diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,5383 +2,2698 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 4.8117,
+      "learning_rate": 2e-08,
+      "loss": 4.7541,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.069,
+      "learning_rate": 4e-08,
+      "loss": 4.8927,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 2e-08,
-      "loss": 4.9247,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.8644,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4e-08,
-      "loss": 4.9264,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.8675,
       "step": 4
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.775,
+      "learning_rate": 8e-08,
+      "loss": 4.7998,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8e-08,
-      "loss": 4.8728,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.8669,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 4.8517,
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.8499,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.7681,
+      "epoch": 0.02,
+      "learning_rate": 1.4e-07,
+      "loss": 4.8967,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-07,
-      "loss": 4.9379,
+      "epoch": 0.02,
+      "learning_rate": 1.6e-07,
+      "loss": 4.8873,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6e-07,
-      "loss": 4.8324,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-07,
+      "loss": 4.7778,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.9107,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.8975,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.885,
+      "epoch": 0.03,
+      "learning_rate": 2.2e-07,
+      "loss": 4.8687,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-07,
-      "loss": 4.8775,
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.8289,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8306,
+      "epoch": 0.03,
+      "learning_rate": 2.6e-07,
+      "loss": 4.8372,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-07,
-      "loss": 4.7829,
+      "epoch": 0.03,
+      "learning_rate": 2.8e-07,
+      "loss": 4.7169,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8e-07,
-      "loss": 4.9149,
+      "epoch": 0.04,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.8774,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.9241,
+      "epoch": 0.04,
+      "learning_rate": 3.2e-07,
+      "loss": 4.8112,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2e-07,
-      "loss": 5.055,
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 4.9211,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 4.765,
+      "epoch": 0.04,
+      "learning_rate": 3.6e-07,
+      "loss": 4.7166,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 4.949,
+      "epoch": 0.04,
+      "learning_rate": 3.8e-07,
+      "loss": 4.9005,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-07,
-      "loss": 4.8868,
+      "epoch": 0.05,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.8957,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.816,
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 4.8022,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 4.8877,
+      "epoch": 0.05,
+      "learning_rate": 4.4e-07,
+      "loss": 4.9258,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-07,
-      "loss": 5.0342,
+      "epoch": 0.05,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 4.9286,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 4.9072,
+      "epoch": 0.06,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.717,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 5.0654,
+      "epoch": 0.06,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.9856,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 4.8262,
+      "epoch": 0.06,
+      "learning_rate": 5.2e-07,
+      "loss": 4.8427,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-07,
-      "loss": 4.7309,
+      "epoch": 0.06,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7164,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-07,
-      "loss": 4.7592,
+      "epoch": 0.07,
+      "learning_rate": 5.6e-07,
+      "loss": 4.6958,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.6e-07,
-      "loss": 4.9317,
+      "epoch": 0.07,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 4.8882,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.800000000000001e-07,
-      "loss": 4.9113,
+      "epoch": 0.07,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8631,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.7046,
+      "epoch": 0.07,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 4.71,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.200000000000001e-07,
-      "loss": 4.8189,
+      "epoch": 0.07,
+      "learning_rate": 6.4e-07,
+      "loss": 4.8112,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.4e-07,
-      "loss": 4.667,
+      "epoch": 0.08,
+      "learning_rate": 6.6e-07,
+      "loss": 4.7364,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.6e-07,
-      "loss": 4.8741,
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 4.8314,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.800000000000001e-07,
-      "loss": 4.8444,
+      "epoch": 0.08,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.7743,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 4.8907,
+      "epoch": 0.08,
+      "learning_rate": 7.2e-07,
+      "loss": 4.8398,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-07,
-      "loss": 4.8745,
+      "epoch": 0.09,
+      "learning_rate": 7.4e-07,
+      "loss": 4.7806,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.4e-07,
-      "loss": 5.0974,
+      "epoch": 0.09,
+      "learning_rate": 7.6e-07,
+      "loss": 4.9238,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.6e-07,
-      "loss": 4.8256,
+      "epoch": 0.09,
+      "learning_rate": 7.8e-07,
+      "loss": 4.701,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.8e-07,
-      "loss": 4.7171,
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.7911,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.913,
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 4.9134,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 4.7189,
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 4.6839,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 4.9229,
+      "epoch": 0.1,
+      "learning_rate": 8.6e-07,
+      "loss": 4.8494,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.6e-07,
-      "loss": 5.0259,
+      "epoch": 0.1,
+      "learning_rate": 8.8e-07,
+      "loss": 4.9975,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.8e-07,
-      "loss": 5.2111,
+      "epoch": 0.1,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 5.1271,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 5.1443,
+      "epoch": 0.11,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.0245,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.5136,
+      "epoch": 0.11,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 5.3001,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.6882,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 5.3946,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 5.5173,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 5.2908,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.5275,
+      "epoch": 0.11,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 4.6459,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.800000000000001e-07,
-      "loss": 4.6764,
+      "epoch": 0.12,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.5063,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.5496,
+      "epoch": 0.12,
+      "learning_rate": 1.02e-06,
+      "loss": 4.6776,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.02e-06,
-      "loss": 4.6228,
+      "epoch": 0.12,
+      "learning_rate": 1.04e-06,
+      "loss": 4.5814,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.04e-06,
-      "loss": 4.6146,
+      "epoch": 0.12,
+      "learning_rate": 1.06e-06,
+      "loss": 4.4042,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.06e-06,
-      "loss": 4.6572,
+      "epoch": 0.13,
+      "learning_rate": 1.08e-06,
+      "loss": 4.6195,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.08e-06,
-      "loss": 4.6467,
+      "epoch": 0.13,
+      "learning_rate": 1.1e-06,
+      "loss": 4.6391,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1e-06,
-      "loss": 4.8707,
+      "epoch": 0.13,
+      "learning_rate": 1.12e-06,
+      "loss": 4.7295,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.12e-06,
-      "loss": 4.6247,
+      "epoch": 0.13,
+      "learning_rate": 1.14e-06,
+      "loss": 4.6633,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.14e-06,
-      "loss": 4.4728,
+      "epoch": 0.13,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 4.4993,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1600000000000001e-06,
-      "loss": 4.6942,
+      "epoch": 0.14,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 4.5966,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1800000000000001e-06,
-      "loss": 4.6197,
+      "epoch": 0.14,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.536,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.5163,
+      "epoch": 0.14,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 4.4949,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 4.6636,
+      "epoch": 0.14,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 4.6052,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 4.417,
+      "epoch": 0.15,
+      "learning_rate": 1.26e-06,
+      "loss": 4.558,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.26e-06,
-      "loss": 4.6925,
+      "epoch": 0.15,
+      "learning_rate": 1.28e-06,
+      "loss": 4.4341,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.28e-06,
-      "loss": 4.4826,
+      "epoch": 0.15,
+      "learning_rate": 1.3e-06,
+      "loss": 4.5926,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3e-06,
-      "loss": 4.5045,
+      "epoch": 0.15,
+      "learning_rate": 1.32e-06,
+      "loss": 4.4335,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.32e-06,
-      "loss": 4.4526,
+      "epoch": 0.15,
+      "learning_rate": 1.34e-06,
+      "loss": 4.5063,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.34e-06,
-      "loss": 4.549,
+      "epoch": 0.16,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 4.4816,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3600000000000001e-06,
-      "loss": 4.7282,
+      "epoch": 0.16,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 4.6217,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 4.6231,
+      "epoch": 0.16,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.5979,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.5543,
+      "epoch": 0.16,
+      "learning_rate": 1.42e-06,
+      "loss": 4.4467,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.42e-06,
-      "loss": 4.6324,
+      "epoch": 0.17,
+      "learning_rate": 1.44e-06,
+      "loss": 4.5226,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.44e-06,
-      "loss": 4.3796,
+      "epoch": 0.17,
+      "learning_rate": 1.46e-06,
+      "loss": 4.5218,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.46e-06,
-      "loss": 4.6923,
+      "epoch": 0.17,
+      "learning_rate": 1.48e-06,
+      "loss": 4.4656,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.48e-06,
-      "loss": 4.5589,
+      "epoch": 0.17,
+      "learning_rate": 1.5e-06,
+      "loss": 4.5227,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5e-06,
-      "loss": 4.5119,
+      "epoch": 0.17,
+      "learning_rate": 1.52e-06,
+      "loss": 4.4625,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.52e-06,
-      "loss": 4.3468,
+      "epoch": 0.18,
+      "learning_rate": 1.54e-06,
+      "loss": 4.4515,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.54e-06,
-      "loss": 4.5559,
+      "epoch": 0.18,
+      "learning_rate": 1.56e-06,
+      "loss": 4.5694,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.56e-06,
-      "loss": 4.6462,
+      "epoch": 0.18,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 4.5223,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5800000000000001e-06,
-      "loss": 4.4336,
+      "epoch": 0.18,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.4574,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.6163,
+      "epoch": 0.19,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 4.5745,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6200000000000002e-06,
-      "loss": 4.5764,
+      "epoch": 0.19,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 4.4455,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 4.5511,
+      "epoch": 0.19,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 4.4914,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 4.4844,
+      "epoch": 0.19,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 4.4821,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6800000000000002e-06,
-      "loss": 4.5007,
+      "epoch": 0.2,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.5791,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 4.4969,
+      "epoch": 0.2,
+      "learning_rate": 1.72e-06,
+      "loss": 4.4611,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.72e-06,
-      "loss": 4.562,
+      "epoch": 0.2,
+      "learning_rate": 1.74e-06,
+      "loss": 4.6041,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.74e-06,
-      "loss": 4.4187,
+      "epoch": 0.2,
+      "learning_rate": 1.76e-06,
+      "loss": 4.5501,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.76e-06,
-      "loss": 4.6965,
+      "epoch": 0.2,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 4.4837,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 4.6423,
+      "epoch": 0.21,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.7265,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.8000000000000001e-06,
-      "loss": 4.3469,
+      "epoch": 0.21,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 4.5814,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 4.5379,
+      "epoch": 0.21,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 4.5799,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 4.8062,
+      "epoch": 0.21,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 4.5869,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8600000000000002e-06,
-      "loss": 4.7967,
+      "epoch": 0.22,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 4.7579,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 4.5775,
+      "epoch": 0.22,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.7943,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9000000000000002e-06,
-      "loss": 4.8837,
+      "epoch": 0.22,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.5681,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.6492,
+      "epoch": 0.22,
+      "learning_rate": 1.94e-06,
+      "loss": 4.8655,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.94e-06,
-      "loss": 4.793,
+      "epoch": 0.22,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 4.6563,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 4.3603,
+      "epoch": 0.23,
+      "learning_rate": 1.98e-06,
+      "loss": 4.4149,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.98e-06,
-      "loss": 4.3455,
+      "epoch": 0.23,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.3519,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.5046,
+      "epoch": 0.23,
+      "learning_rate": 2.02e-06,
+      "loss": 4.2156,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.02e-06,
-      "loss": 4.409,
+      "epoch": 0.23,
+      "learning_rate": 2.04e-06,
+      "loss": 4.2846,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.04e-06,
-      "loss": 4.2052,
+      "epoch": 0.24,
+      "learning_rate": 2.06e-06,
+      "loss": 4.2302,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.06e-06,
-      "loss": 4.4774,
+      "epoch": 0.24,
+      "learning_rate": 2.08e-06,
+      "loss": 4.2765,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.08e-06,
-      "loss": 4.4239,
+      "epoch": 0.24,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.3781,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.5902,
+      "epoch": 0.24,
+      "learning_rate": 2.12e-06,
+      "loss": 4.2992,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.12e-06,
-      "loss": 4.1572,
+      "epoch": 0.24,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 4.2881,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1400000000000003e-06,
-      "loss": 4.301,
+      "epoch": 0.25,
+      "learning_rate": 2.16e-06,
+      "loss": 4.3361,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.16e-06,
-      "loss": 4.5175,
+      "epoch": 0.25,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 4.3737,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 4.347,
+      "epoch": 0.25,
+      "learning_rate": 2.2e-06,
+      "loss": 4.3282,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2e-06,
-      "loss": 4.3326,
+      "epoch": 0.25,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 4.3385,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 4.2646,
+      "epoch": 0.26,
+      "learning_rate": 2.24e-06,
+      "loss": 4.3158,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.24e-06,
-      "loss": 4.3175,
+      "epoch": 0.26,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 4.3688,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2600000000000004e-06,
-      "loss": 4.1262,
+      "epoch": 0.26,
+      "learning_rate": 2.28e-06,
+      "loss": 4.4229,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.28e-06,
-      "loss": 4.4502,
+      "epoch": 0.26,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 4.3013,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3000000000000004e-06,
-      "loss": 4.2503,
+      "epoch": 0.26,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 4.4761,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 4.4425,
+      "epoch": 0.27,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 4.3552,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 4.4808,
+      "epoch": 0.27,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 4.3892,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.3600000000000003e-06,
-      "loss": 4.4266,
+      "epoch": 0.27,
+      "learning_rate": 2.38e-06,
+      "loss": 4.2258,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.38e-06,
-      "loss": 4.3035,
+      "epoch": 0.27,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.3049,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.248,
+      "epoch": 0.28,
+      "learning_rate": 2.42e-06,
+      "loss": 4.3383,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.42e-06,
-      "loss": 4.4539,
+      "epoch": 0.28,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 4.3089,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 4.4127,
+      "epoch": 0.28,
+      "learning_rate": 2.46e-06,
+      "loss": 4.4328,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.46e-06,
-      "loss": 4.4468,
+      "epoch": 0.28,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 4.4379,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4800000000000004e-06,
-      "loss": 4.5366,
+      "epoch": 0.28,
+      "learning_rate": 2.5e-06,
+      "loss": 4.3453,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3051,
+      "epoch": 0.29,
+      "learning_rate": 2.52e-06,
+      "loss": 4.3193,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.52e-06,
-      "loss": 4.2584,
-      "step": 129
+      "epoch": 0.29,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 4.2188,
+      "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5400000000000002e-06,
-      "loss": 4.4583,
+      "epoch": 0.29,
+      "learning_rate": 2.56e-06,
+      "loss": 4.3847,
       "step": 130
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.56e-06,
-      "loss": 4.3395,
+      "epoch": 0.29,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 4.2899,
       "step": 131
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 4.4609,
+      "epoch": 0.3,
+      "learning_rate": 2.6e-06,
+      "loss": 4.331,
       "step": 132
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6e-06,
-      "loss": 4.3807,
+      "epoch": 0.3,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 4.2797,
       "step": 133
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 4.413,
+      "epoch": 0.3,
+      "learning_rate": 2.64e-06,
+      "loss": 4.285,
       "step": 134
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.64e-06,
-      "loss": 4.2847,
+      "epoch": 0.3,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 4.4472,
       "step": 135
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6600000000000004e-06,
-      "loss": 4.2995,
+      "epoch": 0.3,
+      "learning_rate": 2.68e-06,
+      "loss": 4.3031,
       "step": 136
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.68e-06,
-      "loss": 4.6835,
+      "epoch": 0.31,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 4.4124,
       "step": 137
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 4.4243,
+      "epoch": 0.31,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 4.4394,
       "step": 138
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7200000000000002e-06,
-      "loss": 4.4537,
+      "epoch": 0.31,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 4.4579,
       "step": 139
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 4.2517,
+      "epoch": 0.31,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 4.4306,
       "step": 140
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 4.5315,
+      "epoch": 0.32,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 4.5488,
       "step": 141
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7800000000000005e-06,
-      "loss": 4.6131,
+      "epoch": 0.32,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.4791,
       "step": 142
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.5349,
+      "epoch": 0.32,
+      "learning_rate": 2.82e-06,
+      "loss": 4.3989,
       "step": 143
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.82e-06,
-      "loss": 4.4707,
+      "epoch": 0.32,
+      "learning_rate": 2.84e-06,
+      "loss": 4.4766,
       "step": 144
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.84e-06,
-      "loss": 4.5458,
+      "epoch": 0.33,
+      "learning_rate": 2.86e-06,
+      "loss": 4.3723,
       "step": 145
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.86e-06,
-      "loss": 4.5764,
+      "epoch": 0.33,
+      "learning_rate": 2.88e-06,
+      "loss": 4.4711,
       "step": 146
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.88e-06,
-      "loss": 4.6713,
+      "epoch": 0.33,
+      "learning_rate": 2.9e-06,
+      "loss": 4.7235,
       "step": 147
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9e-06,
-      "loss": 4.3646,
+      "epoch": 0.33,
+      "learning_rate": 2.92e-06,
+      "loss": 4.8624,
       "step": 148
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.92e-06,
-      "loss": 4.6639,
+      "epoch": 0.33,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.5986,
       "step": 149
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.3015,
+      "epoch": 0.34,
+      "learning_rate": 2.96e-06,
+      "loss": 4.7065,
       "step": 150
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.96e-06,
-      "loss": 4.4255,
+      "epoch": 0.34,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 4.2124,
       "step": 151
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 4.2821,
+      "epoch": 0.34,
+      "learning_rate": 3e-06,
+      "loss": 4.2878,
       "step": 152
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3e-06,
-      "loss": 4.3454,
+      "epoch": 0.34,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 4.3026,
       "step": 153
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 4.2121,
+      "epoch": 0.35,
+      "learning_rate": 3.04e-06,
+      "loss": 4.2101,
       "step": 154
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.04e-06,
-      "loss": 4.1285,
+      "epoch": 0.35,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 4.2986,
       "step": 155
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.0600000000000003e-06,
-      "loss": 4.2772,
+      "epoch": 0.35,
+      "learning_rate": 3.08e-06,
+      "loss": 4.1734,
       "step": 156
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.08e-06,
-      "loss": 4.4278,
+      "epoch": 0.35,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.2086,
       "step": 157
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1000000000000004e-06,
-      "loss": 4.5082,
+      "epoch": 0.35,
+      "learning_rate": 3.12e-06,
+      "loss": 4.3003,
       "step": 158
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.12e-06,
-      "loss": 4.5615,
+      "epoch": 0.36,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 4.3039,
       "step": 159
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1400000000000004e-06,
-      "loss": 4.2509,
+      "epoch": 0.36,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 4.3001,
       "step": 160
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1600000000000002e-06,
-      "loss": 4.2678,
+      "epoch": 0.36,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 4.3526,
       "step": 161
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 4.3571,
+      "epoch": 0.36,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.312,
       "step": 162
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.2991,
+      "epoch": 0.37,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 4.2679,
       "step": 163
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2200000000000005e-06,
-      "loss": 4.4275,
+      "epoch": 0.37,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 4.333,
       "step": 164
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 4.3572,
+      "epoch": 0.37,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 4.1979,
       "step": 165
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.2600000000000006e-06,
-      "loss": 4.3648,
+      "epoch": 0.37,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 4.2588,
       "step": 166
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 4.3222,
+      "epoch": 0.37,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 4.1736,
       "step": 167
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3000000000000006e-06,
-      "loss": 4.1074,
+      "epoch": 0.38,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 4.2609,
       "step": 168
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 4.095,
+      "epoch": 0.38,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 4.3197,
       "step": 169
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 4.2833,
+      "epoch": 0.38,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 4.3275,
       "step": 170
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3600000000000004e-06,
-      "loss": 4.4821,
+      "epoch": 0.38,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 4.3226,
       "step": 171
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 4.3928,
+      "epoch": 0.39,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.3706,
       "step": 172
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.3295,
+      "epoch": 0.39,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 4.1764,
       "step": 173
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 4.1746,
+      "epoch": 0.39,
+      "learning_rate": 3.44e-06,
+      "loss": 4.3279,
       "step": 174
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.44e-06,
-      "loss": 4.3026,
+      "epoch": 0.39,
+      "learning_rate": 3.46e-06,
+      "loss": 4.3795,
       "step": 175
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.46e-06,
-      "loss": 4.124,
+      "epoch": 0.39,
+      "learning_rate": 3.48e-06,
+      "loss": 4.257,
       "step": 176
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.48e-06,
-      "loss": 4.1713,
+      "epoch": 0.4,
+      "learning_rate": 3.5e-06,
+      "loss": 4.2217,
       "step": 177
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.5e-06,
-      "loss": 4.229,
+      "epoch": 0.4,
+      "learning_rate": 3.52e-06,
+      "loss": 4.3091,
       "step": 178
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.52e-06,
-      "loss": 4.3412,
+      "epoch": 0.4,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3617,
       "step": 179
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.54e-06,
-      "loss": 4.3153,
+      "epoch": 0.4,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 4.3626,
       "step": 180
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 4.3678,
+      "epoch": 0.41,
+      "learning_rate": 3.58e-06,
+      "loss": 4.292,
       "step": 181
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.58e-06,
-      "loss": 4.1591,
+      "epoch": 0.41,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 4.3384,
       "step": 182
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 4.396,
+      "epoch": 0.41,
+      "learning_rate": 3.62e-06,
+      "loss": 4.3467,
       "step": 183
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.62e-06,
-      "loss": 4.3347,
+      "epoch": 0.41,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 4.3526,
       "step": 184
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 4.361,
+      "epoch": 0.41,
+      "learning_rate": 3.66e-06,
+      "loss": 4.3269,
       "step": 185
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.66e-06,
-      "loss": 4.2574,
+      "epoch": 0.42,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 4.1944,
       "step": 186
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6800000000000003e-06,
-      "loss": 4.4495,
+      "epoch": 0.42,
+      "learning_rate": 3.7e-06,
+      "loss": 4.2104,
       "step": 187
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7e-06,
-      "loss": 4.1725,
+      "epoch": 0.42,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 4.1716,
       "step": 188
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7200000000000004e-06,
-      "loss": 4.2552,
+      "epoch": 0.42,
+      "learning_rate": 3.74e-06,
+      "loss": 4.2872,
       "step": 189
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.74e-06,
-      "loss": 4.426,
+      "epoch": 0.43,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 4.4067,
       "step": 190
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 4.418,
+      "epoch": 0.43,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.3407,
       "step": 191
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.375,
+      "epoch": 0.43,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.2671,
       "step": 192
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8000000000000005e-06,
-      "loss": 4.5633,
+      "epoch": 0.43,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 4.2976,
       "step": 193
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.820000000000001e-06,
-      "loss": 4.2975,
+      "epoch": 0.43,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.3868,
       "step": 194
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.3029,
+      "epoch": 0.44,
+      "learning_rate": 3.86e-06,
+      "loss": 4.4837,
       "step": 195
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.86e-06,
-      "loss": 4.5786,
+      "epoch": 0.44,
+      "learning_rate": 3.88e-06,
+      "loss": 4.451,
       "step": 196
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.88e-06,
-      "loss": 4.5452,
+      "epoch": 0.44,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 4.3893,
       "step": 197
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.900000000000001e-06,
-      "loss": 4.536,
+      "epoch": 0.44,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 4.6928,
       "step": 198
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 4.6819,
+      "epoch": 0.45,
+      "learning_rate": 3.94e-06,
+      "loss": 4.6071,
       "step": 199
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.94e-06,
-      "loss": 4.7643,
+      "epoch": 0.45,
+      "learning_rate": 3.96e-06,
+      "loss": 4.5822,
       "step": 200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.96e-06,
-      "loss": 4.1518,
+      "epoch": 0.45,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 4.317,
       "step": 201
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 4.2455,
+      "epoch": 0.45,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.254,
       "step": 202
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.1358,
+      "epoch": 0.46,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.2586,
       "step": 203
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.1243,
+      "epoch": 0.46,
+      "learning_rate": 4.04e-06,
+      "loss": 4.247,
       "step": 204
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.04e-06,
-      "loss": 4.0428,
+      "epoch": 0.46,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 4.3217,
       "step": 205
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.060000000000001e-06,
-      "loss": 4.1063,
+      "epoch": 0.46,
+      "learning_rate": 4.08e-06,
+      "loss": 4.3277,
       "step": 206
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.08e-06,
-      "loss": 4.2254,
+      "epoch": 0.46,
+      "learning_rate": 4.1e-06,
+      "loss": 4.1827,
       "step": 207
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.1e-06,
-      "loss": 4.0463,
+      "epoch": 0.47,
+      "learning_rate": 4.12e-06,
+      "loss": 4.2797,
       "step": 208
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.12e-06,
-      "loss": 4.2815,
+      "epoch": 0.47,
+      "learning_rate": 4.14e-06,
+      "loss": 4.3171,
       "step": 209
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.14e-06,
-      "loss": 4.2311,
+      "epoch": 0.47,
+      "learning_rate": 4.16e-06,
+      "loss": 4.2913,
       "step": 210
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.16e-06,
-      "loss": 4.1705,
+      "epoch": 0.47,
+      "learning_rate": 4.18e-06,
+      "loss": 4.1687,
       "step": 211
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.18e-06,
-      "loss": 4.2728,
+      "epoch": 0.48,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.2174,
       "step": 212
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.248,
+      "epoch": 0.48,
+      "learning_rate": 4.22e-06,
+      "loss": 4.1299,
       "step": 213
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.22e-06,
-      "loss": 4.3013,
+      "epoch": 0.48,
+      "learning_rate": 4.24e-06,
+      "loss": 4.2068,
       "step": 214
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.24e-06,
-      "loss": 4.1759,
+      "epoch": 0.48,
+      "learning_rate": 4.26e-06,
+      "loss": 4.2561,
       "step": 215
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.26e-06,
-      "loss": 4.3035,
+      "epoch": 0.48,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 4.2275,
       "step": 216
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2800000000000005e-06,
-      "loss": 4.2697,
+      "epoch": 0.49,
+      "learning_rate": 4.3e-06,
+      "loss": 4.2821,
       "step": 217
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.3e-06,
-      "loss": 4.1918,
+      "epoch": 0.49,
+      "learning_rate": 4.32e-06,
+      "loss": 4.2398,
       "step": 218
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.32e-06,
-      "loss": 4.2817,
+      "epoch": 0.49,
+      "learning_rate": 4.34e-06,
+      "loss": 4.1627,
       "step": 219
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.34e-06,
-      "loss": 4.2502,
+      "epoch": 0.49,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 4.2745,
       "step": 220
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 4.2403,
+      "epoch": 0.5,
+      "learning_rate": 4.38e-06,
+      "loss": 4.2645,
       "step": 221
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.38e-06,
-      "loss": 4.2914,
+      "epoch": 0.5,
+      "learning_rate": 4.4e-06,
+      "loss": 4.2325,
       "step": 222
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4e-06,
-      "loss": 4.3498,
+      "epoch": 0.5,
+      "learning_rate": 4.42e-06,
+      "loss": 4.2634,
       "step": 223
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.42e-06,
-      "loss": 4.2636,
+      "epoch": 0.5,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 4.1924,
       "step": 224
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 4.3429,
+      "epoch": 0.5,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 4.3595,
       "step": 225
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 4.508,
+      "epoch": 0.51,
+      "learning_rate": 4.48e-06,
+      "loss": 4.2554,
       "step": 226
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.48e-06,
-      "loss": 4.2465,
+      "epoch": 0.51,
+      "learning_rate": 4.5e-06,
+      "loss": 4.3735,
       "step": 227
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5e-06,
-      "loss": 4.1636,
+      "epoch": 0.51,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 4.3111,
       "step": 228
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.520000000000001e-06,
-      "loss": 4.1575,
+      "epoch": 0.51,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 4.1945,
       "step": 229
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 4.254,
+      "epoch": 0.52,
+      "learning_rate": 4.56e-06,
+      "loss": 4.2758,
       "step": 230
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.56e-06,
-      "loss": 4.21,
+      "epoch": 0.52,
+      "learning_rate": 4.58e-06,
+      "loss": 4.3566,
       "step": 231
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.58e-06,
-      "loss": 4.3432,
+      "epoch": 0.52,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.1538,
       "step": 232
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 4.2356,
+      "epoch": 0.52,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 4.171,
       "step": 233
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.620000000000001e-06,
-      "loss": 4.3205,
+      "epoch": 0.52,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 4.2496,
       "step": 234
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.6400000000000005e-06,
-      "loss": 4.4007,
+      "epoch": 0.53,
+      "learning_rate": 4.66e-06,
+      "loss": 4.3324,
       "step": 235
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.66e-06,
-      "loss": 4.2438,
+      "epoch": 0.53,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 4.2674,
       "step": 236
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.680000000000001e-06,
-      "loss": 4.2833,
+      "epoch": 0.53,
+      "learning_rate": 4.7e-06,
+      "loss": 4.2991,
       "step": 237
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.7e-06,
-      "loss": 4.469,
+      "epoch": 0.53,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 4.2457,
       "step": 238
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.7200000000000005e-06,
-      "loss": 4.2948,
+      "epoch": 0.54,
+      "learning_rate": 4.74e-06,
+      "loss": 4.1922,
       "step": 239
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.74e-06,
-      "loss": 4.3899,
+      "epoch": 0.54,
+      "learning_rate": 4.76e-06,
+      "loss": 4.3,
       "step": 240
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.76e-06,
-      "loss": 4.3733,
+      "epoch": 0.54,
+      "learning_rate": 4.78e-06,
+      "loss": 4.274,
       "step": 241
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.78e-06,
-      "loss": 4.5385,
+      "epoch": 0.54,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.3015,
       "step": 242
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3879,
+      "epoch": 0.54,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 4.5195,
       "step": 243
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 4.3414,
+      "epoch": 0.55,
+      "learning_rate": 4.84e-06,
+      "loss": 4.3539,
       "step": 244
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.84e-06,
-      "loss": 4.5273,
+      "epoch": 0.55,
+      "learning_rate": 4.86e-06,
+      "loss": 4.3367,
       "step": 245
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.86e-06,
-      "loss": 4.4047,
+      "epoch": 0.55,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 4.2667,
       "step": 246
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 4.7717,
+      "epoch": 0.55,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.188,
       "step": 247
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 4.7985,
+      "epoch": 0.56,
+      "learning_rate": 4.92e-06,
+      "loss": 4.5605,
       "step": 248
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.92e-06,
-      "loss": 4.6015,
+      "epoch": 0.56,
+      "learning_rate": 4.94e-06,
+      "loss": 4.6662,
       "step": 249
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.94e-06,
-      "loss": 4.2719,
+      "epoch": 0.56,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 4.6607,
       "step": 250
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 4.4124,
+      "epoch": 0.56,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2306,
       "step": 251
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.1304,
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 4.266,
       "step": 252
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5e-06,
-      "loss": 4.1539,
+      "epoch": 0.57,
+      "learning_rate": 5.02e-06,
+      "loss": 4.3091,
       "step": 253
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5.02e-06,
-      "loss": 4.312,
+      "epoch": 0.57,
+      "learning_rate": 5.04e-06,
+      "loss": 4.156,
       "step": 254
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.04e-06,
-      "loss": 4.1491,
+      "epoch": 0.57,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 4.2564,
       "step": 255
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.060000000000001e-06,
-      "loss": 4.1523,
+      "epoch": 0.57,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 4.2243,
       "step": 256
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.0800000000000005e-06,
-      "loss": 4.3502,
+      "epoch": 0.58,
+      "learning_rate": 5.1e-06,
+      "loss": 4.1986,
       "step": 257
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1e-06,
-      "loss": 4.1693,
+      "epoch": 0.58,
+      "learning_rate": 5.12e-06,
+      "loss": 4.2144,
       "step": 258
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.12e-06,
-      "loss": 4.1367,
+      "epoch": 0.58,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 4.2165,
       "step": 259
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 4.2931,
+      "epoch": 0.58,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 4.1611,
       "step": 260
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 4.3277,
+      "epoch": 0.59,
+      "learning_rate": 5.18e-06,
+      "loss": 4.2733,
       "step": 261
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.18e-06,
-      "loss": 4.0501,
+      "epoch": 0.59,
+      "learning_rate": 5.2e-06,
+      "loss": 4.2103,
       "step": 262
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.2e-06,
-      "loss": 4.1875,
+      "epoch": 0.59,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 4.3427,
       "step": 263
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 4.1807,
+      "epoch": 0.59,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 4.2133,
       "step": 264
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 4.3104,
+      "epoch": 0.59,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 4.2544,
       "step": 265
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 4.3502,
+      "epoch": 0.6,
+      "learning_rate": 5.28e-06,
+      "loss": 4.225,
       "step": 266
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.28e-06,
-      "loss": 4.1756,
+      "epoch": 0.6,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 4.254,
       "step": 267
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.300000000000001e-06,
-      "loss": 4.5134,
+      "epoch": 0.6,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 4.204,
       "step": 268
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 4.2337,
+      "epoch": 0.6,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 4.2903,
       "step": 269
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 4.3884,
+      "epoch": 0.61,
+      "learning_rate": 5.36e-06,
+      "loss": 4.1648,
       "step": 270
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.36e-06,
-      "loss": 3.9959,
+      "epoch": 0.61,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 4.0881,
       "step": 271
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 4.1411,
+      "epoch": 0.61,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.272,
       "step": 272
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 4.1706,
+      "epoch": 0.61,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 4.083,
       "step": 273
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 4.0933,
+      "epoch": 0.61,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 4.1777,
       "step": 274
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4400000000000004e-06,
-      "loss": 4.3469,
+      "epoch": 0.62,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 4.2714,
       "step": 275
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 4.261,
+      "epoch": 0.62,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 4.211,
       "step": 276
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 4.2084,
+      "epoch": 0.62,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.1641,
       "step": 277
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 4.1839,
+      "epoch": 0.62,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 4.2538,
       "step": 278
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 4.1806,
+      "epoch": 0.63,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 4.2379,
       "step": 279
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.540000000000001e-06,
-      "loss": 4.3312,
+      "epoch": 0.63,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 4.3645,
       "step": 280
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.560000000000001e-06,
-      "loss": 4.1195,
+      "epoch": 0.63,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 4.2645,
       "step": 281
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 4.2436,
+      "epoch": 0.63,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.12,
       "step": 282
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.015,
+      "epoch": 0.63,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 4.1472,
       "step": 283
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.620000000000001e-06,
-      "loss": 4.2216,
+      "epoch": 0.64,
+      "learning_rate": 5.64e-06,
+      "loss": 4.2229,
       "step": 284
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.64e-06,
-      "loss": 4.3188,
+      "epoch": 0.64,
+      "learning_rate": 5.66e-06,
+      "loss": 4.3255,
       "step": 285
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.66e-06,
-      "loss": 4.2283,
+      "epoch": 0.64,
+      "learning_rate": 5.68e-06,
+      "loss": 4.2164,
       "step": 286
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.68e-06,
-      "loss": 4.3774,
+      "epoch": 0.64,
+      "learning_rate": 5.7e-06,
+      "loss": 4.3277,
       "step": 287
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.7e-06,
-      "loss": 4.3359,
+      "epoch": 0.65,
+      "learning_rate": 5.72e-06,
+      "loss": 4.2827,
       "step": 288
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.72e-06,
-      "loss": 4.3876,
+      "epoch": 0.65,
+      "learning_rate": 5.74e-06,
+      "loss": 4.1908,
       "step": 289
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.74e-06,
-      "loss": 4.376,
+      "epoch": 0.65,
+      "learning_rate": 5.76e-06,
+      "loss": 4.2463,
       "step": 290
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.76e-06,
-      "loss": 4.3976,
+      "epoch": 0.65,
+      "learning_rate": 5.78e-06,
+      "loss": 4.4979,
       "step": 291
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.78e-06,
-      "loss": 4.2478,
+      "epoch": 0.65,
+      "learning_rate": 5.8e-06,
+      "loss": 4.3692,
       "step": 292
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8e-06,
-      "loss": 4.3192,
+      "epoch": 0.66,
+      "learning_rate": 5.82e-06,
+      "loss": 4.4422,
       "step": 293
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.82e-06,
-      "loss": 4.3107,
+      "epoch": 0.66,
+      "learning_rate": 5.84e-06,
+      "loss": 4.4088,
       "step": 294
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.84e-06,
-      "loss": 4.209,
+      "epoch": 0.66,
+      "learning_rate": 5.86e-06,
+      "loss": 4.3877,
       "step": 295
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.86e-06,
-      "loss": 4.4595,
+      "epoch": 0.66,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.3863,
       "step": 296
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.6001,
+      "epoch": 0.67,
+      "learning_rate": 5.9e-06,
+      "loss": 4.291,
       "step": 297
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.9e-06,
-      "loss": 4.3936,
+      "epoch": 0.67,
+      "learning_rate": 5.92e-06,
+      "loss": 4.6069,
       "step": 298
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.92e-06,
-      "loss": 4.8057,
+      "epoch": 0.67,
+      "learning_rate": 5.94e-06,
+      "loss": 4.3674,
       "step": 299
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.94e-06,
-      "loss": 4.8243,
+      "epoch": 0.67,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 4.7813,
       "step": 300
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 4.0976,
+      "epoch": 0.67,
+      "learning_rate": 5.98e-06,
+      "loss": 4.2702,
       "step": 301
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.98e-06,
-      "loss": 4.2088,
+      "epoch": 0.68,
+      "learning_rate": 6e-06,
+      "loss": 4.2331,
       "step": 302
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6e-06,
-      "loss": 4.1036,
+      "epoch": 0.68,
+      "learning_rate": 6.02e-06,
+      "loss": 4.2356,
       "step": 303
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.02e-06,
-      "loss": 4.2705,
+      "epoch": 0.68,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 4.233,
       "step": 304
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.040000000000001e-06,
-      "loss": 4.0958,
+      "epoch": 0.68,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.2104,
       "step": 305
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.1564,
+      "epoch": 0.69,
+      "learning_rate": 6.08e-06,
+      "loss": 4.2141,
       "step": 306
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.08e-06,
-      "loss": 4.1811,
+      "epoch": 0.69,
+      "learning_rate": 6.1e-06,
+      "loss": 4.2022,
       "step": 307
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.1e-06,
-      "loss": 4.2724,
+      "epoch": 0.69,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 4.1967,
       "step": 308
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.120000000000001e-06,
-      "loss": 4.189,
+      "epoch": 0.69,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 4.1975,
       "step": 309
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.1400000000000005e-06,
-      "loss": 4.3119,
+      "epoch": 0.7,
+      "learning_rate": 6.16e-06,
+      "loss": 4.2753,
       "step": 310
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.16e-06,
-      "loss": 4.383,
+      "epoch": 0.7,
+      "learning_rate": 6.18e-06,
+      "loss": 4.0923,
       "step": 311
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.18e-06,
-      "loss": 4.338,
+      "epoch": 0.7,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.2573,
       "step": 312
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.200000000000001e-06,
-      "loss": 4.2291,
+      "epoch": 0.7,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 4.2671,
       "step": 313
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.220000000000001e-06,
-      "loss": 4.2739,
+      "epoch": 0.7,
+      "learning_rate": 6.24e-06,
+      "loss": 4.183,
       "step": 314
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.24e-06,
-      "loss": 4.1138,
+      "epoch": 0.71,
+      "learning_rate": 6.26e-06,
+      "loss": 4.2129,
       "step": 315
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.26e-06,
-      "loss": 4.2754,
+      "epoch": 0.71,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 4.1825,
       "step": 316
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.280000000000001e-06,
-      "loss": 4.1274,
+      "epoch": 0.71,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.1694,
       "step": 317
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 4.1402,
+      "epoch": 0.71,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 4.214,
       "step": 318
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.3200000000000005e-06,
-      "loss": 4.2526,
+      "epoch": 0.72,
+      "learning_rate": 6.34e-06,
+      "loss": 4.201,
       "step": 319
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.34e-06,
-      "loss": 4.18,
+      "epoch": 0.72,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 4.2793,
       "step": 320
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.360000000000001e-06,
-      "loss": 4.286,
+      "epoch": 0.72,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 4.1701,
       "step": 321
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.380000000000001e-06,
-      "loss": 4.1926,
+      "epoch": 0.72,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.0582,
       "step": 322
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.1269,
+      "epoch": 0.72,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2408,
       "step": 323
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.42e-06,
-      "loss": 4.3107,
+      "epoch": 0.73,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 4.1051,
       "step": 324
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.440000000000001e-06,
-      "loss": 4.3192,
+      "epoch": 0.73,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 4.2394,
       "step": 325
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 4.3081,
+      "epoch": 0.73,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 4.197,
       "step": 326
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.480000000000001e-06,
-      "loss": 4.194,
+      "epoch": 0.73,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.1724,
       "step": 327
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 4.1803,
+      "epoch": 0.74,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 4.147,
       "step": 328
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.520000000000001e-06,
-      "loss": 4.3525,
+      "epoch": 0.74,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 4.1868,
       "step": 329
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.540000000000001e-06,
-      "loss": 4.3741,
+      "epoch": 0.74,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 4.2377,
       "step": 330
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.560000000000001e-06,
-      "loss": 4.224,
+      "epoch": 0.74,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 4.2337,
       "step": 331
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.5800000000000005e-06,
-      "loss": 4.2272,
+      "epoch": 0.74,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.136,
       "step": 332
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.600000000000001e-06,
-      "loss": 4.2663,
+      "epoch": 0.75,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 4.2892,
       "step": 333
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.620000000000001e-06,
-      "loss": 4.2808,
+      "epoch": 0.75,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 4.1944,
       "step": 334
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.640000000000001e-06,
-      "loss": 4.4152,
+      "epoch": 0.75,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.37,
       "step": 335
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.1882,
+      "epoch": 0.75,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 4.1538,
       "step": 336
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.680000000000001e-06,
-      "loss": 4.0873,
+      "epoch": 0.76,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.1614,
       "step": 337
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 4.3547,
+      "epoch": 0.76,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 4.2704,
       "step": 338
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.720000000000001e-06,
-      "loss": 4.1204,
+      "epoch": 0.76,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 4.2589,
       "step": 339
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 4.2857,
+      "epoch": 0.76,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 4.3103,
       "step": 340
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.760000000000001e-06,
-      "loss": 4.1946,
+      "epoch": 0.76,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 4.2509,
       "step": 341
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.780000000000001e-06,
-      "loss": 4.3246,
+      "epoch": 0.77,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4144,
       "step": 342
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.2954,
+      "epoch": 0.77,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 4.3132,
       "step": 343
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.820000000000001e-06,
-      "loss": 4.3643,
+      "epoch": 0.77,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 4.3986,
       "step": 344
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.8400000000000014e-06,
-      "loss": 4.3764,
+      "epoch": 0.77,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 4.1276,
       "step": 345
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.860000000000001e-06,
-      "loss": 4.5745,
+      "epoch": 0.78,
+      "learning_rate": 6.88e-06,
+      "loss": 4.3877,
       "step": 346
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.88e-06,
-      "loss": 4.38,
+      "epoch": 0.78,
+      "learning_rate": 6.9e-06,
+      "loss": 4.379,
       "step": 347
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.9e-06,
-      "loss": 4.445,
+      "epoch": 0.78,
+      "learning_rate": 6.92e-06,
+      "loss": 4.318,
       "step": 348
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.92e-06,
-      "loss": 4.6395,
+      "epoch": 0.78,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 4.5441,
       "step": 349
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.9400000000000005e-06,
-      "loss": 4.4864,
+      "epoch": 0.78,
+      "learning_rate": 6.96e-06,
+      "loss": 4.5578,
       "step": 350
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.96e-06,
-      "loss": 4.2038,
+      "epoch": 0.79,
+      "learning_rate": 6.98e-06,
+      "loss": 4.2743,
       "step": 351
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.98e-06,
-      "loss": 4.216,
+      "epoch": 0.79,
+      "learning_rate": 7e-06,
+      "loss": 4.2212,
       "step": 352
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7e-06,
-      "loss": 4.1771,
+      "epoch": 0.79,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.1421,
       "step": 353
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.1654,
+      "epoch": 0.79,
+      "learning_rate": 7.04e-06,
+      "loss": 4.0231,
       "step": 354
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.04e-06,
-      "loss": 4.2884,
+      "epoch": 0.8,
+      "learning_rate": 7.06e-06,
+      "loss": 4.2178,
       "step": 355
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.06e-06,
-      "loss": 4.0349,
+      "epoch": 0.8,
+      "learning_rate": 7.08e-06,
+      "loss": 4.3222,
       "step": 356
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.08e-06,
-      "loss": 4.119,
+      "epoch": 0.8,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 4.1956,
       "step": 357
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 4.1833,
+      "epoch": 0.8,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 4.2345,
       "step": 358
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.1200000000000004e-06,
-      "loss": 4.1186,
+      "epoch": 0.8,
+      "learning_rate": 7.14e-06,
+      "loss": 4.1192,
       "step": 359
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.14e-06,
-      "loss": 4.3081,
+      "epoch": 0.81,
+      "learning_rate": 7.16e-06,
+      "loss": 4.1955,
       "step": 360
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.16e-06,
-      "loss": 4.1604,
+      "epoch": 0.81,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 4.1376,
       "step": 361
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.180000000000001e-06,
-      "loss": 4.162,
+      "epoch": 0.81,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 4.1929,
       "step": 362
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.2000000000000005e-06,
-      "loss": 4.0856,
+      "epoch": 0.81,
+      "learning_rate": 7.22e-06,
+      "loss": 4.115,
       "step": 363
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.22e-06,
-      "loss": 4.3203,
+      "epoch": 0.82,
+      "learning_rate": 7.24e-06,
+      "loss": 4.2139,
       "step": 364
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.24e-06,
-      "loss": 4.0565,
+      "epoch": 0.82,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 4.3332,
       "step": 365
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.260000000000001e-06,
-      "loss": 4.1779,
+      "epoch": 0.82,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 4.1502,
       "step": 366
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 4.1172,
+      "epoch": 0.82,
+      "learning_rate": 7.3e-06,
+      "loss": 4.233,
       "step": 367
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.3e-06,
-      "loss": 4.2462,
+      "epoch": 0.83,
+      "learning_rate": 7.32e-06,
+      "loss": 4.2126,
       "step": 368
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.32e-06,
-      "loss": 4.281,
+      "epoch": 0.83,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 4.1302,
       "step": 369
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.340000000000001e-06,
-      "loss": 4.3191,
+      "epoch": 0.83,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 4.143,
       "step": 370
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.360000000000001e-06,
-      "loss": 4.1227,
+      "epoch": 0.83,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.1357,
       "step": 371
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.4071,
-      "step": 372
-    },
-    {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 7.4e-06,
-      "loss": 4.1831,
-      "step": 373
+      "loss": 4.2163,
+      "step": 372
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.420000000000001e-06,
-      "loss": 4.2501,
-      "step": 374
+      "loss": 4.2068,
+      "step": 373
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 4.1389,
-      "step": 375
+      "loss": 4.0071,
+      "step": 374
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.4600000000000006e-06,
-      "loss": 4.2473,
-      "step": 376
+      "loss": 4.2376,
+      "step": 375
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 7.48e-06,
-      "loss": 3.9824,
-      "step": 377
+      "loss": 4.0878,
+      "step": 376
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 4.3072,
-      "step": 378
+      "loss": 4.2106,
+      "step": 377
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 7.520000000000001e-06,
-      "loss": 4.2326,
-      "step": 379
+      "loss": 4.1972,
+      "step": 378
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.540000000000001e-06,
-      "loss": 4.321,
-      "step": 380
+      "loss": 4.1123,
+      "step": 379
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.2838,
-      "step": 381
+      "loss": 4.1345,
+      "step": 380
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 7.58e-06,
-      "loss": 4.2274,
-      "step": 382
+      "loss": 4.1914,
+      "step": 381
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.600000000000001e-06,
-      "loss": 4.2623,
-      "step": 383
+      "loss": 4.2757,
+      "step": 382
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.620000000000001e-06,
-      "loss": 4.2464,
-      "step": 384
+      "loss": 4.1834,
+      "step": 383
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.640000000000001e-06,
-      "loss": 4.1796,
-      "step": 385
+      "loss": 4.1181,
+      "step": 384
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 7.660000000000001e-06,
-      "loss": 4.2227,
-      "step": 386
+      "loss": 4.2012,
+      "step": 385
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.680000000000001e-06,
-      "loss": 4.1394,
-      "step": 387
+      "loss": 4.077,
+      "step": 386
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 7.7e-06,
-      "loss": 4.0276,
-      "step": 388
+      "loss": 4.2641,
+      "step": 387
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.72e-06,
-      "loss": 4.2015,
-      "step": 389
+      "loss": 4.2954,
+      "step": 388
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.74e-06,
-      "loss": 4.3076,
-      "step": 390
+      "loss": 4.1907,
+      "step": 389
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.87,
       "learning_rate": 7.76e-06,
-      "loss": 4.386,
-      "step": 391
+      "loss": 4.1714,
+      "step": 390
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.78e-06,
-      "loss": 4.0228,
-      "step": 392
+      "loss": 4.2965,
+      "step": 391
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.800000000000002e-06,
-      "loss": 4.2193,
-      "step": 393
+      "loss": 4.1424,
+      "step": 392
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.820000000000001e-06,
-      "loss": 4.3488,
-      "step": 394
+      "loss": 4.139,
+      "step": 393
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 7.840000000000001e-06,
-      "loss": 4.466,
-      "step": 395
+      "loss": 4.2231,
+      "step": 394
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.89,
       "learning_rate": 7.860000000000001e-06,
-      "loss": 4.2375,
-      "step": 396
+      "loss": 4.3462,
+      "step": 395
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.88e-06,
-      "loss": 4.2363,
-      "step": 397
+      "loss": 4.3144,
+      "step": 396
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.9e-06,
-      "loss": 4.7711,
-      "step": 398
+      "loss": 4.4527,
+      "step": 397
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.92e-06,
-      "loss": 4.4802,
-      "step": 399
+      "loss": 4.456,
+      "step": 398
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 7.94e-06,
-      "loss": 4.5695,
-      "step": 400
+      "loss": 4.5934,
+      "step": 399
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.960000000000002e-06,
-      "loss": 4.29,
-      "step": 401
+      "loss": 4.5134,
+      "step": 400
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 7.980000000000002e-06,
-      "loss": 4.1424,
-      "step": 402
+      "loss": 4.217,
+      "step": 401
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 4.2243,
-      "step": 403
+      "loss": 4.0645,
+      "step": 402
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 8.020000000000001e-06,
-      "loss": 4.1423,
-      "step": 404
+      "loss": 4.1291,
+      "step": 403
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 8.040000000000001e-06,
-      "loss": 4.3115,
-      "step": 405
+      "loss": 4.1237,
+      "step": 404
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.06e-06,
-      "loss": 4.198,
-      "step": 406
+      "loss": 4.0902,
+      "step": 405
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.08e-06,
-      "loss": 4.2142,
-      "step": 407
+      "loss": 4.1787,
+      "step": 406
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.1e-06,
-      "loss": 4.1048,
-      "step": 408
+      "loss": 4.1467,
+      "step": 407
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.91,
       "learning_rate": 8.120000000000002e-06,
-      "loss": 4.2956,
-      "step": 409
+      "loss": 4.2247,
+      "step": 408
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.14e-06,
-      "loss": 4.2151,
-      "step": 410
+      "loss": 4.1621,
+      "step": 409
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.16e-06,
-      "loss": 4.0979,
-      "step": 411
+      "loss": 4.141,
+      "step": 410
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.18e-06,
-      "loss": 4.2008,
-      "step": 412
+      "loss": 4.3667,
+      "step": 411
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 8.2e-06,
-      "loss": 4.2033,
-      "step": 413
+      "loss": 4.1049,
+      "step": 412
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 8.220000000000001e-06,
-      "loss": 4.0824,
-      "step": 414
+      "loss": 4.0321,
+      "step": 413
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.24e-06,
-      "loss": 4.0744,
-      "step": 415
+      "loss": 4.1928,
+      "step": 414
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.26e-06,
-      "loss": 4.3098,
-      "step": 416
+      "loss": 4.2578,
+      "step": 415
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.28e-06,
-      "loss": 4.11,
-      "step": 417
+      "loss": 4.1304,
+      "step": 416
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 8.3e-06,
-      "loss": 4.2412,
-      "step": 418
+      "loss": 4.1514,
+      "step": 417
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.32e-06,
-      "loss": 4.1749,
-      "step": 419
+      "loss": 4.0506,
+      "step": 418
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.34e-06,
-      "loss": 4.2086,
-      "step": 420
+      "loss": 4.1579,
+      "step": 419
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.36e-06,
-      "loss": 4.2794,
-      "step": 421
+      "loss": 4.0673,
+      "step": 420
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 8.380000000000001e-06,
-      "loss": 4.2946,
-      "step": 422
+      "loss": 4.2447,
+      "step": 421
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 4.2214,
-      "step": 423
+      "loss": 4.0858,
+      "step": 422
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.42e-06,
-      "loss": 4.2226,
-      "step": 424
+      "loss": 4.1902,
+      "step": 423
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.44e-06,
-      "loss": 4.2759,
-      "step": 425
+      "loss": 4.2478,
+      "step": 424
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.95,
       "learning_rate": 8.46e-06,
-      "loss": 4.3037,
-      "step": 426
+      "loss": 4.1727,
+      "step": 425
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.48e-06,
-      "loss": 4.1417,
-      "step": 427
+      "loss": 4.225,
+      "step": 426
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.5e-06,
-      "loss": 4.3808,
-      "step": 428
+      "loss": 4.1784,
+      "step": 427
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.52e-06,
-      "loss": 4.1689,
-      "step": 429
+      "loss": 4.1503,
+      "step": 428
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.540000000000001e-06,
-      "loss": 4.3288,
-      "step": 430
+      "loss": 4.1028,
+      "step": 429
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 8.560000000000001e-06,
-      "loss": 4.4165,
-      "step": 431
+      "loss": 4.1624,
+      "step": 430
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 8.580000000000001e-06,
-      "loss": 4.2614,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.6e-06,
-      "loss": 3.9755,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.62e-06,
-      "loss": 4.2045,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.64e-06,
-      "loss": 4.1692,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.66e-06,
-      "loss": 4.4135,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.68e-06,
-      "loss": 4.4173,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.700000000000001e-06,
-      "loss": 4.2981,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.720000000000001e-06,
-      "loss": 4.1147,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.740000000000001e-06,
-      "loss": 4.1698,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.76e-06,
-      "loss": 4.3496,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.78e-06,
-      "loss": 4.2193,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.8e-06,
-      "loss": 4.2942,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-06,
-      "loss": 4.4609,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.84e-06,
-      "loss": 4.2708,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.860000000000002e-06,
-      "loss": 4.1075,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.880000000000001e-06,
-      "loss": 4.3124,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.900000000000001e-06,
-      "loss": 4.3807,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.920000000000001e-06,
-      "loss": 4.3406,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.94e-06,
-      "loss": 4.8661,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.96e-06,
-      "loss": 4.2218,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.98e-06,
-      "loss": 4.205,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9e-06,
-      "loss": 4.3621,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.020000000000002e-06,
-      "loss": 4.0638,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.040000000000002e-06,
-      "loss": 4.3311,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.060000000000001e-06,
-      "loss": 4.258,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.080000000000001e-06,
-      "loss": 4.2222,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 4.1564,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.12e-06,
-      "loss": 4.1892,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.14e-06,
-      "loss": 4.3413,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.16e-06,
-      "loss": 4.102,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.180000000000002e-06,
-      "loss": 4.147,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.200000000000002e-06,
-      "loss": 4.1742,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.220000000000002e-06,
-      "loss": 4.1108,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.240000000000001e-06,
-      "loss": 4.1886,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.260000000000001e-06,
-      "loss": 4.2129,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.280000000000001e-06,
-      "loss": 4.1751,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.3e-06,
-      "loss": 4.0679,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.32e-06,
-      "loss": 4.1051,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.340000000000002e-06,
-      "loss": 4.2847,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.360000000000002e-06,
-      "loss": 4.0783,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.38e-06,
-      "loss": 4.074,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.4e-06,
-      "loss": 4.1647,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-06,
-      "loss": 4.2346,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.440000000000001e-06,
-      "loss": 4.1901,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.460000000000001e-06,
-      "loss": 4.2477,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.48e-06,
-      "loss": 4.4027,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.5e-06,
-      "loss": 4.1196,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.52e-06,
-      "loss": 3.9847,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-06,
-      "loss": 4.2909,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.56e-06,
-      "loss": 4.1164,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.58e-06,
-      "loss": 4.0756,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2811,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.620000000000001e-06,
-      "loss": 4.1496,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.640000000000001e-06,
-      "loss": 4.3244,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.66e-06,
-      "loss": 4.1365,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.68e-06,
-      "loss": 4.0356,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.7e-06,
-      "loss": 4.1874,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2132,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.74e-06,
-      "loss": 4.2876,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.760000000000001e-06,
-      "loss": 4.1096,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.3891,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.4212,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.820000000000001e-06,
-      "loss": 4.2193,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.84e-06,
-      "loss": 4.3731,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.86e-06,
-      "loss": 4.153,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.88e-06,
-      "loss": 4.1048,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.9e-06,
-      "loss": 4.7646,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 4.7269,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.940000000000001e-06,
-      "loss": 4.3305,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.228886127471924,
-      "eval_runtime": 974.9232,
-      "eval_samples_per_second": 2.71,
-      "eval_steps_per_second": 0.34,
-      "eval_wer": 1.9632883776279255,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.4624,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 4.1617,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1e-05,
-      "loss": 4.2835,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.974489795918369e-06,
-      "loss": 4.1218,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.948979591836737e-06,
-      "loss": 4.1019,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.923469387755103e-06,
-      "loss": 4.2585,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.89795918367347e-06,
-      "loss": 4.1945,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.872448979591838e-06,
-      "loss": 4.253,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.846938775510205e-06,
-      "loss": 4.1482,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.821428571428573e-06,
-      "loss": 4.1989,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.795918367346939e-06,
-      "loss": 4.198,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.770408163265307e-06,
-      "loss": 4.1666,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.744897959183674e-06,
-      "loss": 4.4225,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.719387755102042e-06,
-      "loss": 4.1743,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.693877551020408e-06,
-      "loss": 4.1603,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.668367346938776e-06,
-      "loss": 4.067,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.642857142857144e-06,
-      "loss": 4.2825,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.617346938775512e-06,
-      "loss": 4.2279,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.591836734693878e-06,
-      "loss": 4.2518,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.566326530612246e-06,
-      "loss": 4.146,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.540816326530612e-06,
-      "loss": 4.1491,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.51530612244898e-06,
-      "loss": 4.1193,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.489795918367348e-06,
-      "loss": 4.0504,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.464285714285714e-06,
-      "loss": 4.2093,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.438775510204082e-06,
-      "loss": 4.2388,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.41326530612245e-06,
-      "loss": 4.0192,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.387755102040818e-06,
-      "loss": 4.2422,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.362244897959184e-06,
-      "loss": 4.2195,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.336734693877552e-06,
-      "loss": 4.1463,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.31122448979592e-06,
-      "loss": 4.4396,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.285714285714288e-06,
-      "loss": 4.2119,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.260204081632654e-06,
-      "loss": 4.0274,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.234693877551022e-06,
-      "loss": 4.2044,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.209183673469388e-06,
-      "loss": 4.0576,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 4.1935,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.158163265306124e-06,
-      "loss": 4.2543,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.13265306122449e-06,
-      "loss": 4.3184,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.107142857142858e-06,
-      "loss": 4.392,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.081632653061225e-06,
-      "loss": 4.2116,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.056122448979593e-06,
-      "loss": 4.3049,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.03061224489796e-06,
-      "loss": 4.3154,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.005102040816327e-06,
-      "loss": 4.3371,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.979591836734695e-06,
-      "loss": 4.5037,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.954081632653061e-06,
-      "loss": 4.3188,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.92857142857143e-06,
-      "loss": 4.4094,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.903061224489795e-06,
-      "loss": 4.4302,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.877551020408163e-06,
-      "loss": 4.3836,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.852040816326531e-06,
-      "loss": 4.7186,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.826530612244899e-06,
-      "loss": 4.1739,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.801020408163265e-06,
-      "loss": 5.1641,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.775510204081633e-06,
-      "loss": 3.9518,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.750000000000001e-06,
-      "loss": 4.3062,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.724489795918369e-06,
-      "loss": 4.2158,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.698979591836737e-06,
-      "loss": 4.1338,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.673469387755103e-06,
-      "loss": 4.1384,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.64795918367347e-06,
-      "loss": 4.1871,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.622448979591837e-06,
-      "loss": 4.0559,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.596938775510205e-06,
-      "loss": 4.1659,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.571428571428571e-06,
-      "loss": 4.1301,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.545918367346939e-06,
-      "loss": 4.054,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.520408163265307e-06,
-      "loss": 4.2714,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.494897959183675e-06,
-      "loss": 4.1198,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.469387755102042e-06,
-      "loss": 4.1379,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.443877551020409e-06,
-      "loss": 4.1508,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.418367346938776e-06,
-      "loss": 4.2795,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.392857142857144e-06,
-      "loss": 4.2692,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.36734693877551e-06,
-      "loss": 4.1255,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.341836734693878e-06,
-      "loss": 4.1056,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.316326530612246e-06,
-      "loss": 4.2612,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.290816326530612e-06,
-      "loss": 4.0514,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.26530612244898e-06,
-      "loss": 4.0331,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.239795918367348e-06,
-      "loss": 4.2712,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.214285714285714e-06,
-      "loss": 4.0528,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.188775510204082e-06,
-      "loss": 4.0658,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.16326530612245e-06,
-      "loss": 4.223,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.137755102040818e-06,
-      "loss": 4.0634,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.112244897959184e-06,
-      "loss": 4.1961,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.086734693877552e-06,
-      "loss": 4.297,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.06122448979592e-06,
-      "loss": 3.8723,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.035714285714286e-06,
-      "loss": 4.3325,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.010204081632654e-06,
-      "loss": 4.3099,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.98469387755102e-06,
-      "loss": 4.0784,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.959183673469388e-06,
-      "loss": 4.0144,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.933673469387756e-06,
-      "loss": 4.2075,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.908163265306124e-06,
-      "loss": 4.3471,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.88265306122449e-06,
-      "loss": 4.214,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.857142857142858e-06,
-      "loss": 4.2309,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.831632653061226e-06,
-      "loss": 4.301,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.806122448979593e-06,
-      "loss": 3.9973,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.78061224489796e-06,
-      "loss": 4.1042,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.755102040816327e-06,
-      "loss": 4.2299,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.729591836734695e-06,
-      "loss": 4.3201,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.704081632653061e-06,
-      "loss": 4.556,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.67857142857143e-06,
-      "loss": 4.3675,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 4.302,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.627551020408163e-06,
-      "loss": 4.1146,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.602040816326531e-06,
-      "loss": 4.2442,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.576530612244899e-06,
-      "loss": 4.3609,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.551020408163265e-06,
-      "loss": 4.3449,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.525510204081633e-06,
-      "loss": 4.2737,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 4.2102,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.474489795918368e-06,
-      "loss": 4.0365,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.448979591836736e-06,
-      "loss": 4.1834,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.423469387755102e-06,
-      "loss": 4.1542,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.39795918367347e-06,
-      "loss": 4.126,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.372448979591838e-06,
-      "loss": 4.1515,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.346938775510205e-06,
-      "loss": 4.1811,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.321428571428572e-06,
-      "loss": 4.1185,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.295918367346939e-06,
-      "loss": 4.1018,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.270408163265307e-06,
-      "loss": 4.1371,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.244897959183675e-06,
-      "loss": 4.0925,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.219387755102042e-06,
-      "loss": 4.2773,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.193877551020409e-06,
-      "loss": 4.2033,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.168367346938776e-06,
-      "loss": 4.2348,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1428571428571436e-06,
-      "loss": 4.1249,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1173469387755114e-06,
-      "loss": 4.1543,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.091836734693878e-06,
-      "loss": 4.0646,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.0663265306122455e-06,
-      "loss": 4.1329,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.0408163265306125e-06,
-      "loss": 4.0905,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.01530612244898e-06,
-      "loss": 4.1415,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.989795918367348e-06,
-      "loss": 4.1331,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.964285714285714e-06,
-      "loss": 4.0551,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.938775510204082e-06,
-      "loss": 4.294,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.91326530612245e-06,
-      "loss": 4.1089,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.887755102040817e-06,
-      "loss": 4.2508,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.862244897959184e-06,
-      "loss": 4.2691,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.836734693877551e-06,
-      "loss": 4.2317,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.811224489795919e-06,
-      "loss": 4.1662,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.785714285714287e-06,
-      "loss": 4.1816,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.760204081632653e-06,
-      "loss": 4.3043,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.734693877551021e-06,
-      "loss": 4.2666,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.709183673469388e-06,
-      "loss": 4.0984,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.683673469387756e-06,
-      "loss": 4.2147,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.658163265306124e-06,
-      "loss": 4.172,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.63265306122449e-06,
-      "loss": 4.2655,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.607142857142858e-06,
-      "loss": 4.2441,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.581632653061225e-06,
-      "loss": 4.1021,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.556122448979593e-06,
-      "loss": 4.0496,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.530612244897959e-06,
-      "loss": 4.2592,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.505102040816327e-06,
-      "loss": 4.0456,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4795918367346946e-06,
-      "loss": 4.2189,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.454081632653062e-06,
-      "loss": 4.5666,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4285714285714295e-06,
-      "loss": 4.3746,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.403061224489796e-06,
-      "loss": 4.1678,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.3775510204081635e-06,
-      "loss": 4.0132,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.352040816326531e-06,
-      "loss": 4.3461,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.326530612244899e-06,
-      "loss": 4.2589,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.301020408163265e-06,
-      "loss": 4.096,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.275510204081633e-06,
-      "loss": 4.5869,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.25e-06,
-      "loss": 4.407,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.224489795918368e-06,
-      "loss": 4.27,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.198979591836736e-06,
-      "loss": 4.2809,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.173469387755102e-06,
-      "loss": 4.2394,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.14795918367347e-06,
-      "loss": 4.2062,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 4.2026,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.096938775510205e-06,
-      "loss": 4.1606,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.071428571428571e-06,
-      "loss": 4.0792,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.045918367346939e-06,
-      "loss": 4.3318,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.020408163265307e-06,
-      "loss": 4.1618,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.994897959183674e-06,
-      "loss": 4.2486,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.969387755102042e-06,
-      "loss": 4.0378,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.943877551020408e-06,
-      "loss": 4.0316,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.918367346938776e-06,
-      "loss": 4.2624,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.892857142857144e-06,
-      "loss": 4.1044,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.867346938775511e-06,
-      "loss": 4.0393,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.841836734693878e-06,
-      "loss": 4.3049,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.816326530612246e-06,
-      "loss": 4.0645,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.790816326530613e-06,
-      "loss": 4.302,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7653061224489805e-06,
-      "loss": 4.2838,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.739795918367348e-06,
-      "loss": 4.2344,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7142857142857145e-06,
-      "loss": 4.1169,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.688775510204082e-06,
-      "loss": 3.9761,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.663265306122449e-06,
-      "loss": 4.0069,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.637755102040817e-06,
-      "loss": 4.0721,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.6122448979591834e-06,
-      "loss": 4.0428,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.586734693877551e-06,
-      "loss": 4.1588,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.561224489795919e-06,
-      "loss": 4.0451,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.535714285714286e-06,
-      "loss": 3.9977,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.510204081632653e-06,
-      "loss": 4.0278,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.48469387755102e-06,
-      "loss": 4.0882,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.459183673469388e-06,
-      "loss": 4.3101,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.433673469387756e-06,
-      "loss": 4.0457,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.408163265306123e-06,
-      "loss": 4.1816,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.38265306122449e-06,
-      "loss": 4.1824,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 4.2306,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.331632653061225e-06,
-      "loss": 4.1195,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.306122448979593e-06,
-      "loss": 4.2259,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.280612244897959e-06,
-      "loss": 4.2941,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.255102040816327e-06,
-      "loss": 4.2482,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.229591836734695e-06,
-      "loss": 4.3791,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.204081632653062e-06,
-      "loss": 4.101,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1785714285714296e-06,
-      "loss": 4.2065,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.153061224489796e-06,
-      "loss": 4.1788,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.127551020408164e-06,
-      "loss": 4.4918,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1020408163265315e-06,
-      "loss": 4.1984,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0765306122448985e-06,
-      "loss": 4.3442,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0510204081632655e-06,
-      "loss": 4.4095,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0255102040816325e-06,
-      "loss": 4.3569,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5e-06,
-      "loss": 4.3541,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 4.974489795918368e-06,
-      "loss": 4.5501,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.948979591836735e-06,
-      "loss": 4.1101,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.923469387755102e-06,
-      "loss": 4.165,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.897959183673469e-06,
-      "loss": 4.0888,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.872448979591837e-06,
-      "loss": 3.9343,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.846938775510204e-06,
-      "loss": 4.0735,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.821428571428572e-06,
-      "loss": 4.2064,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.795918367346939e-06,
-      "loss": 4.169,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.770408163265306e-06,
-      "loss": 4.1919,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.744897959183674e-06,
-      "loss": 4.0701,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.719387755102041e-06,
-      "loss": 4.1478,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.693877551020409e-06,
-      "loss": 4.0602,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.668367346938776e-06,
-      "loss": 4.1501,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.642857142857144e-06,
-      "loss": 4.0528,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.617346938775511e-06,
-      "loss": 4.2488,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 4.2712,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.566326530612245e-06,
-      "loss": 4.0661,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.540816326530613e-06,
-      "loss": 4.3115,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.51530612244898e-06,
-      "loss": 4.1322,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.489795918367348e-06,
-      "loss": 3.8992,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.464285714285715e-06,
-      "loss": 4.1857,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.438775510204082e-06,
-      "loss": 4.0005,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4132653061224495e-06,
-      "loss": 4.0494,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.3877551020408165e-06,
-      "loss": 4.1158,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.362244897959184e-06,
-      "loss": 4.2112,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.336734693877551e-06,
-      "loss": 4.1126,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.3112244897959184e-06,
-      "loss": 4.1853,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.2857142857142855e-06,
-      "loss": 4.1696,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.260204081632653e-06,
-      "loss": 4.1452,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.234693877551021e-06,
-      "loss": 4.0042,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.209183673469388e-06,
-      "loss": 4.2235,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.183673469387755e-06,
-      "loss": 4.2987,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.158163265306123e-06,
-      "loss": 4.1914,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.13265306122449e-06,
-      "loss": 4.1327,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.107142857142857e-06,
-      "loss": 4.2529,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.081632653061225e-06,
-      "loss": 4.0094,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.056122448979592e-06,
-      "loss": 4.1344,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.03061224489796e-06,
-      "loss": 4.3353,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.005102040816327e-06,
-      "loss": 4.1913,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.979591836734694e-06,
-      "loss": 4.0968,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.954081632653062e-06,
-      "loss": 4.2916,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.928571428571429e-06,
-      "loss": 4.2633,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.903061224489797e-06,
-      "loss": 4.1221,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.877551020408164e-06,
-      "loss": 4.0215,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.852040816326531e-06,
-      "loss": 4.1473,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 4.2389,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.8010204081632656e-06,
-      "loss": 4.4309,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7755102040816327e-06,
-      "loss": 4.2204,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7500000000000005e-06,
-      "loss": 4.4819,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.724489795918368e-06,
-      "loss": 4.4102,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.698979591836735e-06,
-      "loss": 4.3648,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6734693877551024e-06,
-      "loss": 4.2377,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6479591836734694e-06,
-      "loss": 4.2337,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6224489795918373e-06,
-      "loss": 4.1717,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5969387755102043e-06,
-      "loss": 4.036,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5714285714285718e-06,
-      "loss": 4.2284,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.545918367346939e-06,
-      "loss": 4.2494,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5204081632653062e-06,
-      "loss": 4.1771,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.494897959183674e-06,
-      "loss": 4.2262,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.469387755102041e-06,
-      "loss": 4.0813,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4438775510204086e-06,
-      "loss": 4.1629,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4183673469387756e-06,
-      "loss": 4.0902,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.3928571428571435e-06,
-      "loss": 4.1826,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.3673469387755105e-06,
-      "loss": 4.0883,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.341836734693878e-06,
-      "loss": 4.1143,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.316326530612245e-06,
-      "loss": 4.1942,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2908163265306124e-06,
-      "loss": 4.1399,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2653061224489794e-06,
-      "loss": 4.1161,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2397959183673473e-06,
-      "loss": 4.1484,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2142857142857147e-06,
-      "loss": 4.1973,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1887755102040818e-06,
-      "loss": 4.065,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1632653061224496e-06,
-      "loss": 4.1393,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.1377551020408166e-06,
-      "loss": 4.2741,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.112244897959184e-06,
-      "loss": 4.2364,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.086734693877551e-06,
-      "loss": 3.9637,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 4.0388,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0357142857142856e-06,
-      "loss": 4.144,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0102040816326534e-06,
-      "loss": 3.9844,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.984693877551021e-06,
-      "loss": 4.1853,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.959183673469388e-06,
-      "loss": 4.0274,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9336734693877553e-06,
-      "loss": 4.0371,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.908163265306123e-06,
-      "loss": 4.0698,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8826530612244902e-06,
-      "loss": 4.183,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 4.1084,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8316326530612247e-06,
-      "loss": 4.1125,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8061224489795917e-06,
-      "loss": 4.0694,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7806122448979596e-06,
-      "loss": 4.0064,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7551020408163266e-06,
-      "loss": 4.1693,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.729591836734694e-06,
-      "loss": 4.1988,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7040816326530615e-06,
-      "loss": 4.0812,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6785714285714285e-06,
-      "loss": 4.1691,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6530612244897964e-06,
-      "loss": 4.1904,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6275510204081634e-06,
-      "loss": 4.0029,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.602040816326531e-06,
-      "loss": 4.1935,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.576530612244898e-06,
-      "loss": 4.2966,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5510204081632657e-06,
-      "loss": 4.3301,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5255102040816328e-06,
-      "loss": 4.2118,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3198,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.4744897959183676e-06,
-      "loss": 4.4169,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.4489795918367347e-06,
-      "loss": 4.764,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.423469387755102e-06,
-      "loss": 4.5849,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3979591836734696e-06,
-      "loss": 4.1708,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.372448979591837e-06,
-      "loss": 4.0805,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3469387755102044e-06,
-      "loss": 4.0916,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.321428571428572e-06,
-      "loss": 4.1451,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 3.9581,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2704081632653064e-06,
-      "loss": 4.1162,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.244897959183674e-06,
-      "loss": 4.2046,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.219387755102041e-06,
-      "loss": 4.0425,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1938775510204083e-06,
-      "loss": 4.2532,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1683673469387757e-06,
-      "loss": 4.0693,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1428571428571427e-06,
-      "loss": 4.3283,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1173469387755106e-06,
-      "loss": 4.148,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0918367346938776e-06,
-      "loss": 3.9872,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.066326530612245e-06,
-      "loss": 4.0553,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0408163265306125e-06,
-      "loss": 4.0599,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.01530612244898e-06,
-      "loss": 4.0409,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.989795918367347e-06,
-      "loss": 4.0849,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9642857142857144e-06,
-      "loss": 4.1077,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.938775510204082e-06,
-      "loss": 4.1032,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.913265306122449e-06,
-      "loss": 4.1298,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8877551020408163e-06,
-      "loss": 4.0518,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.862244897959184e-06,
-      "loss": 4.0592,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8367346938775512e-06,
-      "loss": 3.9919,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8112244897959187e-06,
-      "loss": 4.0972,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.7857142857142859e-06,
-      "loss": 4.348,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7602040816326531e-06,
-      "loss": 4.184,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7346938775510206e-06,
-      "loss": 4.1537,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7091836734693878e-06,
-      "loss": 4.0626,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6836734693877552e-06,
-      "loss": 4.0531,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6581632653061225e-06,
-      "loss": 4.2099,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6326530612244897e-06,
-      "loss": 4.0514,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6071428571428574e-06,
-      "loss": 3.9799,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.5816326530612248e-06,
-      "loss": 4.2617,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.556122448979592e-06,
-      "loss": 4.1059,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 4.209,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5051020408163267e-06,
-      "loss": 4.1347,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.479591836734694e-06,
-      "loss": 4.1282,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4540816326530614e-06,
-      "loss": 4.2145,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 4.0665,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4030612244897959e-06,
-      "loss": 4.1559,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3775510204081633e-06,
-      "loss": 4.3593,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3520408163265307e-06,
-      "loss": 4.2098,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3265306122448982e-06,
-      "loss": 4.3444,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3010204081632654e-06,
-      "loss": 4.4029,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2755102040816329e-06,
-      "loss": 4.4122,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.25e-06,
-      "loss": 4.4857,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2244897959183673e-06,
-      "loss": 4.3104,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1989795918367348e-06,
-      "loss": 4.6066,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1734693877551022e-06,
-      "loss": 4.566,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1479591836734695e-06,
-      "loss": 4.3475,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.122448979591837e-06,
-      "loss": 4.1136,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0969387755102041e-06,
-      "loss": 4.0434,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 4.0987,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0459183673469388e-06,
-      "loss": 4.0302,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0204081632653063e-06,
-      "loss": 4.0521,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.948979591836735e-07,
-      "loss": 4.0754,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.69387755102041e-07,
-      "loss": 4.1819,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.438775510204082e-07,
-      "loss": 4.1869,
-      "step": 858
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.183673469387756e-07,
-      "loss": 4.0316,
-      "step": 859
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.928571428571429e-07,
-      "loss": 4.122,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755103e-07,
-      "loss": 4.3079,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938776e-07,
-      "loss": 4.0366,
-      "step": 862
+      "loss": 4.2611,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.163265306122449e-07,
-      "loss": 4.1618,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306124e-07,
-      "loss": 4.2898,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 4.1907,
-      "step": 865
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-07,
-      "loss": 3.9162,
-      "step": 866
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.142857142857143e-07,
-      "loss": 4.1405,
-      "step": 867
+      "learning_rate": 8.6e-06,
+      "loss": 4.3745,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.887755102040817e-07,
-      "loss": 4.0013,
-      "step": 868
+      "learning_rate": 8.62e-06,
+      "loss": 4.17,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.632653061224491e-07,
-      "loss": 4.0399,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.377551020408164e-07,
-      "loss": 4.2113,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591837e-07,
-      "loss": 4.0378,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.867346938775511e-07,
-      "loss": 4.2345,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.612244897959184e-07,
-      "loss": 4.1941,
-      "step": 873
+      "learning_rate": 8.64e-06,
+      "loss": 4.1702,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-07,
-      "loss": 4.2059,
-      "step": 874
+      "learning_rate": 8.66e-06,
+      "loss": 4.2199,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-07,
-      "loss": 4.2025,
-      "step": 875
+      "learning_rate": 8.68e-06,
+      "loss": 4.1389,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.846938775510205e-07,
-      "loss": 4.0469,
-      "step": 876
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 4.1324,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.591836734693878e-07,
-      "loss": 4.1253,
-      "step": 877
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 4.3086,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.3367346938775514e-07,
-      "loss": 4.1534,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.0816326530612243e-07,
-      "loss": 4.3464,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.826530612244898e-07,
-      "loss": 4.4463,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.5714285714285716e-07,
-      "loss": 4.0558,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3163265306122455e-07,
-      "loss": 4.1566,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0612244897959183e-07,
-      "loss": 4.0943,
-      "step": 883
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 4.2724,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.806122448979592e-07,
-      "loss": 4.0736,
-      "step": 884
+      "learning_rate": 8.76e-06,
+      "loss": 4.3117,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.5510204081632656e-07,
-      "loss": 4.1666,
-      "step": 885
+      "learning_rate": 8.78e-06,
+      "loss": 4.2593,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.295918367346939e-07,
-      "loss": 4.1459,
-      "step": 886
+      "learning_rate": 8.8e-06,
+      "loss": 4.4625,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.0408163265306121e-07,
-      "loss": 4.107,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142858e-07,
-      "loss": 4.308,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979592e-07,
-      "loss": 4.3713,
-      "step": 889
+      "learning_rate": 8.82e-06,
+      "loss": 4.3707,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.2755102040816328e-07,
-      "loss": 4.1827,
-      "step": 890
+      "learning_rate": 8.84e-06,
+      "loss": 4.3694,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.0204081632653061e-07,
-      "loss": 4.4517,
-      "step": 891
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 4.6504,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 7.653061224489796e-08,
-      "loss": 4.0199,
-      "step": 892
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 4.3145,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.297821357913081,
-      "train_runtime": 6364.0567,
-      "train_samples_per_second": 4.484,
-      "train_steps_per_second": 0.14
+      "train_loss": 4.375692403904526,
+      "train_runtime": 5180.8925,
+      "train_samples_per_second": 5.508,
+      "train_steps_per_second": 0.086
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,