss": 0.0731,
+      "loss": 0.0612,
       "step": 4993
     },
     {
       "epoch": 90.8,
       "learning_rate": 5.11e-05,
-      "loss": 0.0862,
+      "loss": 0.069,
       "step": 4994
     },
     {
       "epoch": 90.82,
       "learning_rate": 5.1e-05,
-      "loss": 0.0905,
+      "loss": 0.0673,
       "step": 4995
     },
     {
       "epoch": 90.84,
       "learning_rate": 5.0900000000000004e-05,
-      "loss": 0.084,
+      "loss": 0.0643,
       "step": 4996
     },
     {
       "epoch": 90.85,
       "learning_rate": 5.08e-05,
-      "loss": 0.0967,
+      "loss": 0.0907,
       "step": 4997
     },
     {
       "epoch": 90.87,
       "learning_rate": 5.0700000000000006e-05,
-      "loss": 0.0825,
+      "loss": 0.0657,
       "step": 4998
     },
     {
       "epoch": 90.89,
       "learning_rate": 5.0600000000000003e-05,
-      "loss": 0.0856,
+      "loss": 0.0766,
       "step": 4999
     },
     {
       "epoch": 90.91,
       "learning_rate": 5.05e-05,
-      "loss": 0.0825,
+      "loss": 0.1119,
       "step": 5000
     },
     {
       "epoch": 90.91,
-      "eval_cer": 0.07691269734178924,
-      "eval_loss": 0.4085361361503601,
-      "eval_runtime": 13.9902,
-      "eval_samples_per_second": 117.725,
-      "eval_steps_per_second": 1.858,
-      "eval_wer": 0.31355326320089877,
+      "eval_cer": 0.07550338086028276,
+      "eval_loss": 0.423967570066452,
+      "eval_runtime": 14.2346,
+      "eval_samples_per_second": 115.704,
+      "eval_steps_per_second": 1.827,
+      "eval_wer": 0.31202124399959147,
       "step": 5000
     },
     {
       "epoch": 90.93,
       "learning_rate": 5.04e-05,
-      "loss": 0.0767,
+      "loss": 0.0741,
       "step": 5001
     },
     {
       "epoch": 90.95,
       "learning_rate": 5.0299999999999996e-05,
-      "loss": 0.0851,
+      "loss": 0.0857,
       "step": 5002
     },
     {
       "epoch": 90.96,
       "learning_rate": 5.02e-05,
-      "loss": 0.04,
+      "loss": 0.0394,
       "step": 5003
     },
     {
       "epoch": 90.98,
       "learning_rate": 5.01e-05,
-      "loss": 0.0943,
+      "loss": 0.0888,
       "step": 5004
     },
     {
       "epoch": 91.0,
       "learning_rate": 5e-05,
-      "loss": 0.0387,
+      "loss": 0.027,
       "step": 5005
     },
     {
       "epoch": 91.02,
       "learning_rate": 4.99e-05,
-      "loss": 0.1055,
+      "loss": 0.0753,
       "step": 5006
     },
     {
       "epoch": 91.04,
       "learning_rate": 4.98e-05,
-      "loss": 0.0809,
+      "loss": 0.0728,
       "step": 5007
     },
     {
       "epoch": 91.05,
       "learning_rate": 4.97e-05,
-      "loss": 0.0646,
+      "loss": 0.0677,
       "step": 5008
     },
     {
       "epoch": 91.07,
       "learning_rate": 4.96e-05,
-      "loss": 0.1045,
+      "loss": 0.0856,
       "step": 5009
     },
     {
       "epoch": 91.09,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.0781,
+      "loss": 0.0661,
       "step": 5010
     },
     {
       "epoch": 91.11,
       "learning_rate": 4.94e-05,
-      "loss": 0.0737,
+      "loss": 0.0657,
       "step": 5011
     },
     {
       "epoch": 91.13,
       "learning_rate": 4.93e-05,
-      "loss": 0.0877,
+      "loss": 0.0738,
       "step": 5012
     },
     {
       "epoch": 91.15,
       "learning_rate": 4.92e-05,
-      "loss": 0.074,
+      "loss": 0.0644,
       "step": 5013
     },
     {
       "epoch": 91.16,
       "learning_rate": 4.91e-05,
-      "loss": 0.0724,
+      "loss": 0.0783,
       "step": 5014
     },
     {
       "epoch": 91.18,
       "learning_rate": 4.9000000000000005e-05,
-      "loss": 0.1052,
+      "loss": 0.0869,
       "step": 5015
     },
     {
       "epoch": 91.2,
       "learning_rate": 4.89e-05,
-      "loss": 0.0855,
+      "loss": 0.0774,
       "step": 5016
     },
     {
       "epoch": 91.22,
       "learning_rate": 4.880000000000001e-05,
-      "loss": 0.1007,
+      "loss": 0.0989,
       "step": 5017
     },
     {
       "epoch": 91.24,
       "learning_rate": 4.87e-05,
-      "loss": 0.0766,
+      "loss": 0.0611,
       "step": 5018
     },
     {
       "epoch": 91.25,
       "learning_rate": 4.8599999999999995e-05,
-      "loss": 0.0752,
+      "loss": 0.0832,
       "step": 5019
     },
     {
       "epoch": 91.27,
       "learning_rate": 4.85e-05,
-      "loss": 0.0999,
+      "loss": 0.0928,
       "step": 5020
     },
     {
       "epoch": 91.29,
       "learning_rate": 4.84e-05,
-      "loss": 0.083,
+      "loss": 0.0846,
       "step": 5021
     },
     {
       "epoch": 91.31,
       "learning_rate": 4.83e-05,
-      "loss": 0.0559,
+      "loss": 0.0567,
       "step": 5022
     },
     {
       "epoch": 91.33,
       "learning_rate": 4.82e-05,
-      "loss": 0.0588,
+      "loss": 0.0542,
       "step": 5023
     },
     {
       "epoch": 91.35,
       "learning_rate": 4.81e-05,
-      "loss": 0.0661,
+      "loss": 0.0684,
       "step": 5024
     },
     {
       "epoch": 91.36,
       "learning_rate": 4.8e-05,
-      "loss": 0.0772,
+      "loss": 0.0766,
       "step": 5025
     },
     {
       "epoch": 91.38,
       "learning_rate": 4.79e-05,
-      "loss": 0.0637,
+      "loss": 0.0495,
       "step": 5026
     },
     {
       "epoch": 91.4,
       "learning_rate": 4.78e-05,
-      "loss": 0.0674,
+      "loss": 0.0669,
       "step": 5027
     },
     {
       "epoch": 91.42,
       "learning_rate": 4.77e-05,
-      "loss": 0.1113,
+      "loss": 0.0887,
       "step": 5028
     },
     {
       "epoch": 91.44,
       "learning_rate": 4.7600000000000005e-05,
-      "loss": 0.0601,
+      "loss": 0.0662,
       "step": 5029
     },
     {
       "epoch": 91.45,
       "learning_rate": 4.75e-05,
-      "loss": 0.041,
+      "loss": 0.0348,
       "step": 5030
     },
     {
       "epoch": 91.47,
       "learning_rate": 4.74e-05,
-      "loss": 0.125,
+      "loss": 0.1199,
       "step": 5031
     },
     {
       "epoch": 91.49,
       "learning_rate": 4.7300000000000005e-05,
-      "loss": 0.0576,
+      "loss": 0.0553,
       "step": 5032
     },
     {
       "epoch": 91.51,
       "learning_rate": 4.72e-05,
-      "loss": 0.0708,
+      "loss": 0.0706,
       "step": 5033
     },
     {
       "epoch": 91.53,
       "learning_rate": 4.7100000000000006e-05,
-      "loss": 0.0614,
+      "loss": 0.052,
       "step": 5034
     },
     {
       "epoch": 91.55,
       "learning_rate": 4.7000000000000004e-05,
-      "loss": 0.0706,
+      "loss": 0.0652,
       "step": 5035
     },
     {
       "epoch": 91.56,
       "learning_rate": 4.69e-05,
-      "loss": 0.065,
+      "loss": 0.0435,
       "step": 5036
     },
     {
       "epoch": 91.58,
       "learning_rate": 4.68e-05,
-      "loss": 0.0826,
+      "loss": 0.0776,
       "step": 5037
     },
     {
       "epoch": 91.6,
       "learning_rate": 4.67e-05,
-      "loss": 0.0827,
+      "loss": 0.0726,
       "step": 5038
     },
     {
       "epoch": 91.62,
       "learning_rate": 4.66e-05,
-      "loss": 0.0768,
+      "loss": 0.066,
       "step": 5039
     },
     {
       "epoch": 91.64,
       "learning_rate": 4.65e-05,
-      "loss": 0.0649,
+      "loss": 0.0681,
       "step": 5040
     },
     {
       "epoch": 91.65,
       "learning_rate": 4.6399999999999996e-05,
-      "loss": 0.0529,
+      "loss": 0.0397,
       "step": 5041
     },
     {
       "epoch": 91.67,
       "learning_rate": 4.63e-05,
-      "loss": 0.063,
+      "loss": 0.0463,
       "step": 5042
     },
     {
       "epoch": 91.69,
       "learning_rate": 4.62e-05,
-      "loss": 0.0906,
+      "loss": 0.0936,
       "step": 5043
     },
     {
       "epoch": 91.71,
       "learning_rate": 4.61e-05,
-      "loss": 0.0752,
+      "loss": 0.0685,
       "step": 5044
     },
     {
       "epoch": 91.73,
       "learning_rate": 4.6e-05,
-      "loss": 0.078,
+      "loss": 0.0694,
       "step": 5045
     },
     {
       "epoch": 91.75,
       "learning_rate": 4.5900000000000004e-05,
-      "loss": 0.0959,
+      "loss": 0.0891,
       "step": 5046
     },
     {
       "epoch": 91.76,
       "learning_rate": 4.58e-05,
-      "loss": 0.0833,
+      "loss": 0.0642,
       "step": 5047
     },
     {
       "epoch": 91.78,
       "learning_rate": 4.57e-05,
-      "loss": 0.0606,
+      "loss": 0.0556,
       "step": 5048
     },
     {
       "epoch": 91.8,
       "learning_rate": 4.5600000000000004e-05,
-      "loss": 0.0782,
+      "loss": 0.0669,
       "step": 5049
     },
     {
       "epoch": 91.82,
       "learning_rate": 4.55e-05,
-      "loss": 0.1249,
+      "loss": 0.1192,
       "step": 5050
     },
     {
       "epoch": 91.84,
       "learning_rate": 4.5400000000000006e-05,
-      "loss": 0.077,
+      "loss": 0.0565,
       "step": 5051
     },
     {
       "epoch": 91.85,
       "learning_rate": 4.53e-05,
-      "loss": 0.0973,
+      "loss": 0.069,
       "step": 5052
     },
     {
       "epoch": 91.87,
       "learning_rate": 4.52e-05,
-      "loss": 0.0679,
+      "loss": 0.0465,
       "step": 5053
     },
     {
       "epoch": 91.89,
       "learning_rate": 4.5100000000000005e-05,
-      "loss": 0.0659,
+      "loss": 0.0416,
       "step": 5054
     },
     {
       "epoch": 91.91,
       "learning_rate": 4.4999999999999996e-05,
-      "loss": 0.0546,
+      "loss": 0.0529,
       "step": 5055
     },
     {
       "epoch": 91.93,
       "learning_rate": 4.49e-05,
-      "loss": 0.1089,
+      "loss": 0.0948,
       "step": 5056
     },
     {
       "epoch": 91.95,
       "learning_rate": 4.48e-05,
-      "loss": 0.096,
+      "loss": 0.0746,
       "step": 5057
     },
     {
       "epoch": 91.96,
       "learning_rate": 4.4699999999999996e-05,
-      "loss": 0.081,
+      "loss": 0.067,
       "step": 5058
     },
     {
       "epoch": 91.98,
       "learning_rate": 4.46e-05,
-      "loss": 0.0844,
+      "loss": 0.0749,
       "step": 5059
     },
     {
       "epoch": 92.0,
       "learning_rate": 4.45e-05,
-      "loss": 0.021,
+      "loss": 0.0213,
       "step": 5060
     },
     {
       "epoch": 92.02,
       "learning_rate": 4.44e-05,
-      "loss": 0.1038,
+      "loss": 0.1216,
       "step": 5061
     },
     {
       "epoch": 92.04,
       "learning_rate": 4.43e-05,
-      "loss": 0.1157,
+      "loss": 0.0727,
       "step": 5062
     },
     {
       "epoch": 92.05,
       "learning_rate": 4.4200000000000004e-05,
-      "loss": 0.0618,
+      "loss": 0.0394,
       "step": 5063
     },
     {
       "epoch": 92.07,
       "learning_rate": 4.41e-05,
-      "loss": 0.0711,
+      "loss": 0.0635,
       "step": 5064
     },
     {
       "epoch": 92.09,
       "learning_rate": 4.4e-05,
-      "loss": 0.0697,
+      "loss": 0.0661,
       "step": 5065
     },
     {
       "epoch": 92.11,
       "learning_rate": 4.39e-05,
-      "loss": 0.0825,
+      "loss": 0.0613,
       "step": 5066
     },
     {
       "epoch": 92.13,
       "learning_rate": 4.38e-05,
-      "loss": 0.0684,
+      "loss": 0.0752,
       "step": 5067
     },
     {
       "epoch": 92.15,
       "learning_rate": 4.3700000000000005e-05,
-      "loss": 0.075,
+      "loss": 0.0585,
       "step": 5068
     },
     {
       "epoch": 92.16,
       "learning_rate": 4.36e-05,
-      "loss": 0.0678,
+      "loss": 0.0563,
       "step": 5069
     },
     {
       "epoch": 92.18,
       "learning_rate": 4.35e-05,
-      "loss": 0.0741,
+      "loss": 0.0804,
       "step": 5070
     },
     {
       "epoch": 92.2,
       "learning_rate": 4.3400000000000005e-05,
-      "loss": 0.0823,
+      "loss": 0.0624,
       "step": 5071
     },
     {
       "epoch": 92.22,
       "learning_rate": 4.33e-05,
-      "loss": 0.0658,
+      "loss": 0.0526,
       "step": 5072
     },
     {
       "epoch": 92.24,
       "learning_rate": 4.3200000000000007e-05,
-      "loss": 0.0822,
+      "loss": 0.0893,
       "step": 5073
     },
     {
       "epoch": 92.25,
       "learning_rate": 4.31e-05,
-      "loss": 0.0856,
+      "loss": 0.0804,
       "step": 5074
     },
     {
       "epoch": 92.27,
       "learning_rate": 4.2999999999999995e-05,
-      "loss": 0.0827,
+      "loss": 0.0653,
       "step": 5075
     },
     {
       "epoch": 92.29,
       "learning_rate": 4.29e-05,
-      "loss": 0.0921,
+      "loss": 0.0746,
       "step": 5076
     },
     {
       "epoch": 92.31,
       "learning_rate": 4.28e-05,
-      "loss": 0.0722,
+      "loss": 0.0597,
       "step": 5077
     },
     {
       "epoch": 92.33,
       "learning_rate": 4.27e-05,
-      "loss": 0.0525,
+      "loss": 0.0575,
       "step": 5078
     },
     {
       "epoch": 92.35,
       "learning_rate": 4.26e-05,
-      "loss": 0.082,
+      "loss": 0.0673,
       "step": 5079
     },
     {
       "epoch": 92.36,
       "learning_rate": 4.25e-05,
-      "loss": 0.0885,
+      "loss": 0.0869,
       "step": 5080
     },
     {
       "epoch": 92.38,
       "learning_rate": 4.24e-05,
-      "loss": 0.0791,
+      "loss": 0.0686,
       "step": 5081
     },
     {
       "epoch": 92.4,
       "learning_rate": 4.23e-05,
-      "loss": 0.0646,
+      "loss": 0.0574,
       "step": 5082
     },
     {
       "epoch": 92.42,
       "learning_rate": 4.22e-05,
-      "loss": 0.0463,
+      "loss": 0.046,
       "step": 5083
     },
     {
       "epoch": 92.44,
       "learning_rate": 4.21e-05,
-      "loss": 0.0669,
+      "loss": 0.0694,
       "step": 5084
     },
     {
       "epoch": 92.45,
       "learning_rate": 4.2000000000000004e-05,
-      "loss": 0.0249,
+      "loss": 0.0328,
       "step": 5085
     },
     {
       "epoch": 92.47,
       "learning_rate": 4.19e-05,
-      "loss": 0.0998,
+      "loss": 0.0775,
       "step": 5086
     },
     {
       "epoch": 92.49,
       "learning_rate": 4.18e-05,
-      "loss": 0.0651,
+      "loss": 0.0573,
       "step": 5087
     },
     {
       "epoch": 92.51,
       "learning_rate": 4.1700000000000004e-05,
-      "loss": 0.0722,
+      "loss": 0.0915,
       "step": 5088
     },
     {
       "epoch": 92.53,
       "learning_rate": 4.16e-05,
-      "loss": 0.0855,
+      "loss": 0.0871,
       "step": 5089
     },
     {
       "epoch": 92.55,
       "learning_rate": 4.1500000000000006e-05,
-      "loss": 0.0861,
+      "loss": 0.0505,
       "step": 5090
     },
     {
       "epoch": 92.56,
       "learning_rate": 4.1400000000000003e-05,
-      "loss": 0.0748,
+      "loss": 0.0753,
       "step": 5091
     },
     {
       "epoch": 92.58,
       "learning_rate": 4.13e-05,
-      "loss": 0.0989,
+      "loss": 0.112,
       "step": 5092
     },
     {
       "epoch": 92.6,
       "learning_rate": 4.12e-05,
-      "loss": 0.0998,
+      "loss": 0.0937,
       "step": 5093
     },
     {
       "epoch": 92.62,
       "learning_rate": 4.1099999999999996e-05,
-      "loss": 0.0681,
+      "loss": 0.0439,
       "step": 5094
     },
     {
       "epoch": 92.64,
       "learning_rate": 4.1e-05,
-      "loss": 0.0584,
+      "loss": 0.0671,
       "step": 5095
     },
     {
       "epoch": 92.65,
       "learning_rate": 4.09e-05,
-      "loss": 0.0699,
+      "loss": 0.0746,
       "step": 5096
     },
     {
       "epoch": 92.67,
       "learning_rate": 4.08e-05,
-      "loss": 0.0866,
+      "loss": 0.0664,
       "step": 5097
     },
     {
       "epoch": 92.69,
       "learning_rate": 4.07e-05,
-      "loss": 0.1071,
+      "loss": 0.0981,
       "step": 5098
     },
     {
       "epoch": 92.71,
       "learning_rate": 4.06e-05,
-      "loss": 0.0763,
+      "loss": 0.0681,
       "step": 5099
     },
     {
       "epoch": 92.73,
       "learning_rate": 4.05e-05,
-      "loss": 0.058,
+      "loss": 0.0491,
       "step": 5100
     },
     {
       "epoch": 92.75,
       "learning_rate": 4.04e-05,
-      "loss": 0.0925,
+      "loss": 0.0673,
       "step": 5101
     },
     {
       "epoch": 92.76,
       "learning_rate": 4.0300000000000004e-05,
-      "loss": 0.08,
+      "loss": 0.0882,
       "step": 5102
     },
     {
       "epoch": 92.78,
       "learning_rate": 4.02e-05,
-      "loss": 0.0676,
+      "loss": 0.0685,
       "step": 5103
     },
     {
       "epoch": 92.8,
       "learning_rate": 4.01e-05,
-      "loss": 0.0588,
+      "loss": 0.0848,
       "step": 5104
     },
     {
       "epoch": 92.82,
       "learning_rate": 4e-05,
-      "loss": 0.1047,
+      "loss": 0.0804,
       "step": 5105
     },
     {
       "epoch": 92.84,
       "learning_rate": 3.99e-05,
-      "loss": 0.0711,
+      "loss": 0.0775,
       "step": 5106
     },
     {
       "epoch": 92.85,
       "learning_rate": 3.9800000000000005e-05,
-      "loss": 0.0841,
+      "loss": 0.0875,
       "step": 5107
     },
     {
       "epoch": 92.87,
       "learning_rate": 3.97e-05,
-      "loss": 0.0659,
+      "loss": 0.0571,
       "step": 5108
     },
     {
       "epoch": 92.89,
       "learning_rate": 3.960000000000001e-05,
-      "loss": 0.0819,
+      "loss": 0.0757,
       "step": 5109
     },
     {
       "epoch": 92.91,
       "learning_rate": 3.95e-05,
-      "loss": 0.0879,
+      "loss": 0.0629,
       "step": 5110
     },
     {
       "epoch": 92.93,
       "learning_rate": 3.9399999999999995e-05,
-      "loss": 0.1202,
+      "loss": 0.0865,
       "step": 5111
     },
     {
       "epoch": 92.95,
       "learning_rate": 3.93e-05,
-      "loss": 0.0746,
+      "loss": 0.0714,
       "step": 5112
     },
     {
       "epoch": 92.96,
       "learning_rate": 3.92e-05,
-      "loss": 0.0935,
+      "loss": 0.0706,
       "step": 5113
     },
     {
       "epoch": 92.98,
       "learning_rate": 3.91e-05,
-      "loss": 0.0913,
+      "loss": 0.0977,
       "step": 5114
     },
     {
       "epoch": 93.0,
       "learning_rate": 3.9e-05,
-      "loss": 0.1042,
+      "loss": 0.0528,
       "step": 5115
     },
     {
       "epoch": 93.02,
       "learning_rate": 3.89e-05,
-      "loss": 0.0857,
+      "loss": 0.085,
       "step": 5116
     },
     {
       "epoch": 93.04,
       "learning_rate": 3.88e-05,
-      "loss": 0.0649,
+      "loss": 0.0553,
       "step": 5117
     },
     {
       "epoch": 93.05,
       "learning_rate": 3.87e-05,
-      "loss": 0.0672,
+      "loss": 0.059,
       "step": 5118
     },
     {
       "epoch": 93.07,
       "learning_rate": 3.86e-05,
-      "loss": 0.0841,
+      "loss": 0.0682,
       "step": 5119
     },
     {
       "epoch": 93.09,
       "learning_rate": 3.85e-05,
-      "loss": 0.0819,
+      "loss": 0.0704,
       "step": 5120
     },
     {
       "epoch": 93.11,
       "learning_rate": 3.84e-05,
-      "loss": 0.0892,
+      "loss": 0.0739,
       "step": 5121
     },
     {
       "epoch": 93.13,
       "learning_rate": 3.83e-05,
-      "loss": 0.1674,
+      "loss": 0.1317,
       "step": 5122
     },
     {
       "epoch": 93.15,
       "learning_rate": 3.82e-05,
-      "loss": 0.0751,
+      "loss": 0.0723,
       "step": 5123
     },
     {
       "epoch": 93.16,
       "learning_rate": 3.8100000000000005e-05,
-      "loss": 0.0942,
+      "loss": 0.0748,
       "step": 5124
     },
     {
       "epoch": 93.18,
       "learning_rate": 3.8e-05,
-      "loss": 0.0672,
+      "loss": 0.0654,
       "step": 5125
     },
     {
       "epoch": 93.2,
       "learning_rate": 3.7900000000000006e-05,
-      "loss": 0.0727,
+      "loss": 0.0648,
       "step": 5126
     },
     {
       "epoch": 93.22,
       "learning_rate": 3.7800000000000004e-05,
-      "loss": 0.1115,
+      "loss": 0.0906,
       "step": 5127
     },
     {
       "epoch": 93.24,
       "learning_rate": 3.7699999999999995e-05,
-      "loss": 0.1029,
+      "loss": 0.0624,
       "step": 5128
     },
     {
       "epoch": 93.25,
       "learning_rate": 3.76e-05,
-      "loss": 0.0965,
+      "loss": 0.0807,
       "step": 5129
     },
     {
       "epoch": 93.27,
       "learning_rate": 3.75e-05,
-      "loss": 0.0955,
+      "loss": 0.0716,
       "step": 5130
     },
     {
       "epoch": 93.29,
       "learning_rate": 3.74e-05,
-      "loss": 0.0728,
+      "loss": 0.0611,
       "step": 5131
     },
     {
       "epoch": 93.31,
       "learning_rate": 3.73e-05,
-      "loss": 0.1011,
+      "loss": 0.0801,
       "step": 5132
     },
     {
       "epoch": 93.33,
       "learning_rate": 3.7199999999999996e-05,
-      "loss": 0.1126,
+      "loss": 0.1052,
       "step": 5133
     },
     {
       "epoch": 93.35,
       "learning_rate": 3.71e-05,
-      "loss": 0.0664,
+      "loss": 0.0503,
       "step": 5134
     },
     {
       "epoch": 93.36,
       "learning_rate": 3.7e-05,
-      "loss": 0.0802,
+      "loss": 0.0604,
       "step": 5135
     },
     {
       "epoch": 93.38,
       "learning_rate": 3.69e-05,
-      "loss": 0.0808,
+      "loss": 0.0619,
       "step": 5136
     },
     {
       "epoch": 93.4,
       "learning_rate": 3.68e-05,
-      "loss": 0.0663,
+      "loss": 0.0553,
       "step": 5137
     },
     {
       "epoch": 93.42,
       "learning_rate": 3.6700000000000004e-05,
-      "loss": 0.1023,
+      "loss": 0.0738,
       "step": 5138
     },
     {
       "epoch": 93.44,
-      "learning_rate": 3.6700000000000004e-05,
-      "loss": 0.0866,
+      "learning_rate": 3.66e-05,
+      "loss": 0.068,
       "step": 5139
     },
     {
       "epoch": 93.45,
-      "learning_rate": 3.66e-05,
-      "loss": 0.0405,
+      "learning_rate": 3.65e-05,
+      "loss": 0.0211,
       "step": 5140
     },
     {
       "epoch": 93.47,
-      "learning_rate": 3.65e-05,
-      "loss": 0.0946,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 0.0915,
       "step": 5141
     },
     {
       "epoch": 93.49,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 0.0873,
+      "learning_rate": 3.63e-05,
+      "loss": 0.0843,
       "step": 5142
     },
     {
       "epoch": 93.51,
-      "learning_rate": 3.63e-05,
-      "loss": 0.1108,
+      "learning_rate": 3.6200000000000006e-05,
+      "loss": 0.0836,
       "step": 5143
     },
     {
       "epoch": 93.53,
-      "learning_rate": 3.6200000000000006e-05,
-      "loss": 0.0624,
+      "learning_rate": 3.61e-05,
+      "loss": 0.0556,
       "step": 5144
     },
     {
       "epoch": 93.55,
-      "learning_rate": 3.61e-05,
-      "loss": 0.0745,
+      "learning_rate": 3.6e-05,
+      "loss": 0.065,
       "step": 5145
     },
     {
       "epoch": 93.56,
-      "learning_rate": 3.6e-05,
-      "loss": 0.0508,
+      "learning_rate": 3.5900000000000005e-05,
+      "loss": 0.0387,
       "step": 5146
     },
     {
       "epoch": 93.58,
-      "learning_rate": 3.5900000000000005e-05,
-      "loss": 0.1047,
+      "learning_rate": 3.5799999999999996e-05,
+      "loss": 0.0728,
       "step": 5147
     },
     {
       "epoch": 93.6,
-      "learning_rate": 3.5799999999999996e-05,
-      "loss": 0.0724,
+      "learning_rate": 3.57e-05,
+      "loss": 0.0591,
       "step": 5148
     },
     {
       "epoch": 93.62,
-      "learning_rate": 3.57e-05,
-      "loss": 0.0742,
+      "learning_rate": 3.56e-05,
+      "loss": 0.0551,
       "step": 5149
     },
     {
       "epoch": 93.64,
-      "learning_rate": 3.56e-05,
-      "loss": 0.0764,
+      "learning_rate": 3.5499999999999996e-05,
+      "loss": 0.0683,
       "step": 5150
     },
     {
       "epoch": 93.65,
-      "learning_rate": 3.5499999999999996e-05,
-      "loss": 0.0967,
+      "learning_rate": 3.54e-05,
+      "loss": 0.1048,
       "step": 5151
     },
     {
       "epoch": 93.67,
-      "learning_rate": 3.54e-05,
-      "loss": 0.0838,
+      "learning_rate": 3.53e-05,
+      "loss": 0.0478,
       "step": 5152
     },
     {
       "epoch": 93.69,
-      "learning_rate": 3.53e-05,
-      "loss": 0.0742,
+      "learning_rate": 3.52e-05,
+      "loss": 0.0672,
       "step": 5153
     },
     {
       "epoch": 93.71,
-      "learning_rate": 3.52e-05,
-      "loss": 0.0726,
+      "learning_rate": 3.51e-05,
+      "loss": 0.0735,
       "step": 5154
     },
     {
       "epoch": 93.73,
-      "learning_rate": 3.51e-05,
-      "loss": 0.1103,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.0904,
       "step": 5155
     },
     {
       "epoch": 93.75,
-      "learning_rate": 3.5000000000000004e-05,
-      "loss": 0.1114,
+      "learning_rate": 3.49e-05,
+      "loss": 0.0913,
       "step": 5156
     },
     {
       "epoch": 93.76,
-      "learning_rate": 3.49e-05,
-      "loss": 0.0896,
+      "learning_rate": 3.48e-05,
+      "loss": 0.0972,
       "step": 5157
     },
     {
       "epoch": 93.78,
-      "learning_rate": 3.48e-05,
-      "loss": 0.0748,
+      "learning_rate": 3.47e-05,
+      "loss": 0.0459,
       "step": 5158
     },
     {
       "epoch": 93.8,
-      "learning_rate": 3.47e-05,
-      "loss": 0.0737,
+      "learning_rate": 3.46e-05,
+      "loss": 0.0644,
       "step": 5159
     },
     {
       "epoch": 93.82,
-      "learning_rate": 3.46e-05,
-      "loss": 0.0935,
+      "learning_rate": 3.4500000000000005e-05,
+      "loss": 0.0713,
       "step": 5160
     },
     {
       "epoch": 93.84,
-      "learning_rate": 3.4500000000000005e-05,
-      "loss": 0.09,
+      "learning_rate": 3.44e-05,
+      "loss": 0.0725,
       "step": 5161
     },
     {
       "epoch": 93.85,
-      "learning_rate": 3.44e-05,
-      "loss": 0.081,
+      "learning_rate": 3.43e-05,
+      "loss": 0.0666,
       "step": 5162
     },
     {
       "epoch": 93.87,
-      "learning_rate": 3.43e-05,
-      "loss": 0.0667,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 0.0625,
       "step": 5163
     },
     {
       "epoch": 93.89,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 0.0682,
+      "learning_rate": 3.41e-05,
+      "loss": 0.0656,
       "step": 5164
     },
     {
       "epoch": 93.91,
-      "learning_rate": 3.41e-05,
-      "loss": 0.0581,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.0423,
       "step": 5165
     },
     {
       "epoch": 93.93,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.079,
+      "learning_rate": 3.39e-05,
+      "loss": 0.0837,
       "step": 5166
     },
     {
       "epoch": 93.95,
-      "learning_rate": 3.39e-05,
-      "loss": 0.0728,
+      "learning_rate": 3.3799999999999995e-05,
+      "loss": 0.0787,
       "step": 5167
     },
     {
       "epoch": 93.96,
-      "learning_rate": 3.3799999999999995e-05,
-      "loss": 0.0894,
+      "learning_rate": 3.37e-05,
+      "loss": 0.0811,
       "step": 5168
     },
     {
       "epoch": 93.98,
-      "learning_rate": 3.37e-05,
-      "loss": 0.0764,
+      "learning_rate": 3.36e-05,
+      "loss": 0.0749,
       "step": 5169
     },
     {
       "epoch": 94.0,
-      "learning_rate": 3.36e-05,
-      "loss": 0.0125,
+      "learning_rate": 3.35e-05,
+      "loss": 0.0218,
       "step": 5170
     },
     {
       "epoch": 94.02,
-      "learning_rate": 3.35e-05,
-      "loss": 0.0798,
+      "learning_rate": 3.34e-05,
+      "loss": 0.0689,
       "step": 5171
     },
     {
       "epoch": 94.04,
-      "learning_rate": 3.34e-05,
-      "loss": 0.0744,
+      "learning_rate": 3.33e-05,
+      "loss": 0.0729,
       "step": 5172
     },
     {
       "epoch": 94.05,
-      "learning_rate": 3.33e-05,
-      "loss": 0.0719,
+      "learning_rate": 3.32e-05,
+      "loss": 0.0689,
       "step": 5173
     },
     {
       "epoch": 94.07,
-      "learning_rate": 3.32e-05,
-      "loss": 0.0708,
+      "learning_rate": 3.31e-05,
+      "loss": 0.0463,
       "step": 5174
     },
     {
       "epoch": 94.09,
-      "learning_rate": 3.31e-05,
-      "loss": 0.0734,
+      "learning_rate": 3.3e-05,
+      "loss": 0.0673,
       "step": 5175
     },
     {
       "epoch": 94.11,
-      "learning_rate": 3.3e-05,
-      "loss": 0.0451,
+      "learning_rate": 3.29e-05,
+      "loss": 0.0306,
       "step": 5176
     },
     {
       "epoch": 94.13,
-      "learning_rate": 3.29e-05,
-      "loss": 0.0937,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 0.0875,
       "step": 5177
     },
     {
       "epoch": 94.15,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 0.0698,
+      "learning_rate": 3.27e-05,
+      "loss": 0.0751,
       "step": 5178
     },
     {
       "epoch": 94.16,
-      "learning_rate": 3.27e-05,
-      "loss": 0.0612,
+      "learning_rate": 3.26e-05,
+      "loss": 0.0587,
       "step": 5179
     },
     {
       "epoch": 94.18,
-      "learning_rate": 3.26e-05,
-      "loss": 0.1216,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.095,
       "step": 5180
     },
     {
       "epoch": 94.2,
-      "learning_rate": 3.2500000000000004e-05,
+      "learning_rate": 3.24e-05,
       "loss": 0.0912,
       "step": 5181
     },
     {
       "epoch": 94.22,
-      "learning_rate": 3.24e-05,
-      "loss": 0.0596,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 0.0617,
       "step": 5182
     },
     {
       "epoch": 94.24,
-      "learning_rate": 3.2300000000000006e-05,
-      "loss": 0.0825,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 0.082,
       "step": 5183
     },
     {
       "epoch": 94.25,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 0.0727,
+      "learning_rate": 3.2099999999999994e-05,
+      "loss": 0.0685,
       "step": 5184
     },
     {
       "epoch": 94.27,
-      "learning_rate": 3.2099999999999994e-05,
-      "loss": 0.0778,
+      "learning_rate": 3.2e-05,
+      "loss": 0.0772,
       "step": 5185
     },
     {
       "epoch": 94.29,
-      "learning_rate": 3.2e-05,
-      "loss": 0.0835,
+      "learning_rate": 3.1899999999999996e-05,
+      "loss": 0.0738,
       "step": 5186
     },
     {
       "epoch": 94.31,
-      "learning_rate": 3.1899999999999996e-05,
-      "loss": 0.0823,
+      "learning_rate": 3.18e-05,
+      "loss": 0.0728,
       "step": 5187
     },
     {
       "epoch": 94.33,
-      "learning_rate": 3.18e-05,
-      "loss": 0.0574,
+      "learning_rate": 3.17e-05,
+      "loss": 0.0584,
       "step": 5188
     },
     {
       "epoch": 94.35,
-      "learning_rate": 3.17e-05,
-      "loss": 0.0577,
+      "learning_rate": 3.16e-05,
+      "loss": 0.0702,
       "step": 5189
     },
     {
       "epoch": 94.36,
-      "learning_rate": 3.16e-05,
-      "loss": 0.0737,
+      "learning_rate": 3.15e-05,
+      "loss": 0.0689,
       "step": 5190
     },
     {
       "epoch": 94.38,
-      "learning_rate": 3.15e-05,
-      "loss": 0.0633,
+      "learning_rate": 3.14e-05,
+      "loss": 0.0605,
       "step": 5191
     },
     {
       "epoch": 94.4,
-      "learning_rate": 3.14e-05,
-      "loss": 0.0815,
+      "learning_rate": 3.13e-05,
+      "loss": 0.0525,
       "step": 5192
     },
     {
       "epoch": 94.42,
-      "learning_rate": 3.13e-05,
-      "loss": 0.0995,
+      "learning_rate": 3.12e-05,
+      "loss": 0.0897,
       "step": 5193
     },
     {
       "epoch": 94.44,
-      "learning_rate": 3.12e-05,
-      "loss": 0.0483,
+      "learning_rate": 3.11e-05,
+      "loss": 0.0506,
       "step": 5194
     },
     {
       "epoch": 94.45,
-      "learning_rate": 3.11e-05,
-      "loss": 0.0423,
+      "learning_rate": 3.1e-05,
+      "loss": 0.0442,
       "step": 5195
     },
     {
       "epoch": 94.47,
-      "learning_rate": 3.1e-05,
-      "loss": 0.0886,
+      "learning_rate": 3.09e-05,
+      "loss": 0.0894,
       "step": 5196
     },
     {
       "epoch": 94.49,
-      "learning_rate": 3.09e-05,
-      "loss": 0.0761,
+      "learning_rate": 3.08e-05,
+      "loss": 0.0672,
       "step": 5197
     },
     {
       "epoch": 94.51,
-      "learning_rate": 3.08e-05,
-      "loss": 0.0968,
+      "learning_rate": 3.07e-05,
+      "loss": 0.0862,
       "step": 5198
     },
     {
       "epoch": 94.53,
-      "learning_rate": 3.07e-05,
-      "loss": 0.0775,
+      "learning_rate": 3.06e-05,
+      "loss": 0.0743,
       "step": 5199
     },
     {
       "epoch": 94.55,
-      "learning_rate": 3.06e-05,
-      "loss": 0.0589,
+      "learning_rate": 3.05e-05,
+      "loss": 0.0625,
       "step": 5200
     },
     {
       "epoch": 94.56,
-      "learning_rate": 3.05e-05,
-      "loss": 0.0685,
+      "learning_rate": 3.04e-05,
+      "loss": 0.0562,
       "step": 5201
     },
     {
       "epoch": 94.58,
-      "learning_rate": 3.04e-05,
-      "loss": 0.0632,
+      "learning_rate": 3.03e-05,
+      "loss": 0.0614,
       "step": 5202
     },
     {
       "epoch": 94.6,
-      "learning_rate": 3.03e-05,
-      "loss": 0.0935,
+      "learning_rate": 3.0200000000000002e-05,
+      "loss": 0.0677,
       "step": 5203
     },
     {
       "epoch": 94.62,
-      "learning_rate": 3.0200000000000002e-05,
-      "loss": 0.0764,
+      "learning_rate": 3.01e-05,
+      "loss": 0.0943,
       "step": 5204
     },
     {
       "epoch": 94.64,
-      "learning_rate": 3.01e-05,
-      "loss": 0.0678,
+      "learning_rate": 3e-05,
+      "loss": 0.0445,
       "step": 5205
     },
     {
       "epoch": 94.65,
-      "learning_rate": 3e-05,
-      "loss": 0.0837,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 0.0802,
       "step": 5206
     },
     {
       "epoch": 94.67,
-      "learning_rate": 2.9900000000000002e-05,
-      "loss": 0.0861,
+      "learning_rate": 2.98e-05,
+      "loss": 0.0995,
       "step": 5207
     },
     {
       "epoch": 94.69,
-      "learning_rate": 2.98e-05,
-      "loss": 0.0774,
+      "learning_rate": 2.97e-05,
+      "loss": 0.0608,
       "step": 5208
     },
     {
       "epoch": 94.71,
-      "learning_rate": 2.97e-05,
-      "loss": 0.0983,
+      "learning_rate": 2.96e-05,
+      "loss": 0.081,
       "step": 5209
     },
     {
       "epoch": 94.73,
-      "learning_rate": 2.96e-05,
-      "loss": 0.074,
+      "learning_rate": 2.95e-05,
+      "loss": 0.0577,
       "step": 5210
     },
     {
       "epoch": 94.75,
-      "learning_rate": 2.95e-05,
-      "loss": 0.0825,
+      "learning_rate": 2.94e-05,
+      "loss": 0.0587,
       "step": 5211
     },
     {
       "epoch": 94.76,
-      "learning_rate": 2.94e-05,
-      "loss": 0.0773,
+      "learning_rate": 2.93e-05,
+      "loss": 0.0476,
       "step": 5212
     },
     {
       "epoch": 94.78,
-      "learning_rate": 2.93e-05,
-      "loss": 0.0465,
+      "learning_rate": 2.92e-05,
+      "loss": 0.0374,
       "step": 5213
     },
     {
       "epoch": 94.8,
-      "learning_rate": 2.92e-05,
-      "loss": 0.0562,
+      "learning_rate": 2.9100000000000003e-05,
+      "loss": 0.0375,
       "step": 5214
     },
     {
       "epoch": 94.82,
-      "learning_rate": 2.9100000000000003e-05,
-      "loss": 0.1079,
+      "learning_rate": 2.9000000000000004e-05,
+      "loss": 0.0838,
       "step": 5215
     },
     {
       "epoch": 94.84,
-      "learning_rate": 2.9000000000000004e-05,
-      "loss": 0.0454,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 0.0414,
       "step": 5216
     },
     {
       "epoch": 94.85,
-      "learning_rate": 2.8899999999999998e-05,
-      "loss": 0.068,
+      "learning_rate": 2.88e-05,
+      "loss": 0.0694,
       "step": 5217
     },
     {
       "epoch": 94.87,
-      "learning_rate": 2.88e-05,
-      "loss": 0.0558,
+      "learning_rate": 2.87e-05,
+      "loss": 0.0654,
       "step": 5218
     },
     {
       "epoch": 94.89,
-      "learning_rate": 2.87e-05,
-      "loss": 0.0735,
+      "learning_rate": 2.86e-05,
+      "loss": 0.0676,
       "step": 5219
     },
     {
       "epoch": 94.91,
-      "learning_rate": 2.86e-05,
-      "loss": 0.0484,
+      "learning_rate": 2.85e-05,
+      "loss": 0.0371,
       "step": 5220
     },
     {
       "epoch": 94.93,
-      "learning_rate": 2.85e-05,
-      "loss": 0.1131,
+      "learning_rate": 2.8400000000000003e-05,
+      "loss": 0.0947,
       "step": 5221
     },
     {
       "epoch": 94.95,
-      "learning_rate": 2.8400000000000003e-05,
-      "loss": 0.0891,
+      "learning_rate": 2.83e-05,
+      "loss": 0.0663,
       "step": 5222
     },
     {
       "epoch": 94.96,
-      "learning_rate": 2.83e-05,
-      "loss": 0.0722,
+      "learning_rate": 2.82e-05,
+      "loss": 0.0704,
       "step": 5223
     },
     {
       "epoch": 94.98,
-      "learning_rate": 2.82e-05,
-      "loss": 0.0703,
+      "learning_rate": 2.8100000000000002e-05,
+      "loss": 0.0613,
       "step": 5224
     },
     {
       "epoch": 95.0,
-      "learning_rate": 2.8100000000000002e-05,
-      "loss": 0.0825,
+      "learning_rate": 2.8e-05,
+      "loss": 0.0879,
       "step": 5225
     },
     {
       "epoch": 95.02,
-      "learning_rate": 2.8e-05,
-      "loss": 0.0983,
+      "learning_rate": 2.79e-05,
+      "loss": 0.0781,
       "step": 5226
     },
     {
       "epoch": 95.04,
-      "learning_rate": 2.79e-05,
-      "loss": 0.0868,
+      "learning_rate": 2.7799999999999998e-05,
+      "loss": 0.094,
       "step": 5227
     },
     {
       "epoch": 95.05,
-      "learning_rate": 2.7799999999999998e-05,
-      "loss": 0.0709,
+      "learning_rate": 2.77e-05,
+      "loss": 0.0518,
       "step": 5228
     },
     {
       "epoch": 95.07,
-      "learning_rate": 2.77e-05,
-      "loss": 0.0839,
+      "learning_rate": 2.76e-05,
+      "loss": 0.0829,
       "step": 5229
     },
     {
       "epoch": 95.09,
-      "learning_rate": 2.76e-05,
-      "loss": 0.0641,
+      "learning_rate": 2.75e-05,
+      "loss": 0.0674,
       "step": 5230
     },
     {
       "epoch": 95.11,
-      "learning_rate": 2.75e-05,
-      "loss": 0.0852,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 0.0727,
       "step": 5231
     },
     {
       "epoch": 95.13,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 0.0672,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 0.0757,
       "step": 5232
     },
     {
       "epoch": 95.15,
-      "learning_rate": 2.7300000000000003e-05,
-      "loss": 0.0852,
+      "learning_rate": 2.72e-05,
+      "loss": 0.067,
       "step": 5233
     },
     {
       "epoch": 95.16,
-      "learning_rate": 2.72e-05,
-      "loss": 0.0753,
+      "learning_rate": 2.7099999999999998e-05,
+      "loss": 0.0551,
       "step": 5234
     },
     {
       "epoch": 95.18,
-      "learning_rate": 2.7099999999999998e-05,
-      "loss": 0.0484,
+      "learning_rate": 2.7e-05,
+      "loss": 0.0735,
       "step": 5235
     },
     {
       "epoch": 95.2,
-      "learning_rate": 2.7e-05,
-      "loss": 0.0684,
+      "learning_rate": 2.69e-05,
+      "loss": 0.0797,
       "step": 5236
     },
     {
       "epoch": 95.22,
-      "learning_rate": 2.69e-05,
-      "loss": 0.0664,
+      "learning_rate": 2.68e-05,
+      "loss": 0.0921,
       "step": 5237
     },
     {
       "epoch": 95.24,
-      "learning_rate": 2.68e-05,
-      "loss": 0.0668,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 0.0707,
       "step": 5238
     },
     {
       "epoch": 95.25,
-      "learning_rate": 2.6700000000000002e-05,
-      "loss": 0.1148,
+      "learning_rate": 2.66e-05,
+      "loss": 0.1219,
       "step": 5239
     },
     {
       "epoch": 95.27,
-      "learning_rate": 2.66e-05,
-      "loss": 0.0794,
+      "learning_rate": 2.65e-05,
+      "loss": 0.0658,
       "step": 5240
     },
     {
       "epoch": 95.29,
-      "learning_rate": 2.65e-05,
-      "loss": 0.0701,
+      "learning_rate": 2.64e-05,
+      "loss": 0.0492,
       "step": 5241
     },
     {
       "epoch": 95.31,
-      "learning_rate": 2.64e-05,
-      "loss": 0.0884,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 0.0662,
       "step": 5242
     },
     {
       "epoch": 95.33,
-      "learning_rate": 2.6300000000000002e-05,
-      "loss": 0.059,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 0.0539,
       "step": 5243
     },
     {
       "epoch": 95.35,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 0.0533,
+      "learning_rate": 2.61e-05,
+      "loss": 0.0515,
       "step": 5244
     },
     {
       "epoch": 95.36,
-      "learning_rate": 2.61e-05,
-      "loss": 0.0857,
+      "learning_rate": 2.6e-05,
+      "loss": 0.0564,
       "step": 5245
     },
     {
       "epoch": 95.38,
-      "learning_rate": 2.6e-05,
-      "loss": 0.0673,
+      "learning_rate": 2.59e-05,
+      "loss": 0.0605,
       "step": 5246
     },
     {
       "epoch": 95.4,
-      "learning_rate": 2.59e-05,
-      "loss": 0.0675,
+      "learning_rate": 2.58e-05,
+      "loss": 0.0511,
       "step": 5247
     },
     {
       "epoch": 95.42,
-      "learning_rate": 2.58e-05,
-      "loss": 0.0869,
+      "learning_rate": 2.57e-05,
+      "loss": 0.0703,
       "step": 5248
     },
     {
       "epoch": 95.44,
-      "learning_rate": 2.57e-05,
-      "loss": 0.0653,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.0443,
       "step": 5249
     },
     {
       "epoch": 95.45,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 0.074,
+      "learning_rate": 2.55e-05,
+      "loss": 0.0741,
       "step": 5250
     },
     {
       "epoch": 95.47,
-      "learning_rate": 2.55e-05,
-      "loss": 0.0786,
+      "learning_rate": 2.54e-05,
+      "loss": 0.0777,
       "step": 5251
     },
     {
       "epoch": 95.49,
-      "learning_rate": 2.54e-05,
-      "loss": 0.0888,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 0.0812,
       "step": 5252
     },
     {
       "epoch": 95.51,
-      "learning_rate": 2.5300000000000002e-05,
-      "loss": 0.0791,
+      "learning_rate": 2.52e-05,
+      "loss": 0.0771,
       "step": 5253
     },
     {
       "epoch": 95.53,
-      "learning_rate": 2.52e-05,
-      "loss": 0.0768,
+      "learning_rate": 2.51e-05,
+      "loss": 0.0648,
       "step": 5254
     },
     {
       "epoch": 95.55,
-      "learning_rate": 2.51e-05,
-      "loss": 0.0781,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0639,
       "step": 5255
     },
     {
       "epoch": 95.56,
-      "learning_rate": 2.5e-05,
-      "loss": 0.048,
+      "learning_rate": 2.49e-05,
+      "loss": 0.0458,
       "step": 5256
     },
     {
       "epoch": 95.58,
-      "learning_rate": 2.49e-05,
-      "loss": 0.0821,
+      "learning_rate": 2.48e-05,
+      "loss": 0.0767,
       "step": 5257
     },
     {
       "epoch": 95.6,
-      "learning_rate": 2.48e-05,
-      "loss": 0.0708,
+      "learning_rate": 2.47e-05,
+      "loss": 0.0693,
       "step": 5258
     },
     {
       "epoch": 95.62,
-      "learning_rate": 2.47e-05,
-      "loss": 0.0795,
+      "learning_rate": 2.46e-05,
+      "loss": 0.069,
       "step": 5259
     },
     {
       "epoch": 95.64,
-      "learning_rate": 2.46e-05,
-      "loss": 0.0774,
+      "learning_rate": 2.4500000000000003e-05,
+      "loss": 0.0758,
       "step": 5260
     },
     {
       "epoch": 95.65,
-      "learning_rate": 2.4500000000000003e-05,
-      "loss": 0.0499,
+      "learning_rate": 2.4400000000000004e-05,
+      "loss": 0.0415,
       "step": 5261
     },
     {
       "epoch": 95.67,
-      "learning_rate": 2.4400000000000004e-05,
-      "loss": 0.0781,
+      "learning_rate": 2.4299999999999998e-05,
+      "loss": 0.074,
       "step": 5262
     },
     {
       "epoch": 95.69,
-      "learning_rate": 2.4299999999999998e-05,
-      "loss": 0.0916,
+      "learning_rate": 2.42e-05,
+      "loss": 0.0938,
       "step": 5263
     },
     {
       "epoch": 95.71,
-      "learning_rate": 2.42e-05,
-      "loss": 0.1017,
+      "learning_rate": 2.41e-05,
+      "loss": 0.0953,
       "step": 5264
     },
     {
       "epoch": 95.73,
-      "learning_rate": 2.41e-05,
-      "loss": 0.0574,
+      "learning_rate": 2.4e-05,
+      "loss": 0.049,
       "step": 5265
     },
     {
       "epoch": 95.75,
-      "learning_rate": 2.4e-05,
-      "loss": 0.0776,
+      "learning_rate": 2.39e-05,
+      "loss": 0.0563,
       "step": 5266
     },
     {
       "epoch": 95.76,
-      "learning_rate": 2.39e-05,
-      "loss": 0.0552,
+      "learning_rate": 2.3800000000000003e-05,
+      "loss": 0.0672,
       "step": 5267
     },
     {
       "epoch": 95.78,
-      "learning_rate": 2.3800000000000003e-05,
-      "loss": 0.0677,
+      "learning_rate": 2.37e-05,
+      "loss": 0.0596,
       "step": 5268
     },
     {
       "epoch": 95.8,
-      "learning_rate": 2.37e-05,
-      "loss": 0.0556,
+      "learning_rate": 2.36e-05,
+      "loss": 0.0499,
       "step": 5269
     },
     {
       "epoch": 95.82,
-      "learning_rate": 2.36e-05,
-      "loss": 0.0922,
+      "learning_rate": 2.3500000000000002e-05,
+      "loss": 0.0846,
       "step": 5270
     },
     {
       "epoch": 95.84,
-      "learning_rate": 2.3500000000000002e-05,
-      "loss": 0.0593,
+      "learning_rate": 2.34e-05,
+      "loss": 0.0559,
       "step": 5271
     },
     {
       "epoch": 95.85,
-      "learning_rate": 2.34e-05,
-      "loss": 0.0404,
+      "learning_rate": 2.33e-05,
+      "loss": 0.0531,
       "step": 5272
     },
     {
       "epoch": 95.87,
-      "learning_rate": 2.33e-05,
-      "loss": 0.0961,
+      "learning_rate": 2.3199999999999998e-05,
+      "loss": 0.0701,
       "step": 5273
     },
     {
       "epoch": 95.89,
-      "learning_rate": 2.3199999999999998e-05,
-      "loss": 0.0667,
+      "learning_rate": 2.31e-05,
+      "loss": 0.058,
       "step": 5274
     },
     {
       "epoch": 95.91,
-      "learning_rate": 2.31e-05,
-      "loss": 0.0845,
+      "learning_rate": 2.3e-05,
+      "loss": 0.0615,
       "step": 5275
     },
     {
       "epoch": 95.93,
-      "learning_rate": 2.3e-05,
-      "loss": 0.0757,
+      "learning_rate": 2.29e-05,
+      "loss": 0.0765,
       "step": 5276
     },
     {
       "epoch": 95.95,
-      "learning_rate": 2.29e-05,
-      "loss": 0.0815,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.0744,
       "step": 5277
     },
     {
       "epoch": 95.96,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 0.0558,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 0.0344,
       "step": 5278
     },
     {
       "epoch": 95.98,
-      "learning_rate": 2.2700000000000003e-05,
-      "loss": 0.0578,
+      "learning_rate": 2.26e-05,
+      "loss": 0.0523,
       "step": 5279
     },
     {
       "epoch": 96.0,
-      "learning_rate": 2.26e-05,
-      "loss": 0.0647,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 0.0438,
       "step": 5280
     },
     {
       "epoch": 96.02,
-      "learning_rate": 2.2499999999999998e-05,
-      "loss": 0.0982,
+      "learning_rate": 2.24e-05,
+      "loss": 0.0843,
       "step": 5281
     },
     {
       "epoch": 96.04,
-      "learning_rate": 2.24e-05,
-      "loss": 0.0684,
+      "learning_rate": 2.23e-05,
+      "loss": 0.0458,
       "step": 5282
     },
     {
       "epoch": 96.05,
-      "learning_rate": 2.23e-05,
-      "loss": 0.058,
+      "learning_rate": 2.22e-05,
+      "loss": 0.0685,
       "step": 5283
     },
     {
       "epoch": 96.07,
-      "learning_rate": 2.22e-05,
-      "loss": 0.0716,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 0.0642,
       "step": 5284
     },
     {
       "epoch": 96.09,
-      "learning_rate": 2.2100000000000002e-05,
-      "loss": 0.0658,
+      "learning_rate": 2.2e-05,
+      "loss": 0.0469,
       "step": 5285
     },
     {
       "epoch": 96.11,
-      "learning_rate": 2.2e-05,
-      "loss": 0.0912,
+      "learning_rate": 2.19e-05,
+      "loss": 0.0934,
       "step": 5286
     },
     {
       "epoch": 96.13,
-      "learning_rate": 2.19e-05,
-      "loss": 0.0515,
+      "learning_rate": 2.18e-05,
+      "loss": 0.0398,
       "step": 5287
     },
     {
       "epoch": 96.15,
-      "learning_rate": 2.18e-05,
-      "loss": 0.0503,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 0.065,
       "step": 5288
     },
     {
       "epoch": 96.16,
-      "learning_rate": 2.1700000000000002e-05,
-      "loss": 0.0876,
+      "learning_rate": 2.1600000000000003e-05,
+      "loss": 0.0778,
       "step": 5289
     },
     {
       "epoch": 96.18,
-      "learning_rate": 2.1600000000000003e-05,
-      "loss": 0.1019,
+      "learning_rate": 2.1499999999999997e-05,
+      "loss": 0.0925,
       "step": 5290
     },
     {
       "epoch": 96.2,
-      "learning_rate": 2.1499999999999997e-05,
-      "loss": 0.0813,
+      "learning_rate": 2.14e-05,
+      "loss": 0.0749,
       "step": 5291
     },
     {
       "epoch": 96.22,
-      "learning_rate": 2.14e-05,
-      "loss": 0.0529,
+      "learning_rate": 2.13e-05,
+      "loss": 0.0484,
       "step": 5292
     },
     {
       "epoch": 96.24,
-      "learning_rate": 2.13e-05,
-      "loss": 0.0792,
+      "learning_rate": 2.12e-05,
+      "loss": 0.0682,
       "step": 5293
     },
     {
       "epoch": 96.25,
-      "learning_rate": 2.12e-05,
-      "loss": 0.0616,
+      "learning_rate": 2.11e-05,
+      "loss": 0.07,
       "step": 5294
     },
     {
       "epoch": 96.27,
-      "learning_rate": 2.11e-05,
-      "loss": 0.0584,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 0.0542,
       "step": 5295
     },
     {
       "epoch": 96.29,
-      "learning_rate": 2.1000000000000002e-05,
-      "loss": 0.0724,
+      "learning_rate": 2.09e-05,
+      "loss": 0.0655,
       "step": 5296
     },
     {
       "epoch": 96.31,
-      "learning_rate": 2.09e-05,
-      "loss": 0.0475,
+      "learning_rate": 2.08e-05,
+      "loss": 0.058,
       "step": 5297
     },
     {
       "epoch": 96.33,
-      "learning_rate": 2.08e-05,
-      "loss": 0.0797,
+      "learning_rate": 2.0700000000000002e-05,
+      "loss": 0.0751,
       "step": 5298
     },
     {
       "epoch": 96.35,
-      "learning_rate": 2.0700000000000002e-05,
-      "loss": 0.0607,
+      "learning_rate": 2.06e-05,
+      "loss": 0.0574,
       "step": 5299
     },
     {
       "epoch": 96.36,
-      "learning_rate": 2.06e-05,
-      "loss": 0.0915,
+      "learning_rate": 2.05e-05,
+      "loss": 0.0714,
       "step": 5300
     },
     {
       "epoch": 96.38,
-      "learning_rate": 2.05e-05,
-      "loss": 0.0528,
+      "learning_rate": 2.04e-05,
+      "loss": 0.041,
       "step": 5301
     },
     {
       "epoch": 96.4,
-      "learning_rate": 2.04e-05,
-      "loss": 0.0514,
+      "learning_rate": 2.03e-05,
+      "loss": 0.0394,
       "step": 5302
     },
     {
       "epoch": 96.42,
-      "learning_rate": 2.03e-05,
-      "loss": 0.1005,
+      "learning_rate": 2.02e-05,
+      "loss": 0.0685,
       "step": 5303
     },
     {
       "epoch": 96.44,
-      "learning_rate": 2.02e-05,
-      "loss": 0.0719,
+      "learning_rate": 2.01e-05,
+      "loss": 0.0534,
       "step": 5304
     },
     {
       "epoch": 96.45,
-      "learning_rate": 2.01e-05,
-      "loss": 0.0487,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
       "step": 5305
     },
     {
       "epoch": 96.47,
-      "learning_rate": 2e-05,
-      "loss": 0.0805,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 0.0643,
       "step": 5306
     },
     {
       "epoch": 96.49,
-      "learning_rate": 1.9900000000000003e-05,
-      "loss": 0.0893,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.0803,
       "step": 5307
     },
     {
       "epoch": 96.51,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 0.0841,
+      "learning_rate": 1.9699999999999998e-05,
+      "loss": 0.0801,
       "step": 5308
     },
     {
       "epoch": 96.53,
-      "learning_rate": 1.9699999999999998e-05,
-      "loss": 0.0529,
+      "learning_rate": 1.96e-05,
+      "loss": 0.0493,
       "step": 5309
     },
     {
       "epoch": 96.55,
-      "learning_rate": 1.96e-05,
-      "loss": 0.065,
+      "learning_rate": 1.95e-05,
+      "loss": 0.0507,
       "step": 5310
     },
     {
       "epoch": 96.56,
-      "learning_rate": 1.95e-05,
-      "loss": 0.0698,
+      "learning_rate": 1.94e-05,
+      "loss": 0.0671,
       "step": 5311
     },
     {
       "epoch": 96.58,
-      "learning_rate": 1.94e-05,
-      "loss": 0.1001,
+      "learning_rate": 1.93e-05,
+      "loss": 0.0769,
       "step": 5312
     },
     {
       "epoch": 96.6,
-      "learning_rate": 1.93e-05,
-      "loss": 0.0846,
+      "learning_rate": 1.92e-05,
+      "loss": 0.098,
       "step": 5313
     },
     {
       "epoch": 96.62,
-      "learning_rate": 1.92e-05,
-      "loss": 0.0917,
+      "learning_rate": 1.91e-05,
+      "loss": 0.0894,
       "step": 5314
     },
     {
       "epoch": 96.64,
-      "learning_rate": 1.91e-05,
-      "loss": 0.1144,
+      "learning_rate": 1.9e-05,
+      "loss": 0.0991,
       "step": 5315
     },
     {
       "epoch": 96.65,
-      "learning_rate": 1.9e-05,
-      "loss": 0.0493,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 0.0426,
       "step": 5316
     },
     {
       "epoch": 96.67,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 0.0766,
+      "learning_rate": 1.88e-05,
+      "loss": 0.0557,
       "step": 5317
     },
     {
       "epoch": 96.69,
-      "learning_rate": 1.88e-05,
-      "loss": 0.0891,
+      "learning_rate": 1.87e-05,
+      "loss": 0.0767,
       "step": 5318
     },
     {
       "epoch": 96.71,
-      "learning_rate": 1.87e-05,
-      "loss": 0.0591,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 0.0505,
       "step": 5319
     },
     {
       "epoch": 96.73,
-      "learning_rate": 1.8599999999999998e-05,
-      "loss": 0.0827,
+      "learning_rate": 1.85e-05,
+      "loss": 0.0678,
       "step": 5320
     },
     {
       "epoch": 96.75,
-      "learning_rate": 1.85e-05,
-      "loss": 0.0781,
+      "learning_rate": 1.84e-05,
+      "loss": 0.0643,
       "step": 5321
     },
     {
       "epoch": 96.76,
-      "learning_rate": 1.84e-05,
-      "loss": 0.0834,
+      "learning_rate": 1.83e-05,
+      "loss": 0.094,
       "step": 5322
     },
     {
       "epoch": 96.78,
-      "learning_rate": 1.83e-05,
-      "loss": 0.0857,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.0508,
       "step": 5323
     },
     {
       "epoch": 96.8,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 0.0851,
+      "learning_rate": 1.8100000000000003e-05,
+      "loss": 0.0573,
       "step": 5324
     },
     {
       "epoch": 96.82,
-      "learning_rate": 1.8100000000000003e-05,
-      "loss": 0.1042,
+      "learning_rate": 1.8e-05,
+      "loss": 0.0738,
       "step": 5325
     },
     {
       "epoch": 96.84,
-      "learning_rate": 1.8e-05,
-      "loss": 0.0567,
+      "learning_rate": 1.7899999999999998e-05,
+      "loss": 0.0415,
       "step": 5326
     },
     {
       "epoch": 96.85,
-      "learning_rate": 1.7899999999999998e-05,
-      "loss": 0.0868,
+      "learning_rate": 1.78e-05,
+      "loss": 0.0642,
       "step": 5327
     },
     {
       "epoch": 96.87,
-      "learning_rate": 1.78e-05,
-      "loss": 0.0811,
+      "learning_rate": 1.77e-05,
+      "loss": 0.0789,
       "step": 5328
     },
     {
       "epoch": 96.89,
-      "learning_rate": 1.77e-05,
-      "loss": 0.0501,
+      "learning_rate": 1.76e-05,
+      "loss": 0.0532,
       "step": 5329
     },
     {
       "epoch": 96.91,
-      "learning_rate": 1.76e-05,
-      "loss": 0.035,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.0627,
       "step": 5330
     },
     {
       "epoch": 96.93,
-      "learning_rate": 1.7500000000000002e-05,
-      "loss": 0.0936,
+      "learning_rate": 1.74e-05,
+      "loss": 0.0942,
       "step": 5331
     },
     {
       "epoch": 96.95,
-      "learning_rate": 1.74e-05,
-      "loss": 0.0917,
+      "learning_rate": 1.73e-05,
+      "loss": 0.081,
       "step": 5332
     },
     {
       "epoch": 96.96,
-      "learning_rate": 1.73e-05,
-      "loss": 0.0872,
+      "learning_rate": 1.72e-05,
+      "loss": 0.0706,
       "step": 5333
     },
     {
       "epoch": 96.98,
-      "learning_rate": 1.72e-05,
-      "loss": 0.0658,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.0377,
       "step": 5334
     },
     {
       "epoch": 97.0,
-      "learning_rate": 1.7100000000000002e-05,
-      "loss": 0.1452,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.09,
       "step": 5335
     },
     {
       "epoch": 97.02,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0699,
+      "learning_rate": 1.6899999999999997e-05,
+      "loss": 0.082,
       "step": 5336
     },
     {
       "epoch": 97.04,
-      "learning_rate": 1.6899999999999997e-05,
-      "loss": 0.0758,
+      "learning_rate": 1.68e-05,
+      "loss": 0.0685,
       "step": 5337
     },
     {
       "epoch": 97.05,
-      "learning_rate": 1.68e-05,
-      "loss": 0.1042,
+      "learning_rate": 1.67e-05,
+      "loss": 0.0866,
       "step": 5338
     },
     {
       "epoch": 97.07,
-      "learning_rate": 1.67e-05,
-      "loss": 0.0621,
+      "learning_rate": 1.66e-05,
+      "loss": 0.0587,
       "step": 5339
     },
     {
       "epoch": 97.09,
-      "learning_rate": 1.66e-05,
-      "loss": 0.069,
+      "learning_rate": 1.65e-05,
+      "loss": 0.0667,
       "step": 5340
     },
     {
       "epoch": 97.11,
-      "learning_rate": 1.65e-05,
-      "loss": 0.0477,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 0.0415,
       "step": 5341
     },
     {
       "epoch": 97.13,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 0.0935,
+      "learning_rate": 1.63e-05,
+      "loss": 0.0799,
       "step": 5342
     },
     {
       "epoch": 97.15,
-      "learning_rate": 1.63e-05,
-      "loss": 0.0807,
+      "learning_rate": 1.62e-05,
+      "loss": 0.0784,
       "step": 5343
     },
     {
       "epoch": 97.16,
-      "learning_rate": 1.62e-05,
-      "loss": 0.0595,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.0608,
       "step": 5344
     },
     {
       "epoch": 97.18,
-      "learning_rate": 1.6100000000000002e-05,
-      "loss": 0.0754,
+      "learning_rate": 1.6e-05,
+      "loss": 0.0724,
       "step": 5345
     },
     {
       "epoch": 97.2,
-      "learning_rate": 1.6e-05,
-      "loss": 0.0931,
+      "learning_rate": 1.59e-05,
+      "loss": 0.0652,
       "step": 5346
     },
     {
       "epoch": 97.22,
-      "learning_rate": 1.59e-05,
-      "loss": 0.0529,
+      "learning_rate": 1.58e-05,
+      "loss": 0.0405,
       "step": 5347
     },
     {
       "epoch": 97.24,
-      "learning_rate": 1.58e-05,
-      "loss": 0.0808,
+      "learning_rate": 1.57e-05,
+      "loss": 0.0643,
       "step": 5348
     },
     {
       "epoch": 97.25,
-      "learning_rate": 1.57e-05,
-      "loss": 0.0925,
+      "learning_rate": 1.56e-05,
+      "loss": 0.0647,
       "step": 5349
     },
     {
       "epoch": 97.27,
-      "learning_rate": 1.56e-05,
-      "loss": 0.0877,
+      "learning_rate": 1.55e-05,
+      "loss": 0.0805,
       "step": 5350
     },
     {
       "epoch": 97.29,
-      "learning_rate": 1.55e-05,
-      "loss": 0.0684,
+      "learning_rate": 1.54e-05,
+      "loss": 0.0626,
       "step": 5351
     },
     {
       "epoch": 97.31,
-      "learning_rate": 1.54e-05,
-      "loss": 0.0798,
+      "learning_rate": 1.53e-05,
+      "loss": 0.0708,
       "step": 5352
     },
     {
       "epoch": 97.33,
-      "learning_rate": 1.53e-05,
-      "loss": 0.0591,
+      "learning_rate": 1.52e-05,
+      "loss": 0.0665,
       "step": 5353
     },
     {
       "epoch": 97.35,
-      "learning_rate": 1.52e-05,
-      "loss": 0.058,
+      "learning_rate": 1.5100000000000001e-05,
+      "loss": 0.0519,
       "step": 5354
     },
     {
       "epoch": 97.36,
-      "learning_rate": 1.5100000000000001e-05,
-      "loss": 0.0869,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0825,
       "step": 5355
     },
     {
       "epoch": 97.38,
-      "learning_rate": 1.5e-05,
-      "loss": 0.0655,
+      "learning_rate": 1.49e-05,
+      "loss": 0.0577,
       "step": 5356
     },
     {
       "epoch": 97.4,
-      "learning_rate": 1.49e-05,
-      "loss": 0.0763,
+      "learning_rate": 1.48e-05,
+      "loss": 0.0662,
       "step": 5357
     },
     {
       "epoch": 97.42,
-      "learning_rate": 1.48e-05,
-      "loss": 0.0455,
+      "learning_rate": 1.47e-05,
+      "loss": 0.0444,
       "step": 5358
     },
     {
       "epoch": 97.44,
-      "learning_rate": 1.47e-05,
-      "loss": 0.0638,
+      "learning_rate": 1.46e-05,
+      "loss": 0.0604,
       "step": 5359
     },
     {
       "epoch": 97.45,
-      "learning_rate": 1.46e-05,
-      "loss": 0.079,
+      "learning_rate": 1.4500000000000002e-05,
+      "loss": 0.0699,
       "step": 5360
     },
     {
       "epoch": 97.47,
-      "learning_rate": 1.4500000000000002e-05,
-      "loss": 0.1082,
+      "learning_rate": 1.44e-05,
+      "loss": 0.0982,
       "step": 5361
     },
     {
       "epoch": 97.49,
-      "learning_rate": 1.44e-05,
-      "loss": 0.069,
+      "learning_rate": 1.43e-05,
+      "loss": 0.0569,
       "step": 5362
     },
     {
       "epoch": 97.51,
-      "learning_rate": 1.43e-05,
-      "loss": 0.0921,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.0561,
       "step": 5363
     },
     {
       "epoch": 97.53,
-      "learning_rate": 1.4200000000000001e-05,
-      "loss": 0.0552,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0513,
       "step": 5364
     },
     {
       "epoch": 97.55,
-      "learning_rate": 1.41e-05,
-      "loss": 0.0609,
+      "learning_rate": 1.4e-05,
+      "loss": 0.06,
       "step": 5365
     },
     {
       "epoch": 97.56,
-      "learning_rate": 1.4e-05,
-      "loss": 0.0861,
+      "learning_rate": 1.3899999999999999e-05,
+      "loss": 0.065,
       "step": 5366
     },
     {
       "epoch": 97.58,
-      "learning_rate": 1.3899999999999999e-05,
-      "loss": 0.0953,
+      "learning_rate": 1.38e-05,
+      "loss": 0.0839,
       "step": 5367
     },
     {
       "epoch": 97.6,
-      "learning_rate": 1.38e-05,
-      "loss": 0.0776,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 0.0531,
       "step": 5368
     },
     {
       "epoch": 97.62,
-      "learning_rate": 1.3700000000000001e-05,
-      "loss": 0.0815,
+      "learning_rate": 1.36e-05,
+      "loss": 0.0753,
       "step": 5369
     },
     {
       "epoch": 97.64,
-      "learning_rate": 1.36e-05,
-      "loss": 0.0777,
+      "learning_rate": 1.35e-05,
+      "loss": 0.0618,
       "step": 5370
     },
     {
       "epoch": 97.65,
-      "learning_rate": 1.35e-05,
-      "loss": 0.074,
+      "learning_rate": 1.34e-05,
+      "loss": 0.0894,
       "step": 5371
     },
     {
       "epoch": 97.67,
-      "learning_rate": 1.34e-05,
-      "loss": 0.0685,
+      "learning_rate": 1.33e-05,
+      "loss": 0.0383,
       "step": 5372
     },
     {
       "epoch": 97.69,
-      "learning_rate": 1.33e-05,
-      "loss": 0.0762,
+      "learning_rate": 1.32e-05,
+      "loss": 0.0655,
       "step": 5373
     },
     {
       "epoch": 97.71,
-      "learning_rate": 1.32e-05,
-      "loss": 0.0785,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.0917,
       "step": 5374
     },
     {
       "epoch": 97.73,
-      "learning_rate": 1.3100000000000002e-05,
-      "loss": 0.0539,
+      "learning_rate": 1.3e-05,
+      "loss": 0.0554,
       "step": 5375
     },
     {
       "epoch": 97.75,
-      "learning_rate": 1.3e-05,
-      "loss": 0.0663,
+      "learning_rate": 1.29e-05,
+      "loss": 0.0655,
       "step": 5376
     },
     {
       "epoch": 97.76,
-      "learning_rate": 1.29e-05,
-      "loss": 0.0774,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.0618,
       "step": 5377
     },
     {
       "epoch": 97.78,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 0.0834,
+      "learning_rate": 1.27e-05,
+      "loss": 0.0623,
       "step": 5378
     },
     {
       "epoch": 97.8,
-      "learning_rate": 1.27e-05,
-      "loss": 0.0615,
+      "learning_rate": 1.26e-05,
+      "loss": 0.0559,
       "step": 5379
     },
     {
       "epoch": 97.82,
-      "learning_rate": 1.26e-05,
-      "loss": 0.0746,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0804,
       "step": 5380
     },
     {
       "epoch": 97.84,
-      "learning_rate": 1.25e-05,
-      "loss": 0.0925,
+      "learning_rate": 1.24e-05,
+      "loss": 0.0849,
       "step": 5381
     },
     {
       "epoch": 97.85,
-      "learning_rate": 1.24e-05,
-      "loss": 0.1077,
+      "learning_rate": 1.23e-05,
+      "loss": 0.0812,
       "step": 5382
     },
     {
       "epoch": 97.87,
-      "learning_rate": 1.23e-05,
-      "loss": 0.0484,
+      "learning_rate": 1.2200000000000002e-05,
+      "loss": 0.0471,
       "step": 5383
     },
     {
       "epoch": 97.89,
-      "learning_rate": 1.2200000000000002e-05,
-      "loss": 0.0826,
+      "learning_rate": 1.21e-05,
+      "loss": 0.0612,
       "step": 5384
     },
     {
       "epoch": 97.91,
-      "learning_rate": 1.21e-05,
-      "loss": 0.0644,
+      "learning_rate": 1.2e-05,
+      "loss": 0.0505,
       "step": 5385
     },
     {
       "epoch": 97.93,
-      "learning_rate": 1.2e-05,
-      "loss": 0.0906,
+      "learning_rate": 1.1900000000000001e-05,
+      "loss": 0.0782,
       "step": 5386
     },
     {
       "epoch": 97.95,
-      "learning_rate": 1.1900000000000001e-05,
-      "loss": 0.0824,
+      "learning_rate": 1.18e-05,
+      "loss": 0.0694,
       "step": 5387
     },
     {
       "epoch": 97.96,
-      "learning_rate": 1.18e-05,
-      "loss": 0.0767,
+      "learning_rate": 1.17e-05,
+      "loss": 0.0663,
       "step": 5388
     },
     {
       "epoch": 97.98,
-      "learning_rate": 1.17e-05,
-      "loss": 0.0482,
+      "learning_rate": 1.1599999999999999e-05,
+      "loss": 0.0465,
       "step": 5389
     },
     {
       "epoch": 98.0,
-      "learning_rate": 1.1599999999999999e-05,
-      "loss": 0.0276,
+      "learning_rate": 1.15e-05,
+      "loss": 0.025,
       "step": 5390
     },
     {
       "epoch": 98.02,
-      "learning_rate": 1.15e-05,
-      "loss": 0.0692,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.0569,
       "step": 5391
     },
     {
       "epoch": 98.04,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 0.0802,
+      "learning_rate": 1.13e-05,
+      "loss": 0.0954,
       "step": 5392
     },
     {
       "epoch": 98.05,
-      "learning_rate": 1.13e-05,
-      "loss": 0.0699,
+      "learning_rate": 1.12e-05,
+      "loss": 0.07,
       "step": 5393
     },
     {
       "epoch": 98.07,
-      "learning_rate": 1.12e-05,
-      "loss": 0.061,
+      "learning_rate": 1.11e-05,
+      "loss": 0.0476,
       "step": 5394
     },
     {
       "epoch": 98.09,
-      "learning_rate": 1.11e-05,
-      "loss": 0.0498,
+      "learning_rate": 1.1e-05,
+      "loss": 0.0476,
       "step": 5395
     },
     {
       "epoch": 98.11,
-      "learning_rate": 1.1e-05,
-      "loss": 0.0519,
+      "learning_rate": 1.09e-05,
+      "loss": 0.0459,
       "step": 5396
     },
     {
       "epoch": 98.13,
-      "learning_rate": 1.09e-05,
-      "loss": 0.0784,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.0772,
       "step": 5397
     },
     {
       "epoch": 98.15,
-      "learning_rate": 1.0800000000000002e-05,
-      "loss": 0.0993,
+      "learning_rate": 1.07e-05,
+      "loss": 0.0843,
       "step": 5398
     },
     {
       "epoch": 98.16,
-      "learning_rate": 1.07e-05,
-      "loss": 0.0817,
+      "learning_rate": 1.06e-05,
+      "loss": 0.0843,
       "step": 5399
     },
     {
       "epoch": 98.18,
-      "learning_rate": 1.06e-05,
-      "loss": 0.0818,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.0715,
       "step": 5400
     },
     {
       "epoch": 98.2,
-      "learning_rate": 1.0500000000000001e-05,
-      "loss": 0.0872,
+      "learning_rate": 1.04e-05,
+      "loss": 0.0677,
       "step": 5401
     },
     {
       "epoch": 98.22,
-      "learning_rate": 1.04e-05,
-      "loss": 0.0502,
+      "learning_rate": 1.03e-05,
+      "loss": 0.0501,
       "step": 5402
     },
     {
       "epoch": 98.24,
-      "learning_rate": 1.03e-05,
-      "loss": 0.075,
+      "learning_rate": 1.02e-05,
+      "loss": 0.0664,
       "step": 5403
     },
     {
       "epoch": 98.25,
-      "learning_rate": 1.02e-05,
-      "loss": 0.0639,
+      "learning_rate": 1.01e-05,
+      "loss": 0.0715,
       "step": 5404
     },
     {
       "epoch": 98.27,
-      "learning_rate": 1.01e-05,
-      "loss": 0.0762,
+      "learning_rate": 1e-05,
+      "loss": 0.0503,
       "step": 5405
     },
     {
       "epoch": 98.29,
-      "learning_rate": 1e-05,
-      "loss": 0.0518,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 0.0454,
       "step": 5406
     },
     {
       "epoch": 98.31,
-      "learning_rate": 9.900000000000002e-06,
-      "loss": 0.0756,
+      "learning_rate": 9.8e-06,
+      "loss": 0.0706,
       "step": 5407
     },
     {
       "epoch": 98.33,
-      "learning_rate": 9.8e-06,
-      "loss": 0.0711,
+      "learning_rate": 9.7e-06,
+      "loss": 0.062,
       "step": 5408
     },
     {
       "epoch": 98.35,
-      "learning_rate": 9.7e-06,
-      "loss": 0.0688,
+      "learning_rate": 9.6e-06,
+      "loss": 0.0799,
       "step": 5409
     },
     {
       "epoch": 98.36,
-      "learning_rate": 9.6e-06,
-      "loss": 0.0911,
+      "learning_rate": 9.5e-06,
+      "loss": 0.0817,
       "step": 5410
     },
     {
       "epoch": 98.38,
-      "learning_rate": 9.5e-06,
-      "loss": 0.082,
+      "learning_rate": 9.4e-06,
+      "loss": 0.1021,
       "step": 5411
     },
     {
       "epoch": 98.4,
-      "learning_rate": 9.4e-06,
-      "loss": 0.0547,
+      "learning_rate": 9.299999999999999e-06,
+      "loss": 0.0699,
       "step": 5412
     },
     {
       "epoch": 98.42,
-      "learning_rate": 9.299999999999999e-06,
-      "loss": 0.0623,
+      "learning_rate": 9.2e-06,
+      "loss": 0.077,
       "step": 5413
     },
     {
       "epoch": 98.44,
-      "learning_rate": 9.2e-06,
-      "loss": 0.0654,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.0674,
       "step": 5414
     },
     {
       "epoch": 98.45,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 0.0424,
+      "learning_rate": 9e-06,
+      "loss": 0.0422,
       "step": 5415
     },
     {
       "epoch": 98.47,
-      "learning_rate": 9e-06,
-      "loss": 0.0897,
+      "learning_rate": 8.9e-06,
+      "loss": 0.0665,
       "step": 5416
     },
     {
       "epoch": 98.49,
-      "learning_rate": 8.9e-06,
-      "loss": 0.0809,
+      "learning_rate": 8.8e-06,
+      "loss": 0.068,
       "step": 5417
     },
     {
       "epoch": 98.51,
-      "learning_rate": 8.8e-06,
-      "loss": 0.0702,
+      "learning_rate": 8.7e-06,
+      "loss": 0.0468,
       "step": 5418
     },
     {
       "epoch": 98.53,
-      "learning_rate": 8.7e-06,
-      "loss": 0.0727,
+      "learning_rate": 8.6e-06,
+      "loss": 0.0532,
       "step": 5419
     },
     {
       "epoch": 98.55,
-      "learning_rate": 8.6e-06,
-      "loss": 0.0599,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.0529,
       "step": 5420
     },
     {
       "epoch": 98.56,
-      "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0494,
+      "learning_rate": 8.4e-06,
+      "loss": 0.0753,
       "step": 5421
     },
     {
       "epoch": 98.58,
-      "learning_rate": 8.4e-06,
-      "loss": 0.0828,
+      "learning_rate": 8.3e-06,
+      "loss": 0.0667,
       "step": 5422
     },
     {
       "epoch": 98.6,
-      "learning_rate": 8.3e-06,
-      "loss": 0.0667,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 0.0457,
       "step": 5423
     },
     {
       "epoch": 98.62,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 0.059,
+      "learning_rate": 8.1e-06,
+      "loss": 0.0388,
       "step": 5424
     },
     {
       "epoch": 98.64,
-      "learning_rate": 8.1e-06,
-      "loss": 0.0578,
+      "learning_rate": 8e-06,
+      "loss": 0.0411,
       "step": 5425
     },
     {
       "epoch": 98.65,
-      "learning_rate": 8e-06,
-      "loss": 0.0837,
+      "learning_rate": 7.9e-06,
+      "loss": 0.063,
       "step": 5426
     },
     {
       "epoch": 98.67,
-      "learning_rate": 7.9e-06,
-      "loss": 0.0909,
+      "learning_rate": 7.8e-06,
+      "loss": 0.0642,
       "step": 5427
     },
     {
       "epoch": 98.69,
-      "learning_rate": 7.8e-06,
-      "loss": 0.0578,
+      "learning_rate": 7.7e-06,
+      "loss": 0.0725,
       "step": 5428
     },
     {
       "epoch": 98.71,
-      "learning_rate": 7.7e-06,
-      "loss": 0.067,
+      "learning_rate": 7.6e-06,
+      "loss": 0.0512,
       "step": 5429
     },
     {
       "epoch": 98.73,
-      "learning_rate": 7.6e-06,
-      "loss": 0.0692,
+      "learning_rate": 7.5e-06,
+      "loss": 0.0715,
       "step": 5430
     },
     {
       "epoch": 98.75,
-      "learning_rate": 7.5e-06,
-      "loss": 0.0694,
+      "learning_rate": 7.4e-06,
+      "loss": 0.0667,
       "step": 5431
     },
     {
       "epoch": 98.76,
-      "learning_rate": 7.4e-06,
-      "loss": 0.053,
+      "learning_rate": 7.3e-06,
+      "loss": 0.0482,
       "step": 5432
     },
     {
       "epoch": 98.78,
-      "learning_rate": 7.3e-06,
-      "loss": 0.0937,
+      "learning_rate": 7.2e-06,
+      "loss": 0.078,
       "step": 5433
     },
     {
       "epoch": 98.8,
-      "learning_rate": 7.2e-06,
-      "loss": 0.0697,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.0562,
       "step": 5434
     },
     {
       "epoch": 98.82,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 0.0735,
+      "learning_rate": 7e-06,
+      "loss": 0.0675,
       "step": 5435
     },
     {
       "epoch": 98.84,
-      "learning_rate": 7e-06,
-      "loss": 0.0636,
+      "learning_rate": 6.9e-06,
+      "loss": 0.0656,
       "step": 5436
     },
     {
       "epoch": 98.85,
-      "learning_rate": 6.9e-06,
-      "loss": 0.0445,
+      "learning_rate": 6.8e-06,
+      "loss": 0.0426,
       "step": 5437
     },
     {
       "epoch": 98.87,
-      "learning_rate": 6.8e-06,
-      "loss": 0.0447,
+      "learning_rate": 6.7e-06,
+      "loss": 0.0357,
       "step": 5438
     },
     {
       "epoch": 98.89,
-      "learning_rate": 6.7e-06,
-      "loss": 0.0531,
+      "learning_rate": 6.6e-06,
+      "loss": 0.0625,
       "step": 5439
     },
     {
       "epoch": 98.91,
-      "learning_rate": 6.6e-06,
-      "loss": 0.0369,
+      "learning_rate": 6.5e-06,
+      "loss": 0.0392,
       "step": 5440
     },
     {
       "epoch": 98.93,
-      "learning_rate": 6.5e-06,
-      "loss": 0.09,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.0823,
       "step": 5441
     },
     {
       "epoch": 98.95,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.0815,
+      "learning_rate": 6.3e-06,
+      "loss": 0.06,
       "step": 5442
     },
     {
       "epoch": 98.96,
-      "learning_rate": 6.3e-06,
-      "loss": 0.0557,
+      "learning_rate": 6.2e-06,
+      "loss": 0.0556,
       "step": 5443
     },
     {
       "epoch": 98.98,
-      "learning_rate": 6.2e-06,
-      "loss": 0.0551,
+      "learning_rate": 6.100000000000001e-06,
+      "loss": 0.0627,
       "step": 5444
     },
     {
       "epoch": 99.0,
-      "learning_rate": 6.100000000000001e-06,
-      "loss": 0.0539,
+      "learning_rate": 6e-06,
+      "loss": 0.0273,
       "step": 5445
     },
     {
       "epoch": 99.02,
-      "learning_rate": 6e-06,
-      "loss": 0.0724,
+      "learning_rate": 5.9e-06,
+      "loss": 0.0606,
       "step": 5446
     },
     {
       "epoch": 99.04,
-      "learning_rate": 5.9e-06,
-      "loss": 0.0855,
+      "learning_rate": 5.7999999999999995e-06,
+      "loss": 0.0734,
       "step": 5447
     },
     {
       "epoch": 99.05,
-      "learning_rate": 5.7999999999999995e-06,
-      "loss": 0.1046,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 0.0776,
       "step": 5448
     },
     {
       "epoch": 99.07,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 0.1062,
+      "learning_rate": 5.6e-06,
+      "loss": 0.0933,
       "step": 5449
     },
     {
       "epoch": 99.09,
-      "learning_rate": 5.6e-06,
-      "loss": 0.0645,
+      "learning_rate": 5.5e-06,
+      "loss": 0.0673,
       "step": 5450
     },
     {
       "epoch": 99.11,
-      "learning_rate": 5.5e-06,
-      "loss": 0.0899,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.0778,
       "step": 5451
     },
     {
       "epoch": 99.13,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 0.0762,
+      "learning_rate": 5.3e-06,
+      "loss": 0.0573,
       "step": 5452
     },
     {
       "epoch": 99.15,
-      "learning_rate": 5.3e-06,
-      "loss": 0.0542,
+      "learning_rate": 5.2e-06,
+      "loss": 0.0431,
       "step": 5453
     },
     {
       "epoch": 99.16,
-      "learning_rate": 5.2e-06,
-      "loss": 0.0452,
+      "learning_rate": 5.1e-06,
+      "loss": 0.0339,
       "step": 5454
     },
     {
       "epoch": 99.18,
-      "learning_rate": 5.1e-06,
-      "loss": 0.0705,
+      "learning_rate": 5e-06,
+      "loss": 0.0655,
       "step": 5455
     },
     {
       "epoch": 99.2,
-      "learning_rate": 5e-06,
-      "loss": 0.0925,
+      "learning_rate": 4.9e-06,
+      "loss": 0.0655,
       "step": 5456
     },
     {
       "epoch": 99.22,
-      "learning_rate": 4.9e-06,
-      "loss": 0.1048,
+      "learning_rate": 4.8e-06,
+      "loss": 0.1059,
       "step": 5457
     },
     {
       "epoch": 99.24,
-      "learning_rate": 4.8e-06,
-      "loss": 0.072,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0535,
       "step": 5458
     },
     {
       "epoch": 99.25,
-      "learning_rate": 4.7e-06,
-      "loss": 0.0716,
+      "learning_rate": 4.6e-06,
+      "loss": 0.0556,
       "step": 5459
     },
     {
       "epoch": 99.27,
-      "learning_rate": 4.6e-06,
-      "loss": 0.089,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0656,
       "step": 5460
     },
     {
       "epoch": 99.29,
-      "learning_rate": 4.5e-06,
-      "loss": 0.071,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0587,
       "step": 5461
     },
     {
       "epoch": 99.31,
-      "learning_rate": 4.4e-06,
-      "loss": 0.0611,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0458,
       "step": 5462
     },
     {
       "epoch": 99.33,
-      "learning_rate": 4.3e-06,
-      "loss": 0.0727,
+      "learning_rate": 4.2e-06,
+      "loss": 0.0633,
       "step": 5463
     },
     {
       "epoch": 99.35,
-      "learning_rate": 4.2e-06,
-      "loss": 0.0402,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 0.0462,
       "step": 5464
     },
     {
       "epoch": 99.36,
-      "learning_rate": 4.1000000000000006e-06,
-      "loss": 0.0769,
+      "learning_rate": 4e-06,
+      "loss": 0.0714,
       "step": 5465
     },
     {
       "epoch": 99.38,
-      "learning_rate": 4e-06,
-      "loss": 0.0564,
+      "learning_rate": 3.9e-06,
+      "loss": 0.0556,
       "step": 5466
     },
     {
       "epoch": 99.4,
-      "learning_rate": 3.9e-06,
-      "loss": 0.1047,
+      "learning_rate": 3.8e-06,
+      "loss": 0.0671,
       "step": 5467
     },
     {
       "epoch": 99.42,
-      "learning_rate": 3.8e-06,
-      "loss": 0.0822,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0743,
       "step": 5468
     },
     {
       "epoch": 99.44,
-      "learning_rate": 3.7e-06,
-      "loss": 0.066,
+      "learning_rate": 3.6e-06,
+      "loss": 0.0698,
       "step": 5469
     },
     {
       "epoch": 99.45,
-      "learning_rate": 3.6e-06,
-      "loss": 0.0586,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0815,
       "step": 5470
     },
     {
       "epoch": 99.47,
-      "learning_rate": 3.5e-06,
-      "loss": 0.0706,
+      "learning_rate": 3.4e-06,
+      "loss": 0.0697,
       "step": 5471
     },
     {
       "epoch": 99.49,
-      "learning_rate": 3.4e-06,
-      "loss": 0.0884,
+      "learning_rate": 3.3e-06,
+      "loss": 0.0736,
       "step": 5472
     },
     {
       "epoch": 99.51,
-      "learning_rate": 3.3e-06,
-      "loss": 0.1013,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.1143,
       "step": 5473
     },
     {
       "epoch": 99.53,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 0.0531,
+      "learning_rate": 3.1e-06,
+      "loss": 0.0648,
       "step": 5474
     },
     {
       "epoch": 99.55,
-      "learning_rate": 3.1e-06,
-      "loss": 0.0745,
+      "learning_rate": 3e-06,
+      "loss": 0.0667,
       "step": 5475
     },
     {
       "epoch": 99.56,
-      "learning_rate": 3e-06,
-      "loss": 0.0564,
+      "learning_rate": 2.8999999999999998e-06,
+      "loss": 0.0493,
       "step": 5476
     },
     {
       "epoch": 99.58,
-      "learning_rate": 2.8999999999999998e-06,
-      "loss": 0.0765,
+      "learning_rate": 2.8e-06,
+      "loss": 0.0568,
       "step": 5477
     },
     {
       "epoch": 99.6,
-      "learning_rate": 2.8e-06,
-      "loss": 0.0701,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.0658,
       "step": 5478
     },
     {
       "epoch": 99.62,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 0.1006,
+      "learning_rate": 2.6e-06,
+      "loss": 0.1024,
       "step": 5479
     },
     {
       "epoch": 99.64,
-      "learning_rate": 2.6e-06,
-      "loss": 0.0825,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0768,
       "step": 5480
     },
     {
       "epoch": 99.65,
-      "learning_rate": 2.5e-06,
-      "loss": 0.037,
+      "learning_rate": 2.4e-06,
+      "loss": 0.0264,
       "step": 5481
     },
     {
       "epoch": 99.67,
-      "learning_rate": 2.4e-06,
-      "loss": 0.0951,
+      "learning_rate": 2.3e-06,
+      "loss": 0.074,
       "step": 5482
     },
     {
       "epoch": 99.69,
-      "learning_rate": 2.3e-06,
-      "loss": 0.0715,
+      "learning_rate": 2.2e-06,
+      "loss": 0.0721,
       "step": 5483
     },
     {
       "epoch": 99.71,
-      "learning_rate": 2.2e-06,
-      "loss": 0.0864,
+      "learning_rate": 2.1e-06,
+      "loss": 0.0701,
       "step": 5484
     },
     {
       "epoch": 99.73,
-      "learning_rate": 2.1e-06,
-      "loss": 0.0624,
+      "learning_rate": 2e-06,
+      "loss": 0.0537,
       "step": 5485
     },
     {
       "epoch": 99.75,
-      "learning_rate": 2e-06,
-      "loss": 0.0894,
+      "learning_rate": 1.9e-06,
+      "loss": 0.0719,
       "step": 5486
     },
     {
       "epoch": 99.76,
-      "learning_rate": 1.9e-06,
-      "loss": 0.0682,
+      "learning_rate": 1.8e-06,
+      "loss": 0.0589,
       "step": 5487
     },
     {
       "epoch": 99.78,
-      "learning_rate": 1.8e-06,
-      "loss": 0.0508,
+      "learning_rate": 1.7e-06,
+      "loss": 0.0596,
       "step": 5488
     },
     {
       "epoch": 99.8,
-      "learning_rate": 1.7e-06,
-      "loss": 0.0374,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.0392,
       "step": 5489
     },
     {
       "epoch": 99.82,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 0.0574,
+      "learning_rate": 1.5e-06,
+      "loss": 0.0506,
       "step": 5490
     },
     {
       "epoch": 99.84,
-      "learning_rate": 1.5e-06,
-      "loss": 0.075,
+      "learning_rate": 1.4e-06,
+      "loss": 0.0554,
       "step": 5491
     },
     {
       "epoch": 99.85,
-      "learning_rate": 1.4e-06,
-      "loss": 0.0585,
+      "learning_rate": 1.3e-06,
+      "loss": 0.0434,
       "step": 5492
     },
     {
       "epoch": 99.87,
-      "learning_rate": 1.3e-06,
-      "loss": 0.0677,
+      "learning_rate": 1.2e-06,
+      "loss": 0.059,
       "step": 5493
     },
     {
       "epoch": 99.89,
-      "learning_rate": 1.2e-06,
-      "loss": 0.0802,
+      "learning_rate": 1.1e-06,
+      "loss": 0.0664,
       "step": 5494
     },
     {
       "epoch": 99.91,
-      "learning_rate": 1.1e-06,
-      "loss": 0.059,
+      "learning_rate": 1e-06,
+      "loss": 0.0577,
       "step": 5495
     },
     {
       "epoch": 99.93,
-      "learning_rate": 1e-06,
-      "loss": 0.0846,
+      "learning_rate": 9e-07,
+      "loss": 0.0786,
       "step": 5496
     },
     {
       "epoch": 99.95,
-      "learning_rate": 9e-07,
-      "loss": 0.0616,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.0462,
       "step": 5497
     },
     {
       "epoch": 99.96,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 0.0913,
+      "learning_rate": 7e-07,
+      "loss": 0.0633,
       "step": 5498
     },
     {
       "epoch": 99.98,
-      "learning_rate": 7e-07,
-      "loss": 0.0574,
+      "learning_rate": 6e-07,
+      "loss": 0.0592,
       "step": 5499
     },
     {
       "epoch": 100.0,
-      "learning_rate": 6e-07,
-      "loss": 0.0741,
+      "learning_rate": 5e-07,
+      "loss": 0.0593,
       "step": 5500
     },
     {
       "epoch": 100.0,
-      "eval_cer": 0.07610309000134935,
-      "eval_loss": 0.4067234694957733,
-      "eval_runtime": 14.0515,
-      "eval_samples_per_second": 117.211,
-      "eval_steps_per_second": 1.85,
-      "eval_wer": 0.30977428250434075,
+      "eval_cer": 0.07364428252297636,
+      "eval_loss": 0.41792556643486023,
+      "eval_runtime": 14.2781,
+      "eval_samples_per_second": 115.352,
+      "eval_steps_per_second": 1.821,
+      "eval_wer": 0.30711878255540803,
       "step": 5500
     },
     {
       "epoch": 100.0,
       "step": 5500,
       "total_flos": 4.838135000470022e+19,
-      "train_loss": 0.4563706781716848,
-      "train_runtime": 4209.214,
-      "train_samples_per_second": 82.628,
-      "train_steps_per_second": 1.307
+      "train_loss": 0.45052173870835793,
+      "train_runtime": 4247.5395,
+      "train_samples_per_second": 81.883,
+      "train_steps_per_second": 1.295
     }
   ],
   "max_steps": 5500,