diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 1784,
+  "global_step": 892,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,10745 +10,5375 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.9696,
+      "loss": 9.2246,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.1987,
+      "loss": 9.1373,
       "step": 2
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.113,
+      "loss": 8.8311,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 9.019,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 9.2459,
       "step": 4
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 8.5059,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 9.182,
       "step": 5
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 8.6655,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.8163,
       "step": 6
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 8.6179,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 9.091,
       "step": 7
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.8e-07,
-      "loss": 8.9312,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 8.9724,
       "step": 8
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 9.5973,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 8.9179,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 8.7553,
+      "learning_rate": 3.6e-07,
+      "loss": 9.1465,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.6e-07,
-      "loss": 8.623,
+      "learning_rate": 4.2e-07,
+      "loss": 8.807,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.2e-07,
-      "loss": 9.3041,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 8.769,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 8.7857,
+      "learning_rate": 5.4e-07,
+      "loss": 8.6194,
       "step": 13
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 5.4e-07,
-      "loss": 8.785,
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 8.9798,
       "step": 14
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 8.6603,
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 8.5752,
       "step": 15
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.599999999999999e-07,
-      "loss": 8.2813,
+      "epoch": 0.02,
+      "learning_rate": 7.2e-07,
+      "loss": 8.2835,
       "step": 16
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 7.2e-07,
-      "loss": 8.2457,
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 8.5145,
       "step": 17
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 7.799999999999999e-07,
-      "loss": 8.4808,
+      "epoch": 0.02,
+      "learning_rate": 8.4e-07,
+      "loss": 8.1527,
       "step": 18
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8.4e-07,
-      "loss": 8.2696,
+      "epoch": 0.02,
+      "learning_rate": 9e-07,
+      "loss": 8.2896,
       "step": 19
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 9e-07,
-      "loss": 8.3986,
+      "loss": 8.4004,
       "step": 20
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 9.600000000000001e-07,
-      "loss": 8.17,
+      "loss": 8.184,
       "step": 21
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.0200000000000002e-06,
-      "loss": 8.2959,
+      "loss": 8.1589,
       "step": 22
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 8.5418,
+      "epoch": 0.03,
+      "learning_rate": 1.08e-06,
+      "loss": 8.0137,
       "step": 23
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.08e-06,
-      "loss": 8.324,
+      "epoch": 0.03,
+      "learning_rate": 1.14e-06,
+      "loss": 8.0527,
       "step": 24
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.14e-06,
-      "loss": 8.0825,
+      "epoch": 0.03,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 8.1472,
       "step": 25
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 8.1673,
+      "epoch": 0.03,
+      "learning_rate": 1.26e-06,
+      "loss": 8.3813,
       "step": 26
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.26e-06,
-      "loss": 8.1061,
+      "epoch": 0.03,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 8.0726,
       "step": 27
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.3199999999999999e-06,
-      "loss": 7.933,
+      "epoch": 0.03,
+      "learning_rate": 1.38e-06,
+      "loss": 7.9304,
       "step": 28
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.38e-06,
-      "loss": 8.1098,
+      "epoch": 0.03,
+      "learning_rate": 1.44e-06,
+      "loss": 7.8539,
       "step": 29
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.44e-06,
-      "loss": 8.5941,
+      "epoch": 0.03,
+      "learning_rate": 1.5e-06,
+      "loss": 7.997,
       "step": 30
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.5e-06,
-      "loss": 8.5415,
+      "epoch": 0.03,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 7.9305,
       "step": 31
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.5599999999999999e-06,
-      "loss": 8.1589,
+      "epoch": 0.04,
+      "learning_rate": 1.62e-06,
+      "loss": 7.8678,
       "step": 32
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.62e-06,
-      "loss": 8.0684,
+      "epoch": 0.04,
+      "learning_rate": 1.68e-06,
+      "loss": 7.8208,
       "step": 33
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.68e-06,
-      "loss": 7.88,
+      "epoch": 0.04,
+      "learning_rate": 1.74e-06,
+      "loss": 7.5448,
       "step": 34
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.74e-06,
-      "loss": 7.9476,
+      "epoch": 0.04,
+      "learning_rate": 1.8e-06,
+      "loss": 7.707,
       "step": 35
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.8e-06,
-      "loss": 7.9552,
+      "epoch": 0.04,
+      "learning_rate": 1.86e-06,
+      "loss": 7.6488,
       "step": 36
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.86e-06,
-      "loss": 7.6249,
+      "epoch": 0.04,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 7.6675,
       "step": 37
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 7.917,
+      "epoch": 0.04,
+      "learning_rate": 1.98e-06,
+      "loss": 7.6531,
       "step": 38
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.98e-06,
-      "loss": 8.4142,
+      "epoch": 0.04,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 7.8233,
       "step": 39
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 7.6303,
+      "epoch": 0.04,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 7.5342,
       "step": 40
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 7.9643,
+      "epoch": 0.05,
+      "learning_rate": 2.16e-06,
+      "loss": 7.4604,
       "step": 41
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.16e-06,
-      "loss": 7.6006,
+      "epoch": 0.05,
+      "learning_rate": 2.22e-06,
+      "loss": 7.7178,
       "step": 42
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.22e-06,
-      "loss": 8.1142,
+      "epoch": 0.05,
+      "learning_rate": 2.28e-06,
+      "loss": 7.4701,
       "step": 43
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.28e-06,
-      "loss": 7.6746,
+      "epoch": 0.05,
+      "learning_rate": 2.34e-06,
+      "loss": 7.7004,
       "step": 44
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.34e-06,
-      "loss": 7.9679,
+      "epoch": 0.05,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 7.7114,
       "step": 45
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 7.7575,
+      "epoch": 0.05,
+      "learning_rate": 2.46e-06,
+      "loss": 7.7806,
       "step": 46
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.46e-06,
-      "loss": 7.6456,
+      "epoch": 0.05,
+      "learning_rate": 2.52e-06,
+      "loss": 7.5586,
       "step": 47
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.52e-06,
-      "loss": 8.0129,
+      "epoch": 0.05,
+      "learning_rate": 2.58e-06,
+      "loss": 7.6923,
       "step": 48
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.58e-06,
-      "loss": 7.7983,
+      "epoch": 0.05,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 7.7666,
       "step": 49
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.6399999999999997e-06,
-      "loss": 8.1211,
+      "epoch": 0.06,
+      "learning_rate": 2.7e-06,
+      "loss": 7.8198,
       "step": 50
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.7e-06,
-      "loss": 7.5136,
+      "epoch": 0.06,
+      "learning_rate": 2.76e-06,
+      "loss": 7.3712,
       "step": 51
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.76e-06,
-      "loss": 7.5942,
+      "epoch": 0.06,
+      "learning_rate": 2.82e-06,
+      "loss": 7.3484,
       "step": 52
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.82e-06,
-      "loss": 7.3651,
+      "epoch": 0.06,
+      "learning_rate": 2.88e-06,
+      "loss": 7.2667,
       "step": 53
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.88e-06,
-      "loss": 7.4696,
+      "epoch": 0.06,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 7.4204,
       "step": 54
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 7.722,
+      "epoch": 0.06,
+      "learning_rate": 3e-06,
+      "loss": 7.2713,
       "step": 55
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3e-06,
-      "loss": 7.4008,
+      "epoch": 0.06,
+      "learning_rate": 3.06e-06,
+      "loss": 7.3576,
       "step": 56
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.06e-06,
-      "loss": 7.3648,
+      "epoch": 0.06,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 7.3774,
       "step": 57
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.1199999999999998e-06,
-      "loss": 7.2801,
+      "epoch": 0.07,
+      "learning_rate": 3.18e-06,
+      "loss": 7.4335,
       "step": 58
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.18e-06,
-      "loss": 7.2882,
+      "epoch": 0.07,
+      "learning_rate": 3.24e-06,
+      "loss": 7.2461,
       "step": 59
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.24e-06,
-      "loss": 7.3391,
+      "epoch": 0.07,
+      "learning_rate": 3.3e-06,
+      "loss": 7.3461,
       "step": 60
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.3e-06,
-      "loss": 7.4576,
+      "epoch": 0.07,
+      "learning_rate": 3.36e-06,
+      "loss": 7.3595,
       "step": 61
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.36e-06,
-      "loss": 7.408,
+      "epoch": 0.07,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 7.2852,
       "step": 62
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 7.4581,
+      "epoch": 0.07,
+      "learning_rate": 3.48e-06,
+      "loss": 7.2491,
       "step": 63
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.48e-06,
-      "loss": 7.307,
+      "epoch": 0.07,
+      "learning_rate": 3.54e-06,
+      "loss": 7.3441,
       "step": 64
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.54e-06,
-      "loss": 7.2935,
+      "epoch": 0.07,
+      "learning_rate": 3.6e-06,
+      "loss": 7.1761,
       "step": 65
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.6e-06,
-      "loss": 7.2623,
+      "epoch": 0.07,
+      "learning_rate": 3.66e-06,
+      "loss": 7.4232,
       "step": 66
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.66e-06,
-      "loss": 7.2983,
+      "epoch": 0.08,
+      "learning_rate": 3.72e-06,
+      "loss": 7.2364,
       "step": 67
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.72e-06,
-      "loss": 7.2994,
+      "epoch": 0.08,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 7.1723,
       "step": 68
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 7.3338,
+      "epoch": 0.08,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 7.2477,
       "step": 69
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 7.1561,
+      "epoch": 0.08,
+      "learning_rate": 3.9e-06,
+      "loss": 7.1895,
       "step": 70
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.9e-06,
-      "loss": 7.2455,
+      "epoch": 0.08,
+      "learning_rate": 3.96e-06,
+      "loss": 7.3421,
       "step": 71
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.96e-06,
-      "loss": 7.2355,
+      "epoch": 0.08,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 7.3882,
       "step": 72
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 7.3044,
+      "epoch": 0.08,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 7.1381,
       "step": 73
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 7.1498,
+      "epoch": 0.08,
+      "learning_rate": 4.14e-06,
+      "loss": 7.3217,
       "step": 74
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.14e-06,
-      "loss": 7.4359,
+      "epoch": 0.08,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 7.1776,
       "step": 75
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 7.2435,
+      "epoch": 0.09,
+      "learning_rate": 4.26e-06,
+      "loss": 7.2761,
       "step": 76
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.26e-06,
-      "loss": 7.3642,
+      "epoch": 0.09,
+      "learning_rate": 4.32e-06,
+      "loss": 7.167,
       "step": 77
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.32e-06,
-      "loss": 7.1768,
+      "epoch": 0.09,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 7.1506,
       "step": 78
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.3799999999999996e-06,
-      "loss": 7.1738,
+      "epoch": 0.09,
+      "learning_rate": 4.44e-06,
+      "loss": 7.1386,
       "step": 79
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.44e-06,
-      "loss": 7.1302,
+      "epoch": 0.09,
+      "learning_rate": 4.5e-06,
+      "loss": 7.1926,
       "step": 80
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.5e-06,
-      "loss": 7.315,
+      "epoch": 0.09,
+      "learning_rate": 4.56e-06,
+      "loss": 7.2166,
       "step": 81
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.56e-06,
-      "loss": 7.4078,
+      "epoch": 0.09,
+      "learning_rate": 4.62e-06,
+      "loss": 7.1547,
       "step": 82
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.62e-06,
-      "loss": 7.0844,
+      "epoch": 0.09,
+      "learning_rate": 4.68e-06,
+      "loss": 7.322,
       "step": 83
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.68e-06,
-      "loss": 7.1778,
+      "epoch": 0.09,
+      "learning_rate": 4.74e-06,
+      "loss": 7.094,
       "step": 84
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.74e-06,
-      "loss": 7.2702,
+      "epoch": 0.1,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 7.1213,
       "step": 85
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 7.2671,
+      "epoch": 0.1,
+      "learning_rate": 4.86e-06,
+      "loss": 7.2003,
       "step": 86
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.86e-06,
-      "loss": 7.3253,
+      "epoch": 0.1,
+      "learning_rate": 4.92e-06,
+      "loss": 7.1351,
       "step": 87
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.92e-06,
-      "loss": 7.2238,
+      "epoch": 0.1,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 7.1462,
       "step": 88
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 7.2923,
+      "epoch": 0.1,
+      "learning_rate": 5.04e-06,
+      "loss": 7.1968,
       "step": 89
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.04e-06,
-      "loss": 7.0365,
+      "epoch": 0.1,
+      "learning_rate": 5.1e-06,
+      "loss": 7.2494,
       "step": 90
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.1e-06,
-      "loss": 7.3465,
+      "epoch": 0.1,
+      "learning_rate": 5.16e-06,
+      "loss": 7.3753,
       "step": 91
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.16e-06,
-      "loss": 7.0122,
+      "epoch": 0.1,
+      "learning_rate": 5.22e-06,
+      "loss": 7.2578,
       "step": 92
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.22e-06,
-      "loss": 7.1114,
+      "epoch": 0.1,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 7.0413,
       "step": 93
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.279999999999999e-06,
-      "loss": 7.4159,
+      "epoch": 0.11,
+      "learning_rate": 5.34e-06,
+      "loss": 7.1463,
       "step": 94
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.34e-06,
-      "loss": 7.369,
+      "epoch": 0.11,
+      "learning_rate": 5.4e-06,
+      "loss": 7.2888,
       "step": 95
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.4e-06,
-      "loss": 7.1914,
+      "epoch": 0.11,
+      "learning_rate": 5.46e-06,
+      "loss": 7.3631,
       "step": 96
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.46e-06,
-      "loss": 7.1165,
+      "epoch": 0.11,
+      "learning_rate": 5.52e-06,
+      "loss": 7.1825,
       "step": 97
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 5.52e-06,
-      "loss": 6.9421,
+      "epoch": 0.11,
+      "learning_rate": 5.58e-06,
+      "loss": 7.2189,
       "step": 98
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.58e-06,
-      "loss": 7.2359,
+      "epoch": 0.11,
+      "learning_rate": 5.64e-06,
+      "loss": 6.9729,
       "step": 99
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.64e-06,
-      "loss": 7.1402,
+      "epoch": 0.11,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 6.7578,
       "step": 100
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 7.2906,
+      "epoch": 0.11,
+      "learning_rate": 5.76e-06,
+      "loss": 7.2918,
       "step": 101
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.76e-06,
-      "loss": 7.2212,
+      "epoch": 0.11,
+      "learning_rate": 5.82e-06,
+      "loss": 7.3083,
       "step": 102
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.82e-06,
-      "loss": 7.2242,
+      "epoch": 0.12,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 7.2555,
       "step": 103
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 7.178,
+      "epoch": 0.12,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 7.1407,
       "step": 104
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 7.1151,
+      "epoch": 0.12,
+      "learning_rate": 6e-06,
+      "loss": 7.0634,
       "step": 105
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6e-06,
-      "loss": 7.053,
+      "epoch": 0.12,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 7.0999,
       "step": 106
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 7.5263,
+      "epoch": 0.12,
+      "learning_rate": 6.12e-06,
+      "loss": 7.2116,
       "step": 107
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.12e-06,
-      "loss": 7.1922,
+      "epoch": 0.12,
+      "learning_rate": 6.18e-06,
+      "loss": 7.2409,
       "step": 108
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.18e-06,
-      "loss": 7.0048,
+      "epoch": 0.12,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 6.9523,
       "step": 109
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.2399999999999995e-06,
-      "loss": 7.2121,
+      "epoch": 0.12,
+      "learning_rate": 6.3e-06,
+      "loss": 7.0838,
       "step": 110
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.3e-06,
-      "loss": 7.0303,
+      "epoch": 0.12,
+      "learning_rate": 6.36e-06,
+      "loss": 7.1254,
       "step": 111
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.36e-06,
-      "loss": 6.9278,
+      "epoch": 0.13,
+      "learning_rate": 6.42e-06,
+      "loss": 7.2566,
       "step": 112
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.42e-06,
-      "loss": 7.1768,
+      "epoch": 0.13,
+      "learning_rate": 6.48e-06,
+      "loss": 7.1117,
       "step": 113
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.48e-06,
-      "loss": 7.2352,
+      "epoch": 0.13,
+      "learning_rate": 6.54e-06,
+      "loss": 7.0778,
       "step": 114
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 6.54e-06,
-      "loss": 7.0644,
+      "epoch": 0.13,
+      "learning_rate": 6.6e-06,
+      "loss": 7.0946,
       "step": 115
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.6e-06,
-      "loss": 7.1474,
+      "epoch": 0.13,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 6.9887,
       "step": 116
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 7.2392,
+      "epoch": 0.13,
+      "learning_rate": 6.72e-06,
+      "loss": 7.1303,
       "step": 117
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.72e-06,
-      "loss": 7.0089,
+      "epoch": 0.13,
+      "learning_rate": 6.78e-06,
+      "loss": 7.0789,
       "step": 118
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.78e-06,
-      "loss": 7.1029,
+      "epoch": 0.13,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 7.0958,
       "step": 119
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 7.1292,
+      "epoch": 0.13,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 7.1212,
       "step": 120
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 7.2982,
+      "epoch": 0.14,
+      "learning_rate": 6.96e-06,
+      "loss": 7.033,
       "step": 121
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.96e-06,
-      "loss": 7.2174,
+      "epoch": 0.14,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 7.1919,
       "step": 122
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 7.2169,
+      "epoch": 0.14,
+      "learning_rate": 7.08e-06,
+      "loss": 6.9572,
       "step": 123
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.08e-06,
-      "loss": 7.2,
+      "epoch": 0.14,
+      "learning_rate": 7.14e-06,
+      "loss": 7.0968,
       "step": 124
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.14e-06,
-      "loss": 6.9128,
+      "epoch": 0.14,
+      "learning_rate": 7.2e-06,
+      "loss": 7.1623,
       "step": 125
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.2e-06,
-      "loss": 7.3755,
+      "epoch": 0.14,
+      "learning_rate": 7.26e-06,
+      "loss": 7.235,
       "step": 126
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.26e-06,
-      "loss": 7.0845,
+      "epoch": 0.14,
+      "learning_rate": 7.32e-06,
+      "loss": 7.2622,
       "step": 127
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.32e-06,
-      "loss": 7.0995,
+      "epoch": 0.14,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 7.1798,
       "step": 128
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 7.0067,
+      "epoch": 0.14,
+      "learning_rate": 7.44e-06,
+      "loss": 7.0186,
       "step": 129
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.44e-06,
-      "loss": 7.0221,
+      "epoch": 0.15,
+      "learning_rate": 7.5e-06,
+      "loss": 7.1465,
       "step": 130
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.5e-06,
-      "loss": 7.3019,
+      "epoch": 0.15,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 7.0823,
       "step": 131
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 7.1201,
+      "epoch": 0.15,
+      "learning_rate": 7.62e-06,
+      "loss": 7.1764,
       "step": 132
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 7.62e-06,
-      "loss": 7.2396,
+      "epoch": 0.15,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 7.1002,
       "step": 133
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 7.0189,
+      "epoch": 0.15,
+      "learning_rate": 7.74e-06,
+      "loss": 7.2143,
       "step": 134
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.74e-06,
-      "loss": 6.9969,
+      "epoch": 0.15,
+      "learning_rate": 7.8e-06,
+      "loss": 6.9495,
       "step": 135
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.8e-06,
-      "loss": 7.2609,
+      "epoch": 0.15,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 7.0104,
       "step": 136
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 7.1427,
+      "epoch": 0.15,
+      "learning_rate": 7.92e-06,
+      "loss": 7.2685,
       "step": 137
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.92e-06,
-      "loss": 7.0829,
+      "epoch": 0.15,
+      "learning_rate": 7.98e-06,
+      "loss": 7.1034,
       "step": 138
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.98e-06,
-      "loss": 7.2315,
+      "epoch": 0.16,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 7.0242,
       "step": 139
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 7.1726,
+      "epoch": 0.16,
+      "learning_rate": 8.1e-06,
+      "loss": 7.0288,
       "step": 140
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.1e-06,
-      "loss": 7.3415,
+      "epoch": 0.16,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 7.2486,
       "step": 141
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 7.2772,
+      "epoch": 0.16,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 7.2074,
       "step": 142
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 6.9898,
+      "epoch": 0.16,
+      "learning_rate": 8.28e-06,
+      "loss": 7.1155,
       "step": 143
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.28e-06,
-      "loss": 7.0998,
+      "epoch": 0.16,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 6.9432,
       "step": 144
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.340000000000001e-06,
-      "loss": 7.254,
+      "epoch": 0.16,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 6.9311,
       "step": 145
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 7.1754,
+      "epoch": 0.16,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 6.8801,
       "step": 146
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.459999999999999e-06,
-      "loss": 7.1597,
+      "epoch": 0.16,
+      "learning_rate": 8.52e-06,
+      "loss": 7.1537,
       "step": 147
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.52e-06,
-      "loss": 7.1274,
+      "epoch": 0.17,
+      "learning_rate": 8.58e-06,
+      "loss": 6.6065,
       "step": 148
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.58e-06,
-      "loss": 7.0953,
+      "epoch": 0.17,
+      "learning_rate": 8.64e-06,
+      "loss": 6.9624,
       "step": 149
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.64e-06,
-      "loss": 6.3344,
+      "epoch": 0.17,
+      "learning_rate": 8.7e-06,
+      "loss": 6.4166,
       "step": 150
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 8.7e-06,
-      "loss": 7.0804,
+      "epoch": 0.17,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 7.2044,
       "step": 151
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.759999999999999e-06,
-      "loss": 7.0869,
+      "epoch": 0.17,
+      "learning_rate": 8.82e-06,
+      "loss": 7.1448,
       "step": 152
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.82e-06,
-      "loss": 6.9963,
+      "epoch": 0.17,
+      "learning_rate": 8.88e-06,
+      "loss": 7.1486,
       "step": 153
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.88e-06,
-      "loss": 7.3103,
+      "epoch": 0.17,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 6.9927,
       "step": 154
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 6.919,
+      "epoch": 0.17,
+      "learning_rate": 9e-06,
+      "loss": 7.0003,
       "step": 155
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9e-06,
-      "loss": 7.2369,
+      "epoch": 0.17,
+      "learning_rate": 9.06e-06,
+      "loss": 7.0689,
       "step": 156
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.06e-06,
-      "loss": 7.0502,
+      "epoch": 0.18,
+      "learning_rate": 9.12e-06,
+      "loss": 7.2571,
       "step": 157
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.12e-06,
-      "loss": 7.058,
+      "epoch": 0.18,
+      "learning_rate": 9.18e-06,
+      "loss": 7.2219,
       "step": 158
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.18e-06,
-      "loss": 7.0676,
+      "epoch": 0.18,
+      "learning_rate": 9.24e-06,
+      "loss": 7.1603,
       "step": 159
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.24e-06,
-      "loss": 7.2141,
+      "epoch": 0.18,
+      "learning_rate": 9.3e-06,
+      "loss": 7.1569,
       "step": 160
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.3e-06,
-      "loss": 7.1734,
+      "epoch": 0.18,
+      "learning_rate": 9.36e-06,
+      "loss": 7.0327,
       "step": 161
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.36e-06,
-      "loss": 7.3227,
+      "epoch": 0.18,
+      "learning_rate": 9.42e-06,
+      "loss": 7.0264,
       "step": 162
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.42e-06,
-      "loss": 7.1154,
+      "epoch": 0.18,
+      "learning_rate": 9.48e-06,
+      "loss": 7.0862,
       "step": 163
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.48e-06,
-      "loss": 7.1405,
+      "epoch": 0.18,
+      "learning_rate": 9.54e-06,
+      "loss": 7.0314,
       "step": 164
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.54e-06,
-      "loss": 7.1201,
+      "epoch": 0.18,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 7.0722,
       "step": 165
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 7.1972,
+      "epoch": 0.19,
+      "learning_rate": 9.66e-06,
+      "loss": 7.0334,
       "step": 166
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.66e-06,
-      "loss": 7.1606,
+      "epoch": 0.19,
+      "learning_rate": 9.72e-06,
+      "loss": 6.9734,
       "step": 167
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.72e-06,
-      "loss": 6.9433,
+      "epoch": 0.19,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 7.0998,
       "step": 168
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 6.8733,
+      "epoch": 0.19,
+      "learning_rate": 9.84e-06,
+      "loss": 7.1167,
       "step": 169
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.84e-06,
-      "loss": 7.221,
+      "epoch": 0.19,
+      "learning_rate": 9.9e-06,
+      "loss": 7.0092,
       "step": 170
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.9e-06,
-      "loss": 7.1718,
+      "epoch": 0.19,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 7.202,
       "step": 171
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 7.074,
+      "epoch": 0.19,
+      "learning_rate": 1.002e-05,
+      "loss": 7.1753,
       "step": 172
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.002e-05,
-      "loss": 7.0683,
+      "epoch": 0.19,
+      "learning_rate": 1.008e-05,
+      "loss": 7.19,
       "step": 173
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.008e-05,
-      "loss": 7.0099,
+      "epoch": 0.2,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 7.0812,
       "step": 174
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 7.1668,
+      "epoch": 0.2,
+      "learning_rate": 1.02e-05,
+      "loss": 6.9601,
       "step": 175
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.02e-05,
-      "loss": 6.9405,
+      "epoch": 0.2,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 7.0665,
       "step": 176
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.0260000000000002e-05,
-      "loss": 6.9621,
+      "epoch": 0.2,
+      "learning_rate": 1.032e-05,
+      "loss": 7.0419,
       "step": 177
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.032e-05,
-      "loss": 7.0455,
+      "epoch": 0.2,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 6.9903,
       "step": 178
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.0379999999999999e-05,
-      "loss": 6.9033,
+      "epoch": 0.2,
+      "learning_rate": 1.044e-05,
+      "loss": 7.1019,
       "step": 179
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.044e-05,
-      "loss": 7.108,
+      "epoch": 0.2,
+      "learning_rate": 1.05e-05,
+      "loss": 7.1396,
       "step": 180
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.05e-05,
-      "loss": 6.9378,
+      "epoch": 0.2,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 7.0924,
       "step": 181
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.0559999999999999e-05,
-      "loss": 6.9327,
+      "epoch": 0.2,
+      "learning_rate": 1.062e-05,
+      "loss": 6.9487,
       "step": 182
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.062e-05,
-      "loss": 7.1707,
+      "epoch": 0.21,
+      "learning_rate": 1.068e-05,
+      "loss": 7.0349,
       "step": 183
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.068e-05,
-      "loss": 6.8486,
+      "epoch": 0.21,
+      "learning_rate": 1.074e-05,
+      "loss": 7.1434,
       "step": 184
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.074e-05,
-      "loss": 7.0608,
+      "epoch": 0.21,
+      "learning_rate": 1.08e-05,
+      "loss": 7.0692,
       "step": 185
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.08e-05,
-      "loss": 6.9868,
+      "epoch": 0.21,
+      "learning_rate": 1.086e-05,
+      "loss": 7.1847,
       "step": 186
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.086e-05,
-      "loss": 6.9092,
+      "epoch": 0.21,
+      "learning_rate": 1.092e-05,
+      "loss": 7.1237,
       "step": 187
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.092e-05,
-      "loss": 7.0168,
+      "epoch": 0.21,
+      "learning_rate": 1.098e-05,
+      "loss": 6.9167,
       "step": 188
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.098e-05,
-      "loss": 6.9804,
+      "epoch": 0.21,
+      "learning_rate": 1.104e-05,
+      "loss": 6.9805,
       "step": 189
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.104e-05,
-      "loss": 6.9715,
+      "epoch": 0.21,
+      "learning_rate": 1.11e-05,
+      "loss": 7.0209,
       "step": 190
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.11e-05,
-      "loss": 7.1436,
+      "epoch": 0.21,
+      "learning_rate": 1.116e-05,
+      "loss": 7.0294,
       "step": 191
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.116e-05,
-      "loss": 6.8564,
+      "epoch": 0.22,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 7.1835,
       "step": 192
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1220000000000001e-05,
-      "loss": 6.8947,
+      "epoch": 0.22,
+      "learning_rate": 1.128e-05,
+      "loss": 7.1852,
       "step": 193
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.128e-05,
-      "loss": 7.0262,
+      "epoch": 0.22,
+      "learning_rate": 1.134e-05,
+      "loss": 7.0206,
       "step": 194
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.134e-05,
-      "loss": 7.0312,
+      "epoch": 0.22,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 6.768,
       "step": 195
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 6.9948,
+      "epoch": 0.22,
+      "learning_rate": 1.146e-05,
+      "loss": 7.1724,
       "step": 196
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.146e-05,
-      "loss": 6.9514,
+      "epoch": 0.22,
+      "learning_rate": 1.152e-05,
+      "loss": 6.9272,
       "step": 197
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.152e-05,
-      "loss": 6.7027,
+      "epoch": 0.22,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 6.8322,
       "step": 198
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 6.6922,
+      "epoch": 0.22,
+      "learning_rate": 1.164e-05,
+      "loss": 6.7014,
       "step": 199
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.164e-05,
-      "loss": 6.4467,
+      "epoch": 0.22,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 6.4995,
       "step": 200
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 7.2816,
+      "epoch": 0.23,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 7.1439,
       "step": 201
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 7.0393,
+      "epoch": 0.23,
+      "learning_rate": 1.182e-05,
+      "loss": 7.1787,
       "step": 202
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.182e-05,
-      "loss": 7.3588,
+      "epoch": 0.23,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 7.051,
       "step": 203
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 6.956,
+      "epoch": 0.23,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 7.0691,
       "step": 204
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 6.9184,
+      "epoch": 0.23,
+      "learning_rate": 1.2e-05,
+      "loss": 7.0265,
       "step": 205
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2e-05,
-      "loss": 7.0812,
+      "epoch": 0.23,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 7.01,
       "step": 206
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2060000000000001e-05,
-      "loss": 7.0877,
+      "epoch": 0.23,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 7.0522,
       "step": 207
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 6.9665,
+      "epoch": 0.23,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 6.9969,
       "step": 208
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 7.062,
+      "epoch": 0.23,
+      "learning_rate": 1.224e-05,
+      "loss": 6.9861,
       "step": 209
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.224e-05,
-      "loss": 7.2305,
+      "epoch": 0.24,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 6.9958,
       "step": 210
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2299999999999999e-05,
-      "loss": 7.0124,
+      "epoch": 0.24,
+      "learning_rate": 1.236e-05,
+      "loss": 6.8662,
       "step": 211
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.236e-05,
-      "loss": 7.1488,
+      "epoch": 0.24,
+      "learning_rate": 1.242e-05,
+      "loss": 6.9427,
       "step": 212
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.242e-05,
-      "loss": 7.0427,
+      "epoch": 0.24,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 7.0203,
       "step": 213
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2479999999999999e-05,
-      "loss": 6.8682,
+      "epoch": 0.24,
+      "learning_rate": 1.254e-05,
+      "loss": 7.0491,
       "step": 214
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.254e-05,
-      "loss": 7.1163,
+      "epoch": 0.24,
+      "learning_rate": 1.26e-05,
+      "loss": 7.0124,
       "step": 215
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.26e-05,
-      "loss": 6.887,
+      "epoch": 0.24,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 7.1293,
       "step": 216
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 6.9964,
+      "epoch": 0.24,
+      "learning_rate": 1.272e-05,
+      "loss": 6.9887,
       "step": 217
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.272e-05,
-      "loss": 6.922,
+      "epoch": 0.24,
+      "learning_rate": 1.278e-05,
+      "loss": 7.0182,
       "step": 218
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.278e-05,
-      "loss": 6.9778,
+      "epoch": 0.25,
+      "learning_rate": 1.284e-05,
+      "loss": 6.9398,
       "step": 219
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.284e-05,
-      "loss": 7.0791,
+      "epoch": 0.25,
+      "learning_rate": 1.29e-05,
+      "loss": 7.084,
       "step": 220
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.29e-05,
-      "loss": 7.1201,
+      "epoch": 0.25,
+      "learning_rate": 1.296e-05,
+      "loss": 7.0237,
       "step": 221
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.296e-05,
-      "loss": 7.1932,
+      "epoch": 0.25,
+      "learning_rate": 1.302e-05,
+      "loss": 6.9158,
       "step": 222
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.302e-05,
-      "loss": 6.9148,
+      "epoch": 0.25,
+      "learning_rate": 1.308e-05,
+      "loss": 7.0453,
       "step": 223
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.308e-05,
-      "loss": 7.0437,
+      "epoch": 0.25,
+      "learning_rate": 1.314e-05,
+      "loss": 6.9951,
       "step": 224
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.314e-05,
-      "loss": 7.0932,
+      "epoch": 0.25,
+      "learning_rate": 1.32e-05,
+      "loss": 6.9467,
       "step": 225
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.32e-05,
-      "loss": 7.3096,
+      "epoch": 0.25,
+      "learning_rate": 1.326e-05,
+      "loss": 7.1874,
       "step": 226
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.326e-05,
-      "loss": 7.2123,
+      "epoch": 0.25,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 6.9712,
       "step": 227
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 7.1423,
+      "epoch": 0.26,
+      "learning_rate": 1.338e-05,
+      "loss": 6.9711,
       "step": 228
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.338e-05,
-      "loss": 6.9594,
+      "epoch": 0.26,
+      "learning_rate": 1.344e-05,
+      "loss": 7.0803,
       "step": 229
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.344e-05,
-      "loss": 6.9927,
+      "epoch": 0.26,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 7.0759,
       "step": 230
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 7.0016,
+      "epoch": 0.26,
+      "learning_rate": 1.356e-05,
+      "loss": 6.9329,
       "step": 231
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.356e-05,
-      "loss": 7.1276,
+      "epoch": 0.26,
+      "learning_rate": 1.362e-05,
+      "loss": 7.1158,
       "step": 232
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.362e-05,
-      "loss": 7.2163,
+      "epoch": 0.26,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 7.0195,
       "step": 233
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 7.0341,
+      "epoch": 0.26,
+      "learning_rate": 1.374e-05,
+      "loss": 7.026,
       "step": 234
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.374e-05,
-      "loss": 6.8862,
+      "epoch": 0.26,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 7.0963,
       "step": 235
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 7.2859,
+      "epoch": 0.26,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 6.8661,
       "step": 236
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 7.0918,
+      "epoch": 0.27,
+      "learning_rate": 1.392e-05,
+      "loss": 6.9358,
       "step": 237
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.392e-05,
-      "loss": 7.0758,
+      "epoch": 0.27,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 7.1396,
       "step": 238
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 6.9922,
+      "epoch": 0.27,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 6.9051,
       "step": 239
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 7.0455,
+      "epoch": 0.27,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 6.965,
       "step": 240
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 7.0353,
+      "epoch": 0.27,
+      "learning_rate": 1.416e-05,
+      "loss": 6.9123,
       "step": 241
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.416e-05,
-      "loss": 7.0058,
+      "epoch": 0.27,
+      "learning_rate": 1.422e-05,
+      "loss": 7.0129,
       "step": 242
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.422e-05,
-      "loss": 6.8786,
+      "epoch": 0.27,
+      "learning_rate": 1.428e-05,
+      "loss": 6.884,
       "step": 243
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.428e-05,
-      "loss": 6.9992,
+      "epoch": 0.27,
+      "learning_rate": 1.434e-05,
+      "loss": 6.798,
       "step": 244
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.434e-05,
-      "loss": 6.9791,
+      "epoch": 0.27,
+      "learning_rate": 1.44e-05,
+      "loss": 6.927,
       "step": 245
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.44e-05,
-      "loss": 7.0461,
+      "epoch": 0.28,
+      "learning_rate": 1.446e-05,
+      "loss": 6.906,
       "step": 246
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.446e-05,
-      "loss": 6.5529,
+      "epoch": 0.28,
+      "learning_rate": 1.452e-05,
+      "loss": 6.9235,
       "step": 247
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.452e-05,
-      "loss": 6.886,
+      "epoch": 0.28,
+      "learning_rate": 1.458e-05,
+      "loss": 7.0107,
       "step": 248
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.458e-05,
-      "loss": 6.6048,
+      "epoch": 0.28,
+      "learning_rate": 1.464e-05,
+      "loss": 6.7736,
       "step": 249
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.464e-05,
-      "loss": 6.4621,
+      "epoch": 0.28,
+      "learning_rate": 1.47e-05,
+      "loss": 6.5131,
       "step": 250
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.47e-05,
-      "loss": 7.1444,
+      "epoch": 0.28,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 7.2408,
       "step": 251
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 7.4397,
+      "epoch": 0.28,
+      "learning_rate": 1.482e-05,
+      "loss": 7.1705,
       "step": 252
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.482e-05,
-      "loss": 7.1097,
+      "epoch": 0.28,
+      "learning_rate": 1.488e-05,
+      "loss": 7.0152,
       "step": 253
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.488e-05,
-      "loss": 7.2166,
+      "epoch": 0.28,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 7.0892,
       "step": 254
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.4940000000000001e-05,
-      "loss": 7.0941,
+      "epoch": 0.29,
+      "learning_rate": 1.5e-05,
+      "loss": 6.9197,
       "step": 255
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.5e-05,
-      "loss": 7.092,
+      "epoch": 0.29,
+      "learning_rate": 1.506e-05,
+      "loss": 7.091,
       "step": 256
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.506e-05,
-      "loss": 7.0843,
+      "epoch": 0.29,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 7.0727,
       "step": 257
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.5120000000000001e-05,
-      "loss": 7.0362,
+      "epoch": 0.29,
+      "learning_rate": 1.518e-05,
+      "loss": 6.9958,
       "step": 258
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.518e-05,
-      "loss": 7.2971,
+      "epoch": 0.29,
+      "learning_rate": 1.524e-05,
+      "loss": 6.9301,
       "step": 259
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.524e-05,
-      "loss": 6.7802,
+      "epoch": 0.29,
+      "learning_rate": 1.53e-05,
+      "loss": 7.0663,
       "step": 260
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.53e-05,
-      "loss": 6.9262,
+      "epoch": 0.29,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 6.9489,
       "step": 261
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.5360000000000002e-05,
-      "loss": 6.949,
+      "epoch": 0.29,
+      "learning_rate": 1.542e-05,
+      "loss": 6.8822,
       "step": 262
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.542e-05,
-      "loss": 7.2118,
+      "epoch": 0.29,
+      "learning_rate": 1.548e-05,
+      "loss": 7.0065,
       "step": 263
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.548e-05,
-      "loss": 7.1258,
+      "epoch": 0.3,
+      "learning_rate": 1.554e-05,
+      "loss": 6.9719,
       "step": 264
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.554e-05,
-      "loss": 6.9676,
+      "epoch": 0.3,
+      "learning_rate": 1.56e-05,
+      "loss": 6.9191,
       "step": 265
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.56e-05,
-      "loss": 6.8564,
+      "epoch": 0.3,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 7.0878,
       "step": 266
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.5660000000000003e-05,
-      "loss": 7.1026,
+      "epoch": 0.3,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 7.0698,
       "step": 267
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.5720000000000002e-05,
-      "loss": 7.0552,
+      "epoch": 0.3,
+      "learning_rate": 1.578e-05,
+      "loss": 7.0582,
       "step": 268
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.578e-05,
-      "loss": 7.1273,
+      "epoch": 0.3,
+      "learning_rate": 1.584e-05,
+      "loss": 6.9094,
       "step": 269
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.584e-05,
-      "loss": 6.8998,
+      "epoch": 0.3,
+      "learning_rate": 1.59e-05,
+      "loss": 7.0062,
       "step": 270
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.59e-05,
-      "loss": 6.9583,
+      "epoch": 0.3,
+      "learning_rate": 1.596e-05,
+      "loss": 6.8624,
       "step": 271
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.596e-05,
-      "loss": 6.9907,
+      "epoch": 0.3,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 6.8755,
       "step": 272
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.6020000000000002e-05,
-      "loss": 6.8759,
+      "epoch": 0.31,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 6.923,
       "step": 273
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.6080000000000002e-05,
-      "loss": 6.9947,
+      "epoch": 0.31,
+      "learning_rate": 1.614e-05,
+      "loss": 6.7935,
       "step": 274
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.614e-05,
-      "loss": 7.094,
+      "epoch": 0.31,
+      "learning_rate": 1.62e-05,
+      "loss": 7.053,
       "step": 275
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.62e-05,
-      "loss": 7.1495,
+      "epoch": 0.31,
+      "learning_rate": 1.626e-05,
+      "loss": 6.9004,
       "step": 276
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.626e-05,
-      "loss": 7.0349,
+      "epoch": 0.31,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 7.0005,
       "step": 277
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6320000000000003e-05,
-      "loss": 7.1515,
+      "epoch": 0.31,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 6.9169,
       "step": 278
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6380000000000002e-05,
-      "loss": 6.9614,
+      "epoch": 0.31,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 6.9615,
       "step": 279
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6440000000000002e-05,
-      "loss": 6.8907,
+      "epoch": 0.31,
+      "learning_rate": 1.65e-05,
+      "loss": 7.0077,
       "step": 280
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.65e-05,
-      "loss": 7.0096,
+      "epoch": 0.32,
+      "learning_rate": 1.656e-05,
+      "loss": 6.8053,
       "step": 281
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.656e-05,
-      "loss": 7.1756,
+      "epoch": 0.32,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 7.0004,
       "step": 282
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6620000000000004e-05,
-      "loss": 7.1094,
+      "epoch": 0.32,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 6.7168,
       "step": 283
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6680000000000003e-05,
-      "loss": 6.7193,
+      "epoch": 0.32,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 7.0215,
       "step": 284
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6740000000000002e-05,
-      "loss": 7.1333,
+      "epoch": 0.32,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 6.9857,
       "step": 285
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 6.811,
+      "epoch": 0.32,
+      "learning_rate": 1.686e-05,
+      "loss": 6.7984,
       "step": 286
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.686e-05,
-      "loss": 6.9068,
+      "epoch": 0.32,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 7.0215,
       "step": 287
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.6919999999999997e-05,
-      "loss": 7.1309,
+      "epoch": 0.32,
+      "learning_rate": 1.698e-05,
+      "loss": 6.942,
       "step": 288
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.698e-05,
-      "loss": 6.8141,
+      "epoch": 0.32,
+      "learning_rate": 1.704e-05,
+      "loss": 7.0736,
       "step": 289
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.704e-05,
-      "loss": 7.0745,
+      "epoch": 0.33,
+      "learning_rate": 1.71e-05,
+      "loss": 7.1623,
       "step": 290
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.71e-05,
-      "loss": 7.0307,
+      "epoch": 0.33,
+      "learning_rate": 1.716e-05,
+      "loss": 7.0506,
       "step": 291
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.716e-05,
-      "loss": 7.1498,
+      "epoch": 0.33,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 7.0028,
       "step": 292
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.7219999999999998e-05,
-      "loss": 6.9927,
+      "epoch": 0.33,
+      "learning_rate": 1.728e-05,
+      "loss": 6.8279,
       "step": 293
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.728e-05,
-      "loss": 6.9848,
+      "epoch": 0.33,
+      "learning_rate": 1.734e-05,
+      "loss": 6.8636,
       "step": 294
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.734e-05,
-      "loss": 6.7857,
+      "epoch": 0.33,
+      "learning_rate": 1.74e-05,
+      "loss": 6.7131,
       "step": 295
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.74e-05,
-      "loss": 6.6677,
+      "epoch": 0.33,
+      "learning_rate": 1.746e-05,
+      "loss": 6.7668,
       "step": 296
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.746e-05,
-      "loss": 6.6873,
+      "epoch": 0.33,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 6.7466,
       "step": 297
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.7519999999999998e-05,
-      "loss": 6.9224,
+      "epoch": 0.33,
+      "learning_rate": 1.758e-05,
+      "loss": 6.557,
       "step": 298
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.758e-05,
-      "loss": 6.6368,
+      "epoch": 0.34,
+      "learning_rate": 1.764e-05,
+      "loss": 6.5136,
       "step": 299
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.764e-05,
-      "loss": 6.1771,
+      "epoch": 0.34,
+      "learning_rate": 1.77e-05,
+      "loss": 6.3274,
       "step": 300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.77e-05,
-      "loss": 7.0091,
+      "epoch": 0.34,
+      "learning_rate": 1.776e-05,
+      "loss": 7.0971,
       "step": 301
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.776e-05,
-      "loss": 6.9775,
+      "epoch": 0.34,
+      "learning_rate": 1.782e-05,
+      "loss": 7.1435,
       "step": 302
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.782e-05,
-      "loss": 7.0055,
+      "epoch": 0.34,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 6.9586,
       "step": 303
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.7879999999999998e-05,
-      "loss": 6.8471,
+      "epoch": 0.34,
+      "learning_rate": 1.794e-05,
+      "loss": 7.0416,
       "step": 304
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.794e-05,
-      "loss": 6.9625,
+      "epoch": 0.34,
+      "learning_rate": 1.8e-05,
+      "loss": 6.8186,
       "step": 305
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.8e-05,
-      "loss": 6.9399,
+      "epoch": 0.34,
+      "learning_rate": 1.806e-05,
+      "loss": 6.8905,
       "step": 306
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.806e-05,
-      "loss": 7.0187,
+      "epoch": 0.34,
+      "learning_rate": 1.812e-05,
+      "loss": 6.9319,
       "step": 307
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.812e-05,
-      "loss": 7.1551,
+      "epoch": 0.35,
+      "learning_rate": 1.818e-05,
+      "loss": 6.9557,
       "step": 308
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.818e-05,
-      "loss": 6.9791,
+      "epoch": 0.35,
+      "learning_rate": 1.824e-05,
+      "loss": 6.8847,
       "step": 309
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.824e-05,
-      "loss": 7.098,
+      "epoch": 0.35,
+      "learning_rate": 1.83e-05,
+      "loss": 7.0108,
       "step": 310
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.83e-05,
-      "loss": 6.8741,
+      "epoch": 0.35,
+      "learning_rate": 1.836e-05,
+      "loss": 7.0779,
       "step": 311
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.836e-05,
-      "loss": 7.0331,
+      "epoch": 0.35,
+      "learning_rate": 1.842e-05,
+      "loss": 6.9719,
       "step": 312
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.842e-05,
-      "loss": 6.9652,
+      "epoch": 0.35,
+      "learning_rate": 1.848e-05,
+      "loss": 6.9861,
       "step": 313
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.848e-05,
-      "loss": 6.9758,
+      "epoch": 0.35,
+      "learning_rate": 1.854e-05,
+      "loss": 6.9209,
       "step": 314
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.854e-05,
-      "loss": 7.0413,
+      "epoch": 0.35,
+      "learning_rate": 1.86e-05,
+      "loss": 6.8109,
       "step": 315
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.86e-05,
-      "loss": 6.8393,
+      "epoch": 0.35,
+      "learning_rate": 1.866e-05,
+      "loss": 7.0507,
       "step": 316
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.866e-05,
-      "loss": 6.8837,
+      "epoch": 0.36,
+      "learning_rate": 1.872e-05,
+      "loss": 6.9378,
       "step": 317
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.872e-05,
-      "loss": 6.8909,
+      "epoch": 0.36,
+      "learning_rate": 1.878e-05,
+      "loss": 6.8479,
       "step": 318
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.878e-05,
-      "loss": 7.0392,
+      "epoch": 0.36,
+      "learning_rate": 1.884e-05,
+      "loss": 6.7743,
       "step": 319
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.884e-05,
-      "loss": 7.0512,
+      "epoch": 0.36,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 7.0167,
       "step": 320
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 7.2264,
+      "epoch": 0.36,
+      "learning_rate": 1.896e-05,
+      "loss": 6.9469,
       "step": 321
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.896e-05,
-      "loss": 7.1055,
+      "epoch": 0.36,
+      "learning_rate": 1.902e-05,
+      "loss": 6.8474,
       "step": 322
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.902e-05,
-      "loss": 7.2061,
+      "epoch": 0.36,
+      "learning_rate": 1.908e-05,
+      "loss": 6.9563,
       "step": 323
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.908e-05,
-      "loss": 6.9097,
+      "epoch": 0.36,
+      "learning_rate": 1.914e-05,
+      "loss": 6.9896,
       "step": 324
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.914e-05,
-      "loss": 6.8689,
+      "epoch": 0.36,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 6.9997,
       "step": 325
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 7.0512,
+      "epoch": 0.37,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 6.943,
       "step": 326
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9260000000000002e-05,
-      "loss": 6.9727,
+      "epoch": 0.37,
+      "learning_rate": 1.932e-05,
+      "loss": 6.8517,
       "step": 327
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.932e-05,
-      "loss": 6.9258,
+      "epoch": 0.37,
+      "learning_rate": 1.938e-05,
+      "loss": 6.9522,
       "step": 328
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.938e-05,
-      "loss": 6.8714,
+      "epoch": 0.37,
+      "learning_rate": 1.944e-05,
+      "loss": 7.0001,
       "step": 329
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.944e-05,
-      "loss": 7.0371,
+      "epoch": 0.37,
+      "learning_rate": 1.95e-05,
+      "loss": 6.9991,
       "step": 330
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.95e-05,
-      "loss": 6.9688,
+      "epoch": 0.37,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 7.0175,
       "step": 331
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9560000000000002e-05,
-      "loss": 6.9943,
+      "epoch": 0.37,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 6.739,
       "step": 332
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9620000000000002e-05,
-      "loss": 7.0706,
+      "epoch": 0.37,
+      "learning_rate": 1.968e-05,
+      "loss": 6.873,
       "step": 333
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.968e-05,
-      "loss": 7.0806,
+      "epoch": 0.37,
+      "learning_rate": 1.974e-05,
+      "loss": 6.9507,
       "step": 334
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.974e-05,
-      "loss": 6.9074,
+      "epoch": 0.38,
+      "learning_rate": 1.98e-05,
+      "loss": 7.0429,
       "step": 335
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.98e-05,
-      "loss": 6.9492,
+      "epoch": 0.38,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 6.8541,
       "step": 336
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9860000000000003e-05,
-      "loss": 7.0054,
+      "epoch": 0.38,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 6.8273,
       "step": 337
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9920000000000002e-05,
-      "loss": 6.7917,
+      "epoch": 0.38,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 6.9733,
       "step": 338
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9980000000000002e-05,
-      "loss": 6.8712,
+      "epoch": 0.38,
+      "learning_rate": 2.004e-05,
+      "loss": 6.9591,
       "step": 339
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.004e-05,
-      "loss": 6.9981,
+      "epoch": 0.38,
+      "learning_rate": 2.01e-05,
+      "loss": 6.9074,
       "step": 340
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.01e-05,
-      "loss": 6.739,
+      "epoch": 0.38,
+      "learning_rate": 2.016e-05,
+      "loss": 7.0047,
       "step": 341
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.016e-05,
-      "loss": 7.0512,
+      "epoch": 0.38,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 6.8685,
       "step": 342
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.0220000000000003e-05,
-      "loss": 7.029,
+      "epoch": 0.38,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 6.8809,
       "step": 343
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.0280000000000002e-05,
-      "loss": 6.9331,
+      "epoch": 0.39,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 6.8344,
       "step": 344
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.0340000000000002e-05,
-      "loss": 6.6985,
+      "epoch": 0.39,
+      "learning_rate": 2.04e-05,
+      "loss": 6.7439,
       "step": 345
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.04e-05,
-      "loss": 6.7564,
+      "epoch": 0.39,
+      "learning_rate": 2.046e-05,
+      "loss": 6.8392,
       "step": 346
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.046e-05,
-      "loss": 7.356,
+      "epoch": 0.39,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 6.7882,
       "step": 347
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.0520000000000003e-05,
-      "loss": 6.7162,
+      "epoch": 0.39,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 6.8275,
       "step": 348
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.0580000000000003e-05,
-      "loss": 6.2531,
+      "epoch": 0.39,
+      "learning_rate": 2.064e-05,
+      "loss": 6.7192,
       "step": 349
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.064e-05,
-      "loss": 5.9597,
+      "epoch": 0.39,
+      "learning_rate": 2.07e-05,
+      "loss": 6.5946,
       "step": 350
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.07e-05,
-      "loss": 7.1404,
+      "epoch": 0.39,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 6.9426,
       "step": 351
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.0759999999999998e-05,
-      "loss": 7.2061,
+      "epoch": 0.39,
+      "learning_rate": 2.082e-05,
+      "loss": 6.9544,
       "step": 352
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.082e-05,
-      "loss": 7.0541,
+      "epoch": 0.4,
+      "learning_rate": 2.088e-05,
+      "loss": 6.9189,
       "step": 353
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.088e-05,
-      "loss": 6.9399,
+      "epoch": 0.4,
+      "learning_rate": 2.094e-05,
+      "loss": 6.8795,
       "step": 354
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.094e-05,
-      "loss": 6.8378,
+      "epoch": 0.4,
+      "learning_rate": 2.1e-05,
+      "loss": 7.0007,
       "step": 355
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.1e-05,
-      "loss": 6.9978,
+      "epoch": 0.4,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 6.8779,
       "step": 356
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.1059999999999998e-05,
-      "loss": 7.1521,
+      "epoch": 0.4,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 7.1244,
       "step": 357
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.1119999999999998e-05,
-      "loss": 6.999,
+      "epoch": 0.4,
+      "learning_rate": 2.118e-05,
+      "loss": 6.8622,
       "step": 358
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.118e-05,
-      "loss": 7.1594,
+      "epoch": 0.4,
+      "learning_rate": 2.124e-05,
+      "loss": 6.9796,
       "step": 359
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.124e-05,
-      "loss": 7.0373,
+      "epoch": 0.4,
+      "learning_rate": 2.13e-05,
+      "loss": 6.7798,
       "step": 360
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.13e-05,
-      "loss": 6.787,
+      "epoch": 0.4,
+      "learning_rate": 2.136e-05,
+      "loss": 6.8914,
       "step": 361
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.136e-05,
-      "loss": 6.691,
+      "epoch": 0.41,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 6.9833,
       "step": 362
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.1419999999999998e-05,
+      "epoch": 0.41,
+      "learning_rate": 2.148e-05,
       "loss": 6.8806,
       "step": 363
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.148e-05,
-      "loss": 6.8176,
+      "epoch": 0.41,
+      "learning_rate": 2.154e-05,
+      "loss": 6.9408,
       "step": 364
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.154e-05,
-      "loss": 6.9223,
+      "epoch": 0.41,
+      "learning_rate": 2.16e-05,
+      "loss": 6.8998,
       "step": 365
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 2.16e-05,
-      "loss": 7.0395,
+      "epoch": 0.41,
+      "learning_rate": 2.166e-05,
+      "loss": 6.8415,
       "step": 366
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 2.166e-05,
-      "loss": 6.8747,
+      "epoch": 0.41,
+      "learning_rate": 2.172e-05,
+      "loss": 6.8909,
       "step": 367
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 2.172e-05,
-      "loss": 6.9183,
+      "epoch": 0.41,
+      "learning_rate": 2.178e-05,
+      "loss": 6.8517,
       "step": 368
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 2.178e-05,
-      "loss": 7.0181,
-      "step": 369
-    },
-    {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 2.184e-05,
-      "loss": 6.8826,
-      "step": 370
+      "loss": 6.7895,
+      "step": 369
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 2.19e-05,
-      "loss": 6.8227,
-      "step": 371
+      "loss": 6.9884,
+      "step": 370
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.196e-05,
-      "loss": 7.1325,
-      "step": 372
+      "loss": 6.8805,
+      "step": 371
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.202e-05,
-      "loss": 6.9693,
-      "step": 373
+      "loss": 7.0732,
+      "step": 372
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.208e-05,
-      "loss": 6.8778,
-      "step": 374
+      "loss": 6.9139,
+      "step": 373
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.214e-05,
-      "loss": 6.721,
-      "step": 375
+      "loss": 6.9828,
+      "step": 374
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.22e-05,
-      "loss": 6.9055,
-      "step": 376
+      "loss": 6.7987,
+      "step": 375
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.226e-05,
-      "loss": 6.8857,
-      "step": 377
+      "loss": 6.914,
+      "step": 376
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.232e-05,
-      "loss": 6.8144,
-      "step": 378
+      "loss": 6.7596,
+      "step": 377
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.238e-05,
-      "loss": 6.9794,
-      "step": 379
+      "loss": 6.8845,
+      "step": 378
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 2.2440000000000002e-05,
-      "loss": 7.0523,
-      "step": 380
+      "loss": 6.8886,
+      "step": 379
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 2.25e-05,
-      "loss": 6.831,
-      "step": 381
+      "loss": 6.9432,
+      "step": 380
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 2.256e-05,
-      "loss": 6.76,
-      "step": 382
+      "loss": 6.969,
+      "step": 381
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 2.262e-05,
-      "loss": 6.8863,
-      "step": 383
+      "loss": 6.8245,
+      "step": 382
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.268e-05,
-      "loss": 6.9378,
-      "step": 384
+      "loss": 6.9876,
+      "step": 383
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.274e-05,
-      "loss": 7.0514,
-      "step": 385
+      "loss": 6.9816,
+      "step": 384
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.2800000000000002e-05,
-      "loss": 7.2014,
-      "step": 386
+      "loss": 6.8357,
+      "step": 385
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.286e-05,
-      "loss": 6.9512,
-      "step": 387
+      "loss": 6.8805,
+      "step": 386
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.292e-05,
-      "loss": 6.7428,
-      "step": 388
+      "loss": 6.8696,
+      "step": 387
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 2.298e-05,
-      "loss": 6.8436,
-      "step": 389
+      "loss": 6.8541,
+      "step": 388
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.304e-05,
-      "loss": 7.0127,
-      "step": 390
+      "loss": 6.8526,
+      "step": 389
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.3100000000000002e-05,
-      "loss": 6.8316,
-      "step": 391
+      "loss": 7.0966,
+      "step": 390
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.3160000000000002e-05,
-      "loss": 7.4064,
-      "step": 392
+      "loss": 6.8956,
+      "step": 391
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.322e-05,
-      "loss": 6.9515,
-      "step": 393
+      "loss": 6.8709,
+      "step": 392
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.328e-05,
-      "loss": 6.7482,
-      "step": 394
+      "loss": 6.8491,
+      "step": 393
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.334e-05,
-      "loss": 6.8115,
-      "step": 395
+      "loss": 7.0112,
+      "step": 394
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.3400000000000003e-05,
-      "loss": 6.6293,
-      "step": 396
+      "loss": 6.8404,
+      "step": 395
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 2.3460000000000002e-05,
-      "loss": 6.4929,
-      "step": 397
+      "loss": 6.6512,
+      "step": 396
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 2.3520000000000002e-05,
-      "loss": 6.7504,
-      "step": 398
+      "loss": 6.5857,
+      "step": 397
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 2.358e-05,
-      "loss": 6.8337,
-      "step": 399
+      "loss": 6.6665,
+      "step": 398
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 2.364e-05,
-      "loss": 6.4172,
-      "step": 400
+      "loss": 6.6296,
+      "step": 399
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 2.37e-05,
-      "loss": 7.1933,
-      "step": 401
+      "loss": 6.4654,
+      "step": 400
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 2.3760000000000003e-05,
-      "loss": 6.9968,
-      "step": 402
+      "loss": 7.1123,
+      "step": 401
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 2.3820000000000002e-05,
-      "loss": 6.9228,
-      "step": 403
+      "loss": 7.0724,
+      "step": 402
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 2.3880000000000002e-05,
-      "loss": 7.2128,
-      "step": 404
+      "loss": 6.9767,
+      "step": 403
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 2.394e-05,
-      "loss": 6.92,
-      "step": 405
+      "loss": 6.899,
+      "step": 404
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 2.4e-05,
-      "loss": 7.028,
-      "step": 406
+      "loss": 7.0785,
+      "step": 405
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.4060000000000003e-05,
-      "loss": 6.8878,
-      "step": 407
+      "loss": 6.909,
+      "step": 406
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.4120000000000003e-05,
-      "loss": 6.9468,
-      "step": 408
+      "loss": 6.9077,
+      "step": 407
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.4180000000000002e-05,
-      "loss": 6.878,
-      "step": 409
+      "loss": 6.8431,
+      "step": 408
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.4240000000000002e-05,
-      "loss": 6.8267,
-      "step": 410
+      "loss": 7.0443,
+      "step": 409
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.43e-05,
-      "loss": 6.8701,
-      "step": 411
+      "loss": 6.9833,
+      "step": 410
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.4360000000000004e-05,
-      "loss": 6.7607,
-      "step": 412
+      "loss": 6.9221,
+      "step": 411
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.442e-05,
-      "loss": 7.013,
-      "step": 413
+      "loss": 6.9201,
+      "step": 412
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.448e-05,
-      "loss": 6.8109,
-      "step": 414
+      "loss": 6.9065,
+      "step": 413
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 2.454e-05,
-      "loss": 6.9703,
-      "step": 415
+      "loss": 6.7746,
+      "step": 414
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 2.4599999999999998e-05,
-      "loss": 6.9161,
-      "step": 416
+      "loss": 6.8324,
+      "step": 415
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 2.4659999999999998e-05,
-      "loss": 6.8961,
-      "step": 417
+      "loss": 6.934,
+      "step": 416
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 2.472e-05,
-      "loss": 6.8813,
-      "step": 418
+      "loss": 6.8337,
+      "step": 417
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 2.478e-05,
-      "loss": 6.8783,
-      "step": 419
+      "loss": 6.9123,
+      "step": 418
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 2.484e-05,
-      "loss": 7.0015,
-      "step": 420
+      "loss": 6.8812,
+      "step": 419
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 2.49e-05,
-      "loss": 7.0308,
-      "step": 421
+      "loss": 6.9068,
+      "step": 420
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 2.4959999999999998e-05,
-      "loss": 6.9404,
-      "step": 422
+      "loss": 6.9287,
+      "step": 421
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 2.502e-05,
-      "loss": 6.88,
-      "step": 423
+      "loss": 6.9944,
+      "step": 422
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 2.508e-05,
-      "loss": 7.142,
-      "step": 424
+      "loss": 6.9173,
+      "step": 423
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.514e-05,
-      "loss": 7.1374,
-      "step": 425
+      "loss": 6.9294,
+      "step": 424
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.52e-05,
-      "loss": 6.9695,
-      "step": 426
+      "loss": 6.9498,
+      "step": 425
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.526e-05,
-      "loss": 6.8461,
-      "step": 427
+      "loss": 7.0747,
+      "step": 426
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.5319999999999998e-05,
-      "loss": 6.9249,
-      "step": 428
+      "loss": 6.8528,
+      "step": 427
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.538e-05,
-      "loss": 6.9065,
-      "step": 429
+      "loss": 7.0238,
+      "step": 428
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.544e-05,
-      "loss": 6.8777,
-      "step": 430
+      "loss": 6.8598,
+      "step": 429
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.55e-05,
-      "loss": 6.797,
-      "step": 431
+      "loss": 6.8975,
+      "step": 430
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.556e-05,
-      "loss": 6.9828,
-      "step": 432
+      "loss": 7.0235,
+      "step": 431
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 2.562e-05,
-      "loss": 6.8179,
-      "step": 433
+      "loss": 6.9912,
+      "step": 432
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 2.568e-05,
-      "loss": 7.0432,
-      "step": 434
+      "loss": 6.6638,
+      "step": 433
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 2.574e-05,
-      "loss": 6.8892,
-      "step": 435
+      "loss": 6.8904,
+      "step": 434
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 2.58e-05,
-      "loss": 6.8475,
-      "step": 436
+      "loss": 6.8181,
+      "step": 435
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 2.586e-05,
-      "loss": 6.8647,
-      "step": 437
+      "loss": 7.0223,
+      "step": 436
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 2.592e-05,
-      "loss": 7.0183,
-      "step": 438
+      "loss": 6.9219,
+      "step": 437
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 2.5980000000000002e-05,
-      "loss": 6.7951,
-      "step": 439
+      "loss": 6.9923,
+      "step": 438
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 2.604e-05,
-      "loss": 6.9539,
-      "step": 440
+      "loss": 6.7617,
+      "step": 439
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 2.61e-05,
-      "loss": 6.8903,
-      "step": 441
+      "loss": 6.9326,
+      "step": 440
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 2.616e-05,
-      "loss": 6.8775,
-      "step": 442
+      "loss": 6.9804,
+      "step": 441
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.622e-05,
-      "loss": 6.6416,
-      "step": 443
+      "loss": 6.8346,
+      "step": 442
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.628e-05,
-      "loss": 6.7871,
-      "step": 444
+      "loss": 6.8193,
+      "step": 443
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.6340000000000002e-05,
-      "loss": 6.7176,
-      "step": 445
+      "loss": 6.9711,
+      "step": 444
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.64e-05,
-      "loss": 7.0733,
-      "step": 446
+      "loss": 6.7013,
+      "step": 445
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.646e-05,
-      "loss": 6.7832,
-      "step": 447
+      "loss": 6.7377,
+      "step": 446
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.652e-05,
-      "loss": 6.7244,
-      "step": 448
+      "loss": 6.7358,
+      "step": 447
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.658e-05,
-      "loss": 6.9716,
-      "step": 449
+      "loss": 6.7159,
+      "step": 448
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.6640000000000002e-05,
-      "loss": 6.318,
-      "step": 450
+      "loss": 6.4348,
+      "step": 449
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 2.6700000000000002e-05,
-      "loss": 6.9136,
-      "step": 451
+      "loss": 6.4836,
+      "step": 450
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 2.676e-05,
-      "loss": 6.9698,
-      "step": 452
+      "loss": 6.9618,
+      "step": 451
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 2.682e-05,
-      "loss": 7.1456,
-      "step": 453
+      "loss": 7.0763,
+      "step": 452
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 2.688e-05,
-      "loss": 6.9773,
-      "step": 454
+      "loss": 7.1267,
+      "step": 453
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.6940000000000003e-05,
-      "loss": 6.9118,
-      "step": 455
+      "loss": 6.9226,
+      "step": 454
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 7.0159,
-      "step": 456
+      "loss": 7.0096,
+      "step": 455
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.7060000000000002e-05,
-      "loss": 6.7261,
-      "step": 457
+      "loss": 7.0543,
+      "step": 456
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.712e-05,
-      "loss": 6.897,
-      "step": 458
+      "loss": 6.8422,
+      "step": 457
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.718e-05,
-      "loss": 6.7587,
-      "step": 459
+      "loss": 6.9216,
+      "step": 458
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 2.724e-05,
-      "loss": 6.9831,
-      "step": 460
+      "loss": 6.923,
+      "step": 459
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7300000000000003e-05,
-      "loss": 6.7344,
-      "step": 461
+      "loss": 7.0043,
+      "step": 460
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7360000000000002e-05,
-      "loss": 6.6914,
-      "step": 462
+      "loss": 6.8199,
+      "step": 461
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7420000000000002e-05,
-      "loss": 6.8347,
-      "step": 463
+      "loss": 6.9491,
+      "step": 462
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.748e-05,
-      "loss": 6.9312,
-      "step": 464
+      "loss": 6.8379,
+      "step": 463
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.754e-05,
-      "loss": 6.7597,
-      "step": 465
+      "loss": 6.8074,
+      "step": 464
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7600000000000003e-05,
-      "loss": 7.0332,
-      "step": 466
+      "loss": 6.9184,
+      "step": 465
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7660000000000003e-05,
-      "loss": 6.8458,
-      "step": 467
+      "loss": 6.9016,
+      "step": 466
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.7720000000000002e-05,
-      "loss": 7.0238,
-      "step": 468
+      "loss": 6.8365,
+      "step": 467
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 2.778e-05,
-      "loss": 6.6565,
-      "step": 469
+      "loss": 6.9663,
+      "step": 468
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 2.784e-05,
-      "loss": 6.9542,
-      "step": 470
+      "loss": 6.8732,
+      "step": 469
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 2.79e-05,
-      "loss": 6.9137,
-      "step": 471
+      "loss": 6.955,
+      "step": 470
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 2.7960000000000003e-05,
-      "loss": 6.7256,
-      "step": 472
+      "loss": 6.8268,
+      "step": 471
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.8020000000000003e-05,
-      "loss": 6.7577,
-      "step": 473
+      "loss": 6.8658,
+      "step": 472
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.8080000000000002e-05,
-      "loss": 6.8348,
-      "step": 474
+      "loss": 6.8687,
+      "step": 473
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.8139999999999998e-05,
-      "loss": 6.8269,
-      "step": 475
+      "loss": 6.7876,
+      "step": 474
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.8199999999999998e-05,
-      "loss": 6.8532,
-      "step": 476
+      "loss": 6.9435,
+      "step": 475
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.826e-05,
-      "loss": 6.8558,
-      "step": 477
+      "loss": 6.9562,
+      "step": 476
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 2.832e-05,
-      "loss": 6.8463,
-      "step": 478
+      "loss": 7.0299,
+      "step": 477
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.838e-05,
-      "loss": 6.9432,
-      "step": 479
+      "loss": 6.8044,
+      "step": 478
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.844e-05,
-      "loss": 6.9042,
-      "step": 480
+      "loss": 6.8591,
+      "step": 479
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 7.0396,
-      "step": 481
+      "loss": 6.8271,
+      "step": 480
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.856e-05,
-      "loss": 6.9029,
-      "step": 482
+      "loss": 6.8259,
+      "step": 481
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.862e-05,
-      "loss": 7.2545,
-      "step": 483
+      "loss": 6.8352,
+      "step": 482
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.868e-05,
-      "loss": 6.8826,
-      "step": 484
+      "loss": 6.8316,
+      "step": 483
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.874e-05,
-      "loss": 6.9597,
-      "step": 485
+      "loss": 6.9393,
+      "step": 484
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.88e-05,
-      "loss": 6.9589,
-      "step": 486
+      "loss": 6.9123,
+      "step": 485
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 2.8859999999999998e-05,
-      "loss": 6.7782,
-      "step": 487
+      "loss": 6.8544,
+      "step": 486
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 2.892e-05,
-      "loss": 7.0644,
-      "step": 488
+      "loss": 6.9356,
+      "step": 487
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 2.898e-05,
-      "loss": 6.8763,
-      "step": 489
+      "loss": 6.8117,
+      "step": 488
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 2.904e-05,
-      "loss": 6.8704,
-      "step": 490
+      "loss": 7.0152,
+      "step": 489
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.91e-05,
-      "loss": 6.6573,
-      "step": 491
+      "loss": 7.0389,
+      "step": 490
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.916e-05,
-      "loss": 7.0178,
-      "step": 492
+      "loss": 6.8311,
+      "step": 491
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.922e-05,
-      "loss": 6.9046,
-      "step": 493
+      "loss": 6.8967,
+      "step": 492
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.928e-05,
-      "loss": 7.0029,
-      "step": 494
+      "loss": 6.8858,
+      "step": 493
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.934e-05,
-      "loss": 6.7652,
-      "step": 495
+      "loss": 6.832,
+      "step": 494
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 2.94e-05,
-      "loss": 6.85,
-      "step": 496
+      "loss": 6.921,
+      "step": 495
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 2.946e-05,
-      "loss": 6.6303,
-      "step": 497
+      "loss": 6.6614,
+      "step": 496
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 2.9520000000000002e-05,
-      "loss": 7.0158,
-      "step": 498
+      "loss": 6.3714,
+      "step": 497
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 2.958e-05,
-      "loss": 6.544,
-      "step": 499
+      "loss": 6.9957,
+      "step": 498
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 2.964e-05,
-      "loss": 6.5504,
+      "loss": 6.7645,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.97e-05,
+      "loss": 6.3306,
       "step": 500
     },
     {
-      "epoch": 0.28,
-      "eval_loss": 6.836305141448975,
-      "eval_runtime": 945.4556,
-      "eval_samples_per_second": 2.794,
-      "eval_steps_per_second": 0.35,
-      "eval_wer": 1.8409825981705261,
+      "epoch": 0.56,
+      "eval_loss": 6.914114475250244,
+      "eval_runtime": 952.541,
+      "eval_samples_per_second": 2.774,
+      "eval_steps_per_second": 0.347,
+      "eval_wer": 2.0110324027223845,
       "step": 500
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.97e-05,
-      "loss": 6.9477,
+      "epoch": 0.56,
+      "learning_rate": 2.976e-05,
+      "loss": 7.2242,
       "step": 501
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.976e-05,
-      "loss": 7.0182,
+      "epoch": 0.56,
+      "learning_rate": 2.982e-05,
+      "loss": 6.8443,
       "step": 502
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.982e-05,
-      "loss": 6.8849,
+      "epoch": 0.56,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 7.0131,
       "step": 503
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.9880000000000002e-05,
-      "loss": 7.019,
+      "epoch": 0.57,
+      "learning_rate": 2.994e-05,
+      "loss": 7.0698,
       "step": 504
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.994e-05,
-      "loss": 6.8042,
+      "epoch": 0.57,
+      "learning_rate": 3e-05,
+      "loss": 7.0658,
       "step": 505
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 3e-05,
-      "loss": 7.0466,
+      "epoch": 0.57,
+      "learning_rate": 2.9923469387755103e-05,
+      "loss": 7.0079,
       "step": 506
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.9976635514018694e-05,
-      "loss": 6.7733,
+      "epoch": 0.57,
+      "learning_rate": 2.9846938775510208e-05,
+      "loss": 6.9869,
       "step": 507
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.9953271028037384e-05,
-      "loss": 7.0515,
+      "epoch": 0.57,
+      "learning_rate": 2.9770408163265306e-05,
+      "loss": 6.8591,
       "step": 508
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9929906542056074e-05,
-      "loss": 7.0465,
+      "epoch": 0.57,
+      "learning_rate": 2.969387755102041e-05,
+      "loss": 6.9314,
       "step": 509
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9906542056074768e-05,
-      "loss": 6.9775,
+      "epoch": 0.57,
+      "learning_rate": 2.961734693877551e-05,
+      "loss": 6.9464,
       "step": 510
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.988317757009346e-05,
-      "loss": 6.9379,
+      "epoch": 0.57,
+      "learning_rate": 2.9540816326530616e-05,
+      "loss": 6.8774,
       "step": 511
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.985981308411215e-05,
-      "loss": 6.814,
+      "epoch": 0.57,
+      "learning_rate": 2.9464285714285714e-05,
+      "loss": 6.9161,
       "step": 512
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.983644859813084e-05,
-      "loss": 6.8524,
+      "epoch": 0.58,
+      "learning_rate": 2.9387755102040816e-05,
+      "loss": 7.0163,
       "step": 513
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9813084112149534e-05,
-      "loss": 6.76,
+      "epoch": 0.58,
+      "learning_rate": 2.9311224489795918e-05,
+      "loss": 6.9863,
       "step": 514
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9789719626168224e-05,
-      "loss": 6.964,
+      "epoch": 0.58,
+      "learning_rate": 2.9234693877551023e-05,
+      "loss": 6.9584,
       "step": 515
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9766355140186918e-05,
-      "loss": 6.9627,
+      "epoch": 0.58,
+      "learning_rate": 2.9158163265306125e-05,
+      "loss": 6.7738,
       "step": 516
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9742990654205608e-05,
-      "loss": 6.9889,
+      "epoch": 0.58,
+      "learning_rate": 2.9081632653061224e-05,
+      "loss": 7.0052,
       "step": 517
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.97196261682243e-05,
-      "loss": 6.9928,
+      "epoch": 0.58,
+      "learning_rate": 2.9005102040816326e-05,
+      "loss": 6.8515,
       "step": 518
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.969626168224299e-05,
-      "loss": 6.9118,
+      "epoch": 0.58,
+      "learning_rate": 2.892857142857143e-05,
+      "loss": 7.0348,
       "step": 519
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9672897196261685e-05,
-      "loss": 6.8786,
+      "epoch": 0.58,
+      "learning_rate": 2.8852040816326533e-05,
+      "loss": 7.0086,
       "step": 520
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9649532710280375e-05,
-      "loss": 6.822,
+      "epoch": 0.58,
+      "learning_rate": 2.877551020408163e-05,
+      "loss": 6.8237,
       "step": 521
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9626168224299065e-05,
-      "loss": 6.9571,
+      "epoch": 0.59,
+      "learning_rate": 2.8698979591836733e-05,
+      "loss": 6.8245,
       "step": 522
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.9602803738317758e-05,
-      "loss": 6.7427,
+      "epoch": 0.59,
+      "learning_rate": 2.862244897959184e-05,
+      "loss": 6.7518,
       "step": 523
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.957943925233645e-05,
-      "loss": 6.9174,
+      "epoch": 0.59,
+      "learning_rate": 2.854591836734694e-05,
+      "loss": 6.8415,
       "step": 524
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.955607476635514e-05,
-      "loss": 6.9424,
+      "epoch": 0.59,
+      "learning_rate": 2.8469387755102042e-05,
+      "loss": 6.9844,
       "step": 525
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.953271028037383e-05,
-      "loss": 6.8831,
+      "epoch": 0.59,
+      "learning_rate": 2.8392857142857144e-05,
+      "loss": 6.7874,
       "step": 526
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9509345794392525e-05,
-      "loss": 7.1207,
+      "epoch": 0.59,
+      "learning_rate": 2.8316326530612246e-05,
+      "loss": 6.9059,
       "step": 527
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9485981308411218e-05,
-      "loss": 6.7529,
+      "epoch": 0.59,
+      "learning_rate": 2.8239795918367348e-05,
+      "loss": 6.938,
       "step": 528
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9462616822429905e-05,
-      "loss": 6.948,
+      "epoch": 0.59,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 6.9756,
       "step": 529
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9439252336448598e-05,
-      "loss": 7.056,
+      "epoch": 0.59,
+      "learning_rate": 2.8086734693877552e-05,
+      "loss": 7.0175,
       "step": 530
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.941588785046729e-05,
-      "loss": 6.7203,
+      "epoch": 0.6,
+      "learning_rate": 2.8010204081632654e-05,
+      "loss": 6.8836,
       "step": 531
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9392523364485985e-05,
-      "loss": 6.7475,
+      "epoch": 0.6,
+      "learning_rate": 2.7933673469387756e-05,
+      "loss": 6.8041,
       "step": 532
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.936915887850467e-05,
-      "loss": 6.9221,
+      "epoch": 0.6,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 6.8195,
       "step": 533
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9345794392523365e-05,
-      "loss": 6.7263,
+      "epoch": 0.6,
+      "learning_rate": 2.778061224489796e-05,
+      "loss": 6.8403,
       "step": 534
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.932242990654206e-05,
-      "loss": 7.1025,
+      "epoch": 0.6,
+      "learning_rate": 2.770408163265306e-05,
+      "loss": 6.8532,
       "step": 535
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.929906542056075e-05,
-      "loss": 6.9126,
+      "epoch": 0.6,
+      "learning_rate": 2.7627551020408163e-05,
+      "loss": 7.0077,
       "step": 536
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.927570093457944e-05,
-      "loss": 6.7898,
+      "epoch": 0.6,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 6.8801,
       "step": 537
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9252336448598132e-05,
-      "loss": 6.7788,
+      "epoch": 0.6,
+      "learning_rate": 2.747448979591837e-05,
+      "loss": 6.996,
       "step": 538
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9228971962616825e-05,
-      "loss": 6.9275,
+      "epoch": 0.6,
+      "learning_rate": 2.739795918367347e-05,
+      "loss": 6.9724,
       "step": 539
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9205607476635515e-05,
-      "loss": 7.1243,
+      "epoch": 0.61,
+      "learning_rate": 2.732142857142857e-05,
+      "loss": 6.9039,
       "step": 540
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9182242990654205e-05,
-      "loss": 6.7623,
+      "epoch": 0.61,
+      "learning_rate": 2.7244897959183673e-05,
+      "loss": 7.0277,
       "step": 541
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.91588785046729e-05,
-      "loss": 7.074,
+      "epoch": 0.61,
+      "learning_rate": 2.7168367346938778e-05,
+      "loss": 6.8933,
       "step": 542
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.913551401869159e-05,
-      "loss": 6.7808,
+      "epoch": 0.61,
+      "learning_rate": 2.7091836734693877e-05,
+      "loss": 6.7796,
       "step": 543
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.9112149532710282e-05,
-      "loss": 6.7796,
+      "epoch": 0.61,
+      "learning_rate": 2.701530612244898e-05,
+      "loss": 6.6972,
       "step": 544
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.9088785046728972e-05,
-      "loss": 6.5872,
+      "epoch": 0.61,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 6.7981,
       "step": 545
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.9065420560747665e-05,
-      "loss": 6.5091,
+      "epoch": 0.61,
+      "learning_rate": 2.6862244897959186e-05,
+      "loss": 6.7376,
       "step": 546
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.9042056074766355e-05,
-      "loss": 6.9471,
+      "epoch": 0.61,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 6.9709,
       "step": 547
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.901869158878505e-05,
-      "loss": 6.5288,
+      "epoch": 0.61,
+      "learning_rate": 2.6709183673469386e-05,
+      "loss": 6.7878,
       "step": 548
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.899532710280374e-05,
-      "loss": 6.3195,
+      "epoch": 0.62,
+      "learning_rate": 2.663265306122449e-05,
+      "loss": 6.3952,
       "step": 549
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.897196261682243e-05,
-      "loss": 6.1041,
+      "epoch": 0.62,
+      "learning_rate": 2.6556122448979593e-05,
+      "loss": 6.6438,
       "step": 550
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8948598130841122e-05,
-      "loss": 7.2529,
+      "epoch": 0.62,
+      "learning_rate": 2.6479591836734695e-05,
+      "loss": 6.9375,
       "step": 551
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8925233644859816e-05,
-      "loss": 7.2857,
+      "epoch": 0.62,
+      "learning_rate": 2.6403061224489794e-05,
+      "loss": 7.1262,
       "step": 552
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8901869158878506e-05,
-      "loss": 6.9488,
+      "epoch": 0.62,
+      "learning_rate": 2.63265306122449e-05,
+      "loss": 6.9578,
       "step": 553
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8878504672897196e-05,
-      "loss": 6.9521,
+      "epoch": 0.62,
+      "learning_rate": 2.625e-05,
+      "loss": 6.951,
       "step": 554
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.885514018691589e-05,
-      "loss": 6.9471,
+      "epoch": 0.62,
+      "learning_rate": 2.6173469387755103e-05,
+      "loss": 6.9577,
       "step": 555
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8831775700934582e-05,
-      "loss": 7.0363,
+      "epoch": 0.62,
+      "learning_rate": 2.6096938775510205e-05,
+      "loss": 6.9202,
       "step": 556
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8808411214953272e-05,
-      "loss": 7.0419,
+      "epoch": 0.62,
+      "learning_rate": 2.6020408163265307e-05,
+      "loss": 6.8923,
       "step": 557
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8785046728971962e-05,
-      "loss": 7.0027,
+      "epoch": 0.63,
+      "learning_rate": 2.594387755102041e-05,
+      "loss": 6.9608,
       "step": 558
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8761682242990656e-05,
-      "loss": 6.7093,
+      "epoch": 0.63,
+      "learning_rate": 2.586734693877551e-05,
+      "loss": 6.8879,
       "step": 559
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.873831775700935e-05,
-      "loss": 6.9143,
+      "epoch": 0.63,
+      "learning_rate": 2.5790816326530613e-05,
+      "loss": 6.8141,
       "step": 560
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.8714953271028036e-05,
-      "loss": 6.8941,
+      "epoch": 0.63,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 6.9607,
       "step": 561
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.869158878504673e-05,
-      "loss": 6.8279,
+      "epoch": 0.63,
+      "learning_rate": 2.5637755102040816e-05,
+      "loss": 6.7892,
       "step": 562
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8668224299065423e-05,
-      "loss": 6.9369,
+      "epoch": 0.63,
+      "learning_rate": 2.5561224489795918e-05,
+      "loss": 6.8645,
       "step": 563
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8644859813084116e-05,
-      "loss": 7.1366,
+      "epoch": 0.63,
+      "learning_rate": 2.548469387755102e-05,
+      "loss": 6.8667,
       "step": 564
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8621495327102803e-05,
-      "loss": 6.9355,
+      "epoch": 0.63,
+      "learning_rate": 2.5408163265306125e-05,
+      "loss": 7.0323,
       "step": 565
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8598130841121496e-05,
-      "loss": 6.8462,
+      "epoch": 0.63,
+      "learning_rate": 2.5331632653061224e-05,
+      "loss": 6.7248,
       "step": 566
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.857476635514019e-05,
-      "loss": 7.047,
+      "epoch": 0.64,
+      "learning_rate": 2.5255102040816326e-05,
+      "loss": 6.9148,
       "step": 567
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.855140186915888e-05,
-      "loss": 6.9656,
+      "epoch": 0.64,
+      "learning_rate": 2.517857142857143e-05,
+      "loss": 6.7584,
       "step": 568
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.852803738317757e-05,
-      "loss": 6.9764,
+      "epoch": 0.64,
+      "learning_rate": 2.5102040816326533e-05,
+      "loss": 6.8439,
       "step": 569
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8504672897196263e-05,
-      "loss": 6.9151,
+      "epoch": 0.64,
+      "learning_rate": 2.502551020408163e-05,
+      "loss": 6.9657,
       "step": 570
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8481308411214956e-05,
-      "loss": 6.8525,
+      "epoch": 0.64,
+      "learning_rate": 2.4948979591836734e-05,
+      "loss": 6.8511,
       "step": 571
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8457943925233646e-05,
-      "loss": 6.8168,
+      "epoch": 0.64,
+      "learning_rate": 2.487244897959184e-05,
+      "loss": 6.9668,
       "step": 572
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8434579439252336e-05,
-      "loss": 6.8258,
-      "step": 573
+      "epoch": 0.64,
+      "learning_rate": 2.479591836734694e-05,
+      "loss": 6.823,
+      "step": 573
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.841121495327103e-05,
-      "loss": 6.7106,
+      "epoch": 0.64,
+      "learning_rate": 2.4719387755102043e-05,
+      "loss": 6.7057,
       "step": 574
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.838785046728972e-05,
-      "loss": 6.8401,
+      "epoch": 0.64,
+      "learning_rate": 2.464285714285714e-05,
+      "loss": 6.9239,
       "step": 575
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8364485981308413e-05,
-      "loss": 6.9233,
+      "epoch": 0.65,
+      "learning_rate": 2.4566326530612246e-05,
+      "loss": 6.7678,
       "step": 576
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8341121495327103e-05,
-      "loss": 6.9301,
+      "epoch": 0.65,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 6.9136,
       "step": 577
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8317757009345797e-05,
-      "loss": 6.728,
+      "epoch": 0.65,
+      "learning_rate": 2.441326530612245e-05,
+      "loss": 6.961,
       "step": 578
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8294392523364487e-05,
-      "loss": 6.9819,
+      "epoch": 0.65,
+      "learning_rate": 2.433673469387755e-05,
+      "loss": 6.8128,
       "step": 579
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.827102803738318e-05,
-      "loss": 7.0727,
+      "epoch": 0.65,
+      "learning_rate": 2.4260204081632654e-05,
+      "loss": 6.9925,
       "step": 580
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.824766355140187e-05,
-      "loss": 6.8829,
+      "epoch": 0.65,
+      "learning_rate": 2.4183673469387756e-05,
+      "loss": 7.0088,
       "step": 581
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.822429906542056e-05,
-      "loss": 6.8399,
+      "epoch": 0.65,
+      "learning_rate": 2.4107142857142858e-05,
+      "loss": 6.8783,
       "step": 582
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8200934579439253e-05,
-      "loss": 6.7249,
+      "epoch": 0.65,
+      "learning_rate": 2.403061224489796e-05,
+      "loss": 6.7756,
       "step": 583
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8177570093457947e-05,
-      "loss": 7.0553,
+      "epoch": 0.65,
+      "learning_rate": 2.3954081632653062e-05,
+      "loss": 6.9015,
       "step": 584
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8154205607476637e-05,
-      "loss": 6.9466,
+      "epoch": 0.66,
+      "learning_rate": 2.3877551020408164e-05,
+      "loss": 6.8994,
       "step": 585
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8130841121495327e-05,
-      "loss": 6.7042,
+      "epoch": 0.66,
+      "learning_rate": 2.3801020408163266e-05,
+      "loss": 6.9457,
       "step": 586
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.810747663551402e-05,
-      "loss": 7.0006,
+      "epoch": 0.66,
+      "learning_rate": 2.372448979591837e-05,
+      "loss": 6.9353,
       "step": 587
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8084112149532714e-05,
-      "loss": 6.8928,
+      "epoch": 0.66,
+      "learning_rate": 2.364795918367347e-05,
+      "loss": 6.8825,
       "step": 588
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.80607476635514e-05,
-      "loss": 7.0989,
+      "epoch": 0.66,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 6.783,
       "step": 589
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8037383177570094e-05,
-      "loss": 7.2081,
+      "epoch": 0.66,
+      "learning_rate": 2.3494897959183673e-05,
+      "loss": 6.8247,
       "step": 590
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.8014018691588787e-05,
-      "loss": 6.9274,
+      "epoch": 0.66,
+      "learning_rate": 2.341836734693878e-05,
+      "loss": 6.8881,
       "step": 591
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.799065420560748e-05,
-      "loss": 6.9092,
+      "epoch": 0.66,
+      "learning_rate": 2.3341836734693877e-05,
+      "loss": 6.907,
       "step": 592
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.7967289719626167e-05,
-      "loss": 6.6006,
+      "epoch": 0.66,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 6.9815,
       "step": 593
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.794392523364486e-05,
-      "loss": 6.8673,
+      "epoch": 0.67,
+      "learning_rate": 2.318877551020408e-05,
+      "loss": 6.6994,
       "step": 594
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.7920560747663554e-05,
-      "loss": 6.7209,
+      "epoch": 0.67,
+      "learning_rate": 2.3112244897959186e-05,
+      "loss": 6.7467,
       "step": 595
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.7897196261682244e-05,
-      "loss": 6.5913,
+      "epoch": 0.67,
+      "learning_rate": 2.3035714285714288e-05,
+      "loss": 6.6612,
       "step": 596
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.7873831775700934e-05,
-      "loss": 6.602,
+      "epoch": 0.67,
+      "learning_rate": 2.2959183673469387e-05,
+      "loss": 6.6789,
       "step": 597
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7850467289719627e-05,
-      "loss": 6.6236,
+      "epoch": 0.67,
+      "learning_rate": 2.288265306122449e-05,
+      "loss": 6.5702,
       "step": 598
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.782710280373832e-05,
-      "loss": 6.6432,
+      "epoch": 0.67,
+      "learning_rate": 2.2806122448979594e-05,
+      "loss": 6.4857,
       "step": 599
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.780373831775701e-05,
-      "loss": 6.4442,
+      "epoch": 0.67,
+      "learning_rate": 2.2729591836734696e-05,
+      "loss": 6.2174,
       "step": 600
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.77803738317757e-05,
-      "loss": 6.9091,
+      "epoch": 0.67,
+      "learning_rate": 2.2653061224489794e-05,
+      "loss": 6.9971,
       "step": 601
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7757009345794394e-05,
-      "loss": 7.076,
+      "epoch": 0.67,
+      "learning_rate": 2.2576530612244896e-05,
+      "loss": 6.8156,
       "step": 602
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7733644859813084e-05,
-      "loss": 6.8358,
+      "epoch": 0.68,
+      "learning_rate": 2.25e-05,
+      "loss": 6.9443,
       "step": 603
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7710280373831777e-05,
-      "loss": 6.97,
+      "epoch": 0.68,
+      "learning_rate": 2.2423469387755103e-05,
+      "loss": 6.9254,
       "step": 604
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7686915887850467e-05,
-      "loss": 6.7384,
+      "epoch": 0.68,
+      "learning_rate": 2.2346938775510205e-05,
+      "loss": 7.0202,
       "step": 605
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.766355140186916e-05,
-      "loss": 6.8965,
+      "epoch": 0.68,
+      "learning_rate": 2.2270408163265307e-05,
+      "loss": 6.719,
       "step": 606
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.764018691588785e-05,
-      "loss": 7.1481,
+      "epoch": 0.68,
+      "learning_rate": 2.219387755102041e-05,
+      "loss": 6.8925,
       "step": 607
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7616822429906544e-05,
-      "loss": 6.8106,
+      "epoch": 0.68,
+      "learning_rate": 2.211734693877551e-05,
+      "loss": 6.9246,
       "step": 608
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7593457943925234e-05,
-      "loss": 6.9471,
+      "epoch": 0.68,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 6.8959,
       "step": 609
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7570093457943924e-05,
-      "loss": 6.7231,
+      "epoch": 0.68,
+      "learning_rate": 2.1964285714285715e-05,
+      "loss": 6.8851,
       "step": 610
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7546728971962618e-05,
-      "loss": 7.1053,
+      "epoch": 0.68,
+      "learning_rate": 2.1887755102040817e-05,
+      "loss": 6.8646,
       "step": 611
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.752336448598131e-05,
-      "loss": 6.9046,
+      "epoch": 0.69,
+      "learning_rate": 2.181122448979592e-05,
+      "loss": 6.9031,
       "step": 612
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.75e-05,
-      "loss": 6.9137,
+      "epoch": 0.69,
+      "learning_rate": 2.173469387755102e-05,
+      "loss": 6.8019,
       "step": 613
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.747663551401869e-05,
-      "loss": 6.9535,
+      "epoch": 0.69,
+      "learning_rate": 2.1658163265306126e-05,
+      "loss": 6.8787,
       "step": 614
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.7453271028037384e-05,
-      "loss": 6.903,
+      "epoch": 0.69,
+      "learning_rate": 2.1581632653061224e-05,
+      "loss": 7.0718,
       "step": 615
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7429906542056078e-05,
-      "loss": 6.8893,
+      "epoch": 0.69,
+      "learning_rate": 2.1505102040816326e-05,
+      "loss": 6.7524,
       "step": 616
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7406542056074764e-05,
-      "loss": 6.893,
+      "epoch": 0.69,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 6.8568,
       "step": 617
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7383177570093458e-05,
-      "loss": 6.6302,
+      "epoch": 0.69,
+      "learning_rate": 2.1352040816326533e-05,
+      "loss": 6.712,
       "step": 618
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.735981308411215e-05,
-      "loss": 6.9364,
+      "epoch": 0.69,
+      "learning_rate": 2.1275510204081632e-05,
+      "loss": 6.8248,
       "step": 619
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7336448598130845e-05,
-      "loss": 6.9075,
+      "epoch": 0.7,
+      "learning_rate": 2.1198979591836734e-05,
+      "loss": 6.8799,
       "step": 620
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.731308411214953e-05,
-      "loss": 6.9257,
+      "epoch": 0.7,
+      "learning_rate": 2.1122448979591836e-05,
+      "loss": 6.918,
       "step": 621
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7289719626168225e-05,
-      "loss": 6.9107,
+      "epoch": 0.7,
+      "learning_rate": 2.104591836734694e-05,
+      "loss": 6.8227,
       "step": 622
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7266355140186918e-05,
-      "loss": 6.906,
+      "epoch": 0.7,
+      "learning_rate": 2.0969387755102043e-05,
+      "loss": 6.946,
       "step": 623
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7242990654205608e-05,
-      "loss": 7.1149,
+      "epoch": 0.7,
+      "learning_rate": 2.089285714285714e-05,
+      "loss": 6.7692,
       "step": 624
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7219626168224298e-05,
-      "loss": 6.9155,
+      "epoch": 0.7,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 6.9808,
       "step": 625
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.719626168224299e-05,
-      "loss": 6.9432,
+      "epoch": 0.7,
+      "learning_rate": 2.073979591836735e-05,
+      "loss": 6.8455,
       "step": 626
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7172897196261685e-05,
-      "loss": 6.8205,
+      "epoch": 0.7,
+      "learning_rate": 2.066326530612245e-05,
+      "loss": 6.9412,
       "step": 627
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7149532710280375e-05,
-      "loss": 7.0667,
+      "epoch": 0.7,
+      "learning_rate": 2.058673469387755e-05,
+      "loss": 6.9302,
       "step": 628
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7126168224299065e-05,
-      "loss": 7.0436,
+      "epoch": 0.71,
+      "learning_rate": 2.0510204081632654e-05,
+      "loss": 6.896,
       "step": 629
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7102803738317758e-05,
-      "loss": 6.8381,
+      "epoch": 0.71,
+      "learning_rate": 2.0433673469387756e-05,
+      "loss": 7.0522,
       "step": 630
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.7079439252336448e-05,
-      "loss": 6.8524,
+      "epoch": 0.71,
+      "learning_rate": 2.0357142857142858e-05,
+      "loss": 6.8552,
       "step": 631
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.705607476635514e-05,
-      "loss": 6.7521,
+      "epoch": 0.71,
+      "learning_rate": 2.0280612244897957e-05,
+      "loss": 6.7395,
       "step": 632
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.703271028037383e-05,
-      "loss": 6.8176,
+      "epoch": 0.71,
+      "learning_rate": 2.0204081632653062e-05,
+      "loss": 6.9425,
       "step": 633
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.7009345794392525e-05,
-      "loss": 6.9459,
+      "epoch": 0.71,
+      "learning_rate": 2.0127551020408164e-05,
+      "loss": 6.8416,
       "step": 634
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6985981308411215e-05,
-      "loss": 6.8894,
+      "epoch": 0.71,
+      "learning_rate": 2.0051020408163266e-05,
+      "loss": 6.956,
       "step": 635
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.696261682242991e-05,
-      "loss": 6.7983,
+      "epoch": 0.71,
+      "learning_rate": 1.9974489795918368e-05,
+      "loss": 6.9566,
       "step": 636
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.69392523364486e-05,
-      "loss": 6.7721,
+      "epoch": 0.71,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 6.9165,
       "step": 637
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.691588785046729e-05,
-      "loss": 6.6705,
+      "epoch": 0.72,
+      "learning_rate": 1.982142857142857e-05,
+      "loss": 6.7259,
       "step": 638
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6892523364485982e-05,
-      "loss": 6.9796,
+      "epoch": 0.72,
+      "learning_rate": 1.9744897959183674e-05,
+      "loss": 7.0196,
       "step": 639
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6869158878504675e-05,
-      "loss": 6.9659,
+      "epoch": 0.72,
+      "learning_rate": 1.9668367346938775e-05,
+      "loss": 6.7428,
       "step": 640
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6845794392523365e-05,
-      "loss": 6.7956,
+      "epoch": 0.72,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 6.8471,
       "step": 641
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6822429906542055e-05,
-      "loss": 6.8247,
+      "epoch": 0.72,
+      "learning_rate": 1.951530612244898e-05,
+      "loss": 7.0086,
       "step": 642
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.679906542056075e-05,
-      "loss": 6.7601,
+      "epoch": 0.72,
+      "learning_rate": 1.943877551020408e-05,
+      "loss": 6.897,
       "step": 643
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6775700934579442e-05,
-      "loss": 6.7661,
+      "epoch": 0.72,
+      "learning_rate": 1.9362244897959186e-05,
+      "loss": 6.6539,
       "step": 644
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.675233644859813e-05,
-      "loss": 6.4002,
+      "epoch": 0.72,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 6.7148,
       "step": 645
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6728971962616822e-05,
-      "loss": 6.6197,
+      "epoch": 0.72,
+      "learning_rate": 1.9209183673469387e-05,
+      "loss": 6.8433,
       "step": 646
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6705607476635515e-05,
-      "loss": 6.6521,
+      "epoch": 0.73,
+      "learning_rate": 1.913265306122449e-05,
+      "loss": 6.5996,
       "step": 647
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.668224299065421e-05,
-      "loss": 6.7822,
+      "epoch": 0.73,
+      "learning_rate": 1.9056122448979594e-05,
+      "loss": 6.5503,
       "step": 648
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6658878504672895e-05,
-      "loss": 6.7982,
+      "epoch": 0.73,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 6.7521,
       "step": 649
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.663551401869159e-05,
-      "loss": 6.3804,
+      "epoch": 0.73,
+      "learning_rate": 1.8903061224489795e-05,
+      "loss": 6.4819,
       "step": 650
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.6612149532710282e-05,
-      "loss": 7.0459,
+      "epoch": 0.73,
+      "learning_rate": 1.8826530612244896e-05,
+      "loss": 7.0465,
       "step": 651
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6588785046728972e-05,
-      "loss": 6.9107,
+      "epoch": 0.73,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 7.0229,
       "step": 652
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6565420560747662e-05,
-      "loss": 6.9636,
+      "epoch": 0.73,
+      "learning_rate": 1.8673469387755104e-05,
+      "loss": 7.0164,
       "step": 653
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6542056074766356e-05,
-      "loss": 6.9575,
+      "epoch": 0.73,
+      "learning_rate": 1.8596938775510206e-05,
+      "loss": 6.9124,
       "step": 654
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.651869158878505e-05,
-      "loss": 6.8835,
+      "epoch": 0.73,
+      "learning_rate": 1.8520408163265304e-05,
+      "loss": 6.9706,
       "step": 655
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.649532710280374e-05,
-      "loss": 6.7949,
+      "epoch": 0.74,
+      "learning_rate": 1.844387755102041e-05,
+      "loss": 6.9488,
       "step": 656
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.647196261682243e-05,
-      "loss": 6.7848,
+      "epoch": 0.74,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 6.7632,
       "step": 657
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6448598130841122e-05,
-      "loss": 7.0225,
+      "epoch": 0.74,
+      "learning_rate": 1.8290816326530613e-05,
+      "loss": 7.0554,
       "step": 658
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6425233644859816e-05,
-      "loss": 6.8895,
+      "epoch": 0.74,
+      "learning_rate": 1.8214285714285712e-05,
+      "loss": 6.8021,
       "step": 659
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6401869158878506e-05,
-      "loss": 6.9337,
+      "epoch": 0.74,
+      "learning_rate": 1.8137755102040817e-05,
+      "loss": 6.796,
       "step": 660
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6378504672897196e-05,
-      "loss": 6.9826,
+      "epoch": 0.74,
+      "learning_rate": 1.806122448979592e-05,
+      "loss": 6.8337,
       "step": 661
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.635514018691589e-05,
-      "loss": 6.8631,
+      "epoch": 0.74,
+      "learning_rate": 1.798469387755102e-05,
+      "loss": 6.8948,
       "step": 662
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.633177570093458e-05,
-      "loss": 6.8617,
+      "epoch": 0.74,
+      "learning_rate": 1.7908163265306126e-05,
+      "loss": 6.9559,
       "step": 663
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6308411214953273e-05,
-      "loss": 6.9331,
+      "epoch": 0.74,
+      "learning_rate": 1.7831632653061225e-05,
+      "loss": 6.7223,
       "step": 664
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6285046728971963e-05,
-      "loss": 6.8159,
+      "epoch": 0.75,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 6.8449,
       "step": 665
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6261682242990656e-05,
-      "loss": 6.862,
+      "epoch": 0.75,
+      "learning_rate": 1.767857142857143e-05,
+      "loss": 6.9404,
       "step": 666
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.6238317757009346e-05,
-      "loss": 6.9316,
+      "epoch": 0.75,
+      "learning_rate": 1.7602040816326534e-05,
+      "loss": 6.863,
       "step": 667
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.621495327102804e-05,
-      "loss": 6.8388,
+      "epoch": 0.75,
+      "learning_rate": 1.7525510204081632e-05,
+      "loss": 6.7984,
       "step": 668
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.619158878504673e-05,
-      "loss": 6.8171,
+      "epoch": 0.75,
+      "learning_rate": 1.7448979591836734e-05,
+      "loss": 6.9578,
       "step": 669
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.616822429906542e-05,
-      "loss": 6.7678,
+      "epoch": 0.75,
+      "learning_rate": 1.7372448979591836e-05,
+      "loss": 6.8244,
       "step": 670
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.6144859813084113e-05,
-      "loss": 7.1313,
+      "epoch": 0.75,
+      "learning_rate": 1.729591836734694e-05,
+      "loss": 6.7548,
       "step": 671
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.6121495327102806e-05,
-      "loss": 6.8629,
+      "epoch": 0.75,
+      "learning_rate": 1.7219387755102043e-05,
+      "loss": 6.889,
       "step": 672
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.6098130841121496e-05,
-      "loss": 6.818,
+      "epoch": 0.75,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 6.8286,
       "step": 673
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.6074766355140186e-05,
-      "loss": 6.7714,
+      "epoch": 0.76,
+      "learning_rate": 1.7066326530612244e-05,
+      "loss": 6.797,
       "step": 674
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.605140186915888e-05,
-      "loss": 6.8695,
+      "epoch": 0.76,
+      "learning_rate": 1.698979591836735e-05,
+      "loss": 6.9046,
       "step": 675
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.6028037383177573e-05,
-      "loss": 6.8891,
+      "epoch": 0.76,
+      "learning_rate": 1.691326530612245e-05,
+      "loss": 6.9557,
       "step": 676
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.600467289719626e-05,
-      "loss": 6.707,
+      "epoch": 0.76,
+      "learning_rate": 1.683673469387755e-05,
+      "loss": 6.8563,
       "step": 677
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5981308411214953e-05,
-      "loss": 6.8912,
+      "epoch": 0.76,
+      "learning_rate": 1.676020408163265e-05,
+      "loss": 6.6538,
       "step": 678
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5957943925233647e-05,
-      "loss": 6.8149,
+      "epoch": 0.76,
+      "learning_rate": 1.6683673469387757e-05,
+      "loss": 6.8718,
       "step": 679
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.593457943925234e-05,
-      "loss": 6.9449,
+      "epoch": 0.76,
+      "learning_rate": 1.660714285714286e-05,
+      "loss": 6.8163,
       "step": 680
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5911214953271027e-05,
-      "loss": 7.0771,
+      "epoch": 0.76,
+      "learning_rate": 1.6530612244897957e-05,
+      "loss": 6.9295,
       "step": 681
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.588785046728972e-05,
-      "loss": 6.824,
+      "epoch": 0.76,
+      "learning_rate": 1.6454081632653062e-05,
+      "loss": 6.8405,
       "step": 682
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5864485981308413e-05,
-      "loss": 6.7868,
+      "epoch": 0.77,
+      "learning_rate": 1.6377551020408164e-05,
+      "loss": 6.8496,
       "step": 683
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5841121495327103e-05,
-      "loss": 6.8572,
+      "epoch": 0.77,
+      "learning_rate": 1.6301020408163266e-05,
+      "loss": 6.9388,
       "step": 684
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5817757009345793e-05,
-      "loss": 7.0932,
+      "epoch": 0.77,
+      "learning_rate": 1.6224489795918368e-05,
+      "loss": 6.9675,
       "step": 685
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.5794392523364487e-05,
-      "loss": 6.9244,
+      "epoch": 0.77,
+      "learning_rate": 1.614795918367347e-05,
+      "loss": 6.7805,
       "step": 686
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.577102803738318e-05,
-      "loss": 6.7156,
+      "epoch": 0.77,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 6.7653,
       "step": 687
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.574766355140187e-05,
-      "loss": 6.8624,
+      "epoch": 0.77,
+      "learning_rate": 1.5994897959183674e-05,
+      "loss": 6.999,
       "step": 688
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.572429906542056e-05,
-      "loss": 7.0741,
+      "epoch": 0.77,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 6.8595,
       "step": 689
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5700934579439254e-05,
-      "loss": 6.7927,
+      "epoch": 0.77,
+      "learning_rate": 1.5841836734693878e-05,
+      "loss": 6.8792,
       "step": 690
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5677570093457944e-05,
-      "loss": 6.8285,
+      "epoch": 0.77,
+      "learning_rate": 1.576530612244898e-05,
+      "loss": 6.7378,
       "step": 691
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5654205607476637e-05,
-      "loss": 6.9886,
+      "epoch": 0.78,
+      "learning_rate": 1.568877551020408e-05,
+      "loss": 6.8505,
       "step": 692
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5630841121495327e-05,
-      "loss": 6.8393,
+      "epoch": 0.78,
+      "learning_rate": 1.5612244897959183e-05,
+      "loss": 6.6303,
       "step": 693
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.560747663551402e-05,
-      "loss": 6.8337,
+      "epoch": 0.78,
+      "learning_rate": 1.553571428571429e-05,
+      "loss": 6.8162,
       "step": 694
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.558411214953271e-05,
-      "loss": 6.8241,
+      "epoch": 0.78,
+      "learning_rate": 1.5459183673469387e-05,
+      "loss": 6.8154,
       "step": 695
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5560747663551404e-05,
-      "loss": 6.9435,
+      "epoch": 0.78,
+      "learning_rate": 1.538265306122449e-05,
+      "loss": 6.8108,
       "step": 696
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5537383177570094e-05,
-      "loss": 6.7445,
+      "epoch": 0.78,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 6.685,
       "step": 697
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5514018691588784e-05,
-      "loss": 6.6263,
+      "epoch": 0.78,
+      "learning_rate": 1.5229591836734695e-05,
+      "loss": 6.4358,
       "step": 698
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5490654205607477e-05,
-      "loss": 6.5207,
+      "epoch": 0.78,
+      "learning_rate": 1.5153061224489795e-05,
+      "loss": 6.5531,
       "step": 699
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.546728971962617e-05,
-      "loss": 6.5201,
+      "epoch": 0.78,
+      "learning_rate": 1.5076530612244897e-05,
+      "loss": 6.2196,
       "step": 700
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.544392523364486e-05,
-      "loss": 6.8353,
+      "epoch": 0.79,
+      "learning_rate": 1.5e-05,
+      "loss": 6.9518,
       "step": 701
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.542056074766355e-05,
-      "loss": 6.8489,
+      "epoch": 0.79,
+      "learning_rate": 1.4923469387755104e-05,
+      "loss": 6.9022,
       "step": 702
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5397196261682244e-05,
-      "loss": 6.7908,
+      "epoch": 0.79,
+      "learning_rate": 1.4846938775510204e-05,
+      "loss": 6.8897,
       "step": 703
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.5373831775700937e-05,
-      "loss": 7.1201,
+      "epoch": 0.79,
+      "learning_rate": 1.4770408163265308e-05,
+      "loss": 6.7608,
       "step": 704
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5350467289719624e-05,
-      "loss": 6.8204,
+      "epoch": 0.79,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 6.8291,
       "step": 705
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5327102803738317e-05,
-      "loss": 7.0967,
+      "epoch": 0.79,
+      "learning_rate": 1.4617346938775512e-05,
+      "loss": 6.9625,
       "step": 706
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.530373831775701e-05,
-      "loss": 6.9387,
+      "epoch": 0.79,
+      "learning_rate": 1.4540816326530612e-05,
+      "loss": 6.952,
       "step": 707
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5280373831775704e-05,
-      "loss": 7.0993,
+      "epoch": 0.79,
+      "learning_rate": 1.4464285714285715e-05,
+      "loss": 7.0419,
       "step": 708
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.525700934579439e-05,
-      "loss": 6.7813,
+      "epoch": 0.79,
+      "learning_rate": 1.4387755102040816e-05,
+      "loss": 6.8927,
       "step": 709
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5233644859813084e-05,
-      "loss": 6.8875,
+      "epoch": 0.8,
+      "learning_rate": 1.431122448979592e-05,
+      "loss": 6.849,
       "step": 710
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5210280373831778e-05,
-      "loss": 6.88,
+      "epoch": 0.8,
+      "learning_rate": 1.4234693877551021e-05,
+      "loss": 6.8324,
       "step": 711
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5186915887850468e-05,
-      "loss": 6.9203,
+      "epoch": 0.8,
+      "learning_rate": 1.4158163265306123e-05,
+      "loss": 6.727,
       "step": 712
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5163551401869158e-05,
-      "loss": 6.7676,
+      "epoch": 0.8,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 6.7543,
       "step": 713
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.514018691588785e-05,
-      "loss": 7.0641,
+      "epoch": 0.8,
+      "learning_rate": 1.4005102040816327e-05,
+      "loss": 6.8509,
       "step": 714
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5116822429906544e-05,
-      "loss": 6.9587,
+      "epoch": 0.8,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 6.9081,
       "step": 715
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5093457943925234e-05,
-      "loss": 6.8021,
+      "epoch": 0.8,
+      "learning_rate": 1.385204081632653e-05,
+      "loss": 6.932,
       "step": 716
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5070093457943924e-05,
-      "loss": 6.8006,
+      "epoch": 0.8,
+      "learning_rate": 1.3775510204081633e-05,
+      "loss": 7.0868,
       "step": 717
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5046728971962618e-05,
-      "loss": 6.9343,
+      "epoch": 0.8,
+      "learning_rate": 1.3698979591836735e-05,
+      "loss": 6.8515,
       "step": 718
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5023364485981308e-05,
-      "loss": 6.6841,
+      "epoch": 0.81,
+      "learning_rate": 1.3622448979591836e-05,
+      "loss": 6.738,
       "step": 719
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.5e-05,
-      "loss": 6.9038,
+      "epoch": 0.81,
+      "learning_rate": 1.3545918367346938e-05,
+      "loss": 6.7979,
       "step": 720
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.497663551401869e-05,
-      "loss": 6.9526,
+      "epoch": 0.81,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 6.875,
       "step": 721
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.4953271028037385e-05,
-      "loss": 7.0883,
+      "epoch": 0.81,
+      "learning_rate": 1.3392857142857144e-05,
+      "loss": 6.7573,
       "step": 722
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4929906542056075e-05,
-      "loss": 7.018,
+      "epoch": 0.81,
+      "learning_rate": 1.3316326530612246e-05,
+      "loss": 6.7725,
       "step": 723
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4906542056074768e-05,
-      "loss": 6.9153,
+      "epoch": 0.81,
+      "learning_rate": 1.3239795918367348e-05,
+      "loss": 6.9417,
       "step": 724
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4883177570093458e-05,
-      "loss": 6.8682,
+      "epoch": 0.81,
+      "learning_rate": 1.316326530612245e-05,
+      "loss": 6.8765,
       "step": 725
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4859813084112148e-05,
-      "loss": 6.739,
+      "epoch": 0.81,
+      "learning_rate": 1.3086734693877551e-05,
+      "loss": 6.9876,
       "step": 726
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.483644859813084e-05,
-      "loss": 6.5133,
+      "epoch": 0.82,
+      "learning_rate": 1.3010204081632653e-05,
+      "loss": 6.8864,
       "step": 727
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4813084112149535e-05,
-      "loss": 6.8839,
+      "epoch": 0.82,
+      "learning_rate": 1.2933673469387755e-05,
+      "loss": 6.7862,
       "step": 728
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4789719626168225e-05,
-      "loss": 6.7699,
+      "epoch": 0.82,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 6.8214,
       "step": 729
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4766355140186915e-05,
-      "loss": 6.9781,
+      "epoch": 0.82,
+      "learning_rate": 1.2780612244897959e-05,
+      "loss": 6.9334,
       "step": 730
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4742990654205608e-05,
-      "loss": 6.9263,
+      "epoch": 0.82,
+      "learning_rate": 1.2704081632653063e-05,
+      "loss": 7.0888,
       "step": 731
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.47196261682243e-05,
-      "loss": 6.7494,
+      "epoch": 0.82,
+      "learning_rate": 1.2627551020408163e-05,
+      "loss": 6.9266,
       "step": 732
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4696261682242988e-05,
-      "loss": 6.9572,
+      "epoch": 0.82,
+      "learning_rate": 1.2551020408163267e-05,
+      "loss": 6.8692,
       "step": 733
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.467289719626168e-05,
-      "loss": 6.9119,
+      "epoch": 0.82,
+      "learning_rate": 1.2474489795918367e-05,
+      "loss": 7.0387,
       "step": 734
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4649532710280375e-05,
-      "loss": 6.8606,
+      "epoch": 0.82,
+      "learning_rate": 1.239795918367347e-05,
+      "loss": 6.7814,
       "step": 735
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.462616822429907e-05,
-      "loss": 6.8117,
+      "epoch": 0.83,
+      "learning_rate": 1.232142857142857e-05,
+      "loss": 6.9312,
       "step": 736
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4602803738317755e-05,
-      "loss": 6.8229,
+      "epoch": 0.83,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 6.8926,
       "step": 737
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.457943925233645e-05,
-      "loss": 6.8619,
+      "epoch": 0.83,
+      "learning_rate": 1.2168367346938774e-05,
+      "loss": 6.9204,
       "step": 738
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4556074766355142e-05,
-      "loss": 6.8616,
+      "epoch": 0.83,
+      "learning_rate": 1.2091836734693878e-05,
+      "loss": 6.797,
       "step": 739
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4532710280373832e-05,
-      "loss": 6.881,
+      "epoch": 0.83,
+      "learning_rate": 1.201530612244898e-05,
+      "loss": 6.8846,
       "step": 740
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4509345794392522e-05,
-      "loss": 7.0254,
+      "epoch": 0.83,
+      "learning_rate": 1.1938775510204082e-05,
+      "loss": 6.9006,
       "step": 741
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4485981308411215e-05,
-      "loss": 6.9679,
+      "epoch": 0.83,
+      "learning_rate": 1.1862244897959185e-05,
+      "loss": 6.8803,
       "step": 742
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.446261682242991e-05,
-      "loss": 6.9089,
+      "epoch": 0.83,
+      "learning_rate": 1.1785714285714286e-05,
+      "loss": 6.7397,
       "step": 743
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.44392523364486e-05,
-      "loss": 6.9352,
+      "epoch": 0.83,
+      "learning_rate": 1.170918367346939e-05,
+      "loss": 6.8819,
       "step": 744
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.441588785046729e-05,
-      "loss": 6.8962,
+      "epoch": 0.84,
+      "learning_rate": 1.163265306122449e-05,
+      "loss": 6.7479,
       "step": 745
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4392523364485982e-05,
-      "loss": 6.5153,
+      "epoch": 0.84,
+      "learning_rate": 1.1556122448979593e-05,
+      "loss": 6.9675,
       "step": 746
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4369158878504672e-05,
-      "loss": 6.7076,
+      "epoch": 0.84,
+      "learning_rate": 1.1479591836734693e-05,
+      "loss": 6.6359,
       "step": 747
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4345794392523365e-05,
-      "loss": 6.6634,
+      "epoch": 0.84,
+      "learning_rate": 1.1403061224489797e-05,
+      "loss": 6.5651,
       "step": 748
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4322429906542055e-05,
-      "loss": 6.5065,
+      "epoch": 0.84,
+      "learning_rate": 1.1326530612244897e-05,
+      "loss": 6.5824,
       "step": 749
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.429906542056075e-05,
-      "loss": 6.461,
+      "epoch": 0.84,
+      "learning_rate": 1.125e-05,
+      "loss": 6.4211,
       "step": 750
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.427570093457944e-05,
-      "loss": 6.9578,
+      "epoch": 0.84,
+      "learning_rate": 1.1173469387755103e-05,
+      "loss": 6.8861,
       "step": 751
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4252336448598132e-05,
-      "loss": 6.9136,
+      "epoch": 0.84,
+      "learning_rate": 1.1096938775510205e-05,
+      "loss": 6.9163,
       "step": 752
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4228971962616822e-05,
-      "loss": 6.8486,
+      "epoch": 0.84,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 6.9292,
       "step": 753
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4205607476635516e-05,
-      "loss": 6.8676,
+      "epoch": 0.85,
+      "learning_rate": 1.0943877551020408e-05,
+      "loss": 6.8178,
       "step": 754
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.4182242990654206e-05,
-      "loss": 6.8012,
+      "epoch": 0.85,
+      "learning_rate": 1.086734693877551e-05,
+      "loss": 6.8772,
       "step": 755
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.41588785046729e-05,
-      "loss": 6.9938,
+      "epoch": 0.85,
+      "learning_rate": 1.0790816326530612e-05,
+      "loss": 6.9699,
       "step": 756
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.413551401869159e-05,
-      "loss": 6.9609,
+      "epoch": 0.85,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 6.8686,
       "step": 757
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.411214953271028e-05,
-      "loss": 7.0484,
+      "epoch": 0.85,
+      "learning_rate": 1.0637755102040816e-05,
+      "loss": 6.7655,
       "step": 758
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.4088785046728972e-05,
-      "loss": 6.6264,
+      "epoch": 0.85,
+      "learning_rate": 1.0561224489795918e-05,
+      "loss": 6.8784,
       "step": 759
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.4065420560747666e-05,
-      "loss": 6.7655,
+      "epoch": 0.85,
+      "learning_rate": 1.0484693877551021e-05,
+      "loss": 6.8053,
       "step": 760
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.4042056074766356e-05,
-      "loss": 6.8159,
+      "epoch": 0.85,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 6.8588,
       "step": 761
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.4018691588785046e-05,
-      "loss": 6.8132,
+      "epoch": 0.85,
+      "learning_rate": 1.0331632653061225e-05,
+      "loss": 6.8353,
       "step": 762
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.399532710280374e-05,
-      "loss": 6.9625,
+      "epoch": 0.86,
+      "learning_rate": 1.0255102040816327e-05,
+      "loss": 6.6755,
       "step": 763
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3971962616822433e-05,
-      "loss": 6.7799,
+      "epoch": 0.86,
+      "learning_rate": 1.0178571428571429e-05,
+      "loss": 6.692,
       "step": 764
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.394859813084112e-05,
-      "loss": 6.8399,
+      "epoch": 0.86,
+      "learning_rate": 1.0102040816326531e-05,
+      "loss": 6.8417,
       "step": 765
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3925233644859813e-05,
-      "loss": 6.682,
+      "epoch": 0.86,
+      "learning_rate": 1.0025510204081633e-05,
+      "loss": 6.7841,
       "step": 766
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3901869158878506e-05,
-      "loss": 6.9487,
+      "epoch": 0.86,
+      "learning_rate": 9.948979591836735e-06,
+      "loss": 6.7975,
       "step": 767
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.38785046728972e-05,
-      "loss": 6.7143,
+      "epoch": 0.86,
+      "learning_rate": 9.872448979591837e-06,
+      "loss": 6.871,
       "step": 768
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3855140186915886e-05,
-      "loss": 6.9637,
+      "epoch": 0.86,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 6.9317,
       "step": 769
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.383177570093458e-05,
-      "loss": 6.7947,
+      "epoch": 0.86,
+      "learning_rate": 9.71938775510204e-06,
+      "loss": 6.7614,
       "step": 770
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3808411214953273e-05,
-      "loss": 6.8793,
+      "epoch": 0.86,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 6.9313,
       "step": 771
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3785046728971963e-05,
-      "loss": 6.8757,
+      "epoch": 0.87,
+      "learning_rate": 9.566326530612244e-06,
+      "loss": 6.9925,
       "step": 772
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3761682242990653e-05,
-      "loss": 6.7586,
+      "epoch": 0.87,
+      "learning_rate": 9.489795918367348e-06,
+      "loss": 6.9193,
       "step": 773
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.3738317757009346e-05,
-      "loss": 6.892,
+      "epoch": 0.87,
+      "learning_rate": 9.413265306122448e-06,
+      "loss": 6.7339,
       "step": 774
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.371495327102804e-05,
-      "loss": 6.8216,
+      "epoch": 0.87,
+      "learning_rate": 9.336734693877552e-06,
+      "loss": 6.6202,
       "step": 775
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.369158878504673e-05,
-      "loss": 6.9253,
+      "epoch": 0.87,
+      "learning_rate": 9.260204081632652e-06,
+      "loss": 6.7642,
       "step": 776
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.366822429906542e-05,
-      "loss": 6.8504,
+      "epoch": 0.87,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 6.6974,
       "step": 777
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3644859813084113e-05,
-      "loss": 6.9135,
+      "epoch": 0.87,
+      "learning_rate": 9.107142857142856e-06,
+      "loss": 6.8733,
       "step": 778
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3621495327102803e-05,
-      "loss": 6.9361,
+      "epoch": 0.87,
+      "learning_rate": 9.03061224489796e-06,
+      "loss": 6.8635,
       "step": 779
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3598130841121497e-05,
-      "loss": 6.8002,
+      "epoch": 0.87,
+      "learning_rate": 8.954081632653063e-06,
+      "loss": 6.8125,
       "step": 780
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3574766355140187e-05,
-      "loss": 6.9148,
+      "epoch": 0.88,
+      "learning_rate": 8.877551020408163e-06,
+      "loss": 6.8385,
       "step": 781
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.355140186915888e-05,
-      "loss": 6.8681,
+      "epoch": 0.88,
+      "learning_rate": 8.801020408163267e-06,
+      "loss": 6.7776,
       "step": 782
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.352803738317757e-05,
-      "loss": 6.8964,
+      "epoch": 0.88,
+      "learning_rate": 8.724489795918367e-06,
+      "loss": 6.9029,
       "step": 783
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3504672897196263e-05,
-      "loss": 6.9917,
+      "epoch": 0.88,
+      "learning_rate": 8.64795918367347e-06,
+      "loss": 6.7458,
       "step": 784
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.3481308411214953e-05,
-      "loss": 6.7812,
+      "epoch": 0.88,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 6.8837,
       "step": 785
     },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.3457943925233643e-05,
-      "loss": 6.823,
-      "step": 786
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.3434579439252337e-05,
-      "loss": 6.6291,
-      "step": 787
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.341121495327103e-05,
-      "loss": 7.0222,
-      "step": 788
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.338785046728972e-05,
-      "loss": 6.83,
-      "step": 789
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.336448598130841e-05,
-      "loss": 7.0405,
-      "step": 790
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.3341121495327104e-05,
-      "loss": 6.813,
-      "step": 791
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.3317757009345797e-05,
-      "loss": 6.5597,
-      "step": 792
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.3294392523364484e-05,
-      "loss": 6.8485,
-      "step": 793
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3271028037383177e-05,
-      "loss": 6.8384,
-      "step": 794
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.324766355140187e-05,
-      "loss": 6.7384,
-      "step": 795
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3224299065420564e-05,
-      "loss": 6.7094,
-      "step": 796
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.320093457943925e-05,
-      "loss": 6.3973,
-      "step": 797
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3177570093457944e-05,
-      "loss": 6.6153,
-      "step": 798
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3154205607476637e-05,
-      "loss": 6.6117,
-      "step": 799
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3130841121495327e-05,
-      "loss": 6.4969,
-      "step": 800
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3107476635514017e-05,
-      "loss": 7.0891,
-      "step": 801
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.308411214953271e-05,
-      "loss": 6.9769,
-      "step": 802
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3060747663551404e-05,
-      "loss": 6.9745,
-      "step": 803
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3037383177570094e-05,
-      "loss": 6.861,
-      "step": 804
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.3014018691588784e-05,
-      "loss": 7.0942,
-      "step": 805
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.2990654205607477e-05,
-      "loss": 6.9584,
-      "step": 806
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.2967289719626167e-05,
-      "loss": 6.8102,
-      "step": 807
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.294392523364486e-05,
-      "loss": 6.8368,
-      "step": 808
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.292056074766355e-05,
-      "loss": 6.9359,
-      "step": 809
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.2897196261682244e-05,
-      "loss": 7.0023,
-      "step": 810
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 2.2873831775700934e-05,
-      "loss": 6.9327,
-      "step": 811
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2850467289719628e-05,
-      "loss": 7.0312,
-      "step": 812
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2827102803738318e-05,
-      "loss": 6.9279,
-      "step": 813
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2803738317757008e-05,
-      "loss": 6.8436,
-      "step": 814
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.27803738317757e-05,
-      "loss": 6.9188,
-      "step": 815
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2757009345794394e-05,
-      "loss": 6.8606,
-      "step": 816
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2733644859813084e-05,
-      "loss": 6.8486,
-      "step": 817
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2710280373831774e-05,
-      "loss": 6.9308,
-      "step": 818
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2686915887850468e-05,
-      "loss": 7.0318,
-      "step": 819
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.266355140186916e-05,
-      "loss": 6.8166,
-      "step": 820
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2640186915887848e-05,
-      "loss": 7.0149,
-      "step": 821
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.261682242990654e-05,
-      "loss": 6.8411,
-      "step": 822
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2593457943925235e-05,
-      "loss": 7.0391,
-      "step": 823
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2570093457943928e-05,
-      "loss": 6.9139,
-      "step": 824
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2546728971962615e-05,
-      "loss": 6.9403,
-      "step": 825
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2523364485981308e-05,
-      "loss": 6.8539,
-      "step": 826
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.25e-05,
-      "loss": 6.894,
-      "step": 827
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.247663551401869e-05,
-      "loss": 6.759,
-      "step": 828
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.2453271028037385e-05,
-      "loss": 6.9067,
-      "step": 829
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2429906542056075e-05,
-      "loss": 6.8483,
-      "step": 830
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2406542056074768e-05,
-      "loss": 7.0234,
-      "step": 831
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2383177570093458e-05,
-      "loss": 6.9858,
-      "step": 832
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.235981308411215e-05,
-      "loss": 6.8802,
-      "step": 833
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.233644859813084e-05,
-      "loss": 6.5624,
-      "step": 834
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.231308411214953e-05,
-      "loss": 6.687,
-      "step": 835
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2289719626168225e-05,
-      "loss": 6.8541,
-      "step": 836
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.226635514018692e-05,
-      "loss": 6.8854,
-      "step": 837
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.224299065420561e-05,
-      "loss": 6.7705,
-      "step": 838
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.22196261682243e-05,
-      "loss": 6.7308,
-      "step": 839
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2196261682242992e-05,
-      "loss": 6.8051,
-      "step": 840
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2172897196261685e-05,
-      "loss": 6.9525,
-      "step": 841
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2149532710280372e-05,
-      "loss": 6.8686,
-      "step": 842
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2126168224299065e-05,
-      "loss": 6.7111,
-      "step": 843
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.210280373831776e-05,
-      "loss": 7.0518,
-      "step": 844
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2079439252336452e-05,
-      "loss": 6.5944,
-      "step": 845
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.205607476635514e-05,
-      "loss": 6.5484,
-      "step": 846
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.2032710280373832e-05,
-      "loss": 6.7135,
-      "step": 847
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.2009345794392525e-05,
-      "loss": 6.77,
-      "step": 848
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1985981308411215e-05,
-      "loss": 6.4335,
-      "step": 849
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1962616822429905e-05,
-      "loss": 6.3619,
-      "step": 850
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.19392523364486e-05,
-      "loss": 6.9208,
-      "step": 851
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1915887850467292e-05,
-      "loss": 6.9282,
-      "step": 852
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1892523364485982e-05,
-      "loss": 6.7764,
-      "step": 853
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1869158878504672e-05,
-      "loss": 6.872,
-      "step": 854
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1845794392523366e-05,
-      "loss": 6.9836,
-      "step": 855
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.182242990654206e-05,
-      "loss": 6.874,
-      "step": 856
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.179906542056075e-05,
-      "loss": 6.9393,
-      "step": 857
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.177570093457944e-05,
-      "loss": 6.9515,
-      "step": 858
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1752336448598132e-05,
-      "loss": 6.9122,
-      "step": 859
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1728971962616822e-05,
-      "loss": 6.9423,
-      "step": 860
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1705607476635516e-05,
-      "loss": 6.7799,
-      "step": 861
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1682242990654206e-05,
-      "loss": 6.7063,
-      "step": 862
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.16588785046729e-05,
-      "loss": 6.714,
-      "step": 863
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.163551401869159e-05,
-      "loss": 6.6632,
-      "step": 864
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.1612149532710283e-05,
-      "loss": 6.746,
-      "step": 865
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1588785046728973e-05,
-      "loss": 6.8954,
-      "step": 866
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1565420560747663e-05,
-      "loss": 6.7371,
-      "step": 867
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1542056074766356e-05,
-      "loss": 6.8054,
-      "step": 868
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.151869158878505e-05,
-      "loss": 6.8553,
-      "step": 869
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.149532710280374e-05,
-      "loss": 6.8746,
-      "step": 870
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.147196261682243e-05,
-      "loss": 6.8714,
-      "step": 871
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1448598130841123e-05,
-      "loss": 6.8473,
-      "step": 872
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1425233644859816e-05,
-      "loss": 6.8735,
-      "step": 873
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1401869158878503e-05,
-      "loss": 6.8812,
-      "step": 874
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1378504672897196e-05,
-      "loss": 7.182,
-      "step": 875
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.135514018691589e-05,
-      "loss": 6.9035,
-      "step": 876
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1331775700934583e-05,
-      "loss": 7.041,
-      "step": 877
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.130841121495327e-05,
-      "loss": 6.8413,
-      "step": 878
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1285046728971963e-05,
-      "loss": 6.7065,
-      "step": 879
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1261682242990657e-05,
-      "loss": 6.9946,
-      "step": 880
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1238317757009347e-05,
-      "loss": 6.9583,
-      "step": 881
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.1214953271028037e-05,
-      "loss": 6.6607,
-      "step": 882
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.119158878504673e-05,
-      "loss": 6.9227,
-      "step": 883
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1168224299065423e-05,
-      "loss": 6.8609,
-      "step": 884
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1144859813084113e-05,
-      "loss": 6.9806,
-      "step": 885
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1121495327102803e-05,
-      "loss": 7.0735,
-      "step": 886
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1098130841121497e-05,
-      "loss": 6.9184,
-      "step": 887
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1074766355140187e-05,
-      "loss": 6.7718,
-      "step": 888
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.105140186915888e-05,
-      "loss": 6.7666,
-      "step": 889
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.102803738317757e-05,
-      "loss": 6.8053,
-      "step": 890
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.1004672897196264e-05,
-      "loss": 6.8106,
-      "step": 891
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0981308411214954e-05,
-      "loss": 6.9283,
-      "step": 892
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0957943925233647e-05,
-      "loss": 6.9118,
-      "step": 893
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0934579439252337e-05,
-      "loss": 6.7881,
-      "step": 894
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0911214953271027e-05,
-      "loss": 6.6612,
-      "step": 895
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.088785046728972e-05,
-      "loss": 6.7019,
-      "step": 896
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0864485981308414e-05,
-      "loss": 6.4215,
-      "step": 897
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0841121495327104e-05,
-      "loss": 6.7619,
-      "step": 898
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0817757009345794e-05,
-      "loss": 6.1893,
-      "step": 899
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 2.0794392523364487e-05,
-      "loss": 6.3476,
-      "step": 900
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.077102803738318e-05,
-      "loss": 6.7679,
-      "step": 901
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0747663551401867e-05,
-      "loss": 7.0297,
-      "step": 902
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.072429906542056e-05,
-      "loss": 7.087,
-      "step": 903
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0700934579439254e-05,
-      "loss": 6.8482,
-      "step": 904
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0677570093457947e-05,
-      "loss": 7.0867,
-      "step": 905
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0654205607476634e-05,
-      "loss": 7.0951,
-      "step": 906
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0630841121495327e-05,
-      "loss": 6.8813,
-      "step": 907
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.060747663551402e-05,
-      "loss": 6.7417,
-      "step": 908
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.058411214953271e-05,
-      "loss": 6.9576,
-      "step": 909
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.05607476635514e-05,
-      "loss": 6.8562,
-      "step": 910
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0537383177570094e-05,
-      "loss": 7.0367,
-      "step": 911
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0514018691588788e-05,
-      "loss": 6.8002,
-      "step": 912
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0490654205607478e-05,
-      "loss": 6.7203,
-      "step": 913
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0467289719626168e-05,
-      "loss": 6.9329,
-      "step": 914
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.044392523364486e-05,
-      "loss": 6.8467,
-      "step": 915
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.042056074766355e-05,
-      "loss": 6.715,
-      "step": 916
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0397196261682244e-05,
-      "loss": 6.8809,
-      "step": 917
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.0373831775700934e-05,
-      "loss": 6.9604,
-      "step": 918
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0350467289719628e-05,
-      "loss": 6.8344,
-      "step": 919
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0327102803738318e-05,
-      "loss": 6.7132,
-      "step": 920
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.030373831775701e-05,
-      "loss": 6.8926,
-      "step": 921
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.02803738317757e-05,
-      "loss": 6.651,
-      "step": 922
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.025700934579439e-05,
-      "loss": 6.6979,
-      "step": 923
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0233644859813085e-05,
-      "loss": 6.9543,
-      "step": 924
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0210280373831778e-05,
-      "loss": 6.8831,
-      "step": 925
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0186915887850468e-05,
-      "loss": 6.9815,
-      "step": 926
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0163551401869158e-05,
-      "loss": 6.7378,
-      "step": 927
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.014018691588785e-05,
-      "loss": 6.7603,
-      "step": 928
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0116822429906545e-05,
-      "loss": 6.8225,
-      "step": 929
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.009345794392523e-05,
-      "loss": 6.9021,
-      "step": 930
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0070093457943925e-05,
-      "loss": 6.9059,
-      "step": 931
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.0046728971962618e-05,
-      "loss": 7.0147,
-      "step": 932
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 2.002336448598131e-05,
-      "loss": 6.6675,
-      "step": 933
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 1.9999999999999998e-05,
-      "loss": 6.9065,
-      "step": 934
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 1.997663551401869e-05,
-      "loss": 6.8607,
-      "step": 935
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 1.9953271028037385e-05,
-      "loss": 6.938,
-      "step": 936
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9929906542056075e-05,
-      "loss": 6.9035,
-      "step": 937
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9906542056074765e-05,
-      "loss": 6.7782,
-      "step": 938
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.988317757009346e-05,
-      "loss": 6.614,
-      "step": 939
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9859813084112152e-05,
-      "loss": 6.9619,
-      "step": 940
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9836448598130842e-05,
-      "loss": 6.9196,
-      "step": 941
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9813084112149532e-05,
-      "loss": 6.7751,
-      "step": 942
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9789719626168225e-05,
-      "loss": 7.0537,
-      "step": 943
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9766355140186915e-05,
-      "loss": 6.7198,
-      "step": 944
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.974299065420561e-05,
-      "loss": 6.9298,
-      "step": 945
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.97196261682243e-05,
-      "loss": 6.7528,
-      "step": 946
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9696261682242992e-05,
-      "loss": 6.356,
-      "step": 947
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9672897196261682e-05,
-      "loss": 6.6354,
-      "step": 948
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9649532710280375e-05,
-      "loss": 6.9587,
-      "step": 949
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9626168224299065e-05,
-      "loss": 6.0676,
-      "step": 950
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.960280373831776e-05,
-      "loss": 6.9761,
-      "step": 951
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.957943925233645e-05,
-      "loss": 6.9882,
-      "step": 952
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9556074766355142e-05,
-      "loss": 7.0076,
-      "step": 953
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 1.9532710280373832e-05,
-      "loss": 6.8262,
-      "step": 954
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9509345794392522e-05,
-      "loss": 6.8917,
-      "step": 955
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9485981308411216e-05,
-      "loss": 6.9115,
-      "step": 956
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.946261682242991e-05,
-      "loss": 6.9221,
-      "step": 957
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.94392523364486e-05,
-      "loss": 6.8089,
-      "step": 958
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.941588785046729e-05,
-      "loss": 7.0238,
-      "step": 959
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9392523364485982e-05,
-      "loss": 6.8093,
-      "step": 960
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9369158878504676e-05,
-      "loss": 6.7606,
-      "step": 961
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9345794392523363e-05,
-      "loss": 6.9127,
-      "step": 962
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9322429906542056e-05,
-      "loss": 6.772,
-      "step": 963
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.929906542056075e-05,
-      "loss": 6.7633,
-      "step": 964
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9275700934579443e-05,
-      "loss": 6.8137,
-      "step": 965
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.925233644859813e-05,
-      "loss": 6.7548,
-      "step": 966
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9228971962616823e-05,
-      "loss": 6.8991,
-      "step": 967
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9205607476635516e-05,
-      "loss": 6.8567,
-      "step": 968
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9182242990654206e-05,
-      "loss": 7.0149,
-      "step": 969
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9158878504672896e-05,
-      "loss": 6.7471,
-      "step": 970
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.913551401869159e-05,
-      "loss": 6.9262,
-      "step": 971
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 1.9112149532710283e-05,
-      "loss": 6.8414,
-      "step": 972
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.9088785046728973e-05,
-      "loss": 6.9483,
-      "step": 973
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.9065420560747663e-05,
-      "loss": 7.0209,
-      "step": 974
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.9042056074766356e-05,
-      "loss": 6.836,
-      "step": 975
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.9018691588785046e-05,
-      "loss": 6.704,
-      "step": 976
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.899532710280374e-05,
-      "loss": 7.1502,
-      "step": 977
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.897196261682243e-05,
-      "loss": 6.8727,
-      "step": 978
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8948598130841123e-05,
-      "loss": 6.775,
-      "step": 979
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8925233644859813e-05,
-      "loss": 6.7186,
-      "step": 980
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8901869158878507e-05,
-      "loss": 6.5942,
-      "step": 981
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8878504672897197e-05,
-      "loss": 6.7332,
-      "step": 982
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8855140186915887e-05,
-      "loss": 7.0078,
-      "step": 983
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.883177570093458e-05,
-      "loss": 6.8427,
-      "step": 984
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8808411214953273e-05,
-      "loss": 6.7045,
-      "step": 985
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8785046728971963e-05,
-      "loss": 6.9084,
-      "step": 986
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8761682242990653e-05,
-      "loss": 6.9645,
-      "step": 987
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8738317757009347e-05,
-      "loss": 6.7983,
-      "step": 988
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.871495327102804e-05,
-      "loss": 7.131,
-      "step": 989
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 1.8691588785046727e-05,
-      "loss": 6.7642,
-      "step": 990
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.866822429906542e-05,
-      "loss": 6.6372,
-      "step": 991
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8644859813084114e-05,
-      "loss": 7.0229,
-      "step": 992
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8621495327102807e-05,
-      "loss": 6.6841,
-      "step": 993
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8598130841121494e-05,
-      "loss": 6.7718,
-      "step": 994
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8574766355140187e-05,
-      "loss": 6.6926,
-      "step": 995
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.855140186915888e-05,
-      "loss": 6.3464,
-      "step": 996
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.852803738317757e-05,
-      "loss": 6.3116,
-      "step": 997
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.850467289719626e-05,
-      "loss": 7.1582,
-      "step": 998
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8481308411214954e-05,
-      "loss": 6.6423,
-      "step": 999
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8457943925233647e-05,
-      "loss": 6.5148,
-      "step": 1000
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 6.737658977508545,
-      "eval_runtime": 941.4412,
-      "eval_samples_per_second": 2.806,
-      "eval_steps_per_second": 0.352,
-      "eval_wer": 1.501637002202512,
-      "step": 1000
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8434579439252337e-05,
-      "loss": 7.1038,
-      "step": 1001
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8411214953271027e-05,
-      "loss": 6.8758,
-      "step": 1002
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.838785046728972e-05,
-      "loss": 6.9171,
-      "step": 1003
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.836448598130841e-05,
-      "loss": 6.9867,
-      "step": 1004
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8341121495327104e-05,
-      "loss": 7.0943,
-      "step": 1005
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8317757009345794e-05,
-      "loss": 7.0355,
-      "step": 1006
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1.8294392523364487e-05,
-      "loss": 6.8502,
-      "step": 1007
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8271028037383177e-05,
-      "loss": 6.8548,
-      "step": 1008
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.824766355140187e-05,
-      "loss": 6.8544,
-      "step": 1009
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.822429906542056e-05,
-      "loss": 6.8583,
-      "step": 1010
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.820093457943925e-05,
-      "loss": 6.8146,
-      "step": 1011
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8177570093457944e-05,
-      "loss": 6.9968,
-      "step": 1012
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8154205607476638e-05,
-      "loss": 6.7158,
-      "step": 1013
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8130841121495328e-05,
-      "loss": 6.9972,
-      "step": 1014
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8107476635514018e-05,
-      "loss": 6.9357,
-      "step": 1015
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.808411214953271e-05,
-      "loss": 6.7677,
-      "step": 1016
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8060747663551404e-05,
-      "loss": 6.8397,
-      "step": 1017
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.803738317757009e-05,
-      "loss": 6.8022,
-      "step": 1018
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.8014018691588784e-05,
-      "loss": 7.1418,
-      "step": 1019
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.7990654205607478e-05,
-      "loss": 6.9871,
-      "step": 1020
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.796728971962617e-05,
-      "loss": 6.8691,
-      "step": 1021
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.7943925233644858e-05,
-      "loss": 6.7965,
-      "step": 1022
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.792056074766355e-05,
-      "loss": 6.6605,
-      "step": 1023
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.7897196261682245e-05,
-      "loss": 6.7858,
-      "step": 1024
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 1.7873831775700935e-05,
-      "loss": 6.9358,
-      "step": 1025
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7850467289719625e-05,
-      "loss": 6.7704,
-      "step": 1026
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7827102803738318e-05,
-      "loss": 6.9262,
-      "step": 1027
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.780373831775701e-05,
-      "loss": 7.0061,
-      "step": 1028
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.77803738317757e-05,
-      "loss": 6.8394,
-      "step": 1029
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.775700934579439e-05,
-      "loss": 6.92,
-      "step": 1030
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7733644859813085e-05,
-      "loss": 6.8467,
-      "step": 1031
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7710280373831775e-05,
-      "loss": 6.8494,
-      "step": 1032
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7686915887850468e-05,
-      "loss": 6.8035,
-      "step": 1033
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7663551401869158e-05,
-      "loss": 6.6572,
-      "step": 1034
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.764018691588785e-05,
-      "loss": 6.8449,
-      "step": 1035
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.761682242990654e-05,
-      "loss": 6.9249,
-      "step": 1036
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7593457943925235e-05,
-      "loss": 6.6205,
-      "step": 1037
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7570093457943925e-05,
-      "loss": 6.8802,
-      "step": 1038
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7546728971962615e-05,
-      "loss": 6.8796,
-      "step": 1039
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.752336448598131e-05,
-      "loss": 6.8767,
-      "step": 1040
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7500000000000002e-05,
-      "loss": 7.0034,
-      "step": 1041
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7476635514018692e-05,
-      "loss": 6.6806,
-      "step": 1042
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 1.7453271028037382e-05,
-      "loss": 6.673,
-      "step": 1043
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7429906542056075e-05,
-      "loss": 6.7023,
-      "step": 1044
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.740654205607477e-05,
-      "loss": 6.8465,
-      "step": 1045
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.738317757009346e-05,
-      "loss": 6.7385,
-      "step": 1046
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.735981308411215e-05,
-      "loss": 6.696,
-      "step": 1047
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7336448598130842e-05,
-      "loss": 6.3817,
-      "step": 1048
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7313084112149535e-05,
-      "loss": 6.3691,
-      "step": 1049
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7289719626168222e-05,
-      "loss": 6.7965,
-      "step": 1050
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7266355140186915e-05,
-      "loss": 7.1985,
-      "step": 1051
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.724299065420561e-05,
-      "loss": 6.7922,
-      "step": 1052
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.72196261682243e-05,
-      "loss": 6.8542,
-      "step": 1053
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.719626168224299e-05,
-      "loss": 6.927,
-      "step": 1054
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7172897196261682e-05,
-      "loss": 6.9802,
-      "step": 1055
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7149532710280376e-05,
-      "loss": 6.8942,
-      "step": 1056
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7126168224299066e-05,
-      "loss": 7.057,
-      "step": 1057
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7102803738317756e-05,
-      "loss": 6.8625,
-      "step": 1058
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.707943925233645e-05,
-      "loss": 6.9211,
-      "step": 1059
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7056074766355142e-05,
-      "loss": 6.8816,
-      "step": 1060
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 1.7032710280373833e-05,
-      "loss": 6.9357,
-      "step": 1061
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.7009345794392523e-05,
-      "loss": 6.835,
-      "step": 1062
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6985981308411216e-05,
-      "loss": 7.113,
-      "step": 1063
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6962616822429906e-05,
-      "loss": 6.7592,
-      "step": 1064
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.69392523364486e-05,
-      "loss": 7.0179,
-      "step": 1065
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.691588785046729e-05,
-      "loss": 6.6335,
-      "step": 1066
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6892523364485983e-05,
-      "loss": 7.1261,
-      "step": 1067
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6869158878504673e-05,
-      "loss": 6.8373,
-      "step": 1068
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6845794392523366e-05,
-      "loss": 6.8848,
-      "step": 1069
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6822429906542056e-05,
-      "loss": 6.9108,
-      "step": 1070
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6799065420560746e-05,
-      "loss": 6.7475,
-      "step": 1071
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.677570093457944e-05,
-      "loss": 6.6454,
-      "step": 1072
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6752336448598133e-05,
-      "loss": 6.8809,
-      "step": 1073
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6728971962616823e-05,
-      "loss": 6.873,
-      "step": 1074
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6705607476635513e-05,
-      "loss": 6.7922,
-      "step": 1075
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6682242990654206e-05,
-      "loss": 6.9001,
-      "step": 1076
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.66588785046729e-05,
-      "loss": 6.7483,
-      "step": 1077
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.6635514018691586e-05,
-      "loss": 6.8103,
-      "step": 1078
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 1.661214953271028e-05,
-      "loss": 6.9621,
-      "step": 1079
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6588785046728973e-05,
-      "loss": 6.7952,
-      "step": 1080
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6565420560747667e-05,
-      "loss": 6.8259,
-      "step": 1081
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6542056074766353e-05,
-      "loss": 6.8101,
-      "step": 1082
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6518691588785047e-05,
-      "loss": 6.7697,
-      "step": 1083
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.649532710280374e-05,
-      "loss": 6.9418,
-      "step": 1084
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.647196261682243e-05,
-      "loss": 6.7724,
-      "step": 1085
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.644859813084112e-05,
-      "loss": 6.956,
-      "step": 1086
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6425233644859813e-05,
-      "loss": 7.0609,
-      "step": 1087
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6401869158878507e-05,
-      "loss": 6.9645,
-      "step": 1088
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6378504672897197e-05,
-      "loss": 6.9978,
-      "step": 1089
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6355140186915887e-05,
-      "loss": 6.7865,
-      "step": 1090
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.633177570093458e-05,
-      "loss": 6.9561,
-      "step": 1091
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.630841121495327e-05,
-      "loss": 6.9596,
-      "step": 1092
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6285046728971964e-05,
-      "loss": 6.7403,
-      "step": 1093
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6261682242990654e-05,
-      "loss": 6.4898,
-      "step": 1094
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6238317757009347e-05,
-      "loss": 6.671,
-      "step": 1095
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.6214953271028037e-05,
-      "loss": 6.6625,
-      "step": 1096
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 1.619158878504673e-05,
-      "loss": 6.759,
-      "step": 1097
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.616822429906542e-05,
-      "loss": 7.0088,
-      "step": 1098
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.614485981308411e-05,
-      "loss": 6.5444,
-      "step": 1099
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.6121495327102804e-05,
-      "loss": 6.2035,
-      "step": 1100
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.6098130841121497e-05,
-      "loss": 6.789,
-      "step": 1101
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.6074766355140187e-05,
-      "loss": 7.0043,
-      "step": 1102
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.6051401869158877e-05,
-      "loss": 6.9837,
-      "step": 1103
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.602803738317757e-05,
-      "loss": 6.867,
-      "step": 1104
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.6004672897196264e-05,
-      "loss": 6.8113,
-      "step": 1105
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.598130841121495e-05,
-      "loss": 6.718,
-      "step": 1106
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5957943925233644e-05,
-      "loss": 6.8854,
-      "step": 1107
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5934579439252337e-05,
-      "loss": 6.8536,
-      "step": 1108
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.591121495327103e-05,
-      "loss": 6.908,
-      "step": 1109
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5887850467289717e-05,
-      "loss": 6.8715,
-      "step": 1110
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.586448598130841e-05,
-      "loss": 6.7722,
-      "step": 1111
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5841121495327104e-05,
-      "loss": 6.9108,
-      "step": 1112
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5817757009345794e-05,
-      "loss": 6.8096,
-      "step": 1113
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5794392523364484e-05,
-      "loss": 6.9669,
-      "step": 1114
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 1.5771028037383178e-05,
-      "loss": 6.9777,
-      "step": 1115
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.574766355140187e-05,
-      "loss": 6.6319,
-      "step": 1116
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.572429906542056e-05,
-      "loss": 6.7477,
-      "step": 1117
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.570093457943925e-05,
-      "loss": 6.8733,
-      "step": 1118
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5677570093457944e-05,
-      "loss": 6.9132,
-      "step": 1119
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5654205607476634e-05,
-      "loss": 6.8032,
-      "step": 1120
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5630841121495328e-05,
-      "loss": 6.8404,
-      "step": 1121
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5607476635514018e-05,
-      "loss": 6.8208,
-      "step": 1122
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.558411214953271e-05,
-      "loss": 6.6568,
-      "step": 1123
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.55607476635514e-05,
-      "loss": 6.7614,
-      "step": 1124
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5537383177570095e-05,
-      "loss": 6.8104,
-      "step": 1125
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5514018691588785e-05,
-      "loss": 6.7202,
-      "step": 1126
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5490654205607475e-05,
-      "loss": 6.8003,
-      "step": 1127
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5467289719626168e-05,
-      "loss": 7.1313,
-      "step": 1128
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.544392523364486e-05,
-      "loss": 6.8845,
-      "step": 1129
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.542056074766355e-05,
-      "loss": 6.9188,
-      "step": 1130
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.539719626168224e-05,
-      "loss": 6.9601,
-      "step": 1131
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 1.5373831775700935e-05,
-      "loss": 6.9369,
-      "step": 1132
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5350467289719628e-05,
-      "loss": 6.7624,
-      "step": 1133
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5327102803738315e-05,
-      "loss": 6.8038,
-      "step": 1134
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5303738317757008e-05,
-      "loss": 6.9014,
-      "step": 1135
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.52803738317757e-05,
-      "loss": 6.8223,
-      "step": 1136
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5257009345794393e-05,
-      "loss": 6.8039,
-      "step": 1137
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5233644859813083e-05,
-      "loss": 6.9394,
-      "step": 1138
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5210280373831775e-05,
-      "loss": 6.6821,
-      "step": 1139
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5186915887850468e-05,
-      "loss": 6.8601,
-      "step": 1140
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.516355140186916e-05,
-      "loss": 6.7954,
-      "step": 1141
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5140186915887848e-05,
-      "loss": 7.0452,
-      "step": 1142
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5116822429906542e-05,
-      "loss": 6.9025,
-      "step": 1143
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5093457943925234e-05,
-      "loss": 6.8907,
-      "step": 1144
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5070093457943927e-05,
-      "loss": 6.6486,
-      "step": 1145
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5046728971962615e-05,
-      "loss": 6.7866,
-      "step": 1146
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5023364485981309e-05,
-      "loss": 6.7144,
-      "step": 1147
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.5e-05,
-      "loss": 6.5127,
-      "step": 1148
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.4976635514018692e-05,
-      "loss": 6.4288,
-      "step": 1149
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.4953271028037384e-05,
-      "loss": 6.4314,
-      "step": 1150
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4929906542056075e-05,
-      "loss": 6.9166,
-      "step": 1151
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4906542056074767e-05,
-      "loss": 7.0363,
-      "step": 1152
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4883177570093459e-05,
-      "loss": 6.8633,
-      "step": 1153
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.485981308411215e-05,
-      "loss": 6.9493,
-      "step": 1154
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4836448598130842e-05,
-      "loss": 7.0066,
-      "step": 1155
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4813084112149532e-05,
-      "loss": 6.9363,
-      "step": 1156
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4789719626168226e-05,
-      "loss": 6.8766,
-      "step": 1157
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4766355140186916e-05,
-      "loss": 7.0088,
-      "step": 1158
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4742990654205609e-05,
-      "loss": 6.7353,
-      "step": 1159
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4719626168224299e-05,
-      "loss": 6.7643,
-      "step": 1160
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4696261682242993e-05,
-      "loss": 6.9577,
-      "step": 1161
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4672897196261683e-05,
-      "loss": 6.7012,
-      "step": 1162
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4649532710280374e-05,
-      "loss": 6.7994,
-      "step": 1163
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4626168224299066e-05,
-      "loss": 6.7012,
-      "step": 1164
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4602803738317758e-05,
-      "loss": 6.8767,
-      "step": 1165
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.457943925233645e-05,
-      "loss": 6.8347,
-      "step": 1166
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4556074766355141e-05,
-      "loss": 7.0028,
-      "step": 1167
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.4532710280373833e-05,
-      "loss": 6.5566,
-      "step": 1168
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4509345794392524e-05,
-      "loss": 6.7657,
-      "step": 1169
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4485981308411214e-05,
-      "loss": 6.8744,
-      "step": 1170
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4462616822429908e-05,
-      "loss": 6.8628,
-      "step": 1171
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4439252336448598e-05,
-      "loss": 6.7861,
-      "step": 1172
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4415887850467291e-05,
-      "loss": 7.0183,
-      "step": 1173
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4392523364485981e-05,
-      "loss": 6.7683,
-      "step": 1174
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4369158878504675e-05,
-      "loss": 6.7431,
-      "step": 1175
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4345794392523365e-05,
-      "loss": 6.8766,
-      "step": 1176
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4322429906542058e-05,
-      "loss": 6.8354,
-      "step": 1177
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4299065420560748e-05,
-      "loss": 6.7087,
-      "step": 1178
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.427570093457944e-05,
-      "loss": 6.6534,
-      "step": 1179
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4252336448598131e-05,
-      "loss": 7.0726,
-      "step": 1180
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4228971962616823e-05,
-      "loss": 6.8476,
-      "step": 1181
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4205607476635515e-05,
-      "loss": 6.7303,
-      "step": 1182
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4182242990654207e-05,
-      "loss": 6.7329,
-      "step": 1183
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.4158878504672898e-05,
-      "loss": 6.7736,
-      "step": 1184
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.413551401869159e-05,
-      "loss": 7.1533,
-      "step": 1185
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.411214953271028e-05,
-      "loss": 6.7005,
-      "step": 1186
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.4088785046728973e-05,
-      "loss": 6.8716,
-      "step": 1187
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.4065420560747663e-05,
-      "loss": 6.8559,
-      "step": 1188
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.4042056074766357e-05,
-      "loss": 6.7757,
-      "step": 1189
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.4018691588785047e-05,
-      "loss": 6.7642,
-      "step": 1190
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.399532710280374e-05,
-      "loss": 6.7972,
-      "step": 1191
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.397196261682243e-05,
-      "loss": 6.5915,
-      "step": 1192
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3948598130841122e-05,
-      "loss": 6.6523,
-      "step": 1193
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3925233644859814e-05,
-      "loss": 6.6908,
-      "step": 1194
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3901869158878505e-05,
-      "loss": 6.5408,
-      "step": 1195
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3878504672897197e-05,
-      "loss": 6.5524,
-      "step": 1196
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3855140186915889e-05,
-      "loss": 6.4311,
-      "step": 1197
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.383177570093458e-05,
-      "loss": 6.622,
-      "step": 1198
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3808411214953272e-05,
-      "loss": 6.1649,
-      "step": 1199
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3785046728971962e-05,
-      "loss": 6.0119,
-      "step": 1200
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3761682242990655e-05,
-      "loss": 6.9058,
-      "step": 1201
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3738317757009345e-05,
-      "loss": 6.8021,
-      "step": 1202
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3714953271028039e-05,
-      "loss": 6.8847,
-      "step": 1203
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.3691588785046729e-05,
-      "loss": 6.7415,
-      "step": 1204
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3668224299065422e-05,
-      "loss": 6.8442,
-      "step": 1205
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3644859813084112e-05,
-      "loss": 6.7987,
-      "step": 1206
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3621495327102804e-05,
-      "loss": 6.9022,
-      "step": 1207
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3598130841121496e-05,
-      "loss": 6.9534,
-      "step": 1208
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3574766355140187e-05,
-      "loss": 6.835,
-      "step": 1209
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3551401869158879e-05,
-      "loss": 6.7285,
-      "step": 1210
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.352803738317757e-05,
-      "loss": 6.9321,
-      "step": 1211
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3504672897196263e-05,
-      "loss": 6.6922,
-      "step": 1212
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3481308411214954e-05,
-      "loss": 6.8289,
-      "step": 1213
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3457943925233644e-05,
-      "loss": 7.0545,
-      "step": 1214
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3434579439252338e-05,
-      "loss": 6.8402,
-      "step": 1215
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3411214953271028e-05,
-      "loss": 6.7454,
-      "step": 1216
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3387850467289721e-05,
-      "loss": 6.7598,
-      "step": 1217
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3364485981308411e-05,
-      "loss": 6.4891,
-      "step": 1218
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3341121495327104e-05,
-      "loss": 6.7329,
-      "step": 1219
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3317757009345794e-05,
-      "loss": 6.9508,
-      "step": 1220
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3294392523364486e-05,
-      "loss": 6.8433,
-      "step": 1221
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.3271028037383178e-05,
-      "loss": 6.9852,
-      "step": 1222
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.324766355140187e-05,
-      "loss": 7.0215,
-      "step": 1223
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3224299065420561e-05,
-      "loss": 6.7673,
-      "step": 1224
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3200934579439253e-05,
-      "loss": 6.8057,
-      "step": 1225
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3177570093457945e-05,
-      "loss": 6.7798,
-      "step": 1226
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3154205607476636e-05,
-      "loss": 6.8299,
-      "step": 1227
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3130841121495328e-05,
-      "loss": 6.9276,
-      "step": 1228
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.310747663551402e-05,
-      "loss": 6.8587,
-      "step": 1229
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.308411214953271e-05,
-      "loss": 7.1469,
-      "step": 1230
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3060747663551403e-05,
-      "loss": 6.9,
-      "step": 1231
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3037383177570093e-05,
-      "loss": 6.5898,
-      "step": 1232
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3014018691588787e-05,
-      "loss": 6.9599,
-      "step": 1233
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.2990654205607477e-05,
-      "loss": 6.7468,
-      "step": 1234
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.296728971962617e-05,
-      "loss": 7.0251,
-      "step": 1235
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.294392523364486e-05,
-      "loss": 6.8813,
-      "step": 1236
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.2920560747663552e-05,
-      "loss": 6.8648,
-      "step": 1237
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.2897196261682243e-05,
-      "loss": 6.6525,
-      "step": 1238
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.2873831775700935e-05,
-      "loss": 6.996,
-      "step": 1239
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2850467289719627e-05,
-      "loss": 6.93,
-      "step": 1240
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2827102803738318e-05,
-      "loss": 6.7741,
-      "step": 1241
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.280373831775701e-05,
-      "loss": 6.9528,
-      "step": 1242
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2780373831775702e-05,
-      "loss": 6.8065,
-      "step": 1243
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2757009345794392e-05,
-      "loss": 6.6115,
-      "step": 1244
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2733644859813085e-05,
-      "loss": 6.6697,
-      "step": 1245
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2710280373831775e-05,
-      "loss": 6.9055,
-      "step": 1246
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2686915887850469e-05,
-      "loss": 6.779,
-      "step": 1247
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2663551401869159e-05,
-      "loss": 6.6926,
-      "step": 1248
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2640186915887852e-05,
-      "loss": 6.5221,
-      "step": 1249
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2616822429906542e-05,
-      "loss": 6.7396,
-      "step": 1250
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2593457943925234e-05,
-      "loss": 6.9528,
-      "step": 1251
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2570093457943925e-05,
-      "loss": 6.8506,
-      "step": 1252
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2546728971962617e-05,
-      "loss": 6.8744,
-      "step": 1253
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2523364485981309e-05,
-      "loss": 6.9843,
-      "step": 1254
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.25e-05,
-      "loss": 7.1224,
-      "step": 1255
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2476635514018692e-05,
-      "loss": 6.7583,
-      "step": 1256
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2453271028037384e-05,
-      "loss": 6.7441,
-      "step": 1257
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2429906542056074e-05,
-      "loss": 6.8587,
-      "step": 1258
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2406542056074767e-05,
-      "loss": 6.8435,
-      "step": 1259
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2383177570093457e-05,
-      "loss": 6.9438,
-      "step": 1260
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.235981308411215e-05,
-      "loss": 6.875,
-      "step": 1261
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.233644859813084e-05,
-      "loss": 6.8595,
-      "step": 1262
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2313084112149534e-05,
-      "loss": 6.8138,
-      "step": 1263
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2289719626168224e-05,
-      "loss": 6.7342,
-      "step": 1264
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2266355140186916e-05,
-      "loss": 6.8854,
-      "step": 1265
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2242990654205608e-05,
-      "loss": 6.9195,
-      "step": 1266
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.22196261682243e-05,
-      "loss": 6.7517,
-      "step": 1267
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2196261682242991e-05,
-      "loss": 6.8759,
-      "step": 1268
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2172897196261683e-05,
-      "loss": 6.6875,
-      "step": 1269
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2149532710280374e-05,
-      "loss": 6.7925,
-      "step": 1270
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2126168224299066e-05,
-      "loss": 6.6688,
-      "step": 1271
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2102803738317758e-05,
-      "loss": 6.6145,
-      "step": 1272
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.207943925233645e-05,
-      "loss": 6.8868,
-      "step": 1273
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.205607476635514e-05,
-      "loss": 6.7076,
-      "step": 1274
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.2032710280373833e-05,
-      "loss": 6.824,
-      "step": 1275
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.2009345794392523e-05,
-      "loss": 6.9852,
-      "step": 1276
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1985981308411216e-05,
-      "loss": 7.0516,
-      "step": 1277
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1962616822429906e-05,
-      "loss": 6.8803,
-      "step": 1278
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.19392523364486e-05,
-      "loss": 6.9159,
-      "step": 1279
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.191588785046729e-05,
-      "loss": 6.7933,
-      "step": 1280
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1892523364485981e-05,
-      "loss": 6.6927,
-      "step": 1281
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1869158878504673e-05,
-      "loss": 6.7607,
-      "step": 1282
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1845794392523365e-05,
-      "loss": 6.9396,
-      "step": 1283
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1822429906542057e-05,
-      "loss": 6.7364,
-      "step": 1284
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1799065420560748e-05,
-      "loss": 6.8395,
-      "step": 1285
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.177570093457944e-05,
-      "loss": 6.957,
-      "step": 1286
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1752336448598132e-05,
-      "loss": 6.8491,
-      "step": 1287
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1728971962616822e-05,
-      "loss": 6.6642,
-      "step": 1288
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1705607476635515e-05,
-      "loss": 6.8575,
-      "step": 1289
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1682242990654205e-05,
-      "loss": 6.6707,
-      "step": 1290
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1658878504672898e-05,
-      "loss": 6.8011,
-      "step": 1291
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1635514018691588e-05,
-      "loss": 6.8822,
-      "step": 1292
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1612149532710282e-05,
-      "loss": 6.8102,
-      "step": 1293
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1588785046728972e-05,
-      "loss": 6.5879,
-      "step": 1294
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1565420560747664e-05,
-      "loss": 6.6685,
-      "step": 1295
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1542056074766355e-05,
-      "loss": 6.5262,
-      "step": 1296
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1518691588785047e-05,
-      "loss": 6.4487,
-      "step": 1297
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1495327102803739e-05,
-      "loss": 6.2968,
-      "step": 1298
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.147196261682243e-05,
-      "loss": 6.7793,
-      "step": 1299
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1448598130841122e-05,
-      "loss": 6.1899,
-      "step": 1300
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1425233644859814e-05,
-      "loss": 6.9,
-      "step": 1301
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1401869158878504e-05,
-      "loss": 6.9417,
-      "step": 1302
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1378504672897197e-05,
-      "loss": 6.9338,
-      "step": 1303
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1355140186915887e-05,
-      "loss": 6.9187,
-      "step": 1304
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.133177570093458e-05,
-      "loss": 6.8079,
-      "step": 1305
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.130841121495327e-05,
-      "loss": 6.9654,
-      "step": 1306
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1285046728971964e-05,
-      "loss": 6.7002,
-      "step": 1307
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1261682242990654e-05,
-      "loss": 6.9376,
-      "step": 1308
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1238317757009346e-05,
-      "loss": 6.6259,
-      "step": 1309
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1214953271028037e-05,
-      "loss": 6.7826,
-      "step": 1310
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.1191588785046729e-05,
-      "loss": 6.9096,
-      "step": 1311
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.116822429906542e-05,
-      "loss": 6.8284,
-      "step": 1312
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.1144859813084113e-05,
-      "loss": 6.7662,
-      "step": 1313
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.1121495327102804e-05,
-      "loss": 6.6924,
-      "step": 1314
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.1098130841121496e-05,
-      "loss": 7.0971,
-      "step": 1315
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.1074766355140186e-05,
-      "loss": 6.7222,
-      "step": 1316
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.105140186915888e-05,
-      "loss": 6.6248,
-      "step": 1317
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.102803738317757e-05,
-      "loss": 6.9189,
-      "step": 1318
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.1004672897196263e-05,
-      "loss": 6.7038,
-      "step": 1319
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0981308411214953e-05,
-      "loss": 6.7651,
-      "step": 1320
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0957943925233646e-05,
-      "loss": 6.6245,
-      "step": 1321
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0934579439252336e-05,
-      "loss": 6.7014,
-      "step": 1322
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.091121495327103e-05,
-      "loss": 6.917,
-      "step": 1323
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.088785046728972e-05,
-      "loss": 6.9843,
-      "step": 1324
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0864485981308411e-05,
-      "loss": 6.9012,
-      "step": 1325
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0841121495327103e-05,
-      "loss": 6.7189,
-      "step": 1326
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0817757009345795e-05,
-      "loss": 6.5935,
-      "step": 1327
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0794392523364486e-05,
-      "loss": 6.8825,
-      "step": 1328
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.0771028037383178e-05,
-      "loss": 6.9153,
-      "step": 1329
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.074766355140187e-05,
-      "loss": 6.7837,
-      "step": 1330
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0724299065420561e-05,
-      "loss": 6.6775,
-      "step": 1331
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0700934579439251e-05,
-      "loss": 6.7742,
-      "step": 1332
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0677570093457945e-05,
-      "loss": 6.9887,
-      "step": 1333
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0654205607476635e-05,
-      "loss": 6.9344,
-      "step": 1334
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0630841121495328e-05,
-      "loss": 6.625,
-      "step": 1335
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0607476635514018e-05,
-      "loss": 6.783,
-      "step": 1336
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0584112149532712e-05,
-      "loss": 6.9321,
-      "step": 1337
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0560747663551402e-05,
-      "loss": 6.792,
-      "step": 1338
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0537383177570093e-05,
-      "loss": 6.8872,
-      "step": 1339
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0514018691588785e-05,
-      "loss": 6.895,
-      "step": 1340
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0490654205607477e-05,
-      "loss": 6.9005,
-      "step": 1341
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0467289719626168e-05,
-      "loss": 6.726,
-      "step": 1342
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.044392523364486e-05,
-      "loss": 6.7971,
-      "step": 1343
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0420560747663552e-05,
-      "loss": 6.6783,
-      "step": 1344
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0397196261682244e-05,
-      "loss": 6.6607,
-      "step": 1345
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.0373831775700934e-05,
-      "loss": 6.7,
-      "step": 1346
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0350467289719627e-05,
-      "loss": 6.7139,
-      "step": 1347
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0327102803738317e-05,
-      "loss": 6.2283,
-      "step": 1348
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.030373831775701e-05,
-      "loss": 6.5001,
-      "step": 1349
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.02803738317757e-05,
-      "loss": 6.0824,
-      "step": 1350
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0257009345794394e-05,
-      "loss": 7.0136,
-      "step": 1351
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0233644859813084e-05,
-      "loss": 6.9984,
-      "step": 1352
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0210280373831776e-05,
-      "loss": 6.8829,
-      "step": 1353
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0186915887850467e-05,
-      "loss": 6.7388,
-      "step": 1354
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0163551401869159e-05,
-      "loss": 6.7791,
-      "step": 1355
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.014018691588785e-05,
-      "loss": 6.9606,
-      "step": 1356
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0116822429906542e-05,
-      "loss": 6.7186,
-      "step": 1357
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0093457943925234e-05,
-      "loss": 6.9715,
-      "step": 1358
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0070093457943926e-05,
-      "loss": 7.0262,
-      "step": 1359
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0046728971962616e-05,
-      "loss": 6.9211,
-      "step": 1360
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.0023364485981309e-05,
-      "loss": 6.8823,
-      "step": 1361
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 9.999999999999999e-06,
-      "loss": 6.8308,
-      "step": 1362
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 9.976635514018693e-06,
-      "loss": 6.7276,
-      "step": 1363
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 9.953271028037383e-06,
-      "loss": 7.0151,
-      "step": 1364
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.929906542056076e-06,
-      "loss": 6.6719,
-      "step": 1365
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.906542056074766e-06,
-      "loss": 6.8039,
-      "step": 1366
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.883177570093458e-06,
-      "loss": 6.7744,
-      "step": 1367
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.85981308411215e-06,
-      "loss": 6.7465,
-      "step": 1368
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.836448598130841e-06,
-      "loss": 6.8745,
-      "step": 1369
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.813084112149533e-06,
-      "loss": 6.8596,
-      "step": 1370
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.789719626168224e-06,
-      "loss": 7.0121,
-      "step": 1371
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.766355140186916e-06,
-      "loss": 6.5877,
-      "step": 1372
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.742990654205608e-06,
-      "loss": 6.8829,
-      "step": 1373
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.7196261682243e-06,
-      "loss": 6.9915,
-      "step": 1374
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.696261682242991e-06,
-      "loss": 6.7518,
-      "step": 1375
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.672897196261681e-06,
-      "loss": 6.6339,
-      "step": 1376
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.649532710280375e-06,
-      "loss": 6.8386,
-      "step": 1377
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.626168224299065e-06,
-      "loss": 6.8796,
-      "step": 1378
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.602803738317758e-06,
-      "loss": 6.7376,
-      "step": 1379
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.579439252336448e-06,
-      "loss": 6.9024,
-      "step": 1380
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.556074766355141e-06,
-      "loss": 6.6556,
-      "step": 1381
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.532710280373831e-06,
-      "loss": 6.8581,
-      "step": 1382
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.509345794392523e-06,
-      "loss": 6.78,
-      "step": 1383
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.485981308411215e-06,
-      "loss": 6.9171,
-      "step": 1384
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.462616822429907e-06,
-      "loss": 6.8258,
-      "step": 1385
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.439252336448598e-06,
-      "loss": 6.919,
-      "step": 1386
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.41588785046729e-06,
-      "loss": 6.829,
-      "step": 1387
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.392523364485982e-06,
-      "loss": 6.716,
-      "step": 1388
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.369158878504673e-06,
-      "loss": 6.8679,
-      "step": 1389
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.345794392523363e-06,
-      "loss": 6.7308,
-      "step": 1390
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.322429906542057e-06,
-      "loss": 6.5672,
-      "step": 1391
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.299065420560747e-06,
-      "loss": 6.827,
-      "step": 1392
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.27570093457944e-06,
-      "loss": 6.6068,
-      "step": 1393
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.25233644859813e-06,
-      "loss": 6.5838,
-      "step": 1394
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.228971962616824e-06,
-      "loss": 6.8024,
-      "step": 1395
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.205607476635514e-06,
-      "loss": 6.7667,
-      "step": 1396
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.182242990654205e-06,
-      "loss": 6.5138,
-      "step": 1397
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.158878504672897e-06,
-      "loss": 6.5048,
-      "step": 1398
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.135514018691589e-06,
-      "loss": 6.3922,
-      "step": 1399
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.11214953271028e-06,
-      "loss": 6.4021,
-      "step": 1400
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 9.088785046728972e-06,
-      "loss": 6.9025,
-      "step": 1401
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 9.065420560747664e-06,
-      "loss": 6.8046,
-      "step": 1402
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 9.042056074766356e-06,
-      "loss": 6.7137,
-      "step": 1403
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 9.018691588785046e-06,
-      "loss": 6.7114,
-      "step": 1404
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.995327102803739e-06,
-      "loss": 6.7813,
-      "step": 1405
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.971962616822429e-06,
-      "loss": 7.1083,
-      "step": 1406
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.948598130841122e-06,
-      "loss": 6.9838,
-      "step": 1407
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.925233644859812e-06,
-      "loss": 6.8504,
-      "step": 1408
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.901869158878506e-06,
-      "loss": 6.7504,
-      "step": 1409
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.878504672897196e-06,
-      "loss": 6.7097,
-      "step": 1410
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.855140186915887e-06,
-      "loss": 6.6734,
-      "step": 1411
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.831775700934579e-06,
-      "loss": 6.7109,
-      "step": 1412
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.80841121495327e-06,
-      "loss": 6.9027,
-      "step": 1413
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.785046728971963e-06,
-      "loss": 6.7525,
-      "step": 1414
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.761682242990654e-06,
-      "loss": 6.8815,
-      "step": 1415
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.738317757009346e-06,
-      "loss": 7.0992,
-      "step": 1416
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.714953271028038e-06,
-      "loss": 6.9811,
-      "step": 1417
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 8.69158878504673e-06,
-      "loss": 6.8584,
-      "step": 1418
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.668224299065421e-06,
-      "loss": 6.8395,
-      "step": 1419
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.644859813084111e-06,
-      "loss": 6.7214,
-      "step": 1420
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.621495327102804e-06,
-      "loss": 6.9589,
-      "step": 1421
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.598130841121494e-06,
-      "loss": 6.9133,
-      "step": 1422
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.574766355140188e-06,
-      "loss": 6.7229,
-      "step": 1423
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.551401869158878e-06,
-      "loss": 6.7995,
-      "step": 1424
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.528037383177571e-06,
-      "loss": 6.9087,
-      "step": 1425
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.504672897196261e-06,
-      "loss": 6.9328,
-      "step": 1426
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.481308411214953e-06,
-      "loss": 6.7455,
-      "step": 1427
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.457943925233645e-06,
-      "loss": 6.8455,
-      "step": 1428
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.434579439252336e-06,
-      "loss": 6.884,
-      "step": 1429
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.411214953271028e-06,
-      "loss": 7.0846,
-      "step": 1430
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.38785046728972e-06,
-      "loss": 6.9135,
-      "step": 1431
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.364485981308411e-06,
-      "loss": 6.9324,
-      "step": 1432
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.341121495327103e-06,
-      "loss": 6.9555,
-      "step": 1433
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.317757009345793e-06,
-      "loss": 6.8336,
-      "step": 1434
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.294392523364487e-06,
-      "loss": 6.9833,
-      "step": 1435
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 8.271028037383177e-06,
-      "loss": 6.6424,
-      "step": 1436
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.24766355140187e-06,
-      "loss": 6.9918,
-      "step": 1437
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.22429906542056e-06,
-      "loss": 6.8168,
-      "step": 1438
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.200934579439253e-06,
-      "loss": 6.7281,
-      "step": 1439
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.177570093457943e-06,
-      "loss": 6.7251,
-      "step": 1440
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.154205607476635e-06,
-      "loss": 6.9147,
-      "step": 1441
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.130841121495327e-06,
-      "loss": 6.8296,
-      "step": 1442
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.107476635514018e-06,
-      "loss": 6.7516,
-      "step": 1443
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.08411214953271e-06,
-      "loss": 6.7435,
-      "step": 1444
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.060747663551402e-06,
-      "loss": 6.6818,
-      "step": 1445
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.037383177570094e-06,
-      "loss": 6.8852,
-      "step": 1446
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 8.014018691588785e-06,
-      "loss": 6.6071,
-      "step": 1447
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.990654205607475e-06,
-      "loss": 6.4095,
-      "step": 1448
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.967289719626169e-06,
-      "loss": 6.4623,
-      "step": 1449
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.943925233644859e-06,
-      "loss": 6.4755,
-      "step": 1450
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.920560747663552e-06,
-      "loss": 6.9414,
-      "step": 1451
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.897196261682242e-06,
-      "loss": 6.9657,
-      "step": 1452
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 7.873831775700936e-06,
-      "loss": 6.7071,
-      "step": 1453
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.850467289719626e-06,
-      "loss": 6.7686,
-      "step": 1454
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.827102803738317e-06,
-      "loss": 6.7602,
-      "step": 1455
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.803738317757009e-06,
-      "loss": 6.8466,
-      "step": 1456
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.7803738317757e-06,
-      "loss": 6.9463,
-      "step": 1457
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.757009345794392e-06,
-      "loss": 6.9118,
-      "step": 1458
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.733644859813084e-06,
-      "loss": 6.8144,
-      "step": 1459
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.710280373831776e-06,
-      "loss": 6.8214,
-      "step": 1460
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.686915887850467e-06,
-      "loss": 6.8199,
-      "step": 1461
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.663551401869157e-06,
-      "loss": 6.8728,
-      "step": 1462
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.64018691588785e-06,
-      "loss": 6.7393,
-      "step": 1463
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.616822429906542e-06,
-      "loss": 6.7191,
-      "step": 1464
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.593457943925234e-06,
-      "loss": 6.8913,
-      "step": 1465
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.570093457943924e-06,
-      "loss": 6.7908,
-      "step": 1466
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.546728971962617e-06,
-      "loss": 6.9624,
-      "step": 1467
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.523364485981308e-06,
-      "loss": 6.7987,
-      "step": 1468
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.5e-06,
-      "loss": 6.6072,
-      "step": 1469
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.476635514018692e-06,
-      "loss": 6.6908,
-      "step": 1470
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 7.453271028037384e-06,
-      "loss": 6.8686,
-      "step": 1471
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.429906542056075e-06,
-      "loss": 6.7008,
-      "step": 1472
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.406542056074766e-06,
-      "loss": 6.6635,
-      "step": 1473
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.383177570093458e-06,
-      "loss": 6.5574,
-      "step": 1474
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.3598130841121496e-06,
-      "loss": 6.9079,
-      "step": 1475
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.336448598130841e-06,
-      "loss": 6.874,
-      "step": 1476
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.313084112149533e-06,
-      "loss": 7.0416,
-      "step": 1477
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.289719626168225e-06,
-      "loss": 6.638,
-      "step": 1478
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.266355140186916e-06,
-      "loss": 6.7438,
-      "step": 1479
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.242990654205607e-06,
-      "loss": 6.677,
-      "step": 1480
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.219626168224299e-06,
-      "loss": 6.9239,
-      "step": 1481
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.196261682242991e-06,
-      "loss": 6.9341,
-      "step": 1482
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.172897196261682e-06,
-      "loss": 6.8849,
-      "step": 1483
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.149532710280374e-06,
-      "loss": 6.5439,
-      "step": 1484
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.126168224299066e-06,
-      "loss": 6.9419,
-      "step": 1485
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.1028037383177574e-06,
-      "loss": 6.8689,
-      "step": 1486
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.079439252336449e-06,
-      "loss": 6.7977,
-      "step": 1487
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.05607476635514e-06,
-      "loss": 6.7809,
-      "step": 1488
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.032710280373832e-06,
-      "loss": 6.9552,
-      "step": 1489
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 7.009345794392523e-06,
-      "loss": 6.5815,
-      "step": 1490
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.985981308411215e-06,
-      "loss": 6.8619,
-      "step": 1491
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.962616822429907e-06,
-      "loss": 6.9539,
-      "step": 1492
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.9392523364485985e-06,
-      "loss": 6.593,
-      "step": 1493
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.91588785046729e-06,
-      "loss": 6.8966,
-      "step": 1494
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.892523364485981e-06,
-      "loss": 6.6498,
-      "step": 1495
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.869158878504673e-06,
-      "loss": 6.961,
-      "step": 1496
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.8457943925233644e-06,
-      "loss": 6.6218,
-      "step": 1497
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.822429906542056e-06,
-      "loss": 6.6435,
-      "step": 1498
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.799065420560748e-06,
-      "loss": 6.4273,
-      "step": 1499
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.7757009345794396e-06,
-      "loss": 6.3715,
-      "step": 1500
-    },
-    {
-      "epoch": 0.84,
-      "eval_loss": 6.701315402984619,
-      "eval_runtime": 947.3483,
-      "eval_samples_per_second": 2.789,
-      "eval_steps_per_second": 0.349,
-      "eval_wer": 1.7750858186003136,
-      "step": 1500
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.752336448598131e-06,
-      "loss": 6.7157,
-      "step": 1501
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.728971962616822e-06,
-      "loss": 6.9714,
-      "step": 1502
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.705607476635514e-06,
-      "loss": 6.8154,
-      "step": 1503
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.6822429906542055e-06,
-      "loss": 6.8313,
-      "step": 1504
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.658878504672897e-06,
-      "loss": 6.8636,
-      "step": 1505
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.635514018691589e-06,
-      "loss": 6.9083,
-      "step": 1506
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.612149532710281e-06,
-      "loss": 6.8204,
-      "step": 1507
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.588785046728972e-06,
-      "loss": 6.7485,
-      "step": 1508
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.565420560747664e-06,
-      "loss": 6.9086,
-      "step": 1509
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.542056074766355e-06,
-      "loss": 6.7954,
-      "step": 1510
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.518691588785047e-06,
-      "loss": 6.7352,
-      "step": 1511
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.495327102803738e-06,
-      "loss": 6.7133,
-      "step": 1512
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.47196261682243e-06,
-      "loss": 6.558,
-      "step": 1513
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.448598130841122e-06,
-      "loss": 6.8093,
-      "step": 1514
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.425233644859813e-06,
-      "loss": 6.6,
-      "step": 1515
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.401869158878505e-06,
-      "loss": 6.8491,
-      "step": 1516
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.378504672897196e-06,
-      "loss": 6.747,
-      "step": 1517
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.355140186915888e-06,
-      "loss": 6.759,
-      "step": 1518
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.331775700934579e-06,
-      "loss": 6.7585,
-      "step": 1519
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.308411214953271e-06,
-      "loss": 6.8533,
-      "step": 1520
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.285046728971963e-06,
-      "loss": 6.6984,
-      "step": 1521
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.2616822429906544e-06,
-      "loss": 6.855,
-      "step": 1522
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.238317757009346e-06,
-      "loss": 6.8599,
-      "step": 1523
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.214953271028037e-06,
-      "loss": 6.8842,
-      "step": 1524
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.191588785046729e-06,
-      "loss": 6.7022,
-      "step": 1525
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.16822429906542e-06,
-      "loss": 6.585,
-      "step": 1526
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.144859813084112e-06,
-      "loss": 6.8371,
-      "step": 1527
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.121495327102804e-06,
-      "loss": 6.6685,
-      "step": 1528
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.0981308411214955e-06,
-      "loss": 6.739,
-      "step": 1529
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.074766355140187e-06,
-      "loss": 6.7389,
-      "step": 1530
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.051401869158879e-06,
-      "loss": 6.6918,
-      "step": 1531
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.02803738317757e-06,
-      "loss": 6.5078,
-      "step": 1532
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 6.0046728971962615e-06,
-      "loss": 6.8115,
-      "step": 1533
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.981308411214953e-06,
-      "loss": 6.7498,
-      "step": 1534
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.957943925233645e-06,
-      "loss": 6.8758,
-      "step": 1535
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.934579439252337e-06,
-      "loss": 6.7364,
-      "step": 1536
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.911214953271028e-06,
-      "loss": 6.7019,
-      "step": 1537
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.88785046728972e-06,
-      "loss": 6.9665,
-      "step": 1538
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.864485981308411e-06,
-      "loss": 6.7439,
-      "step": 1539
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.8411214953271025e-06,
-      "loss": 6.7318,
-      "step": 1540
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.817757009345794e-06,
-      "loss": 6.6489,
-      "step": 1541
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.794392523364486e-06,
-      "loss": 6.5396,
-      "step": 1542
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.771028037383178e-06,
-      "loss": 6.7614,
-      "step": 1543
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.747663551401869e-06,
-      "loss": 6.7203,
-      "step": 1544
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.724299065420561e-06,
-      "loss": 6.7767,
-      "step": 1545
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.700934579439252e-06,
-      "loss": 6.6129,
-      "step": 1546
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.677570093457944e-06,
-      "loss": 6.7072,
-      "step": 1547
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.654205607476635e-06,
-      "loss": 6.5945,
-      "step": 1548
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.630841121495327e-06,
-      "loss": 6.7242,
-      "step": 1549
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.607476635514019e-06,
-      "loss": 6.1861,
-      "step": 1550
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.58411214953271e-06,
-      "loss": 6.9487,
-      "step": 1551
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.560747663551402e-06,
-      "loss": 6.9354,
-      "step": 1552
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.537383177570093e-06,
-      "loss": 6.7221,
-      "step": 1553
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.514018691588785e-06,
-      "loss": 6.5671,
-      "step": 1554
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.490654205607476e-06,
-      "loss": 6.8678,
-      "step": 1555
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.467289719626168e-06,
-      "loss": 6.937,
-      "step": 1556
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.44392523364486e-06,
-      "loss": 6.6915,
-      "step": 1557
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.4205607476635515e-06,
-      "loss": 6.7696,
-      "step": 1558
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.397196261682243e-06,
-      "loss": 6.686,
-      "step": 1559
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.373831775700935e-06,
-      "loss": 6.9102,
-      "step": 1560
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.350467289719626e-06,
-      "loss": 6.875,
-      "step": 1561
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.3271028037383174e-06,
-      "loss": 6.864,
-      "step": 1562
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.303738317757009e-06,
-      "loss": 6.7382,
-      "step": 1563
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.280373831775701e-06,
-      "loss": 6.5864,
-      "step": 1564
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.2570093457943925e-06,
-      "loss": 6.7846,
-      "step": 1565
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.233644859813084e-06,
-      "loss": 6.7557,
-      "step": 1566
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.210280373831776e-06,
-      "loss": 6.6975,
-      "step": 1567
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.186915887850467e-06,
-      "loss": 6.8594,
-      "step": 1568
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.1635514018691585e-06,
-      "loss": 6.8348,
-      "step": 1569
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.14018691588785e-06,
-      "loss": 6.87,
-      "step": 1570
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.116822429906542e-06,
-      "loss": 6.8449,
-      "step": 1571
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.093457943925234e-06,
-      "loss": 7.0682,
-      "step": 1572
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.070093457943925e-06,
-      "loss": 6.7918,
-      "step": 1573
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.046728971962617e-06,
-      "loss": 6.5322,
-      "step": 1574
-    },
     {
       "epoch": 0.88,
-      "learning_rate": 5.023364485981308e-06,
-      "loss": 6.4974,
-      "step": 1575
+      "learning_rate": 8.494897959183675e-06,
+      "loss": 6.8629,
+      "step": 786
     },
     {
       "epoch": 0.88,
-      "learning_rate": 4.9999999999999996e-06,
-      "loss": 6.6932,
-      "step": 1576
+      "learning_rate": 8.418367346938775e-06,
+      "loss": 6.7897,
+      "step": 787
     },
     {
       "epoch": 0.88,
-      "learning_rate": 4.976635514018691e-06,
-      "loss": 6.5953,
-      "step": 1577
+      "learning_rate": 8.341836734693878e-06,
+      "loss": 6.8748,
+      "step": 788
     },
     {
       "epoch": 0.88,
-      "learning_rate": 4.953271028037383e-06,
-      "loss": 6.7909,
-      "step": 1578
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.929906542056075e-06,
-      "loss": 6.7514,
-      "step": 1579
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.906542056074766e-06,
-      "loss": 6.824,
-      "step": 1580
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.883177570093458e-06,
-      "loss": 6.9313,
-      "step": 1581
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.85981308411215e-06,
-      "loss": 6.9126,
-      "step": 1582
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.836448598130841e-06,
-      "loss": 6.8406,
-      "step": 1583
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.813084112149532e-06,
-      "loss": 6.6236,
-      "step": 1584
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.789719626168224e-06,
-      "loss": 6.7743,
-      "step": 1585
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.766355140186916e-06,
-      "loss": 6.8217,
-      "step": 1586
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.7429906542056074e-06,
-      "loss": 6.7736,
-      "step": 1587
+      "learning_rate": 8.265306122448979e-06,
+      "loss": 6.7965,
+      "step": 789
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.719626168224299e-06,
-      "loss": 6.7384,
-      "step": 1588
+      "learning_rate": 8.188775510204082e-06,
+      "loss": 6.7865,
+      "step": 790
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.696261682242991e-06,
-      "loss": 6.7265,
-      "step": 1589
+      "learning_rate": 8.112244897959184e-06,
+      "loss": 6.7277,
+      "step": 791
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.672897196261682e-06,
-      "loss": 6.7732,
-      "step": 1590
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 6.6408,
+      "step": 792
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.649532710280373e-06,
-      "loss": 6.6609,
-      "step": 1591
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 6.7423,
+      "step": 793
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.626168224299065e-06,
-      "loss": 6.7198,
-      "step": 1592
+      "learning_rate": 7.88265306122449e-06,
+      "loss": 6.7093,
+      "step": 794
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.602803738317757e-06,
-      "loss": 6.5661,
-      "step": 1593
+      "learning_rate": 7.806122448979592e-06,
+      "loss": 6.8157,
+      "step": 795
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.5794392523364485e-06,
-      "loss": 6.5046,
-      "step": 1594
+      "learning_rate": 7.729591836734694e-06,
+      "loss": 6.7312,
+      "step": 796
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.55607476635514e-06,
-      "loss": 6.7806,
-      "step": 1595
+      "learning_rate": 7.653061224489796e-06,
+      "loss": 6.7111,
+      "step": 797
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.532710280373832e-06,
-      "loss": 6.6918,
-      "step": 1596
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.509345794392523e-06,
-      "loss": 6.5291,
-      "step": 1597
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.4859813084112145e-06,
-      "loss": 6.7057,
-      "step": 1598
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.462616822429906e-06,
-      "loss": 7.0633,
-      "step": 1599
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.439252336448598e-06,
-      "loss": 6.3798,
-      "step": 1600
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.4158878504672896e-06,
-      "loss": 6.9662,
-      "step": 1601
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.392523364485981e-06,
-      "loss": 6.6787,
-      "step": 1602
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.369158878504673e-06,
-      "loss": 6.9373,
-      "step": 1603
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.345794392523365e-06,
-      "loss": 6.9744,
-      "step": 1604
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.3224299065420555e-06,
-      "loss": 6.6731,
-      "step": 1605
+      "learning_rate": 7.5765306122448974e-06,
+      "loss": 6.7191,
+      "step": 798
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.299065420560747e-06,
-      "loss": 6.8186,
-      "step": 1606
+      "learning_rate": 7.5e-06,
+      "loss": 6.8694,
+      "step": 799
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.275700934579439e-06,
-      "loss": 6.8788,
-      "step": 1607
+      "learning_rate": 7.423469387755102e-06,
+      "loss": 6.3313,
+      "step": 800
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.252336448598131e-06,
-      "loss": 6.6214,
-      "step": 1608
+      "learning_rate": 7.346938775510204e-06,
+      "loss": 6.91,
+      "step": 801
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.228971962616822e-06,
-      "loss": 6.9421,
-      "step": 1609
+      "learning_rate": 7.270408163265306e-06,
+      "loss": 6.849,
+      "step": 802
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.205607476635514e-06,
-      "loss": 6.7981,
-      "step": 1610
+      "learning_rate": 7.193877551020408e-06,
+      "loss": 6.9105,
+      "step": 803
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.182242990654206e-06,
-      "loss": 6.8711,
-      "step": 1611
+      "learning_rate": 7.117346938775511e-06,
+      "loss": 6.9051,
+      "step": 804
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.158878504672897e-06,
-      "loss": 6.9173,
-      "step": 1612
+      "learning_rate": 7.0408163265306125e-06,
+      "loss": 6.6685,
+      "step": 805
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.135514018691588e-06,
-      "loss": 6.9506,
-      "step": 1613
+      "learning_rate": 6.964285714285714e-06,
+      "loss": 6.9164,
+      "step": 806
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.11214953271028e-06,
-      "loss": 6.7826,
-      "step": 1614
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 4.088785046728972e-06,
-      "loss": 6.7785,
-      "step": 1615
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 4.065420560747663e-06,
-      "loss": 6.7141,
-      "step": 1616
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 4.042056074766355e-06,
-      "loss": 6.8168,
-      "step": 1617
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 4.018691588785047e-06,
-      "loss": 6.8274,
-      "step": 1618
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 3.995327102803738e-06,
-      "loss": 6.7787,
-      "step": 1619
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 3.971962616822429e-06,
-      "loss": 6.7911,
-      "step": 1620
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 3.948598130841121e-06,
-      "loss": 6.9676,
-      "step": 1621
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 3.925233644859813e-06,
-      "loss": 6.7094,
-      "step": 1622
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 3.9018691588785045e-06,
-      "loss": 6.8304,
-      "step": 1623
+      "learning_rate": 6.887755102040816e-06,
+      "loss": 6.8981,
+      "step": 807
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.878504672897196e-06,
-      "loss": 6.5296,
-      "step": 1624
+      "learning_rate": 6.811224489795918e-06,
+      "loss": 6.7168,
+      "step": 808
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.855140186915888e-06,
-      "loss": 6.9151,
-      "step": 1625
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 6.8595,
+      "step": 809
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.831775700934579e-06,
-      "loss": 6.9664,
-      "step": 1626
+      "learning_rate": 6.658163265306123e-06,
+      "loss": 6.8056,
+      "step": 810
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.808411214953271e-06,
-      "loss": 6.8585,
-      "step": 1627
+      "learning_rate": 6.581632653061225e-06,
+      "loss": 6.9164,
+      "step": 811
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.785046728971962e-06,
-      "loss": 6.7189,
-      "step": 1628
+      "learning_rate": 6.505102040816327e-06,
+      "loss": 6.9122,
+      "step": 812
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.761682242990654e-06,
-      "loss": 6.8238,
-      "step": 1629
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 6.7648,
+      "step": 813
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.738317757009346e-06,
-      "loss": 6.7295,
-      "step": 1630
+      "learning_rate": 6.352040816326531e-06,
+      "loss": 6.845,
+      "step": 814
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.7149532710280376e-06,
-      "loss": 6.711,
-      "step": 1631
+      "learning_rate": 6.275510204081633e-06,
+      "loss": 6.8755,
+      "step": 815
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.691588785046729e-06,
-      "loss": 7.0368,
-      "step": 1632
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.6682242990654206e-06,
-      "loss": 6.6051,
-      "step": 1633
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.6448598130841123e-06,
-      "loss": 6.8239,
-      "step": 1634
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.6214953271028036e-06,
-      "loss": 6.791,
-      "step": 1635
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.5981308411214953e-06,
-      "loss": 6.7491,
-      "step": 1636
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.574766355140187e-06,
-      "loss": 6.6947,
-      "step": 1637
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.5514018691588787e-06,
-      "loss": 6.8842,
-      "step": 1638
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.52803738317757e-06,
-      "loss": 6.5936,
-      "step": 1639
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.5046728971962617e-06,
-      "loss": 6.7034,
-      "step": 1640
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.4813084112149534e-06,
-      "loss": 6.8694,
-      "step": 1641
+      "learning_rate": 6.198979591836735e-06,
+      "loss": 6.7565,
+      "step": 816
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.457943925233645e-06,
-      "loss": 6.6715,
-      "step": 1642
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 6.9427,
+      "step": 817
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.4345794392523364e-06,
-      "loss": 6.7285,
-      "step": 1643
+      "learning_rate": 6.045918367346939e-06,
+      "loss": 6.8183,
+      "step": 818
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.411214953271028e-06,
-      "loss": 6.5949,
-      "step": 1644
+      "learning_rate": 5.969387755102041e-06,
+      "loss": 6.7349,
+      "step": 819
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.3878504672897198e-06,
-      "loss": 6.7281,
-      "step": 1645
+      "learning_rate": 5.892857142857143e-06,
+      "loss": 6.8932,
+      "step": 820
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.364485981308411e-06,
-      "loss": 6.665,
-      "step": 1646
+      "learning_rate": 5.816326530612245e-06,
+      "loss": 6.9249,
+      "step": 821
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.3411214953271028e-06,
-      "loss": 6.4592,
-      "step": 1647
+      "learning_rate": 5.739795918367347e-06,
+      "loss": 6.8097,
+      "step": 822
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.3177570093457945e-06,
-      "loss": 6.7938,
-      "step": 1648
+      "learning_rate": 5.6632653061224486e-06,
+      "loss": 6.9243,
+      "step": 823
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.294392523364486e-06,
-      "loss": 6.6691,
-      "step": 1649
+      "learning_rate": 5.586734693877551e-06,
+      "loss": 6.7965,
+      "step": 824
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.2710280373831774e-06,
-      "loss": 6.1677,
-      "step": 1650
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.247663551401869e-06,
-      "loss": 6.7899,
-      "step": 1651
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.224299065420561e-06,
-      "loss": 6.8457,
-      "step": 1652
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.2009345794392525e-06,
-      "loss": 6.8378,
-      "step": 1653
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.177570093457944e-06,
-      "loss": 6.7115,
-      "step": 1654
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.1542056074766355e-06,
-      "loss": 6.5828,
-      "step": 1655
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.1308411214953272e-06,
-      "loss": 6.8242,
-      "step": 1656
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.1074766355140185e-06,
-      "loss": 6.9079,
-      "step": 1657
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.08411214953271e-06,
-      "loss": 6.7004,
-      "step": 1658
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.060747663551402e-06,
-      "loss": 6.6191,
-      "step": 1659
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.0373831775700936e-06,
-      "loss": 6.8245,
-      "step": 1660
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.014018691588785e-06,
-      "loss": 6.7542,
-      "step": 1661
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 2.9906542056074766e-06,
-      "loss": 6.5764,
-      "step": 1662
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 2.9672897196261683e-06,
-      "loss": 6.674,
-      "step": 1663
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 2.94392523364486e-06,
-      "loss": 6.861,
-      "step": 1664
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 2.9205607476635513e-06,
-      "loss": 6.8747,
-      "step": 1665
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 2.897196261682243e-06,
-      "loss": 6.7927,
-      "step": 1666
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 7.0408,
+      "step": 825
     },
     {
       "epoch": 0.93,
-      "learning_rate": 2.8738317757009347e-06,
-      "loss": 6.8287,
-      "step": 1667
+      "learning_rate": 5.433673469387755e-06,
+      "loss": 6.9394,
+      "step": 826
     },
     {
       "epoch": 0.93,
-      "learning_rate": 2.850467289719626e-06,
-      "loss": 6.7287,
-      "step": 1668
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 2.8271028037383177e-06,
-      "loss": 6.6719,
-      "step": 1669
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 2.8037383177570094e-06,
-      "loss": 6.949,
-      "step": 1670
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 2.780373831775701e-06,
-      "loss": 6.7625,
-      "step": 1671
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 2.7570093457943923e-06,
-      "loss": 7.008,
-      "step": 1672
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 6.8263,
+      "step": 827
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.733644859813084e-06,
-      "loss": 6.8049,
-      "step": 1673
+      "epoch": 0.93,
+      "learning_rate": 5.280612244897959e-06,
+      "loss": 6.7378,
+      "step": 828
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.7102803738317757e-06,
-      "loss": 6.9483,
-      "step": 1674
+      "epoch": 0.93,
+      "learning_rate": 5.204081632653062e-06,
+      "loss": 6.7529,
+      "step": 829
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.6869158878504674e-06,
-      "loss": 7.0996,
-      "step": 1675
+      "epoch": 0.93,
+      "learning_rate": 5.127551020408164e-06,
+      "loss": 6.8206,
+      "step": 830
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.6635514018691587e-06,
-      "loss": 6.6412,
-      "step": 1676
+      "epoch": 0.93,
+      "learning_rate": 5.0510204081632655e-06,
+      "loss": 6.7545,
+      "step": 831
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.6401869158878504e-06,
-      "loss": 6.658,
-      "step": 1677
+      "epoch": 0.93,
+      "learning_rate": 4.9744897959183674e-06,
+      "loss": 6.7889,
+      "step": 832
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.616822429906542e-06,
-      "loss": 6.6648,
-      "step": 1678
+      "epoch": 0.93,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 6.8035,
+      "step": 833
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.5934579439252334e-06,
-      "loss": 6.8104,
-      "step": 1679
+      "epoch": 0.93,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 6.8525,
+      "step": 834
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.570093457943925e-06,
-      "loss": 6.7633,
-      "step": 1680
+      "learning_rate": 4.744897959183674e-06,
+      "loss": 6.8054,
+      "step": 835
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.546728971962617e-06,
-      "loss": 6.62,
-      "step": 1681
+      "learning_rate": 4.668367346938776e-06,
+      "loss": 6.9522,
+      "step": 836
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.5233644859813085e-06,
-      "loss": 6.5355,
-      "step": 1682
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 6.8324,
+      "step": 837
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.4999999999999998e-06,
-      "loss": 6.788,
-      "step": 1683
+      "learning_rate": 4.51530612244898e-06,
+      "loss": 6.9383,
+      "step": 838
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.4766355140186915e-06,
-      "loss": 6.7666,
-      "step": 1684
+      "learning_rate": 4.438775510204082e-06,
+      "loss": 6.7394,
+      "step": 839
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.453271028037383e-06,
-      "loss": 6.8733,
-      "step": 1685
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.429906542056075e-06,
-      "loss": 6.8013,
-      "step": 1686
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.406542056074766e-06,
-      "loss": 6.8757,
-      "step": 1687
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.383177570093458e-06,
-      "loss": 6.9649,
-      "step": 1688
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.3598130841121496e-06,
-      "loss": 6.8738,
-      "step": 1689
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.336448598130841e-06,
-      "loss": 6.5257,
-      "step": 1690
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.3130841121495325e-06,
-      "loss": 6.8036,
-      "step": 1691
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.2897196261682242e-06,
-      "loss": 6.7962,
-      "step": 1692
+      "learning_rate": 4.3622448979591836e-06,
+      "loss": 6.7162,
+      "step": 840
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.266355140186916e-06,
-      "loss": 6.951,
-      "step": 1693
+      "epoch": 0.94,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 6.8783,
+      "step": 841
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.2429906542056072e-06,
-      "loss": 6.8359,
-      "step": 1694
+      "epoch": 0.94,
+      "learning_rate": 4.209183673469387e-06,
+      "loss": 6.7736,
+      "step": 842
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.219626168224299e-06,
-      "loss": 6.719,
-      "step": 1695
+      "learning_rate": 4.132653061224489e-06,
+      "loss": 6.976,
+      "step": 843
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.1962616822429906e-06,
-      "loss": 6.6708,
-      "step": 1696
+      "learning_rate": 4.056122448979592e-06,
+      "loss": 6.7993,
+      "step": 844
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.1728971962616823e-06,
-      "loss": 6.418,
-      "step": 1697
+      "learning_rate": 3.979591836734694e-06,
+      "loss": 6.807,
+      "step": 845
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.1495327102803736e-06,
-      "loss": 6.6458,
-      "step": 1698
+      "learning_rate": 3.903061224489796e-06,
+      "loss": 6.7076,
+      "step": 846
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.1261682242990653e-06,
-      "loss": 6.5149,
-      "step": 1699
+      "learning_rate": 3.826530612244898e-06,
+      "loss": 6.5819,
+      "step": 847
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.102803738317757e-06,
-      "loss": 6.5905,
-      "step": 1700
+      "learning_rate": 3.75e-06,
+      "loss": 6.8109,
+      "step": 848
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.0794392523364483e-06,
-      "loss": 6.9096,
-      "step": 1701
+      "learning_rate": 3.673469387755102e-06,
+      "loss": 6.5935,
+      "step": 849
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.05607476635514e-06,
-      "loss": 6.8825,
-      "step": 1702
+      "learning_rate": 3.596938775510204e-06,
+      "loss": 6.3915,
+      "step": 850
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.0327102803738317e-06,
-      "loss": 6.8664,
-      "step": 1703
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.0093457943925234e-06,
-      "loss": 6.893,
-      "step": 1704
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.9859813084112147e-06,
-      "loss": 6.8631,
-      "step": 1705
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.9626168224299064e-06,
-      "loss": 6.731,
-      "step": 1706
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.939252336448598e-06,
-      "loss": 6.8009,
-      "step": 1707
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.9158878504672894e-06,
-      "loss": 6.9211,
-      "step": 1708
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.892523364485981e-06,
-      "loss": 7.0041,
-      "step": 1709
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.869158878504673e-06,
-      "loss": 7.0845,
-      "step": 1710
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.8457943925233645e-06,
-      "loss": 6.6433,
-      "step": 1711
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.8224299065420562e-06,
-      "loss": 6.8527,
-      "step": 1712
+      "learning_rate": 3.5204081632653062e-06,
+      "loss": 6.8856,
+      "step": 851
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7990654205607477e-06,
-      "loss": 6.9782,
-      "step": 1713
+      "learning_rate": 3.443877551020408e-06,
+      "loss": 6.8172,
+      "step": 852
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7757009345794394e-06,
-      "loss": 6.8788,
-      "step": 1714
+      "learning_rate": 3.3673469387755105e-06,
+      "loss": 7.0145,
+      "step": 853
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7523364485981308e-06,
-      "loss": 6.6813,
-      "step": 1715
+      "learning_rate": 3.2908163265306124e-06,
+      "loss": 6.8908,
+      "step": 854
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7289719626168225e-06,
-      "loss": 6.8344,
-      "step": 1716
+      "learning_rate": 3.2142857142857143e-06,
+      "loss": 6.8144,
+      "step": 855
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.705607476635514e-06,
-      "loss": 7.07,
-      "step": 1717
+      "learning_rate": 3.1377551020408166e-06,
+      "loss": 6.9178,
+      "step": 856
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.6822429906542055e-06,
-      "loss": 6.9329,
-      "step": 1718
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 6.9697,
+      "step": 857
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.6588785046728972e-06,
-      "loss": 6.6725,
-      "step": 1719
+      "learning_rate": 2.9846938775510205e-06,
+      "loss": 6.902,
+      "step": 858
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.6355140186915887e-06,
-      "loss": 6.992,
-      "step": 1720
+      "learning_rate": 2.9081632653061224e-06,
+      "loss": 6.7325,
+      "step": 859
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.6121495327102804e-06,
-      "loss": 6.6669,
-      "step": 1721
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.588785046728972e-06,
-      "loss": 6.7484,
-      "step": 1722
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.5654205607476636e-06,
-      "loss": 6.8997,
-      "step": 1723
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.542056074766355e-06,
-      "loss": 6.8578,
-      "step": 1724
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.5186915887850468e-06,
-      "loss": 6.8508,
-      "step": 1725
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.4953271028037383e-06,
-      "loss": 6.7163,
-      "step": 1726
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.47196261682243e-06,
-      "loss": 6.8112,
-      "step": 1727
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.4485981308411215e-06,
-      "loss": 7.0273,
-      "step": 1728
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.425233644859813e-06,
-      "loss": 6.9244,
-      "step": 1729
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.4018691588785047e-06,
-      "loss": 6.8438,
-      "step": 1730
+      "learning_rate": 2.8316326530612243e-06,
+      "loss": 6.8723,
+      "step": 860
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3785046728971962e-06,
-      "loss": 6.558,
-      "step": 1731
+      "learning_rate": 2.7551020408163266e-06,
+      "loss": 7.0073,
+      "step": 861
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3551401869158879e-06,
-      "loss": 6.6695,
-      "step": 1732
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 6.9034,
+      "step": 862
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3317757009345794e-06,
-      "loss": 6.7048,
-      "step": 1733
+      "learning_rate": 2.602040816326531e-06,
+      "loss": 6.8114,
+      "step": 863
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.308411214953271e-06,
-      "loss": 6.8208,
-      "step": 1734
+      "learning_rate": 2.5255102040816328e-06,
+      "loss": 7.0092,
+      "step": 864
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2850467289719625e-06,
-      "loss": 6.9385,
-      "step": 1735
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 6.9172,
+      "step": 865
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2616822429906543e-06,
-      "loss": 6.908,
-      "step": 1736
+      "learning_rate": 2.372448979591837e-06,
+      "loss": 6.7697,
+      "step": 866
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2383177570093457e-06,
-      "loss": 6.4836,
-      "step": 1737
+      "learning_rate": 2.295918367346939e-06,
+      "loss": 6.9497,
+      "step": 867
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2149532710280374e-06,
-      "loss": 6.8577,
-      "step": 1738
+      "learning_rate": 2.219387755102041e-06,
+      "loss": 6.7357,
+      "step": 868
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.191588785046729e-06,
-      "loss": 6.7293,
-      "step": 1739
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.1682242990654204e-06,
-      "loss": 6.7548,
-      "step": 1740
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.1448598130841121e-06,
-      "loss": 6.6983,
-      "step": 1741
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.1214953271028036e-06,
-      "loss": 6.7675,
-      "step": 1742
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.0981308411214953e-06,
-      "loss": 6.7107,
-      "step": 1743
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.0747663551401868e-06,
-      "loss": 6.4002,
-      "step": 1744
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.0514018691588785e-06,
-      "loss": 6.5271,
-      "step": 1745
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.02803738317757e-06,
-      "loss": 6.6567,
-      "step": 1746
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.0046728971962617e-06,
-      "loss": 6.7087,
-      "step": 1747
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 9.813084112149532e-07,
-      "loss": 6.3945,
-      "step": 1748
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 6.853,
+      "step": 869
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.579439252336447e-07,
-      "loss": 6.333,
-      "step": 1749
+      "learning_rate": 2.0663265306122446e-06,
+      "loss": 6.8345,
+      "step": 870
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.345794392523365e-07,
-      "loss": 6.5274,
-      "step": 1750
+      "learning_rate": 1.989795918367347e-06,
+      "loss": 6.7548,
+      "step": 871
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.112149532710281e-07,
-      "loss": 6.743,
-      "step": 1751
+      "learning_rate": 1.913265306122449e-06,
+      "loss": 6.8532,
+      "step": 872
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.878504672897197e-07,
-      "loss": 6.6863,
-      "step": 1752
+      "learning_rate": 1.836734693877551e-06,
+      "loss": 6.8114,
+      "step": 873
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.644859813084113e-07,
-      "loss": 6.9507,
-      "step": 1753
+      "learning_rate": 1.7602040816326531e-06,
+      "loss": 6.9239,
+      "step": 874
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.411214953271028e-07,
-      "loss": 6.7592,
-      "step": 1754
+      "learning_rate": 1.6836734693877552e-06,
+      "loss": 6.9835,
+      "step": 875
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.177570093457944e-07,
-      "loss": 7.0295,
-      "step": 1755
+      "learning_rate": 1.6071428571428572e-06,
+      "loss": 6.7284,
+      "step": 876
     },
     {
       "epoch": 0.98,
-      "learning_rate": 7.94392523364486e-07,
-      "loss": 6.6155,
-      "step": 1756
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 6.7125,
+      "step": 877
     },
     {
       "epoch": 0.98,
-      "learning_rate": 7.710280373831776e-07,
-      "loss": 6.6454,
-      "step": 1757
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 7.476635514018691e-07,
-      "loss": 6.7024,
-      "step": 1758
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 7.242990654205607e-07,
-      "loss": 6.9926,
-      "step": 1759
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 7.009345794392523e-07,
-      "loss": 6.8504,
-      "step": 1760
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.775700934579439e-07,
-      "loss": 6.8889,
-      "step": 1761
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.542056074766355e-07,
-      "loss": 6.7199,
-      "step": 1762
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.308411214953271e-07,
-      "loss": 6.8088,
-      "step": 1763
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 6.074766355140187e-07,
-      "loss": 6.8932,
-      "step": 1764
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.841121495327102e-07,
-      "loss": 6.6074,
-      "step": 1765
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.607476635514018e-07,
-      "loss": 6.8411,
-      "step": 1766
+      "learning_rate": 1.4540816326530612e-06,
+      "loss": 6.8183,
+      "step": 878
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.373831775700934e-07,
-      "loss": 6.645,
-      "step": 1767
+      "learning_rate": 1.3775510204081633e-06,
+      "loss": 6.8289,
+      "step": 879
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.14018691588785e-07,
-      "loss": 6.5784,
-      "step": 1768
+      "learning_rate": 1.3010204081632654e-06,
+      "loss": 6.9183,
+      "step": 880
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.906542056074766e-07,
-      "loss": 6.731,
-      "step": 1769
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 6.6072,
+      "step": 881
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.6728971962616824e-07,
-      "loss": 6.6866,
-      "step": 1770
+      "learning_rate": 1.1479591836734695e-06,
+      "loss": 6.8833,
+      "step": 882
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.4392523364485984e-07,
-      "loss": 6.858,
-      "step": 1771
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 6.7338,
+      "step": 883
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.205607476635514e-07,
-      "loss": 6.6769,
-      "step": 1772
+      "learning_rate": 9.948979591836735e-07,
+      "loss": 6.8157,
+      "step": 884
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.97196261682243e-07,
-      "loss": 6.7402,
-      "step": 1773
+      "learning_rate": 9.183673469387755e-07,
+      "loss": 6.8095,
+      "step": 885
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.7383177570093457e-07,
-      "loss": 6.7128,
-      "step": 1774
+      "learning_rate": 8.418367346938776e-07,
+      "loss": 6.8044,
+      "step": 886
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.5046728971962617e-07,
-      "loss": 6.5639,
-      "step": 1775
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.2710280373831776e-07,
-      "loss": 6.7678,
-      "step": 1776
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.0373831775700936e-07,
-      "loss": 6.6281,
-      "step": 1777
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.803738317757009e-07,
-      "loss": 6.827,
-      "step": 1778
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.570093457943925e-07,
-      "loss": 6.8963,
-      "step": 1779
+      "learning_rate": 7.653061224489796e-07,
+      "loss": 6.5781,
+      "step": 887
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.3364485981308412e-07,
-      "loss": 6.633,
-      "step": 1780
+      "learning_rate": 6.887755102040817e-07,
+      "loss": 6.7909,
+      "step": 888
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.102803738317757e-07,
-      "loss": 6.4977,
-      "step": 1781
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 6.6131,
+      "step": 889
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.8691588785046729e-07,
-      "loss": 6.6355,
-      "step": 1782
+      "learning_rate": 5.357142857142857e-07,
+      "loss": 6.6244,
+      "step": 890
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.6355140186915888e-07,
-      "loss": 6.2981,
-      "step": 1783
+      "learning_rate": 4.5918367346938775e-07,
+      "loss": 6.4836,
+      "step": 891
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.4018691588785045e-07,
-      "loss": 5.8992,
-      "step": 1784
+      "learning_rate": 3.826530612244898e-07,
+      "loss": 6.2152,
+      "step": 892
     },
     {
       "epoch": 1.0,
-      "step": 1784,
+      "step": 892,
       "total_flos": 0.0,
-      "train_loss": 6.904792482542884,
-      "train_runtime": 8426.689,
-      "train_samples_per_second": 3.387,
-      "train_steps_per_second": 0.212
+      "train_loss": 6.990358243074118,
+      "train_runtime": 6205.6335,
+      "train_samples_per_second": 4.599,
+      "train_steps_per_second": 0.144
     }
   ],
-  "max_steps": 1784,
+  "max_steps": 892,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,