diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2699 +1,7175 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2e-06,
-      "loss": 4.7448,
+      "learning_rate": 0.0,
+      "loss": 4.8639,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4e-06,
-      "loss": 4.8883,
+      "learning_rate": 0.0,
+      "loss": 5.0562,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6e-06,
-      "loss": 4.8436,
+      "epoch": 0.0,
+      "learning_rate": 2e-06,
+      "loss": 5.06,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8e-06,
-      "loss": 4.7855,
+      "epoch": 0.0,
+      "learning_rate": 4e-06,
+      "loss": 4.8812,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1e-05,
-      "loss": 4.6616,
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 4.5954,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2e-05,
-      "loss": 4.6613,
+      "learning_rate": 8e-06,
+      "loss": 4.9116,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.4e-05,
-      "loss": 4.5803,
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 4.6206,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.6e-05,
-      "loss": 4.5786,
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 4.5554,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.8e-05,
-      "loss": 4.5486,
+      "epoch": 0.01,
+      "learning_rate": 1.4e-05,
+      "loss": 4.6959,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2e-05,
-      "loss": 4.3926,
+      "epoch": 0.01,
+      "learning_rate": 1.6e-05,
+      "loss": 4.479,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2e-05,
-      "loss": 4.4643,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 4.5237,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.2e-05,
-      "loss": 4.4572,
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 4.3893,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.4e-05,
-      "loss": 4.373,
+      "epoch": 0.01,
+      "learning_rate": 2.2e-05,
+      "loss": 4.5701,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.6e-05,
-      "loss": 4.3855,
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 4.3773,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.8e-05,
-      "loss": 4.2546,
+      "epoch": 0.01,
+      "learning_rate": 2.6e-05,
+      "loss": 4.2756,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3e-05,
-      "loss": 4.395,
+      "epoch": 0.01,
+      "learning_rate": 2.8e-05,
+      "loss": 4.4677,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.2e-05,
-      "loss": 4.2986,
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 4.396,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.4178,
+      "epoch": 0.02,
+      "learning_rate": 3.2e-05,
+      "loss": 4.5728,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.6e-05,
-      "loss": 4.222,
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.2803,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.8e-05,
-      "loss": 4.3886,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-05,
+      "loss": 4.4825,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4e-05,
-      "loss": 4.359,
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 4.4547,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.2000000000000004e-05,
-      "loss": 4.3137,
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 4.3269,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.4e-05,
-      "loss": 4.3939,
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 4.4523,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.6e-05,
-      "loss": 4.354,
+      "epoch": 0.02,
+      "learning_rate": 4.4e-05,
+      "loss": 4.4519,
       "step": 24
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.8e-05,
-      "loss": 4.1825,
+      "epoch": 0.02,
+      "learning_rate": 4.6e-05,
+      "loss": 4.268,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5e-05,
-      "loss": 4.413,
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 4.4331,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.2e-05,
-      "loss": 4.2932,
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 4.3092,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.4e-05,
-      "loss": 4.155,
+      "epoch": 0.02,
+      "learning_rate": 5.2e-05,
+      "loss": 4.1246,
       "step": 28
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.6e-05,
-      "loss": 4.1171,
+      "epoch": 0.02,
+      "learning_rate": 5.4e-05,
+      "loss": 4.1848,
       "step": 29
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.800000000000001e-05,
-      "loss": 4.3741,
+      "epoch": 0.03,
+      "learning_rate": 5.6e-05,
+      "loss": 4.2948,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6e-05,
-      "loss": 4.3261,
+      "epoch": 0.03,
+      "learning_rate": 5.800000000000001e-05,
+      "loss": 4.4211,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.2e-05,
-      "loss": 4.1959,
+      "epoch": 0.03,
+      "learning_rate": 6e-05,
+      "loss": 4.2275,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.4e-05,
-      "loss": 4.2738,
+      "epoch": 0.03,
+      "learning_rate": 6.2e-05,
+      "loss": 4.3839,
       "step": 33
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.6e-05,
-      "loss": 4.2401,
+      "epoch": 0.03,
+      "learning_rate": 6.4e-05,
+      "loss": 4.0573,
       "step": 34
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 4.2822,
+      "epoch": 0.03,
+      "learning_rate": 6.6e-05,
+      "loss": 4.2317,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.000000000000001e-05,
-      "loss": 4.2956,
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.585,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.2e-05,
-      "loss": 4.3229,
+      "epoch": 0.03,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 4.3714,
       "step": 37
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.4e-05,
-      "loss": 4.2568,
+      "epoch": 0.03,
+      "learning_rate": 7.2e-05,
+      "loss": 4.2535,
       "step": 38
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.6e-05,
-      "loss": 4.373,
+      "epoch": 0.03,
+      "learning_rate": 7.4e-05,
+      "loss": 4.7745,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.8e-05,
-      "loss": 4.2255,
+      "epoch": 0.03,
+      "learning_rate": 7.6e-05,
+      "loss": 4.4277,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8e-05,
-      "loss": 4.2638,
+      "epoch": 0.03,
+      "learning_rate": 7.8e-05,
+      "loss": 4.1859,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.2e-05,
-      "loss": 4.3446,
+      "epoch": 0.04,
+      "learning_rate": 8e-05,
+      "loss": 4.2726,
       "step": 42
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.400000000000001e-05,
-      "loss": 4.1355,
+      "epoch": 0.04,
+      "learning_rate": 8.2e-05,
+      "loss": 4.2361,
       "step": 43
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.599999999999999e-05,
-      "loss": 4.3231,
+      "epoch": 0.04,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 4.2978,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.8e-05,
-      "loss": 4.4,
+      "epoch": 0.04,
+      "learning_rate": 8.599999999999999e-05,
+      "loss": 4.493,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 4.4553,
+      "epoch": 0.04,
+      "learning_rate": 8.8e-05,
+      "loss": 4.4094,
       "step": 46
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.2e-05,
-      "loss": 4.3854,
+      "epoch": 0.04,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.6894,
       "step": 47
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.400000000000001e-05,
-      "loss": 4.6031,
+      "epoch": 0.04,
+      "learning_rate": 9.2e-05,
+      "loss": 4.7486,
       "step": 48
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.6e-05,
-      "loss": 4.629,
+      "epoch": 0.04,
+      "learning_rate": 9.400000000000001e-05,
+      "loss": 4.9082,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.800000000000001e-05,
-      "loss": 4.3993,
+      "epoch": 0.04,
+      "learning_rate": 9.400000000000001e-05,
+      "loss": 4.7752,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.0001,
-      "loss": 4.3064,
+      "epoch": 0.04,
+      "learning_rate": 9.6e-05,
+      "loss": 4.3463,
       "step": 51
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000102,
-      "loss": 4.1792,
+      "epoch": 0.04,
+      "learning_rate": 9.800000000000001e-05,
+      "loss": 4.3752,
       "step": 52
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000104,
-      "loss": 4.338,
+      "epoch": 0.04,
+      "learning_rate": 0.0001,
+      "loss": 4.1995,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000106,
-      "loss": 4.2367,
+      "epoch": 0.05,
+      "learning_rate": 0.000102,
+      "loss": 4.2569,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000108,
-      "loss": 4.0805,
+      "epoch": 0.05,
+      "learning_rate": 0.000104,
+      "loss": 4.4036,
       "step": 55
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00011,
-      "loss": 4.2757,
+      "epoch": 0.05,
+      "learning_rate": 0.000106,
+      "loss": 4.1833,
       "step": 56
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000112,
-      "loss": 4.2968,
+      "epoch": 0.05,
+      "learning_rate": 0.000108,
+      "loss": 4.453,
       "step": 57
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000114,
-      "loss": 4.414,
+      "epoch": 0.05,
+      "learning_rate": 0.00011,
+      "loss": 4.395,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00011600000000000001,
-      "loss": 4.2925,
+      "epoch": 0.05,
+      "learning_rate": 0.000112,
+      "loss": 4.2395,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000118,
-      "loss": 4.1971,
+      "epoch": 0.05,
+      "learning_rate": 0.000114,
+      "loss": 4.4837,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00012,
-      "loss": 4.2551,
+      "epoch": 0.05,
+      "learning_rate": 0.00011600000000000001,
+      "loss": 4.2339,
       "step": 61
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000122,
-      "loss": 4.2037,
+      "epoch": 0.05,
+      "learning_rate": 0.000118,
+      "loss": 4.3999,
       "step": 62
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000124,
-      "loss": 4.1967,
+      "epoch": 0.05,
+      "learning_rate": 0.00012,
+      "loss": 4.2847,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000126,
-      "loss": 4.3095,
+      "epoch": 0.05,
+      "learning_rate": 0.000122,
+      "loss": 4.1678,
       "step": 64
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000128,
-      "loss": 4.2633,
+      "epoch": 0.05,
+      "learning_rate": 0.000124,
+      "loss": 4.1681,
       "step": 65
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00013000000000000002,
-      "loss": 4.144,
+      "epoch": 0.06,
+      "learning_rate": 0.000126,
+      "loss": 4.2683,
       "step": 66
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000132,
-      "loss": 4.2906,
+      "epoch": 0.06,
+      "learning_rate": 0.000128,
+      "loss": 4.3899,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000134,
-      "loss": 4.1689,
+      "epoch": 0.06,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 4.2539,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00013600000000000003,
-      "loss": 4.2172,
+      "epoch": 0.06,
+      "learning_rate": 0.000132,
+      "loss": 4.2929,
       "step": 69
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00013800000000000002,
-      "loss": 4.1874,
+      "epoch": 0.06,
+      "learning_rate": 0.000134,
+      "loss": 4.3351,
       "step": 70
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00014000000000000001,
-      "loss": 4.3269,
+      "epoch": 0.06,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 4.526,
       "step": 71
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00014199999999999998,
-      "loss": 4.3836,
+      "epoch": 0.06,
+      "learning_rate": 0.00013800000000000002,
+      "loss": 4.4187,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.000144,
-      "loss": 4.1776,
+      "epoch": 0.06,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 4.29,
       "step": 73
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000146,
-      "loss": 4.2561,
+      "epoch": 0.06,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 4.4053,
       "step": 74
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000148,
-      "loss": 4.2782,
+      "epoch": 0.06,
+      "learning_rate": 0.000144,
+      "loss": 4.2215,
       "step": 75
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00015,
-      "loss": 4.2432,
+      "epoch": 0.06,
+      "learning_rate": 0.000146,
+      "loss": 4.5723,
       "step": 76
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000152,
-      "loss": 4.2544,
+      "epoch": 0.06,
+      "learning_rate": 0.000148,
+      "loss": 4.4546,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000154,
-      "loss": 4.2034,
+      "epoch": 0.07,
+      "learning_rate": 0.00015,
+      "loss": 4.2738,
       "step": 78
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000156,
-      "loss": 4.2487,
+      "epoch": 0.07,
+      "learning_rate": 0.000152,
+      "loss": 4.3245,
       "step": 79
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000158,
-      "loss": 4.3555,
+      "epoch": 0.07,
+      "learning_rate": 0.000154,
+      "loss": 4.4287,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00016,
-      "loss": 4.3519,
+      "epoch": 0.07,
+      "learning_rate": 0.000156,
+      "loss": 4.4625,
       "step": 81
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000162,
-      "loss": 4.2654,
+      "epoch": 0.07,
+      "learning_rate": 0.000158,
+      "loss": 4.382,
       "step": 82
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000164,
-      "loss": 4.3638,
+      "epoch": 0.07,
+      "learning_rate": 0.00016,
+      "loss": 4.3249,
       "step": 83
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00016600000000000002,
-      "loss": 4.3315,
+      "epoch": 0.07,
+      "learning_rate": 0.000162,
+      "loss": 4.456,
       "step": 84
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00016800000000000002,
-      "loss": 4.3381,
+      "epoch": 0.07,
+      "learning_rate": 0.000164,
+      "loss": 4.4607,
       "step": 85
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00017,
-      "loss": 4.2955,
+      "epoch": 0.07,
+      "learning_rate": 0.00016600000000000002,
+      "loss": 4.35,
       "step": 86
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00017199999999999998,
-      "loss": 4.3315,
+      "epoch": 0.07,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 4.2586,
       "step": 87
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000174,
-      "loss": 4.2362,
+      "epoch": 0.07,
+      "learning_rate": 0.00017,
+      "loss": 4.4717,
       "step": 88
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000176,
-      "loss": 4.4432,
+      "epoch": 0.07,
+      "learning_rate": 0.00017199999999999998,
+      "loss": 4.5826,
       "step": 89
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000178,
-      "loss": 4.4031,
+      "epoch": 0.08,
+      "learning_rate": 0.000174,
+      "loss": 4.4634,
       "step": 90
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 4.2638,
+      "epoch": 0.08,
+      "learning_rate": 0.000176,
+      "loss": 4.6529,
       "step": 91
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000182,
-      "loss": 4.5215,
+      "epoch": 0.08,
+      "learning_rate": 0.000178,
+      "loss": 4.5446,
       "step": 92
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000184,
-      "loss": 4.4342,
+      "epoch": 0.08,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.2002,
       "step": 93
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000186,
-      "loss": 4.6104,
+      "epoch": 0.08,
+      "learning_rate": 0.000182,
+      "loss": 4.5535,
       "step": 94
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00018800000000000002,
-      "loss": 4.321,
+      "epoch": 0.08,
+      "learning_rate": 0.000184,
+      "loss": 4.7888,
       "step": 95
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00019,
-      "loss": 4.6963,
+      "epoch": 0.08,
+      "learning_rate": 0.000186,
+      "loss": 4.6889,
       "step": 96
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000192,
-      "loss": 4.4766,
+      "epoch": 0.08,
+      "learning_rate": 0.00018800000000000002,
+      "loss": 4.6501,
       "step": 97
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000194,
-      "loss": 4.6106,
+      "epoch": 0.08,
+      "learning_rate": 0.00019,
+      "loss": 4.7487,
       "step": 98
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00019600000000000002,
-      "loss": 4.7505,
+      "epoch": 0.08,
+      "learning_rate": 0.000192,
+      "loss": 4.7805,
       "step": 99
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00019800000000000002,
-      "loss": 4.5221,
+      "epoch": 0.08,
+      "learning_rate": 0.000194,
+      "loss": 4.6956,
       "step": 100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0002,
-      "loss": 4.4953,
+      "epoch": 0.08,
+      "learning_rate": 0.00019600000000000002,
+      "loss": 4.3338,
       "step": 101
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000202,
-      "loss": 4.5015,
+      "epoch": 0.09,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 4.719,
       "step": 102
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000204,
-      "loss": 4.2782,
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 4.4967,
       "step": 103
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000206,
-      "loss": 4.3432,
+      "epoch": 0.09,
+      "learning_rate": 0.000202,
+      "loss": 4.7219,
       "step": 104
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000208,
-      "loss": 4.2491,
+      "epoch": 0.09,
+      "learning_rate": 0.000204,
+      "loss": 4.7372,
       "step": 105
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.00021,
-      "loss": 4.343,
+      "epoch": 0.09,
+      "learning_rate": 0.000206,
+      "loss": 4.6261,
       "step": 106
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000212,
-      "loss": 4.3491,
+      "epoch": 0.09,
+      "learning_rate": 0.000208,
+      "loss": 4.7148,
       "step": 107
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000214,
-      "loss": 4.2156,
+      "epoch": 0.09,
+      "learning_rate": 0.00021,
+      "loss": 4.6218,
       "step": 108
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000216,
-      "loss": 4.2983,
+      "epoch": 0.09,
+      "learning_rate": 0.000212,
+      "loss": 4.7761,
       "step": 109
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000218,
-      "loss": 4.315,
+      "epoch": 0.09,
+      "learning_rate": 0.000214,
+      "loss": 4.4707,
       "step": 110
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00022,
-      "loss": 4.4117,
+      "epoch": 0.09,
+      "learning_rate": 0.000216,
+      "loss": 4.6993,
       "step": 111
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000222,
-      "loss": 4.3823,
+      "epoch": 0.09,
+      "learning_rate": 0.000218,
+      "loss": 4.8377,
       "step": 112
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000224,
-      "loss": 4.3341,
+      "epoch": 0.09,
+      "learning_rate": 0.00022,
+      "loss": 4.4853,
       "step": 113
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00022600000000000002,
-      "loss": 4.2731,
+      "epoch": 0.1,
+      "learning_rate": 0.000222,
+      "loss": 4.6009,
       "step": 114
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000228,
-      "loss": 4.3656,
+      "epoch": 0.1,
+      "learning_rate": 0.000224,
+      "loss": 4.6293,
       "step": 115
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00023,
-      "loss": 4.4402,
+      "epoch": 0.1,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 4.8498,
       "step": 116
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00023200000000000003,
-      "loss": 4.3582,
+      "epoch": 0.1,
+      "learning_rate": 0.000228,
+      "loss": 4.5136,
       "step": 117
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00023400000000000002,
-      "loss": 4.4894,
+      "epoch": 0.1,
+      "learning_rate": 0.00023,
+      "loss": 4.3641,
       "step": 118
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000236,
-      "loss": 4.4289,
+      "epoch": 0.1,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 4.4533,
       "step": 119
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00023799999999999998,
-      "loss": 4.4637,
+      "epoch": 0.1,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 4.4244,
       "step": 120
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00024,
-      "loss": 4.2816,
+      "epoch": 0.1,
+      "learning_rate": 0.000236,
+      "loss": 4.934,
       "step": 121
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000242,
-      "loss": 4.3416,
+      "epoch": 0.1,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 4.5533,
       "step": 122
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000244,
-      "loss": 4.3645,
+      "epoch": 0.1,
+      "learning_rate": 0.00024,
+      "loss": 4.5435,
       "step": 123
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000246,
-      "loss": 4.3001,
+      "epoch": 0.1,
+      "learning_rate": 0.000242,
+      "loss": 4.6954,
       "step": 124
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000248,
-      "loss": 4.4611,
+      "epoch": 0.11,
+      "learning_rate": 0.000244,
+      "loss": 4.4366,
       "step": 125
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.00025,
-      "loss": 4.5181,
+      "epoch": 0.11,
+      "learning_rate": 0.000246,
+      "loss": 4.5332,
       "step": 126
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000252,
-      "loss": 4.4458,
+      "epoch": 0.11,
+      "learning_rate": 0.000248,
+      "loss": 4.673,
       "step": 127
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000254,
-      "loss": 4.3131,
+      "epoch": 0.11,
+      "learning_rate": 0.00025,
+      "loss": 4.4091,
       "step": 128
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000256,
-      "loss": 4.27,
+      "epoch": 0.11,
+      "learning_rate": 0.000252,
+      "loss": 4.3541,
       "step": 129
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00025800000000000004,
-      "loss": 4.6221,
+      "epoch": 0.11,
+      "learning_rate": 0.000254,
+      "loss": 4.658,
       "step": 130
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00026000000000000003,
-      "loss": 4.3875,
+      "epoch": 0.11,
+      "learning_rate": 0.000256,
+      "loss": 4.5905,
       "step": 131
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000262,
-      "loss": 4.406,
+      "epoch": 0.11,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 4.494,
       "step": 132
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000264,
-      "loss": 4.2984,
+      "epoch": 0.11,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 4.8793,
       "step": 133
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000266,
-      "loss": 4.4539,
+      "epoch": 0.11,
+      "learning_rate": 0.000262,
+      "loss": 4.5163,
       "step": 134
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000268,
-      "loss": 4.5082,
+      "epoch": 0.11,
+      "learning_rate": 0.000264,
+      "loss": 4.7546,
       "step": 135
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.00027,
-      "loss": 4.3125,
+      "epoch": 0.11,
+      "learning_rate": 0.000266,
+      "loss": 4.3953,
       "step": 136
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00027200000000000005,
-      "loss": 4.4509,
+      "epoch": 0.12,
+      "learning_rate": 0.000268,
+      "loss": 4.6186,
       "step": 137
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00027400000000000005,
-      "loss": 4.5328,
+      "epoch": 0.12,
+      "learning_rate": 0.00027,
+      "loss": 4.6296,
       "step": 138
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00027600000000000004,
-      "loss": 4.4671,
+      "epoch": 0.12,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 4.8838,
       "step": 139
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00027800000000000004,
-      "loss": 4.478,
+      "epoch": 0.12,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 4.8756,
       "step": 140
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00028000000000000003,
-      "loss": 4.5525,
+      "epoch": 0.12,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 4.7281,
       "step": 141
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 4.7511,
+      "epoch": 0.12,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 4.4655,
       "step": 142
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00028399999999999996,
-      "loss": 5.0134,
+      "epoch": 0.12,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 4.6666,
       "step": 143
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00028599999999999996,
-      "loss": 4.7465,
+      "epoch": 0.12,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 4.798,
       "step": 144
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000288,
-      "loss": 4.4685,
+      "epoch": 0.12,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 4.8074,
       "step": 145
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.00029,
-      "loss": 4.6673,
+      "epoch": 0.12,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 4.8072,
       "step": 146
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000292,
-      "loss": 5.0644,
+      "epoch": 0.12,
+      "learning_rate": 0.000288,
+      "loss": 4.7666,
       "step": 147
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000294,
-      "loss": 4.8156,
+      "epoch": 0.12,
+      "learning_rate": 0.00029,
+      "loss": 4.4023,
       "step": 148
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000296,
-      "loss": 4.6348,
+      "epoch": 0.13,
+      "learning_rate": 0.000292,
+      "loss": 5.6534,
       "step": 149
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000298,
-      "loss": 4.905,
+      "epoch": 0.13,
+      "learning_rate": 0.000294,
+      "loss": 4.8126,
       "step": 150
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0003,
-      "loss": 4.5051,
+      "epoch": 0.13,
+      "learning_rate": 0.000296,
+      "loss": 4.8123,
       "step": 151
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000302,
-      "loss": 4.4782,
+      "epoch": 0.13,
+      "learning_rate": 0.000298,
+      "loss": 4.9249,
       "step": 152
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000304,
-      "loss": 4.563,
+      "epoch": 0.13,
+      "learning_rate": 0.0003,
+      "loss": 4.6911,
       "step": 153
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000306,
-      "loss": 4.3117,
+      "epoch": 0.13,
+      "learning_rate": 0.000302,
+      "loss": 4.8323,
       "step": 154
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000308,
-      "loss": 4.4834,
+      "epoch": 0.13,
+      "learning_rate": 0.000304,
+      "loss": 4.625,
       "step": 155
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.00031,
-      "loss": 4.3141,
+      "epoch": 0.13,
+      "learning_rate": 0.000306,
+      "loss": 4.5068,
       "step": 156
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000312,
-      "loss": 4.4631,
+      "epoch": 0.13,
+      "learning_rate": 0.000308,
+      "loss": 4.6335,
       "step": 157
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000314,
-      "loss": 4.5245,
+      "epoch": 0.13,
+      "learning_rate": 0.00031,
+      "loss": 5.077,
       "step": 158
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000316,
-      "loss": 4.5088,
+      "epoch": 0.13,
+      "learning_rate": 0.000312,
+      "loss": 4.8097,
       "step": 159
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00031800000000000003,
-      "loss": 4.5542,
+      "epoch": 0.13,
+      "learning_rate": 0.000314,
+      "loss": 4.3888,
       "step": 160
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00032,
-      "loss": 4.5309,
+      "epoch": 0.14,
+      "learning_rate": 0.000316,
+      "loss": 4.9162,
       "step": 161
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000322,
-      "loss": 4.448,
+      "epoch": 0.14,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 4.9361,
       "step": 162
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000324,
-      "loss": 4.4301,
+      "epoch": 0.14,
+      "learning_rate": 0.00032,
+      "loss": 4.8107,
       "step": 163
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000326,
-      "loss": 4.5063,
+      "epoch": 0.14,
+      "learning_rate": 0.000322,
+      "loss": 4.6171,
       "step": 164
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000328,
-      "loss": 4.3681,
+      "epoch": 0.14,
+      "learning_rate": 0.000324,
+      "loss": 5.0743,
       "step": 165
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00033,
-      "loss": 4.5243,
+      "epoch": 0.14,
+      "learning_rate": 0.000326,
+      "loss": 4.3022,
       "step": 166
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00033200000000000005,
-      "loss": 4.3789,
+      "epoch": 0.14,
+      "learning_rate": 0.000328,
+      "loss": 4.6418,
       "step": 167
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00033400000000000004,
-      "loss": 4.457,
+      "epoch": 0.14,
+      "learning_rate": 0.00033,
+      "loss": 4.6373,
       "step": 168
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00033600000000000004,
-      "loss": 4.599,
+      "epoch": 0.14,
+      "learning_rate": 0.00033200000000000005,
+      "loss": 4.6817,
       "step": 169
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00033800000000000003,
-      "loss": 4.4211,
+      "epoch": 0.14,
+      "learning_rate": 0.00033400000000000004,
+      "loss": 4.5453,
       "step": 170
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00034,
-      "loss": 4.4048,
+      "epoch": 0.14,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 4.6601,
       "step": 171
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000342,
-      "loss": 4.5022,
+      "epoch": 0.14,
+      "learning_rate": 0.00033800000000000003,
+      "loss": 4.6025,
       "step": 172
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00034399999999999996,
-      "loss": 4.4207,
+      "epoch": 0.15,
+      "learning_rate": 0.00034,
+      "loss": 4.4849,
       "step": 173
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000346,
-      "loss": 4.6248,
+      "epoch": 0.15,
+      "learning_rate": 0.000342,
+      "loss": 5.4156,
       "step": 174
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000348,
-      "loss": 4.6288,
+      "epoch": 0.15,
+      "learning_rate": 0.00034399999999999996,
+      "loss": 4.8372,
       "step": 175
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00035,
-      "loss": 4.3465,
+      "epoch": 0.15,
+      "learning_rate": 0.000346,
+      "loss": 5.8255,
       "step": 176
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000352,
-      "loss": 4.2846,
+      "epoch": 0.15,
+      "learning_rate": 0.000348,
+      "loss": 4.9039,
       "step": 177
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000354,
-      "loss": 4.5639,
+      "epoch": 0.15,
+      "learning_rate": 0.00035,
+      "loss": 4.856,
       "step": 178
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000356,
-      "loss": 4.5932,
+      "epoch": 0.15,
+      "learning_rate": 0.000352,
+      "loss": 4.6854,
       "step": 179
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000358,
-      "loss": 4.6008,
+      "epoch": 0.15,
+      "learning_rate": 0.000354,
+      "loss": 4.9355,
       "step": 180
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.00035999999999999997,
-      "loss": 4.5307,
+      "epoch": 0.15,
+      "learning_rate": 0.000356,
+      "loss": 4.824,
       "step": 181
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000362,
-      "loss": 4.5309,
+      "epoch": 0.15,
+      "learning_rate": 0.000358,
+      "loss": 5.1254,
       "step": 182
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000364,
-      "loss": 5.0206,
+      "epoch": 0.15,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 4.6313,
       "step": 183
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000366,
-      "loss": 4.5031,
+      "epoch": 0.15,
+      "learning_rate": 0.000362,
+      "loss": 4.8208,
       "step": 184
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000368,
-      "loss": 4.5873,
+      "epoch": 0.16,
+      "learning_rate": 0.000364,
+      "loss": 4.387,
       "step": 185
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00037,
-      "loss": 4.4111,
+      "epoch": 0.16,
+      "learning_rate": 0.000366,
+      "loss": 4.797,
       "step": 186
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000372,
-      "loss": 4.419,
+      "epoch": 0.16,
+      "learning_rate": 0.000368,
+      "loss": 4.8683,
       "step": 187
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000374,
-      "loss": 4.4443,
+      "epoch": 0.16,
+      "learning_rate": 0.00037,
+      "loss": 4.745,
       "step": 188
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00037600000000000003,
-      "loss": 4.6531,
+      "epoch": 0.16,
+      "learning_rate": 0.000372,
+      "loss": 5.0061,
       "step": 189
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000378,
-      "loss": 4.6183,
+      "epoch": 0.16,
+      "learning_rate": 0.000374,
+      "loss": 4.8935,
       "step": 190
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.00038,
-      "loss": 4.8718,
+      "epoch": 0.16,
+      "learning_rate": 0.00037600000000000003,
+      "loss": 4.8115,
       "step": 191
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000382,
-      "loss": 4.7412,
+      "epoch": 0.16,
+      "learning_rate": 0.000378,
+      "loss": 5.1207,
       "step": 192
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000384,
-      "loss": 4.4731,
+      "epoch": 0.16,
+      "learning_rate": 0.00038,
+      "loss": 4.8936,
       "step": 193
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000386,
-      "loss": 4.8124,
+      "epoch": 0.16,
+      "learning_rate": 0.000382,
+      "loss": 4.7928,
       "step": 194
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000388,
-      "loss": 5.2314,
+      "epoch": 0.16,
+      "learning_rate": 0.000384,
+      "loss": 5.1036,
       "step": 195
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.00039000000000000005,
-      "loss": 4.739,
+      "epoch": 0.16,
+      "learning_rate": 0.000386,
+      "loss": 5.7075,
       "step": 196
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.00039200000000000004,
-      "loss": 5.8253,
+      "epoch": 0.17,
+      "learning_rate": 0.000388,
+      "loss": 4.8485,
       "step": 197
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.00039400000000000004,
-      "loss": 5.141,
+      "epoch": 0.17,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 5.0071,
       "step": 198
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.00039600000000000003,
-      "loss": 4.8066,
+      "epoch": 0.17,
+      "learning_rate": 0.00039200000000000004,
+      "loss": 5.2219,
       "step": 199
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000398,
-      "loss": 5.3003,
+      "epoch": 0.17,
+      "learning_rate": 0.00039400000000000004,
+      "loss": 4.7529,
       "step": 200
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0004,
-      "loss": 4.8144,
+      "epoch": 0.17,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 4.9263,
       "step": 201
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000402,
-      "loss": 4.759,
+      "epoch": 0.17,
+      "learning_rate": 0.000398,
+      "loss": 5.0138,
       "step": 202
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000404,
-      "loss": 4.9561,
+      "epoch": 0.17,
+      "learning_rate": 0.0004,
+      "loss": 5.1898,
       "step": 203
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.00040600000000000006,
-      "loss": 4.8247,
+      "epoch": 0.17,
+      "learning_rate": 0.000402,
+      "loss": 4.7526,
       "step": 204
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000408,
-      "loss": 4.6602,
+      "epoch": 0.17,
+      "learning_rate": 0.000404,
+      "loss": 5.1598,
       "step": 205
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.00041,
-      "loss": 4.7296,
+      "epoch": 0.17,
+      "learning_rate": 0.00040600000000000006,
+      "loss": 4.9076,
       "step": 206
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000412,
-      "loss": 4.4467,
+      "epoch": 0.17,
+      "learning_rate": 0.000408,
+      "loss": 4.7115,
       "step": 207
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000414,
-      "loss": 4.7099,
+      "epoch": 0.17,
+      "learning_rate": 0.00041,
+      "loss": 5.9574,
       "step": 208
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000416,
-      "loss": 4.6622,
+      "epoch": 0.18,
+      "learning_rate": 0.000412,
+      "loss": 5.0494,
       "step": 209
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.00041799999999999997,
-      "loss": 4.9508,
+      "epoch": 0.18,
+      "learning_rate": 0.000414,
+      "loss": 5.4179,
       "step": 210
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.00042,
-      "loss": 4.4373,
+      "epoch": 0.18,
+      "learning_rate": 0.000416,
+      "loss": 4.7074,
       "step": 211
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.000422,
-      "loss": 4.748,
+      "epoch": 0.18,
+      "learning_rate": 0.00041799999999999997,
+      "loss": 5.0304,
       "step": 212
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.000424,
-      "loss": 4.623,
+      "epoch": 0.18,
+      "learning_rate": 0.00042,
+      "loss": 4.7222,
       "step": 213
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.000426,
-      "loss": 4.5733,
+      "epoch": 0.18,
+      "learning_rate": 0.000422,
+      "loss": 5.3001,
       "step": 214
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.000428,
-      "loss": 4.6245,
+      "epoch": 0.18,
+      "learning_rate": 0.000424,
+      "loss": 4.7676,
       "step": 215
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.00043,
-      "loss": 4.6892,
+      "epoch": 0.18,
+      "learning_rate": 0.000426,
+      "loss": 4.7473,
       "step": 216
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.000432,
-      "loss": 4.6396,
+      "epoch": 0.18,
+      "learning_rate": 0.000428,
+      "loss": 5.0886,
       "step": 217
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.00043400000000000003,
-      "loss": 4.6174,
+      "epoch": 0.18,
+      "learning_rate": 0.00043,
+      "loss": 4.8381,
       "step": 218
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.000436,
-      "loss": 4.7913,
+      "epoch": 0.18,
+      "learning_rate": 0.000432,
+      "loss": 4.5595,
       "step": 219
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.000438,
-      "loss": 4.6375,
+      "epoch": 0.18,
+      "learning_rate": 0.00043400000000000003,
+      "loss": 5.412,
       "step": 220
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.00044,
-      "loss": 4.5069,
+      "epoch": 0.19,
+      "learning_rate": 0.000436,
+      "loss": 4.8637,
       "step": 221
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.000442,
-      "loss": 4.7579,
+      "epoch": 0.19,
+      "learning_rate": 0.000438,
+      "loss": 5.3239,
       "step": 222
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.000444,
-      "loss": 4.7951,
+      "epoch": 0.19,
+      "learning_rate": 0.00044,
+      "loss": 5.1594,
       "step": 223
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.000446,
-      "loss": 4.9794,
+      "epoch": 0.19,
+      "learning_rate": 0.000442,
+      "loss": 4.8977,
       "step": 224
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.000448,
-      "loss": 4.8345,
+      "epoch": 0.19,
+      "learning_rate": 0.000444,
+      "loss": 5.0385,
       "step": 225
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.00045000000000000004,
-      "loss": 4.6178,
+      "epoch": 0.19,
+      "learning_rate": 0.000446,
+      "loss": 4.7914,
       "step": 226
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.00045200000000000004,
-      "loss": 4.9773,
+      "epoch": 0.19,
+      "learning_rate": 0.000448,
+      "loss": 5.0119,
       "step": 227
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.00045400000000000003,
-      "loss": 4.6117,
+      "epoch": 0.19,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 5.3607,
       "step": 228
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.000456,
-      "loss": 4.5106,
+      "epoch": 0.19,
+      "learning_rate": 0.00045200000000000004,
+      "loss": 5.1186,
       "step": 229
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.000458,
-      "loss": 4.5805,
+      "epoch": 0.19,
+      "learning_rate": 0.00045400000000000003,
+      "loss": 4.9873,
       "step": 230
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.00046,
-      "loss": 4.634,
+      "epoch": 0.19,
+      "learning_rate": 0.000456,
+      "loss": 4.9773,
       "step": 231
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.000462,
-      "loss": 4.4508,
+      "epoch": 0.2,
+      "learning_rate": 0.000458,
+      "loss": 4.8306,
       "step": 232
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.00046400000000000006,
-      "loss": 4.7686,
+      "epoch": 0.2,
+      "learning_rate": 0.00046,
+      "loss": 5.3968,
       "step": 233
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.00046600000000000005,
-      "loss": 4.6839,
+      "epoch": 0.2,
+      "learning_rate": 0.000462,
+      "loss": 4.779,
       "step": 234
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.00046800000000000005,
-      "loss": 4.8609,
+      "epoch": 0.2,
+      "learning_rate": 0.00046400000000000006,
+      "loss": 4.9465,
       "step": 235
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.00047,
-      "loss": 4.5689,
+      "epoch": 0.2,
+      "learning_rate": 0.00046600000000000005,
+      "loss": 5.0305,
       "step": 236
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.000472,
-      "loss": 4.5626,
+      "epoch": 0.2,
+      "learning_rate": 0.00046800000000000005,
+      "loss": 4.8854,
       "step": 237
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.000474,
-      "loss": 4.4552,
+      "epoch": 0.2,
+      "learning_rate": 0.00047,
+      "loss": 4.9381,
       "step": 238
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.00047599999999999997,
-      "loss": 4.6341,
+      "epoch": 0.2,
+      "learning_rate": 0.000472,
+      "loss": 4.7391,
       "step": 239
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.00047799999999999996,
-      "loss": 4.7347,
+      "epoch": 0.2,
+      "learning_rate": 0.000474,
+      "loss": 4.7947,
       "step": 240
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.00048,
-      "loss": 4.5554,
+      "epoch": 0.2,
+      "learning_rate": 0.00047599999999999997,
+      "loss": 5.3118,
       "step": 241
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.000482,
-      "loss": 5.1093,
+      "epoch": 0.2,
+      "learning_rate": 0.00047799999999999996,
+      "loss": 4.6798,
       "step": 242
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.000484,
-      "loss": 4.8549,
+      "epoch": 0.2,
+      "learning_rate": 0.00048,
+      "loss": 5.0058,
       "step": 243
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 0.000486,
-      "loss": 4.7953,
+      "epoch": 0.21,
+      "learning_rate": 0.000482,
+      "loss": 4.8296,
       "step": 244
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 0.000488,
-      "loss": 5.1484,
+      "epoch": 0.21,
+      "learning_rate": 0.000484,
+      "loss": 6.0175,
       "step": 245
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 0.00049,
-      "loss": 4.5329,
+      "epoch": 0.21,
+      "learning_rate": 0.000486,
+      "loss": 4.7253,
       "step": 246
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 0.000492,
-      "loss": 4.4645,
+      "epoch": 0.21,
+      "learning_rate": 0.000488,
+      "loss": 5.8907,
       "step": 247
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.000494,
-      "loss": 5.3223,
+      "epoch": 0.21,
+      "learning_rate": 0.00049,
+      "loss": 5.0653,
       "step": 248
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.000496,
-      "loss": 6.4257,
+      "epoch": 0.21,
+      "learning_rate": 0.000492,
+      "loss": 7.6408,
       "step": 249
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.000498,
-      "loss": 6.3633,
+      "epoch": 0.21,
+      "learning_rate": 0.000494,
+      "loss": 5.0257,
       "step": 250
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.0005,
-      "loss": 6.1391,
+      "epoch": 0.21,
+      "learning_rate": 0.000496,
+      "loss": 5.3533,
       "step": 251
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.0005020000000000001,
-      "loss": 5.3127,
+      "epoch": 0.21,
+      "learning_rate": 0.000498,
+      "loss": 4.7404,
       "step": 252
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.000504,
-      "loss": 4.912,
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 5.0286,
       "step": 253
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.000506,
-      "loss": 4.6873,
+      "epoch": 0.21,
+      "learning_rate": 0.0005020000000000001,
+      "loss": 4.9493,
       "step": 254
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.000508,
-      "loss": 4.9215,
+      "epoch": 0.21,
+      "learning_rate": 0.000504,
+      "loss": 4.8882,
       "step": 255
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.00051,
-      "loss": 4.6735,
+      "epoch": 0.22,
+      "learning_rate": 0.000506,
+      "loss": 5.1059,
       "step": 256
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.000512,
-      "loss": 5.0567,
+      "epoch": 0.22,
+      "learning_rate": 0.000508,
+      "loss": 4.8053,
       "step": 257
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.000514,
-      "loss": 5.2453,
+      "epoch": 0.22,
+      "learning_rate": 0.00051,
+      "loss": 5.2623,
       "step": 258
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.0005160000000000001,
-      "loss": 4.8018,
+      "epoch": 0.22,
+      "learning_rate": 0.000512,
+      "loss": 5.0566,
       "step": 259
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.000518,
-      "loss": 5.2815,
+      "epoch": 0.22,
+      "learning_rate": 0.000514,
+      "loss": 4.8121,
       "step": 260
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.0005200000000000001,
-      "loss": 4.7621,
+      "epoch": 0.22,
+      "learning_rate": 0.0005160000000000001,
+      "loss": 4.9033,
       "step": 261
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.000522,
-      "loss": 4.6185,
+      "epoch": 0.22,
+      "learning_rate": 0.000518,
+      "loss": 5.1741,
       "step": 262
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.000524,
-      "loss": 4.9622,
+      "epoch": 0.22,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 5.0122,
       "step": 263
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.000526,
-      "loss": 4.9574,
+      "epoch": 0.22,
+      "learning_rate": 0.000522,
+      "loss": 5.4247,
       "step": 264
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.000528,
-      "loss": 4.6961,
+      "epoch": 0.22,
+      "learning_rate": 0.000524,
+      "loss": 5.0175,
       "step": 265
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.0005300000000000001,
-      "loss": 4.8911,
+      "epoch": 0.22,
+      "learning_rate": 0.000526,
+      "loss": 5.2525,
       "step": 266
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.000532,
-      "loss": 4.8782,
+      "epoch": 0.22,
+      "learning_rate": 0.000528,
+      "loss": 4.8613,
       "step": 267
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.0005340000000000001,
-      "loss": 4.9493,
+      "epoch": 0.23,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 4.9458,
       "step": 268
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.000536,
-      "loss": 4.8858,
+      "epoch": 0.23,
+      "learning_rate": 0.000532,
+      "loss": 5.0877,
       "step": 269
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.0005380000000000001,
-      "loss": 5.1246,
+      "epoch": 0.23,
+      "learning_rate": 0.0005340000000000001,
+      "loss": 4.8471,
       "step": 270
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.00054,
-      "loss": 4.5835,
+      "epoch": 0.23,
+      "learning_rate": 0.000536,
+      "loss": 5.1964,
       "step": 271
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.0005420000000000001,
-      "loss": 4.6087,
+      "epoch": 0.23,
+      "learning_rate": 0.0005380000000000001,
+      "loss": 4.7942,
       "step": 272
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.0005440000000000001,
-      "loss": 4.7457,
+      "epoch": 0.23,
+      "learning_rate": 0.00054,
+      "loss": 4.8114,
       "step": 273
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.000546,
-      "loss": 4.8951,
+      "epoch": 0.23,
+      "learning_rate": 0.0005420000000000001,
+      "loss": 4.8993,
       "step": 274
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.0005480000000000001,
-      "loss": 5.3403,
+      "epoch": 0.23,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 4.8725,
       "step": 275
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.00055,
-      "loss": 4.6695,
+      "epoch": 0.23,
+      "learning_rate": 0.000546,
+      "loss": 4.6347,
       "step": 276
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.0005520000000000001,
-      "loss": 4.783,
+      "epoch": 0.23,
+      "learning_rate": 0.0005480000000000001,
+      "loss": 5.6725,
       "step": 277
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.000554,
-      "loss": 4.7711,
+      "epoch": 0.23,
+      "learning_rate": 0.00055,
+      "loss": 5.1446,
       "step": 278
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0005560000000000001,
-      "loss": 5.0676,
+      "epoch": 0.23,
+      "learning_rate": 0.0005520000000000001,
+      "loss": 5.1579,
       "step": 279
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.000558,
-      "loss": 4.8579,
+      "epoch": 0.24,
+      "learning_rate": 0.000554,
+      "loss": 4.9103,
       "step": 280
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0005600000000000001,
-      "loss": 5.0367,
+      "epoch": 0.24,
+      "learning_rate": 0.0005560000000000001,
+      "loss": 4.9805,
       "step": 281
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0005620000000000001,
-      "loss": 4.9462,
+      "epoch": 0.24,
+      "learning_rate": 0.000558,
+      "loss": 4.8509,
       "step": 282
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0005639999999999999,
-      "loss": 4.7645,
+      "epoch": 0.24,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 5.5893,
       "step": 283
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.000566,
-      "loss": 4.6745,
+      "epoch": 0.24,
+      "learning_rate": 0.0005620000000000001,
+      "loss": 5.1396,
       "step": 284
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.0005679999999999999,
-      "loss": 4.9527,
+      "epoch": 0.24,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 4.9078,
       "step": 285
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.00057,
-      "loss": 4.9766,
+      "epoch": 0.24,
+      "learning_rate": 0.000566,
+      "loss": 4.7542,
       "step": 286
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.0005719999999999999,
-      "loss": 5.3699,
+      "epoch": 0.24,
+      "learning_rate": 0.0005679999999999999,
+      "loss": 4.8776,
       "step": 287
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.000574,
-      "loss": 4.7668,
+      "epoch": 0.24,
+      "learning_rate": 0.00057,
+      "loss": 4.6694,
       "step": 288
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.000576,
-      "loss": 4.6789,
+      "epoch": 0.24,
+      "learning_rate": 0.0005719999999999999,
+      "loss": 4.8497,
       "step": 289
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.000578,
-      "loss": 5.0322,
+      "epoch": 0.24,
+      "learning_rate": 0.000574,
+      "loss": 4.8594,
       "step": 290
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.00058,
-      "loss": 5.2736,
+      "epoch": 0.24,
+      "learning_rate": 0.000576,
+      "loss": 5.1021,
       "step": 291
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.0005819999999999999,
-      "loss": 5.0443,
+      "epoch": 0.25,
+      "learning_rate": 0.000578,
+      "loss": 5.8118,
       "step": 292
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.000584,
-      "loss": 4.8732,
+      "epoch": 0.25,
+      "learning_rate": 0.00058,
+      "loss": 4.7193,
       "step": 293
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.0005859999999999999,
-      "loss": 5.166,
+      "epoch": 0.25,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 4.9685,
       "step": 294
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.000588,
-      "loss": 4.8355,
+      "epoch": 0.25,
+      "learning_rate": 0.000584,
+      "loss": 5.0434,
       "step": 295
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.00059,
-      "loss": 4.8877,
+      "epoch": 0.25,
+      "learning_rate": 0.0005859999999999999,
+      "loss": 6.1779,
       "step": 296
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.000592,
-      "loss": 5.2269,
+      "epoch": 0.25,
+      "learning_rate": 0.000588,
+      "loss": 5.5563,
       "step": 297
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.000594,
-      "loss": 17.9702,
+      "epoch": 0.25,
+      "learning_rate": 0.00059,
+      "loss": 5.2643,
       "step": 298
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.000596,
-      "loss": 5.7343,
+      "epoch": 0.25,
+      "learning_rate": 0.000592,
+      "loss": 5.9004,
       "step": 299
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.000598,
-      "loss": 5.515,
+      "epoch": 0.25,
+      "learning_rate": 0.000594,
+      "loss": 5.223,
       "step": 300
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.0006,
-      "loss": 5.3824,
+      "epoch": 0.25,
+      "learning_rate": 0.000596,
+      "loss": 5.1204,
       "step": 301
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.000602,
-      "loss": 5.148,
+      "epoch": 0.25,
+      "learning_rate": 0.000598,
+      "loss": 5.3329,
       "step": 302
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.000604,
-      "loss": 4.8959,
+      "epoch": 0.25,
+      "learning_rate": 0.0006,
+      "loss": 5.4375,
       "step": 303
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.000606,
-      "loss": 5.0647,
+      "epoch": 0.26,
+      "learning_rate": 0.000602,
+      "loss": 4.8753,
       "step": 304
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.000608,
-      "loss": 4.8941,
+      "epoch": 0.26,
+      "learning_rate": 0.000604,
+      "loss": 5.0302,
       "step": 305
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.00061,
-      "loss": 4.8608,
+      "epoch": 0.26,
+      "learning_rate": 0.000606,
+      "loss": 5.1604,
       "step": 306
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.000612,
-      "loss": 5.2265,
+      "epoch": 0.26,
+      "learning_rate": 0.000608,
+      "loss": 4.968,
       "step": 307
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.000614,
-      "loss": 5.0575,
+      "epoch": 0.26,
+      "learning_rate": 0.00061,
+      "loss": 4.9019,
       "step": 308
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.000616,
-      "loss": 4.9254,
+      "epoch": 0.26,
+      "learning_rate": 0.000612,
+      "loss": 4.6643,
       "step": 309
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.0006180000000000001,
-      "loss": 4.9891,
+      "epoch": 0.26,
+      "learning_rate": 0.000614,
+      "loss": 5.2494,
       "step": 310
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.00062,
-      "loss": 4.5658,
+      "epoch": 0.26,
+      "learning_rate": 0.000616,
+      "loss": 5.8753,
       "step": 311
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.000622,
-      "loss": 4.9779,
+      "epoch": 0.26,
+      "learning_rate": 0.0006180000000000001,
+      "loss": 4.9128,
       "step": 312
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.000624,
-      "loss": 5.3951,
+      "epoch": 0.26,
+      "learning_rate": 0.00062,
+      "loss": 4.9045,
       "step": 313
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.000626,
-      "loss": 4.796,
-      "step": 314
+      "epoch": 0.26,
+      "learning_rate": 0.000622,
+      "loss": 5.2783,
+      "step": 314
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.000628,
-      "loss": 4.8412,
+      "epoch": 0.26,
+      "learning_rate": 0.000624,
+      "loss": 4.7307,
       "step": 315
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.00063,
-      "loss": 4.6783,
+      "epoch": 0.27,
+      "learning_rate": 0.000626,
+      "loss": 5.1032,
       "step": 316
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.000632,
-      "loss": 4.7405,
+      "epoch": 0.27,
+      "learning_rate": 0.000628,
+      "loss": 5.4256,
       "step": 317
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.000634,
-      "loss": 4.6025,
+      "epoch": 0.27,
+      "learning_rate": 0.00063,
+      "loss": 5.1252,
       "step": 318
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.0006360000000000001,
-      "loss": 4.8142,
+      "epoch": 0.27,
+      "learning_rate": 0.000632,
+      "loss": 5.1324,
       "step": 319
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.000638,
-      "loss": 4.8876,
+      "epoch": 0.27,
+      "learning_rate": 0.000634,
+      "loss": 4.8646,
       "step": 320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.00064,
-      "loss": 4.8069,
+      "epoch": 0.27,
+      "learning_rate": 0.0006360000000000001,
+      "loss": 4.9822,
       "step": 321
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.000642,
-      "loss": 4.8133,
+      "epoch": 0.27,
+      "learning_rate": 0.000638,
+      "loss": 4.7422,
       "step": 322
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.000644,
-      "loss": 4.7322,
+      "epoch": 0.27,
+      "learning_rate": 0.00064,
+      "loss": 4.7852,
       "step": 323
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.000646,
-      "loss": 4.5794,
+      "epoch": 0.27,
+      "learning_rate": 0.000642,
+      "loss": 5.3777,
       "step": 324
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.000648,
-      "loss": 5.0187,
+      "epoch": 0.27,
+      "learning_rate": 0.000644,
+      "loss": 4.7624,
       "step": 325
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.0006500000000000001,
-      "loss": 5.1333,
+      "epoch": 0.27,
+      "learning_rate": 0.000646,
+      "loss": 5.4472,
       "step": 326
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.000652,
-      "loss": 4.7612,
+      "epoch": 0.27,
+      "learning_rate": 0.000648,
+      "loss": 5.2027,
       "step": 327
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.0006540000000000001,
-      "loss": 4.6401,
+      "epoch": 0.28,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 5.1171,
       "step": 328
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.000656,
-      "loss": 4.7183,
+      "epoch": 0.28,
+      "learning_rate": 0.000652,
+      "loss": 4.8381,
       "step": 329
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.0006580000000000001,
-      "loss": 4.8948,
+      "epoch": 0.28,
+      "learning_rate": 0.0006540000000000001,
+      "loss": 5.2469,
       "step": 330
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.00066,
-      "loss": 4.7955,
+      "epoch": 0.28,
+      "learning_rate": 0.000656,
+      "loss": 5.0761,
       "step": 331
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.000662,
-      "loss": 4.7045,
+      "epoch": 0.28,
+      "learning_rate": 0.0006580000000000001,
+      "loss": 5.9278,
       "step": 332
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.0006640000000000001,
-      "loss": 4.9097,
+      "epoch": 0.28,
+      "learning_rate": 0.00066,
+      "loss": 5.1369,
       "step": 333
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.000666,
-      "loss": 4.924,
+      "epoch": 0.28,
+      "learning_rate": 0.000662,
+      "loss": 5.2111,
       "step": 334
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.0006680000000000001,
-      "loss": 4.8816,
+      "epoch": 0.28,
+      "learning_rate": 0.0006640000000000001,
+      "loss": 5.3243,
       "step": 335
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.00067,
-      "loss": 4.6617,
+      "epoch": 0.28,
+      "learning_rate": 0.000666,
+      "loss": 5.3962,
       "step": 336
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.0006720000000000001,
-      "loss": 4.8668,
+      "epoch": 0.28,
+      "learning_rate": 0.0006680000000000001,
+      "loss": 4.895,
       "step": 337
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.000674,
-      "loss": 4.813,
+      "epoch": 0.28,
+      "learning_rate": 0.00067,
+      "loss": 5.2441,
       "step": 338
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.0006760000000000001,
-      "loss": 4.869,
+      "epoch": 0.28,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 5.0612,
       "step": 339
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.0006780000000000001,
-      "loss": 4.9609,
+      "epoch": 0.29,
+      "learning_rate": 0.000674,
+      "loss": 5.0606,
       "step": 340
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.00068,
-      "loss": 4.735,
+      "epoch": 0.29,
+      "learning_rate": 0.0006760000000000001,
+      "loss": 5.3363,
       "step": 341
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.0006820000000000001,
-      "loss": 6.541,
+      "epoch": 0.29,
+      "learning_rate": 0.0006780000000000001,
+      "loss": 5.3404,
       "step": 342
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.000684,
-      "loss": 5.0304,
+      "epoch": 0.29,
+      "learning_rate": 0.00068,
+      "loss": 5.4006,
       "step": 343
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.0006860000000000001,
-      "loss": 4.9253,
+      "epoch": 0.29,
+      "learning_rate": 0.0006820000000000001,
+      "loss": 5.3602,
       "step": 344
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.0006879999999999999,
-      "loss": 4.7216,
+      "epoch": 0.29,
+      "learning_rate": 0.000684,
+      "loss": 4.8384,
       "step": 345
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.00069,
-      "loss": 4.8354,
+      "epoch": 0.29,
+      "learning_rate": 0.0006860000000000001,
+      "loss": 5.4087,
       "step": 346
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.000692,
-      "loss": 4.8476,
+      "epoch": 0.29,
+      "learning_rate": 0.0006879999999999999,
+      "loss": 5.179,
       "step": 347
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.000694,
-      "loss": 4.8727,
+      "epoch": 0.29,
+      "learning_rate": 0.00069,
+      "loss": 5.6626,
       "step": 348
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.000696,
-      "loss": 5.0211,
+      "epoch": 0.29,
+      "learning_rate": 0.000692,
+      "loss": 5.3192,
       "step": 349
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.0006979999999999999,
-      "loss": 4.7231,
+      "epoch": 0.29,
+      "learning_rate": 0.000694,
+      "loss": 5.1726,
       "step": 350
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.0007,
-      "loss": 6.3094,
+      "epoch": 0.3,
+      "learning_rate": 0.000696,
+      "loss": 6.0533,
       "step": 351
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.0007019999999999999,
-      "loss": 5.0654,
+      "epoch": 0.3,
+      "learning_rate": 0.0006979999999999999,
+      "loss": 5.3456,
       "step": 352
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.000704,
-      "loss": 4.8125,
+      "epoch": 0.3,
+      "learning_rate": 0.0007,
+      "loss": 5.7693,
       "step": 353
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.0007059999999999999,
-      "loss": 4.9841,
+      "epoch": 0.3,
+      "learning_rate": 0.0007019999999999999,
+      "loss": 5.147,
       "step": 354
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.000708,
-      "loss": 4.9406,
+      "epoch": 0.3,
+      "learning_rate": 0.000704,
+      "loss": 5.052,
       "step": 355
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.00071,
-      "loss": 5.3482,
+      "epoch": 0.3,
+      "learning_rate": 0.0007059999999999999,
+      "loss": 5.25,
       "step": 356
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.000712,
-      "loss": 5.2098,
+      "epoch": 0.3,
+      "learning_rate": 0.000708,
+      "loss": 5.1822,
       "step": 357
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.000714,
-      "loss": 5.0179,
+      "epoch": 0.3,
+      "learning_rate": 0.00071,
+      "loss": 4.8395,
       "step": 358
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.000716,
-      "loss": 4.8699,
+      "epoch": 0.3,
+      "learning_rate": 0.000712,
+      "loss": 5.0995,
       "step": 359
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.000718,
-      "loss": 4.8703,
+      "epoch": 0.3,
+      "learning_rate": 0.000714,
+      "loss": 5.1604,
       "step": 360
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.0007199999999999999,
-      "loss": 4.8909,
+      "epoch": 0.3,
+      "learning_rate": 0.000716,
+      "loss": 4.9228,
       "step": 361
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.000722,
-      "loss": 4.8246,
+      "epoch": 0.3,
+      "learning_rate": 0.000718,
+      "loss": 5.1524,
       "step": 362
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.000724,
-      "loss": 4.6949,
+      "epoch": 0.31,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 5.0732,
       "step": 363
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.000726,
-      "loss": 5.0433,
+      "epoch": 0.31,
+      "learning_rate": 0.000722,
+      "loss": 5.0136,
       "step": 364
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.000728,
-      "loss": 5.055,
+      "epoch": 0.31,
+      "learning_rate": 0.000724,
+      "loss": 6.0669,
       "step": 365
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.00073,
-      "loss": 4.7938,
+      "epoch": 0.31,
+      "learning_rate": 0.000726,
+      "loss": 5.4818,
       "step": 366
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.000732,
-      "loss": 4.8905,
+      "epoch": 0.31,
+      "learning_rate": 0.000728,
+      "loss": 5.0185,
       "step": 367
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.000734,
-      "loss": 4.7919,
+      "epoch": 0.31,
+      "learning_rate": 0.00073,
+      "loss": 5.243,
       "step": 368
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.000736,
-      "loss": 4.7992,
+      "epoch": 0.31,
+      "learning_rate": 0.000732,
+      "loss": 5.1703,
       "step": 369
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.000738,
-      "loss": 4.7454,
+      "epoch": 0.31,
+      "learning_rate": 0.000734,
+      "loss": 5.2142,
       "step": 370
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.00074,
-      "loss": 4.8729,
+      "epoch": 0.31,
+      "learning_rate": 0.000736,
+      "loss": 5.3355,
       "step": 371
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.000742,
-      "loss": 4.8837,
+      "epoch": 0.31,
+      "learning_rate": 0.000738,
+      "loss": 5.1462,
       "step": 372
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.000744,
-      "loss": 4.7922,
+      "epoch": 0.31,
+      "learning_rate": 0.00074,
+      "loss": 4.9194,
       "step": 373
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.000746,
-      "loss": 4.6167,
+      "epoch": 0.31,
+      "learning_rate": 0.000742,
+      "loss": 4.9267,
       "step": 374
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.000748,
-      "loss": 4.7867,
-      "step": 375
+      "epoch": 0.32,
+      "learning_rate": 0.000744,
+      "loss": 4.9824,
+      "step": 375
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.00075,
-      "loss": 4.708,
+      "epoch": 0.32,
+      "learning_rate": 0.000746,
+      "loss": 5.2742,
       "step": 376
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.0007520000000000001,
-      "loss": 4.7856,
+      "epoch": 0.32,
+      "learning_rate": 0.000748,
+      "loss": 4.8618,
       "step": 377
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.000754,
-      "loss": 4.8291,
+      "epoch": 0.32,
+      "learning_rate": 0.00075,
+      "loss": 4.868,
       "step": 378
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.000756,
-      "loss": 4.8428,
+      "epoch": 0.32,
+      "learning_rate": 0.0007520000000000001,
+      "loss": 5.1074,
       "step": 379
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.000758,
-      "loss": 4.7221,
+      "epoch": 0.32,
+      "learning_rate": 0.000754,
+      "loss": 5.1586,
       "step": 380
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.00076,
-      "loss": 5.227,
+      "epoch": 0.32,
+      "learning_rate": 0.000756,
+      "loss": 5.0592,
       "step": 381
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.000762,
-      "loss": 5.304,
+      "epoch": 0.32,
+      "learning_rate": 0.000758,
+      "loss": 5.0195,
       "step": 382
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.000764,
-      "loss": 4.8953,
+      "epoch": 0.32,
+      "learning_rate": 0.00076,
+      "loss": 5.0721,
       "step": 383
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.0007660000000000001,
-      "loss": 4.7784,
+      "epoch": 0.32,
+      "learning_rate": 0.000762,
+      "loss": 5.4243,
       "step": 384
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.000768,
-      "loss": 4.874,
+      "epoch": 0.32,
+      "learning_rate": 0.000764,
+      "loss": 5.1279,
       "step": 385
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.0007700000000000001,
-      "loss": 4.727,
+      "epoch": 0.32,
+      "learning_rate": 0.0007660000000000001,
+      "loss": 5.0395,
       "step": 386
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.000772,
-      "loss": 4.9384,
+      "epoch": 0.33,
+      "learning_rate": 0.000768,
+      "loss": 5.1286,
       "step": 387
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.0007740000000000001,
-      "loss": 4.8338,
+      "epoch": 0.33,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 5.0282,
       "step": 388
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.000776,
-      "loss": 4.8328,
+      "epoch": 0.33,
+      "learning_rate": 0.000772,
+      "loss": 5.2939,
       "step": 389
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.000778,
-      "loss": 4.7213,
+      "epoch": 0.33,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 5.3163,
       "step": 390
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.0007800000000000001,
-      "loss": 4.9069,
+      "epoch": 0.33,
+      "learning_rate": 0.000776,
+      "loss": 5.1947,
       "step": 391
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.000782,
-      "loss": 4.97,
+      "epoch": 0.33,
+      "learning_rate": 0.000778,
+      "loss": 5.9107,
       "step": 392
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.0007840000000000001,
-      "loss": 4.7095,
+      "epoch": 0.33,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 5.1329,
       "step": 393
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.000786,
-      "loss": 4.7421,
+      "epoch": 0.33,
+      "learning_rate": 0.000782,
+      "loss": 5.2657,
       "step": 394
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.0007880000000000001,
-      "loss": 4.8849,
+      "epoch": 0.33,
+      "learning_rate": 0.0007840000000000001,
+      "loss": 4.8574,
       "step": 395
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.00079,
-      "loss": 5.2645,
+      "epoch": 0.33,
+      "learning_rate": 0.000786,
+      "loss": 6.4753,
       "step": 396
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.0007920000000000001,
-      "loss": 5.1189,
+      "epoch": 0.33,
+      "learning_rate": 0.0007880000000000001,
+      "loss": 5.6222,
       "step": 397
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.0007940000000000001,
-      "loss": 4.8777,
+      "epoch": 0.33,
+      "learning_rate": 0.00079,
+      "loss": 5.3745,
       "step": 398
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.000796,
-      "loss": 5.253,
+      "epoch": 0.34,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 5.4114,
       "step": 399
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.0007980000000000001,
-      "loss": 4.937,
+      "epoch": 0.34,
+      "learning_rate": 0.0007940000000000001,
+      "loss": 6.4515,
       "step": 400
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.0008,
-      "loss": 5.2032,
+      "epoch": 0.34,
+      "learning_rate": 0.000796,
+      "loss": 6.6052,
       "step": 401
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.0008020000000000001,
-      "loss": 5.0266,
+      "epoch": 0.34,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 5.3085,
       "step": 402
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.000804,
-      "loss": 5.1012,
+      "epoch": 0.34,
+      "learning_rate": 0.0008,
+      "loss": 5.7514,
       "step": 403
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0008060000000000001,
-      "loss": 4.9038,
+      "epoch": 0.34,
+      "learning_rate": 0.0008020000000000001,
+      "loss": 5.3295,
       "step": 404
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.000808,
-      "loss": 4.8848,
+      "epoch": 0.34,
+      "learning_rate": 0.000804,
+      "loss": 5.3105,
       "step": 405
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0008100000000000001,
-      "loss": 5.2038,
+      "epoch": 0.34,
+      "learning_rate": 0.0008060000000000001,
+      "loss": 5.1785,
       "step": 406
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0008120000000000001,
-      "loss": 4.9554,
+      "epoch": 0.34,
+      "learning_rate": 0.000808,
+      "loss": 5.1525,
       "step": 407
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0008139999999999999,
-      "loss": 5.1101,
+      "epoch": 0.34,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 6.1956,
       "step": 408
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.000816,
-      "loss": 4.8516,
+      "epoch": 0.34,
+      "learning_rate": 0.0008120000000000001,
+      "loss": 5.7512,
       "step": 409
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.0008179999999999999,
-      "loss": 4.6971,
+      "epoch": 0.34,
+      "learning_rate": 0.0008139999999999999,
+      "loss": 5.4321,
       "step": 410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.00082,
-      "loss": 5.1527,
+      "epoch": 0.35,
+      "learning_rate": 0.000816,
+      "loss": 5.5366,
       "step": 411
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.0008219999999999999,
-      "loss": 4.9715,
+      "epoch": 0.35,
+      "learning_rate": 0.0008179999999999999,
+      "loss": 5.2462,
       "step": 412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.000824,
-      "loss": 4.6715,
+      "epoch": 0.35,
+      "learning_rate": 0.00082,
+      "loss": 5.1589,
       "step": 413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.000826,
-      "loss": 5.0357,
+      "epoch": 0.35,
+      "learning_rate": 0.0008219999999999999,
+      "loss": 4.8115,
       "step": 414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.000828,
-      "loss": 5.0545,
+      "epoch": 0.35,
+      "learning_rate": 0.000824,
+      "loss": 5.1286,
       "step": 415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.00083,
-      "loss": 4.9108,
+      "epoch": 0.35,
+      "learning_rate": 0.000826,
+      "loss": 5.4853,
       "step": 416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.000832,
-      "loss": 5.2884,
+      "epoch": 0.35,
+      "learning_rate": 0.000828,
+      "loss": 4.7102,
       "step": 417
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.000834,
-      "loss": 4.8856,
+      "epoch": 0.35,
+      "learning_rate": 0.00083,
+      "loss": 5.1715,
       "step": 418
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.0008359999999999999,
-      "loss": 5.253,
+      "epoch": 0.35,
+      "learning_rate": 0.000832,
+      "loss": 5.1337,
       "step": 419
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.000838,
-      "loss": 5.1539,
+      "epoch": 0.35,
+      "learning_rate": 0.000834,
+      "loss": 5.0089,
       "step": 420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.00084,
-      "loss": 5.2202,
+      "epoch": 0.35,
+      "learning_rate": 0.0008359999999999999,
+      "loss": 5.382,
       "step": 421
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.000842,
-      "loss": 4.7612,
+      "epoch": 0.35,
+      "learning_rate": 0.000838,
+      "loss": 5.008,
       "step": 422
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.000844,
-      "loss": 4.8588,
+      "epoch": 0.36,
+      "learning_rate": 0.00084,
+      "loss": 5.1232,
       "step": 423
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.000846,
-      "loss": 5.0295,
+      "epoch": 0.36,
+      "learning_rate": 0.000842,
+      "loss": 5.0953,
       "step": 424
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.000848,
-      "loss": 4.7428,
+      "epoch": 0.36,
+      "learning_rate": 0.000844,
+      "loss": 5.5967,
       "step": 425
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.00085,
-      "loss": 4.7598,
+      "epoch": 0.36,
+      "learning_rate": 0.000846,
+      "loss": 5.0625,
       "step": 426
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.000852,
-      "loss": 4.7621,
+      "epoch": 0.36,
+      "learning_rate": 0.000848,
+      "loss": 4.8947,
       "step": 427
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.000854,
-      "loss": 4.7438,
+      "epoch": 0.36,
+      "learning_rate": 0.00085,
+      "loss": 4.9733,
       "step": 428
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.000856,
-      "loss": 4.5182,
+      "epoch": 0.36,
+      "learning_rate": 0.000852,
+      "loss": 5.9625,
       "step": 429
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.000858,
-      "loss": 5.0404,
+      "epoch": 0.36,
+      "learning_rate": 0.000854,
+      "loss": 5.401,
       "step": 430
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.00086,
-      "loss": 4.814,
+      "epoch": 0.36,
+      "learning_rate": 0.000856,
+      "loss": 5.1208,
       "step": 431
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.000862,
-      "loss": 5.1218,
+      "epoch": 0.36,
+      "learning_rate": 0.000858,
+      "loss": 5.0253,
       "step": 432
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.000864,
-      "loss": 4.7879,
+      "epoch": 0.36,
+      "learning_rate": 0.00086,
+      "loss": 5.2245,
       "step": 433
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.000866,
-      "loss": 4.8229,
+      "epoch": 0.36,
+      "learning_rate": 0.000862,
+      "loss": 5.0729,
       "step": 434
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 0.0008680000000000001,
-      "loss": 5.1322,
+      "epoch": 0.37,
+      "learning_rate": 0.000864,
+      "loss": 5.2137,
       "step": 435
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 0.00087,
-      "loss": 4.6759,
+      "epoch": 0.37,
+      "learning_rate": 0.000866,
+      "loss": 5.029,
       "step": 436
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 0.000872,
-      "loss": 4.9387,
+      "epoch": 0.37,
+      "learning_rate": 0.0008680000000000001,
+      "loss": 4.8951,
       "step": 437
     },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00087,
+      "loss": 4.9978,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000872,
+      "loss": 5.3774,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000874,
+      "loss": 5.0731,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000876,
+      "loss": 6.0037,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000878,
+      "loss": 5.2605,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00088,
+      "loss": 5.0199,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000882,
+      "loss": 5.0762,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000884,
+      "loss": 4.847,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0008860000000000001,
+      "loss": 5.0762,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000888,
+      "loss": 7.0061,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0008900000000000001,
+      "loss": 6.2748,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000892,
+      "loss": 5.3964,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000894,
+      "loss": 5.2459,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000896,
+      "loss": 5.7893,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000898,
+      "loss": 5.1035,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 5.2482,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000902,
+      "loss": 5.1441,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009040000000000001,
+      "loss": 5.9311,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000906,
+      "loss": 6.0411,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009080000000000001,
+      "loss": 5.4916,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00091,
+      "loss": 6.1491,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000912,
+      "loss": 5.3559,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009140000000000001,
+      "loss": 5.4161,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000916,
+      "loss": 5.4883,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009180000000000001,
+      "loss": 5.2702,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00092,
+      "loss": 5.2641,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009220000000000001,
+      "loss": 5.9715,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000924,
+      "loss": 5.3146,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009260000000000001,
+      "loss": 5.5785,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009280000000000001,
+      "loss": 5.0883,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00093,
+      "loss": 5.1286,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009320000000000001,
+      "loss": 5.0579,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000934,
+      "loss": 5.3605,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009360000000000001,
+      "loss": 5.1274,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 5.2791,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00094,
+      "loss": 5.037,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000942,
+      "loss": 5.3842,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000944,
+      "loss": 5.2689,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000946,
+      "loss": 5.3079,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000948,
+      "loss": 5.0145,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00095,
+      "loss": 5.0121,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009519999999999999,
+      "loss": 5.0444,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000954,
+      "loss": 5.0832,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009559999999999999,
+      "loss": 4.9396,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000958,
+      "loss": 5.0183,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00096,
+      "loss": 5.7342,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000962,
+      "loss": 5.1733,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000964,
+      "loss": 5.0985,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000966,
+      "loss": 5.3892,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000968,
+      "loss": 4.9622,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009699999999999999,
+      "loss": 5.0768,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000972,
+      "loss": 5.5292,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000974,
+      "loss": 4.9231,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000976,
+      "loss": 5.5424,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000978,
+      "loss": 5.4956,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00098,
+      "loss": 5.6218,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000982,
+      "loss": 6.0528,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000984,
+      "loss": 6.9273,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009860000000000001,
+      "loss": 5.5907,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000988,
+      "loss": 5.3488,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00099,
+      "loss": 5.4343,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000992,
+      "loss": 5.2812,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000994,
+      "loss": 5.4469,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 5.590612411499023,
+      "eval_runtime": 787.3401,
+      "eval_samples_per_second": 3.356,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 1.9976993256644189,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000996,
+      "loss": 5.5798,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000998,
+      "loss": 5.6459,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001,
+      "loss": 5.5149,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009985486211901307,
+      "loss": 5.2738,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009970972423802612,
+      "loss": 5.3517,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000995645863570392,
+      "loss": 5.6229,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009941944847605226,
+      "loss": 5.2317,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009927431059506531,
+      "loss": 5.2242,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009912917271407838,
+      "loss": 5.1385,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009898403483309143,
+      "loss": 5.4837,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000988388969521045,
+      "loss": 5.3382,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009869375907111755,
+      "loss": 5.0682,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009854862119013062,
+      "loss": 5.4414,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000984034833091437,
+      "loss": 5.0219,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009825834542815674,
+      "loss": 5.1192,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009811320754716981,
+      "loss": 5.166,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009796806966618288,
+      "loss": 5.7212,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009782293178519593,
+      "loss": 4.9735,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00097677793904209,
+      "loss": 5.7979,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009753265602322206,
+      "loss": 5.0325,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009738751814223512,
+      "loss": 5.1862,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009724238026124818,
+      "loss": 5.1709,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009709724238026125,
+      "loss": 6.1005,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000969521044992743,
+      "loss": 5.4103,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009680696661828737,
+      "loss": 5.1801,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009666182873730044,
+      "loss": 4.9993,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000965166908563135,
+      "loss": 5.256,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009637155297532656,
+      "loss": 5.3794,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009622641509433962,
+      "loss": 5.3789,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009608127721335269,
+      "loss": 5.2737,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009593613933236574,
+      "loss": 5.3525,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009579100145137881,
+      "loss": 5.7573,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009564586357039187,
+      "loss": 5.1226,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009550072568940493,
+      "loss": 5.4122,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009535558780841799,
+      "loss": 5.1807,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009521044992743106,
+      "loss": 5.3875,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009506531204644411,
+      "loss": 5.2093,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009492017416545718,
+      "loss": 5.3129,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009477503628447026,
+      "loss": 5.2434,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009462989840348331,
+      "loss": 5.5791,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009448476052249637,
+      "loss": 5.3577,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009433962264150943,
+      "loss": 5.4667,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000941944847605225,
+      "loss": 5.0717,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009404934687953555,
+      "loss": 5.3747,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009390420899854863,
+      "loss": 5.4907,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009375907111756169,
+      "loss": 5.132,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009361393323657474,
+      "loss": 5.5721,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000934687953555878,
+      "loss": 6.7533,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009332365747460088,
+      "loss": 5.292,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009317851959361394,
+      "loss": 6.7222,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00093033381712627,
+      "loss": 5.4979,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009288824383164007,
+      "loss": 5.6428,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009274310595065311,
+      "loss": 5.3726,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009259796806966619,
+      "loss": 5.3097,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009245283018867925,
+      "loss": 5.4151,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009230769230769232,
+      "loss": 5.3612,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009216255442670537,
+      "loss": 5.2058,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009201741654571844,
+      "loss": 5.0532,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000918722786647315,
+      "loss": 5.4266,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009172714078374456,
+      "loss": 5.5015,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009158200290275763,
+      "loss": 5.4375,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009143686502177069,
+      "loss": 5.3005,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009129172714078375,
+      "loss": 6.1749,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009114658925979681,
+      "loss": 4.9101,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009100145137880988,
+      "loss": 5.2097,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009085631349782293,
+      "loss": 5.1512,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00090711175616836,
+      "loss": 5.2671,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009056603773584906,
+      "loss": 5.324,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009042089985486213,
+      "loss": 5.0224,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009027576197387518,
+      "loss": 5.1475,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009013062409288825,
+      "loss": 5.1307,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008998548621190132,
+      "loss": 5.3267,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008984034833091437,
+      "loss": 5.1535,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008969521044992744,
+      "loss": 5.1622,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000895500725689405,
+      "loss": 5.343,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008940493468795356,
+      "loss": 5.5425,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008925979680696662,
+      "loss": 5.5587,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008911465892597969,
+      "loss": 5.037,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008896952104499274,
+      "loss": 5.1311,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008882438316400581,
+      "loss": 5.3214,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008867924528301887,
+      "loss": 5.2571,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008853410740203193,
+      "loss": 4.9891,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00088388969521045,
+      "loss": 5.3721,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008824383164005806,
+      "loss": 5.146,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008809869375907113,
+      "loss": 5.3849,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008795355587808418,
+      "loss": 5.2897,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008780841799709725,
+      "loss": 5.4393,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008766328011611031,
+      "loss": 5.0352,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008751814223512337,
+      "loss": 5.2357,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008737300435413643,
+      "loss": 5.3962,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000872278664731495,
+      "loss": 5.131,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008708272859216255,
+      "loss": 5.0494,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008693759071117562,
+      "loss": 5.6948,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008679245283018869,
+      "loss": 5.4695,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008664731494920174,
+      "loss": 5.0781,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008650217706821481,
+      "loss": 4.9336,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008635703918722787,
+      "loss": 5.1719,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008621190130624093,
+      "loss": 6.7856,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008606676342525399,
+      "loss": 6.1423,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008592162554426706,
+      "loss": 6.4766,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008577648766328012,
+      "loss": 5.2018,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008563134978229318,
+      "loss": 5.3829,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008548621190130624,
+      "loss": 5.5641,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008534107402031931,
+      "loss": 5.226,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008519593613933237,
+      "loss": 5.327,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008505079825834543,
+      "loss": 5.3499,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000849056603773585,
+      "loss": 5.1473,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008476052249637155,
+      "loss": 5.3517,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008461538461538462,
+      "loss": 5.1707,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008447024673439768,
+      "loss": 5.1805,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008432510885341074,
+      "loss": 5.2304,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000841799709724238,
+      "loss": 5.0367,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008403483309143687,
+      "loss": 5.9651,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008388969521044993,
+      "loss": 4.9595,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008374455732946299,
+      "loss": 5.1965,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008359941944847606,
+      "loss": 5.2899,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008345428156748912,
+      "loss": 5.282,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008330914368650218,
+      "loss": 5.6795,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008316400580551524,
+      "loss": 5.3106,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008301886792452831,
+      "loss": 5.1533,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008287373004354136,
+      "loss": 4.9877,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008272859216255443,
+      "loss": 4.8672,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008258345428156749,
+      "loss": 5.9693,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008243831640058055,
+      "loss": 5.3869,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008229317851959361,
+      "loss": 5.3817,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008214804063860668,
+      "loss": 5.357,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008200290275761973,
+      "loss": 5.185,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000818577648766328,
+      "loss": 4.8705,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008171262699564587,
+      "loss": 5.2537,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008156748911465893,
+      "loss": 5.1341,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008142235123367199,
+      "loss": 5.1746,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008127721335268505,
+      "loss": 5.1372,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008113207547169812,
+      "loss": 5.4583,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008098693759071117,
+      "loss": 5.0465,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008084179970972424,
+      "loss": 5.4324,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000806966618287373,
+      "loss": 5.0843,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008055152394775036,
+      "loss": 4.8874,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008040638606676342,
+      "loss": 4.8573,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008026124818577649,
+      "loss": 4.9093,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008011611030478955,
+      "loss": 5.1895,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007997097242380261,
+      "loss": 5.1419,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007982583454281568,
+      "loss": 5.538,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007968069666182874,
+      "loss": 5.4555,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000795355587808418,
+      "loss": 5.1396,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007939042089985486,
+      "loss": 5.2019,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007924528301886793,
+      "loss": 4.9881,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007910014513788098,
+      "loss": 4.8018,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007895500725689405,
+      "loss": 5.5678,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007880986937590711,
+      "loss": 5.4444,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007866473149492017,
+      "loss": 4.8348,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007851959361393324,
+      "loss": 5.4409,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000783744557329463,
+      "loss": 5.1152,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007822931785195936,
+      "loss": 5.1351,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007808417997097242,
+      "loss": 5.3607,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007793904208998549,
+      "loss": 5.1377,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007779390420899854,
+      "loss": 4.9976,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007764876632801161,
+      "loss": 5.1782,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007750362844702467,
+      "loss": 5.3511,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007735849056603774,
+      "loss": 5.3262,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007721335268505079,
+      "loss": 5.1326,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007706821480406386,
+      "loss": 5.1635,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007692307692307693,
+      "loss": 5.0094,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007677793904208998,
+      "loss": 5.0631,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007663280116110305,
+      "loss": 5.0579,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007648766328011611,
+      "loss": 5.6685,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007634252539912917,
+      "loss": 5.0147,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007619738751814223,
+      "loss": 5.1623,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000760522496371553,
+      "loss": 4.9782,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007590711175616835,
+      "loss": 5.3223,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007576197387518142,
+      "loss": 5.2677,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007561683599419448,
+      "loss": 5.2954,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007547169811320755,
+      "loss": 5.0404,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007532656023222061,
+      "loss": 5.2872,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007518142235123367,
+      "loss": 5.0307,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007503628447024674,
+      "loss": 5.11,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007489114658925979,
+      "loss": 4.9515,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007474600870827286,
+      "loss": 5.2947,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007460087082728592,
+      "loss": 5.1966,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007445573294629898,
+      "loss": 5.1349,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007431059506531204,
+      "loss": 5.2762,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007416545718432511,
+      "loss": 4.9943,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007402031930333816,
+      "loss": 4.7472,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007387518142235123,
+      "loss": 5.1081,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000737300435413643,
+      "loss": 4.9711,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007358490566037735,
+      "loss": 5.089,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007343976777939043,
+      "loss": 5.2889,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007329462989840348,
+      "loss": 5.0058,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007314949201741656,
+      "loss": 5.2081,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000730043541364296,
+      "loss": 5.2121,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007285921625544268,
+      "loss": 5.2403,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007271407837445574,
+      "loss": 5.5316,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000725689404934688,
+      "loss": 5.1785,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007242380261248185,
+      "loss": 5.2335,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007227866473149493,
+      "loss": 5.6821,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007213352685050797,
+      "loss": 5.2331,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007198838896952105,
+      "loss": 5.1521,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007184325108853412,
+      "loss": 5.1556,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007169811320754717,
+      "loss": 5.5863,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007155297532656024,
+      "loss": 4.9558,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000714078374455733,
+      "loss": 5.9564,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007126269956458637,
+      "loss": 5.5535,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007111756168359942,
+      "loss": 4.9877,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007097242380261249,
+      "loss": 5.4046,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007082728592162555,
+      "loss": 4.9359,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007068214804063861,
+      "loss": 5.0701,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007053701015965167,
+      "loss": 5.1551,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007039187227866474,
+      "loss": 5.1423,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000702467343976778,
+      "loss": 5.123,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007010159651669086,
+      "loss": 5.1392,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006995645863570393,
+      "loss": 5.4456,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006981132075471698,
+      "loss": 4.7891,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006966618287373005,
+      "loss": 5.2811,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006952104499274311,
+      "loss": 5.2269,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006937590711175617,
+      "loss": 4.905,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006923076923076923,
+      "loss": 5.2503,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000690856313497823,
+      "loss": 4.8439,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006894049346879536,
+      "loss": 4.9781,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006879535558780842,
+      "loss": 5.1939,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0006865021770682149,
+      "loss": 4.9898,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006850507982583455,
+      "loss": 4.9892,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006835994194484761,
+      "loss": 4.8861,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006821480406386067,
+      "loss": 4.776,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006806966618287374,
+      "loss": 5.1766,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006792452830188679,
+      "loss": 4.7434,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006777939042089986,
+      "loss": 5.311,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006763425253991292,
+      "loss": 5.593,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006748911465892598,
+      "loss": 4.9293,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006734397677793904,
+      "loss": 5.1939,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006719883889695211,
+      "loss": 4.7093,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006705370101596517,
+      "loss": 5.2601,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006690856313497823,
+      "loss": 5.468,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000667634252539913,
+      "loss": 5.3929,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006661828737300436,
+      "loss": 5.0651,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006647314949201742,
+      "loss": 5.0296,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006632801161103048,
+      "loss": 5.05,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006618287373004355,
+      "loss": 4.8995,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000660377358490566,
+      "loss": 5.1864,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006589259796806967,
+      "loss": 5.4455,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006574746008708273,
+      "loss": 5.1694,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006560232220609579,
+      "loss": 5.0285,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006545718432510886,
+      "loss": 5.1161,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006531204644412192,
+      "loss": 5.7698,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006516690856313498,
+      "loss": 5.1096,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006502177068214804,
+      "loss": 5.042,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006487663280116111,
+      "loss": 5.3352,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006473149492017417,
+      "loss": 5.6569,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006458635703918723,
+      "loss": 5.4589,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006444121915820029,
+      "loss": 6.0816,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006429608127721336,
+      "loss": 5.3323,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006415094339622641,
+      "loss": 5.0095,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006400580551523948,
+      "loss": 5.2505,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006386066763425255,
+      "loss": 5.4166,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000637155297532656,
+      "loss": 5.1596,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006357039187227867,
+      "loss": 4.9306,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006342525399129173,
+      "loss": 5.3309,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006328011611030479,
+      "loss": 5.0184,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006313497822931785,
+      "loss": 4.9704,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006298984034833092,
+      "loss": 5.0853,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006284470246734397,
+      "loss": 4.8743,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006269956458635704,
+      "loss": 5.3263,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000625544267053701,
+      "loss": 4.8698,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006240928882438317,
+      "loss": 4.9983,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006226415094339623,
+      "loss": 5.0317,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006211901306240929,
+      "loss": 4.9205,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006197387518142236,
+      "loss": 5.2878,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006182873730043541,
+      "loss": 4.9686,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006168359941944848,
+      "loss": 5.0436,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006153846153846154,
+      "loss": 4.787,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000613933236574746,
+      "loss": 4.9172,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006124818577648766,
+      "loss": 5.1994,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006110304789550073,
+      "loss": 5.0562,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006095791001451378,
+      "loss": 5.095,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006081277213352685,
+      "loss": 4.9692,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006066763425253992,
+      "loss": 5.3101,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006052249637155298,
+      "loss": 5.3721,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006037735849056604,
+      "loss": 4.857,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000602322206095791,
+      "loss": 4.9941,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0006008708272859217,
+      "loss": 5.1053,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005994194484760522,
+      "loss": 5.0145,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005979680696661829,
+      "loss": 5.1952,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005965166908563135,
+      "loss": 5.0277,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005950653120464441,
+      "loss": 5.1151,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005936139332365747,
+      "loss": 4.7625,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005921625544267054,
+      "loss": 5.0227,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005907111756168359,
+      "loss": 5.1045,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005892597968069666,
+      "loss": 5.1363,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005878084179970973,
+      "loss": 5.162,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005863570391872278,
+      "loss": 5.0185,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005849056603773585,
+      "loss": 4.9179,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005834542815674891,
+      "loss": 5.0604,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005820029027576198,
+      "loss": 5.2606,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005805515239477503,
+      "loss": 5.4544,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000579100145137881,
+      "loss": 5.3557,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005776487663280116,
+      "loss": 4.9155,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005761973875181422,
+      "loss": 5.0743,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005747460087082728,
+      "loss": 4.6857,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005732946298984035,
+      "loss": 4.8694,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005718432510885341,
+      "loss": 5.5002,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005703918722786647,
+      "loss": 5.8095,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005689404934687954,
+      "loss": 4.9876,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005674891146589259,
+      "loss": 5.0183,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005660377358490566,
+      "loss": 4.9917,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005645863570391872,
+      "loss": 4.9777,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005631349782293179,
+      "loss": 5.1874,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005616835994194484,
+      "loss": 4.8542,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005602322206095791,
+      "loss": 4.9569,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005587808417997097,
+      "loss": 4.9964,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005573294629898403,
+      "loss": 4.9433,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000555878084179971,
+      "loss": 5.003,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005544267053701016,
+      "loss": 4.8614,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005529753265602322,
+      "loss": 4.9952,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005515239477503628,
+      "loss": 5.1134,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005500725689404935,
+      "loss": 4.892,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000548621190130624,
+      "loss": 5.3991,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005471698113207547,
+      "loss": 5.0198,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005457184325108853,
+      "loss": 5.0104,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005442670537010159,
+      "loss": 4.9483,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005428156748911465,
+      "loss": 4.6876,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005413642960812772,
+      "loss": 4.7779,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000539912917271408,
+      "loss": 4.9751,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005384615384615384,
+      "loss": 4.9923,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005370101596516691,
+      "loss": 4.7491,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005355587808417997,
+      "loss": 5.0279,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005341074020319303,
+      "loss": 5.2695,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005326560232220609,
+      "loss": 4.7669,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005312046444121916,
+      "loss": 5.2179,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005297532656023221,
+      "loss": 5.0492,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005283018867924528,
+      "loss": 5.0599,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005268505079825834,
+      "loss": 4.8054,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000525399129172714,
+      "loss": 4.9806,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005239477503628448,
+      "loss": 5.1529,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005224963715529754,
+      "loss": 5.0402,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005210449927431061,
+      "loss": 4.9572,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005195936139332365,
+      "loss": 5.1013,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005181422351233673,
+      "loss": 5.0028,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005166908563134979,
+      "loss": 5.1219,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005152394775036285,
+      "loss": 4.8937,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000513788098693759,
+      "loss": 4.669,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005123367198838898,
+      "loss": 5.0989,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005108853410740202,
+      "loss": 4.8334,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000509433962264151,
+      "loss": 4.9753,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005079825834542817,
+      "loss": 5.2866,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005065312046444122,
+      "loss": 5.4865,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005050798258345429,
+      "loss": 5.1335,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005036284470246735,
+      "loss": 4.9465,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005021770682148041,
+      "loss": 4.8108,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005007256894049347,
+      "loss": 5.0448,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004992743105950654,
+      "loss": 4.6395,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000497822931785196,
+      "loss": 5.2402,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004963715529753266,
+      "loss": 5.1744,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004949201741654572,
+      "loss": 5.4538,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004934687953555878,
+      "loss": 5.0876,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004920174165457185,
+      "loss": 4.9978,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004905660377358491,
+      "loss": 5.0183,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004891146589259797,
+      "loss": 5.0703,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004876632801161103,
+      "loss": 5.0315,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004862119013062409,
+      "loss": 4.9009,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004847605224963715,
+      "loss": 4.7226,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004833091436865022,
+      "loss": 4.8724,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004818577648766328,
+      "loss": 5.0809,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00048040638606676347,
+      "loss": 4.6434,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00047895500725689407,
+      "loss": 4.9967,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00047750362844702467,
+      "loss": 4.9087,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004760522496371553,
+      "loss": 4.9569,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004746008708272859,
+      "loss": 4.9037,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004731494920174166,
+      "loss": 4.8452,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004716981132075472,
+      "loss": 4.909,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00047024673439767777,
+      "loss": 4.8391,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004687953555878084,
+      "loss": 5.071,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000467343976777939,
+      "loss": 4.5401,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004658925979680697,
+      "loss": 4.5879,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046444121915820033,
+      "loss": 4.7276,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046298984034833093,
+      "loss": 5.0536,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004615384615384616,
+      "loss": 4.8388,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004600870827285922,
+      "loss": 5.114,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004586357039187228,
+      "loss": 4.7428,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00045718432510885343,
+      "loss": 4.6002,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00045573294629898403,
+      "loss": 5.3256,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00045428156748911463,
+      "loss": 4.8627,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004528301886792453,
+      "loss": 5.0031,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004513788098693759,
+      "loss": 4.6507,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004499274310595066,
+      "loss": 5.052,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004484760522496372,
+      "loss": 4.8907,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004470246734397678,
+      "loss": 4.9239,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00044557329462989844,
+      "loss": 4.7405,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00044412191582002904,
+      "loss": 4.9512,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00044267053701015964,
+      "loss": 4.9561,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0004412191582002903,
+      "loss": 4.8431,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0004397677793904209,
+      "loss": 4.9315,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043831640058055154,
+      "loss": 5.2878,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043686502177068214,
+      "loss": 5.5832,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043541364296081274,
+      "loss": 4.8962,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043396226415094345,
+      "loss": 5.0005,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043251088534107404,
+      "loss": 4.9839,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00043105950653120464,
+      "loss": 4.8642,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0004296081277213353,
+      "loss": 5.1954,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0004281567489114659,
+      "loss": 5.1071,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00042670537010159655,
+      "loss": 4.8691,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00042525399129172715,
+      "loss": 4.8602,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00042380261248185774,
+      "loss": 4.8208,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004223512336719884,
+      "loss": 4.9217,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000420899854862119,
+      "loss": 4.9919,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00041944847605224965,
+      "loss": 4.9691,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004179970972423803,
+      "loss": 5.0758,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004165457184325109,
+      "loss": 4.8835,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00041509433962264155,
+      "loss": 5.066,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00041364296081277215,
+      "loss": 4.8244,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00041219158200290275,
+      "loss": 5.3287,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004107402031930334,
+      "loss": 4.9491,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000409288824383164,
+      "loss": 5.0155,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00040783744557329466,
+      "loss": 4.9867,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00040638606676342525,
+      "loss": 4.7942,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00040493468795355585,
+      "loss": 5.8642,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004034833091436865,
+      "loss": 4.7684,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004020319303338171,
+      "loss": 5.0727,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00040058055152394776,
+      "loss": 4.8537,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003991291727140784,
+      "loss": 4.9414,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000397677793904209,
+      "loss": 4.9024,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039622641509433966,
+      "loss": 5.0548,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039477503628447026,
+      "loss": 5.0909,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039332365747460086,
+      "loss": 4.7496,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003918722786647315,
+      "loss": 5.2255,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003904208998548621,
+      "loss": 4.8146,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003889695210449927,
+      "loss": 4.8293,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00038751814223512336,
+      "loss": 4.6352,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00038606676342525396,
+      "loss": 4.7872,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00038461538461538467,
+      "loss": 4.9817,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00038316400580551527,
+      "loss": 4.8392,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00038171262699564587,
+      "loss": 4.6469,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003802612481857765,
+      "loss": 4.7641,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003788098693759071,
+      "loss": 4.9705,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00037735849056603777,
+      "loss": 5.0269,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00037590711175616837,
+      "loss": 4.9725,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00037445573294629897,
+      "loss": 4.8754,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003730043541364296,
+      "loss": 5.0898,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003715529753265602,
+      "loss": 4.8642,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003701015965166908,
+      "loss": 4.9223,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003686502177068215,
+      "loss": 4.9794,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003671988388969521,
+      "loss": 4.9581,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003657474600870828,
+      "loss": 4.9206,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003642960812772134,
+      "loss": 4.7632,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000362844702467344,
+      "loss": 5.0981,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00036139332365747463,
+      "loss": 4.6014,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00035994194484760523,
+      "loss": 4.6544,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003584905660377358,
+      "loss": 4.983,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003570391872278665,
+      "loss": 4.8756,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003555878084179971,
+      "loss": 4.9037,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00035413642960812773,
+      "loss": 4.6167,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00035268505079825833,
+      "loss": 4.7624,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000351233671988389,
+      "loss": 4.8791,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034978229317851964,
+      "loss": 4.8372,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034833091436865023,
+      "loss": 5.0622,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034687953555878083,
+      "loss": 4.654,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003454281567489115,
+      "loss": 4.6066,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003439767779390421,
+      "loss": 4.8342,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034252539912917274,
+      "loss": 5.0489,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034107402031930334,
+      "loss": 5.1657,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00033962264150943393,
+      "loss": 4.8363,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003381712626995646,
+      "loss": 4.9493,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003367198838896952,
+      "loss": 4.7621,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00033526850507982584,
+      "loss": 4.7409,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003338171262699565,
+      "loss": 4.9177,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003323657474600871,
+      "loss": 4.619,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00033091436865021774,
+      "loss": 5.0449,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00032946298984034834,
+      "loss": 4.9689,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00032801161103047894,
+      "loss": 4.6766,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003265602322206096,
+      "loss": 5.0324,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003251088534107402,
+      "loss": 4.9344,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00032365747460087085,
+      "loss": 4.5898,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00032220609579100144,
+      "loss": 4.8141,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00032075471698113204,
+      "loss": 4.7134,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00031930333817126275,
+      "loss": 4.709,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00031785195936139335,
+      "loss": 4.8181,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00031640058055152395,
+      "loss": 4.7667,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003149492017416546,
+      "loss": 4.8282,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003134978229317852,
+      "loss": 4.8374,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00031204644412191585,
+      "loss": 4.6868,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00031059506531204645,
+      "loss": 4.778,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00030914368650217705,
+      "loss": 4.628,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003076923076923077,
+      "loss": 4.9878,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003062409288824383,
+      "loss": 5.3127,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003047895500725689,
+      "loss": 5.2205,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003033381712626996,
+      "loss": 4.6811,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003018867924528302,
+      "loss": 4.9154,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00030043541364296086,
+      "loss": 4.6451,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029898403483309146,
+      "loss": 4.8432,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029753265602322206,
+      "loss": 4.8732,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002960812772133527,
+      "loss": 4.8714,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002946298984034833,
+      "loss": 4.8009,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002931785195936139,
+      "loss": 4.7758,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029172714078374456,
+      "loss": 5.0523,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029027576197387516,
+      "loss": 4.7362,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002888243831640058,
+      "loss": 4.5611,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002873730043541364,
+      "loss": 4.8776,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028592162554426706,
+      "loss": 4.7802,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002844702467343977,
+      "loss": 5.3845,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002830188679245283,
+      "loss": 4.9866,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028156748911465897,
+      "loss": 4.7694,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028011611030478957,
+      "loss": 4.8902,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00027866473149492017,
+      "loss": 5.0456,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 5.038694858551025,
+      "eval_runtime": 799.2667,
+      "eval_samples_per_second": 3.306,
+      "eval_steps_per_second": 0.277,
+      "eval_wer": 1.9940103133677112,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002772133526850508,
+      "loss": 4.8059,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002757619738751814,
+      "loss": 4.8468,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000274310595065312,
+      "loss": 4.7108,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00027285921625544267,
+      "loss": 5.0852,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00027140783744557327,
+      "loss": 4.8766,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000269956458635704,
+      "loss": 4.9789,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002685050798258346,
+      "loss": 4.6763,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00026705370101596517,
+      "loss": 4.7252,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002656023222060958,
+      "loss": 4.7792,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002641509433962264,
+      "loss": 4.6629,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000262699564586357,
+      "loss": 4.6113,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002612481857764877,
+      "loss": 4.6525,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002597968069666183,
+      "loss": 4.6593,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002583454281567489,
+      "loss": 4.6083,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002568940493468795,
+      "loss": 4.7946,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002554426705370101,
+      "loss": 4.9219,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00025399129172714083,
+      "loss": 4.5965,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00025253991291727143,
+      "loss": 4.8996,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00025108853410740203,
+      "loss": 4.9836,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002496371552975327,
+      "loss": 4.625,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002481857764876633,
+      "loss": 4.77,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002467343976777939,
+      "loss": 4.9939,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00024528301886792453,
+      "loss": 5.0383,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00024383164005805516,
+      "loss": 4.5553,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00024238026124818576,
+      "loss": 4.7566,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002409288824383164,
+      "loss": 4.6564,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023947750362844704,
+      "loss": 4.7374,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023802612481857766,
+      "loss": 4.6311,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002365747460087083,
+      "loss": 4.6473,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023512336719883889,
+      "loss": 4.5614,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002336719883889695,
+      "loss": 4.7444,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023222060957910016,
+      "loss": 4.5839,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002307692307692308,
+      "loss": 5.014,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002293178519593614,
+      "loss": 4.5252,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00022786647314949202,
+      "loss": 4.6645,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00022641509433962264,
+      "loss": 4.6178,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002249637155297533,
+      "loss": 4.815,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002235123367198839,
+      "loss": 4.8939,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00022206095791001452,
+      "loss": 4.7946,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00022060957910014514,
+      "loss": 4.6519,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021915820029027577,
+      "loss": 4.777,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021770682148040637,
+      "loss": 4.8815,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021625544267053702,
+      "loss": 4.9162,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021480406386066765,
+      "loss": 4.7677,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021335268505079827,
+      "loss": 4.9261,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021190130624092887,
+      "loss": 4.6152,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002104499274310595,
+      "loss": 4.6767,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00020899854862119015,
+      "loss": 5.0953,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00020754716981132078,
+      "loss": 5.3704,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00020609579100145138,
+      "loss": 5.0476,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000204644412191582,
+      "loss": 4.9475,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00020319303338171263,
+      "loss": 4.652,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00020174165457184325,
+      "loss": 4.7119,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00020029027576197388,
+      "loss": 4.7762,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001988388969521045,
+      "loss": 4.8432,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019738751814223513,
+      "loss": 4.9744,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019593613933236576,
+      "loss": 4.8143,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019448476052249636,
+      "loss": 4.7354,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019303338171262698,
+      "loss": 4.8452,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019158200290275763,
+      "loss": 4.8321,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019013062409288826,
+      "loss": 4.7938,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018867924528301889,
+      "loss": 4.6422,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018722786647314948,
+      "loss": 4.9425,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001857764876632801,
+      "loss": 4.7413,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018432510885341076,
+      "loss": 4.5155,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001828737300435414,
+      "loss": 4.5478,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000181422351233672,
+      "loss": 4.718,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017997097242380261,
+      "loss": 4.6345,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017851959361393324,
+      "loss": 4.7027,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017706821480406387,
+      "loss": 4.8152,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001756168359941945,
+      "loss": 4.615,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017416545718432512,
+      "loss": 4.7067,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017271407837445574,
+      "loss": 4.3144,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017126269956458637,
+      "loss": 4.6724,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016981132075471697,
+      "loss": 4.5228,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001683599419448476,
+      "loss": 4.8212,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016690856313497825,
+      "loss": 4.9697,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016545718432510887,
+      "loss": 4.7846,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016400580551523947,
+      "loss": 4.7131,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001625544267053701,
+      "loss": 4.5873,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016110304789550072,
+      "loss": 4.8464,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015965166908563138,
+      "loss": 4.8257,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015820029027576197,
+      "loss": 4.916,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001567489114658926,
+      "loss": 4.5864,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015529753265602323,
+      "loss": 4.6401,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 4.8297,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015239477503628445,
+      "loss": 4.6124,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001509433962264151,
+      "loss": 4.8153,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014949201741654573,
+      "loss": 4.6562,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014804063860667635,
+      "loss": 4.805,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014658925979680695,
+      "loss": 4.5369,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014513788098693758,
+      "loss": 4.4222,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001436865021770682,
+      "loss": 4.6304,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014223512336719886,
+      "loss": 4.7839,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00014078374455732948,
+      "loss": 4.7115,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013933236574746008,
+      "loss": 5.1023,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001378809869375907,
+      "loss": 4.7742,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013642960812772133,
+      "loss": 5.1808,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000134978229317852,
+      "loss": 5.1768,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013352685050798259,
+      "loss": 4.6172,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001320754716981132,
+      "loss": 5.3709,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013062409288824384,
+      "loss": 4.7415,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012917271407837446,
+      "loss": 4.7487,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012772133526850506,
+      "loss": 4.6315,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012626995645863572,
+      "loss": 4.6744,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012481857764876634,
+      "loss": 4.4679,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012336719883889694,
+      "loss": 4.9871,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012191582002902758,
+      "loss": 4.7613,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001204644412191582,
+      "loss": 4.7534,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011901306240928883,
+      "loss": 4.555,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011756168359941944,
+      "loss": 4.6237,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011611030478955008,
+      "loss": 4.9051,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001146589259796807,
+      "loss": 4.67,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011320754716981132,
+      "loss": 4.3835,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011175616835994195,
+      "loss": 4.7945,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011030478955007257,
+      "loss": 4.8047,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010885341074020318,
+      "loss": 4.8833,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010740203193033382,
+      "loss": 4.5767,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010595065312046444,
+      "loss": 4.2787,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010449927431059508,
+      "loss": 4.8757,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010304789550072569,
+      "loss": 4.8329,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010159651669085631,
+      "loss": 4.9161,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010014513788098694,
+      "loss": 4.581,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.869375907111757e-05,
+      "loss": 4.6677,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.724238026124818e-05,
+      "loss": 5.0114,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.579100145137882e-05,
+      "loss": 4.5642,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433962264150944e-05,
+      "loss": 4.7539,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.288824383164006e-05,
+      "loss": 4.6364,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.14368650217707e-05,
+      "loss": 4.6833,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.998548621190131e-05,
+      "loss": 4.7156,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.853410740203193e-05,
+      "loss": 4.3974,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.708272859216256e-05,
+      "loss": 4.3917,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.563134978229318e-05,
+      "loss": 4.748,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.41799709724238e-05,
+      "loss": 4.6152,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.272859216255444e-05,
+      "loss": 4.8699,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.127721335268505e-05,
+      "loss": 4.5895,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.982583454281569e-05,
+      "loss": 4.6202,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.83744557329463e-05,
+      "loss": 4.6373,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 4.5034,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547169811320755e-05,
+      "loss": 4.5172,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402031930333818e-05,
+      "loss": 4.8678,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.256894049346879e-05,
+      "loss": 4.6158,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111756168359943e-05,
+      "loss": 4.7612,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966618287373004e-05,
+      "loss": 4.8563,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.821480406386067e-05,
+      "loss": 4.4897,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.676342525399129e-05,
+      "loss": 4.8093,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.531204644412192e-05,
+      "loss": 6.4519,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.386066763425253e-05,
+      "loss": 4.6793,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.240928882438317e-05,
+      "loss": 5.1183,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.095791001451379e-05,
+      "loss": 4.7116,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9506531204644415e-05,
+      "loss": 4.5322,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805515239477504e-05,
+      "loss": 4.3579,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 4.6153,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5152394775036286e-05,
+      "loss": 4.4632,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370101596516691e-05,
+      "loss": 4.8472,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224963715529754e-05,
+      "loss": 4.6387,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.079825834542816e-05,
+      "loss": 4.7494,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934687953555878e-05,
+      "loss": 4.5634,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.789550072568941e-05,
+      "loss": 4.3732,
+      "step": 1159
+    },
     {
       "epoch": 0.98,
-      "learning_rate": 0.000874,
-      "loss": 5.0383,
-      "step": 438
+      "learning_rate": 4.644412191582003e-05,
+      "loss": 4.6898,
+      "step": 1160
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.000876,
-      "loss": 4.9599,
-      "step": 439
+      "learning_rate": 4.4992743105950653e-05,
+      "loss": 4.6292,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.354136429608128e-05,
+      "loss": 4.971,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.20899854862119e-05,
+      "loss": 4.7939,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0638606676342524e-05,
+      "loss": 4.5,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.918722786647315e-05,
+      "loss": 4.8009,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 4.813,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6284470246734395e-05,
+      "loss": 4.4776,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483309143686502e-05,
+      "loss": 4.523,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3381712626995646e-05,
+      "loss": 4.6047,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1930333817126266e-05,
+      "loss": 4.5341,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0478955007256895e-05,
+      "loss": 4.5799,
+      "step": 1171
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.000878,
-      "loss": 4.9917,
-      "step": 440
+      "learning_rate": 2.902757619738752e-05,
+      "loss": 4.4194,
+      "step": 1172
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00088,
-      "loss": 4.7285,
-      "step": 441
+      "learning_rate": 2.7576197387518143e-05,
+      "loss": 4.6734,
+      "step": 1173
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.000882,
-      "loss": 5.0258,
-      "step": 442
+      "learning_rate": 2.612481857764877e-05,
+      "loss": 4.5512,
+      "step": 1174
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.000884,
-      "loss": 6.6035,
-      "step": 443
+      "learning_rate": 2.467343976777939e-05,
+      "loss": 4.3157,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3222060957910014e-05,
+      "loss": 4.643,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.177068214804064e-05,
+      "loss": 4.5303,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0319303338171262e-05,
+      "loss": 5.094,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 4.4242,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 4.7599,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5965166908563133e-05,
+      "loss": 4.6907,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.451378809869376e-05,
+      "loss": 4.6596,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3062409288824384e-05,
+      "loss": 4.5321,
+      "step": 1183
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0008860000000000001,
-      "loss": 5.1825,
-      "step": 444
+      "learning_rate": 1.1611030478955007e-05,
+      "loss": 4.7853,
+      "step": 1184
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.000888,
-      "loss": 5.0386,
-      "step": 445
+      "learning_rate": 1.0159651669085631e-05,
+      "loss": 4.7172,
+      "step": 1185
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0008900000000000001,
-      "loss": 4.8899,
-      "step": 446
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 4.7867,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25689404934688e-06,
+      "loss": 4.5683,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8055152394775034e-06,
+      "loss": 4.7779,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 4.843,
+      "step": 1189
     },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 4.734800568610563,
-      "train_runtime": 5167.9675,
-      "train_samples_per_second": 5.522,
-      "train_steps_per_second": 0.086
+      "train_loss": 4.993681282832704,
+      "train_runtime": 6750.5443,
+      "train_samples_per_second": 4.228,
+      "train_steps_per_second": 0.176
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,