diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,2690 +10,7166 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 4.7746,
+      "loss": 4.8482,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.9289,
+      "learning_rate": 0.0,
+      "loss": 5.0423,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.8883,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 5.0389,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.8789,
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.8622,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8296,
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 4.5971,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.8764,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.9547,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 4.8601,
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.735,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.2e-07,
-      "loss": 4.9133,
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 4.7239,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.916,
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 4.9385,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.4e-07,
-      "loss": 4.7759,
+      "epoch": 0.01,
+      "learning_rate": 4.2e-07,
+      "loss": 4.7796,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8958,
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.8294,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.599999999999999e-07,
-      "loss": 4.8645,
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 4.6947,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.2e-07,
-      "loss": 4.8147,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.9776,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.799999999999999e-07,
-      "loss": 4.8259,
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 4.7534,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.4e-07,
-      "loss": 4.7034,
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 4.6509,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9e-07,
-      "loss": 4.8451,
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 4.8797,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.7678,
+      "epoch": 0.01,
+      "learning_rate": 8.4e-07,
+      "loss": 4.8165,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 4.8715,
+      "epoch": 0.02,
+      "learning_rate": 9e-07,
+      "loss": 4.9677,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.08e-06,
-      "loss": 4.6656,
+      "epoch": 0.02,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.7116,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.14e-06,
-      "loss": 4.8439,
+      "epoch": 0.02,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 4.9459,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.8241,
+      "epoch": 0.02,
+      "learning_rate": 1.08e-06,
+      "loss": 4.9045,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.26e-06,
-      "loss": 4.7418,
+      "epoch": 0.02,
+      "learning_rate": 1.14e-06,
+      "loss": 4.6776,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.3199999999999999e-06,
-      "loss": 4.8523,
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.8528,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.38e-06,
-      "loss": 4.8442,
+      "epoch": 0.02,
+      "learning_rate": 1.26e-06,
+      "loss": 4.8938,
       "step": 24
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.44e-06,
-      "loss": 4.6269,
+      "epoch": 0.02,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 4.7033,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.5e-06,
-      "loss": 4.8907,
+      "epoch": 0.02,
+      "learning_rate": 1.38e-06,
+      "loss": 4.8629,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.5599999999999999e-06,
-      "loss": 4.7369,
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 4.728,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.62e-06,
-      "loss": 4.5892,
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 4.569,
       "step": 28
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.68e-06,
-      "loss": 4.564,
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 4.6476,
       "step": 29
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.74e-06,
-      "loss": 4.7634,
+      "epoch": 0.03,
+      "learning_rate": 1.62e-06,
+      "loss": 4.6812,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.8e-06,
-      "loss": 4.735,
+      "epoch": 0.03,
+      "learning_rate": 1.68e-06,
+      "loss": 4.8084,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.86e-06,
-      "loss": 4.5768,
+      "epoch": 0.03,
+      "learning_rate": 1.74e-06,
+      "loss": 4.6104,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.6719,
+      "epoch": 0.03,
+      "learning_rate": 1.8e-06,
+      "loss": 4.7272,
       "step": 33
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.98e-06,
-      "loss": 4.5934,
+      "epoch": 0.03,
+      "learning_rate": 1.86e-06,
+      "loss": 4.4005,
       "step": 34
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 4.6764,
+      "epoch": 0.03,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.5867,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.6146,
+      "epoch": 0.03,
+      "learning_rate": 1.98e-06,
+      "loss": 4.9098,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.16e-06,
-      "loss": 4.6754,
+      "epoch": 0.03,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 4.7237,
       "step": 37
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.22e-06,
-      "loss": 4.6119,
+      "epoch": 0.03,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.5612,
       "step": 38
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.28e-06,
-      "loss": 4.7453,
+      "epoch": 0.03,
+      "learning_rate": 2.16e-06,
+      "loss": 5.0326,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.34e-06,
-      "loss": 4.5306,
+      "epoch": 0.03,
+      "learning_rate": 2.22e-06,
+      "loss": 4.6771,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.595,
+      "epoch": 0.03,
+      "learning_rate": 2.28e-06,
+      "loss": 4.5023,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.46e-06,
-      "loss": 4.7049,
+      "epoch": 0.04,
+      "learning_rate": 2.34e-06,
+      "loss": 4.5602,
       "step": 42
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.52e-06,
-      "loss": 4.4633,
+      "epoch": 0.04,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.5688,
       "step": 43
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.58e-06,
-      "loss": 4.6159,
+      "epoch": 0.04,
+      "learning_rate": 2.46e-06,
+      "loss": 4.6182,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.6399999999999997e-06,
-      "loss": 4.7318,
+      "epoch": 0.04,
+      "learning_rate": 2.52e-06,
+      "loss": 4.8205,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.7e-06,
-      "loss": 4.8312,
+      "epoch": 0.04,
+      "learning_rate": 2.58e-06,
+      "loss": 4.7995,
       "step": 46
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.76e-06,
-      "loss": 4.7198,
+      "epoch": 0.04,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 4.9812,
       "step": 47
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.82e-06,
-      "loss": 4.9727,
+      "epoch": 0.04,
+      "learning_rate": 2.7e-06,
+      "loss": 5.1944,
       "step": 48
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.88e-06,
-      "loss": 5.057,
+      "epoch": 0.04,
+      "learning_rate": 2.76e-06,
+      "loss": 5.3073,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.88e-06,
-      "loss": 4.837,
+      "epoch": 0.04,
+      "learning_rate": 2.76e-06,
+      "loss": 5.1714,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.4908,
+      "epoch": 0.04,
+      "learning_rate": 2.82e-06,
+      "loss": 4.453,
       "step": 51
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3e-06,
-      "loss": 4.3346,
+      "epoch": 0.04,
+      "learning_rate": 2.88e-06,
+      "loss": 4.4799,
       "step": 52
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.06e-06,
-      "loss": 4.5022,
+      "epoch": 0.04,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.3157,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.1199999999999998e-06,
-      "loss": 4.4011,
+      "epoch": 0.05,
+      "learning_rate": 3e-06,
+      "loss": 4.3796,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.18e-06,
-      "loss": 4.2133,
+      "epoch": 0.05,
+      "learning_rate": 3.06e-06,
+      "loss": 4.5439,
       "step": 55
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.24e-06,
-      "loss": 4.4502,
+      "epoch": 0.05,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 4.3019,
       "step": 56
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.3e-06,
-      "loss": 4.4597,
+      "epoch": 0.05,
+      "learning_rate": 3.18e-06,
+      "loss": 4.5605,
       "step": 57
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.36e-06,
-      "loss": 4.5472,
+      "epoch": 0.05,
+      "learning_rate": 3.24e-06,
+      "loss": 4.4849,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 4.4784,
+      "epoch": 0.05,
+      "learning_rate": 3.3e-06,
+      "loss": 4.3597,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.48e-06,
-      "loss": 4.3192,
+      "epoch": 0.05,
+      "learning_rate": 3.36e-06,
+      "loss": 4.5394,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.54e-06,
-      "loss": 4.4132,
+      "epoch": 0.05,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 4.293,
       "step": 61
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.6e-06,
-      "loss": 4.3723,
+      "epoch": 0.05,
+      "learning_rate": 3.48e-06,
+      "loss": 4.4734,
       "step": 62
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.66e-06,
-      "loss": 4.3143,
+      "epoch": 0.05,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3956,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.72e-06,
-      "loss": 4.4111,
+      "epoch": 0.05,
+      "learning_rate": 3.6e-06,
+      "loss": 4.2579,
       "step": 64
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.381,
+      "epoch": 0.05,
+      "learning_rate": 3.66e-06,
+      "loss": 4.235,
       "step": 65
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.2726,
+      "epoch": 0.06,
+      "learning_rate": 3.72e-06,
+      "loss": 4.373,
       "step": 66
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.9e-06,
-      "loss": 4.4058,
+      "epoch": 0.06,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.4967,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.96e-06,
-      "loss": 4.269,
+      "epoch": 0.06,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.3602,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.3351,
+      "epoch": 0.06,
+      "learning_rate": 3.9e-06,
+      "loss": 4.3296,
       "step": 69
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 4.2911,
+      "epoch": 0.06,
+      "learning_rate": 3.96e-06,
+      "loss": 4.3836,
       "step": 70
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.14e-06,
-      "loss": 4.434,
+      "epoch": 0.06,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.5421,
       "step": 71
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.4303,
+      "epoch": 0.06,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 4.4315,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.26e-06,
-      "loss": 4.2569,
+      "epoch": 0.06,
+      "learning_rate": 4.14e-06,
+      "loss": 4.377,
       "step": 73
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.32e-06,
-      "loss": 4.3597,
+      "epoch": 0.06,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.4473,
       "step": 74
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.3799999999999996e-06,
-      "loss": 4.3462,
+      "epoch": 0.06,
+      "learning_rate": 4.26e-06,
+      "loss": 4.2584,
       "step": 75
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.44e-06,
-      "loss": 4.3001,
+      "epoch": 0.06,
+      "learning_rate": 4.32e-06,
+      "loss": 4.7131,
       "step": 76
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.5e-06,
-      "loss": 4.3445,
+      "epoch": 0.06,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 4.4143,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.56e-06,
-      "loss": 4.2932,
+      "epoch": 0.07,
+      "learning_rate": 4.44e-06,
+      "loss": 4.2887,
       "step": 78
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.62e-06,
-      "loss": 4.267,
+      "epoch": 0.07,
+      "learning_rate": 4.5e-06,
+      "loss": 4.286,
       "step": 79
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.68e-06,
-      "loss": 4.4182,
+      "epoch": 0.07,
+      "learning_rate": 4.56e-06,
+      "loss": 4.3957,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.74e-06,
-      "loss": 4.3713,
+      "epoch": 0.07,
+      "learning_rate": 4.62e-06,
+      "loss": 4.4273,
       "step": 81
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3061,
+      "epoch": 0.07,
+      "learning_rate": 4.68e-06,
+      "loss": 4.4089,
       "step": 82
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.86e-06,
-      "loss": 4.4143,
+      "epoch": 0.07,
+      "learning_rate": 4.74e-06,
+      "loss": 4.2635,
       "step": 83
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.92e-06,
-      "loss": 4.2805,
+      "epoch": 0.07,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.4184,
       "step": 84
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.3571,
+      "epoch": 0.07,
+      "learning_rate": 4.86e-06,
+      "loss": 4.3043,
       "step": 85
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 5.04e-06,
-      "loss": 4.3305,
+      "epoch": 0.07,
+      "learning_rate": 4.92e-06,
+      "loss": 4.3185,
       "step": 86
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.1e-06,
-      "loss": 4.4209,
+      "epoch": 0.07,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2074,
       "step": 87
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.16e-06,
-      "loss": 4.2915,
+      "epoch": 0.07,
+      "learning_rate": 5.04e-06,
+      "loss": 4.4158,
       "step": 88
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.22e-06,
-      "loss": 4.4348,
+      "epoch": 0.07,
+      "learning_rate": 5.1e-06,
+      "loss": 4.4677,
       "step": 89
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.279999999999999e-06,
-      "loss": 4.4178,
+      "epoch": 0.08,
+      "learning_rate": 5.16e-06,
+      "loss": 4.368,
       "step": 90
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.34e-06,
-      "loss": 4.3064,
+      "epoch": 0.08,
+      "learning_rate": 5.22e-06,
+      "loss": 4.4315,
       "step": 91
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.4e-06,
-      "loss": 4.5617,
+      "epoch": 0.08,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 4.4858,
       "step": 92
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.46e-06,
-      "loss": 4.4422,
+      "epoch": 0.08,
+      "learning_rate": 5.34e-06,
+      "loss": 4.1103,
       "step": 93
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.52e-06,
-      "loss": 4.4235,
+      "epoch": 0.08,
+      "learning_rate": 5.4e-06,
+      "loss": 4.4791,
       "step": 94
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.58e-06,
-      "loss": 4.3727,
+      "epoch": 0.08,
+      "learning_rate": 5.46e-06,
+      "loss": 4.6979,
       "step": 95
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.64e-06,
-      "loss": 4.5854,
+      "epoch": 0.08,
+      "learning_rate": 5.52e-06,
+      "loss": 4.506,
       "step": 96
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 4.5272,
+      "epoch": 0.08,
+      "learning_rate": 5.58e-06,
+      "loss": 4.4602,
       "step": 97
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.76e-06,
-      "loss": 4.3841,
+      "epoch": 0.08,
+      "learning_rate": 5.64e-06,
+      "loss": 4.6356,
       "step": 98
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.82e-06,
-      "loss": 4.711,
+      "epoch": 0.08,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 4.7009,
       "step": 99
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.4718,
+      "epoch": 0.08,
+      "learning_rate": 5.76e-06,
+      "loss": 4.45,
       "step": 100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 4.3046,
+      "epoch": 0.08,
+      "learning_rate": 5.82e-06,
+      "loss": 4.0906,
       "step": 101
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 6e-06,
-      "loss": 4.2435,
+      "epoch": 0.09,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.3836,
       "step": 102
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.113,
+      "epoch": 0.09,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 4.2678,
       "step": 103
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 6.12e-06,
-      "loss": 4.1906,
+      "epoch": 0.09,
+      "learning_rate": 6e-06,
+      "loss": 4.3507,
       "step": 104
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.18e-06,
-      "loss": 4.1371,
+      "epoch": 0.09,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.2486,
       "step": 105
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.2399999999999995e-06,
-      "loss": 4.1835,
+      "epoch": 0.09,
+      "learning_rate": 6.12e-06,
+      "loss": 4.3941,
       "step": 106
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.3e-06,
-      "loss": 4.28,
+      "epoch": 0.09,
+      "learning_rate": 6.18e-06,
+      "loss": 4.4976,
       "step": 107
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.36e-06,
-      "loss": 4.1821,
+      "epoch": 0.09,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 4.4349,
       "step": 108
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.42e-06,
-      "loss": 4.1884,
+      "epoch": 0.09,
+      "learning_rate": 6.3e-06,
+      "loss": 4.3553,
       "step": 109
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.48e-06,
-      "loss": 4.2333,
+      "epoch": 0.09,
+      "learning_rate": 6.36e-06,
+      "loss": 4.0747,
       "step": 110
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.54e-06,
-      "loss": 4.2675,
+      "epoch": 0.09,
+      "learning_rate": 6.42e-06,
+      "loss": 4.385,
       "step": 111
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.6e-06,
-      "loss": 4.2355,
+      "epoch": 0.09,
+      "learning_rate": 6.48e-06,
+      "loss": 4.4549,
       "step": 112
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.2424,
+      "epoch": 0.09,
+      "learning_rate": 6.54e-06,
+      "loss": 4.2384,
       "step": 113
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.72e-06,
-      "loss": 4.2007,
+      "epoch": 0.1,
+      "learning_rate": 6.6e-06,
+      "loss": 4.4148,
       "step": 114
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.78e-06,
-      "loss": 4.2562,
+      "epoch": 0.1,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.3352,
       "step": 115
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 4.3384,
+      "epoch": 0.1,
+      "learning_rate": 6.72e-06,
+      "loss": 4.4855,
       "step": 116
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 4.217,
+      "epoch": 0.1,
+      "learning_rate": 6.78e-06,
+      "loss": 4.2227,
       "step": 117
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.96e-06,
-      "loss": 4.37,
+      "epoch": 0.1,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 4.0853,
       "step": 118
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.2666,
+      "epoch": 0.1,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.189,
       "step": 119
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.08e-06,
-      "loss": 4.292,
+      "epoch": 0.1,
+      "learning_rate": 6.96e-06,
+      "loss": 4.2727,
       "step": 120
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.14e-06,
-      "loss": 4.1314,
+      "epoch": 0.1,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.5555,
       "step": 121
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.2e-06,
-      "loss": 4.2254,
+      "epoch": 0.1,
+      "learning_rate": 7.08e-06,
+      "loss": 4.2642,
       "step": 122
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.26e-06,
-      "loss": 4.236,
+      "epoch": 0.1,
+      "learning_rate": 7.14e-06,
+      "loss": 4.3601,
       "step": 123
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.32e-06,
-      "loss": 4.2277,
+      "epoch": 0.1,
+      "learning_rate": 7.2e-06,
+      "loss": 4.3532,
       "step": 124
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.3383,
+      "epoch": 0.11,
+      "learning_rate": 7.26e-06,
+      "loss": 4.1326,
       "step": 125
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.44e-06,
-      "loss": 4.3372,
+      "epoch": 0.11,
+      "learning_rate": 7.32e-06,
+      "loss": 4.3052,
       "step": 126
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.5e-06,
-      "loss": 4.2553,
+      "epoch": 0.11,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.5223,
       "step": 127
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.2319,
+      "epoch": 0.11,
+      "learning_rate": 7.44e-06,
+      "loss": 4.1606,
       "step": 128
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.62e-06,
-      "loss": 4.1354,
+      "epoch": 0.11,
+      "learning_rate": 7.5e-06,
+      "loss": 4.0518,
       "step": 129
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 4.2859,
+      "epoch": 0.11,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.348,
       "step": 130
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.74e-06,
-      "loss": 4.1919,
+      "epoch": 0.11,
+      "learning_rate": 7.62e-06,
+      "loss": 4.293,
       "step": 131
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.8e-06,
-      "loss": 4.2501,
+      "epoch": 0.11,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.2408,
       "step": 132
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 4.1723,
+      "epoch": 0.11,
+      "learning_rate": 7.74e-06,
+      "loss": 4.4414,
       "step": 133
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.92e-06,
-      "loss": 4.2081,
+      "epoch": 0.11,
+      "learning_rate": 7.8e-06,
+      "loss": 4.308,
       "step": 134
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.98e-06,
-      "loss": 4.3765,
+      "epoch": 0.11,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.4456,
       "step": 135
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 4.2239,
+      "epoch": 0.11,
+      "learning_rate": 7.92e-06,
+      "loss": 4.1247,
       "step": 136
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.1e-06,
-      "loss": 4.3212,
+      "epoch": 0.12,
+      "learning_rate": 7.98e-06,
+      "loss": 4.4136,
       "step": 137
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 4.3565,
+      "epoch": 0.12,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.4033,
       "step": 138
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 4.3689,
+      "epoch": 0.12,
+      "learning_rate": 8.1e-06,
+      "loss": 4.1514,
       "step": 139
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.28e-06,
-      "loss": 4.3316,
+      "epoch": 0.12,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 4.3848,
       "step": 140
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.340000000000001e-06,
-      "loss": 4.4513,
+      "epoch": 0.12,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.4218,
       "step": 141
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.3836,
+      "epoch": 0.12,
+      "learning_rate": 8.28e-06,
+      "loss": 4.261,
       "step": 142
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.459999999999999e-06,
-      "loss": 4.3245,
+      "epoch": 0.12,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 4.1636,
       "step": 143
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.52e-06,
-      "loss": 4.3805,
+      "epoch": 0.12,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.5385,
       "step": 144
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.58e-06,
-      "loss": 4.2924,
+      "epoch": 0.12,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 4.2805,
       "step": 145
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.64e-06,
-      "loss": 4.3797,
+      "epoch": 0.12,
+      "learning_rate": 8.52e-06,
+      "loss": 4.6044,
       "step": 146
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.7e-06,
-      "loss": 4.6306,
+      "epoch": 0.12,
+      "learning_rate": 8.58e-06,
+      "loss": 4.3167,
       "step": 147
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.759999999999999e-06,
-      "loss": 4.77,
+      "epoch": 0.12,
+      "learning_rate": 8.64e-06,
+      "loss": 4.2034,
       "step": 148
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.82e-06,
-      "loss": 4.5055,
+      "epoch": 0.13,
+      "learning_rate": 8.7e-06,
+      "loss": 4.4271,
       "step": 149
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.88e-06,
-      "loss": 4.6005,
+      "epoch": 0.13,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 4.2278,
       "step": 150
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 4.1378,
+      "epoch": 0.13,
+      "learning_rate": 8.82e-06,
+      "loss": 4.2755,
       "step": 151
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9e-06,
-      "loss": 4.2224,
+      "epoch": 0.13,
+      "learning_rate": 8.88e-06,
+      "loss": 4.1613,
       "step": 152
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.06e-06,
-      "loss": 4.2385,
+      "epoch": 0.13,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 4.2435,
       "step": 153
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.12e-06,
-      "loss": 4.1296,
+      "epoch": 0.13,
+      "learning_rate": 9e-06,
+      "loss": 4.3511,
       "step": 154
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.18e-06,
-      "loss": 4.2382,
+      "epoch": 0.13,
+      "learning_rate": 9.06e-06,
+      "loss": 4.2418,
       "step": 155
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.24e-06,
-      "loss": 4.0975,
+      "epoch": 0.13,
+      "learning_rate": 9.12e-06,
+      "loss": 4.0787,
       "step": 156
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.3e-06,
-      "loss": 4.1335,
+      "epoch": 0.13,
+      "learning_rate": 9.18e-06,
+      "loss": 4.2462,
       "step": 157
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.36e-06,
-      "loss": 4.2322,
+      "epoch": 0.13,
+      "learning_rate": 9.24e-06,
+      "loss": 4.4351,
       "step": 158
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.42e-06,
-      "loss": 4.2269,
+      "epoch": 0.13,
+      "learning_rate": 9.3e-06,
+      "loss": 4.1828,
       "step": 159
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.48e-06,
-      "loss": 4.2434,
+      "epoch": 0.13,
+      "learning_rate": 9.36e-06,
+      "loss": 3.98,
       "step": 160
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.54e-06,
-      "loss": 4.2971,
+      "epoch": 0.14,
+      "learning_rate": 9.42e-06,
+      "loss": 4.2868,
       "step": 161
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2479,
+      "epoch": 0.14,
+      "learning_rate": 9.48e-06,
+      "loss": 4.2953,
       "step": 162
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.66e-06,
-      "loss": 4.2019,
+      "epoch": 0.14,
+      "learning_rate": 9.54e-06,
+      "loss": 4.2162,
       "step": 163
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2686,
+      "epoch": 0.14,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.0061,
       "step": 164
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.1306,
+      "epoch": 0.14,
+      "learning_rate": 9.66e-06,
+      "loss": 4.2366,
       "step": 165
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.84e-06,
-      "loss": 4.1961,
+      "epoch": 0.14,
+      "learning_rate": 9.72e-06,
+      "loss": 3.9201,
       "step": 166
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.9e-06,
-      "loss": 4.1072,
+      "epoch": 0.14,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.2515,
       "step": 167
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.1856,
+      "epoch": 0.14,
+      "learning_rate": 9.84e-06,
+      "loss": 4.1252,
       "step": 168
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.002e-05,
-      "loss": 4.2457,
+      "epoch": 0.14,
+      "learning_rate": 9.9e-06,
+      "loss": 4.3622,
       "step": 169
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.008e-05,
-      "loss": 4.2701,
+      "epoch": 0.14,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.2236,
       "step": 170
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 4.2464,
+      "epoch": 0.14,
+      "learning_rate": 1.002e-05,
+      "loss": 4.2973,
       "step": 171
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.02e-05,
-      "loss": 4.3068,
+      "epoch": 0.14,
+      "learning_rate": 1.008e-05,
+      "loss": 4.116,
       "step": 172
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.0260000000000002e-05,
-      "loss": 4.1162,
+      "epoch": 0.15,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 4.0488,
       "step": 173
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.032e-05,
-      "loss": 4.2632,
+      "epoch": 0.15,
+      "learning_rate": 1.02e-05,
+      "loss": 4.3851,
       "step": 174
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.0379999999999999e-05,
-      "loss": 4.3134,
+      "epoch": 0.15,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 4.2722,
       "step": 175
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.044e-05,
-      "loss": 4.1883,
+      "epoch": 0.15,
+      "learning_rate": 1.032e-05,
+      "loss": 4.2434,
       "step": 176
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.05e-05,
-      "loss": 4.1563,
+      "epoch": 0.15,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 4.3755,
       "step": 177
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.0559999999999999e-05,
-      "loss": 4.2408,
+      "epoch": 0.15,
+      "learning_rate": 1.044e-05,
+      "loss": 4.1772,
       "step": 178
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.062e-05,
-      "loss": 4.3123,
+      "epoch": 0.15,
+      "learning_rate": 1.05e-05,
+      "loss": 4.2407,
       "step": 179
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.068e-05,
-      "loss": 4.301,
+      "epoch": 0.15,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 4.3553,
       "step": 180
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.074e-05,
-      "loss": 4.2258,
+      "epoch": 0.15,
+      "learning_rate": 1.062e-05,
+      "loss": 4.327,
       "step": 181
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.08e-05,
-      "loss": 4.2896,
+      "epoch": 0.15,
+      "learning_rate": 1.068e-05,
+      "loss": 4.3378,
       "step": 182
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.086e-05,
-      "loss": 4.2775,
+      "epoch": 0.15,
+      "learning_rate": 1.074e-05,
+      "loss": 4.2012,
       "step": 183
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.092e-05,
-      "loss": 4.2849,
+      "epoch": 0.15,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3031,
       "step": 184
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.098e-05,
-      "loss": 4.2654,
+      "epoch": 0.16,
+      "learning_rate": 1.086e-05,
+      "loss": 4.056,
       "step": 185
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.104e-05,
-      "loss": 4.128,
+      "epoch": 0.16,
+      "learning_rate": 1.092e-05,
+      "loss": 4.2039,
       "step": 186
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.11e-05,
-      "loss": 4.1517,
+      "epoch": 0.16,
+      "learning_rate": 1.098e-05,
+      "loss": 4.5027,
       "step": 187
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.116e-05,
-      "loss": 4.1204,
+      "epoch": 0.16,
+      "learning_rate": 1.104e-05,
+      "loss": 4.282,
       "step": 188
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.1220000000000001e-05,
-      "loss": 4.2358,
+      "epoch": 0.16,
+      "learning_rate": 1.11e-05,
+      "loss": 4.422,
       "step": 189
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.128e-05,
-      "loss": 4.3392,
+      "epoch": 0.16,
+      "learning_rate": 1.116e-05,
+      "loss": 4.0707,
       "step": 190
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.134e-05,
-      "loss": 4.254,
+      "epoch": 0.16,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 4.3089,
       "step": 191
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.2131,
+      "epoch": 0.16,
+      "learning_rate": 1.128e-05,
+      "loss": 4.5055,
       "step": 192
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.146e-05,
-      "loss": 4.2087,
+      "epoch": 0.16,
+      "learning_rate": 1.134e-05,
+      "loss": 4.5268,
       "step": 193
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.152e-05,
-      "loss": 4.3236,
+      "epoch": 0.16,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2839,
       "step": 194
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 4.4341,
+      "epoch": 0.16,
+      "learning_rate": 1.146e-05,
+      "loss": 4.4614,
       "step": 195
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.164e-05,
-      "loss": 4.3805,
+      "epoch": 0.16,
+      "learning_rate": 1.152e-05,
+      "loss": 4.3277,
       "step": 196
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 4.3194,
+      "epoch": 0.17,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 4.457,
       "step": 197
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 4.61,
+      "epoch": 0.17,
+      "learning_rate": 1.164e-05,
+      "loss": 4.5566,
       "step": 198
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.182e-05,
-      "loss": 4.5467,
+      "epoch": 0.17,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 4.5523,
       "step": 199
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 4.4941,
+      "epoch": 0.17,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 4.3267,
       "step": 200
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 4.2598,
+      "epoch": 0.17,
+      "learning_rate": 1.182e-05,
+      "loss": 4.3091,
       "step": 201
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2e-05,
-      "loss": 4.1943,
+      "epoch": 0.17,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 4.1833,
       "step": 202
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2060000000000001e-05,
-      "loss": 4.2136,
+      "epoch": 0.17,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 4.2306,
       "step": 203
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 4.209,
+      "epoch": 0.17,
+      "learning_rate": 1.2e-05,
+      "loss": 4.1851,
       "step": 204
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 4.2688,
+      "epoch": 0.17,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 4.1231,
       "step": 205
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.224e-05,
-      "loss": 4.2527,
+      "epoch": 0.17,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 4.1966,
       "step": 206
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2299999999999999e-05,
-      "loss": 4.1229,
+      "epoch": 0.17,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 4.2891,
       "step": 207
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.236e-05,
-      "loss": 4.2149,
+      "epoch": 0.17,
+      "learning_rate": 1.224e-05,
+      "loss": 4.4791,
       "step": 208
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.242e-05,
-      "loss": 4.2757,
+      "epoch": 0.18,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.3371,
       "step": 209
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.2479999999999999e-05,
-      "loss": 4.2353,
+      "epoch": 0.18,
+      "learning_rate": 1.236e-05,
+      "loss": 4.3748,
       "step": 210
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.254e-05,
-      "loss": 4.1108,
+      "epoch": 0.18,
+      "learning_rate": 1.242e-05,
+      "loss": 4.217,
       "step": 211
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.26e-05,
-      "loss": 4.16,
+      "epoch": 0.18,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.2961,
       "step": 212
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 4.0885,
+      "epoch": 0.18,
+      "learning_rate": 1.254e-05,
+      "loss": 4.2113,
       "step": 213
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.272e-05,
-      "loss": 4.1628,
+      "epoch": 0.18,
+      "learning_rate": 1.26e-05,
+      "loss": 4.2395,
       "step": 214
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.278e-05,
-      "loss": 4.209,
+      "epoch": 0.18,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.3181,
       "step": 215
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.284e-05,
-      "loss": 4.1769,
+      "epoch": 0.18,
+      "learning_rate": 1.272e-05,
+      "loss": 4.2523,
       "step": 216
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.29e-05,
-      "loss": 4.2236,
+      "epoch": 0.18,
+      "learning_rate": 1.278e-05,
+      "loss": 4.3477,
       "step": 217
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.296e-05,
-      "loss": 4.175,
+      "epoch": 0.18,
+      "learning_rate": 1.284e-05,
+      "loss": 4.2062,
       "step": 218
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.302e-05,
-      "loss": 4.1064,
+      "epoch": 0.18,
+      "learning_rate": 1.29e-05,
+      "loss": 3.9255,
       "step": 219
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.308e-05,
-      "loss": 4.2232,
+      "epoch": 0.18,
+      "learning_rate": 1.296e-05,
+      "loss": 4.2456,
       "step": 220
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.314e-05,
-      "loss": 4.1993,
+      "epoch": 0.19,
+      "learning_rate": 1.302e-05,
+      "loss": 4.212,
       "step": 221
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.32e-05,
-      "loss": 4.1825,
+      "epoch": 0.19,
+      "learning_rate": 1.308e-05,
+      "loss": 4.5293,
       "step": 222
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.326e-05,
-      "loss": 4.2186,
+      "epoch": 0.19,
+      "learning_rate": 1.314e-05,
+      "loss": 4.2978,
       "step": 223
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 4.1414,
+      "epoch": 0.19,
+      "learning_rate": 1.32e-05,
+      "loss": 4.3409,
       "step": 224
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.338e-05,
-      "loss": 4.299,
+      "epoch": 0.19,
+      "learning_rate": 1.326e-05,
+      "loss": 4.1193,
       "step": 225
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.344e-05,
-      "loss": 4.1934,
+      "epoch": 0.19,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.0577,
       "step": 226
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 4.3162,
+      "epoch": 0.19,
+      "learning_rate": 1.338e-05,
+      "loss": 4.2107,
       "step": 227
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.356e-05,
-      "loss": 4.262,
+      "epoch": 0.19,
+      "learning_rate": 1.344e-05,
+      "loss": 4.2579,
       "step": 228
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.362e-05,
-      "loss": 4.1437,
+      "epoch": 0.19,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.1397,
       "step": 229
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 4.2367,
+      "epoch": 0.19,
+      "learning_rate": 1.356e-05,
+      "loss": 4.3277,
       "step": 230
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.374e-05,
-      "loss": 4.2988,
+      "epoch": 0.19,
+      "learning_rate": 1.362e-05,
+      "loss": 4.3096,
       "step": 231
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.0968,
+      "epoch": 0.2,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.261,
       "step": 232
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 4.1227,
+      "epoch": 0.2,
+      "learning_rate": 1.374e-05,
+      "loss": 4.2361,
       "step": 233
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.392e-05,
-      "loss": 4.2133,
+      "epoch": 0.2,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.1582,
       "step": 234
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 4.2843,
+      "epoch": 0.2,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 4.2623,
       "step": 235
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 4.2125,
+      "epoch": 0.2,
+      "learning_rate": 1.392e-05,
+      "loss": 4.4154,
       "step": 236
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 4.2384,
+      "epoch": 0.2,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 4.1272,
       "step": 237
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.416e-05,
-      "loss": 4.1947,
+      "epoch": 0.2,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 4.2669,
       "step": 238
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.422e-05,
-      "loss": 4.1355,
+      "epoch": 0.2,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.245,
       "step": 239
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.428e-05,
-      "loss": 4.2412,
+      "epoch": 0.2,
+      "learning_rate": 1.416e-05,
+      "loss": 4.3029,
       "step": 240
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.434e-05,
-      "loss": 4.2237,
+      "epoch": 0.2,
+      "learning_rate": 1.422e-05,
+      "loss": 4.3906,
       "step": 241
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.44e-05,
-      "loss": 4.2662,
+      "epoch": 0.2,
+      "learning_rate": 1.428e-05,
+      "loss": 4.2033,
       "step": 242
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.446e-05,
-      "loss": 4.4641,
+      "epoch": 0.2,
+      "learning_rate": 1.434e-05,
+      "loss": 4.4401,
       "step": 243
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.452e-05,
-      "loss": 4.2923,
+      "epoch": 0.21,
+      "learning_rate": 1.44e-05,
+      "loss": 4.2719,
       "step": 244
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.458e-05,
-      "loss": 4.2855,
+      "epoch": 0.21,
+      "learning_rate": 1.446e-05,
+      "loss": 4.2957,
       "step": 245
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.464e-05,
-      "loss": 4.1766,
+      "epoch": 0.21,
+      "learning_rate": 1.452e-05,
+      "loss": 4.1976,
       "step": 246
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.47e-05,
-      "loss": 4.1232,
+      "epoch": 0.21,
+      "learning_rate": 1.458e-05,
+      "loss": 4.773,
       "step": 247
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 4.4921,
+      "epoch": 0.21,
+      "learning_rate": 1.464e-05,
+      "loss": 4.402,
       "step": 248
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.482e-05,
-      "loss": 4.5888,
+      "epoch": 0.21,
+      "learning_rate": 1.47e-05,
+      "loss": 4.332,
       "step": 249
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.488e-05,
-      "loss": 4.582,
+      "epoch": 0.21,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 4.4351,
       "step": 250
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.4940000000000001e-05,
-      "loss": 4.1951,
+      "epoch": 0.21,
+      "learning_rate": 1.482e-05,
+      "loss": 4.3525,
       "step": 251
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2269,
+      "epoch": 0.21,
+      "learning_rate": 1.488e-05,
+      "loss": 4.0294,
       "step": 252
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.506e-05,
-      "loss": 4.2679,
+      "epoch": 0.21,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 4.3153,
       "step": 253
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.5120000000000001e-05,
-      "loss": 4.1101,
+      "epoch": 0.21,
+      "learning_rate": 1.5e-05,
+      "loss": 4.0302,
       "step": 254
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.518e-05,
-      "loss": 4.1986,
+      "epoch": 0.21,
+      "learning_rate": 1.506e-05,
+      "loss": 4.0512,
       "step": 255
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.524e-05,
-      "loss": 4.179,
+      "epoch": 0.22,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 4.2336,
       "step": 256
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.53e-05,
-      "loss": 4.1623,
+      "epoch": 0.22,
+      "learning_rate": 1.518e-05,
+      "loss": 4.0394,
       "step": 257
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.5360000000000002e-05,
-      "loss": 4.1636,
+      "epoch": 0.22,
+      "learning_rate": 1.524e-05,
+      "loss": 4.4877,
       "step": 258
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.542e-05,
-      "loss": 4.1744,
+      "epoch": 0.22,
+      "learning_rate": 1.53e-05,
+      "loss": 4.269,
       "step": 259
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.548e-05,
-      "loss": 4.1098,
+      "epoch": 0.22,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 4.2188,
       "step": 260
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.554e-05,
-      "loss": 4.2371,
+      "epoch": 0.22,
+      "learning_rate": 1.542e-05,
+      "loss": 4.2159,
       "step": 261
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.56e-05,
-      "loss": 4.1562,
+      "epoch": 0.22,
+      "learning_rate": 1.548e-05,
+      "loss": 4.19,
       "step": 262
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.5660000000000003e-05,
-      "loss": 4.294,
+      "epoch": 0.22,
+      "learning_rate": 1.554e-05,
+      "loss": 4.3155,
       "step": 263
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.5720000000000002e-05,
-      "loss": 4.1616,
+      "epoch": 0.22,
+      "learning_rate": 1.56e-05,
+      "loss": 4.3023,
       "step": 264
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.578e-05,
-      "loss": 4.2108,
+      "epoch": 0.22,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 4.2151,
       "step": 265
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.584e-05,
-      "loss": 4.1689,
+      "epoch": 0.22,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 4.2791,
       "step": 266
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.59e-05,
-      "loss": 4.2019,
+      "epoch": 0.22,
+      "learning_rate": 1.578e-05,
+      "loss": 4.1519,
       "step": 267
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.596e-05,
-      "loss": 4.1509,
+      "epoch": 0.23,
+      "learning_rate": 1.584e-05,
+      "loss": 4.19,
       "step": 268
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6020000000000002e-05,
-      "loss": 4.2425,
+      "epoch": 0.23,
+      "learning_rate": 1.59e-05,
+      "loss": 4.188,
       "step": 269
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6080000000000002e-05,
-      "loss": 4.1185,
+      "epoch": 0.23,
+      "learning_rate": 1.596e-05,
+      "loss": 4.1215,
       "step": 270
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.614e-05,
-      "loss": 4.0561,
+      "epoch": 0.23,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 4.4414,
       "step": 271
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.62e-05,
-      "loss": 4.2334,
+      "epoch": 0.23,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 4.0973,
       "step": 272
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.626e-05,
-      "loss": 4.03,
+      "epoch": 0.23,
+      "learning_rate": 1.614e-05,
+      "loss": 4.1955,
       "step": 273
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6320000000000003e-05,
-      "loss": 4.1374,
+      "epoch": 0.23,
+      "learning_rate": 1.62e-05,
+      "loss": 4.249,
       "step": 274
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6380000000000002e-05,
-      "loss": 4.2188,
+      "epoch": 0.23,
+      "learning_rate": 1.626e-05,
+      "loss": 4.2878,
       "step": 275
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6440000000000002e-05,
-      "loss": 4.1667,
+      "epoch": 0.23,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 4.0312,
       "step": 276
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.65e-05,
-      "loss": 4.1154,
+      "epoch": 0.23,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 4.0002,
       "step": 277
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.656e-05,
-      "loss": 4.1921,
+      "epoch": 0.23,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 4.1866,
       "step": 278
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6620000000000004e-05,
-      "loss": 4.1726,
+      "epoch": 0.23,
+      "learning_rate": 1.65e-05,
+      "loss": 4.2889,
       "step": 279
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6680000000000003e-05,
-      "loss": 4.3145,
+      "epoch": 0.24,
+      "learning_rate": 1.656e-05,
+      "loss": 4.1737,
       "step": 280
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6740000000000002e-05,
-      "loss": 4.2087,
+      "epoch": 0.24,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 4.1932,
       "step": 281
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.0653,
+      "epoch": 0.24,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 4.307,
       "step": 282
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.686e-05,
-      "loss": 4.0902,
+      "epoch": 0.24,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 4.3193,
       "step": 283
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.6919999999999997e-05,
-      "loss": 4.1814,
+      "epoch": 0.24,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.1298,
       "step": 284
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.698e-05,
-      "loss": 4.2828,
+      "epoch": 0.24,
+      "learning_rate": 1.686e-05,
+      "loss": 4.3486,
       "step": 285
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.704e-05,
-      "loss": 4.1693,
+      "epoch": 0.24,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 4.1724,
       "step": 286
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.71e-05,
-      "loss": 4.2818,
+      "epoch": 0.24,
+      "learning_rate": 1.698e-05,
+      "loss": 4.115,
       "step": 287
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.716e-05,
-      "loss": 4.2331,
+      "epoch": 0.24,
+      "learning_rate": 1.704e-05,
+      "loss": 4.1832,
       "step": 288
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.7219999999999998e-05,
-      "loss": 4.1357,
+      "epoch": 0.24,
+      "learning_rate": 1.71e-05,
+      "loss": 4.4367,
       "step": 289
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.728e-05,
-      "loss": 4.1985,
+      "epoch": 0.24,
+      "learning_rate": 1.716e-05,
+      "loss": 4.3428,
       "step": 290
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.734e-05,
-      "loss": 4.4431,
+      "epoch": 0.24,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 4.3138,
       "step": 291
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.74e-05,
-      "loss": 4.3144,
+      "epoch": 0.25,
+      "learning_rate": 1.728e-05,
+      "loss": 4.566,
       "step": 292
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.746e-05,
-      "loss": 4.3839,
+      "epoch": 0.25,
+      "learning_rate": 1.734e-05,
+      "loss": 4.1059,
       "step": 293
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.7519999999999998e-05,
-      "loss": 4.3518,
+      "epoch": 0.25,
+      "learning_rate": 1.74e-05,
+      "loss": 4.4551,
       "step": 294
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.758e-05,
-      "loss": 4.3501,
+      "epoch": 0.25,
+      "learning_rate": 1.746e-05,
+      "loss": 4.3844,
       "step": 295
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.764e-05,
-      "loss": 4.3526,
+      "epoch": 0.25,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 4.9761,
       "step": 296
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.77e-05,
-      "loss": 4.2403,
+      "epoch": 0.25,
+      "learning_rate": 1.758e-05,
+      "loss": 4.2972,
       "step": 297
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.776e-05,
-      "loss": 4.5569,
+      "epoch": 0.25,
+      "learning_rate": 1.764e-05,
+      "loss": 4.8831,
       "step": 298
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.782e-05,
-      "loss": 4.3044,
+      "epoch": 0.25,
+      "learning_rate": 1.77e-05,
+      "loss": 4.5865,
       "step": 299
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.7879999999999998e-05,
-      "loss": 4.7419,
+      "epoch": 0.25,
+      "learning_rate": 1.776e-05,
+      "loss": 4.7354,
       "step": 300
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.794e-05,
-      "loss": 4.2255,
+      "epoch": 0.25,
+      "learning_rate": 1.782e-05,
+      "loss": 4.0228,
       "step": 301
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.8e-05,
-      "loss": 4.1911,
+      "epoch": 0.25,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 4.1383,
       "step": 302
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.806e-05,
-      "loss": 4.1913,
+      "epoch": 0.25,
+      "learning_rate": 1.794e-05,
+      "loss": 4.2188,
       "step": 303
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.812e-05,
-      "loss": 4.2035,
+      "epoch": 0.26,
+      "learning_rate": 1.8e-05,
+      "loss": 3.8823,
       "step": 304
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.818e-05,
-      "loss": 4.1659,
+      "epoch": 0.26,
+      "learning_rate": 1.806e-05,
+      "loss": 4.0361,
       "step": 305
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.824e-05,
-      "loss": 4.1522,
+      "epoch": 0.26,
+      "learning_rate": 1.812e-05,
+      "loss": 4.0446,
       "step": 306
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.83e-05,
-      "loss": 4.1422,
+      "epoch": 0.26,
+      "learning_rate": 1.818e-05,
+      "loss": 4.2072,
       "step": 307
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.836e-05,
-      "loss": 4.148,
+      "epoch": 0.26,
+      "learning_rate": 1.824e-05,
+      "loss": 3.9454,
       "step": 308
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.842e-05,
-      "loss": 4.1458,
+      "epoch": 0.26,
+      "learning_rate": 1.83e-05,
+      "loss": 3.9268,
       "step": 309
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.848e-05,
-      "loss": 4.2421,
+      "epoch": 0.26,
+      "learning_rate": 1.836e-05,
+      "loss": 4.1389,
       "step": 310
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.854e-05,
-      "loss": 4.0367,
+      "epoch": 0.26,
+      "learning_rate": 1.842e-05,
+      "loss": 4.058,
       "step": 311
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.86e-05,
-      "loss": 4.2137,
+      "epoch": 0.26,
+      "learning_rate": 1.848e-05,
+      "loss": 4.0705,
       "step": 312
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.866e-05,
-      "loss": 4.2195,
+      "epoch": 0.26,
+      "learning_rate": 1.854e-05,
+      "loss": 4.0209,
       "step": 313
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.872e-05,
-      "loss": 4.1313,
+      "epoch": 0.26,
+      "learning_rate": 1.86e-05,
+      "loss": 4.1257,
       "step": 314
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.878e-05,
-      "loss": 4.165,
+      "epoch": 0.26,
+      "learning_rate": 1.866e-05,
+      "loss": 4.0073,
       "step": 315
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.884e-05,
-      "loss": 4.1387,
+      "epoch": 0.27,
+      "learning_rate": 1.872e-05,
+      "loss": 4.3839,
       "step": 316
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 4.1316,
+      "epoch": 0.27,
+      "learning_rate": 1.878e-05,
+      "loss": 4.3243,
       "step": 317
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.896e-05,
-      "loss": 4.164,
+      "epoch": 0.27,
+      "learning_rate": 1.884e-05,
+      "loss": 4.2505,
       "step": 318
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.902e-05,
-      "loss": 4.1583,
+      "epoch": 0.27,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 4.0626,
       "step": 319
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.908e-05,
-      "loss": 4.2331,
+      "epoch": 0.27,
+      "learning_rate": 1.896e-05,
+      "loss": 4.2428,
       "step": 320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.914e-05,
-      "loss": 4.1241,
+      "epoch": 0.27,
+      "learning_rate": 1.902e-05,
+      "loss": 4.0947,
       "step": 321
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 3.9994,
+      "epoch": 0.27,
+      "learning_rate": 1.908e-05,
+      "loss": 4.1172,
       "step": 322
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.9260000000000002e-05,
-      "loss": 4.1963,
+      "epoch": 0.27,
+      "learning_rate": 1.914e-05,
+      "loss": 4.048,
       "step": 323
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.932e-05,
-      "loss": 4.05,
+      "epoch": 0.27,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.27,
       "step": 324
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.938e-05,
-      "loss": 4.2003,
+      "epoch": 0.27,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 4.1968,
       "step": 325
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.944e-05,
-      "loss": 4.1707,
+      "epoch": 0.27,
+      "learning_rate": 1.932e-05,
+      "loss": 4.3657,
       "step": 326
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.95e-05,
-      "loss": 4.1238,
+      "epoch": 0.27,
+      "learning_rate": 1.938e-05,
+      "loss": 4.3547,
       "step": 327
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.9560000000000002e-05,
-      "loss": 4.0829,
+      "epoch": 0.28,
+      "learning_rate": 1.944e-05,
+      "loss": 4.2394,
       "step": 328
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.9620000000000002e-05,
-      "loss": 4.1401,
+      "epoch": 0.28,
+      "learning_rate": 1.95e-05,
+      "loss": 4.127,
       "step": 329
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.968e-05,
-      "loss": 4.1984,
+      "epoch": 0.28,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 4.3617,
       "step": 330
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.974e-05,
-      "loss": 4.2043,
+      "epoch": 0.28,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 4.0746,
       "step": 331
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.98e-05,
-      "loss": 4.0911,
+      "epoch": 0.28,
+      "learning_rate": 1.968e-05,
+      "loss": 4.2261,
       "step": 332
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.9860000000000003e-05,
-      "loss": 4.2525,
+      "epoch": 0.28,
+      "learning_rate": 1.974e-05,
+      "loss": 4.2513,
       "step": 333
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.9920000000000002e-05,
-      "loss": 4.1515,
+      "epoch": 0.28,
+      "learning_rate": 1.98e-05,
+      "loss": 4.029,
       "step": 334
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.9980000000000002e-05,
-      "loss": 4.3075,
+      "epoch": 0.28,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 4.2843,
       "step": 335
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 2.004e-05,
-      "loss": 4.1172,
+      "epoch": 0.28,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 4.2978,
       "step": 336
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.01e-05,
-      "loss": 4.1193,
+      "epoch": 0.28,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 4.2217,
       "step": 337
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.016e-05,
-      "loss": 4.2335,
+      "epoch": 0.28,
+      "learning_rate": 2.004e-05,
+      "loss": 4.4017,
       "step": 338
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.0220000000000003e-05,
-      "loss": 4.2131,
+      "epoch": 0.28,
+      "learning_rate": 2.01e-05,
+      "loss": 4.3868,
       "step": 339
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.0280000000000002e-05,
-      "loss": 4.264,
+      "epoch": 0.29,
+      "learning_rate": 2.016e-05,
+      "loss": 4.2188,
       "step": 340
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.0340000000000002e-05,
-      "loss": 4.1833,
+      "epoch": 0.29,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 4.2795,
       "step": 341
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.04e-05,
-      "loss": 4.3401,
+      "epoch": 0.29,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 4.3846,
       "step": 342
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.046e-05,
-      "loss": 4.2661,
+      "epoch": 0.29,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 4.3164,
       "step": 343
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.0520000000000003e-05,
-      "loss": 4.3547,
+      "epoch": 0.29,
+      "learning_rate": 2.04e-05,
+      "loss": 4.4459,
       "step": 344
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.0580000000000003e-05,
-      "loss": 4.096,
+      "epoch": 0.29,
+      "learning_rate": 2.046e-05,
+      "loss": 4.256,
       "step": 345
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.064e-05,
-      "loss": 4.339,
+      "epoch": 0.29,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 4.2019,
       "step": 346
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.07e-05,
-      "loss": 4.3258,
+      "epoch": 0.29,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 4.4257,
       "step": 347
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.0759999999999998e-05,
-      "loss": 4.2641,
+      "epoch": 0.29,
+      "learning_rate": 2.064e-05,
+      "loss": 4.6603,
       "step": 348
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.082e-05,
-      "loss": 4.4812,
+      "epoch": 0.29,
+      "learning_rate": 2.07e-05,
+      "loss": 4.857,
       "step": 349
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.088e-05,
-      "loss": 4.4627,
+      "epoch": 0.29,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 4.4458,
       "step": 350
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.094e-05,
-      "loss": 4.2444,
+      "epoch": 0.3,
+      "learning_rate": 2.082e-05,
+      "loss": 4.4633,
       "step": 351
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1e-05,
-      "loss": 4.1637,
+      "epoch": 0.3,
+      "learning_rate": 2.088e-05,
+      "loss": 4.1266,
       "step": 352
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1059999999999998e-05,
-      "loss": 4.1012,
+      "epoch": 0.3,
+      "learning_rate": 2.094e-05,
+      "loss": 4.14,
       "step": 353
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1119999999999998e-05,
-      "loss": 3.9923,
+      "epoch": 0.3,
+      "learning_rate": 2.1e-05,
+      "loss": 4.3007,
       "step": 354
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.118e-05,
-      "loss": 4.175,
+      "epoch": 0.3,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 4.1424,
       "step": 355
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.124e-05,
-      "loss": 4.2816,
+      "epoch": 0.3,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 4.1509,
       "step": 356
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.13e-05,
-      "loss": 4.1508,
+      "epoch": 0.3,
+      "learning_rate": 2.118e-05,
+      "loss": 4.2986,
       "step": 357
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.136e-05,
-      "loss": 4.2011,
+      "epoch": 0.3,
+      "learning_rate": 2.124e-05,
+      "loss": 4.0543,
       "step": 358
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.1419999999999998e-05,
-      "loss": 4.0786,
+      "epoch": 0.3,
+      "learning_rate": 2.13e-05,
+      "loss": 4.2081,
       "step": 359
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.148e-05,
-      "loss": 4.1572,
+      "epoch": 0.3,
+      "learning_rate": 2.136e-05,
+      "loss": 4.3371,
       "step": 360
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.154e-05,
-      "loss": 4.0844,
+      "epoch": 0.3,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 4.2052,
       "step": 361
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.16e-05,
-      "loss": 4.1476,
+      "epoch": 0.3,
+      "learning_rate": 2.148e-05,
+      "loss": 4.0907,
       "step": 362
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.166e-05,
-      "loss": 4.0828,
+      "epoch": 0.31,
+      "learning_rate": 2.154e-05,
+      "loss": 4.126,
       "step": 363
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.172e-05,
-      "loss": 4.1726,
+      "epoch": 0.31,
+      "learning_rate": 2.16e-05,
+      "loss": 4.2623,
       "step": 364
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.178e-05,
-      "loss": 4.2892,
+      "epoch": 0.31,
+      "learning_rate": 2.166e-05,
+      "loss": 4.2171,
       "step": 365
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.184e-05,
-      "loss": 4.0957,
+      "epoch": 0.31,
+      "learning_rate": 2.172e-05,
+      "loss": 4.1573,
       "step": 366
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.19e-05,
-      "loss": 4.1822,
+      "epoch": 0.31,
+      "learning_rate": 2.178e-05,
+      "loss": 4.0573,
       "step": 367
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.196e-05,
-      "loss": 4.1647,
+      "epoch": 0.31,
+      "learning_rate": 2.184e-05,
+      "loss": 4.4715,
       "step": 368
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.202e-05,
-      "loss": 4.0817,
+      "epoch": 0.31,
+      "learning_rate": 2.19e-05,
+      "loss": 4.3253,
       "step": 369
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.208e-05,
-      "loss": 4.0894,
+      "epoch": 0.31,
+      "learning_rate": 2.196e-05,
+      "loss": 4.2055,
       "step": 370
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.214e-05,
-      "loss": 4.079,
+      "epoch": 0.31,
+      "learning_rate": 2.202e-05,
+      "loss": 4.1163,
       "step": 371
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.22e-05,
-      "loss": 4.1652,
+      "epoch": 0.31,
+      "learning_rate": 2.208e-05,
+      "loss": 4.0746,
       "step": 372
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.226e-05,
-      "loss": 4.1596,
+      "epoch": 0.31,
+      "learning_rate": 2.214e-05,
+      "loss": 4.0822,
       "step": 373
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.232e-05,
-      "loss": 3.9787,
+      "epoch": 0.31,
+      "learning_rate": 2.22e-05,
+      "loss": 4.0934,
       "step": 374
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.238e-05,
-      "loss": 4.1849,
+      "epoch": 0.32,
+      "learning_rate": 2.226e-05,
+      "loss": 4.1601,
       "step": 375
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.2440000000000002e-05,
-      "loss": 4.0419,
+      "epoch": 0.32,
+      "learning_rate": 2.232e-05,
+      "loss": 4.2061,
       "step": 376
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.25e-05,
-      "loss": 4.145,
+      "epoch": 0.32,
+      "learning_rate": 2.238e-05,
+      "loss": 4.136,
       "step": 377
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.256e-05,
-      "loss": 4.1548,
+      "epoch": 0.32,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 4.1212,
       "step": 378
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.262e-05,
-      "loss": 4.0737,
+      "epoch": 0.32,
+      "learning_rate": 2.25e-05,
+      "loss": 4.091,
       "step": 379
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.268e-05,
-      "loss": 4.096,
+      "epoch": 0.32,
+      "learning_rate": 2.256e-05,
+      "loss": 4.2768,
       "step": 380
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.274e-05,
-      "loss": 4.1361,
+      "epoch": 0.32,
+      "learning_rate": 2.262e-05,
+      "loss": 4.0621,
       "step": 381
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.2105,
+      "epoch": 0.32,
+      "learning_rate": 2.268e-05,
+      "loss": 4.2477,
       "step": 382
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.286e-05,
-      "loss": 4.1431,
+      "epoch": 0.32,
+      "learning_rate": 2.274e-05,
+      "loss": 4.0139,
       "step": 383
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.292e-05,
-      "loss": 4.0593,
+      "epoch": 0.32,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.3174,
       "step": 384
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.298e-05,
-      "loss": 4.16,
+      "epoch": 0.32,
+      "learning_rate": 2.286e-05,
+      "loss": 4.3421,
       "step": 385
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.304e-05,
-      "loss": 4.0336,
+      "epoch": 0.32,
+      "learning_rate": 2.292e-05,
+      "loss": 4.0962,
       "step": 386
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.3100000000000002e-05,
-      "loss": 4.2153,
+      "epoch": 0.33,
+      "learning_rate": 2.298e-05,
+      "loss": 4.322,
       "step": 387
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.3160000000000002e-05,
-      "loss": 4.2426,
+      "epoch": 0.33,
+      "learning_rate": 2.304e-05,
+      "loss": 4.2927,
       "step": 388
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.322e-05,
-      "loss": 4.1324,
+      "epoch": 0.33,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 4.365,
       "step": 389
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.328e-05,
-      "loss": 4.126,
+      "epoch": 0.33,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 4.3296,
       "step": 390
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.334e-05,
-      "loss": 4.2552,
+      "epoch": 0.33,
+      "learning_rate": 2.322e-05,
+      "loss": 4.3811,
       "step": 391
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.0773,
+      "epoch": 0.33,
+      "learning_rate": 2.328e-05,
+      "loss": 4.2609,
       "step": 392
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.3460000000000002e-05,
-      "loss": 4.0848,
+      "epoch": 0.33,
+      "learning_rate": 2.334e-05,
+      "loss": 4.2523,
       "step": 393
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.3520000000000002e-05,
-      "loss": 4.1795,
+      "epoch": 0.33,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.3928,
       "step": 394
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.358e-05,
-      "loss": 4.3042,
+      "epoch": 0.33,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 4.1325,
       "step": 395
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.364e-05,
-      "loss": 4.2524,
+      "epoch": 0.33,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 4.5287,
       "step": 396
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.37e-05,
-      "loss": 4.4209,
+      "epoch": 0.33,
+      "learning_rate": 2.358e-05,
+      "loss": 4.4165,
       "step": 397
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.3760000000000003e-05,
-      "loss": 4.3868,
+      "epoch": 0.33,
+      "learning_rate": 2.364e-05,
+      "loss": 4.3961,
       "step": 398
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.3820000000000002e-05,
-      "loss": 4.5116,
+      "epoch": 0.34,
+      "learning_rate": 2.37e-05,
+      "loss": 4.6827,
       "step": 399
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.3880000000000002e-05,
-      "loss": 4.4561,
+      "epoch": 0.34,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 4.5686,
       "step": 400
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.394e-05,
-      "loss": 4.1715,
+      "epoch": 0.34,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 4.0445,
       "step": 401
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.4e-05,
-      "loss": 4.018,
+      "epoch": 0.34,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 4.1595,
       "step": 402
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.4060000000000003e-05,
-      "loss": 4.0975,
+      "epoch": 0.34,
+      "learning_rate": 2.394e-05,
+      "loss": 4.1869,
       "step": 403
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4120000000000003e-05,
-      "loss": 4.0795,
+      "epoch": 0.34,
+      "learning_rate": 2.4e-05,
+      "loss": 4.2626,
       "step": 404
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4180000000000002e-05,
-      "loss": 4.0454,
+      "epoch": 0.34,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 4.1379,
       "step": 405
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4240000000000002e-05,
-      "loss": 4.1319,
+      "epoch": 0.34,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 4.1241,
       "step": 406
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.43e-05,
-      "loss": 4.1041,
+      "epoch": 0.34,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 4.0632,
       "step": 407
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4360000000000004e-05,
-      "loss": 4.1619,
+      "epoch": 0.34,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 4.1654,
       "step": 408
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.442e-05,
-      "loss": 4.1178,
+      "epoch": 0.34,
+      "learning_rate": 2.43e-05,
+      "loss": 4.1701,
       "step": 409
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.448e-05,
-      "loss": 4.0931,
+      "epoch": 0.34,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 4.327,
       "step": 410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.454e-05,
-      "loss": 4.314,
+      "epoch": 0.35,
+      "learning_rate": 2.442e-05,
+      "loss": 4.3407,
       "step": 411
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 4.0508,
+      "epoch": 0.35,
+      "learning_rate": 2.448e-05,
+      "loss": 4.2136,
       "step": 412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.4659999999999998e-05,
-      "loss": 4.003,
+      "epoch": 0.35,
+      "learning_rate": 2.454e-05,
+      "loss": 4.2112,
       "step": 413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.472e-05,
-      "loss": 4.1375,
+      "epoch": 0.35,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.0556,
       "step": 414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.478e-05,
-      "loss": 4.2206,
+      "epoch": 0.35,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 4.2036,
       "step": 415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.484e-05,
-      "loss": 4.0785,
+      "epoch": 0.35,
+      "learning_rate": 2.472e-05,
+      "loss": 4.2702,
       "step": 416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.49e-05,
-      "loss": 4.1078,
+      "epoch": 0.35,
+      "learning_rate": 2.478e-05,
+      "loss": 3.9489,
       "step": 417
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.4959999999999998e-05,
-      "loss": 4.0014,
+      "epoch": 0.35,
+      "learning_rate": 2.484e-05,
+      "loss": 4.0711,
       "step": 418
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.502e-05,
-      "loss": 4.0941,
+      "epoch": 0.35,
+      "learning_rate": 2.49e-05,
+      "loss": 4.2809,
       "step": 419
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.508e-05,
-      "loss": 4.0181,
+      "epoch": 0.35,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 4.176,
       "step": 420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.514e-05,
-      "loss": 4.1846,
+      "epoch": 0.35,
+      "learning_rate": 2.502e-05,
+      "loss": 4.1256,
       "step": 421
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.52e-05,
-      "loss": 4.0393,
-      "step": 422
+      "epoch": 0.35,
+      "learning_rate": 2.508e-05,
+      "loss": 4.2436,
+      "step": 422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.514e-05,
+      "loss": 4.2221,
+      "step": 423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.52e-05,
+      "loss": 4.1591,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.526e-05,
+      "loss": 4.4836,
+      "step": 425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 4.1937,
+      "step": 426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.538e-05,
+      "loss": 4.1032,
+      "step": 427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.544e-05,
+      "loss": 4.1523,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.55e-05,
+      "loss": 4.419,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.556e-05,
+      "loss": 4.3495,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.562e-05,
+      "loss": 4.2146,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.568e-05,
+      "loss": 4.1366,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.574e-05,
+      "loss": 4.2631,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.58e-05,
+      "loss": 4.1508,
+      "step": 434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.586e-05,
+      "loss": 4.3975,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.592e-05,
+      "loss": 4.0812,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 4.0992,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.604e-05,
+      "loss": 4.084,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.61e-05,
+      "loss": 4.3237,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.616e-05,
+      "loss": 4.1893,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.622e-05,
+      "loss": 4.298,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.628e-05,
+      "loss": 4.2912,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 4.1628,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.64e-05,
+      "loss": 4.2758,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.646e-05,
+      "loss": 4.2355,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.652e-05,
+      "loss": 4.5063,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.658e-05,
+      "loss": 4.753,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 4.4271,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 4.5121,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.676e-05,
+      "loss": 4.5602,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.682e-05,
+      "loss": 4.0981,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.688e-05,
+      "loss": 4.0877,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 4.1266,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2176,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 4.2328,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.712e-05,
+      "loss": 4.2785,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.718e-05,
+      "loss": 4.0902,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.724e-05,
+      "loss": 4.3384,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 4.1156,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 4.2531,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 4.387,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.748e-05,
+      "loss": 4.2243,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.754e-05,
+      "loss": 4.1874,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.3125,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 4.3213,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 4.0718,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.778e-05,
+      "loss": 4.2087,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.784e-05,
+      "loss": 4.2517,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79e-05,
+      "loss": 4.0324,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 4.3931,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 4.2216,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 4.3404,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 3.9645,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.306,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.826e-05,
+      "loss": 4.3816,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.832e-05,
+      "loss": 4.3026,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.838e-05,
+      "loss": 4.0065,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.844e-05,
+      "loss": 4.2526,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 4.2118,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.856e-05,
+      "loss": 4.2084,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.862e-05,
+      "loss": 4.1341,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.868e-05,
+      "loss": 4.2129,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.874e-05,
+      "loss": 4.4199,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.88e-05,
+      "loss": 4.372,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 4.2345,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.892e-05,
+      "loss": 4.2636,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.898e-05,
+      "loss": 4.0573,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.904e-05,
+      "loss": 4.1976,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.91e-05,
+      "loss": 4.0942,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.916e-05,
+      "loss": 4.1982,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.922e-05,
+      "loss": 4.2939,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.928e-05,
+      "loss": 4.1793,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.934e-05,
+      "loss": 4.4447,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.94e-05,
+      "loss": 4.3125,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.946e-05,
+      "loss": 4.654,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 4.39,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.958e-05,
+      "loss": 4.3429,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.964e-05,
+      "loss": 4.5546,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.97e-05,
+      "loss": 4.3525,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.976e-05,
+      "loss": 4.3963,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 4.31812047958374,
+      "eval_runtime": 787.054,
+      "eval_samples_per_second": 3.357,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 1.8917096390321302,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.982e-05,
+      "loss": 4.2787,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 4.3156,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.994e-05,
+      "loss": 4.2268,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3e-05,
+      "loss": 3.9769,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.995645863570392e-05,
+      "loss": 4.2368,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9912917271407838e-05,
+      "loss": 4.3288,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9869375907111758e-05,
+      "loss": 4.0865,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9825834542815675e-05,
+      "loss": 4.1491,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9782293178519592e-05,
+      "loss": 4.2185,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9738751814223512e-05,
+      "loss": 4.4993,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9695210449927432e-05,
+      "loss": 4.2993,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651669085631353e-05,
+      "loss": 4.0531,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.960812772133527e-05,
+      "loss": 4.3561,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.956458635703919e-05,
+      "loss": 4.0485,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9521044992743107e-05,
+      "loss": 4.0402,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9477503628447024e-05,
+      "loss": 4.1555,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9433962264150944e-05,
+      "loss": 4.2702,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9390420899854864e-05,
+      "loss": 4.1078,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.934687953555878e-05,
+      "loss": 4.5942,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.93033381712627e-05,
+      "loss": 4.1278,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925979680696662e-05,
+      "loss": 4.2122,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9216255442670535e-05,
+      "loss": 4.2119,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9172714078374455e-05,
+      "loss": 4.2674,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9129172714078376e-05,
+      "loss": 4.2288,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9085631349782292e-05,
+      "loss": 4.2191,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9042089985486213e-05,
+      "loss": 3.983,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8998548621190133e-05,
+      "loss": 4.254,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8955007256894053e-05,
+      "loss": 4.2543,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8911465892597967e-05,
+      "loss": 4.4138,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8867924528301887e-05,
+      "loss": 4.3523,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8824383164005807e-05,
+      "loss": 4.3558,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8780841799709724e-05,
+      "loss": 4.3372,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8737300435413644e-05,
+      "loss": 4.2213,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8693759071117565e-05,
+      "loss": 4.2444,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.865021770682148e-05,
+      "loss": 4.1607,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.86066763425254e-05,
+      "loss": 4.098,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.856313497822932e-05,
+      "loss": 4.2933,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8519593613933236e-05,
+      "loss": 3.9004,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8476052249637156e-05,
+      "loss": 4.3056,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8432510885341076e-05,
+      "loss": 4.2941,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8388969521044996e-05,
+      "loss": 4.3823,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.834542815674891e-05,
+      "loss": 4.2052,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 4.1273,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.825834542815675e-05,
+      "loss": 4.325,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8214804063860667e-05,
+      "loss": 4.4019,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8171262699564588e-05,
+      "loss": 4.2391,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8127721335268508e-05,
+      "loss": 4.1486,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8084179970972425e-05,
+      "loss": 4.4644,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.804063860667634e-05,
+      "loss": 4.536,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7997097242380262e-05,
+      "loss": 4.9046,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.795355587808418e-05,
+      "loss": 4.3163,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.79100145137881e-05,
+      "loss": 4.2158,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.786647314949202e-05,
+      "loss": 4.2927,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7822931785195936e-05,
+      "loss": 4.3969,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7779390420899856e-05,
+      "loss": 4.3762,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7735849056603773e-05,
+      "loss": 4.1186,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 4.1002,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.764876632801161e-05,
+      "loss": 4.1572,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.760522496371553e-05,
+      "loss": 4.3174,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.756168359941945e-05,
+      "loss": 4.1761,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7518142235123368e-05,
+      "loss": 4.1761,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7474600870827288e-05,
+      "loss": 4.0344,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7431059506531205e-05,
+      "loss": 4.36,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7387518142235122e-05,
+      "loss": 3.9784,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7343976777939042e-05,
+      "loss": 4.0456,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7300435413642962e-05,
+      "loss": 4.2407,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.725689404934688e-05,
+      "loss": 4.2338,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.72133526850508e-05,
+      "loss": 4.3216,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.716981132075472e-05,
+      "loss": 4.0335,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7126269956458637e-05,
+      "loss": 4.1294,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7082728592162554e-05,
+      "loss": 4.1418,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7039187227866474e-05,
+      "loss": 4.3184,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6995645863570394e-05,
+      "loss": 4.1821,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.695210449927431e-05,
+      "loss": 4.3712,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.690856313497823e-05,
+      "loss": 4.3184,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.686502177068215e-05,
+      "loss": 4.3245,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6821480406386065e-05,
+      "loss": 4.3666,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6777939042089985e-05,
+      "loss": 4.1713,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6734397677793905e-05,
+      "loss": 4.2383,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6690856313497822e-05,
+      "loss": 4.2657,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6647314949201743e-05,
+      "loss": 4.4122,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6603773584905663e-05,
+      "loss": 4.0174,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656023222060958e-05,
+      "loss": 4.1298,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6516690856313497e-05,
+      "loss": 4.2765,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6473149492017417e-05,
+      "loss": 4.3216,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6429608127721337e-05,
+      "loss": 4.2649,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6386066763425254e-05,
+      "loss": 4.3561,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6342525399129174e-05,
+      "loss": 4.2005,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6298984034833095e-05,
+      "loss": 4.1685,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.625544267053701e-05,
+      "loss": 4.194,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.621190130624093e-05,
+      "loss": 4.2619,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.616835994194485e-05,
+      "loss": 4.1578,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6124818577648765e-05,
+      "loss": 4.6091,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6081277213352686e-05,
+      "loss": 4.224,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6037735849056606e-05,
+      "loss": 4.3454,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5994194484760523e-05,
+      "loss": 4.0568,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5950653120464443e-05,
+      "loss": 4.2776,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.590711175616836e-05,
+      "loss": 4.5071,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5863570391872277e-05,
+      "loss": 4.4919,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5820029027576197e-05,
+      "loss": 4.7298,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5776487663280117e-05,
+      "loss": 4.1329,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5732946298984038e-05,
+      "loss": 4.2925,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5689404934687955e-05,
+      "loss": 4.4772,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.564586357039187e-05,
+      "loss": 4.0057,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5602322206095792e-05,
+      "loss": 4.2021,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.555878084179971e-05,
+      "loss": 4.3052,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.551523947750363e-05,
+      "loss": 4.1938,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 4.3898,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5428156748911466e-05,
+      "loss": 4.2565,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 4.2614,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5341074020319303e-05,
+      "loss": 4.2616,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.529753265602322e-05,
+      "loss": 4.0055,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.525399129172714e-05,
+      "loss": 4.2784,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.521044992743106e-05,
+      "loss": 4.0664,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.516690856313498e-05,
+      "loss": 4.1826,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5123367198838898e-05,
+      "loss": 4.2279,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5079825834542818e-05,
+      "loss": 4.1516,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5036284470246735e-05,
+      "loss": 4.1771,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4992743105950652e-05,
+      "loss": 4.2239,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4949201741654572e-05,
+      "loss": 4.1977,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4905660377358492e-05,
+      "loss": 3.9837,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.486211901306241e-05,
+      "loss": 3.8723,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.481857764876633e-05,
+      "loss": 4.3487,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.477503628447025e-05,
+      "loss": 4.2749,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4731494920174163e-05,
+      "loss": 4.2056,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4687953555878083e-05,
+      "loss": 4.4245,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4644412191582004e-05,
+      "loss": 4.3677,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.460087082728592e-05,
+      "loss": 3.9882,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.455732946298984e-05,
+      "loss": 4.1001,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.451378809869376e-05,
+      "loss": 4.2242,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.447024673439768e-05,
+      "loss": 4.4039,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4426705370101595e-05,
+      "loss": 4.1222,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4383164005805515e-05,
+      "loss": 4.2713,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4339622641509435e-05,
+      "loss": 4.1611,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4296081277213352e-05,
+      "loss": 4.3203,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4252539912917273e-05,
+      "loss": 4.1684,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4208998548621193e-05,
+      "loss": 3.8974,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.416545718432511e-05,
+      "loss": 4.117,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4121915820029027e-05,
+      "loss": 4.1567,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4078374455732947e-05,
+      "loss": 4.2168,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4034833091436864e-05,
+      "loss": 4.161,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3991291727140784e-05,
+      "loss": 4.2408,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3947750362844704e-05,
+      "loss": 4.4083,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3904208998548625e-05,
+      "loss": 4.2046,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.386066763425254e-05,
+      "loss": 4.2269,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3817126269956458e-05,
+      "loss": 4.1862,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.377358490566038e-05,
+      "loss": 4.0266,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3730043541364295e-05,
+      "loss": 4.5443,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3686502177068216e-05,
+      "loss": 4.6949,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3642960812772136e-05,
+      "loss": 4.6649,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3599419448476053e-05,
+      "loss": 4.3659,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3555878084179973e-05,
+      "loss": 4.1146,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.351233671988389e-05,
+      "loss": 4.3208,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3468795355587807e-05,
+      "loss": 4.228,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3425253991291727e-05,
+      "loss": 4.2048,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3381712626995647e-05,
+      "loss": 4.186,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3338171262699564e-05,
+      "loss": 4.1094,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3294629898403485e-05,
+      "loss": 4.2268,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3251088534107405e-05,
+      "loss": 4.1989,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.320754716981132e-05,
+      "loss": 4.1938,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.316400580551524e-05,
+      "loss": 4.2022,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.312046444121916e-05,
+      "loss": 4.0532,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 4.2375,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3033381712626996e-05,
+      "loss": 4.1408,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2989840348330916e-05,
+      "loss": 4.4725,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2946298984034833e-05,
+      "loss": 4.2757,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.290275761973875e-05,
+      "loss": 4.3083,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.285921625544267e-05,
+      "loss": 4.1377,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.281567489114659e-05,
+      "loss": 4.1596,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2772133526850507e-05,
+      "loss": 4.1153,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2728592162554428e-05,
+      "loss": 4.2722,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2685050798258348e-05,
+      "loss": 4.1801,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 4.2868,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.259796806966618e-05,
+      "loss": 4.2577,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2554426705370102e-05,
+      "loss": 4.1359,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2510885341074022e-05,
+      "loss": 3.9345,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.246734397677794e-05,
+      "loss": 4.3573,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.242380261248186e-05,
+      "loss": 4.1647,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.238026124818578e-05,
+      "loss": 4.1087,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2336719883889693e-05,
+      "loss": 4.3817,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2293178519593613e-05,
+      "loss": 3.9119,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2249637155297534e-05,
+      "loss": 4.0308,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.220609579100145e-05,
+      "loss": 4.2724,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.216255442670537e-05,
+      "loss": 4.0064,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.211901306240929e-05,
+      "loss": 4.1809,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2075471698113208e-05,
+      "loss": 4.2179,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2031930333817125e-05,
+      "loss": 4.1127,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1988388969521045e-05,
+      "loss": 4.1502,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1944847605224965e-05,
+      "loss": 4.3217,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1901306240928882e-05,
+      "loss": 4.3293,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1857764876632802e-05,
+      "loss": 4.5591,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1814223512336723e-05,
+      "loss": 4.3901,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.177068214804064e-05,
+      "loss": 4.4107,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1727140783744556e-05,
+      "loss": 4.4495,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1683599419448477e-05,
+      "loss": 4.189,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1640058055152394e-05,
+      "loss": 4.1249,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1596516690856314e-05,
+      "loss": 4.2779,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1552975326560234e-05,
+      "loss": 4.6044,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.150943396226415e-05,
+      "loss": 4.0719,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.146589259796807e-05,
+      "loss": 4.8907,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1422351233671988e-05,
+      "loss": 4.3704,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.137880986937591e-05,
+      "loss": 4.0879,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1335268505079825e-05,
+      "loss": 4.1881,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1291727140783746e-05,
+      "loss": 3.9277,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1248185776487666e-05,
+      "loss": 4.0999,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1204644412191583e-05,
+      "loss": 4.1482,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1161103047895503e-05,
+      "loss": 4.1645,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.111756168359942e-05,
+      "loss": 4.129,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1074020319303337e-05,
+      "loss": 4.1714,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1030478955007257e-05,
+      "loss": 4.2756,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0986937590711177e-05,
+      "loss": 4.0141,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0943396226415094e-05,
+      "loss": 4.3078,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0899854862119014e-05,
+      "loss": 4.165,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0856313497822935e-05,
+      "loss": 4.0277,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0812772133526848e-05,
+      "loss": 4.1871,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.076923076923077e-05,
+      "loss": 4.0522,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.072568940493469e-05,
+      "loss": 4.0718,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.068214804063861e-05,
+      "loss": 4.3626,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0638606676342526e-05,
+      "loss": 4.1405,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0595065312046446e-05,
+      "loss": 4.0301,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0551523947750363e-05,
+      "loss": 4.111,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.050798258345428e-05,
+      "loss": 3.9836,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.04644412191582e-05,
+      "loss": 4.1762,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.042089985486212e-05,
+      "loss": 4.0696,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0377358490566037e-05,
+      "loss": 4.217,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0333817126269958e-05,
+      "loss": 4.2628,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0290275761973878e-05,
+      "loss": 4.1171,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.024673439767779e-05,
+      "loss": 4.2818,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.020319303338171e-05,
+      "loss": 3.9628,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0159651669085632e-05,
+      "loss": 4.308,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.011611030478955e-05,
+      "loss": 4.2634,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.007256894049347e-05,
+      "loss": 4.4191,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.002902757619739e-05,
+      "loss": 3.9985,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.998548621190131e-05,
+      "loss": 4.246,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9941944847605223e-05,
+      "loss": 4.1971,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9898403483309143e-05,
+      "loss": 4.1536,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9854862119013064e-05,
+      "loss": 4.1814,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.981132075471698e-05,
+      "loss": 4.3588,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97677793904209e-05,
+      "loss": 4.2966,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.972423802612482e-05,
+      "loss": 4.0711,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9680696661828738e-05,
+      "loss": 4.3214,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9637155297532655e-05,
+      "loss": 4.1784,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9593613933236575e-05,
+      "loss": 4.2183,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9550072568940492e-05,
+      "loss": 4.3362,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9506531204644412e-05,
+      "loss": 4.4376,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9462989840348332e-05,
+      "loss": 4.4175,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9419448476052253e-05,
+      "loss": 4.5554,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.937590711175617e-05,
+      "loss": 4.6952,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9332365747460086e-05,
+      "loss": 4.2484,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9288824383164007e-05,
+      "loss": 4.534,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9245283018867924e-05,
+      "loss": 4.1023,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9201741654571844e-05,
+      "loss": 4.2307,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9158200290275764e-05,
+      "loss": 4.2812,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.911465892597968e-05,
+      "loss": 4.0733,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.90711175616836e-05,
+      "loss": 4.279,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9027576197387518e-05,
+      "loss": 4.1167,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8984034833091435e-05,
+      "loss": 4.1397,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8940493468795355e-05,
+      "loss": 3.9688,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8896952104499276e-05,
+      "loss": 4.1564,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8853410740203192e-05,
+      "loss": 4.3238,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8809869375907113e-05,
+      "loss": 4.0137,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8766328011611033e-05,
+      "loss": 4.1269,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.872278664731495e-05,
+      "loss": 4.1683,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8679245283018867e-05,
+      "loss": 4.057,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8635703918722787e-05,
+      "loss": 4.2914,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8592162554426707e-05,
+      "loss": 4.1258,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8548621190130624e-05,
+      "loss": 4.1764,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8505079825834544e-05,
+      "loss": 3.9162,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 4.1707,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8417997097242378e-05,
+      "loss": 4.3203,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.83744557329463e-05,
+      "loss": 3.8903,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.833091436865022e-05,
+      "loss": 4.2466,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8287373004354136e-05,
+      "loss": 4.0481,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243831640058056e-05,
+      "loss": 4.0653,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200290275761976e-05,
+      "loss": 4.1995,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8156748911465896e-05,
+      "loss": 4.1336,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.811320754716981e-05,
+      "loss": 4.1521,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.806966618287373e-05,
+      "loss": 4.1294,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.802612481857765e-05,
+      "loss": 4.023,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7982583454281567e-05,
+      "loss": 4.2847,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7939042089985487e-05,
+      "loss": 4.052,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7895500725689408e-05,
+      "loss": 4.1232,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7851959361393325e-05,
+      "loss": 3.9496,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.780841799709724e-05,
+      "loss": 4.2887,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7764876632801162e-05,
+      "loss": 4.3544,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.772133526850508e-05,
+      "loss": 4.0901,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7677793904209e-05,
+      "loss": 4.3868,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.763425253991292e-05,
+      "loss": 4.0689,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7590711175616836e-05,
+      "loss": 4.0721,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7547169811320753e-05,
+      "loss": 3.9722,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7503628447024673e-05,
+      "loss": 4.5048,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7460087082728593e-05,
+      "loss": 4.4479,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 4.6251,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.737300435413643e-05,
+      "loss": 4.1257,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.732946298984035e-05,
+      "loss": 4.2497,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7285921625544268e-05,
+      "loss": 3.9545,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7242380261248185e-05,
+      "loss": 4.1237,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7198838896952105e-05,
+      "loss": 4.4149,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7155297532656022e-05,
+      "loss": 4.4912,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7111756168359942e-05,
+      "loss": 4.2902,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7068214804063862e-05,
+      "loss": 4.0741,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.702467343976778e-05,
+      "loss": 4.1417,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 4.0497,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6937590711175616e-05,
+      "loss": 4.1369,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6894049346879537e-05,
+      "loss": 3.9978,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6850507982583453e-05,
+      "loss": 4.1982,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6806966618287374e-05,
+      "loss": 4.185,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6763425253991294e-05,
+      "loss": 4.0565,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.671988388969521e-05,
+      "loss": 4.0964,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.667634252539913e-05,
+      "loss": 4.0345,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6632801161103048e-05,
+      "loss": 4.1398,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6589259796806965e-05,
+      "loss": 4.2538,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6545718432510885e-05,
+      "loss": 4.1444,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6502177068214805e-05,
+      "loss": 4.4523,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6458635703918722e-05,
+      "loss": 4.1366,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 4.2139,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6371552975326563e-05,
+      "loss": 4.0953,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6328011611030476e-05,
+      "loss": 4.0001,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6284470246734397e-05,
+      "loss": 4.0284,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6240928882438317e-05,
+      "loss": 4.2043,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6197387518142237e-05,
+      "loss": 4.1797,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 3.9392,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6110304789550074e-05,
+      "loss": 4.218,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6066763425253995e-05,
+      "loss": 4.4048,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6023222060957908e-05,
+      "loss": 4.1203,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5979680696661828e-05,
+      "loss": 4.4287,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.593613933236575e-05,
+      "loss": 4.1276,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5892597968069665e-05,
+      "loss": 4.2576,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 4.0118,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5805515239477506e-05,
+      "loss": 4.2405,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5761973875181423e-05,
+      "loss": 4.1189,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.571843251088534e-05,
+      "loss": 4.1439,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.567489114658926e-05,
+      "loss": 3.9951,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.563134978229318e-05,
+      "loss": 4.2016,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5587808417997097e-05,
+      "loss": 4.2312,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5544267053701017e-05,
+      "loss": 4.3068,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5500725689404938e-05,
+      "loss": 4.0535,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5457184325108855e-05,
+      "loss": 3.9266,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.541364296081277e-05,
+      "loss": 4.2312,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.537010159651669e-05,
+      "loss": 3.9523,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.532656023222061e-05,
+      "loss": 4.2394,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.528301886792453e-05,
+      "loss": 4.2336,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5239477503628449e-05,
+      "loss": 4.4971,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5195936139332364e-05,
+      "loss": 4.3696,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5152394775036285e-05,
+      "loss": 4.1565,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5108853410740205e-05,
+      "loss": 4.0786,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.506531204644412e-05,
+      "loss": 4.4144,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.502177068214804e-05,
+      "loss": 3.9351,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.497822931785196e-05,
+      "loss": 4.4926,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4934687953555879e-05,
+      "loss": 4.4327,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4891146589259796e-05,
+      "loss": 4.2866,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4847605224963716e-05,
+      "loss": 4.1438,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4804063860667635e-05,
+      "loss": 4.2642,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4760522496371553e-05,
+      "loss": 4.1644,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 4.1274,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.467343976777939e-05,
+      "loss": 4.1132,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.462989840348331e-05,
+      "loss": 4.1611,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4586357039187228e-05,
+      "loss": 3.987,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4542815674891146e-05,
+      "loss": 4.1212,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4499274310595066e-05,
+      "loss": 4.1824,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4455732946298983e-05,
+      "loss": 3.9273,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4412191582002904e-05,
+      "loss": 4.2457,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4368650217706822e-05,
+      "loss": 4.2206,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.432510885341074e-05,
+      "loss": 4.1049,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.428156748911466e-05,
+      "loss": 4.0865,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4238026124818578e-05,
+      "loss": 4.0894,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4194484760522498e-05,
+      "loss": 4.0147,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 4.1724,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4107402031930334e-05,
+      "loss": 4.2902,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4063860667634254e-05,
+      "loss": 3.8653,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.402031930333817e-05,
+      "loss": 3.8608,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.397677793904209e-05,
+      "loss": 4.0124,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.393323657474601e-05,
+      "loss": 4.1798,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3889695210449928e-05,
+      "loss": 3.9899,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 4.1595,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3802612481857765e-05,
+      "loss": 3.9731,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3759071117561684e-05,
+      "loss": 3.9706,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3715529753265602e-05,
+      "loss": 4.2816,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3671988388969521e-05,
+      "loss": 4.0969,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.362844702467344e-05,
+      "loss": 4.3128,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.358490566037736e-05,
+      "loss": 3.9761,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3541364296081277e-05,
+      "loss": 4.2406,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3497822931785197e-05,
+      "loss": 4.1138,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3454281567489116e-05,
+      "loss": 4.1666,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3410740203193032e-05,
+      "loss": 4.0527,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3367198838896953e-05,
+      "loss": 4.1166,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3323657474600871e-05,
+      "loss": 4.3572,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.328011611030479e-05,
+      "loss": 4.2719,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3236574746008708e-05,
+      "loss": 4.3508,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3193033381712627e-05,
+      "loss": 4.3323,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3149492017416547e-05,
+      "loss": 4.3149,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3105950653120464e-05,
+      "loss": 4.197,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3062409288824383e-05,
+      "loss": 4.1674,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 4.1212,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2975326560232222e-05,
+      "loss": 4.238,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2931785195936138e-05,
+      "loss": 4.2748,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2888243831640059e-05,
+      "loss": 4.2855,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2844702467343977e-05,
+      "loss": 4.1303,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2801161103047896e-05,
+      "loss": 4.2541,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2757619738751814e-05,
+      "loss": 4.3779,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2714078374455733e-05,
+      "loss": 4.1411,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2670537010159652e-05,
+      "loss": 4.2128,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.262699564586357e-05,
+      "loss": 4.1722,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.258345428156749e-05,
+      "loss": 4.2009,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2539912917271409e-05,
+      "loss": 4.0748,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2496371552975326e-05,
+      "loss": 4.2604,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 4.1866,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2409288824383165e-05,
+      "loss": 4.475,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2365747460087082e-05,
+      "loss": 4.1919,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2322206095791002e-05,
+      "loss": 4.3414,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.227866473149492e-05,
+      "loss": 4.0319,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.223512336719884e-05,
+      "loss": 4.0561,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2191582002902758e-05,
+      "loss": 4.1455,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2148040638606676e-05,
+      "loss": 4.1583,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2104499274310596e-05,
+      "loss": 4.2151,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2060957910014513e-05,
+      "loss": 4.108,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2017416545718432e-05,
+      "loss": 4.1194,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1973875181422352e-05,
+      "loss": 4.0818,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.193033381712627e-05,
+      "loss": 4.3585,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 4.2281,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1843251088534108e-05,
+      "loss": 4.0967,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1799709724238026e-05,
+      "loss": 4.2058,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1756168359941945e-05,
+      "loss": 4.0207,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712626995645864e-05,
+      "loss": 4.0735,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1669085631349782e-05,
+      "loss": 3.9778,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1625544267053702e-05,
+      "loss": 4.0005,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.158200290275762e-05,
+      "loss": 4.1877,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 4.223,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1494920174165458e-05,
+      "loss": 4.0355,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1451378809869375e-05,
+      "loss": 3.9679,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1407837445573295e-05,
+      "loss": 4.2022,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1364296081277214e-05,
+      "loss": 4.2689,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 4.1395,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1277213352685051e-05,
+      "loss": 4.2064,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.123367198838897e-05,
+      "loss": 4.2839,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119013062409289e-05,
+      "loss": 4.2356,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1146589259796807e-05,
+      "loss": 4.1739,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1103047895500725e-05,
+      "loss": 4.269,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1059506531204646e-05,
+      "loss": 4.1956,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1015965166908562e-05,
+      "loss": 4.1559,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0972423802612483e-05,
+      "loss": 3.997,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0928882438316401e-05,
+      "loss": 4.2529,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088534107402032e-05,
+      "loss": 4.0163,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0841799709724238e-05,
+      "loss": 4.2044,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0798258345428157e-05,
+      "loss": 4.3213,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 4.0306,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0711175616835994e-05,
+      "loss": 4.2202,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0667634252539913e-05,
+      "loss": 4.3561,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0624092888243833e-05,
+      "loss": 4.3484,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0580551523947752e-05,
+      "loss": 4.1115,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0537010159651668e-05,
+      "loss": 4.0041,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0493468795355589e-05,
+      "loss": 4.2221,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0449927431059507e-05,
+      "loss": 3.9924,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0406386066763424e-05,
+      "loss": 3.8731,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0362844702467344e-05,
+      "loss": 4.0766,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0319303338171263e-05,
+      "loss": 4.2045,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0275761973875182e-05,
+      "loss": 4.2499,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.02322206095791e-05,
+      "loss": 4.0216,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0188679245283019e-05,
+      "loss": 4.1061,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0145137880986939e-05,
+      "loss": 4.0627,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0101596516690856e-05,
+      "loss": 4.0515,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0058055152394774e-05,
+      "loss": 4.1712,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0014513788098695e-05,
+      "loss": 3.948,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.970972423802612e-06,
+      "loss": 4.2556,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.927431059506532e-06,
+      "loss": 4.265,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.88388969521045e-06,
+      "loss": 3.9826,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.840348330914369e-06,
+      "loss": 4.1475,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.796806966618287e-06,
+      "loss": 4.2088,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.753265602322206e-06,
+      "loss": 3.8011,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.709724238026126e-06,
+      "loss": 4.0959,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.666182873730043e-06,
+      "loss": 3.961,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.622641509433962e-06,
+      "loss": 3.9991,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 4.1277,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.5355587808418e-06,
+      "loss": 4.0237,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.492017416545717e-06,
+      "loss": 4.1151,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.448476052249638e-06,
+      "loss": 4.1315,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.404934687953556e-06,
+      "loss": 3.9966,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.361393323657475e-06,
+      "loss": 4.1165,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.317851959361393e-06,
+      "loss": 4.014,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.274310595065312e-06,
+      "loss": 4.1335,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 4.3056,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.18722786647315e-06,
+      "loss": 4.2319,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.143686502177068e-06,
+      "loss": 4.0098,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.100145137880988e-06,
+      "loss": 4.2086,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.056603773584905e-06,
+      "loss": 3.9627,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.013062409288825e-06,
+      "loss": 4.0762,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.969521044992744e-06,
+      "loss": 4.2008,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.925979680696662e-06,
+      "loss": 4.0906,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.882438316400581e-06,
+      "loss": 4.139,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.8388969521045e-06,
+      "loss": 4.1577,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.795355587808418e-06,
+      "loss": 4.2921,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.751814223512337e-06,
+      "loss": 4.0552,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 3.8965,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.664731494920175e-06,
+      "loss": 4.1591,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.621190130624092e-06,
+      "loss": 4.2888,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.577648766328011e-06,
+      "loss": 4.6172,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.534107402031931e-06,
+      "loss": 4.3232,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 4.2875,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.447024673439768e-06,
+      "loss": 4.3191,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.403483309143687e-06,
+      "loss": 4.4804,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.192998886108398,
+      "eval_runtime": 786.9912,
+      "eval_samples_per_second": 3.357,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 1.903669178897263,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.359941944847605e-06,
+      "loss": 4.2412,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.316400580551524e-06,
+      "loss": 4.1815,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.272859216255443e-06,
+      "loss": 4.1036,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.229317851959361e-06,
+      "loss": 4.1248,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.185776487663281e-06,
+      "loss": 4.2296,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.142235123367198e-06,
+      "loss": 4.0708,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.098693759071119e-06,
+      "loss": 4.0667,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.055152394775037e-06,
+      "loss": 4.0618,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.011611030478954e-06,
+      "loss": 4.0827,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.968069666182874e-06,
+      "loss": 4.0883,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 4.0383,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.880986937590711e-06,
+      "loss": 4.1827,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.83744557329463e-06,
+      "loss": 3.999,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.793904208998549e-06,
+      "loss": 4.0975,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.750362844702469e-06,
+      "loss": 4.1168,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.706821480406386e-06,
+      "loss": 4.2072,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.663280116110304e-06,
+      "loss": 4.0171,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.6197387518142245e-06,
+      "loss": 4.277,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.576197387518142e-06,
+      "loss": 4.2476,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.53265602322206e-06,
+      "loss": 3.9786,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.48911465892598e-06,
+      "loss": 4.0455,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.445573294629898e-06,
+      "loss": 4.3022,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.402031930333817e-06,
+      "loss": 4.2723,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 3.9124,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.314949201741655e-06,
+      "loss": 4.0956,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.271407837445573e-06,
+      "loss": 3.9983,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.227866473149492e-06,
+      "loss": 4.1068,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.184325108853411e-06,
+      "loss": 4.0294,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.14078374455733e-06,
+      "loss": 3.9695,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.097242380261249e-06,
+      "loss": 3.8736,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.053701015965167e-06,
+      "loss": 4.1041,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.010159651669085e-06,
+      "loss": 4.0364,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.966618287373005e-06,
+      "loss": 4.091,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 3.9847,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.879535558780842e-06,
+      "loss": 3.9737,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.8359941944847605e-06,
+      "loss": 3.9495,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.79245283018868e-06,
+      "loss": 4.0117,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7489114658925985e-06,
+      "loss": 4.3146,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.705370101596516e-06,
+      "loss": 4.0996,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.661828737300436e-06,
+      "loss": 4.0728,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.618287373004354e-06,
+      "loss": 4.2033,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.574746008708274e-06,
+      "loss": 4.0457,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.531204644412191e-06,
+      "loss": 4.0855,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.487663280116111e-06,
+      "loss": 4.2396,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.444121915820029e-06,
+      "loss": 4.2983,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.400580551523948e-06,
+      "loss": 4.0847,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.3570391872278665e-06,
+      "loss": 4.1117,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.313497822931785e-06,
+      "loss": 4.3389,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2699564586357045e-06,
+      "loss": 4.6018,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 4.3337,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.182873730043541e-06,
+      "loss": 4.1061,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.13933236574746e-06,
+      "loss": 4.0278,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095791001451379e-06,
+      "loss": 4.0041,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052249637155298e-06,
+      "loss": 4.117,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008708272859216e-06,
+      "loss": 4.1773,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.965166908563135e-06,
+      "loss": 4.2456,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921625544267054e-06,
+      "loss": 4.2032,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8780841799709725e-06,
+      "loss": 3.9239,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.834542815674891e-06,
+      "loss": 4.193,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79100145137881e-06,
+      "loss": 4.1571,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.747460087082729e-06,
+      "loss": 4.2011,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.703918722786648e-06,
+      "loss": 4.0113,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 4.1774,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.616835994194485e-06,
+      "loss": 3.9938,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573294629898403e-06,
+      "loss": 3.8992,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529753265602323e-06,
+      "loss": 4.0046,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.486211901306241e-06,
+      "loss": 4.0542,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44267053701016e-06,
+      "loss": 4.0333,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3991291727140785e-06,
+      "loss": 4.0606,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355587808417997e-06,
+      "loss": 4.261,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3120464441219165e-06,
+      "loss": 4.0941,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.268505079825834e-06,
+      "loss": 4.0718,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 3.8236,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181422351233672e-06,
+      "loss": 3.9818,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137880986937591e-06,
+      "loss": 3.9589,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.094339622641509e-06,
+      "loss": 4.0444,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.050798258345428e-06,
+      "loss": 4.4012,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.007256894049347e-06,
+      "loss": 4.1033,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.963715529753266e-06,
+      "loss": 4.1105,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9201741654571845e-06,
+      "loss": 3.9048,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.876632801161103e-06,
+      "loss": 4.2026,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.833091436865022e-06,
+      "loss": 4.1863,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 4.0533,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.746008708272859e-06,
+      "loss": 4.0973,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702467343976778e-06,
+      "loss": 4.1554,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658925979680697e-06,
+      "loss": 4.0183,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 4.1119,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.571843251088534e-06,
+      "loss": 4.0165,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5283018867924524e-06,
+      "loss": 4.0389,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.484760522496372e-06,
+      "loss": 4.0758,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4412191582002904e-06,
+      "loss": 4.0803,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.397677793904209e-06,
+      "loss": 4.0622,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 4.0949,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.310595065312046e-06,
+      "loss": 4.1664,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2670537010159656e-06,
+      "loss": 4.1606,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.223512336719884e-06,
+      "loss": 4.6159,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.179970972423803e-06,
+      "loss": 4.3542,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.136429608127721e-06,
+      "loss": 4.4235,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.092888243831641e-06,
+      "loss": 4.5862,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.049346879535559e-06,
+      "loss": 4.2408,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.005805515239477e-06,
+      "loss": 4.3313,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 4.1019,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.918722786647315e-06,
+      "loss": 4.0683,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.875181422351234e-06,
+      "loss": 3.9354,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.831640058055152e-06,
+      "loss": 4.0619,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.788098693759071e-06,
+      "loss": 3.9154,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.74455732946299e-06,
+      "loss": 4.4415,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7010159651669087e-06,
+      "loss": 4.2145,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6574746008708277e-06,
+      "loss": 4.1533,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.613933236574746e-06,
+      "loss": 4.0206,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.570391872278665e-06,
+      "loss": 4.0455,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5268505079825834e-06,
+      "loss": 4.3857,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4833091436865024e-06,
+      "loss": 4.1454,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439767779390421e-06,
+      "loss": 3.8138,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.39622641509434e-06,
+      "loss": 4.1425,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.352685050798258e-06,
+      "loss": 4.2171,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.309143686502177e-06,
+      "loss": 4.2071,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2656023222060957e-06,
+      "loss": 3.9879,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2220609579100147e-06,
+      "loss": 3.8181,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1785195936139333e-06,
+      "loss": 4.1376,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1349782293178522e-06,
+      "loss": 4.1359,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0914368650217704e-06,
+      "loss": 4.2224,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0478955007256894e-06,
+      "loss": 3.9595,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.004354136429608e-06,
+      "loss": 4.1257,
+      "step": 1124
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.526e-05,
-      "loss": 4.1369,
-      "step": 423
+      "learning_rate": 2.960812772133527e-06,
+      "loss": 4.3169,
+      "step": 1125
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.5319999999999998e-05,
-      "loss": 4.1978,
-      "step": 424
+      "learning_rate": 2.9172714078374455e-06,
+      "loss": 4.1508,
+      "step": 1126
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.538e-05,
-      "loss": 4.1363,
-      "step": 425
+      "learning_rate": 2.8737300435413645e-06,
+      "loss": 4.2589,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 4.0514,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7866473149492017e-06,
+      "loss": 4.1067,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7431059506531207e-06,
+      "loss": 4.1904,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6995645863570392e-06,
+      "loss": 3.8716,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6560232220609582e-06,
+      "loss": 3.939,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 4.2239,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5689404934687954e-06,
+      "loss": 4.1028,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.525399129172714e-06,
+      "loss": 4.2637,
+      "step": 1135
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.544e-05,
-      "loss": 4.1793,
-      "step": 426
+      "learning_rate": 2.481857764876633e-06,
+      "loss": 4.0135,
+      "step": 1136
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.55e-05,
-      "loss": 4.1114,
-      "step": 427
+      "learning_rate": 2.4383164005805515e-06,
+      "loss": 4.0396,
+      "step": 1137
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.556e-05,
-      "loss": 4.1015,
-      "step": 428
+      "learning_rate": 2.3947750362844705e-06,
+      "loss": 4.0986,
+      "step": 1138
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.562e-05,
-      "loss": 4.0305,
-      "step": 429
+      "learning_rate": 2.351233671988389e-06,
+      "loss": 4.0424,
+      "step": 1139
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.568e-05,
-      "loss": 4.1179,
-      "step": 430
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 3.9483,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2641509433962262e-06,
+      "loss": 4.3512,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2206095791001452e-06,
+      "loss": 4.0812,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 4.2587,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1335268505079828e-06,
+      "loss": 4.3133,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0899854862119014e-06,
+      "loss": 4.1824,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0464441219158204e-06,
+      "loss": 4.3971,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0029027576197385e-06,
+      "loss": 4.7171,
+      "step": 1147
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.574e-05,
-      "loss": 4.1957,
-      "step": 431
+      "learning_rate": 1.9593613933236575e-06,
+      "loss": 4.3438,
+      "step": 1148
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.58e-05,
-      "loss": 4.3161,
-      "step": 432
+      "learning_rate": 1.915820029027576e-06,
+      "loss": 4.5947,
+      "step": 1149
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.586e-05,
-      "loss": 4.1238,
-      "step": 433
+      "learning_rate": 1.872278664731495e-06,
+      "loss": 4.2554,
+      "step": 1150
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.592e-05,
-      "loss": 4.1152,
-      "step": 434
+      "learning_rate": 1.8287373004354138e-06,
+      "loss": 4.0357,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7851959361393324e-06,
+      "loss": 3.9277,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7416545718432512e-06,
+      "loss": 3.9927,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.69811320754717e-06,
+      "loss": 3.9426,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6545718432510886e-06,
+      "loss": 4.0832,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6110304789550073e-06,
+      "loss": 4.1312,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5674891146589261e-06,
+      "loss": 4.0275,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5239477503628447e-06,
+      "loss": 4.0904,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4804063860667635e-06,
+      "loss": 3.9514,
+      "step": 1159
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.5980000000000002e-05,
-      "loss": 4.1593,
-      "step": 435
+      "learning_rate": 1.4368650217706823e-06,
+      "loss": 4.1812,
+      "step": 1160
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.604e-05,
-      "loss": 4.0797,
-      "step": 436
+      "learning_rate": 1.3933236574746008e-06,
+      "loss": 3.9168,
+      "step": 1161
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.61e-05,
-      "loss": 4.0625,
-      "step": 437
+      "learning_rate": 1.3497822931785196e-06,
+      "loss": 4.1842,
+      "step": 1162
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.616e-05,
-      "loss": 4.2553,
-      "step": 438
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 4.2224,
+      "step": 1163
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.622e-05,
-      "loss": 4.1998,
-      "step": 439
+      "learning_rate": 1.262699564586357e-06,
+      "loss": 3.9669,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2191582002902758e-06,
+      "loss": 4.0835,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1756168359941945e-06,
+      "loss": 4.19,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1320754716981131e-06,
+      "loss": 3.8765,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.088534107402032e-06,
+      "loss": 4.0299,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0449927431059507e-06,
+      "loss": 4.0296,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0014513788098693e-06,
+      "loss": 4.147,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.57910014513788e-07,
+      "loss": 4.0634,
+      "step": 1171
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.628e-05,
-      "loss": 4.2637,
-      "step": 440
+      "learning_rate": 9.143686502177069e-07,
+      "loss": 3.9348,
+      "step": 1172
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.6340000000000002e-05,
-      "loss": 4.2109,
-      "step": 441
+      "learning_rate": 8.708272859216256e-07,
+      "loss": 4.1515,
+      "step": 1173
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.64e-05,
-      "loss": 4.4059,
-      "step": 442
+      "learning_rate": 8.272859216255443e-07,
+      "loss": 4.0434,
+      "step": 1174
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.646e-05,
-      "loss": 4.3216,
-      "step": 443
+      "learning_rate": 7.837445573294631e-07,
+      "loss": 3.9705,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.402031930333817e-07,
+      "loss": 4.0213,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.966618287373004e-07,
+      "loss": 4.149,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.531204644412192e-07,
+      "loss": 4.4807,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.095791001451379e-07,
+      "loss": 3.9835,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.660377358490566e-07,
+      "loss": 4.1959,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.224963715529753e-07,
+      "loss": 4.1525,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.78955007256894e-07,
+      "loss": 4.0851,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.354136429608128e-07,
+      "loss": 4.1103,
+      "step": 1183
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.652e-05,
-      "loss": 4.2793,
-      "step": 444
+      "learning_rate": 3.9187227866473153e-07,
+      "loss": 4.1269,
+      "step": 1184
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.658e-05,
-      "loss": 4.5485,
-      "step": 445
+      "learning_rate": 3.483309143686502e-07,
+      "loss": 4.1185,
+      "step": 1185
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.6640000000000002e-05,
-      "loss": 4.2807,
-      "step": 446
+      "learning_rate": 3.0478955007256894e-07,
+      "loss": 4.2908,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6124818577648767e-07,
+      "loss": 4.2001,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.177068214804064e-07,
+      "loss": 4.228,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.741654571843251e-07,
+      "loss": 4.3516,
+      "step": 1189
     },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 4.298295111399595,
-      "train_runtime": 5163.9194,
-      "train_samples_per_second": 5.526,
-      "train_steps_per_second": 0.086
+      "train_loss": 4.242165094268132,
+      "train_runtime": 6714.5583,
+      "train_samples_per_second": 4.25,
+      "train_steps_per_second": 0.177
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,