diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9995796553173603,
-  "global_step": 1189,
+  "epoch": 0.9987389659520807,
+  "global_step": 594,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,7166 +10,3587 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.1784,
+      "loss": 9.2391,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.5725,
+      "loss": 9.104,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 2e-06,
-      "loss": 9.3285,
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.9021,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 2e-06,
-      "loss": 9.1371,
+      "loss": 8.8682,
       "step": 4
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 4e-06,
-      "loss": 8.9227,
+      "loss": 8.8262,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 6e-06,
-      "loss": 8.7957,
+      "loss": 8.4047,
       "step": 6
     },
     {
       "epoch": 0.01,
       "learning_rate": 8e-06,
-      "loss": 8.6135,
+      "loss": 8.3502,
       "step": 7
     },
     {
       "epoch": 0.01,
       "learning_rate": 1e-05,
-      "loss": 8.479,
+      "loss": 7.9771,
       "step": 8
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.2e-05,
-      "loss": 8.2944,
+      "loss": 7.8414,
       "step": 9
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.4e-05,
-      "loss": 8.06,
+      "loss": 7.6606,
       "step": 10
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.6e-05,
-      "loss": 7.929,
+      "loss": 7.5532,
       "step": 11
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.8e-05,
-      "loss": 7.7736,
+      "loss": 7.5309,
       "step": 12
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 2e-05,
-      "loss": 7.5648,
+      "loss": 7.3389,
       "step": 13
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 2.2e-05,
-      "loss": 7.6847,
+      "loss": 7.5473,
       "step": 14
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.03,
       "learning_rate": 2.4e-05,
-      "loss": 7.9066,
+      "loss": 7.3029,
       "step": 15
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.03,
       "learning_rate": 2.6e-05,
-      "loss": 7.4937,
+      "loss": 7.3217,
       "step": 16
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.03,
       "learning_rate": 2.8e-05,
-      "loss": 7.4857,
+      "loss": 7.3244,
       "step": 17
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.03,
       "learning_rate": 3e-05,
-      "loss": 7.3758,
+      "loss": 7.2319,
       "step": 18
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.03,
       "learning_rate": 3.2e-05,
-      "loss": 7.2432,
+      "loss": 7.1778,
       "step": 19
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.03,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 7.4421,
+      "loss": 7.257,
       "step": 20
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 3.6e-05,
-      "loss": 7.2182,
+      "loss": 7.3211,
       "step": 21
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 3.8e-05,
-      "loss": 7.3485,
+      "loss": 7.2754,
       "step": 22
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 4e-05,
-      "loss": 7.3024,
+      "loss": 7.268,
       "step": 23
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 4.2000000000000004e-05,
-      "loss": 7.2753,
+      "loss": 7.2223,
       "step": 24
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 4.4e-05,
-      "loss": 7.1561,
+      "loss": 7.0232,
       "step": 25
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 4.6e-05,
-      "loss": 7.3336,
+      "loss": 7.377,
       "step": 26
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.05,
       "learning_rate": 4.8e-05,
-      "loss": 7.1281,
+      "loss": 7.1471,
       "step": 27
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.05,
       "learning_rate": 5e-05,
-      "loss": 7.0395,
+      "loss": 7.0311,
       "step": 28
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.05,
       "learning_rate": 5.2e-05,
-      "loss": 7.2563,
+      "loss": 7.1356,
       "step": 29
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.05,
       "learning_rate": 5.4e-05,
-      "loss": 7.3104,
+      "loss": 7.1118,
       "step": 30
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.05,
       "learning_rate": 5.6e-05,
-      "loss": 7.1971,
+      "loss": 7.2032,
       "step": 31
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.05,
       "learning_rate": 5.800000000000001e-05,
-      "loss": 7.1307,
+      "loss": 7.1625,
       "step": 32
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 6e-05,
-      "loss": 7.3249,
+      "loss": 7.0952,
       "step": 33
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 6.2e-05,
-      "loss": 6.9528,
+      "loss": 7.0539,
       "step": 34
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 6.4e-05,
-      "loss": 7.0472,
+      "loss": 7.0215,
       "step": 35
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 6.6e-05,
-      "loss": 7.2907,
+      "loss": 7.1635,
       "step": 36
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 7.1711,
+      "loss": 7.118,
       "step": 37
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 7.000000000000001e-05,
-      "loss": 7.1423,
+      "loss": 7.1143,
       "step": 38
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 7.2e-05,
-      "loss": 7.2415,
+      "loss": 7.1712,
       "step": 39
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 7.4e-05,
-      "loss": 7.0673,
+      "loss": 7.0529,
       "step": 40
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 7.6e-05,
-      "loss": 7.0163,
+      "loss": 7.0855,
       "step": 41
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 7.8e-05,
-      "loss": 7.1791,
+      "loss": 7.1638,
       "step": 42
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 8e-05,
-      "loss": 7.0275,
+      "loss": 6.9741,
       "step": 43
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 8.2e-05,
-      "loss": 7.0427,
+      "loss": 7.0521,
       "step": 44
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 8.400000000000001e-05,
-      "loss": 7.1336,
+      "loss": 7.1212,
       "step": 45
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 8.599999999999999e-05,
-      "loss": 7.0856,
+      "loss": 7.0775,
       "step": 46
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 8.8e-05,
-      "loss": 7.0557,
+      "loss": 6.9064,
       "step": 47
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 7.0549,
+      "loss": 7.0588,
       "step": 48
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 9.2e-05,
-      "loss": 6.824,
+      "loss": 6.7153,
       "step": 49
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 9.400000000000001e-05,
-      "loss": 6.6725,
+      "loss": 6.5244,
       "step": 50
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 9.6e-05,
-      "loss": 7.4345,
+      "loss": 7.3138,
       "step": 51
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 9.800000000000001e-05,
-      "loss": 7.1991,
+      "loss": 7.2796,
       "step": 52
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 0.0001,
-      "loss": 7.0812,
+      "loss": 7.0741,
       "step": 53
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 0.000102,
-      "loss": 7.1414,
+      "loss": 7.1467,
       "step": 54
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 0.000104,
-      "loss": 7.2373,
+      "loss": 7.0,
       "step": 55
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 0.000106,
-      "loss": 6.9596,
+      "loss": 7.1375,
       "step": 56
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.000108,
-      "loss": 7.3832,
+      "loss": 6.9579,
       "step": 57
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.00011,
-      "loss": 6.9846,
+      "loss": 7.1664,
       "step": 58
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.000112,
-      "loss": 6.976,
+      "loss": 7.1162,
       "step": 59
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.000114,
-      "loss": 7.1303,
+      "loss": 6.9863,
       "step": 60
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.00011600000000000001,
-      "loss": 6.9403,
+      "loss": 7.1279,
       "step": 61
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 0.000118,
-      "loss": 7.0249,
+      "loss": 7.0931,
       "step": 62
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 0.00012,
-      "loss": 7.1265,
+      "loss": 7.0619,
       "step": 63
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 0.000122,
-      "loss": 7.0689,
+      "loss": 7.0442,
       "step": 64
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 0.000124,
-      "loss": 7.0175,
+      "loss": 7.0437,
       "step": 65
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 0.000126,
-      "loss": 7.0994,
+      "loss": 7.0038,
       "step": 66
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 0.000128,
-      "loss": 7.0589,
+      "loss": 7.05,
       "step": 67
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 0.00013000000000000002,
-      "loss": 6.9998,
+      "loss": 6.863,
       "step": 68
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.000132,
-      "loss": 6.9596,
+      "loss": 7.0607,
       "step": 69
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.000134,
-      "loss": 7.0045,
+      "loss": 7.0143,
       "step": 70
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.00013600000000000003,
-      "loss": 7.1545,
+      "loss": 7.041,
       "step": 71
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.00013800000000000002,
-      "loss": 7.2349,
+      "loss": 7.151,
       "step": 72
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.00014000000000000001,
-      "loss": 7.0776,
+      "loss": 6.8324,
       "step": 73
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 0.00014199999999999998,
-      "loss": 7.1223,
+      "loss": 6.9986,
       "step": 74
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 0.000144,
-      "loss": 6.8991,
+      "loss": 7.0604,
       "step": 75
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 0.000146,
-      "loss": 7.2667,
+      "loss": 7.0742,
       "step": 76
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.13,
       "learning_rate": 0.000148,
-      "loss": 7.0611,
+      "loss": 7.0835,
       "step": 77
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 0.00015,
-      "loss": 6.8979,
+      "loss": 7.0913,
       "step": 78
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 0.000152,
-      "loss": 6.927,
+      "loss": 6.9772,
       "step": 79
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 0.000154,
-      "loss": 7.0384,
+      "loss": 7.025,
       "step": 80
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.000156,
-      "loss": 7.1154,
+      "loss": 7.0169,
       "step": 81
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.000158,
-      "loss": 7.2001,
+      "loss": 7.0588,
       "step": 82
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.00016,
-      "loss": 6.965,
+      "loss": 7.0161,
       "step": 83
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.000162,
-      "loss": 7.1127,
+      "loss": 7.0474,
       "step": 84
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.000164,
-      "loss": 7.0066,
+      "loss": 6.9451,
       "step": 85
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 0.00016600000000000002,
-      "loss": 7.076,
+      "loss": 6.8814,
       "step": 86
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 0.00016800000000000002,
-      "loss": 6.9905,
+      "loss": 7.0686,
       "step": 87
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 0.00017,
-      "loss": 7.0706,
+      "loss": 7.0055,
       "step": 88
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 0.00017199999999999998,
-      "loss": 7.2036,
+      "loss": 6.9607,
       "step": 89
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 0.000174,
-      "loss": 7.0382,
+      "loss": 7.0966,
       "step": 90
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 0.000176,
-      "loss": 7.2465,
+      "loss": 6.9914,
       "step": 91
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.15,
       "learning_rate": 0.000178,
-      "loss": 7.1257,
+      "loss": 7.0491,
       "step": 92
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.00017999999999999998,
-      "loss": 6.8238,
+      "loss": 6.9837,
       "step": 93
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.000182,
-      "loss": 7.1199,
+      "loss": 6.9912,
       "step": 94
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.000184,
-      "loss": 7.2177,
+      "loss": 6.8828,
       "step": 95
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.000186,
-      "loss": 6.8903,
+      "loss": 6.8851,
       "step": 96
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.00018800000000000002,
-      "loss": 7.0236,
+      "loss": 6.9958,
       "step": 97
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 0.00019,
-      "loss": 6.9423,
+      "loss": 6.6052,
       "step": 98
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 0.000192,
-      "loss": 6.6384,
+      "loss": 6.8389,
       "step": 99
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 0.000194,
-      "loss": 6.4692,
+      "loss": 6.4098,
       "step": 100
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 0.00019600000000000002,
-      "loss": 7.0916,
+      "loss": 7.1685,
       "step": 101
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 0.00019800000000000002,
-      "loss": 7.0911,
+      "loss": 7.0531,
       "step": 102
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 0.0002,
-      "loss": 7.0966,
+      "loss": 7.0706,
       "step": 103
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 0.000202,
-      "loss": 7.0376,
+      "loss": 6.9564,
       "step": 104
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.000204,
-      "loss": 6.9953,
+      "loss": 6.9889,
       "step": 105
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.000206,
-      "loss": 7.1316,
+      "loss": 7.0398,
       "step": 106
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.000208,
-      "loss": 7.0401,
+      "loss": 7.1539,
       "step": 107
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.00021,
-      "loss": 7.1044,
+      "loss": 7.0754,
       "step": 108
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.000212,
-      "loss": 7.2254,
+      "loss": 7.0787,
       "step": 109
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 0.000214,
-      "loss": 7.1079,
+      "loss": 7.044,
       "step": 110
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 0.000216,
-      "loss": 7.1528,
+      "loss": 6.9558,
       "step": 111
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 0.000218,
-      "loss": 7.1297,
+      "loss": 6.9957,
       "step": 112
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 0.00022,
-      "loss": 7.0263,
+      "loss": 6.9355,
       "step": 113
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 0.000222,
-      "loss": 22.7629,
+      "loss": 7.0677,
       "step": 114
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.19,
       "learning_rate": 0.000224,
-      "loss": 14.4324,
+      "loss": 7.0284,
       "step": 115
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000224,
-      "loss": 15.1333,
+      "epoch": 0.2,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 6.9798,
       "step": 116
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000224,
-      "loss": 15.5222,
+      "epoch": 0.2,
+      "learning_rate": 0.000228,
+      "loss": 7.0059,
       "step": 117
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000224,
-      "loss": 15.1973,
+      "epoch": 0.2,
+      "learning_rate": 0.00023,
+      "loss": 6.9782,
       "step": 118
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000224,
-      "loss": 15.3933,
+      "epoch": 0.2,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 7.0379,
       "step": 119
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00022600000000000002,
-      "loss": 15.2458,
+      "epoch": 0.2,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 7.0151,
       "step": 120
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000228,
-      "loss": 7.3314,
+      "epoch": 0.2,
+      "learning_rate": 0.000236,
+      "loss": 7.1422,
       "step": 121
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00023,
-      "loss": 7.0395,
+      "epoch": 0.21,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 7.1631,
       "step": 122
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00023200000000000003,
-      "loss": 7.1346,
+      "epoch": 0.21,
+      "learning_rate": 0.00024,
+      "loss": 7.0311,
       "step": 123
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00023400000000000002,
-      "loss": 7.0888,
+      "epoch": 0.21,
+      "learning_rate": 0.000242,
+      "loss": 7.0971,
       "step": 124
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000236,
-      "loss": 7.0469,
+      "epoch": 0.21,
+      "learning_rate": 0.000244,
+      "loss": 6.9513,
       "step": 125
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00023799999999999998,
-      "loss": 6.9381,
+      "epoch": 0.21,
+      "learning_rate": 0.000246,
+      "loss": 7.0559,
       "step": 126
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00024,
-      "loss": 7.252,
+      "epoch": 0.21,
+      "learning_rate": 0.000248,
+      "loss": 6.963,
       "step": 127
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000242,
-      "loss": 6.9697,
+      "epoch": 0.22,
+      "learning_rate": 0.00025,
+      "loss": 7.0279,
       "step": 128
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000244,
-      "loss": 7.0154,
+      "epoch": 0.22,
+      "learning_rate": 0.000252,
+      "loss": 7.0081,
       "step": 129
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000246,
-      "loss": 6.9832,
+      "epoch": 0.22,
+      "learning_rate": 0.000254,
+      "loss": 7.1089,
       "step": 130
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000248,
-      "loss": 6.8832,
+      "epoch": 0.22,
+      "learning_rate": 0.000256,
+      "loss": 7.0063,
       "step": 131
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00025,
-      "loss": 7.0214,
+      "epoch": 0.22,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 6.9755,
       "step": 132
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000252,
-      "loss": 7.2379,
+      "epoch": 0.22,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 7.0143,
       "step": 133
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000254,
-      "loss": 6.9883,
+      "epoch": 0.23,
+      "learning_rate": 0.000262,
+      "loss": 7.1515,
       "step": 134
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000256,
-      "loss": 7.0613,
+      "epoch": 0.23,
+      "learning_rate": 0.000264,
+      "loss": 7.0614,
       "step": 135
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00025800000000000004,
-      "loss": 6.9666,
+      "epoch": 0.23,
+      "learning_rate": 0.000266,
+      "loss": 7.1099,
       "step": 136
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00026000000000000003,
-      "loss": 6.9942,
+      "epoch": 0.23,
+      "learning_rate": 0.000268,
+      "loss": 6.9085,
       "step": 137
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000262,
-      "loss": 7.0373,
+      "epoch": 0.23,
+      "learning_rate": 0.00027,
+      "loss": 7.0223,
       "step": 138
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000264,
-      "loss": 6.9944,
+      "epoch": 0.23,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 6.9545,
       "step": 139
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000266,
-      "loss": 7.1287,
+      "epoch": 0.24,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 6.993,
       "step": 140
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000268,
-      "loss": 7.0394,
+      "epoch": 0.24,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 7.0001,
       "step": 141
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00027,
-      "loss": 6.9535,
+      "epoch": 0.24,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 7.1251,
       "step": 142
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00027200000000000005,
-      "loss": 6.9182,
+      "epoch": 0.24,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 7.1047,
       "step": 143
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00027400000000000005,
-      "loss": 7.0463,
+      "epoch": 0.24,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.7781,
       "step": 144
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00027600000000000004,
-      "loss": 6.8786,
+      "epoch": 0.24,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 6.8235,
       "step": 145
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00027800000000000004,
-      "loss": 7.0804,
+      "epoch": 0.25,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 7.1905,
       "step": 146
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00028000000000000003,
-      "loss": 6.7952,
+      "epoch": 0.25,
+      "learning_rate": 0.000288,
+      "loss": 6.8899,
       "step": 147
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 6.5721,
+      "epoch": 0.25,
+      "learning_rate": 0.00029,
+      "loss": 6.7851,
       "step": 148
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00028399999999999996,
-      "loss": 6.7277,
+      "epoch": 0.25,
+      "learning_rate": 0.000292,
+      "loss": 6.8372,
       "step": 149
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00028599999999999996,
-      "loss": 6.6187,
+      "epoch": 0.25,
+      "learning_rate": 0.000294,
+      "loss": 6.4942,
       "step": 150
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000288,
-      "loss": 7.3634,
+      "epoch": 0.25,
+      "learning_rate": 0.000296,
+      "loss": 7.1939,
       "step": 151
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00029,
-      "loss": 7.408,
+      "epoch": 0.26,
+      "learning_rate": 0.000298,
+      "loss": 7.1799,
       "step": 152
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000292,
-      "loss": 7.1431,
+      "epoch": 0.26,
+      "learning_rate": 0.0003,
+      "loss": 7.1226,
       "step": 153
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000294,
-      "loss": 7.1948,
+      "epoch": 0.26,
+      "learning_rate": 0.000302,
+      "loss": 6.963,
       "step": 154
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000296,
-      "loss": 7.0712,
+      "epoch": 0.26,
+      "learning_rate": 0.000304,
+      "loss": 7.0494,
       "step": 155
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000298,
-      "loss": 7.0385,
+      "epoch": 0.26,
+      "learning_rate": 0.000306,
+      "loss": 7.0654,
       "step": 156
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.0003,
-      "loss": 7.083,
+      "epoch": 0.26,
+      "learning_rate": 0.000308,
+      "loss": 7.0491,
       "step": 157
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000302,
-      "loss": 7.2521,
+      "epoch": 0.27,
+      "learning_rate": 0.00031,
+      "loss": 7.0751,
       "step": 158
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000304,
-      "loss": 6.987,
+      "epoch": 0.27,
+      "learning_rate": 0.000312,
+      "loss": 6.9384,
       "step": 159
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000306,
-      "loss": 6.951,
+      "epoch": 0.27,
+      "learning_rate": 0.000314,
+      "loss": 7.1189,
       "step": 160
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000308,
-      "loss": 7.0044,
+      "epoch": 0.27,
+      "learning_rate": 0.000316,
+      "loss": 6.9326,
       "step": 161
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00031,
-      "loss": 7.151,
+      "epoch": 0.27,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 6.9468,
       "step": 162
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000312,
-      "loss": 7.2072,
+      "epoch": 0.27,
+      "learning_rate": 0.00032,
+      "loss": 7.0409,
       "step": 163
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000314,
-      "loss": 7.05,
+      "epoch": 0.28,
+      "learning_rate": 0.000322,
+      "loss": 7.0507,
       "step": 164
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000316,
-      "loss": 7.0243,
+      "epoch": 0.28,
+      "learning_rate": 0.000324,
+      "loss": 6.9408,
       "step": 165
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00031800000000000003,
-      "loss": 6.8716,
+      "epoch": 0.28,
+      "learning_rate": 0.000326,
+      "loss": 7.1385,
       "step": 166
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00032,
-      "loss": 7.0433,
+      "epoch": 0.28,
+      "learning_rate": 0.000328,
+      "loss": 7.0965,
       "step": 167
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000322,
-      "loss": 7.0543,
+      "epoch": 0.28,
+      "learning_rate": 0.00033,
+      "loss": 7.0572,
       "step": 168
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000324,
-      "loss": 7.0754,
+      "epoch": 0.28,
+      "learning_rate": 0.00033200000000000005,
+      "loss": 6.9633,
       "step": 169
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000326,
-      "loss": 7.0035,
+      "epoch": 0.29,
+      "learning_rate": 0.00033400000000000004,
+      "loss": 7.0541,
       "step": 170
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000328,
-      "loss": 7.0058,
-      "step": 171
+      "epoch": 0.29,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 6.9672,
+      "step": 171
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00033,
-      "loss": 7.03,
+      "epoch": 0.29,
+      "learning_rate": 0.00033800000000000003,
+      "loss": 6.8879,
       "step": 172
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00033200000000000005,
-      "loss": 6.8019,
+      "epoch": 0.29,
+      "learning_rate": 0.00034,
+      "loss": 6.9559,
       "step": 173
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00033400000000000004,
-      "loss": 7.0817,
+      "epoch": 0.29,
+      "learning_rate": 0.000342,
+      "loss": 6.8725,
       "step": 174
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00033600000000000004,
-      "loss": 7.0984,
+      "epoch": 0.29,
+      "learning_rate": 0.00034399999999999996,
+      "loss": 6.9745,
       "step": 175
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00033800000000000003,
-      "loss": 7.2515,
+      "epoch": 0.3,
+      "learning_rate": 0.000346,
+      "loss": 7.0163,
       "step": 176
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00034,
-      "loss": 7.1763,
+      "epoch": 0.3,
+      "learning_rate": 0.000348,
+      "loss": 7.0641,
       "step": 177
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000342,
-      "loss": 7.1449,
+      "epoch": 0.3,
+      "learning_rate": 0.00035,
+      "loss": 6.9334,
       "step": 178
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00034399999999999996,
-      "loss": 7.0247,
+      "epoch": 0.3,
+      "learning_rate": 0.000352,
+      "loss": 7.0079,
       "step": 179
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000346,
-      "loss": 7.0918,
+      "epoch": 0.3,
+      "learning_rate": 0.000354,
+      "loss": 7.0947,
       "step": 180
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000348,
-      "loss": 7.1613,
+      "epoch": 0.3,
+      "learning_rate": 0.000356,
+      "loss": 7.0143,
       "step": 181
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00035,
-      "loss": 7.2377,
+      "epoch": 0.31,
+      "learning_rate": 0.000358,
+      "loss": 7.087,
       "step": 182
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000352,
-      "loss": 6.9927,
+      "epoch": 0.31,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 7.0131,
       "step": 183
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000354,
-      "loss": 7.0281,
+      "epoch": 0.31,
+      "learning_rate": 0.000362,
+      "loss": 6.8973,
       "step": 184
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000356,
-      "loss": 6.8501,
+      "epoch": 0.31,
+      "learning_rate": 0.000364,
+      "loss": 7.1537,
       "step": 185
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000358,
-      "loss": 6.9226,
+      "epoch": 0.31,
+      "learning_rate": 0.000366,
+      "loss": 6.9218,
       "step": 186
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00035999999999999997,
-      "loss": 7.1811,
+      "epoch": 0.31,
+      "learning_rate": 0.000368,
+      "loss": 7.0113,
       "step": 187
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000362,
-      "loss": 6.9784,
+      "epoch": 0.32,
+      "learning_rate": 0.00037,
+      "loss": 6.9827,
       "step": 188
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000364,
-      "loss": 7.0126,
+      "epoch": 0.32,
+      "learning_rate": 0.000372,
+      "loss": 7.0805,
       "step": 189
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000366,
-      "loss": 6.9648,
+      "epoch": 0.32,
+      "learning_rate": 0.000374,
+      "loss": 7.0587,
       "step": 190
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000368,
-      "loss": 7.178,
+      "epoch": 0.32,
+      "learning_rate": 0.00037600000000000003,
+      "loss": 7.0688,
       "step": 191
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00037,
-      "loss": 7.1971,
+      "epoch": 0.32,
+      "learning_rate": 0.000378,
+      "loss": 7.0879,
       "step": 192
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000372,
-      "loss": 7.0049,
+      "epoch": 0.32,
+      "learning_rate": 0.00038,
+      "loss": 6.9182,
       "step": 193
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000374,
-      "loss": 6.9962,
+      "epoch": 0.33,
+      "learning_rate": 0.000382,
+      "loss": 6.9707,
       "step": 194
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00037600000000000003,
-      "loss": 6.9091,
+      "epoch": 0.33,
+      "learning_rate": 0.000384,
+      "loss": 6.9151,
       "step": 195
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000378,
-      "loss": 6.7908,
+      "epoch": 0.33,
+      "learning_rate": 0.000386,
+      "loss": 6.7898,
       "step": 196
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00038,
-      "loss": 7.199,
+      "epoch": 0.33,
+      "learning_rate": 0.000388,
+      "loss": 6.9734,
       "step": 197
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000382,
-      "loss": 6.892,
+      "epoch": 0.33,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 6.9121,
       "step": 198
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000384,
-      "loss": 6.8718,
+      "epoch": 0.33,
+      "learning_rate": 0.00039200000000000004,
+      "loss": 6.6485,
       "step": 199
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000386,
-      "loss": 6.4145,
+      "epoch": 0.34,
+      "learning_rate": 0.00039400000000000004,
+      "loss": 6.6154,
       "step": 200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000388,
-      "loss": 7.1194,
+      "epoch": 0.34,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 7.0778,
       "step": 201
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00039000000000000005,
-      "loss": 7.0583,
+      "epoch": 0.34,
+      "learning_rate": 0.000398,
+      "loss": 7.0853,
       "step": 202
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00039200000000000004,
-      "loss": 7.1171,
+      "epoch": 0.34,
+      "learning_rate": 0.0004,
+      "loss": 7.1074,
       "step": 203
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00039400000000000004,
-      "loss": 6.9629,
+      "epoch": 0.34,
+      "learning_rate": 0.000402,
+      "loss": 6.8628,
       "step": 204
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00039600000000000003,
-      "loss": 6.98,
+      "epoch": 0.34,
+      "learning_rate": 0.000404,
+      "loss": 6.9699,
       "step": 205
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000398,
-      "loss": 7.0801,
+      "epoch": 0.35,
+      "learning_rate": 0.00040600000000000006,
+      "loss": 6.9795,
       "step": 206
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.0004,
-      "loss": 7.1003,
+      "epoch": 0.35,
+      "learning_rate": 0.000408,
+      "loss": 7.0912,
       "step": 207
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000402,
-      "loss": 7.2571,
+      "epoch": 0.35,
+      "learning_rate": 0.00041,
+      "loss": 7.0847,
       "step": 208
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000404,
-      "loss": 7.0932,
+      "epoch": 0.35,
+      "learning_rate": 0.000412,
+      "loss": 7.0106,
       "step": 209
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00040600000000000006,
-      "loss": 7.2553,
+      "epoch": 0.35,
+      "learning_rate": 0.000414,
+      "loss": 7.1056,
       "step": 210
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000408,
-      "loss": 7.0012,
+      "epoch": 0.35,
+      "learning_rate": 0.000416,
+      "loss": 7.0423,
       "step": 211
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00041,
-      "loss": 6.9778,
+      "epoch": 0.36,
+      "learning_rate": 0.00041799999999999997,
+      "loss": 7.0226,
       "step": 212
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000412,
-      "loss": 7.0422,
+      "epoch": 0.36,
+      "learning_rate": 0.00042,
+      "loss": 7.0216,
       "step": 213
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000414,
-      "loss": 7.0387,
+      "epoch": 0.36,
+      "learning_rate": 0.000422,
+      "loss": 7.0556,
       "step": 214
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000416,
-      "loss": 7.0497,
+      "epoch": 0.36,
+      "learning_rate": 0.000424,
+      "loss": 7.0019,
       "step": 215
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00041799999999999997,
-      "loss": 6.9891,
+      "epoch": 0.36,
+      "learning_rate": 0.000426,
+      "loss": 7.0848,
       "step": 216
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00042,
-      "loss": 7.0476,
+      "epoch": 0.36,
+      "learning_rate": 0.000428,
+      "loss": 6.8998,
       "step": 217
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000422,
-      "loss": 7.0848,
+      "epoch": 0.37,
+      "learning_rate": 0.00043,
+      "loss": 6.9194,
       "step": 218
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000424,
-      "loss": 6.9996,
+      "epoch": 0.37,
+      "learning_rate": 0.000432,
+      "loss": 6.9024,
       "step": 219
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000426,
-      "loss": 7.1819,
+      "epoch": 0.37,
+      "learning_rate": 0.00043400000000000003,
+      "loss": 7.1083,
       "step": 220
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000428,
-      "loss": 6.9997,
+      "epoch": 0.37,
+      "learning_rate": 0.000436,
+      "loss": 6.9852,
       "step": 221
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00043,
-      "loss": 7.3022,
+      "epoch": 0.37,
+      "learning_rate": 0.000438,
+      "loss": 7.0352,
       "step": 222
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000432,
-      "loss": 7.191,
+      "epoch": 0.37,
+      "learning_rate": 0.00044,
+      "loss": 7.0216,
       "step": 223
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00043400000000000003,
-      "loss": 7.1603,
+      "epoch": 0.38,
+      "learning_rate": 0.000442,
+      "loss": 7.0737,
       "step": 224
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000436,
-      "loss": 7.0109,
+      "epoch": 0.38,
+      "learning_rate": 0.000444,
+      "loss": 7.0748,
       "step": 225
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000438,
-      "loss": 6.8553,
+      "epoch": 0.38,
+      "learning_rate": 0.000446,
+      "loss": 6.9234,
       "step": 226
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00044,
-      "loss": 7.1697,
+      "epoch": 0.38,
+      "learning_rate": 0.000448,
+      "loss": 6.9104,
       "step": 227
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000442,
-      "loss": 7.0937,
+      "epoch": 0.38,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 7.0043,
       "step": 228
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000444,
-      "loss": 7.0137,
+      "epoch": 0.39,
+      "learning_rate": 0.00045200000000000004,
+      "loss": 7.0571,
       "step": 229
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000446,
-      "loss": 7.0811,
+      "epoch": 0.39,
+      "learning_rate": 0.00045400000000000003,
+      "loss": 7.0416,
       "step": 230
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000448,
-      "loss": 7.1647,
+      "epoch": 0.39,
+      "learning_rate": 0.000456,
+      "loss": 7.0239,
       "step": 231
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00045000000000000004,
-      "loss": 7.071,
+      "epoch": 0.39,
+      "learning_rate": 0.000458,
+      "loss": 6.9304,
       "step": 232
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00045200000000000004,
-      "loss": 7.0748,
+      "epoch": 0.39,
+      "learning_rate": 0.00046,
+      "loss": 7.0238,
       "step": 233
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00045400000000000003,
-      "loss": 6.9141,
+      "epoch": 0.39,
+      "learning_rate": 0.000462,
+      "loss": 7.0501,
       "step": 234
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000456,
-      "loss": 7.139,
+      "epoch": 0.4,
+      "learning_rate": 0.00046400000000000006,
+      "loss": 7.0763,
       "step": 235
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000458,
-      "loss": 7.1141,
+      "epoch": 0.4,
+      "learning_rate": 0.00046600000000000005,
+      "loss": 6.893,
       "step": 236
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00046,
-      "loss": 7.0931,
+      "epoch": 0.4,
+      "learning_rate": 0.00046800000000000005,
+      "loss": 6.9541,
       "step": 237
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000462,
-      "loss": 7.0072,
+      "epoch": 0.4,
+      "learning_rate": 0.00047,
+      "loss": 6.9877,
       "step": 238
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00046400000000000006,
-      "loss": 7.0934,
+      "epoch": 0.4,
+      "learning_rate": 0.000472,
+      "loss": 7.0107,
       "step": 239
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00046600000000000005,
-      "loss": 6.9072,
+      "epoch": 0.4,
+      "learning_rate": 0.000474,
+      "loss": 7.0396,
       "step": 240
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00046800000000000005,
-      "loss": 7.0331,
+      "epoch": 0.41,
+      "learning_rate": 0.00047599999999999997,
+      "loss": 7.1206,
       "step": 241
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00047,
-      "loss": 7.0503,
+      "epoch": 0.41,
+      "learning_rate": 0.00047799999999999996,
+      "loss": 7.0092,
       "step": 242
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000472,
-      "loss": 7.1425,
+      "epoch": 0.41,
+      "learning_rate": 0.00048,
+      "loss": 7.0429,
       "step": 243
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000474,
-      "loss": 7.1577,
+      "epoch": 0.41,
+      "learning_rate": 0.000482,
+      "loss": 6.9709,
       "step": 244
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00047599999999999997,
-      "loss": 6.8863,
+      "epoch": 0.41,
+      "learning_rate": 0.000484,
+      "loss": 6.8753,
       "step": 245
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00047799999999999996,
-      "loss": 6.844,
+      "epoch": 0.41,
+      "learning_rate": 0.000486,
+      "loss": 6.8831,
       "step": 246
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00048,
-      "loss": 7.4113,
+      "epoch": 0.42,
+      "learning_rate": 0.000488,
+      "loss": 6.8629,
       "step": 247
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000482,
-      "loss": 6.7754,
+      "epoch": 0.42,
+      "learning_rate": 0.00049,
+      "loss": 6.905,
       "step": 248
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000484,
-      "loss": 6.6743,
+      "epoch": 0.42,
+      "learning_rate": 0.000492,
+      "loss": 6.7441,
       "step": 249
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000486,
-      "loss": 6.4596,
+      "epoch": 0.42,
+      "learning_rate": 0.000494,
+      "loss": 6.5621,
       "step": 250
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000488,
-      "loss": 7.2881,
+      "epoch": 0.42,
+      "learning_rate": 0.000496,
+      "loss": 7.2098,
       "step": 251
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00049,
-      "loss": 7.0383,
+      "epoch": 0.42,
+      "learning_rate": 0.000498,
+      "loss": 7.2084,
       "step": 252
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000492,
-      "loss": 7.2685,
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 7.0214,
       "step": 253
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000494,
-      "loss": 6.9829,
+      "epoch": 0.43,
+      "learning_rate": 0.0005020000000000001,
+      "loss": 7.0158,
       "step": 254
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000496,
-      "loss": 7.0622,
+      "epoch": 0.43,
+      "learning_rate": 0.000504,
+      "loss": 7.1246,
       "step": 255
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000498,
-      "loss": 7.1413,
+      "epoch": 0.43,
+      "learning_rate": 0.000506,
+      "loss": 7.0502,
       "step": 256
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.0005,
-      "loss": 7.0308,
+      "epoch": 0.43,
+      "learning_rate": 0.000508,
+      "loss": 7.0624,
       "step": 257
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.0005020000000000001,
-      "loss": 7.183,
+      "epoch": 0.43,
+      "learning_rate": 0.00051,
+      "loss": 6.9986,
       "step": 258
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000504,
-      "loss": 7.0245,
+      "epoch": 0.44,
+      "learning_rate": 0.000512,
+      "loss": 7.1072,
       "step": 259
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000506,
-      "loss": 6.9604,
+      "epoch": 0.44,
+      "learning_rate": 0.000514,
+      "loss": 7.0238,
       "step": 260
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000508,
-      "loss": 7.0466,
+      "epoch": 0.44,
+      "learning_rate": 0.0005160000000000001,
+      "loss": 7.0155,
       "step": 261
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00051,
-      "loss": 7.0001,
+      "epoch": 0.44,
+      "learning_rate": 0.000518,
+      "loss": 6.9996,
       "step": 262
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000512,
-      "loss": 6.9583,
+      "epoch": 0.44,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 7.0465,
       "step": 263
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000514,
-      "loss": 7.0768,
+      "epoch": 0.44,
+      "learning_rate": 0.000522,
+      "loss": 6.8355,
       "step": 264
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.0005160000000000001,
-      "loss": 7.0325,
+      "epoch": 0.45,
+      "learning_rate": 0.000524,
+      "loss": 6.9596,
       "step": 265
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000518,
-      "loss": 7.0483,
+      "epoch": 0.45,
+      "learning_rate": 0.000526,
+      "loss": 6.9608,
       "step": 266
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.0005200000000000001,
-      "loss": 6.9711,
+      "epoch": 0.45,
+      "learning_rate": 0.000528,
+      "loss": 7.0486,
       "step": 267
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000522,
-      "loss": 6.9877,
+      "epoch": 0.45,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 6.9112,
       "step": 268
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000524,
-      "loss": 7.0728,
+      "epoch": 0.45,
+      "learning_rate": 0.000532,
+      "loss": 7.0594,
       "step": 269
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000526,
-      "loss": 7.0505,
+      "epoch": 0.45,
+      "learning_rate": 0.0005340000000000001,
+      "loss": 6.9925,
       "step": 270
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000528,
-      "loss": 7.1279,
+      "epoch": 0.46,
+      "learning_rate": 0.000536,
+      "loss": 7.0339,
       "step": 271
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0005300000000000001,
-      "loss": 6.9928,
+      "epoch": 0.46,
+      "learning_rate": 0.0005380000000000001,
+      "loss": 7.0785,
       "step": 272
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000532,
-      "loss": 7.0933,
+      "epoch": 0.46,
+      "learning_rate": 0.00054,
+      "loss": 7.0008,
       "step": 273
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0005340000000000001,
-      "loss": 7.1199,
+      "epoch": 0.46,
+      "learning_rate": 0.0005420000000000001,
+      "loss": 6.976,
       "step": 274
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000536,
-      "loss": 7.1138,
+      "epoch": 0.46,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 7.0215,
       "step": 275
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0005380000000000001,
-      "loss": 6.9262,
+      "epoch": 0.46,
+      "learning_rate": 0.000546,
+      "loss": 7.0251,
       "step": 276
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00054,
-      "loss": 7.0241,
+      "epoch": 0.47,
+      "learning_rate": 0.0005480000000000001,
+      "loss": 7.0722,
       "step": 277
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0005420000000000001,
-      "loss": 7.0357,
+      "epoch": 0.47,
+      "learning_rate": 0.00055,
+      "loss": 7.0051,
       "step": 278
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0005440000000000001,
-      "loss": 7.1868,
+      "epoch": 0.47,
+      "learning_rate": 0.0005520000000000001,
+      "loss": 7.0019,
       "step": 279
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000546,
-      "loss": 6.9221,
+      "epoch": 0.47,
+      "learning_rate": 0.000554,
+      "loss": 7.0538,
       "step": 280
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005480000000000001,
-      "loss": 6.9587,
+      "epoch": 0.47,
+      "learning_rate": 0.0005560000000000001,
+      "loss": 7.0292,
       "step": 281
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.00055,
-      "loss": 7.015,
+      "epoch": 0.47,
+      "learning_rate": 0.000558,
+      "loss": 7.0564,
       "step": 282
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005520000000000001,
-      "loss": 7.2622,
+      "epoch": 0.48,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 6.8884,
       "step": 283
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000554,
-      "loss": 7.1459,
+      "epoch": 0.48,
+      "learning_rate": 0.0005620000000000001,
+      "loss": 7.0051,
       "step": 284
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005560000000000001,
-      "loss": 7.072,
+      "epoch": 0.48,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 7.0457,
       "step": 285
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000558,
-      "loss": 7.0269,
+      "epoch": 0.48,
+      "learning_rate": 0.000566,
+      "loss": 6.9338,
       "step": 286
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005600000000000001,
-      "loss": 6.9213,
+      "epoch": 0.48,
+      "learning_rate": 0.0005679999999999999,
+      "loss": 7.1379,
       "step": 287
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005620000000000001,
-      "loss": 6.9281,
+      "epoch": 0.48,
+      "learning_rate": 0.00057,
+      "loss": 6.8807,
       "step": 288
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005639999999999999,
-      "loss": 7.0964,
+      "epoch": 0.49,
+      "learning_rate": 0.0005719999999999999,
+      "loss": 6.9174,
       "step": 289
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000566,
-      "loss": 6.9473,
+      "epoch": 0.49,
+      "learning_rate": 0.000574,
+      "loss": 7.0579,
       "step": 290
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.0005679999999999999,
-      "loss": 7.2092,
+      "epoch": 0.49,
+      "learning_rate": 0.000576,
+      "loss": 7.0882,
       "step": 291
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00057,
-      "loss": 7.1037,
+      "epoch": 0.49,
+      "learning_rate": 0.000578,
+      "loss": 6.9218,
       "step": 292
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.0005719999999999999,
-      "loss": 6.7986,
+      "epoch": 0.49,
+      "learning_rate": 0.00058,
+      "loss": 6.9869,
       "step": 293
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000574,
-      "loss": 6.8426,
+      "epoch": 0.49,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 7.0547,
       "step": 294
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000576,
-      "loss": 7.0094,
+      "epoch": 0.5,
+      "learning_rate": 0.000584,
+      "loss": 6.8407,
       "step": 295
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000578,
-      "loss": 7.1792,
+      "epoch": 0.5,
+      "learning_rate": 0.0005859999999999999,
+      "loss": 6.8458,
       "step": 296
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00058,
-      "loss": 6.8114,
+      "epoch": 0.5,
+      "learning_rate": 0.000588,
+      "loss": 6.7859,
       "step": 297
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.0005819999999999999,
-      "loss": 6.8439,
+      "epoch": 0.5,
+      "learning_rate": 0.00059,
+      "loss": 6.8682,
       "step": 298
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000584,
-      "loss": 6.9772,
+      "epoch": 0.5,
+      "learning_rate": 0.000592,
+      "loss": 6.6934,
       "step": 299
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.0005859999999999999,
-      "loss": 6.5952,
+      "epoch": 0.5,
+      "learning_rate": 0.000594,
+      "loss": 6.5845,
       "step": 300
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000588,
-      "loss": 7.1037,
+      "epoch": 0.51,
+      "learning_rate": 0.000596,
+      "loss": 7.236,
       "step": 301
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00059,
-      "loss": 7.229,
+      "epoch": 0.51,
+      "learning_rate": 0.000598,
+      "loss": 7.2218,
       "step": 302
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000592,
-      "loss": 7.2436,
+      "epoch": 0.51,
+      "learning_rate": 0.0006,
+      "loss": 7.1656,
       "step": 303
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000594,
-      "loss": 6.9555,
+      "epoch": 0.51,
+      "learning_rate": 0.000602,
+      "loss": 7.0663,
       "step": 304
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000596,
-      "loss": 7.0729,
+      "epoch": 0.51,
+      "learning_rate": 0.000604,
+      "loss": 7.1259,
       "step": 305
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000598,
-      "loss": 7.0202,
+      "epoch": 0.51,
+      "learning_rate": 0.000606,
+      "loss": 7.1706,
       "step": 306
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.0006,
-      "loss": 7.1313,
+      "epoch": 0.52,
+      "learning_rate": 0.000608,
+      "loss": 7.0302,
       "step": 307
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000602,
-      "loss": 6.9928,
+      "epoch": 0.52,
+      "learning_rate": 0.00061,
+      "loss": 7.132,
       "step": 308
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000604,
-      "loss": 6.8953,
+      "epoch": 0.52,
+      "learning_rate": 0.000612,
+      "loss": 7.0026,
       "step": 309
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000606,
-      "loss": 7.1579,
+      "epoch": 0.52,
+      "learning_rate": 0.000614,
+      "loss": 7.0629,
       "step": 310
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000608,
-      "loss": 6.8321,
+      "epoch": 0.52,
+      "learning_rate": 0.000616,
+      "loss": 6.9223,
       "step": 311
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00061,
-      "loss": 6.9062,
+      "epoch": 0.52,
+      "learning_rate": 0.0006180000000000001,
+      "loss": 7.16,
       "step": 312
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000612,
-      "loss": 7.0055,
+      "epoch": 0.53,
+      "learning_rate": 0.00062,
+      "loss": 6.9756,
       "step": 313
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000614,
-      "loss": 7.0492,
+      "epoch": 0.53,
+      "learning_rate": 0.000622,
+      "loss": 6.9914,
       "step": 314
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000616,
-      "loss": 6.9698,
+      "epoch": 0.53,
+      "learning_rate": 0.000624,
+      "loss": 7.018,
       "step": 315
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.0006180000000000001,
-      "loss": 7.1945,
+      "epoch": 0.53,
+      "learning_rate": 0.000626,
+      "loss": 6.948,
       "step": 316
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00062,
-      "loss": 7.0507,
+      "epoch": 0.53,
+      "learning_rate": 0.000628,
+      "loss": 6.9204,
       "step": 317
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000622,
-      "loss": 7.07,
+      "epoch": 0.53,
+      "learning_rate": 0.00063,
+      "loss": 7.1444,
       "step": 318
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000624,
-      "loss": 6.9557,
+      "epoch": 0.54,
+      "learning_rate": 0.000632,
+      "loss": 7.0219,
       "step": 319
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000626,
-      "loss": 6.956,
+      "epoch": 0.54,
+      "learning_rate": 0.000634,
+      "loss": 7.0609,
       "step": 320
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000628,
-      "loss": 6.9846,
+      "epoch": 0.54,
+      "learning_rate": 0.0006360000000000001,
+      "loss": 6.9684,
       "step": 321
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00063,
-      "loss": 6.9683,
+      "epoch": 0.54,
+      "learning_rate": 0.000638,
+      "loss": 6.9689,
       "step": 322
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000632,
-      "loss": 6.9356,
+      "epoch": 0.54,
+      "learning_rate": 0.00064,
+      "loss": 6.9515,
       "step": 323
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000634,
-      "loss": 7.0013,
+      "epoch": 0.54,
+      "learning_rate": 0.000642,
+      "loss": 6.9158,
       "step": 324
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.0006360000000000001,
-      "loss": 6.8627,
+      "epoch": 0.55,
+      "learning_rate": 0.000644,
+      "loss": 7.1327,
       "step": 325
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000638,
-      "loss": 7.0897,
+      "epoch": 0.55,
+      "learning_rate": 0.000646,
+      "loss": 7.0171,
       "step": 326
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00064,
-      "loss": 7.0129,
+      "epoch": 0.55,
+      "learning_rate": 0.000648,
+      "loss": 7.1963,
       "step": 327
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000642,
-      "loss": 7.0737,
+      "epoch": 0.55,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 7.0119,
       "step": 328
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000644,
-      "loss": 6.9392,
+      "epoch": 0.55,
+      "learning_rate": 0.000652,
+      "loss": 6.9315,
       "step": 329
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000646,
-      "loss": 7.1556,
+      "epoch": 0.55,
+      "learning_rate": 0.0006540000000000001,
+      "loss": 7.0296,
       "step": 330
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000648,
-      "loss": 7.0289,
+      "epoch": 0.56,
+      "learning_rate": 0.000656,
+      "loss": 6.9909,
       "step": 331
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0006500000000000001,
-      "loss": 7.0916,
+      "epoch": 0.56,
+      "learning_rate": 0.0006580000000000001,
+      "loss": 6.9743,
       "step": 332
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000652,
-      "loss": 7.1824,
+      "epoch": 0.56,
+      "learning_rate": 0.00066,
+      "loss": 6.9316,
       "step": 333
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0006540000000000001,
-      "loss": 6.9691,
+      "epoch": 0.56,
+      "learning_rate": 0.000662,
+      "loss": 7.0359,
       "step": 334
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000656,
-      "loss": 7.0728,
+      "epoch": 0.56,
+      "learning_rate": 0.0006640000000000001,
+      "loss": 7.0403,
       "step": 335
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0006580000000000001,
-      "loss": 7.121,
+      "epoch": 0.56,
+      "learning_rate": 0.000666,
+      "loss": 6.9493,
       "step": 336
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.00066,
-      "loss": 6.8791,
+      "epoch": 0.57,
+      "learning_rate": 0.0006680000000000001,
+      "loss": 6.991,
       "step": 337
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000662,
-      "loss": 7.1717,
+      "epoch": 0.57,
+      "learning_rate": 0.00067,
+      "loss": 6.9762,
       "step": 338
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0006640000000000001,
-      "loss": 6.9594,
+      "epoch": 0.57,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 7.0749,
       "step": 339
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000666,
-      "loss": 7.0685,
+      "epoch": 0.57,
+      "learning_rate": 0.000674,
+      "loss": 7.1129,
       "step": 340
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006680000000000001,
-      "loss": 6.9809,
+      "epoch": 0.57,
+      "learning_rate": 0.0006760000000000001,
+      "loss": 6.9891,
       "step": 341
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00067,
-      "loss": 7.1153,
+      "epoch": 0.58,
+      "learning_rate": 0.0006780000000000001,
+      "loss": 6.9923,
       "step": 342
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006720000000000001,
-      "loss": 6.9467,
+      "epoch": 0.58,
+      "learning_rate": 0.00068,
+      "loss": 6.8966,
       "step": 343
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000674,
-      "loss": 7.0976,
+      "epoch": 0.58,
+      "learning_rate": 0.0006820000000000001,
+      "loss": 7.0609,
       "step": 344
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006760000000000001,
-      "loss": 6.8509,
+      "epoch": 0.58,
+      "learning_rate": 0.000684,
+      "loss": 6.9641,
       "step": 345
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006780000000000001,
-      "loss": 6.8434,
+      "epoch": 0.58,
+      "learning_rate": 0.0006860000000000001,
+      "loss": 6.7934,
       "step": 346
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00068,
-      "loss": 7.018,
+      "epoch": 0.58,
+      "learning_rate": 0.0006879999999999999,
+      "loss": 6.814,
       "step": 347
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006820000000000001,
+      "epoch": 0.59,
+      "learning_rate": 0.00069,
       "loss": 7.1134,
       "step": 348
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000684,
-      "loss": 6.7391,
+      "epoch": 0.59,
+      "learning_rate": 0.000692,
+      "loss": 6.7713,
       "step": 349
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006860000000000001,
-      "loss": 6.9186,
+      "epoch": 0.59,
+      "learning_rate": 0.000694,
+      "loss": 6.7095,
       "step": 350
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0006879999999999999,
-      "loss": 7.3938,
+      "epoch": 0.59,
+      "learning_rate": 0.000696,
+      "loss": 7.2116,
       "step": 351
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.00069,
-      "loss": 7.2696,
+      "epoch": 0.59,
+      "learning_rate": 0.0006979999999999999,
+      "loss": 7.1415,
       "step": 352
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000692,
-      "loss": 7.1135,
+      "epoch": 0.59,
+      "learning_rate": 0.0007,
+      "loss": 7.0851,
       "step": 353
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000694,
-      "loss": 7.1692,
+      "epoch": 0.6,
+      "learning_rate": 0.0007019999999999999,
+      "loss": 7.0795,
       "step": 354
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000696,
-      "loss": 7.0229,
+      "epoch": 0.6,
+      "learning_rate": 0.000704,
+      "loss": 7.0815,
       "step": 355
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0006979999999999999,
-      "loss": 7.1764,
+      "epoch": 0.6,
+      "learning_rate": 0.0007059999999999999,
+      "loss": 7.1496,
       "step": 356
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0007,
-      "loss": 7.1834,
+      "epoch": 0.6,
+      "learning_rate": 0.000708,
+      "loss": 7.0391,
       "step": 357
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0007019999999999999,
-      "loss": 6.9568,
+      "epoch": 0.6,
+      "learning_rate": 0.00071,
+      "loss": 7.0427,
       "step": 358
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000704,
-      "loss": 7.0929,
+      "epoch": 0.6,
+      "learning_rate": 0.000712,
+      "loss": 7.006,
       "step": 359
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0007059999999999999,
-      "loss": 7.1641,
+      "epoch": 0.61,
+      "learning_rate": 0.000714,
+      "loss": 7.0213,
       "step": 360
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000708,
-      "loss": 6.9673,
+      "epoch": 0.61,
+      "learning_rate": 0.000716,
+      "loss": 7.0563,
       "step": 361
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.00071,
-      "loss": 6.9279,
+      "epoch": 0.61,
+      "learning_rate": 0.000718,
+      "loss": 6.9326,
       "step": 362
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000712,
-      "loss": 7.1738,
+      "epoch": 0.61,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 285.9326,
       "step": 363
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000714,
-      "loss": 7.1017,
+      "epoch": 0.61,
+      "learning_rate": 0.000722,
+      "loss": 7.0385,
       "step": 364
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000716,
-      "loss": 6.9808,
+      "epoch": 0.61,
+      "learning_rate": 0.000724,
+      "loss": 7.1396,
       "step": 365
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000718,
-      "loss": 7.0973,
+      "epoch": 0.62,
+      "learning_rate": 0.000726,
+      "loss": 6.8595,
       "step": 366
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0007199999999999999,
-      "loss": 7.0794,
+      "epoch": 0.62,
+      "learning_rate": 0.000728,
+      "loss": 7.1262,
       "step": 367
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000722,
-      "loss": 7.1846,
+      "epoch": 0.62,
+      "learning_rate": 0.00073,
+      "loss": 6.9156,
       "step": 368
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000724,
-      "loss": 7.0808,
+      "epoch": 0.62,
+      "learning_rate": 0.000732,
+      "loss": 6.9526,
       "step": 369
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000726,
-      "loss": 7.1264,
+      "epoch": 0.62,
+      "learning_rate": 0.000734,
+      "loss": 7.0985,
       "step": 370
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000728,
-      "loss": 6.9534,
+      "epoch": 0.62,
+      "learning_rate": 0.000736,
+      "loss": 6.993,
       "step": 371
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00073,
-      "loss": 6.9788,
+      "epoch": 0.63,
+      "learning_rate": 0.000738,
+      "loss": 6.9229,
       "step": 372
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000732,
-      "loss": 6.9074,
+      "epoch": 0.63,
+      "learning_rate": 0.00074,
+      "loss": 6.9861,
       "step": 373
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000734,
-      "loss": 6.9939,
+      "epoch": 0.63,
+      "learning_rate": 0.000742,
+      "loss": 6.911,
       "step": 374
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000736,
-      "loss": 6.9457,
+      "epoch": 0.63,
+      "learning_rate": 0.000744,
+      "loss": 7.0069,
       "step": 375
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000738,
-      "loss": 7.111,
+      "epoch": 0.63,
+      "learning_rate": 0.000746,
+      "loss": 6.9277,
       "step": 376
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00074,
-      "loss": 6.9439,
+      "epoch": 0.63,
+      "learning_rate": 0.000748,
+      "loss": 6.956,
       "step": 377
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000742,
-      "loss": 7.0072,
+      "epoch": 0.64,
+      "learning_rate": 0.00075,
+      "loss": 7.0714,
       "step": 378
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000744,
-      "loss": 7.0979,
+      "epoch": 0.64,
+      "learning_rate": 0.0007520000000000001,
+      "loss": 6.9807,
       "step": 379
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000746,
-      "loss": 7.0279,
+      "epoch": 0.64,
+      "learning_rate": 0.000754,
+      "loss": 7.108,
       "step": 380
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000748,
-      "loss": 7.0169,
+      "epoch": 0.64,
+      "learning_rate": 0.000756,
+      "loss": 7.0734,
       "step": 381
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00075,
-      "loss": 7.0453,
+      "epoch": 0.64,
+      "learning_rate": 0.000758,
+      "loss": 7.0057,
       "step": 382
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0007520000000000001,
-      "loss": 6.8604,
+      "epoch": 0.64,
+      "learning_rate": 0.00076,
+      "loss": 6.9243,
       "step": 383
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000754,
-      "loss": 7.2136,
+      "epoch": 0.65,
+      "learning_rate": 0.000762,
+      "loss": 7.031,
       "step": 384
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000756,
-      "loss": 7.0006,
+      "epoch": 0.65,
+      "learning_rate": 0.000764,
+      "loss": 7.0799,
       "step": 385
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000758,
-      "loss": 6.8976,
+      "epoch": 0.65,
+      "learning_rate": 0.0007660000000000001,
+      "loss": 7.0769,
       "step": 386
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.00076,
-      "loss": 6.9871,
+      "epoch": 0.65,
+      "learning_rate": 0.000768,
+      "loss": 7.1266,
       "step": 387
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000762,
-      "loss": 7.0362,
+      "epoch": 0.65,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 7.0542,
       "step": 388
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000764,
-      "loss": 7.1718,
+      "epoch": 0.65,
+      "learning_rate": 0.000772,
+      "loss": 6.965,
       "step": 389
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0007660000000000001,
-      "loss": 7.3146,
+      "epoch": 0.66,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 7.0061,
       "step": 390
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000768,
-      "loss": 7.1342,
+      "epoch": 0.66,
+      "learning_rate": 0.000776,
+      "loss": 7.089,
       "step": 391
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0007700000000000001,
-      "loss": 6.9868,
+      "epoch": 0.66,
+      "learning_rate": 0.000778,
+      "loss": 7.0489,
       "step": 392
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000772,
-      "loss": 6.8952,
+      "epoch": 0.66,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 7.0362,
       "step": 393
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0007740000000000001,
-      "loss": 7.0074,
+      "epoch": 0.66,
+      "learning_rate": 0.000782,
+      "loss": 6.9978,
       "step": 394
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000776,
-      "loss": 6.8717,
+      "epoch": 0.66,
+      "learning_rate": 0.0007840000000000001,
+      "loss": 6.7938,
       "step": 395
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000778,
-      "loss": 6.9778,
+      "epoch": 0.67,
+      "learning_rate": 0.000786,
+      "loss": 6.7785,
       "step": 396
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0007800000000000001,
-      "loss": 6.7461,
+      "epoch": 0.67,
+      "learning_rate": 0.0007880000000000001,
+      "loss": 6.9116,
       "step": 397
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000782,
-      "loss": 6.7326,
+      "epoch": 0.67,
+      "learning_rate": 0.00079,
+      "loss": 7.0112,
       "step": 398
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0007840000000000001,
-      "loss": 6.6938,
+      "epoch": 0.67,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 6.6276,
       "step": 399
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000786,
-      "loss": 6.4652,
+      "epoch": 0.67,
+      "learning_rate": 0.0007940000000000001,
+      "loss": 6.4829,
       "step": 400
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0007880000000000001,
-      "loss": 7.2434,
+      "epoch": 0.67,
+      "learning_rate": 0.000796,
+      "loss": 7.3835,
       "step": 401
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.00079,
-      "loss": 7.1463,
+      "epoch": 0.68,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 7.2009,
       "step": 402
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0007920000000000001,
-      "loss": 7.1114,
+      "epoch": 0.68,
+      "learning_rate": 0.0008,
+      "loss": 7.1478,
       "step": 403
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0007940000000000001,
-      "loss": 7.2073,
+      "epoch": 0.68,
+      "learning_rate": 0.0008020000000000001,
+      "loss": 7.112,
       "step": 404
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000796,
-      "loss": 7.0008,
+      "epoch": 0.68,
+      "learning_rate": 0.000804,
+      "loss": 7.0844,
       "step": 405
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0007980000000000001,
-      "loss": 7.0475,
+      "epoch": 0.68,
+      "learning_rate": 0.0008060000000000001,
+      "loss": 6.9843,
       "step": 406
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0008,
-      "loss": 7.0595,
+      "epoch": 0.68,
+      "learning_rate": 0.000808,
+      "loss": 6.9488,
       "step": 407
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0008020000000000001,
-      "loss": 7.1738,
+      "epoch": 0.69,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 7.1111,
       "step": 408
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000804,
-      "loss": 8.4566,
+      "epoch": 0.69,
+      "learning_rate": 0.0008120000000000001,
+      "loss": 6.9722,
       "step": 409
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0008060000000000001,
-      "loss": 7.151,
+      "epoch": 0.69,
+      "learning_rate": 0.0008139999999999999,
+      "loss": 6.9885,
       "step": 410
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000808,
-      "loss": 7.1658,
+      "epoch": 0.69,
+      "learning_rate": 0.000816,
+      "loss": 7.0014,
       "step": 411
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0008100000000000001,
-      "loss": 7.1185,
+      "epoch": 0.69,
+      "learning_rate": 0.0008179999999999999,
+      "loss": 7.0313,
       "step": 412
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0008120000000000001,
-      "loss": 7.061,
+      "epoch": 0.69,
+      "learning_rate": 0.00082,
+      "loss": 6.8849,
       "step": 413
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0008139999999999999,
-      "loss": 6.9386,
+      "epoch": 0.7,
+      "learning_rate": 0.0008219999999999999,
+      "loss": 7.0015,
       "step": 414
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000816,
-      "loss": 7.0055,
+      "epoch": 0.7,
+      "learning_rate": 0.000824,
+      "loss": 7.1869,
       "step": 415
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0008179999999999999,
-      "loss": 7.2785,
+      "epoch": 0.7,
+      "learning_rate": 0.000826,
+      "loss": 6.9322,
       "step": 416
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.00082,
-      "loss": 6.9835,
+      "epoch": 0.7,
+      "learning_rate": 0.000828,
+      "loss": 7.0145,
       "step": 417
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0008219999999999999,
-      "loss": 6.8751,
+      "epoch": 0.7,
+      "learning_rate": 0.00083,
+      "loss": 6.812,
       "step": 418
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000824,
-      "loss": 6.9674,
+      "epoch": 0.7,
+      "learning_rate": 0.000832,
+      "loss": 6.9888,
       "step": 419
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000826,
-      "loss": 7.1833,
+      "epoch": 0.71,
+      "learning_rate": 0.000834,
+      "loss": 7.0677,
       "step": 420
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000828,
-      "loss": 7.0289,
+      "epoch": 0.71,
+      "learning_rate": 0.0008359999999999999,
+      "loss": 6.9418,
       "step": 421
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.00083,
-      "loss": 6.9144,
+      "epoch": 0.71,
+      "learning_rate": 0.000838,
+      "loss": 6.9906,
       "step": 422
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000832,
-      "loss": 7.0525,
+      "epoch": 0.71,
+      "learning_rate": 0.00084,
+      "loss": 7.0385,
       "step": 423
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000834,
-      "loss": 6.9794,
+      "epoch": 0.71,
+      "learning_rate": 0.000842,
+      "loss": 7.0314,
       "step": 424
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.0008359999999999999,
-      "loss": 7.2218,
+      "epoch": 0.71,
+      "learning_rate": 0.000844,
+      "loss": 6.9856,
       "step": 425
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000838,
-      "loss": 7.031,
+      "epoch": 0.72,
+      "learning_rate": 0.000846,
+      "loss": 7.0841,
       "step": 426
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00084,
-      "loss": 7.0031,
+      "epoch": 0.72,
+      "learning_rate": 0.000848,
+      "loss": 6.9897,
       "step": 427
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000842,
-      "loss": 7.0494,
+      "epoch": 0.72,
+      "learning_rate": 0.00085,
+      "loss": 6.9838,
       "step": 428
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000844,
-      "loss": 7.1181,
+      "epoch": 0.72,
+      "learning_rate": 0.000852,
+      "loss": 7.1931,
       "step": 429
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000846,
-      "loss": 7.0683,
+      "epoch": 0.72,
+      "learning_rate": 0.000854,
+      "loss": 6.9795,
       "step": 430
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000848,
-      "loss": 7.1466,
+      "epoch": 0.72,
+      "learning_rate": 0.000856,
+      "loss": 6.9566,
       "step": 431
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00085,
-      "loss": 6.8249,
+      "epoch": 0.73,
+      "learning_rate": 0.000858,
+      "loss": 6.8935,
       "step": 432
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000852,
-      "loss": 7.0882,
+      "epoch": 0.73,
+      "learning_rate": 0.00086,
+      "loss": 7.0856,
       "step": 433
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000854,
-      "loss": 7.0593,
+      "epoch": 0.73,
+      "learning_rate": 0.000862,
+      "loss": 7.0538,
       "step": 434
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000856,
-      "loss": 7.088,
+      "epoch": 0.73,
+      "learning_rate": 0.000864,
+      "loss": 7.1257,
       "step": 435
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000858,
-      "loss": 6.8876,
+      "epoch": 0.73,
+      "learning_rate": 0.000866,
+      "loss": 6.9539,
       "step": 436
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00086,
-      "loss": 6.9609,
+      "epoch": 0.73,
+      "learning_rate": 0.0008680000000000001,
+      "loss": 6.9312,
       "step": 437
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000862,
-      "loss": 6.9152,
+      "epoch": 0.74,
+      "learning_rate": 0.00087,
+      "loss": 7.0382,
       "step": 438
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000864,
-      "loss": 7.2055,
+      "epoch": 0.74,
+      "learning_rate": 0.000872,
+      "loss": 7.0195,
       "step": 439
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000866,
-      "loss": 6.99,
+      "epoch": 0.74,
+      "learning_rate": 0.000874,
+      "loss": 7.1437,
       "step": 440
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.0008680000000000001,
-      "loss": 7.0483,
+      "epoch": 0.74,
+      "learning_rate": 0.000876,
+      "loss": 6.9629,
       "step": 441
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00087,
-      "loss": 6.9639,
+      "epoch": 0.74,
+      "learning_rate": 0.000878,
+      "loss": 7.1108,
       "step": 442
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000872,
-      "loss": 7.0258,
+      "epoch": 0.74,
+      "learning_rate": 0.00088,
+      "loss": 6.9873,
       "step": 443
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000874,
-      "loss": 6.9092,
+      "epoch": 0.75,
+      "learning_rate": 0.000882,
+      "loss": 6.8853,
       "step": 444
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000876,
-      "loss": 6.7097,
+      "epoch": 0.75,
+      "learning_rate": 0.000884,
+      "loss": 6.8583,
       "step": 445
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000878,
-      "loss": 6.9751,
+      "epoch": 0.75,
+      "learning_rate": 0.0008860000000000001,
+      "loss": 7.0631,
       "step": 446
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00088,
-      "loss": 6.9166,
+      "epoch": 0.75,
+      "learning_rate": 0.000888,
+      "loss": 6.8263,
       "step": 447
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000882,
-      "loss": 6.9943,
+      "epoch": 0.75,
+      "learning_rate": 0.0008900000000000001,
+      "loss": 6.6845,
       "step": 448
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000884,
-      "loss": 6.7797,
+      "epoch": 0.75,
+      "learning_rate": 0.000892,
+      "loss": 6.8607,
       "step": 449
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0008860000000000001,
-      "loss": 6.5635,
+      "epoch": 0.76,
+      "learning_rate": 0.000894,
+      "loss": 6.5526,
       "step": 450
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000888,
-      "loss": 7.0689,
+      "epoch": 0.76,
+      "learning_rate": 0.000896,
+      "loss": 7.4182,
       "step": 451
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0008900000000000001,
-      "loss": 7.0351,
+      "epoch": 0.76,
+      "learning_rate": 0.000898,
+      "loss": 7.2457,
       "step": 452
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000892,
-      "loss": 31.0659,
+      "epoch": 0.76,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 7.1327,
       "step": 453
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000894,
-      "loss": 7.03,
+      "epoch": 0.76,
+      "learning_rate": 0.000902,
+      "loss": 6.9056,
       "step": 454
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000896,
-      "loss": 7.0468,
+      "epoch": 0.77,
+      "learning_rate": 0.0009040000000000001,
+      "loss": 7.174,
       "step": 455
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000898,
-      "loss": 7.057,
+      "epoch": 0.77,
+      "learning_rate": 0.000906,
+      "loss": 7.0875,
       "step": 456
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0009000000000000001,
-      "loss": 7.0597,
+      "epoch": 0.77,
+      "learning_rate": 0.0009080000000000001,
+      "loss": 7.0482,
       "step": 457
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000902,
-      "loss": 7.1754,
+      "epoch": 0.77,
+      "learning_rate": 0.00091,
+      "loss": 7.2428,
       "step": 458
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0009040000000000001,
-      "loss": 7.1328,
+      "epoch": 0.77,
+      "learning_rate": 0.000912,
+      "loss": 6.9443,
       "step": 459
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000906,
-      "loss": 7.1022,
+      "epoch": 0.77,
+      "learning_rate": 0.0009140000000000001,
+      "loss": 6.978,
       "step": 460
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0009080000000000001,
-      "loss": 7.0483,
+      "epoch": 0.78,
+      "learning_rate": 0.000916,
+      "loss": 7.0511,
       "step": 461
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00091,
-      "loss": 10.5709,
+      "epoch": 0.78,
+      "learning_rate": 0.0009180000000000001,
+      "loss": 6.9723,
       "step": 462
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000912,
-      "loss": 7.0752,
+      "epoch": 0.78,
+      "learning_rate": 0.00092,
+      "loss": 6.9377,
       "step": 463
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0009140000000000001,
-      "loss": 7.1188,
+      "epoch": 0.78,
+      "learning_rate": 0.0009220000000000001,
+      "loss": 6.9787,
       "step": 464
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000916,
-      "loss": 7.0777,
+      "epoch": 0.78,
+      "learning_rate": 0.000924,
+      "loss": 7.0386,
       "step": 465
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0009180000000000001,
-      "loss": 7.0153,
+      "epoch": 0.78,
+      "learning_rate": 0.0009260000000000001,
+      "loss": 7.041,
       "step": 466
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00092,
-      "loss": 6.8818,
-      "step": 467
+      "epoch": 0.79,
+      "learning_rate": 0.0009280000000000001,
+      "loss": 7.1201,
+      "step": 467
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0009220000000000001,
-      "loss": 6.9739,
+      "epoch": 0.79,
+      "learning_rate": 0.00093,
+      "loss": 6.9869,
       "step": 468
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000924,
-      "loss": 6.8447,
+      "epoch": 0.79,
+      "learning_rate": 0.0009320000000000001,
+      "loss": 6.9156,
       "step": 469
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0009260000000000001,
-      "loss": 7.2424,
+      "epoch": 0.79,
+      "learning_rate": 0.000934,
+      "loss": 6.9832,
       "step": 470
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0009280000000000001,
-      "loss": 7.0031,
+      "epoch": 0.79,
+      "learning_rate": 0.0009360000000000001,
+      "loss": 6.9666,
       "step": 471
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.00093,
-      "loss": 7.1577,
+      "epoch": 0.79,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 7.0176,
       "step": 472
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0009320000000000001,
-      "loss": 7.0784,
+      "epoch": 0.8,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 473
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000934,
-      "loss": 7.1943,
+      "epoch": 0.8,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 474
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0009360000000000001,
-      "loss": 6.9866,
+      "epoch": 0.8,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 475
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 0.0009379999999999999,
-      "loss": 6.8905,
+      "loss": 0.0,
       "step": 476
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.00094,
-      "loss": 6.8424,
+      "epoch": 0.8,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 477
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000942,
-      "loss": 7.0373,
+      "epoch": 0.8,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 478
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000944,
-      "loss": 6.9945,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 479
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000946,
-      "loss": 7.1459,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 480
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000948,
-      "loss": 6.942,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 481
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.00095,
-      "loss": 6.9965,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 482
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0009519999999999999,
-      "loss": 7.0274,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 483
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000954,
-      "loss": 7.0962,
+      "epoch": 0.81,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 484
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0009559999999999999,
-      "loss": 7.0712,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 485
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000958,
-      "loss": 6.9201,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 486
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.00096,
-      "loss": 6.9673,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 487
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000962,
-      "loss": 7.0785,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 488
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000964,
-      "loss": 7.0721,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 489
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000966,
-      "loss": 6.9475,
+      "epoch": 0.82,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 490
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000968,
-      "loss": 7.1285,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 491
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0009699999999999999,
-      "loss": 7.1993,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 492
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000972,
-      "loss": 7.1318,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 493
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000974,
-      "loss": 7.1,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 494
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000976,
-      "loss": 6.9925,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 495
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000978,
-      "loss": 6.9424,
+      "epoch": 0.83,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 496
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00098,
-      "loss": 6.9446,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 497
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000982,
-      "loss": 6.8137,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 498
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000984,
-      "loss": 6.7793,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 499
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.0009860000000000001,
-      "loss": 6.6679,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 500
     },
     {
-      "epoch": 0.42,
-      "eval_loss": 7.067996978759766,
-      "eval_runtime": 346.2794,
-      "eval_samples_per_second": 7.63,
-      "eval_steps_per_second": 0.638,
-      "eval_wer": 1.0,
+      "epoch": 0.84,
+      "eval_loss": NaN,
+      "eval_runtime": 772.1245,
+      "eval_samples_per_second": 3.422,
+      "eval_steps_per_second": 0.286,
+      "eval_wer": 2.0097029585094353,
       "step": 500
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000988,
-      "loss": 7.1796,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 501
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00099,
-      "loss": 23.0601,
+      "epoch": 0.84,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 502
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000992,
-      "loss": 7.1719,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 503
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000994,
-      "loss": 7.1177,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 504
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000996,
-      "loss": 7.0817,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 505
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000998,
-      "loss": 7.2064,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 506
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.001,
-      "loss": 7.048,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 507
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009985486211901307,
-      "loss": 7.0498,
+      "epoch": 0.85,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 508
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009970972423802612,
-      "loss": 7.019,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 509
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000995645863570392,
-      "loss": 7.0718,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 510
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009941944847605226,
-      "loss": 6.8736,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 511
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009927431059506531,
-      "loss": 7.0358,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 512
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009912917271407838,
-      "loss": 7.0267,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 513
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009898403483309143,
-      "loss": 6.9751,
+      "epoch": 0.86,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 514
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000988388969521045,
-      "loss": 7.0864,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 515
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009869375907111755,
-      "loss": 6.9706,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 516
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0009854862119013062,
-      "loss": 7.0958,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 517
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000984034833091437,
-      "loss": 6.9314,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 518
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009825834542815674,
-      "loss": 7.2066,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 519
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009811320754716981,
-      "loss": 6.9461,
+      "epoch": 0.87,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 520
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009796806966618288,
-      "loss": 7.0822,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 521
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009782293178519593,
-      "loss": 7.0513,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 522
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.00097677793904209,
-      "loss": 7.0695,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 523
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009753265602322206,
-      "loss": 6.9845,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 524
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009738751814223512,
-      "loss": 6.9712,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 525
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009724238026124818,
-      "loss": 6.8825,
+      "epoch": 0.88,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 526
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009709724238026125,
-      "loss": 7.0213,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 527
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000969521044992743,
-      "loss": 7.0573,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 528
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0009680696661828737,
-      "loss": 7.0032,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 529
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009666182873730044,
-      "loss": 7.0469,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 530
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000965166908563135,
-      "loss": 7.011,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 531
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009637155297532656,
-      "loss": 7.0697,
+      "epoch": 0.89,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 532
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009622641509433962,
-      "loss": 6.8978,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 533
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009608127721335269,
-      "loss": 6.994,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 534
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009593613933236574,
-      "loss": 7.099,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 535
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009579100145137881,
-      "loss": 6.9542,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 536
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009564586357039187,
-      "loss": 6.9809,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 537
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009550072568940493,
-      "loss": 6.8488,
+      "epoch": 0.9,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 538
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009535558780841799,
-      "loss": 6.9842,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 539
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009521044992743106,
-      "loss": 7.1183,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 540
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0009506531204644411,
-      "loss": 7.1515,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 541
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0009492017416545718,
-      "loss": 7.0061,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 542
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0009477503628447026,
-      "loss": 6.8293,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 543
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0009462989840348331,
-      "loss": 7.1905,
+      "epoch": 0.91,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
       "step": 544
     },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009448476052249637,
-      "loss": 7.0211,
-      "step": 545
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009433962264150943,
-      "loss": 6.9183,
-      "step": 546
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.000941944847605225,
-      "loss": 6.6965,
-      "step": 547
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009404934687953555,
-      "loss": 6.9138,
-      "step": 548
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009390420899854863,
-      "loss": 6.7907,
-      "step": 549
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009375907111756169,
-      "loss": 6.7276,
-      "step": 550
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009361393323657474,
-      "loss": 7.237,
-      "step": 551
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.000934687953555878,
-      "loss": 7.0571,
-      "step": 552
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 0.0009332365747460088,
-      "loss": 7.1393,
-      "step": 553
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009317851959361394,
-      "loss": 7.0599,
-      "step": 554
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.00093033381712627,
-      "loss": 7.169,
-      "step": 555
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009288824383164007,
-      "loss": 7.0099,
-      "step": 556
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009274310595065311,
-      "loss": 7.0857,
-      "step": 557
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009259796806966619,
-      "loss": 6.9762,
-      "step": 558
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009245283018867925,
-      "loss": 7.2202,
-      "step": 559
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009230769230769232,
-      "loss": 6.9869,
-      "step": 560
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009216255442670537,
-      "loss": 7.2204,
-      "step": 561
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009201741654571844,
-      "loss": 6.9178,
-      "step": 562
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.000918722786647315,
-      "loss": 7.1024,
-      "step": 563
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009172714078374456,
-      "loss": 6.7413,
-      "step": 564
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0009158200290275763,
-      "loss": 6.9166,
-      "step": 565
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009143686502177069,
-      "loss": 7.025,
-      "step": 566
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009129172714078375,
-      "loss": 6.9701,
-      "step": 567
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009114658925979681,
-      "loss": 6.9994,
-      "step": 568
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009100145137880988,
-      "loss": 7.0043,
-      "step": 569
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009085631349782293,
-      "loss": 7.0049,
-      "step": 570
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.00090711175616836,
-      "loss": 7.0456,
-      "step": 571
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009056603773584906,
-      "loss": 7.0374,
-      "step": 572
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009042089985486213,
-      "loss": 6.9867,
-      "step": 573
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009027576197387518,
-      "loss": 7.0539,
-      "step": 574
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0009013062409288825,
-      "loss": 7.231,
-      "step": 575
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.0008998548621190132,
-      "loss": 7.1384,
-      "step": 576
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008984034833091437,
-      "loss": 7.1052,
-      "step": 577
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008969521044992744,
-      "loss": 7.0373,
-      "step": 578
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000895500725689405,
-      "loss": 6.9778,
-      "step": 579
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008940493468795356,
-      "loss": 7.1991,
-      "step": 580
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008925979680696662,
-      "loss": 7.1533,
-      "step": 581
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008911465892597969,
-      "loss": 6.951,
-      "step": 582
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008896952104499274,
-      "loss": 7.0072,
-      "step": 583
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008882438316400581,
-      "loss": 6.914,
-      "step": 584
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008867924528301887,
-      "loss": 7.1663,
-      "step": 585
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008853410740203193,
-      "loss": 7.01,
-      "step": 586
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00088388969521045,
-      "loss": 7.1636,
-      "step": 587
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008824383164005806,
-      "loss": 6.9596,
-      "step": 588
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008809869375907113,
-      "loss": 6.929,
-      "step": 589
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008795355587808418,
-      "loss": 7.0193,
-      "step": 590
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008780841799709725,
-      "loss": 7.077,
-      "step": 591
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008766328011611031,
-      "loss": 6.9883,
-      "step": 592
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008751814223512337,
-      "loss": 7.0503,
-      "step": 593
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008737300435413643,
-      "loss": 6.8605,
-      "step": 594
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000872278664731495,
-      "loss": 6.769,
-      "step": 595
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008708272859216255,
-      "loss": 6.8794,
-      "step": 596
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008693759071117562,
-      "loss": 6.7565,
-      "step": 597
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008679245283018869,
-      "loss": 6.8621,
-      "step": 598
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008664731494920174,
-      "loss": 6.6671,
-      "step": 599
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008650217706821481,
-      "loss": 6.5365,
-      "step": 600
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008635703918722787,
-      "loss": 7.0766,
-      "step": 601
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008621190130624093,
-      "loss": 7.2011,
-      "step": 602
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008606676342525399,
-      "loss": 7.2054,
-      "step": 603
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008592162554426706,
-      "loss": 7.0521,
-      "step": 604
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008577648766328012,
-      "loss": 7.1367,
-      "step": 605
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008563134978229318,
-      "loss": 7.21,
-      "step": 606
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008548621190130624,
-      "loss": 6.9841,
-      "step": 607
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008534107402031931,
-      "loss": 7.1412,
-      "step": 608
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008519593613933237,
-      "loss": 7.0282,
-      "step": 609
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008505079825834543,
-      "loss": 7.0733,
-      "step": 610
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000849056603773585,
-      "loss": 7.1754,
-      "step": 611
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0008476052249637155,
-      "loss": 6.9809,
-      "step": 612
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008461538461538462,
-      "loss": 6.9781,
-      "step": 613
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008447024673439768,
-      "loss": 6.9147,
-      "step": 614
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008432510885341074,
-      "loss": 7.0614,
-      "step": 615
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000841799709724238,
-      "loss": 6.9196,
-      "step": 616
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008403483309143687,
-      "loss": 6.9836,
-      "step": 617
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008388969521044993,
-      "loss": 6.94,
-      "step": 618
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008374455732946299,
-      "loss": 7.0487,
-      "step": 619
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008359941944847606,
-      "loss": 7.1344,
-      "step": 620
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008345428156748912,
-      "loss": 6.8899,
-      "step": 621
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008330914368650218,
-      "loss": 6.9095,
-      "step": 622
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008316400580551524,
-      "loss": 6.9635,
-      "step": 623
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0008301886792452831,
-      "loss": 7.0661,
-      "step": 624
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008287373004354136,
-      "loss": 7.1297,
-      "step": 625
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008272859216255443,
-      "loss": 7.1099,
-      "step": 626
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008258345428156749,
-      "loss": 7.1122,
-      "step": 627
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008243831640058055,
-      "loss": 6.9014,
-      "step": 628
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008229317851959361,
-      "loss": 6.8948,
-      "step": 629
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008214804063860668,
-      "loss": 6.9861,
-      "step": 630
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008200290275761973,
-      "loss": 7.0249,
-      "step": 631
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000818577648766328,
-      "loss": 7.05,
-      "step": 632
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008171262699564587,
-      "loss": 6.9883,
-      "step": 633
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008156748911465893,
-      "loss": 7.1293,
-      "step": 634
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008142235123367199,
-      "loss": 7.0649,
-      "step": 635
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0008127721335268505,
-      "loss": 7.0533,
-      "step": 636
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008113207547169812,
-      "loss": 6.9667,
-      "step": 637
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008098693759071117,
-      "loss": 6.9186,
-      "step": 638
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008084179970972424,
-      "loss": 6.9798,
-      "step": 639
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000806966618287373,
-      "loss": 7.2642,
-      "step": 640
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008055152394775036,
-      "loss": 6.9569,
-      "step": 641
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008040638606676342,
-      "loss": 6.9103,
-      "step": 642
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008026124818577649,
-      "loss": 7.0987,
-      "step": 643
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0008011611030478955,
-      "loss": 7.0009,
-      "step": 644
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0007997097242380261,
-      "loss": 6.9996,
-      "step": 645
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0007982583454281568,
-      "loss": 6.8922,
-      "step": 646
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0007968069666182874,
-      "loss": 6.584,
-      "step": 647
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000795355587808418,
-      "loss": 7.1011,
-      "step": 648
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007939042089985486,
-      "loss": 6.9417,
-      "step": 649
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007924528301886793,
-      "loss": 6.5543,
-      "step": 650
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007910014513788098,
-      "loss": 7.2658,
-      "step": 651
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007895500725689405,
-      "loss": 7.0368,
-      "step": 652
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007880986937590711,
-      "loss": 7.1025,
-      "step": 653
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007866473149492017,
-      "loss": 7.1689,
-      "step": 654
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007851959361393324,
-      "loss": 7.1003,
-      "step": 655
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000783744557329463,
-      "loss": 7.0462,
-      "step": 656
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007822931785195936,
-      "loss": 6.9565,
-      "step": 657
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007808417997097242,
-      "loss": 7.1247,
-      "step": 658
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007793904208998549,
-      "loss": 6.9882,
-      "step": 659
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0007779390420899854,
-      "loss": 7.0008,
-      "step": 660
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007764876632801161,
-      "loss": 6.9634,
-      "step": 661
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007750362844702467,
-      "loss": 7.0588,
-      "step": 662
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007735849056603774,
-      "loss": 6.9029,
-      "step": 663
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007721335268505079,
-      "loss": 7.0907,
-      "step": 664
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007706821480406386,
-      "loss": 7.0847,
-      "step": 665
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007692307692307693,
-      "loss": 6.8848,
-      "step": 666
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007677793904208998,
-      "loss": 7.0483,
-      "step": 667
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007663280116110305,
-      "loss": 6.979,
-      "step": 668
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007648766328011611,
-      "loss": 7.2513,
-      "step": 669
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007634252539912917,
-      "loss": 7.1019,
-      "step": 670
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0007619738751814223,
-      "loss": 7.0031,
-      "step": 671
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000760522496371553,
-      "loss": 7.0034,
-      "step": 672
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007590711175616835,
-      "loss": 6.9702,
-      "step": 673
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007576197387518142,
-      "loss": 6.8878,
-      "step": 674
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007561683599419448,
-      "loss": 7.1037,
-      "step": 675
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007547169811320755,
-      "loss": 7.0139,
-      "step": 676
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007532656023222061,
-      "loss": 7.1821,
-      "step": 677
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007518142235123367,
-      "loss": 7.1849,
-      "step": 678
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007503628447024674,
-      "loss": 7.0322,
-      "step": 679
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007489114658925979,
-      "loss": 7.1286,
-      "step": 680
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007474600870827286,
-      "loss": 6.9087,
-      "step": 681
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007460087082728592,
-      "loss": 6.9388,
-      "step": 682
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0007445573294629898,
-      "loss": 6.9481,
-      "step": 683
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007431059506531204,
-      "loss": 6.9418,
-      "step": 684
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007416545718432511,
-      "loss": 6.9699,
-      "step": 685
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007402031930333816,
-      "loss": 6.9794,
-      "step": 686
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007387518142235123,
-      "loss": 7.0891,
-      "step": 687
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.000737300435413643,
-      "loss": 7.0207,
-      "step": 688
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007358490566037735,
-      "loss": 7.1929,
-      "step": 689
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007343976777939043,
-      "loss": 6.9594,
-      "step": 690
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007329462989840348,
-      "loss": 7.2063,
-      "step": 691
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007314949201741656,
-      "loss": 6.9044,
-      "step": 692
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.000730043541364296,
-      "loss": 6.9193,
-      "step": 693
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007285921625544268,
-      "loss": 7.0464,
-      "step": 694
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0007271407837445574,
-      "loss": 6.964,
-      "step": 695
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.000725689404934688,
-      "loss": 6.6928,
-      "step": 696
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007242380261248185,
-      "loss": 7.1109,
-      "step": 697
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007227866473149493,
-      "loss": 7.0787,
-      "step": 698
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007213352685050797,
-      "loss": 6.5761,
-      "step": 699
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007198838896952105,
-      "loss": 6.9642,
-      "step": 700
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007184325108853412,
-      "loss": 7.2433,
-      "step": 701
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007169811320754717,
-      "loss": 6.9795,
-      "step": 702
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007155297532656024,
-      "loss": 7.1225,
-      "step": 703
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.000714078374455733,
-      "loss": 7.0223,
-      "step": 704
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007126269956458637,
-      "loss": 7.0501,
-      "step": 705
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007111756168359942,
-      "loss": 7.1019,
-      "step": 706
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0007097242380261249,
-      "loss": 7.056,
-      "step": 707
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0007082728592162555,
-      "loss": 6.9328,
-      "step": 708
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0007068214804063861,
-      "loss": 7.0541,
-      "step": 709
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0007053701015965167,
-      "loss": 7.0901,
-      "step": 710
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0007039187227866474,
-      "loss": 6.8945,
-      "step": 711
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.000702467343976778,
-      "loss": 7.2194,
-      "step": 712
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0007010159651669086,
-      "loss": 7.039,
-      "step": 713
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006995645863570393,
-      "loss": 7.051,
-      "step": 714
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006981132075471698,
-      "loss": 6.9684,
-      "step": 715
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006966618287373005,
-      "loss": 6.9407,
-      "step": 716
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006952104499274311,
-      "loss": 7.0311,
-      "step": 717
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006937590711175617,
-      "loss": 7.0686,
-      "step": 718
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0006923076923076923,
-      "loss": 7.026,
-      "step": 719
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.000690856313497823,
-      "loss": 6.9705,
-      "step": 720
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006894049346879536,
-      "loss": 6.8694,
-      "step": 721
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006879535558780842,
-      "loss": 6.8761,
-      "step": 722
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006865021770682149,
-      "loss": 6.946,
-      "step": 723
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006850507982583455,
-      "loss": 6.9311,
-      "step": 724
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006835994194484761,
-      "loss": 7.0254,
-      "step": 725
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006821480406386067,
-      "loss": 6.9334,
-      "step": 726
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006806966618287374,
-      "loss": 6.908,
-      "step": 727
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006792452830188679,
-      "loss": 7.0937,
-      "step": 728
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006777939042089986,
-      "loss": 6.972,
-      "step": 729
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006763425253991292,
-      "loss": 7.0818,
-      "step": 730
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0006748911465892598,
-      "loss": 6.9788,
-      "step": 731
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006734397677793904,
-      "loss": 7.0501,
-      "step": 732
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006719883889695211,
-      "loss": 7.0429,
-      "step": 733
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006705370101596517,
-      "loss": 6.9314,
-      "step": 734
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006690856313497823,
-      "loss": 7.05,
-      "step": 735
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000667634252539913,
-      "loss": 6.9731,
-      "step": 736
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006661828737300436,
-      "loss": 7.1809,
-      "step": 737
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006647314949201742,
-      "loss": 7.1111,
-      "step": 738
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006632801161103048,
-      "loss": 7.1377,
-      "step": 739
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006618287373004355,
-      "loss": 6.9592,
-      "step": 740
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000660377358490566,
-      "loss": 7.1619,
-      "step": 741
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006589259796806967,
-      "loss": 7.0776,
-      "step": 742
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0006574746008708273,
-      "loss": 7.1051,
-      "step": 743
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006560232220609579,
-      "loss": 6.9187,
-      "step": 744
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006545718432510886,
-      "loss": 6.9202,
-      "step": 745
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006531204644412192,
-      "loss": 6.7895,
-      "step": 746
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006516690856313498,
-      "loss": 6.958,
-      "step": 747
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006502177068214804,
-      "loss": 7.14,
-      "step": 748
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006487663280116111,
-      "loss": 6.7628,
-      "step": 749
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006473149492017417,
-      "loss": 6.4782,
-      "step": 750
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006458635703918723,
-      "loss": 7.0834,
-      "step": 751
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006444121915820029,
-      "loss": 7.1919,
-      "step": 752
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006429608127721336,
-      "loss": 7.0338,
-      "step": 753
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006415094339622641,
-      "loss": 6.9988,
-      "step": 754
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0006400580551523948,
-      "loss": 7.2172,
-      "step": 755
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006386066763425255,
-      "loss": 7.0916,
-      "step": 756
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.000637155297532656,
-      "loss": 7.0179,
-      "step": 757
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006357039187227867,
-      "loss": 7.1129,
-      "step": 758
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006342525399129173,
-      "loss": 6.8909,
-      "step": 759
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006328011611030479,
-      "loss": 7.0784,
-      "step": 760
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006313497822931785,
-      "loss": 7.0147,
-      "step": 761
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006298984034833092,
-      "loss": 6.9088,
-      "step": 762
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006284470246734397,
-      "loss": 6.913,
-      "step": 763
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006269956458635704,
-      "loss": 6.9156,
-      "step": 764
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.000625544267053701,
-      "loss": 7.2477,
-      "step": 765
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006240928882438317,
-      "loss": 7.0455,
-      "step": 766
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0006226415094339623,
-      "loss": 7.0375,
-      "step": 767
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006211901306240929,
-      "loss": 6.835,
-      "step": 768
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006197387518142236,
-      "loss": 6.9587,
-      "step": 769
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006182873730043541,
-      "loss": 7.0557,
-      "step": 770
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006168359941944848,
-      "loss": 7.0017,
-      "step": 771
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006153846153846154,
-      "loss": 7.1627,
-      "step": 772
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.000613933236574746,
-      "loss": 7.0343,
-      "step": 773
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006124818577648766,
-      "loss": 6.8341,
-      "step": 774
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006110304789550073,
-      "loss": 7.0343,
-      "step": 775
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006095791001451378,
-      "loss": 6.9326,
-      "step": 776
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006081277213352685,
-      "loss": 6.8837,
-      "step": 777
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006066763425253992,
-      "loss": 6.9924,
-      "step": 778
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0006052249637155298,
-      "loss": 7.0209,
-      "step": 779
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0006037735849056604,
-      "loss": 7.2092,
-      "step": 780
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.000602322206095791,
-      "loss": 7.0569,
-      "step": 781
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0006008708272859217,
-      "loss": 6.9737,
-      "step": 782
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005994194484760522,
-      "loss": 6.8599,
-      "step": 783
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005979680696661829,
-      "loss": 7.0745,
-      "step": 784
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005965166908563135,
-      "loss": 7.158,
-      "step": 785
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005950653120464441,
-      "loss": 6.8415,
-      "step": 786
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005936139332365747,
-      "loss": 7.1236,
-      "step": 787
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005921625544267054,
-      "loss": 6.9651,
-      "step": 788
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005907111756168359,
-      "loss": 6.9907,
-      "step": 789
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005892597968069666,
-      "loss": 6.9473,
-      "step": 790
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0005878084179970973,
-      "loss": 7.1449,
-      "step": 791
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005863570391872278,
-      "loss": 6.9635,
-      "step": 792
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005849056603773585,
-      "loss": 6.9782,
-      "step": 793
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005834542815674891,
-      "loss": 6.8888,
-      "step": 794
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005820029027576198,
-      "loss": 6.8093,
-      "step": 795
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005805515239477503,
-      "loss": 6.708,
-      "step": 796
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.000579100145137881,
-      "loss": 6.7831,
-      "step": 797
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005776487663280116,
-      "loss": 6.8203,
-      "step": 798
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005761973875181422,
-      "loss": 6.7705,
-      "step": 799
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005747460087082728,
-      "loss": 6.4364,
-      "step": 800
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005732946298984035,
-      "loss": 7.09,
-      "step": 801
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0005718432510885341,
-      "loss": 6.9635,
-      "step": 802
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005703918722786647,
-      "loss": 6.9889,
-      "step": 803
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005689404934687954,
-      "loss": 7.0255,
-      "step": 804
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005674891146589259,
-      "loss": 7.0305,
-      "step": 805
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005660377358490566,
-      "loss": 6.9456,
-      "step": 806
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005645863570391872,
-      "loss": 6.9681,
-      "step": 807
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005631349782293179,
-      "loss": 7.0481,
-      "step": 808
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005616835994194484,
-      "loss": 6.9558,
-      "step": 809
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005602322206095791,
-      "loss": 7.0219,
-      "step": 810
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005587808417997097,
-      "loss": 6.9424,
-      "step": 811
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005573294629898403,
-      "loss": 6.9673,
-      "step": 812
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.000555878084179971,
-      "loss": 6.8765,
-      "step": 813
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0005544267053701016,
-      "loss": 7.1636,
-      "step": 814
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005529753265602322,
-      "loss": 7.1162,
-      "step": 815
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005515239477503628,
-      "loss": 6.8999,
-      "step": 816
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005500725689404935,
-      "loss": 7.0049,
-      "step": 817
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.000548621190130624,
-      "loss": 6.8031,
-      "step": 818
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005471698113207547,
-      "loss": 7.0143,
-      "step": 819
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005457184325108853,
-      "loss": 7.0933,
-      "step": 820
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005442670537010159,
-      "loss": 6.9953,
-      "step": 821
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005428156748911465,
-      "loss": 6.9575,
-      "step": 822
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005413642960812772,
-      "loss": 7.0271,
-      "step": 823
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.000539912917271408,
-      "loss": 7.1252,
-      "step": 824
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005384615384615384,
-      "loss": 6.9604,
-      "step": 825
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0005370101596516691,
-      "loss": 6.9373,
-      "step": 826
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005355587808417997,
-      "loss": 7.0669,
-      "step": 827
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005341074020319303,
-      "loss": 7.0643,
-      "step": 828
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005326560232220609,
-      "loss": 7.0197,
-      "step": 829
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005312046444121916,
-      "loss": 7.1988,
-      "step": 830
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005297532656023221,
-      "loss": 7.0213,
-      "step": 831
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005283018867924528,
-      "loss": 6.7809,
-      "step": 832
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005268505079825834,
-      "loss": 6.8629,
-      "step": 833
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.000525399129172714,
-      "loss": 7.0814,
-      "step": 834
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005239477503628448,
-      "loss": 7.019,
-      "step": 835
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005224963715529754,
-      "loss": 7.1263,
-      "step": 836
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005210449927431061,
-      "loss": 7.0374,
-      "step": 837
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0005195936139332365,
-      "loss": 6.7339,
-      "step": 838
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005181422351233673,
-      "loss": 7.1797,
-      "step": 839
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005166908563134979,
-      "loss": 7.0079,
-      "step": 840
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005152394775036285,
-      "loss": 7.0764,
-      "step": 841
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.000513788098693759,
-      "loss": 6.9529,
-      "step": 842
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005123367198838898,
-      "loss": 7.1791,
-      "step": 843
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005108853410740202,
-      "loss": 6.8321,
-      "step": 844
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.000509433962264151,
-      "loss": 6.9381,
-      "step": 845
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005079825834542817,
-      "loss": 6.8807,
-      "step": 846
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005065312046444122,
-      "loss": 6.9491,
-      "step": 847
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005050798258345429,
-      "loss": 6.6411,
-      "step": 848
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005036284470246735,
-      "loss": 6.981,
-      "step": 849
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0005021770682148041,
-      "loss": 6.9311,
-      "step": 850
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0005007256894049347,
-      "loss": 7.2351,
-      "step": 851
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004992743105950654,
-      "loss": 7.0825,
-      "step": 852
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.000497822931785196,
-      "loss": 7.0775,
-      "step": 853
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004963715529753266,
-      "loss": 7.1195,
-      "step": 854
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004949201741654572,
-      "loss": 7.0512,
-      "step": 855
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004934687953555878,
-      "loss": 7.0764,
-      "step": 856
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004920174165457185,
-      "loss": 6.9077,
-      "step": 857
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004905660377358491,
-      "loss": 7.0929,
-      "step": 858
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004891146589259797,
-      "loss": 6.9194,
-      "step": 859
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004876632801161103,
-      "loss": 6.9869,
-      "step": 860
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004862119013062409,
-      "loss": 7.0333,
-      "step": 861
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0004847605224963715,
-      "loss": 7.0011,
-      "step": 862
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004833091436865022,
-      "loss": 6.908,
-      "step": 863
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004818577648766328,
-      "loss": 7.0157,
-      "step": 864
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00048040638606676347,
-      "loss": 7.031,
-      "step": 865
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00047895500725689407,
-      "loss": 7.0082,
-      "step": 866
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00047750362844702467,
-      "loss": 6.8794,
-      "step": 867
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004760522496371553,
-      "loss": 6.9123,
-      "step": 868
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004746008708272859,
-      "loss": 6.9084,
-      "step": 869
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004731494920174166,
-      "loss": 6.8418,
-      "step": 870
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004716981132075472,
-      "loss": 6.9268,
-      "step": 871
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.00047024673439767777,
-      "loss": 6.8352,
-      "step": 872
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0004687953555878084,
-      "loss": 7.1646,
-      "step": 873
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.000467343976777939,
-      "loss": 7.0545,
-      "step": 874
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004658925979680697,
-      "loss": 7.0751,
-      "step": 875
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00046444121915820033,
-      "loss": 6.9282,
-      "step": 876
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00046298984034833093,
-      "loss": 6.8993,
-      "step": 877
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004615384615384616,
-      "loss": 7.1428,
-      "step": 878
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004600870827285922,
-      "loss": 6.9436,
-      "step": 879
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004586357039187228,
-      "loss": 7.0862,
-      "step": 880
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00045718432510885343,
-      "loss": 6.9531,
-      "step": 881
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00045573294629898403,
-      "loss": 7.0204,
-      "step": 882
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.00045428156748911463,
-      "loss": 7.0929,
-      "step": 883
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004528301886792453,
-      "loss": 7.102,
-      "step": 884
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004513788098693759,
-      "loss": 6.8697,
-      "step": 885
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0004499274310595066,
-      "loss": 7.0026,
-      "step": 886
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0004484760522496372,
-      "loss": 7.031,
-      "step": 887
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0004470246734397678,
-      "loss": 6.9972,
-      "step": 888
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00044557329462989844,
-      "loss": 7.0602,
-      "step": 889
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00044412191582002904,
-      "loss": 6.9149,
-      "step": 890
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00044267053701015964,
-      "loss": 7.1231,
-      "step": 891
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0004412191582002903,
-      "loss": 7.0089,
-      "step": 892
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0004397677793904209,
-      "loss": 6.9217,
-      "step": 893
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00043831640058055154,
-      "loss": 6.8635,
-      "step": 894
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00043686502177068214,
-      "loss": 6.8953,
-      "step": 895
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00043541364296081274,
-      "loss": 6.8953,
-      "step": 896
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00043396226415094345,
-      "loss": 6.7635,
-      "step": 897
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.00043251088534107404,
-      "loss": 6.5849,
-      "step": 898
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00043105950653120464,
-      "loss": 6.7803,
-      "step": 899
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0004296081277213353,
-      "loss": 6.3055,
-      "step": 900
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0004281567489114659,
-      "loss": 7.1556,
-      "step": 901
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00042670537010159655,
-      "loss": 7.1128,
-      "step": 902
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00042525399129172715,
-      "loss": 7.0628,
-      "step": 903
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00042380261248185774,
-      "loss": 6.8395,
-      "step": 904
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0004223512336719884,
-      "loss": 7.0728,
-      "step": 905
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.000420899854862119,
-      "loss": 7.0456,
-      "step": 906
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.00041944847605224965,
-      "loss": 6.9815,
-      "step": 907
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0004179970972423803,
-      "loss": 7.3142,
-      "step": 908
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0004165457184325109,
-      "loss": 6.9824,
-      "step": 909
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00041509433962264155,
-      "loss": 7.0251,
-      "step": 910
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00041364296081277215,
-      "loss": 6.957,
-      "step": 911
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00041219158200290275,
-      "loss": 7.011,
-      "step": 912
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0004107402031930334,
-      "loss": 6.9846,
-      "step": 913
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.000409288824383164,
-      "loss": 6.9361,
-      "step": 914
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00040783744557329466,
-      "loss": 7.012,
-      "step": 915
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00040638606676342525,
-      "loss": 7.0764,
-      "step": 916
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00040493468795355585,
-      "loss": 7.0074,
-      "step": 917
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0004034833091436865,
-      "loss": 6.9982,
-      "step": 918
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0004020319303338171,
-      "loss": 7.0356,
-      "step": 919
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.00040058055152394776,
-      "loss": 7.024,
-      "step": 920
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0003991291727140784,
-      "loss": 6.9512,
-      "step": 921
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.000397677793904209,
-      "loss": 7.1374,
-      "step": 922
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00039622641509433966,
-      "loss": 7.0076,
-      "step": 923
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00039477503628447026,
-      "loss": 7.0693,
-      "step": 924
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00039332365747460086,
-      "loss": 6.8384,
-      "step": 925
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0003918722786647315,
-      "loss": 6.993,
-      "step": 926
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0003904208998548621,
-      "loss": 7.0704,
-      "step": 927
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0003889695210449927,
-      "loss": 7.0208,
-      "step": 928
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00038751814223512336,
-      "loss": 6.9332,
-      "step": 929
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00038606676342525396,
-      "loss": 6.9932,
-      "step": 930
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00038461538461538467,
-      "loss": 6.9625,
-      "step": 931
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00038316400580551527,
-      "loss": 7.0556,
-      "step": 932
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.00038171262699564587,
-      "loss": 7.1128,
-      "step": 933
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003802612481857765,
-      "loss": 6.9821,
-      "step": 934
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003788098693759071,
-      "loss": 6.9837,
-      "step": 935
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00037735849056603777,
-      "loss": 7.0444,
-      "step": 936
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00037590711175616837,
-      "loss": 6.9861,
-      "step": 937
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00037445573294629897,
-      "loss": 7.1477,
-      "step": 938
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003730043541364296,
-      "loss": 7.0205,
-      "step": 939
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003715529753265602,
-      "loss": 6.9423,
-      "step": 940
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003701015965166908,
-      "loss": 6.9051,
-      "step": 941
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003686502177068215,
-      "loss": 7.1341,
-      "step": 942
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003671988388969521,
-      "loss": 6.7609,
-      "step": 943
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003657474600870828,
-      "loss": 6.6581,
-      "step": 944
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0003642960812772134,
-      "loss": 7.0643,
-      "step": 945
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.000362844702467344,
-      "loss": 6.8123,
-      "step": 946
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00036139332365747463,
-      "loss": 6.6869,
-      "step": 947
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00035994194484760523,
-      "loss": 6.6426,
-      "step": 948
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0003584905660377358,
-      "loss": 6.6798,
-      "step": 949
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0003570391872278665,
-      "loss": 6.6849,
-      "step": 950
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0003555878084179971,
-      "loss": 7.1134,
-      "step": 951
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00035413642960812773,
-      "loss": 7.0725,
-      "step": 952
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00035268505079825833,
-      "loss": 6.9904,
-      "step": 953
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.000351233671988389,
-      "loss": 6.8351,
-      "step": 954
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00034978229317851964,
-      "loss": 6.963,
-      "step": 955
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00034833091436865023,
-      "loss": 7.1159,
-      "step": 956
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00034687953555878083,
-      "loss": 6.9889,
-      "step": 957
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003454281567489115,
-      "loss": 7.1882,
-      "step": 958
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003439767779390421,
-      "loss": 6.9888,
-      "step": 959
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00034252539912917274,
-      "loss": 6.9362,
-      "step": 960
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00034107402031930334,
-      "loss": 6.976,
-      "step": 961
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00033962264150943393,
-      "loss": 6.9001,
-      "step": 962
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003381712626995646,
-      "loss": 6.8253,
-      "step": 963
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003367198838896952,
-      "loss": 6.9968,
-      "step": 964
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00033526850507982584,
-      "loss": 7.0632,
-      "step": 965
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003338171262699565,
-      "loss": 6.9839,
-      "step": 966
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0003323657474600871,
-      "loss": 7.2008,
-      "step": 967
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00033091436865021774,
-      "loss": 6.9924,
-      "step": 968
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00032946298984034834,
-      "loss": 6.7617,
-      "step": 969
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00032801161103047894,
-      "loss": 6.9878,
-      "step": 970
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0003265602322206096,
-      "loss": 6.9036,
-      "step": 971
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0003251088534107402,
-      "loss": 6.8653,
-      "step": 972
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00032365747460087085,
-      "loss": 6.937,
-      "step": 973
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00032220609579100144,
-      "loss": 6.9683,
-      "step": 974
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00032075471698113204,
-      "loss": 7.0596,
-      "step": 975
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00031930333817126275,
-      "loss": 7.0847,
-      "step": 976
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00031785195936139335,
-      "loss": 6.9152,
-      "step": 977
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00031640058055152395,
-      "loss": 6.9945,
-      "step": 978
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0003149492017416546,
-      "loss": 6.9241,
-      "step": 979
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0003134978229317852,
-      "loss": 6.9992,
-      "step": 980
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00031204644412191585,
-      "loss": 7.2573,
-      "step": 981
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00031059506531204645,
-      "loss": 6.9963,
-      "step": 982
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00030914368650217705,
-      "loss": 7.021,
-      "step": 983
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003076923076923077,
-      "loss": 7.1762,
-      "step": 984
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003062409288824383,
-      "loss": 6.9364,
-      "step": 985
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003047895500725689,
-      "loss": 6.957,
-      "step": 986
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003033381712626996,
-      "loss": 7.0227,
-      "step": 987
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003018867924528302,
-      "loss": 6.9943,
-      "step": 988
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00030043541364296086,
-      "loss": 6.9493,
-      "step": 989
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00029898403483309146,
-      "loss": 6.9182,
-      "step": 990
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00029753265602322206,
-      "loss": 6.9994,
-      "step": 991
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0002960812772133527,
-      "loss": 6.9722,
-      "step": 992
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0002946298984034833,
-      "loss": 6.9006,
-      "step": 993
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002931785195936139,
-      "loss": 7.0978,
-      "step": 994
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00029172714078374456,
-      "loss": 7.015,
-      "step": 995
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00029027576197387516,
-      "loss": 7.2474,
-      "step": 996
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002888243831640058,
-      "loss": 6.9951,
-      "step": 997
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002873730043541364,
-      "loss": 6.7982,
-      "step": 998
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00028592162554426706,
-      "loss": 6.8156,
-      "step": 999
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002844702467343977,
-      "loss": 6.6049,
-      "step": 1000
-    },
-    {
-      "epoch": 0.84,
-      "eval_loss": 6.955667972564697,
-      "eval_runtime": 758.0162,
-      "eval_samples_per_second": 3.485,
-      "eval_steps_per_second": 0.292,
-      "eval_wer": 2.0097029585094353,
-      "step": 1000
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002830188679245283,
-      "loss": 6.9633,
-      "step": 1001
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00028156748911465897,
-      "loss": 7.0516,
-      "step": 1002
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00028011611030478957,
-      "loss": 7.019,
-      "step": 1003
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00027866473149492017,
-      "loss": 7.0578,
-      "step": 1004
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0002772133526850508,
-      "loss": 6.9281,
-      "step": 1005
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002757619738751814,
-      "loss": 7.0005,
-      "step": 1006
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.000274310595065312,
-      "loss": 6.9486,
-      "step": 1007
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00027285921625544267,
-      "loss": 6.8204,
-      "step": 1008
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00027140783744557327,
-      "loss": 7.0691,
-      "step": 1009
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.000269956458635704,
-      "loss": 6.8514,
-      "step": 1010
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002685050798258346,
-      "loss": 6.9034,
-      "step": 1011
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00026705370101596517,
-      "loss": 6.9737,
-      "step": 1012
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002656023222060958,
-      "loss": 6.8929,
-      "step": 1013
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002641509433962264,
-      "loss": 6.8821,
-      "step": 1014
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.000262699564586357,
-      "loss": 6.9074,
-      "step": 1015
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002612481857764877,
-      "loss": 7.0795,
-      "step": 1016
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0002597968069666183,
-      "loss": 6.8741,
-      "step": 1017
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002583454281567489,
-      "loss": 6.925,
-      "step": 1018
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002568940493468795,
-      "loss": 7.0645,
-      "step": 1019
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002554426705370101,
-      "loss": 6.8357,
-      "step": 1020
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00025399129172714083,
-      "loss": 7.0222,
-      "step": 1021
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00025253991291727143,
-      "loss": 7.099,
-      "step": 1022
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00025108853410740203,
-      "loss": 7.1473,
-      "step": 1023
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002496371552975327,
-      "loss": 6.9464,
-      "step": 1024
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002481857764876633,
-      "loss": 6.8569,
-      "step": 1025
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0002467343976777939,
-      "loss": 6.8892,
-      "step": 1026
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00024528301886792453,
-      "loss": 6.9002,
-      "step": 1027
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00024383164005805516,
-      "loss": 6.9339,
-      "step": 1028
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00024238026124818576,
-      "loss": 6.966,
-      "step": 1029
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002409288824383164,
-      "loss": 6.9522,
-      "step": 1030
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00023947750362844704,
-      "loss": 7.0329,
-      "step": 1031
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00023802612481857766,
-      "loss": 6.8151,
-      "step": 1032
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002365747460087083,
-      "loss": 7.0338,
-      "step": 1033
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00023512336719883889,
-      "loss": 6.9186,
-      "step": 1034
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002336719883889695,
-      "loss": 6.9835,
-      "step": 1035
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00023222060957910016,
-      "loss": 7.1005,
-      "step": 1036
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002307692307692308,
-      "loss": 6.8181,
-      "step": 1037
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002293178519593614,
-      "loss": 7.0983,
-      "step": 1038
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00022786647314949202,
-      "loss": 6.9616,
-      "step": 1039
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00022641509433962264,
-      "loss": 6.8104,
-      "step": 1040
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002249637155297533,
-      "loss": 6.9079,
-      "step": 1041
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002235123367198839,
-      "loss": 6.8955,
-      "step": 1042
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00022206095791001452,
-      "loss": 6.9046,
-      "step": 1043
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00022060957910014514,
-      "loss": 6.9043,
-      "step": 1044
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021915820029027577,
-      "loss": 6.9388,
-      "step": 1045
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021770682148040637,
-      "loss": 6.9023,
-      "step": 1046
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021625544267053702,
-      "loss": 6.7615,
-      "step": 1047
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021480406386066765,
-      "loss": 7.0216,
-      "step": 1048
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021335268505079827,
-      "loss": 7.0565,
-      "step": 1049
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00021190130624092887,
-      "loss": 6.6157,
-      "step": 1050
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002104499274310595,
-      "loss": 7.0615,
-      "step": 1051
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00020899854862119015,
-      "loss": 6.931,
-      "step": 1052
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00020754716981132078,
-      "loss": 6.9521,
-      "step": 1053
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00020609579100145138,
-      "loss": 7.0459,
-      "step": 1054
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.000204644412191582,
-      "loss": 7.0397,
-      "step": 1055
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00020319303338171263,
-      "loss": 7.0696,
-      "step": 1056
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00020174165457184325,
-      "loss": 6.9833,
-      "step": 1057
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00020029027576197388,
-      "loss": 6.9062,
-      "step": 1058
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0001988388969521045,
-      "loss": 7.0576,
-      "step": 1059
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00019738751814223513,
-      "loss": 7.0124,
-      "step": 1060
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00019593613933236576,
-      "loss": 7.039,
-      "step": 1061
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00019448476052249636,
-      "loss": 6.9322,
-      "step": 1062
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00019303338171262698,
-      "loss": 7.0308,
-      "step": 1063
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00019158200290275763,
-      "loss": 6.8833,
-      "step": 1064
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00019013062409288826,
-      "loss": 6.9581,
-      "step": 1065
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00018867924528301889,
-      "loss": 6.9187,
-      "step": 1066
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00018722786647314948,
-      "loss": 6.9337,
-      "step": 1067
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0001857764876632801,
-      "loss": 6.9806,
-      "step": 1068
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00018432510885341076,
-      "loss": 7.1115,
-      "step": 1069
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0001828737300435414,
-      "loss": 7.0317,
-      "step": 1070
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.000181422351233672,
-      "loss": 6.8814,
-      "step": 1071
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00017997097242380261,
-      "loss": 7.0167,
-      "step": 1072
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00017851959361393324,
-      "loss": 6.7661,
-      "step": 1073
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00017706821480406387,
-      "loss": 6.9122,
-      "step": 1074
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0001756168359941945,
-      "loss": 6.6888,
-      "step": 1075
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00017416545718432512,
-      "loss": 7.0122,
-      "step": 1076
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00017271407837445574,
-      "loss": 7.1287,
-      "step": 1077
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00017126269956458637,
-      "loss": 7.0498,
-      "step": 1078
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00016981132075471697,
-      "loss": 6.8909,
-      "step": 1079
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.0001683599419448476,
-      "loss": 6.9292,
-      "step": 1080
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00016690856313497825,
-      "loss": 6.9645,
-      "step": 1081
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00016545718432510887,
-      "loss": 7.2057,
-      "step": 1082
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00016400580551523947,
-      "loss": 6.8021,
-      "step": 1083
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.0001625544267053701,
-      "loss": 6.8902,
-      "step": 1084
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00016110304789550072,
-      "loss": 6.936,
-      "step": 1085
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00015965166908563138,
-      "loss": 6.961,
-      "step": 1086
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00015820029027576197,
-      "loss": 6.933,
-      "step": 1087
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.0001567489114658926,
-      "loss": 6.9899,
-      "step": 1088
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00015529753265602323,
-      "loss": 6.9847,
-      "step": 1089
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00015384615384615385,
-      "loss": 6.8568,
-      "step": 1090
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00015239477503628445,
-      "loss": 6.9409,
-      "step": 1091
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.0001509433962264151,
-      "loss": 6.8721,
-      "step": 1092
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00014949201741654573,
-      "loss": 6.8431,
-      "step": 1093
-    },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00014804063860667635,
-      "loss": 6.9966,
-      "step": 1094
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00014658925979680695,
-      "loss": 6.812,
-      "step": 1095
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 545
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00014513788098693758,
-      "loss": 7.0126,
-      "step": 1096
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 546
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0001436865021770682,
-      "loss": 6.8241,
-      "step": 1097
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 547
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00014223512336719886,
-      "loss": 7.0255,
-      "step": 1098
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 548
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00014078374455732948,
-      "loss": 7.0514,
-      "step": 1099
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 549
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00013933236574746008,
-      "loss": 6.2709,
-      "step": 1100
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.0001378809869375907,
-      "loss": 6.9508,
-      "step": 1101
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00013642960812772133,
-      "loss": 7.0238,
-      "step": 1102
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.000134978229317852,
-      "loss": 7.0654,
-      "step": 1103
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00013352685050798259,
-      "loss": 6.9235,
-      "step": 1104
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.0001320754716981132,
-      "loss": 6.8528,
-      "step": 1105
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00013062409288824384,
-      "loss": 6.8695,
-      "step": 1106
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 550
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012917271407837446,
-      "loss": 7.1841,
-      "step": 1107
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 551
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012772133526850506,
-      "loss": 7.095,
-      "step": 1108
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 552
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012626995645863572,
-      "loss": 6.9615,
-      "step": 1109
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 553
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012481857764876634,
-      "loss": 6.9125,
-      "step": 1110
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 554
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012336719883889694,
-      "loss": 6.949,
-      "step": 1111
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 555
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012191582002902758,
-      "loss": 6.976,
-      "step": 1112
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.0001204644412191582,
-      "loss": 6.8741,
-      "step": 1113
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00011901306240928883,
-      "loss": 6.8775,
-      "step": 1114
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00011756168359941944,
-      "loss": 7.0588,
-      "step": 1115
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00011611030478955008,
-      "loss": 7.0187,
-      "step": 1116
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.0001146589259796807,
-      "loss": 7.0744,
-      "step": 1117
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00011320754716981132,
-      "loss": 7.0031,
-      "step": 1118
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 556
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00011175616835994195,
-      "loss": 6.7731,
-      "step": 1119
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 557
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00011030478955007257,
-      "loss": 7.0852,
-      "step": 1120
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 558
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00010885341074020318,
-      "loss": 6.9801,
-      "step": 1121
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 559
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00010740203193033382,
-      "loss": 7.1414,
-      "step": 1122
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 560
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00010595065312046444,
-      "loss": 6.9665,
-      "step": 1123
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 561
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00010449927431059508,
-      "loss": 7.0696,
-      "step": 1124
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00010304789550072569,
-      "loss": 7.1454,
-      "step": 1125
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00010159651669085631,
-      "loss": 6.9548,
-      "step": 1126
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00010014513788098694,
-      "loss": 7.0,
-      "step": 1127
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.869375907111757e-05,
-      "loss": 6.9365,
-      "step": 1128
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.724238026124818e-05,
-      "loss": 7.0163,
-      "step": 1129
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.579100145137882e-05,
-      "loss": 6.9703,
-      "step": 1130
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 562
     },
     {
       "epoch": 0.95,
-      "learning_rate": 9.433962264150944e-05,
-      "loss": 6.8396,
-      "step": 1131
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 563
     },
     {
       "epoch": 0.95,
-      "learning_rate": 9.288824383164006e-05,
-      "loss": 6.8502,
-      "step": 1132
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 564
     },
     {
       "epoch": 0.95,
-      "learning_rate": 9.14368650217707e-05,
-      "loss": 7.0567,
-      "step": 1133
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 565
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.998548621190131e-05,
-      "loss": 6.9886,
-      "step": 1134
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 566
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.853410740203193e-05,
-      "loss": 7.099,
-      "step": 1135
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.708272859216256e-05,
-      "loss": 6.8939,
-      "step": 1136
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.563134978229318e-05,
-      "loss": 6.9841,
-      "step": 1137
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.41799709724238e-05,
-      "loss": 6.9891,
-      "step": 1138
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.272859216255444e-05,
-      "loss": 6.9023,
-      "step": 1139
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.127721335268505e-05,
-      "loss": 6.805,
-      "step": 1140
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.982583454281569e-05,
-      "loss": 7.0568,
-      "step": 1141
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 567
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.83744557329463e-05,
-      "loss": 6.983,
-      "step": 1142
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 568
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.692307692307693e-05,
-      "loss": 6.9293,
-      "step": 1143
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 569
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.547169811320755e-05,
-      "loss": 7.0366,
-      "step": 1144
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 570
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.402031930333818e-05,
-      "loss": 6.7063,
-      "step": 1145
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 571
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.256894049346879e-05,
-      "loss": 7.1163,
-      "step": 1146
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 572
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.111756168359943e-05,
-      "loss": 6.7976,
-      "step": 1147
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.966618287373004e-05,
-      "loss": 6.6924,
-      "step": 1148
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.821480406386067e-05,
-      "loss": 6.8346,
-      "step": 1149
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.676342525399129e-05,
-      "loss": 6.6107,
-      "step": 1150
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.531204644412192e-05,
-      "loss": 7.0302,
-      "step": 1151
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.386066763425253e-05,
-      "loss": 6.8901,
-      "step": 1152
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.240928882438317e-05,
-      "loss": 7.083,
-      "step": 1153
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 573
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.095791001451379e-05,
-      "loss": 6.977,
-      "step": 1154
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 574
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.9506531204644415e-05,
-      "loss": 6.9382,
-      "step": 1155
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 575
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.805515239477504e-05,
-      "loss": 7.1438,
-      "step": 1156
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 576
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.660377358490566e-05,
-      "loss": 6.9432,
-      "step": 1157
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 577
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.5152394775036286e-05,
-      "loss": 6.8917,
-      "step": 1158
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 578
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.370101596516691e-05,
-      "loss": 6.9139,
-      "step": 1159
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.224963715529754e-05,
-      "loss": 6.9881,
-      "step": 1160
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.079825834542816e-05,
-      "loss": 7.1474,
-      "step": 1161
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.934687953555878e-05,
-      "loss": 7.0312,
-      "step": 1162
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.789550072568941e-05,
-      "loss": 7.0814,
-      "step": 1163
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.644412191582003e-05,
-      "loss": 6.9351,
-      "step": 1164
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.4992743105950653e-05,
-      "loss": 6.9329,
-      "step": 1165
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 579
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.354136429608128e-05,
-      "loss": 7.139,
-      "step": 1166
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 580
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.20899854862119e-05,
-      "loss": 6.7964,
-      "step": 1167
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 581
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.0638606676342524e-05,
-      "loss": 6.9832,
-      "step": 1168
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 582
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.918722786647315e-05,
-      "loss": 6.8912,
-      "step": 1169
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 583
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.7735849056603776e-05,
-      "loss": 6.8612,
-      "step": 1170
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 584
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.6284470246734395e-05,
-      "loss": 6.9285,
-      "step": 1171
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.483309143686502e-05,
-      "loss": 6.8886,
-      "step": 1172
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3381712626995646e-05,
-      "loss": 7.0357,
-      "step": 1173
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.1930333817126266e-05,
-      "loss": 7.0646,
-      "step": 1174
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0478955007256895e-05,
-      "loss": 6.7593,
-      "step": 1175
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.902757619738752e-05,
-      "loss": 6.7634,
-      "step": 1176
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.7576197387518143e-05,
-      "loss": 6.9193,
-      "step": 1177
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 585
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.612481857764877e-05,
-      "loss": 7.066,
-      "step": 1178
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 586
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.467343976777939e-05,
-      "loss": 6.7822,
-      "step": 1179
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 587
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.3222060957910014e-05,
-      "loss": 7.1183,
-      "step": 1180
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 588
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.177068214804064e-05,
-      "loss": 6.8649,
-      "step": 1181
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 589
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.0319303338171262e-05,
-      "loss": 7.0857,
-      "step": 1182
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 590
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.8867924528301888e-05,
-      "loss": 6.8863,
-      "step": 1183
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.741654571843251e-05,
-      "loss": 6.8633,
-      "step": 1184
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5965166908563133e-05,
-      "loss": 6.8773,
-      "step": 1185
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.451378809869376e-05,
-      "loss": 6.8021,
-      "step": 1186
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 591
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.3062409288824384e-05,
-      "loss": 6.8164,
-      "step": 1187
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 592
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.1611030478955007e-05,
-      "loss": 6.646,
-      "step": 1188
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 593
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.0159651669085631e-05,
-      "loss": 6.5812,
-      "step": 1189
+      "learning_rate": 0.0009379999999999999,
+      "loss": 0.0,
+      "step": 594
     },
     {
       "epoch": 1.0,
-      "step": 1189,
+      "step": 594,
       "total_flos": 0.0,
-      "train_loss": 7.117300854900287,
-      "train_runtime": 6046.5433,
-      "train_samples_per_second": 4.72,
-      "train_steps_per_second": 0.197
+      "train_loss": 6.071190901476927,
+      "train_runtime": 5557.8497,
+      "train_samples_per_second": 5.135,
+      "train_steps_per_second": 0.107
     }
   ],
-  "max_steps": 1189,
+  "max_steps": 594,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,