diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,1805 +1,10754 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9987389659520807,
-  "global_step": 297,
+  "epoch": 1.0,
+  "global_step": 1784,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.7416,
+      "learning_rate": 0.0,
+      "loss": 4.8987,
       "step": 1
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.822,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.0275,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8917,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.1495,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.833,
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 4.6853,
       "step": 4
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.7162,
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 4.7796,
       "step": 5
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.8449,
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 4.8181,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.8237,
+      "epoch": 0.0,
+      "learning_rate": 2.4e-06,
+      "loss": 4.668,
       "step": 7
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.8885,
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 4.8098,
       "step": 8
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.7995,
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 5.0043,
       "step": 9
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.8e-06,
-      "loss": 4.6334,
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 4.6714,
       "step": 10
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.765,
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 4.7061,
       "step": 11
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.2e-06,
-      "loss": 4.7846,
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 4.6284,
       "step": 12
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.6955,
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 4.7973,
       "step": 13
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6e-06,
-      "loss": 4.6238,
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 4.4247,
       "step": 14
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.6308,
+      "epoch": 0.01,
+      "learning_rate": 7.2e-06,
+      "loss": 4.5414,
       "step": 15
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 3e-06,
-      "loss": 4.5522,
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 4.6507,
       "step": 16
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.6306,
+      "epoch": 0.01,
+      "learning_rate": 8.4e-06,
+      "loss": 4.6245,
       "step": 17
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.6244,
+      "epoch": 0.01,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 4.4725,
       "step": 18
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.6e-06,
-      "loss": 4.5672,
+      "epoch": 0.01,
+      "learning_rate": 9.6e-06,
+      "loss": 4.4363,
       "step": 19
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.8e-06,
-      "loss": 4.6278,
+      "epoch": 0.01,
+      "learning_rate": 1.02e-05,
+      "loss": 4.7024,
       "step": 20
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.6397,
+      "epoch": 0.01,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 4.3845,
       "step": 21
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.5934,
+      "epoch": 0.01,
+      "learning_rate": 1.14e-05,
+      "loss": 4.7751,
       "step": 22
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.4e-06,
-      "loss": 4.4719,
+      "epoch": 0.01,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 4.5633,
       "step": 23
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6e-06,
-      "loss": 4.6009,
+      "epoch": 0.01,
+      "learning_rate": 1.26e-05,
+      "loss": 4.315,
       "step": 24
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.4169,
+      "epoch": 0.01,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 4.4592,
       "step": 25
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5e-06,
-      "loss": 4.7091,
+      "epoch": 0.01,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 4.584,
       "step": 26
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5.2e-06,
-      "loss": 4.4467,
+      "epoch": 0.02,
+      "learning_rate": 1.44e-05,
+      "loss": 4.3579,
       "step": 27
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 5.4e-06,
-      "loss": 4.4314,
+      "epoch": 0.02,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 4.0938,
       "step": 28
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.3324,
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 4.4154,
       "step": 29
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 5.8e-06,
-      "loss": 4.4922,
+      "epoch": 0.02,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 4.3189,
       "step": 30
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 6e-06,
-      "loss": 4.4856,
+      "epoch": 0.02,
+      "learning_rate": 1.68e-05,
+      "loss": 4.2552,
       "step": 31
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.2e-06,
-      "loss": 4.3944,
+      "epoch": 0.02,
+      "learning_rate": 1.74e-05,
+      "loss": 4.5623,
       "step": 32
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.4152,
+      "epoch": 0.02,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 4.2836,
       "step": 33
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 6.6e-06,
-      "loss": 4.3853,
+      "epoch": 0.02,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 4.1511,
       "step": 34
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.3918,
+      "epoch": 0.02,
+      "learning_rate": 1.92e-05,
+      "loss": 4.4044,
       "step": 35
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.3543,
+      "epoch": 0.02,
+      "learning_rate": 1.98e-05,
+      "loss": 4.6801,
       "step": 36
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 7.2e-06,
-      "loss": 4.4801,
+      "epoch": 0.02,
+      "learning_rate": 2.04e-05,
+      "loss": 4.3039,
       "step": 37
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.4e-06,
-      "loss": 4.4362,
+      "epoch": 0.02,
+      "learning_rate": 2.1e-05,
+      "loss": 4.4422,
       "step": 38
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.6e-06,
-      "loss": 4.4816,
+      "epoch": 0.02,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 4.5188,
       "step": 39
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 7.8e-06,
-      "loss": 4.347,
+      "epoch": 0.02,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 4.5726,
       "step": 40
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.4076,
+      "epoch": 0.02,
+      "learning_rate": 2.28e-05,
+      "loss": 4.3135,
       "step": 41
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.4635,
+      "epoch": 0.02,
+      "learning_rate": 2.34e-05,
+      "loss": 4.1679,
       "step": 42
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.2502,
+      "epoch": 0.02,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 4.4289,
       "step": 43
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.5461,
+      "epoch": 0.02,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.2429,
       "step": 44
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.8e-06,
-      "loss": 4.4963,
+      "epoch": 0.03,
+      "learning_rate": 2.52e-05,
+      "loss": 4.5192,
       "step": 45
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 9e-06,
-      "loss": 4.5238,
+      "epoch": 0.03,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 4.5163,
       "step": 46
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.2e-06,
-      "loss": 4.5026,
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 4.5619,
       "step": 47
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.4e-06,
-      "loss": 4.5945,
+      "epoch": 0.03,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 4.8743,
       "step": 48
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.7031,
+      "epoch": 0.03,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 4.8881,
       "step": 49
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.415,
+      "epoch": 0.03,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.8063,
       "step": 50
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1e-05,
-      "loss": 4.3481,
+      "epoch": 0.03,
+      "learning_rate": 2.88e-05,
+      "loss": 4.2264,
       "step": 51
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.02e-05,
-      "loss": 4.2441,
+      "epoch": 0.03,
+      "learning_rate": 2.94e-05,
+      "loss": 4.4088,
       "step": 52
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.04e-05,
-      "loss": 4.2067,
+      "epoch": 0.03,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 4.3993,
       "step": 53
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.06e-05,
-      "loss": 4.206,
+      "epoch": 0.03,
+      "learning_rate": 3.06e-05,
+      "loss": 4.2467,
       "step": 54
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.08e-05,
-      "loss": 4.1688,
+      "epoch": 0.03,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 4.3273,
       "step": 55
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.1921,
+      "epoch": 0.03,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 4.419,
       "step": 56
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.3179,
+      "epoch": 0.03,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 4.3186,
       "step": 57
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.3096,
+      "epoch": 0.03,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 4.2985,
       "step": 58
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.16e-05,
-      "loss": 4.2819,
+      "epoch": 0.03,
+      "learning_rate": 3.36e-05,
+      "loss": 4.3233,
       "step": 59
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.18e-05,
-      "loss": 4.2195,
+      "epoch": 0.03,
+      "learning_rate": 3.42e-05,
+      "loss": 4.2305,
       "step": 60
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.2e-05,
-      "loss": 4.2763,
+      "epoch": 0.03,
+      "learning_rate": 3.48e-05,
+      "loss": 4.3408,
       "step": 61
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.22e-05,
-      "loss": 4.1643,
+      "epoch": 0.03,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 4.2246,
       "step": 62
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.24e-05,
-      "loss": 4.2968,
+      "epoch": 0.04,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 4.601,
       "step": 63
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.2292,
+      "epoch": 0.04,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 4.0831,
       "step": 64
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.2935,
+      "epoch": 0.04,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 4.2891,
       "step": 65
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.3136,
+      "epoch": 0.04,
+      "learning_rate": 3.78e-05,
+      "loss": 4.4117,
       "step": 66
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.32e-05,
-      "loss": 4.2581,
+      "epoch": 0.04,
+      "learning_rate": 3.84e-05,
+      "loss": 4.3885,
       "step": 67
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.2987,
+      "epoch": 0.04,
+      "learning_rate": 3.9e-05,
+      "loss": 4.4074,
       "step": 68
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.2362,
+      "epoch": 0.04,
+      "learning_rate": 3.96e-05,
+      "loss": 4.3227,
       "step": 69
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.1762,
+      "epoch": 0.04,
+      "learning_rate": 4.02e-05,
+      "loss": 4.2538,
       "step": 70
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.1934,
+      "epoch": 0.04,
+      "learning_rate": 4.08e-05,
+      "loss": 4.3241,
       "step": 71
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.42e-05,
-      "loss": 4.309,
+      "epoch": 0.04,
+      "learning_rate": 4.14e-05,
+      "loss": 4.0814,
       "step": 72
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.44e-05,
-      "loss": 4.2195,
+      "epoch": 0.04,
+      "learning_rate": 4.2e-05,
+      "loss": 4.3634,
       "step": 73
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.2148,
+      "epoch": 0.04,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 4.173,
       "step": 74
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.48e-05,
-      "loss": 4.3228,
+      "epoch": 0.04,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 4.6653,
       "step": 75
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.5e-05,
-      "loss": 4.1966,
+      "epoch": 0.04,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 4.4373,
       "step": 76
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.52e-05,
-      "loss": 4.1663,
+      "epoch": 0.04,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.253,
       "step": 77
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.54e-05,
-      "loss": 4.2249,
+      "epoch": 0.04,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.1852,
       "step": 78
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.56e-05,
-      "loss": 4.1044,
+      "epoch": 0.04,
+      "learning_rate": 4.56e-05,
+      "loss": 4.2115,
       "step": 79
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.58e-05,
-      "loss": 4.2918,
+      "epoch": 0.04,
+      "learning_rate": 4.62e-05,
+      "loss": 4.541,
       "step": 80
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.215,
+      "epoch": 0.05,
+      "learning_rate": 4.68e-05,
+      "loss": 4.1658,
       "step": 81
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.62e-05,
-      "loss": 4.2246,
+      "epoch": 0.05,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.3241,
       "step": 82
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.1984,
+      "epoch": 0.05,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 4.1587,
       "step": 83
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.66e-05,
-      "loss": 4.2291,
+      "epoch": 0.05,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 4.3266,
       "step": 84
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.337,
+      "epoch": 0.05,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 4.2094,
       "step": 85
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.2152,
+      "epoch": 0.05,
+      "learning_rate": 4.98e-05,
+      "loss": 4.1966,
       "step": 86
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.2546,
+      "epoch": 0.05,
+      "learning_rate": 5.04e-05,
+      "loss": 4.5955,
       "step": 87
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.74e-05,
-      "loss": 4.2966,
+      "epoch": 0.05,
+      "learning_rate": 5.1e-05,
+      "loss": 4.1396,
       "step": 88
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.76e-05,
-      "loss": 4.3031,
+      "epoch": 0.05,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 4.5032,
       "step": 89
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.78e-05,
-      "loss": 4.3245,
+      "epoch": 0.05,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 4.0923,
       "step": 90
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.8e-05,
-      "loss": 4.3341,
+      "epoch": 0.05,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 4.4655,
       "step": 91
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.4108,
+      "epoch": 0.05,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 4.1286,
       "step": 92
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.84e-05,
-      "loss": 4.3563,
+      "epoch": 0.05,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 3.9898,
       "step": 93
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.86e-05,
-      "loss": 4.294,
+      "epoch": 0.05,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 4.4139,
       "step": 94
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.88e-05,
-      "loss": 4.2507,
+      "epoch": 0.05,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 4.6238,
       "step": 95
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9e-05,
-      "loss": 4.4097,
+      "epoch": 0.05,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 4.8141,
       "step": 96
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.6024,
+      "epoch": 0.05,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.5297,
       "step": 97
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.94e-05,
-      "loss": 4.5406,
+      "epoch": 0.05,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.4874,
       "step": 98
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.5267,
+      "epoch": 0.06,
+      "learning_rate": 5.76e-05,
+      "loss": 4.7581,
       "step": 99
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.5753,
+      "epoch": 0.06,
+      "learning_rate": 5.82e-05,
+      "loss": 4.7329,
       "step": 100
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2e-05,
-      "loss": 4.1584,
+      "epoch": 0.06,
+      "learning_rate": 5.88e-05,
+      "loss": 4.4531,
       "step": 101
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.2193,
+      "epoch": 0.06,
+      "learning_rate": 5.94e-05,
+      "loss": 4.3324,
       "step": 102
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.04e-05,
-      "loss": 4.2235,
+      "epoch": 0.06,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 4.3949,
       "step": 103
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.06e-05,
-      "loss": 4.1147,
+      "epoch": 0.06,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 4.2711,
       "step": 104
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.08e-05,
-      "loss": 4.2271,
+      "epoch": 0.06,
+      "learning_rate": 6.12e-05,
+      "loss": 4.294,
       "step": 105
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.1e-05,
-      "loss": 4.14,
+      "epoch": 0.06,
+      "learning_rate": 6.18e-05,
+      "loss": 4.2988,
       "step": 106
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.12e-05,
-      "loss": 4.1039,
+      "epoch": 0.06,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 4.3157,
       "step": 107
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.2112,
+      "epoch": 0.06,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 4.4341,
       "step": 108
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.16e-05,
-      "loss": 4.2219,
+      "epoch": 0.06,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 4.3025,
       "step": 109
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.18e-05,
-      "loss": 4.2005,
+      "epoch": 0.06,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 4.3431,
       "step": 110
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.2685,
+      "epoch": 0.06,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 4.2753,
       "step": 111
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.22e-05,
-      "loss": 4.1964,
+      "epoch": 0.06,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 4.1674,
       "step": 112
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.1327,
+      "epoch": 0.06,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 4.3238,
       "step": 113
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 2.26e-05,
-      "loss": 4.1996,
+      "epoch": 0.06,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 4.0734,
       "step": 114
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.1549,
+      "epoch": 0.06,
+      "learning_rate": 6.72e-05,
+      "loss": 4.0799,
       "step": 115
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.1736,
+      "epoch": 0.07,
+      "learning_rate": 6.78e-05,
+      "loss": 4.1991,
       "step": 116
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.32e-05,
-      "loss": 4.1134,
+      "epoch": 0.07,
+      "learning_rate": 6.84e-05,
+      "loss": 4.4108,
       "step": 117
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.1522,
+      "epoch": 0.07,
+      "learning_rate": 6.9e-05,
+      "loss": 4.2132,
       "step": 118
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.36e-05,
-      "loss": 4.2019,
+      "epoch": 0.07,
+      "learning_rate": 6.96e-05,
+      "loss": 4.3399,
       "step": 119
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.38e-05,
-      "loss": 4.208,
+      "epoch": 0.07,
+      "learning_rate": 7.02e-05,
+      "loss": 4.4068,
       "step": 120
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4e-05,
-      "loss": 4.2208,
+      "epoch": 0.07,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 4.4561,
       "step": 121
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 4.2966,
+      "epoch": 0.07,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 4.3531,
       "step": 122
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 2.44e-05,
-      "loss": 4.1012,
+      "epoch": 0.07,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 4.3516,
       "step": 123
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.46e-05,
-      "loss": 4.2317,
+      "epoch": 0.07,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 4.3174,
       "step": 124
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.48e-05,
-      "loss": 4.2931,
+      "epoch": 0.07,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 4.0512,
       "step": 125
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 2.5e-05,
-      "loss": 4.1984,
+      "epoch": 0.07,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 4.5767,
       "step": 126
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.2407,
+      "epoch": 0.07,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 4.345,
       "step": 127
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.54e-05,
-      "loss": 4.111,
+      "epoch": 0.07,
+      "learning_rate": 7.5e-05,
+      "loss": 4.3119,
       "step": 128
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2664,
+      "epoch": 0.07,
+      "learning_rate": 7.56e-05,
+      "loss": 4.2563,
       "step": 129
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.58e-05,
-      "loss": 4.3569,
+      "epoch": 0.07,
+      "learning_rate": 7.62e-05,
+      "loss": 4.4141,
       "step": 130
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.2576,
+      "epoch": 0.07,
+      "learning_rate": 7.68e-05,
+      "loss": 4.6154,
       "step": 131
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.1036,
+      "epoch": 0.07,
+      "learning_rate": 7.74e-05,
+      "loss": 4.4004,
       "step": 132
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.64e-05,
-      "loss": 4.2504,
+      "epoch": 0.07,
+      "learning_rate": 7.8e-05,
+      "loss": 4.3886,
       "step": 133
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.1388,
+      "epoch": 0.08,
+      "learning_rate": 7.86e-05,
+      "loss": 4.3843,
       "step": 134
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.3289,
+      "epoch": 0.08,
+      "learning_rate": 7.92e-05,
+      "loss": 4.2755,
       "step": 135
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.2359,
+      "epoch": 0.08,
+      "learning_rate": 7.98e-05,
+      "loss": 4.5099,
       "step": 136
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.1676,
+      "epoch": 0.08,
+      "learning_rate": 8.04e-05,
+      "loss": 4.2549,
       "step": 137
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.1007,
+      "epoch": 0.08,
+      "learning_rate": 8.1e-05,
+      "loss": 4.4867,
       "step": 138
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.1463,
+      "epoch": 0.08,
+      "learning_rate": 8.16e-05,
+      "loss": 4.6586,
       "step": 139
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.1603,
+      "epoch": 0.08,
+      "learning_rate": 8.22e-05,
+      "loss": 4.3022,
       "step": 140
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.3578,
+      "epoch": 0.08,
+      "learning_rate": 8.28e-05,
+      "loss": 4.5577,
       "step": 141
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.1484,
+      "epoch": 0.08,
+      "learning_rate": 8.34e-05,
+      "loss": 4.6748,
       "step": 142
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.84e-05,
-      "loss": 4.2449,
+      "epoch": 0.08,
+      "learning_rate": 8.4e-05,
+      "loss": 4.3893,
       "step": 143
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.86e-05,
-      "loss": 4.2826,
+      "epoch": 0.08,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 4.5241,
       "step": 144
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.88e-05,
-      "loss": 4.3742,
+      "epoch": 0.08,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 5.097,
       "step": 145
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.9e-05,
-      "loss": 4.1835,
+      "epoch": 0.08,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 4.6905,
       "step": 146
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.3268,
+      "epoch": 0.08,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 4.7197,
       "step": 147
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.94e-05,
-      "loss": 4.454,
+      "epoch": 0.08,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 4.7719,
       "step": 148
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.96e-05,
-      "loss": 4.4426,
+      "epoch": 0.08,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 4.8233,
       "step": 149
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.98e-05,
-      "loss": 4.5606,
+      "epoch": 0.08,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 4.228,
       "step": 150
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3e-05,
-      "loss": 4.2171,
+      "epoch": 0.08,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 4.1789,
       "step": 151
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3.02e-05,
-      "loss": 4.1916,
+      "epoch": 0.09,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 4.607,
       "step": 152
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 3.04e-05,
-      "loss": 4.2203,
+      "epoch": 0.09,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.152,
       "step": 153
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.06e-05,
-      "loss": 4.1566,
+      "epoch": 0.09,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 4.6215,
       "step": 154
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.08e-05,
-      "loss": 4.1767,
+      "epoch": 0.09,
+      "learning_rate": 9.12e-05,
+      "loss": 4.4727,
       "step": 155
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 3.1e-05,
-      "loss": 4.1967,
+      "epoch": 0.09,
+      "learning_rate": 9.18e-05,
+      "loss": 4.7536,
       "step": 156
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.12e-05,
-      "loss": 4.1163,
+      "epoch": 0.09,
+      "learning_rate": 9.24e-05,
+      "loss": 4.5145,
       "step": 157
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.1687,
+      "epoch": 0.09,
+      "learning_rate": 9.3e-05,
+      "loss": 4.4542,
       "step": 158
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 3.16e-05,
-      "loss": 4.1796,
+      "epoch": 0.09,
+      "learning_rate": 9.36e-05,
+      "loss": 4.4696,
       "step": 159
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.18e-05,
-      "loss": 4.1479,
+      "epoch": 0.09,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 4.9688,
       "step": 160
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.1677,
+      "epoch": 0.09,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 4.4859,
       "step": 161
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.1325,
+      "epoch": 0.09,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 4.741,
       "step": 162
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.24e-05,
-      "loss": 4.203,
+      "epoch": 0.09,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 4.6048,
       "step": 163
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.26e-05,
-      "loss": 4.1248,
+      "epoch": 0.09,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 4.5725,
       "step": 164
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.1852,
+      "epoch": 0.09,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 4.4576,
       "step": 165
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.3e-05,
-      "loss": 4.1591,
+      "epoch": 0.09,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 4.5854,
       "step": 166
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.32e-05,
-      "loss": 4.1522,
+      "epoch": 0.09,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 4.331,
       "step": 167
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.1526,
+      "epoch": 0.09,
+      "learning_rate": 9.9e-05,
+      "loss": 4.1595,
       "step": 168
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.1457,
+      "epoch": 0.09,
+      "learning_rate": 9.96e-05,
+      "loss": 4.2601,
       "step": 169
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 3.38e-05,
-      "loss": 4.0866,
+      "epoch": 0.1,
+      "learning_rate": 0.0001002,
+      "loss": 4.3851,
       "step": 170
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.0714,
+      "epoch": 0.1,
+      "learning_rate": 0.0001008,
+      "loss": 4.5368,
       "step": 171
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.1798,
+      "epoch": 0.1,
+      "learning_rate": 0.0001014,
+      "loss": 5.5086,
       "step": 172
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.032,
+      "epoch": 0.1,
+      "learning_rate": 0.000102,
+      "loss": 4.3675,
       "step": 173
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.46e-05,
-      "loss": 4.1041,
+      "epoch": 0.1,
+      "learning_rate": 0.0001026,
+      "loss": 4.5266,
       "step": 174
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.48e-05,
-      "loss": 4.2255,
+      "epoch": 0.1,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 4.7312,
       "step": 175
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 3.5e-05,
-      "loss": 4.1027,
+      "epoch": 0.1,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 4.3899,
       "step": 176
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.52e-05,
-      "loss": 4.2227,
+      "epoch": 0.1,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 4.4549,
       "step": 177
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.54e-05,
-      "loss": 4.071,
+      "epoch": 0.1,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 4.3311,
       "step": 178
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.56e-05,
-      "loss": 4.1006,
+      "epoch": 0.1,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 4.1532,
       "step": 179
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.58e-05,
-      "loss": 4.2902,
+      "epoch": 0.1,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 4.6106,
       "step": 180
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.6e-05,
-      "loss": 4.1315,
+      "epoch": 0.1,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 4.5489,
       "step": 181
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.62e-05,
-      "loss": 4.0543,
+      "epoch": 0.1,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 4.3362,
       "step": 182
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.0681,
+      "epoch": 0.1,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 4.7618,
       "step": 183
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.66e-05,
-      "loss": 4.1713,
+      "epoch": 0.1,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 4.6088,
       "step": 184
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.68e-05,
-      "loss": 4.2008,
+      "epoch": 0.1,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 4.6477,
       "step": 185
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.7e-05,
-      "loss": 4.1546,
+      "epoch": 0.1,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 4.4798,
       "step": 186
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.72e-05,
-      "loss": 4.1709,
+      "epoch": 0.1,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 4.4516,
       "step": 187
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.74e-05,
-      "loss": 4.1473,
+      "epoch": 0.11,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 4.6847,
       "step": 188
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.76e-05,
-      "loss": 4.1097,
+      "epoch": 0.11,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 4.5905,
       "step": 189
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.2103,
+      "epoch": 0.11,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 4.4689,
       "step": 190
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.8e-05,
-      "loss": 4.2618,
+      "epoch": 0.11,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 4.6269,
       "step": 191
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.82e-05,
-      "loss": 4.2977,
+      "epoch": 0.11,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 4.4034,
       "step": 192
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.3615,
+      "epoch": 0.11,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 4.6338,
       "step": 193
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.86e-05,
-      "loss": 4.2861,
+      "epoch": 0.11,
+      "learning_rate": 0.0001146,
+      "loss": 4.4695,
       "step": 194
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.88e-05,
-      "loss": 4.2774,
+      "epoch": 0.11,
+      "learning_rate": 0.0001152,
+      "loss": 4.7839,
       "step": 195
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.231,
+      "epoch": 0.11,
+      "learning_rate": 0.0001158,
+      "loss": 5.239,
       "step": 196
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.0982,
+      "epoch": 0.11,
+      "learning_rate": 0.0001164,
+      "loss": 5.0441,
       "step": 197
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.94e-05,
-      "loss": 4.6038,
+      "epoch": 0.11,
+      "learning_rate": 0.000117,
+      "loss": 4.5253,
       "step": 198
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 4.3139,
+      "epoch": 0.11,
+      "learning_rate": 0.0001176,
+      "loss": 4.8081,
       "step": 199
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.5138,
+      "epoch": 0.11,
+      "learning_rate": 0.0001182,
+      "loss": 4.772,
       "step": 200
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4e-05,
-      "loss": 4.2164,
+      "epoch": 0.11,
+      "learning_rate": 0.0001188,
+      "loss": 4.6347,
       "step": 201
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.02e-05,
-      "loss": 4.1559,
+      "epoch": 0.11,
+      "learning_rate": 0.0001194,
+      "loss": 4.4527,
       "step": 202
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.1452,
+      "epoch": 0.11,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 4.5706,
       "step": 203
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.0741,
+      "epoch": 0.11,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 4.591,
       "step": 204
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.08e-05,
-      "loss": 4.1215,
+      "epoch": 0.11,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 4.4277,
       "step": 205
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.1e-05,
-      "loss": 4.0996,
+      "epoch": 0.12,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 4.8654,
       "step": 206
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.12e-05,
-      "loss": 4.164,
+      "epoch": 0.12,
+      "learning_rate": 0.0001224,
+      "loss": 4.6914,
       "step": 207
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.14e-05,
-      "loss": 4.1406,
+      "epoch": 0.12,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 4.6337,
       "step": 208
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.16e-05,
-      "loss": 4.1043,
+      "epoch": 0.12,
+      "learning_rate": 0.0001236,
+      "loss": 4.6592,
       "step": 209
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.18e-05,
-      "loss": 4.2183,
+      "epoch": 0.12,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 4.6443,
       "step": 210
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.2e-05,
-      "loss": 3.9963,
+      "epoch": 0.12,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 4.5077,
       "step": 211
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.22e-05,
-      "loss": 4.1352,
+      "epoch": 0.12,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 4.8719,
       "step": 212
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.24e-05,
-      "loss": 4.1756,
+      "epoch": 0.12,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 4.6238,
       "step": 213
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.26e-05,
-      "loss": 4.1339,
+      "epoch": 0.12,
+      "learning_rate": 0.0001266,
+      "loss": 4.316,
       "step": 214
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.1696,
+      "epoch": 0.12,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 5.0334,
       "step": 215
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.3e-05,
-      "loss": 4.1555,
+      "epoch": 0.12,
+      "learning_rate": 0.0001278,
+      "loss": 4.1594,
       "step": 216
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.32e-05,
-      "loss": 4.104,
+      "epoch": 0.12,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 4.616,
       "step": 217
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.0657,
+      "epoch": 0.12,
+      "learning_rate": 0.000129,
+      "loss": 4.3934,
       "step": 218
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.36e-05,
-      "loss": 4.0738,
+      "epoch": 0.12,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 4.7009,
       "step": 219
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.38e-05,
-      "loss": 4.1514,
+      "epoch": 0.12,
+      "learning_rate": 0.0001302,
+      "loss": 5.0774,
       "step": 220
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.0798,
+      "epoch": 0.12,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 4.5736,
       "step": 221
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.0354,
+      "epoch": 0.12,
+      "learning_rate": 0.0001314,
+      "loss": 4.8936,
       "step": 222
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.44e-05,
-      "loss": 4.0835,
+      "epoch": 0.12,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 4.3545,
       "step": 223
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.46e-05,
-      "loss": 4.1241,
+      "epoch": 0.13,
+      "learning_rate": 0.0001326,
+      "loss": 4.4807,
       "step": 224
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.1139,
+      "epoch": 0.13,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 4.67,
       "step": 225
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.5e-05,
-      "loss": 4.1373,
+      "epoch": 0.13,
+      "learning_rate": 0.0001338,
+      "loss": 4.7084,
       "step": 226
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.52e-05,
-      "loss": 4.1314,
+      "epoch": 0.13,
+      "learning_rate": 0.0001344,
+      "loss": 4.6215,
       "step": 227
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.0515,
+      "epoch": 0.13,
+      "learning_rate": 0.000135,
+      "loss": 4.3859,
       "step": 228
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.0908,
+      "epoch": 0.13,
+      "learning_rate": 0.0001356,
+      "loss": 4.4294,
       "step": 229
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.58e-05,
-      "loss": 4.1007,
+      "epoch": 0.13,
+      "learning_rate": 0.0001362,
+      "loss": 4.6407,
       "step": 230
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 4.2356,
+      "epoch": 0.13,
+      "learning_rate": 0.0001368,
+      "loss": 4.4371,
       "step": 231
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.0633,
+      "epoch": 0.13,
+      "learning_rate": 0.0001374,
+      "loss": 4.647,
       "step": 232
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.64e-05,
-      "loss": 4.1935,
+      "epoch": 0.13,
+      "learning_rate": 0.000138,
+      "loss": 4.6561,
       "step": 233
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 4.1731,
+      "epoch": 0.13,
+      "learning_rate": 0.0001386,
+      "loss": 4.6821,
       "step": 234
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.2143,
+      "epoch": 0.13,
+      "learning_rate": 0.0001392,
+      "loss": 4.4471,
       "step": 235
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.7e-05,
-      "loss": 4.0391,
+      "epoch": 0.13,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 4.9001,
       "step": 236
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.72e-05,
-      "loss": 4.1392,
+      "epoch": 0.13,
+      "learning_rate": 0.0001404,
+      "loss": 4.5673,
       "step": 237
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.74e-05,
-      "loss": 4.2716,
+      "epoch": 0.13,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 4.8226,
       "step": 238
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.76e-05,
-      "loss": 4.0908,
+      "epoch": 0.13,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 4.6531,
       "step": 239
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.78e-05,
-      "loss": 4.1365,
+      "epoch": 0.13,
+      "learning_rate": 0.0001422,
+      "loss": 4.6618,
       "step": 240
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.8e-05,
-      "loss": 4.2666,
+      "epoch": 0.14,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 4.6119,
       "step": 241
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.82e-05,
-      "loss": 4.1715,
+      "epoch": 0.14,
+      "learning_rate": 0.0001434,
+      "loss": 4.4451,
       "step": 242
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.1233,
+      "epoch": 0.14,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 6.0049,
       "step": 243
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.86e-05,
-      "loss": 4.1594,
+      "epoch": 0.14,
+      "learning_rate": 0.0001446,
+      "loss": 5.359,
       "step": 244
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.88e-05,
-      "loss": 4.1508,
+      "epoch": 0.14,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 4.7498,
       "step": 245
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.9e-05,
-      "loss": 4.3186,
+      "epoch": 0.14,
+      "learning_rate": 0.0001458,
+      "loss": 4.7383,
       "step": 246
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.92e-05,
-      "loss": 4.2761,
+      "epoch": 0.14,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 5.379,
       "step": 247
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.94e-05,
-      "loss": 4.2375,
+      "epoch": 0.14,
+      "learning_rate": 0.000147,
+      "loss": 5.0694,
       "step": 248
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.96e-05,
-      "loss": 4.3505,
+      "epoch": 0.14,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 4.5803,
       "step": 249
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.3046,
+      "epoch": 0.14,
+      "learning_rate": 0.0001482,
+      "loss": 5.1926,
       "step": 250
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 5e-05,
-      "loss": 4.23,
+      "epoch": 0.14,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 4.5571,
       "step": 251
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.02e-05,
-      "loss": 4.1051,
+      "epoch": 0.14,
+      "learning_rate": 0.0001494,
+      "loss": 5.7546,
       "step": 252
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.0697,
+      "epoch": 0.14,
+      "learning_rate": 0.00015,
+      "loss": 4.9364,
       "step": 253
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.0968,
+      "epoch": 0.14,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 4.7004,
       "step": 254
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.08e-05,
-      "loss": 4.1046,
+      "epoch": 0.14,
+      "learning_rate": 0.0001512,
+      "loss": 5.0182,
       "step": 255
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.0778,
+      "epoch": 0.14,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 4.7166,
       "step": 256
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.2512,
+      "epoch": 0.14,
+      "learning_rate": 0.0001524,
+      "loss": 4.5697,
       "step": 257
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.14e-05,
-      "loss": 4.0971,
+      "epoch": 0.14,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 5.2672,
       "step": 258
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.16e-05,
-      "loss": 4.1317,
+      "epoch": 0.15,
+      "learning_rate": 0.0001536,
+      "loss": 4.8173,
       "step": 259
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.0438,
+      "epoch": 0.15,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 4.329,
       "step": 260
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.1585,
+      "epoch": 0.15,
+      "learning_rate": 0.0001548,
+      "loss": 4.9393,
       "step": 261
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.22e-05,
-      "loss": 4.1922,
+      "epoch": 0.15,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 4.7213,
       "step": 262
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.0355,
+      "epoch": 0.15,
+      "learning_rate": 0.000156,
+      "loss": 4.9763,
       "step": 263
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.0942,
+      "epoch": 0.15,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 4.6642,
       "step": 264
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.28e-05,
-      "loss": 4.0823,
+      "epoch": 0.15,
+      "learning_rate": 0.0001572,
+      "loss": 4.8533,
       "step": 265
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 4.2057,
+      "epoch": 0.15,
+      "learning_rate": 0.0001578,
+      "loss": 4.3969,
       "step": 266
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.1053,
+      "epoch": 0.15,
+      "learning_rate": 0.0001584,
+      "loss": 4.7146,
       "step": 267
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.0612,
+      "epoch": 0.15,
+      "learning_rate": 0.000159,
+      "loss": 4.6821,
       "step": 268
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 4.0192,
+      "epoch": 0.15,
+      "learning_rate": 0.0001596,
+      "loss": 4.788,
       "step": 269
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 4.1746,
+      "epoch": 0.15,
+      "learning_rate": 0.0001602,
+      "loss": 4.2517,
       "step": 270
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.0217,
+      "epoch": 0.15,
+      "learning_rate": 0.0001608,
+      "loss": 4.5119,
       "step": 271
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 4.0175,
+      "epoch": 0.15,
+      "learning_rate": 0.0001614,
+      "loss": 4.8131,
       "step": 272
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 4.0711,
+      "epoch": 0.15,
+      "learning_rate": 0.000162,
+      "loss": 4.6499,
       "step": 273
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 4.0552,
+      "epoch": 0.15,
+      "learning_rate": 0.0001626,
+      "loss": 5.3647,
       "step": 274
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.1794,
+      "epoch": 0.15,
+      "learning_rate": 0.0001632,
+      "loss": 4.6299,
       "step": 275
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 4.1307,
+      "epoch": 0.15,
+      "learning_rate": 0.0001638,
+      "loss": 5.581,
       "step": 276
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 4.1143,
+      "epoch": 0.16,
+      "learning_rate": 0.0001644,
+      "loss": 4.5606,
       "step": 277
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.0236,
+      "epoch": 0.16,
+      "learning_rate": 0.000165,
+      "loss": 4.7095,
       "step": 278
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 4.0805,
+      "epoch": 0.16,
+      "learning_rate": 0.0001656,
+      "loss": 4.3864,
       "step": 279
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 4.0284,
+      "epoch": 0.16,
+      "learning_rate": 0.0001662,
+      "loss": 4.5545,
       "step": 280
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 4.1211,
+      "epoch": 0.16,
+      "learning_rate": 0.0001668,
+      "loss": 5.0875,
       "step": 281
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 4.0672,
+      "epoch": 0.16,
+      "learning_rate": 0.0001674,
+      "loss": 5.2147,
       "step": 282
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.2571,
+      "epoch": 0.16,
+      "learning_rate": 0.000168,
+      "loss": 4.8058,
       "step": 283
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.66e-05,
-      "loss": 4.0389,
+      "epoch": 0.16,
+      "learning_rate": 0.0001686,
+      "loss": 4.4864,
       "step": 284
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.68e-05,
-      "loss": 4.1248,
+      "epoch": 0.16,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 4.9827,
       "step": 285
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.1542,
+      "epoch": 0.16,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 4.7352,
       "step": 286
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.72e-05,
-      "loss": 4.0784,
+      "epoch": 0.16,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 4.7162,
       "step": 287
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.74e-05,
-      "loss": 4.0661,
+      "epoch": 0.16,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 4.8099,
       "step": 288
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.76e-05,
-      "loss": 4.0908,
+      "epoch": 0.16,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 4.5027,
       "step": 289
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.179,
+      "epoch": 0.16,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 5.066,
       "step": 290
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.8e-05,
-      "loss": 4.17,
+      "epoch": 0.16,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 4.5522,
       "step": 291
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 5.82e-05,
-      "loss": 4.2625,
+      "epoch": 0.16,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 5.6078,
       "step": 292
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 4.1965,
+      "epoch": 0.16,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 4.7149,
       "step": 293
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.86e-05,
-      "loss": 4.5008,
+      "epoch": 0.16,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 4.8358,
       "step": 294
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.88e-05,
-      "loss": 4.1809,
+      "epoch": 0.17,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 4.6981,
       "step": 295
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 5.9e-05,
-      "loss": 4.5448,
+      "epoch": 0.17,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 7.6051,
       "step": 296
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 5.92e-05,
-      "loss": 4.4749,
+      "epoch": 0.17,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 5.2167,
       "step": 297
     },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 5.1343,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 4.683,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 5.2883,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 5.2891,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 5.4221,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.9102,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 4.5784,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 5.5962,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 4.7911,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001824,
+      "loss": 4.673,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 6.4645,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001836,
+      "loss": 4.8379,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 5.2851,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001848,
+      "loss": 4.7361,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 5.0055,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000186,
+      "loss": 4.7797,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 4.9896,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001872,
+      "loss": 4.7066,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 4.5315,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 4.5292,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 4.5059,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 5.0415,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001902,
+      "loss": 5.0882,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 5.3482,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001914,
+      "loss": 5.0749,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 4.9192,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001926,
+      "loss": 4.5151,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 4.8147,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938,
+      "loss": 4.8692,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 5.3922,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000195,
+      "loss": 5.1391,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 4.8038,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001962,
+      "loss": 5.0244,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 4.8476,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001974,
+      "loss": 5.0352,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000198,
+      "loss": 5.9123,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001986,
+      "loss": 5.1061,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001992,
+      "loss": 5.058,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001998,
+      "loss": 5.1996,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002004,
+      "loss": 4.6912,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000201,
+      "loss": 4.772,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002016,
+      "loss": 4.6496,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002022,
+      "loss": 4.9221,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002028,
+      "loss": 4.6008,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 4.7957,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000204,
+      "loss": 4.9876,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 4.3914,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002052,
+      "loss": 5.8961,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002058,
+      "loss": 4.5013,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 5.2834,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 5.1501,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 5.3556,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 5.5409,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 4.9931,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 5.1523,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 4.9868,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 4.6934,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 4.5688,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 4.8821,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 5.2561,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 4.9584,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 5.2996,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 4.6552,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 4.9308,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 4.4198,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 5.0918,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 5.079,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 5.2672,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 5.1869,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 4.7009,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 4.6603,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 5.1922,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 4.9497,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 4.5666,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002214,
+      "loss": 5.4305,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 4.7896,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002226,
+      "loss": 5.4258,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 4.4568,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002238,
+      "loss": 4.7348,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 4.6844,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000225,
+      "loss": 4.5267,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 4.7165,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 5.2593,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 4.3269,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 4.7356,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 4.5989,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 5.4692,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002292,
+      "loss": 5.0367,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 4.7516,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002304,
+      "loss": 5.9294,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 4.5957,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002316,
+      "loss": 4.7088,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 5.8763,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002328,
+      "loss": 4.4181,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 5.2992,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000234,
+      "loss": 5.0853,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 4.912,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002352,
+      "loss": 5.1396,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 5.4524,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002364,
+      "loss": 5.438,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000237,
+      "loss": 7.1359,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002376,
+      "loss": 10.3571,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002382,
+      "loss": 5.2293,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002388,
+      "loss": 4.9652,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002394,
+      "loss": 4.883,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.8143,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002406,
+      "loss": 5.0558,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 4.6178,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002418,
+      "loss": 5.0112,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 4.9684,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000243,
+      "loss": 5.8897,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 5.3244,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 5.1039,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002448,
+      "loss": 5.0326,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 5.2395,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 5.1967,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002466,
+      "loss": 5.2712,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002472,
+      "loss": 5.3194,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 4.7941,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 5.6131,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000249,
+      "loss": 4.8572,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 4.7795,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 4.9344,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 4.6473,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002514,
+      "loss": 4.8754,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 4.9176,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 5.5264,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002532,
+      "loss": 4.7957,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002538,
+      "loss": 4.7374,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 5.3779,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 6.1729,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002556,
+      "loss": 4.7028,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002562,
+      "loss": 4.7961,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 5.1186,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 4.7296,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000258,
+      "loss": 4.7054,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002586,
+      "loss": 4.7615,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 4.8103,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 4.4192,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604,
+      "loss": 4.9114,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000261,
+      "loss": 4.9089,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 4.5249,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002622,
+      "loss": 4.8355,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002628,
+      "loss": 5.442,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 6.2262,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 4.7432,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002646,
+      "loss": 5.8577,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002652,
+      "loss": 4.7698,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 7.5969,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 5.9009,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000267,
+      "loss": 5.2798,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002676,
+      "loss": 6.5801,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 4.91,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002688,
+      "loss": 8.2594,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002694,
+      "loss": 5.2204,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027,
+      "loss": 6.2953,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 5.6008,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002712,
+      "loss": 4.9293,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002718,
+      "loss": 4.8321,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002724,
+      "loss": 4.6382,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 5.6477,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002736,
+      "loss": 4.5272,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002742,
+      "loss": 5.0871,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002748,
+      "loss": 5.669,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 4.8767,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000276,
+      "loss": 5.1443,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002766,
+      "loss": 4.8547,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002772,
+      "loss": 4.6289,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002778,
+      "loss": 5.0307,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002784,
+      "loss": 4.8807,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000279,
+      "loss": 5.6747,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 4.6908,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002802,
+      "loss": 4.9543,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002808,
+      "loss": 5.3293,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 4.7712,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 4.7345,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002826,
+      "loss": 5.2528,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 5.1536,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 5.2428,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002844,
+      "loss": 5.6762,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000285,
+      "loss": 4.6718,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 5.4173,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 5.4427,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002868,
+      "loss": 4.7836,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 6.2387,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 5.3981,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 4.9966,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002892,
+      "loss": 5.5966,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 5.8143,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 5.4283,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 5.1403,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002916,
+      "loss": 5.253,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 5.1972,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 5.3527,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002934,
+      "loss": 5.4327,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000294,
+      "loss": 4.9888,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 6.1206,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 5.533,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002958,
+      "loss": 4.9783,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002964,
+      "loss": 4.6209,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 5.9688,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 5.1729,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002982,
+      "loss": 4.5471,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 5.581608295440674,
+      "eval_runtime": 900.7705,
+      "eval_samples_per_second": 2.933,
+      "eval_steps_per_second": 0.367,
+      "eval_wer": 1.9376041253470846,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002988,
+      "loss": 5.933,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 5.1431,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003,
+      "loss": 5.2846,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002997663551401869,
+      "loss": 4.9981,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029953271028037383,
+      "loss": 5.6008,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029929906542056073,
+      "loss": 5.0449,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029906542056074763,
+      "loss": 4.8612,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002988317757009346,
+      "loss": 4.804,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002985981308411215,
+      "loss": 5.7526,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983644859813084,
+      "loss": 5.2698,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002981308411214953,
+      "loss": 4.894,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029789719626168224,
+      "loss": 5.1537,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029766355140186914,
+      "loss": 5.4263,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029742990654205604,
+      "loss": 4.4888,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029719626168224294,
+      "loss": 4.6535,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002969626168224299,
+      "loss": 5.9521,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002967289719626168,
+      "loss": 4.8879,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002964953271028037,
+      "loss": 5.3439,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029626168224299065,
+      "loss": 4.9201,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029602803738317755,
+      "loss": 4.9309,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002957943925233645,
+      "loss": 4.6364,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029556074766355135,
+      "loss": 4.8853,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002953271028037383,
+      "loss": 5.3484,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002950934579439252,
+      "loss": 4.9527,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029485981308411215,
+      "loss": 6.347,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029462616822429905,
+      "loss": 5.2694,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029439252336448596,
+      "loss": 5.2538,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029415887850467286,
+      "loss": 4.54,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002939252336448598,
+      "loss": 5.0422,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002936915887850467,
+      "loss": 6.0477,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002934579439252336,
+      "loss": 4.5828,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029322429906542056,
+      "loss": 4.5899,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029299065420560746,
+      "loss": 5.068,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029275700934579436,
+      "loss": 4.5531,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029252336448598126,
+      "loss": 5.2529,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002922897196261682,
+      "loss": 4.7926,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002920560747663551,
+      "loss": 4.8982,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000291822429906542,
+      "loss": 5.4155,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002915887850467289,
+      "loss": 5.0782,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029135514018691587,
+      "loss": 5.2239,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002911214953271028,
+      "loss": 5.1793,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002908878504672897,
+      "loss": 4.9732,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029065420560747663,
+      "loss": 5.5249,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029042056074766353,
+      "loss": 4.8778,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002901869158878505,
+      "loss": 5.0025,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028995327102803733,
+      "loss": 6.4381,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002897196261682243,
+      "loss": 5.6839,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002894859813084112,
+      "loss": 5.835,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028925233644859814,
+      "loss": 5.7503,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028901869158878504,
+      "loss": 4.9375,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028878504672897194,
+      "loss": 5.2374,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028855140186915884,
+      "loss": 5.7827,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002883177570093458,
+      "loss": 5.9669,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002880841121495327,
+      "loss": 4.8548,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002878504672897196,
+      "loss": 4.9032,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028761682242990655,
+      "loss": 5.6123,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028738317757009345,
+      "loss": 5.4261,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028714953271028035,
+      "loss": 5.0758,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00028691588785046725,
+      "loss": 4.534,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002866822429906542,
+      "loss": 4.7744,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002864485981308411,
+      "loss": 5.2177,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000286214953271028,
+      "loss": 6.0423,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002859813084112149,
+      "loss": 4.5625,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028574766355140185,
+      "loss": 5.242,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028551401869158875,
+      "loss": 5.9316,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028528037383177565,
+      "loss": 4.6705,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002850467289719626,
+      "loss": 4.8982,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002848130841121495,
+      "loss": 5.2801,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028457943925233646,
+      "loss": 4.8128,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002843457943925233,
+      "loss": 5.0719,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028411214953271026,
+      "loss": 5.2668,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028387850467289716,
+      "loss": 5.0351,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002836448598130841,
+      "loss": 5.8277,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000283411214953271,
+      "loss": 4.7576,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002831775700934579,
+      "loss": 4.9129,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002829439252336448,
+      "loss": 5.08,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028271028037383177,
+      "loss": 4.928,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028247663551401867,
+      "loss": 4.8651,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028224299065420557,
+      "loss": 4.9735,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002820093457943925,
+      "loss": 5.5964,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002817757009345794,
+      "loss": 4.6658,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002815420560747663,
+      "loss": 5.1068,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002813084112149532,
+      "loss": 4.9917,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002810747663551402,
+      "loss": 5.1377,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002808411214953271,
+      "loss": 6.1242,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000280607476635514,
+      "loss": 4.6956,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002803738317757009,
+      "loss": 5.6856,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00028014018691588784,
+      "loss": 4.8435,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027990654205607474,
+      "loss": 5.375,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027967289719626164,
+      "loss": 5.1088,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002794392523364486,
+      "loss": 5.0008,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002792056074766355,
+      "loss": 5.0166,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027897196261682244,
+      "loss": 5.1512,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002787383177570093,
+      "loss": 4.9898,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027850467289719624,
+      "loss": 4.821,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027827102803738314,
+      "loss": 4.9955,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002780373831775701,
+      "loss": 7.4815,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000277803738317757,
+      "loss": 6.8128,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002775700934579439,
+      "loss": 5.2313,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002773364485981308,
+      "loss": 5.3007,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027710280373831775,
+      "loss": 4.9507,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027686915887850465,
+      "loss": 5.0799,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027663551401869155,
+      "loss": 4.7422,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002764018691588785,
+      "loss": 4.7912,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002761682242990654,
+      "loss": 4.9524,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002759345794392523,
+      "loss": 4.6615,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002757009345794392,
+      "loss": 5.3252,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027546728971962616,
+      "loss": 5.1719,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027523364485981306,
+      "loss": 6.0747,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027499999999999996,
+      "loss": 4.9708,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027476635514018686,
+      "loss": 5.6665,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002745327102803738,
+      "loss": 4.6714,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002742990654205607,
+      "loss": 4.8929,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002740654205607476,
+      "loss": 4.9788,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00027383177570093457,
+      "loss": 5.2246,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00027359813084112147,
+      "loss": 4.7412,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002733644859813084,
+      "loss": 5.2217,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00027313084112149527,
+      "loss": 4.6002,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002728971962616822,
+      "loss": 4.6342,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002726635514018691,
+      "loss": 5.5986,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002724299065420561,
+      "loss": 4.9237,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000272196261682243,
+      "loss": 5.1799,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002719626168224299,
+      "loss": 5.0955,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002717289719626168,
+      "loss": 5.6877,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00027149532710280373,
+      "loss": 5.0829,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00027126168224299063,
+      "loss": 4.7393,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00027102803738317753,
+      "loss": 4.7369,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002707943925233645,
+      "loss": 5.3679,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002705607476635514,
+      "loss": 6.4361,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002703271028037383,
+      "loss": 4.7886,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002700934579439252,
+      "loss": 4.9049,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00026985981308411214,
+      "loss": 5.1811,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00026962616822429904,
+      "loss": 4.7211,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026939252336448594,
+      "loss": 4.9558,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026915887850467284,
+      "loss": 4.738,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002689252336448598,
+      "loss": 4.8307,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002686915887850467,
+      "loss": 4.6711,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002684579439252336,
+      "loss": 4.4691,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026822429906542055,
+      "loss": 4.8473,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026799065420560745,
+      "loss": 6.5801,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002677570093457944,
+      "loss": 6.0626,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026752336448598125,
+      "loss": 5.4207,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002672897196261682,
+      "loss": 4.8942,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002670560747663551,
+      "loss": 4.9021,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026682242990654206,
+      "loss": 4.7175,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026658878504672896,
+      "loss": 4.7728,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026635514018691586,
+      "loss": 5.7365,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026612149532710276,
+      "loss": 6.7046,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002658878504672897,
+      "loss": 5.2938,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002656542056074766,
+      "loss": 5.2759,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002654205607476635,
+      "loss": 6.0766,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026518691588785047,
+      "loss": 5.2083,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026495327102803737,
+      "loss": 5.1199,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026471962616822427,
+      "loss": 4.78,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026448598130841117,
+      "loss": 5.0556,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002642523364485981,
+      "loss": 4.7203,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000264018691588785,
+      "loss": 4.7321,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002637850467289719,
+      "loss": 6.6024,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002635514018691588,
+      "loss": 5.0324,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002633177570093458,
+      "loss": 4.9096,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002630841121495327,
+      "loss": 5.4461,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002628504672897196,
+      "loss": 4.8457,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026261682242990653,
+      "loss": 4.8117,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026238317757009343,
+      "loss": 5.1213,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002621495327102804,
+      "loss": 5.0574,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026191588785046723,
+      "loss": 5.1551,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002616822429906542,
+      "loss": 4.7985,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002614485981308411,
+      "loss": 5.4033,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026121495327102804,
+      "loss": 4.612,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026098130841121494,
+      "loss": 4.9412,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026074766355140184,
+      "loss": 4.8942,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026051401869158874,
+      "loss": 5.4892,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002602803738317757,
+      "loss": 4.7117,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002600467289719626,
+      "loss": 4.6364,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002598130841121495,
+      "loss": 4.8958,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025957943925233645,
+      "loss": 4.7709,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025934579439252335,
+      "loss": 4.5826,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025911214953271025,
+      "loss": 4.6646,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025887850467289715,
+      "loss": 4.8181,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002586448598130841,
+      "loss": 5.2085,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000258411214953271,
+      "loss": 4.9686,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002581775700934579,
+      "loss": 4.7409,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002579439252336448,
+      "loss": 5.1911,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025771028037383176,
+      "loss": 5.0611,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025747663551401866,
+      "loss": 5.1322,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00025724299065420556,
+      "loss": 4.9188,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002570093457943925,
+      "loss": 4.7283,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002567757009345794,
+      "loss": 4.7484,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025654205607476637,
+      "loss": 4.927,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002563084112149532,
+      "loss": 4.749,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025607476635514017,
+      "loss": 4.5379,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025584112149532707,
+      "loss": 5.0992,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000255607476635514,
+      "loss": 5.0194,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002553738317757009,
+      "loss": 4.7863,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002551401869158878,
+      "loss": 4.9485,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002549065420560747,
+      "loss": 4.9278,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002546728971962617,
+      "loss": 4.4361,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002544392523364486,
+      "loss": 4.9515,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002542056074766355,
+      "loss": 5.1618,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025397196261682243,
+      "loss": 4.8154,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025373831775700933,
+      "loss": 4.929,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025350467289719623,
+      "loss": 4.6813,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00025327102803738313,
+      "loss": 4.897,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002530373831775701,
+      "loss": 5.4599,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000252803738317757,
+      "loss": 5.6072,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002525700934579439,
+      "loss": 5.3548,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002523364485981308,
+      "loss": 5.0716,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025210280373831774,
+      "loss": 5.1283,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025186915887850464,
+      "loss": 4.6845,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025163551401869154,
+      "loss": 5.046,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002514018691588785,
+      "loss": 5.2705,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002511682242990654,
+      "loss": 4.8732,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025093457943925235,
+      "loss": 4.9474,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002507009345794392,
+      "loss": 5.0853,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025046728971962615,
+      "loss": 5.0843,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025023364485981305,
+      "loss": 4.5215,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025,
+      "loss": 4.7131,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002497663551401869,
+      "loss": 5.1258,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002495327102803738,
+      "loss": 4.6722,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002492990654205607,
+      "loss": 4.6464,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00024906542056074766,
+      "loss": 4.9733,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00024883177570093456,
+      "loss": 4.9711,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024859813084112146,
+      "loss": 4.7633,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002483644859813084,
+      "loss": 5.4998,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002481308411214953,
+      "loss": 4.9303,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002478971962616822,
+      "loss": 5.1392,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002476635514018691,
+      "loss": 4.3532,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024742990654205607,
+      "loss": 5.0674,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024719626168224297,
+      "loss": 4.6014,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024696261682242987,
+      "loss": 4.7569,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024672897196261677,
+      "loss": 4.6892,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002464953271028037,
+      "loss": 4.6065,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002462616822429906,
+      "loss": 6.3043,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002460280373831775,
+      "loss": 4.7317,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002457943925233645,
+      "loss": 4.8874,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002455607476635514,
+      "loss": 4.7145,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024532710280373833,
+      "loss": 4.6753,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002450934579439252,
+      "loss": 4.6223,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024485981308411213,
+      "loss": 5.2338,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00024462616822429903,
+      "loss": 4.7033,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000244392523364486,
+      "loss": 5.5446,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002441588785046729,
+      "loss": 4.7961,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002439252336448598,
+      "loss": 5.8382,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024369158878504671,
+      "loss": 5.0871,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024345794392523364,
+      "loss": 6.1311,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024322429906542051,
+      "loss": 5.2221,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024299065420560744,
+      "loss": 4.9638,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024275700934579437,
+      "loss": 5.1487,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002425233644859813,
+      "loss": 4.647,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002422897196261682,
+      "loss": 5.0244,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024205607476635512,
+      "loss": 5.1103,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024182242990654205,
+      "loss": 4.8882,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024158878504672895,
+      "loss": 4.8263,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024135514018691585,
+      "loss": 4.8644,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024112149532710278,
+      "loss": 4.9268,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002408878504672897,
+      "loss": 4.9957,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00024065420560747663,
+      "loss": 5.0076,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002404205607476635,
+      "loss": 5.1385,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024018691588785043,
+      "loss": 4.8003,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023995327102803736,
+      "loss": 4.7509,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023971962616822429,
+      "loss": 5.3028,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023948598130841119,
+      "loss": 4.7583,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002392523364485981,
+      "loss": 4.7361,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023901869158878504,
+      "loss": 4.4698,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023878504672897194,
+      "loss": 4.5434,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023855140186915884,
+      "loss": 4.7767,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023831775700934577,
+      "loss": 5.0727,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002380841121495327,
+      "loss": 4.951,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023785046728971962,
+      "loss": 5.0511,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002376168224299065,
+      "loss": 4.7732,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023738317757009342,
+      "loss": 5.1477,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023714953271028035,
+      "loss": 4.6437,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023691588785046728,
+      "loss": 4.7255,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023668224299065418,
+      "loss": 4.9444,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002364485981308411,
+      "loss": 4.581,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023621495327102803,
+      "loss": 4.5809,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023598130841121493,
+      "loss": 4.8372,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023574766355140183,
+      "loss": 5.2178,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023551401869158876,
+      "loss": 4.8039,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023528037383177569,
+      "loss": 4.8313,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002350467289719626,
+      "loss": 5.0778,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023481308411214949,
+      "loss": 4.9378,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002345794392523364,
+      "loss": 4.7003,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023434579439252334,
+      "loss": 4.5985,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023411214953271027,
+      "loss": 4.6101,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023387850467289717,
+      "loss": 5.1548,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002336448598130841,
+      "loss": 5.0424,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023341121495327102,
+      "loss": 5.1229,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023317757009345792,
+      "loss": 5.0449,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023294392523364482,
+      "loss": 5.2281,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023271028037383175,
+      "loss": 4.9368,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023247663551401868,
+      "loss": 4.7728,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002322429906542056,
+      "loss": 5.0216,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023200934579439248,
+      "loss": 4.8232,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002317757009345794,
+      "loss": 5.4885,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023154205607476633,
+      "loss": 5.4549,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023130841121495326,
+      "loss": 5.1114,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023107476635514016,
+      "loss": 5.061,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023084112149532709,
+      "loss": 4.9769,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000230607476635514,
+      "loss": 4.9299,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002303738317757009,
+      "loss": 5.6714,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002301401869158878,
+      "loss": 5.0961,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022990654205607474,
+      "loss": 5.1892,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022967289719626167,
+      "loss": 4.7421,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002294392523364486,
+      "loss": 4.9102,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022920560747663547,
+      "loss": 4.9065,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002289719626168224,
+      "loss": 4.704,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022873831775700932,
+      "loss": 4.584,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022850467289719625,
+      "loss": 4.818,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022827102803738315,
+      "loss": 4.9744,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00022803738317757008,
+      "loss": 5.1315,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000227803738317757,
+      "loss": 5.0164,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002275700934579439,
+      "loss": 4.8607,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002273364485981308,
+      "loss": 4.5535,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022710280373831773,
+      "loss": 4.6626,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022686915887850466,
+      "loss": 4.6536,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022663551401869158,
+      "loss": 5.2567,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022640186915887846,
+      "loss": 4.7812,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022616822429906539,
+      "loss": 4.666,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002259345794392523,
+      "loss": 4.7216,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022570093457943924,
+      "loss": 4.8084,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022546728971962614,
+      "loss": 4.6977,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022523364485981307,
+      "loss": 5.9018,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000225,
+      "loss": 4.8642,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002247663551401869,
+      "loss": 5.131,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022453271028037382,
+      "loss": 4.7648,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022429906542056072,
+      "loss": 5.4213,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022406542056074765,
+      "loss": 4.8285,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022383177570093458,
+      "loss": 5.0544,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002235981308411215,
+      "loss": 4.6938,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022336448598130838,
+      "loss": 5.0009,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002231308411214953,
+      "loss": 5.077,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022289719626168223,
+      "loss": 4.9157,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022266355140186916,
+      "loss": 4.471,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022242990654205606,
+      "loss": 4.7155,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022219626168224296,
+      "loss": 4.7203,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022196261682242988,
+      "loss": 4.9135,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002217289719626168,
+      "loss": 4.9905,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002214953271028037,
+      "loss": 5.0028,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022126168224299064,
+      "loss": 5.0444,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022102803738317757,
+      "loss": 5.6934,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002207943925233645,
+      "loss": 4.6798,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022056074766355137,
+      "loss": 4.5773,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002203271028037383,
+      "loss": 4.8751,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022009345794392522,
+      "loss": 4.4002,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00021985981308411215,
+      "loss": 4.5314,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00021962616822429905,
+      "loss": 4.4511,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021939252336448595,
+      "loss": 5.081,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021915887850467288,
+      "loss": 4.9182,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002189252336448598,
+      "loss": 6.4733,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002186915887850467,
+      "loss": 4.9352,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021845794392523363,
+      "loss": 4.7182,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021822429906542056,
+      "loss": 4.9097,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021799065420560748,
+      "loss": 4.8379,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021775700934579436,
+      "loss": 4.8323,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021752336448598128,
+      "loss": 5.0515,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002172897196261682,
+      "loss": 4.6746,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021705607476635514,
+      "loss": 4.9048,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021682242990654204,
+      "loss": 5.1008,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021658878504672894,
+      "loss": 5.2501,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021635514018691587,
+      "loss": 4.8127,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002161214953271028,
+      "loss": 5.1189,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002158878504672897,
+      "loss": 5.127,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021565420560747662,
+      "loss": 4.8071,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021542056074766355,
+      "loss": 5.7193,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021518691588785047,
+      "loss": 4.803,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021495327102803735,
+      "loss": 4.8942,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021471962616822427,
+      "loss": 5.032,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002144859813084112,
+      "loss": 4.7539,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021425233644859813,
+      "loss": 4.6788,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021401869158878503,
+      "loss": 4.6206,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021378504672897193,
+      "loss": 5.1516,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021355140186915886,
+      "loss": 4.9009,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021331775700934578,
+      "loss": 4.7841,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021308411214953268,
+      "loss": 4.9146,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002128504672897196,
+      "loss": 5.0889,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021261682242990654,
+      "loss": 5.0272,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021238317757009347,
+      "loss": 4.6575,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021214953271028034,
+      "loss": 4.6819,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021191588785046727,
+      "loss": 4.7811,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002116822429906542,
+      "loss": 5.0642,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021144859813084112,
+      "loss": 4.5642,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021121495327102802,
+      "loss": 5.1497,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021098130841121492,
+      "loss": 4.8615,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021074766355140185,
+      "loss": 4.9596,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021051401869158877,
+      "loss": 5.1478,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021028037383177567,
+      "loss": 5.4373,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002100467289719626,
+      "loss": 4.5459,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020981308411214953,
+      "loss": 5.1358,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020957943925233646,
+      "loss": 5.0646,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020934579439252333,
+      "loss": 4.6713,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020911214953271026,
+      "loss": 4.8727,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020887850467289718,
+      "loss": 5.5929,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002086448598130841,
+      "loss": 4.7889,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000208411214953271,
+      "loss": 4.9456,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002081775700934579,
+      "loss": 5.0594,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020794392523364484,
+      "loss": 4.6649,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020771028037383176,
+      "loss": 6.2458,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020747663551401867,
+      "loss": 6.1455,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002072429906542056,
+      "loss": 8.3342,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020700934579439252,
+      "loss": 4.8155,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020677570093457945,
+      "loss": 4.9064,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020654205607476632,
+      "loss": 5.468,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020630841121495325,
+      "loss": 4.964,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020607476635514017,
+      "loss": 5.0216,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002058411214953271,
+      "loss": 5.0149,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000205607476635514,
+      "loss": 4.8836,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002053738317757009,
+      "loss": 4.5883,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020514018691588783,
+      "loss": 4.8891,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020490654205607476,
+      "loss": 4.7208,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020467289719626166,
+      "loss": 5.1443,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020443925233644858,
+      "loss": 4.6899,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002042056074766355,
+      "loss": 4.7846,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020397196261682244,
+      "loss": 4.7023,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002037383177570093,
+      "loss": 4.8883,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020350467289719624,
+      "loss": 5.2402,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020327102803738316,
+      "loss": 4.5024,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002030373831775701,
+      "loss": 5.3522,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000202803738317757,
+      "loss": 4.784,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002025700934579439,
+      "loss": 4.3737,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020233644859813082,
+      "loss": 4.6135,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020210280373831775,
+      "loss": 4.9948,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020186915887850465,
+      "loss": 4.7271,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020163551401869157,
+      "loss": 4.5752,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002014018691588785,
+      "loss": 4.8277,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020116822429906543,
+      "loss": 5.2676,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002009345794392523,
+      "loss": 4.5385,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020070093457943923,
+      "loss": 4.6389,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020046728971962616,
+      "loss": 4.5535,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020023364485981308,
+      "loss": 4.8505,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 4.9484,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019976635514018688,
+      "loss": 4.9759,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001995327102803738,
+      "loss": 4.6066,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019929906542056074,
+      "loss": 4.8184,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019906542056074764,
+      "loss": 5.2863,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019883177570093456,
+      "loss": 4.8743,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001985981308411215,
+      "loss": 4.8111,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019836448598130842,
+      "loss": 4.46,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001981308411214953,
+      "loss": 4.6367,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019789719626168222,
+      "loss": 4.8193,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019766355140186915,
+      "loss": 4.7381,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019742990654205607,
+      "loss": 4.6991,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019719626168224297,
+      "loss": 5.7858,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019696261682242987,
+      "loss": 4.8903,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001967289719626168,
+      "loss": 4.8371,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019649532710280373,
+      "loss": 4.7936,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019626168224299063,
+      "loss": 4.5183,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019602803738317755,
+      "loss": 4.8386,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019579439252336448,
+      "loss": 5.5206,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001955607476635514,
+      "loss": 4.3584,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019532710280373828,
+      "loss": 4.7839,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001950934579439252,
+      "loss": 4.8174,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019485981308411214,
+      "loss": 4.7915,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019462616822429906,
+      "loss": 4.406,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019439252336448596,
+      "loss": 4.7956,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019415887850467286,
+      "loss": 4.4982,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001939252336448598,
+      "loss": 4.892,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019369158878504672,
+      "loss": 4.8088,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019345794392523362,
+      "loss": 5.399,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019322429906542055,
+      "loss": 4.9746,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019299065420560747,
+      "loss": 4.5356,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001927570093457944,
+      "loss": 4.7247,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019252336448598127,
+      "loss": 5.0142,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001922897196261682,
+      "loss": 4.6553,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019205607476635513,
+      "loss": 4.5916,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019182242990654205,
+      "loss": 4.5721,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019158878504672895,
+      "loss": 5.2081,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019135514018691585,
+      "loss": 4.624,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019112149532710278,
+      "loss": 5.0746,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001908878504672897,
+      "loss": 4.4659,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001906542056074766,
+      "loss": 4.8608,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019042056074766354,
+      "loss": 4.8085,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019018691588785046,
+      "loss": 4.7054,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001899532710280374,
+      "loss": 5.526,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018971962616822426,
+      "loss": 4.6951,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001894859813084112,
+      "loss": 5.1382,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018925233644859812,
+      "loss": 5.1781,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018901869158878504,
+      "loss": 4.6907,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018878504672897195,
+      "loss": 4.7877,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018855140186915885,
+      "loss": 4.8763,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018831775700934577,
+      "loss": 4.3011,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001880841121495327,
+      "loss": 4.3919,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001878504672897196,
+      "loss": 4.9187,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018761682242990653,
+      "loss": 4.909,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018738317757009345,
+      "loss": 4.4464,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018714953271028038,
+      "loss": 4.8126,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018691588785046725,
+      "loss": 4.705,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018668224299065418,
+      "loss": 4.6552,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001864485981308411,
+      "loss": 4.9665,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018621495327102804,
+      "loss": 4.5219,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018598130841121494,
+      "loss": 4.9879,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018574766355140184,
+      "loss": 4.9641,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018551401869158876,
+      "loss": 5.1829,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001852803738317757,
+      "loss": 5.1299,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001850467289719626,
+      "loss": 4.7448,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018481308411214952,
+      "loss": 4.4289,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018457943925233644,
+      "loss": 4.7672,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018434579439252337,
+      "loss": 5.3529,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018411214953271024,
+      "loss": 4.8228,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018387850467289717,
+      "loss": 4.6377,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 5.072754859924316,
+      "eval_runtime": 903.6611,
+      "eval_samples_per_second": 2.924,
+      "eval_steps_per_second": 0.366,
+      "eval_wer": 1.9177905593018643,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001836448598130841,
+      "loss": 4.9653,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018341121495327103,
+      "loss": 4.6313,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018317757009345793,
+      "loss": 4.6301,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018294392523364483,
+      "loss": 5.1457,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018271028037383175,
+      "loss": 4.7742,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018247663551401868,
+      "loss": 4.7146,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018224299065420558,
+      "loss": 5.1547,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001820093457943925,
+      "loss": 5.2403,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018177570093457944,
+      "loss": 5.126,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018154205607476636,
+      "loss": 4.5986,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018130841121495324,
+      "loss": 4.8227,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018107476635514016,
+      "loss": 4.6896,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001808411214953271,
+      "loss": 4.7199,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018060747663551402,
+      "loss": 5.0175,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018037383177570092,
+      "loss": 5.6869,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018014018691588782,
+      "loss": 4.8219,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017990654205607474,
+      "loss": 4.7156,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017967289719626167,
+      "loss": 4.7404,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017943925233644857,
+      "loss": 5.3779,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001792056074766355,
+      "loss": 5.596,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017897196261682243,
+      "loss": 5.3479,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017873831775700935,
+      "loss": 4.8651,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017850467289719623,
+      "loss": 4.7978,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017827102803738315,
+      "loss": 4.783,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017803738317757008,
+      "loss": 4.9094,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000177803738317757,
+      "loss": 4.6664,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001775700934579439,
+      "loss": 4.7607,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001773364485981308,
+      "loss": 5.0478,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017710280373831773,
+      "loss": 4.8521,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017686915887850466,
+      "loss": 5.113,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017663551401869156,
+      "loss": 4.6649,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001764018691588785,
+      "loss": 4.8997,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017616822429906542,
+      "loss": 4.8152,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017593457943925234,
+      "loss": 4.3392,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017570093457943922,
+      "loss": 4.5973,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017546728971962614,
+      "loss": 4.8074,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017523364485981307,
+      "loss": 4.4612,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000175,
+      "loss": 4.6995,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001747663551401869,
+      "loss": 4.738,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001745327102803738,
+      "loss": 5.6264,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017429906542056073,
+      "loss": 5.0616,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017406542056074765,
+      "loss": 4.74,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017383177570093455,
+      "loss": 5.2807,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017359813084112148,
+      "loss": 5.0314,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001733644859813084,
+      "loss": 5.445,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017313084112149533,
+      "loss": 4.777,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001728971962616822,
+      "loss": 4.6379,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017266355140186913,
+      "loss": 4.6174,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017242990654205606,
+      "loss": 4.9705,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000172196261682243,
+      "loss": 5.6703,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001719626168224299,
+      "loss": 5.6907,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001717289719626168,
+      "loss": 4.4903,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017149532710280372,
+      "loss": 4.7383,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017126168224299064,
+      "loss": 4.4566,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017102803738317754,
+      "loss": 4.7452,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017079439252336447,
+      "loss": 4.5495,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001705607476635514,
+      "loss": 5.0378,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017032710280373833,
+      "loss": 4.6649,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001700934579439252,
+      "loss": 5.0953,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016985981308411213,
+      "loss": 4.4979,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016962616822429905,
+      "loss": 4.7325,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016939252336448598,
+      "loss": 4.6278,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016915887850467288,
+      "loss": 5.0017,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016892523364485978,
+      "loss": 4.364,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001686915887850467,
+      "loss": 5.0188,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016845794392523363,
+      "loss": 4.2547,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016822429906542053,
+      "loss": 4.7862,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016799065420560746,
+      "loss": 4.7286,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001677570093457944,
+      "loss": 4.7978,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016752336448598132,
+      "loss": 4.5335,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001672897196261682,
+      "loss": 4.5555,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016705607476635512,
+      "loss": 4.3638,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016682242990654204,
+      "loss": 4.508,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016658878504672897,
+      "loss": 5.2232,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016635514018691587,
+      "loss": 4.7749,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016612149532710277,
+      "loss": 5.182,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001658878504672897,
+      "loss": 4.9489,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016565420560747662,
+      "loss": 4.659,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016542056074766352,
+      "loss": 4.6653,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016518691588785045,
+      "loss": 4.874,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016495327102803738,
+      "loss": 4.9778,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001647196261682243,
+      "loss": 4.8992,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016448598130841118,
+      "loss": 4.3147,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001642523364485981,
+      "loss": 4.7119,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016401869158878503,
+      "loss": 4.7404,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016378504672897196,
+      "loss": 4.8604,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016355140186915886,
+      "loss": 4.8493,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016331775700934576,
+      "loss": 5.0921,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001630841121495327,
+      "loss": 4.937,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016285046728971962,
+      "loss": 4.3971,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016261682242990652,
+      "loss": 4.5391,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016238317757009344,
+      "loss": 4.8539,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016214953271028037,
+      "loss": 4.9911,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001619158878504673,
+      "loss": 5.5408,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016168224299065417,
+      "loss": 4.7559,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001614485981308411,
+      "loss": 5.7429,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016121495327102802,
+      "loss": 5.2865,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016098130841121495,
+      "loss": 5.9881,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016074766355140185,
+      "loss": 4.863,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016051401869158875,
+      "loss": 5.001,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016028037383177568,
+      "loss": 4.4963,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001600467289719626,
+      "loss": 4.7383,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001598130841121495,
+      "loss": 4.771,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015957943925233643,
+      "loss": 4.3538,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015934579439252336,
+      "loss": 4.5339,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001591121495327103,
+      "loss": 4.4357,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015887850467289716,
+      "loss": 4.3471,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586448598130841,
+      "loss": 4.4512,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015841121495327101,
+      "loss": 4.5906,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015817757009345794,
+      "loss": 4.7666,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015794392523364484,
+      "loss": 4.5706,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015771028037383174,
+      "loss": 4.7768,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015747663551401867,
+      "loss": 4.7465,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001572429906542056,
+      "loss": 5.0073,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001570093457943925,
+      "loss": 5.1489,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015677570093457942,
+      "loss": 4.6082,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015654205607476635,
+      "loss": 4.7613,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015630841121495328,
+      "loss": 4.6944,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015607476635514015,
+      "loss": 4.5798,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015584112149532708,
+      "loss": 4.564,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000155607476635514,
+      "loss": 4.6052,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015537383177570093,
+      "loss": 4.5484,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015514018691588783,
+      "loss": 4.5938,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015490654205607473,
+      "loss": 4.4018,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015467289719626166,
+      "loss": 5.1897,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001544392523364486,
+      "loss": 4.3624,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001542056074766355,
+      "loss": 4.3703,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015397196261682241,
+      "loss": 5.0461,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015373831775700934,
+      "loss": 4.3255,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015350467289719627,
+      "loss": 5.0283,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015327102803738314,
+      "loss": 4.823,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015303738317757007,
+      "loss": 4.5654,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000152803738317757,
+      "loss": 4.9188,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015257009345794392,
+      "loss": 4.5866,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015233644859813082,
+      "loss": 5.1037,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015210280373831772,
+      "loss": 4.4932,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015186915887850465,
+      "loss": 4.9721,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015163551401869158,
+      "loss": 4.7306,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015140186915887848,
+      "loss": 4.4082,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001511682242990654,
+      "loss": 4.7153,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015093457943925233,
+      "loss": 4.566,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015070093457943926,
+      "loss": 4.9826,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015046728971962613,
+      "loss": 5.618,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015023364485981306,
+      "loss": 4.8602,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015,
+      "loss": 4.494,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014976635514018691,
+      "loss": 4.8555,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014953271028037381,
+      "loss": 4.8804,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014929906542056074,
+      "loss": 4.9153,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014906542056074764,
+      "loss": 5.914,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014883177570093457,
+      "loss": 4.6331,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014859813084112147,
+      "loss": 4.8212,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001483644859813084,
+      "loss": 5.0512,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014813084112149532,
+      "loss": 4.592,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014789719626168225,
+      "loss": 4.6896,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014766355140186915,
+      "loss": 4.6735,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014742990654205608,
+      "loss": 4.9311,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014719626168224298,
+      "loss": 4.4262,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001469626168224299,
+      "loss": 4.5413,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001467289719626168,
+      "loss": 4.421,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014649532710280373,
+      "loss": 4.5932,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014626168224299063,
+      "loss": 4.6563,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014602803738317756,
+      "loss": 4.6683,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014579439252336446,
+      "loss": 4.7811,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001455607476635514,
+      "loss": 4.3987,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014532710280373831,
+      "loss": 4.4767,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014509345794392524,
+      "loss": 4.7227,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014485981308411214,
+      "loss": 4.6941,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014462616822429907,
+      "loss": 4.3973,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014439252336448597,
+      "loss": 4.6582,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001441588785046729,
+      "loss": 4.7733,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001439252336448598,
+      "loss": 4.9257,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014369158878504672,
+      "loss": 4.3115,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014345794392523362,
+      "loss": 4.8651,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014322429906542055,
+      "loss": 4.4354,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014299065420560745,
+      "loss": 4.7014,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014275700934579438,
+      "loss": 4.788,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001425233644859813,
+      "loss": 4.7997,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014228971962616823,
+      "loss": 5.054,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014205607476635513,
+      "loss": 4.3161,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014182242990654206,
+      "loss": 4.5677,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014158878504672896,
+      "loss": 4.4084,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014135514018691589,
+      "loss": 4.9911,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014112149532710279,
+      "loss": 4.4559,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001408878504672897,
+      "loss": 4.5061,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001406542056074766,
+      "loss": 5.236,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014042056074766354,
+      "loss": 4.4686,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014018691588785044,
+      "loss": 4.8665,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013995327102803737,
+      "loss": 4.6375,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001397196261682243,
+      "loss": 4.5114,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013948598130841122,
+      "loss": 4.8559,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013925233644859812,
+      "loss": 4.7208,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013901869158878505,
+      "loss": 5.0585,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013878504672897195,
+      "loss": 4.9908,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013855140186915888,
+      "loss": 4.6023,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013831775700934578,
+      "loss": 4.9435,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001380841121495327,
+      "loss": 4.6312,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001378504672897196,
+      "loss": 4.3065,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013761682242990653,
+      "loss": 4.8449,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013738317757009343,
+      "loss": 4.7368,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013714953271028036,
+      "loss": 4.3137,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013691588785046729,
+      "loss": 4.7101,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001366822429906542,
+      "loss": 4.6089,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001364485981308411,
+      "loss": 4.5011,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013621495327102804,
+      "loss": 4.5113,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013598130841121494,
+      "loss": 4.2968,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013574766355140187,
+      "loss": 4.79,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013551401869158877,
+      "loss": 4.8039,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001352803738317757,
+      "loss": 4.5847,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001350467289719626,
+      "loss": 4.553,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013481308411214952,
+      "loss": 4.4305,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013457943925233642,
+      "loss": 4.7628,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013434579439252335,
+      "loss": 4.7811,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013411214953271028,
+      "loss": 4.6595,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001338785046728972,
+      "loss": 4.4837,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001336448598130841,
+      "loss": 4.5001,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013341121495327103,
+      "loss": 4.6376,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013317757009345793,
+      "loss": 4.3984,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013294392523364486,
+      "loss": 4.2217,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013271028037383176,
+      "loss": 4.3163,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013247663551401869,
+      "loss": 4.8204,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013224299065420559,
+      "loss": 4.4814,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001320093457943925,
+      "loss": 4.6491,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001317757009345794,
+      "loss": 4.7224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013154205607476634,
+      "loss": 4.6153,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013130841121495327,
+      "loss": 5.1254,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001310747663551402,
+      "loss": 4.6913,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001308411214953271,
+      "loss": 4.4284,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013060747663551402,
+      "loss": 4.4841,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013037383177570092,
+      "loss": 4.3891,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013014018691588785,
+      "loss": 5.0009,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012990654205607475,
+      "loss": 4.694,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012967289719626168,
+      "loss": 5.124,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012943925233644858,
+      "loss": 4.7175,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001292056074766355,
+      "loss": 4.4971,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001289719626168224,
+      "loss": 4.7049,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012873831775700933,
+      "loss": 4.6269,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012850467289719626,
+      "loss": 4.6521,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012827102803738318,
+      "loss": 4.3151,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012803738317757008,
+      "loss": 4.6416,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000127803738317757,
+      "loss": 4.9035,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001275700934579439,
+      "loss": 4.5872,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012733644859813084,
+      "loss": 4.9876,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012710280373831774,
+      "loss": 4.964,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012686915887850467,
+      "loss": 4.4855,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012663551401869157,
+      "loss": 4.1817,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001264018691588785,
+      "loss": 4.8327,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001261682242990654,
+      "loss": 4.7371,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012593457943925232,
+      "loss": 4.6397,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012570093457943925,
+      "loss": 4.7792,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012546728971962618,
+      "loss": 4.8694,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012523364485981308,
+      "loss": 4.7286,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000125,
+      "loss": 4.8146,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001247663551401869,
+      "loss": 4.7488,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012453271028037383,
+      "loss": 4.7155,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012429906542056073,
+      "loss": 4.9373,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012406542056074766,
+      "loss": 4.4673,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012383177570093456,
+      "loss": 4.3828,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012359813084112148,
+      "loss": 4.3086,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012336448598130838,
+      "loss": 4.5156,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001231308411214953,
+      "loss": 5.0298,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012289719626168224,
+      "loss": 4.3251,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012266355140186917,
+      "loss": 4.6947,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012242990654205607,
+      "loss": 4.6539,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000122196261682243,
+      "loss": 4.3909,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001219626168224299,
+      "loss": 5.1204,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012172897196261682,
+      "loss": 4.6467,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012149532710280372,
+      "loss": 4.7147,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012126168224299065,
+      "loss": 4.6279,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012102803738317756,
+      "loss": 4.6076,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012079439252336448,
+      "loss": 4.4556,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012056074766355139,
+      "loss": 4.5697,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012032710280373832,
+      "loss": 4.3969,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012009345794392522,
+      "loss": 4.8635,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011985981308411214,
+      "loss": 4.4007,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011962616822429906,
+      "loss": 4.4937,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011939252336448597,
+      "loss": 4.9556,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011915887850467288,
+      "loss": 4.6646,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011892523364485981,
+      "loss": 4.7646,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011869158878504671,
+      "loss": 4.7527,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011845794392523364,
+      "loss": 4.6587,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011822429906542055,
+      "loss": 4.5145,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011799065420560747,
+      "loss": 4.5064,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011775700934579438,
+      "loss": 4.7615,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001175233644859813,
+      "loss": 4.4567,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001172897196261682,
+      "loss": 4.5432,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011705607476635513,
+      "loss": 4.5347,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011682242990654205,
+      "loss": 4.7786,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011658878504672896,
+      "loss": 4.5019,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011635514018691587,
+      "loss": 4.5446,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001161214953271028,
+      "loss": 4.5309,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001158878504672897,
+      "loss": 4.6369,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011565420560747663,
+      "loss": 5.2667,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011542056074766354,
+      "loss": 6.2945,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011518691588785046,
+      "loss": 4.3663,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011495327102803737,
+      "loss": 4.3191,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001147196261682243,
+      "loss": 4.1791,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001144859813084112,
+      "loss": 4.237,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011425233644859812,
+      "loss": 4.0311,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011401869158878504,
+      "loss": 6.0924,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011378504672897195,
+      "loss": 4.2445,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011355140186915887,
+      "loss": 4.8752,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011331775700934579,
+      "loss": 4.7493,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011308411214953269,
+      "loss": 4.6395,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011285046728971962,
+      "loss": 4.5663,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011261682242990653,
+      "loss": 4.4646,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011238317757009345,
+      "loss": 4.6283,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011214953271028036,
+      "loss": 4.5374,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011191588785046729,
+      "loss": 4.2707,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011168224299065419,
+      "loss": 4.4542,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011144859813084112,
+      "loss": 4.3583,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011121495327102803,
+      "loss": 4.3053,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011098130841121494,
+      "loss": 4.7099,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011074766355140186,
+      "loss": 4.3973,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011051401869158878,
+      "loss": 4.3032,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011028037383177568,
+      "loss": 4.8801,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011004672897196261,
+      "loss": 4.7108,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010981308411214952,
+      "loss": 4.3154,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010957943925233644,
+      "loss": 5.1603,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010934579439252335,
+      "loss": 4.4641,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010911214953271028,
+      "loss": 4.1332,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010887850467289718,
+      "loss": 4.2032,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001086448598130841,
+      "loss": 4.4041,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010841121495327102,
+      "loss": 4.5958,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010817757009345793,
+      "loss": 4.3467,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010794392523364485,
+      "loss": 4.6209,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010771028037383177,
+      "loss": 4.2929,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010747663551401867,
+      "loss": 4.1587,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001072429906542056,
+      "loss": 4.7793,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010700934579439251,
+      "loss": 4.7722,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010677570093457943,
+      "loss": 4.8571,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010654205607476634,
+      "loss": 4.2917,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010630841121495327,
+      "loss": 4.9375,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010607476635514017,
+      "loss": 4.5551,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001058411214953271,
+      "loss": 4.6761,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010560747663551401,
+      "loss": 4.3065,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010537383177570092,
+      "loss": 4.6009,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010514018691588784,
+      "loss": 4.7881,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010490654205607476,
+      "loss": 4.6032,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010467289719626166,
+      "loss": 4.806,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010443925233644859,
+      "loss": 5.0689,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001042056074766355,
+      "loss": 4.8513,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010397196261682242,
+      "loss": 4.7286,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010373831775700933,
+      "loss": 4.6413,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010350467289719626,
+      "loss": 5.067,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010327102803738316,
+      "loss": 4.7387,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010303738317757009,
+      "loss": 4.5827,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000102803738317757,
+      "loss": 5.3836,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010257009345794391,
+      "loss": 4.4282,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010233644859813083,
+      "loss": 4.5781,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010210280373831776,
+      "loss": 4.6178,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010186915887850466,
+      "loss": 4.6899,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010163551401869158,
+      "loss": 4.6505,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001014018691588785,
+      "loss": 4.6107,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010116822429906541,
+      "loss": 4.8563,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010093457943925232,
+      "loss": 4.4199,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010070093457943925,
+      "loss": 4.7724,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010046728971962615,
+      "loss": 4.1903,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010023364485981308,
+      "loss": 4.6997,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 4.6855,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.97663551401869e-05,
+      "loss": 4.997,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.953271028037382e-05,
+      "loss": 4.4632,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.929906542056075e-05,
+      "loss": 4.4651,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.906542056074765e-05,
+      "loss": 4.2949,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.883177570093457e-05,
+      "loss": 4.9577,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.859813084112149e-05,
+      "loss": 4.8504,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.83644859813084e-05,
+      "loss": 4.3042,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.813084112149531e-05,
+      "loss": 4.581,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.789719626168224e-05,
+      "loss": 4.3119,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.766355140186914e-05,
+      "loss": 4.6041,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.742990654205607e-05,
+      "loss": 4.6708,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.719626168224298e-05,
+      "loss": 4.7258,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.69626168224299e-05,
+      "loss": 4.2403,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.672897196261681e-05,
+      "loss": 4.9588,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.649532710280374e-05,
+      "loss": 4.6089,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.626168224299064e-05,
+      "loss": 4.4039,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.602803738317756e-05,
+      "loss": 4.3359,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.579439252336448e-05,
+      "loss": 4.6864,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.556074766355139e-05,
+      "loss": 4.5197,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.53271028037383e-05,
+      "loss": 4.4822,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.509345794392523e-05,
+      "loss": 4.4383,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.485981308411213e-05,
+      "loss": 4.0791,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.462616822429906e-05,
+      "loss": 4.5552,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.439252336448597e-05,
+      "loss": 4.4294,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.415887850467289e-05,
+      "loss": 4.6212,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.39252336448598e-05,
+      "loss": 4.8217,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.369158878504673e-05,
+      "loss": 4.4626,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.345794392523363e-05,
+      "loss": 4.9503,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.322429906542055e-05,
+      "loss": 4.6515,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.299065420560747e-05,
+      "loss": 4.4508,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.275700934579438e-05,
+      "loss": 4.7594,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.25233644859813e-05,
+      "loss": 4.2903,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.228971962616822e-05,
+      "loss": 4.349,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.205607476635512e-05,
+      "loss": 4.4674,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.182242990654205e-05,
+      "loss": 4.6315,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.158878504672896e-05,
+      "loss": 4.9254,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.135514018691588e-05,
+      "loss": 4.3552,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.112149532710279e-05,
+      "loss": 4.8721,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.088785046728972e-05,
+      "loss": 4.47,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.065420560747662e-05,
+      "loss": 4.5815,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.042056074766354e-05,
+      "loss": 4.7547,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.018691588785046e-05,
+      "loss": 4.2618,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.995327102803737e-05,
+      "loss": 4.7009,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.971962616822429e-05,
+      "loss": 4.339,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.948598130841121e-05,
+      "loss": 4.5113,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.925233644859811e-05,
+      "loss": 4.2392,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.901869158878504e-05,
+      "loss": 5.1986,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.878504672897195e-05,
+      "loss": 4.5835,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.855140186915887e-05,
+      "loss": 4.4664,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.831775700934578e-05,
+      "loss": 4.4967,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.808411214953271e-05,
+      "loss": 4.3904,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.785046728971961e-05,
+      "loss": 4.1992,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.761682242990654e-05,
+      "loss": 4.2952,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.738317757009345e-05,
+      "loss": 4.7417,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.714953271028036e-05,
+      "loss": 4.9638,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.691588785046728e-05,
+      "loss": 4.522,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.66822429906542e-05,
+      "loss": 4.6829,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.64485981308411e-05,
+      "loss": 4.6061,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.621495327102803e-05,
+      "loss": 4.6179,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.598130841121494e-05,
+      "loss": 4.8331,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.574766355140186e-05,
+      "loss": 4.3792,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.551401869158877e-05,
+      "loss": 4.5551,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.52803738317757e-05,
+      "loss": 4.7392,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.50467289719626e-05,
+      "loss": 4.3609,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.481308411214953e-05,
+      "loss": 4.4721,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.457943925233644e-05,
+      "loss": 5.0827,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.434579439252335e-05,
+      "loss": 4.5881,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.411214953271027e-05,
+      "loss": 4.2625,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.38785046728972e-05,
+      "loss": 4.5462,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.36448598130841e-05,
+      "loss": 4.6827,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.341121495327102e-05,
+      "loss": 4.9506,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.317757009345794e-05,
+      "loss": 5.4046,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.294392523364485e-05,
+      "loss": 4.555,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.271028037383176e-05,
+      "loss": 5.0098,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.247663551401869e-05,
+      "loss": 4.2566,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.224299065420559e-05,
+      "loss": 4.4053,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.200934579439252e-05,
+      "loss": 4.2827,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.177570093457943e-05,
+      "loss": 4.6128,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.154205607476634e-05,
+      "loss": 4.4264,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.130841121495326e-05,
+      "loss": 4.6697,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.107476635514018e-05,
+      "loss": 4.5781,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.084112149532708e-05,
+      "loss": 5.115,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.060747663551401e-05,
+      "loss": 4.3891,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.037383177570093e-05,
+      "loss": 4.3513,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.014018691588784e-05,
+      "loss": 4.3578,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.990654205607475e-05,
+      "loss": 4.5404,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.967289719626168e-05,
+      "loss": 4.839,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.943925233644858e-05,
+      "loss": 4.5189,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.920560747663551e-05,
+      "loss": 4.6335,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.897196261682242e-05,
+      "loss": 4.8941,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.873831775700933e-05,
+      "loss": 4.4223,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.850467289719625e-05,
+      "loss": 4.3988,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.827102803738318e-05,
+      "loss": 4.5387,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.803738317757008e-05,
+      "loss": 4.3465,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.7803738317757e-05,
+      "loss": 4.1828,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.757009345794392e-05,
+      "loss": 4.3629,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.733644859813083e-05,
+      "loss": 4.4213,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.710280373831774e-05,
+      "loss": 4.9567,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.686915887850467e-05,
+      "loss": 4.5918,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.663551401869157e-05,
+      "loss": 4.6807,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.64018691588785e-05,
+      "loss": 4.9541,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.616822429906541e-05,
+      "loss": 4.5968,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.593457943925233e-05,
+      "loss": 4.6322,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.570093457943924e-05,
+      "loss": 4.4,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.546728971962617e-05,
+      "loss": 4.4597,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.523364485981307e-05,
+      "loss": 4.8758,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.5e-05,
+      "loss": 4.3989,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.476635514018691e-05,
+      "loss": 4.3877,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.453271028037382e-05,
+      "loss": 4.5695,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.429906542056073e-05,
+      "loss": 4.3639,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.406542056074766e-05,
+      "loss": 4.1633,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.383177570093458e-05,
+      "loss": 4.5091,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.359813084112149e-05,
+      "loss": 4.1899,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.33644859813084e-05,
+      "loss": 4.3316,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.313084112149532e-05,
+      "loss": 4.2494,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.289719626168223e-05,
+      "loss": 4.4593,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.266355140186916e-05,
+      "loss": 4.6835,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.242990654205607e-05,
+      "loss": 4.5329,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.219626168224298e-05,
+      "loss": 4.0649,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.19626168224299e-05,
+      "loss": 4.4632,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172897196261681e-05,
+      "loss": 4.3951,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.149532710280372e-05,
+      "loss": 4.4505,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.126168224299065e-05,
+      "loss": 4.9738,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.102803738317757e-05,
+      "loss": 4.4743,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.079439252336448e-05,
+      "loss": 4.11,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.056074766355139e-05,
+      "loss": 4.3696,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.03271028037383e-05,
+      "loss": 4.7327,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.009345794392522e-05,
+      "loss": 4.6291,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.985981308411215e-05,
+      "loss": 4.668,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.962616822429906e-05,
+      "loss": 4.5731,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.939252336448597e-05,
+      "loss": 4.2246,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.915887850467289e-05,
+      "loss": 4.8637,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.89252336448598e-05,
+      "loss": 4.5471,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.869158878504672e-05,
+      "loss": 4.011,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.845794392523364e-05,
+      "loss": 4.3462,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822429906542056e-05,
+      "loss": 4.3227,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.799065420560747e-05,
+      "loss": 5.5984,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.775700934579438e-05,
+      "loss": 4.7689,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.75233644859813e-05,
+      "loss": 4.7399,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.728971962616821e-05,
+      "loss": 4.2169,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.705607476635514e-05,
+      "loss": 4.5932,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.684563636779785,
+      "eval_runtime": 902.9412,
+      "eval_samples_per_second": 2.926,
+      "eval_steps_per_second": 0.367,
+      "eval_wer": 1.962951209837366,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682242990654205e-05,
+      "loss": 4.5794,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658878504672897e-05,
+      "loss": 4.408,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635514018691588e-05,
+      "loss": 4.5533,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.612149532710279e-05,
+      "loss": 4.6865,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.58878504672897e-05,
+      "loss": 4.7087,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.565420560747663e-05,
+      "loss": 4.5525,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.542056074766355e-05,
+      "loss": 4.3097,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.518691588785046e-05,
+      "loss": 4.4776,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.495327102803737e-05,
+      "loss": 4.4956,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.471962616822429e-05,
+      "loss": 4.4221,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.44859813084112e-05,
+      "loss": 4.2104,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.425233644859813e-05,
+      "loss": 4.4226,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.401869158878504e-05,
+      "loss": 4.1319,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.378504672897196e-05,
+      "loss": 4.3302,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.355140186915887e-05,
+      "loss": 4.3782,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.331775700934578e-05,
+      "loss": 4.7651,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.30841121495327e-05,
+      "loss": 4.2555,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.285046728971962e-05,
+      "loss": 4.5066,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.261682242990654e-05,
+      "loss": 4.6007,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.238317757009345e-05,
+      "loss": 4.5272,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.214953271028036e-05,
+      "loss": 4.4501,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.191588785046728e-05,
+      "loss": 4.6648,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.168224299065419e-05,
+      "loss": 4.8737,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144859813084112e-05,
+      "loss": 4.248,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.121495327102803e-05,
+      "loss": 4.2556,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.098130841121495e-05,
+      "loss": 4.1853,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.074766355140186e-05,
+      "loss": 4.3537,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.051401869158878e-05,
+      "loss": 4.3641,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0280373831775694e-05,
+      "loss": 4.1877,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.004672897196261e-05,
+      "loss": 4.3508,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.981308411214953e-05,
+      "loss": 4.3853,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.957943925233644e-05,
+      "loss": 4.3307,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.9345794392523356e-05,
+      "loss": 4.8142,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.9112149532710276e-05,
+      "loss": 4.3588,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.887850467289719e-05,
+      "loss": 4.4076,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.86448598130841e-05,
+      "loss": 4.09,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8411214953271024e-05,
+      "loss": 4.7277,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.817757009345794e-05,
+      "loss": 4.655,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.794392523364485e-05,
+      "loss": 4.3764,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.771028037383177e-05,
+      "loss": 4.5638,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.7476635514018685e-05,
+      "loss": 4.7757,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.72429906542056e-05,
+      "loss": 4.1836,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.700934579439252e-05,
+      "loss": 4.6114,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.677570093457943e-05,
+      "loss": 4.7632,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.6542056074766346e-05,
+      "loss": 5.0162,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.630841121495327e-05,
+      "loss": 4.1637,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.607476635514018e-05,
+      "loss": 4.6761,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.5841121495327094e-05,
+      "loss": 4.6081,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.5607476635514014e-05,
+      "loss": 4.9931,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.537383177570093e-05,
+      "loss": 4.278,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.514018691588784e-05,
+      "loss": 4.6468,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.490654205607476e-05,
+      "loss": 4.4786,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.4672897196261676e-05,
+      "loss": 4.3608,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.443925233644859e-05,
+      "loss": 4.3711,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.420560747663551e-05,
+      "loss": 4.5756,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.397196261682242e-05,
+      "loss": 4.9567,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.373831775700934e-05,
+      "loss": 4.5372,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.350467289719626e-05,
+      "loss": 4.6352,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.327102803738317e-05,
+      "loss": 4.356,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3037383177570085e-05,
+      "loss": 4.6309,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.2803738317757005e-05,
+      "loss": 4.444,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.257009345794392e-05,
+      "loss": 4.3915,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.233644859813083e-05,
+      "loss": 4.557,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.210280373831775e-05,
+      "loss": 4.4178,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.1869158878504666e-05,
+      "loss": 4.7344,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.163551401869158e-05,
+      "loss": 4.2944,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.14018691588785e-05,
+      "loss": 4.0712,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.1168224299065414e-05,
+      "loss": 4.4221,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.093457943925233e-05,
+      "loss": 4.4791,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.070093457943925e-05,
+      "loss": 4.3037,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.046728971962616e-05,
+      "loss": 4.3704,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.0233644859813075e-05,
+      "loss": 4.6434,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 4.5905,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.976635514018691e-05,
+      "loss": 4.2027,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.953271028037382e-05,
+      "loss": 4.2243,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.929906542056074e-05,
+      "loss": 4.4988,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.906542056074766e-05,
+      "loss": 4.4496,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.883177570093457e-05,
+      "loss": 4.7679,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.859813084112149e-05,
+      "loss": 4.5786,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8364485981308405e-05,
+      "loss": 4.9594,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.813084112149532e-05,
+      "loss": 4.5915,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.789719626168224e-05,
+      "loss": 4.5056,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.766355140186915e-05,
+      "loss": 4.3929,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7429906542056066e-05,
+      "loss": 4.2318,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7196261682242986e-05,
+      "loss": 4.3119,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.69626168224299e-05,
+      "loss": 4.3862,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6728971962616814e-05,
+      "loss": 4.2969,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6495327102803734e-05,
+      "loss": 4.4787,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.626168224299065e-05,
+      "loss": 4.3504,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.602803738317756e-05,
+      "loss": 4.2523,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.579439252336448e-05,
+      "loss": 4.3615,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5560747663551395e-05,
+      "loss": 4.2794,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.532710280373831e-05,
+      "loss": 4.291,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.509345794392523e-05,
+      "loss": 4.3676,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.485981308411214e-05,
+      "loss": 4.4046,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4626168224299057e-05,
+      "loss": 4.5139,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439252336448598e-05,
+      "loss": 4.57,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.415887850467289e-05,
+      "loss": 4.6402,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3925233644859804e-05,
+      "loss": 5.634,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3691588785046725e-05,
+      "loss": 4.7402,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.345794392523364e-05,
+      "loss": 4.4763,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.322429906542055e-05,
+      "loss": 4.1862,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.299065420560747e-05,
+      "loss": 4.3362,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2757009345794386e-05,
+      "loss": 4.5014,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.25233644859813e-05,
+      "loss": 4.155,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.228971962616822e-05,
+      "loss": 4.4603,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2056074766355134e-05,
+      "loss": 4.5002,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.182242990654205e-05,
+      "loss": 4.3655,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.158878504672897e-05,
+      "loss": 4.5439,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.135514018691588e-05,
+      "loss": 4.2401,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1121495327102795e-05,
+      "loss": 4.6235,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.0887850467289715e-05,
+      "loss": 4.4492,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.065420560747663e-05,
+      "loss": 4.6652,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.042056074766354e-05,
+      "loss": 4.2725,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.018691588785046e-05,
+      "loss": 4.3769,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9953271028037377e-05,
+      "loss": 4.3812,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.971962616822429e-05,
+      "loss": 4.5223,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.948598130841121e-05,
+      "loss": 4.3296,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9252336448598124e-05,
+      "loss": 4.3042,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.901869158878504e-05,
+      "loss": 4.3898,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878504672897196e-05,
+      "loss": 4.8097,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.855140186915887e-05,
+      "loss": 4.096,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8317757009345785e-05,
+      "loss": 4.4866,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8084112149532706e-05,
+      "loss": 4.2791,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785046728971962e-05,
+      "loss": 4.4026,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.761682242990653e-05,
+      "loss": 4.6418,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7383177570093454e-05,
+      "loss": 4.4598,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.714953271028037e-05,
+      "loss": 4.2184,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.691588785046729e-05,
+      "loss": 4.4338,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.66822429906542e-05,
+      "loss": 4.3216,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6448598130841115e-05,
+      "loss": 4.281,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6214953271028035e-05,
+      "loss": 4.4902,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.598130841121495e-05,
+      "loss": 4.5964,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574766355140186e-05,
+      "loss": 4.2606,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.551401869158878e-05,
+      "loss": 4.4497,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5280373831775697e-05,
+      "loss": 4.3028,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.504672897196261e-05,
+      "loss": 4.2589,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481308411214953e-05,
+      "loss": 4.941,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4579439252336444e-05,
+      "loss": 4.2086,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.434579439252336e-05,
+      "loss": 4.643,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411214953271028e-05,
+      "loss": 4.5205,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387850467289719e-05,
+      "loss": 4.4125,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3644859813084105e-05,
+      "loss": 4.4828,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3411214953271026e-05,
+      "loss": 4.4454,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.317757009345794e-05,
+      "loss": 4.7431,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294392523364485e-05,
+      "loss": 4.7787,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 4.4746,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247663551401869e-05,
+      "loss": 5.2549,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.22429906542056e-05,
+      "loss": 4.6796,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.200934579439252e-05,
+      "loss": 4.4343,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1775700934579435e-05,
+      "loss": 4.4162,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.154205607476635e-05,
+      "loss": 4.3306,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.130841121495327e-05,
+      "loss": 4.4044,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.107476635514018e-05,
+      "loss": 4.3585,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0841121495327096e-05,
+      "loss": 4.2039,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0607476635514017e-05,
+      "loss": 4.3417,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.037383177570093e-05,
+      "loss": 4.5688,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0140186915887847e-05,
+      "loss": 4.3957,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9906542056074764e-05,
+      "loss": 4.204,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9672897196261678e-05,
+      "loss": 4.3833,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9439252336448595e-05,
+      "loss": 4.4503,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9205607476635512e-05,
+      "loss": 4.392,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8971962616822425e-05,
+      "loss": 4.0731,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8738317757009342e-05,
+      "loss": 4.4957,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.850467289719626e-05,
+      "loss": 4.1702,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8271028037383173e-05,
+      "loss": 4.4144,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.803738317757009e-05,
+      "loss": 4.5028,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7803738317757007e-05,
+      "loss": 4.1839,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.757009345794392e-05,
+      "loss": 4.2156,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7336448598130838e-05,
+      "loss": 4.4568,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7102803738317755e-05,
+      "loss": 4.4087,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.686915887850467e-05,
+      "loss": 4.6661,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6635514018691585e-05,
+      "loss": 4.2527,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6401869158878502e-05,
+      "loss": 4.4357,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6168224299065416e-05,
+      "loss": 4.6565,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5934579439252333e-05,
+      "loss": 4.206,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.570093457943925e-05,
+      "loss": 4.1303,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5467289719626164e-05,
+      "loss": 4.337,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.523364485981308e-05,
+      "loss": 4.6861,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4999999999999998e-05,
+      "loss": 4.8476,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.476635514018691e-05,
+      "loss": 4.0134,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.453271028037383e-05,
+      "loss": 4.098,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4299065420560745e-05,
+      "loss": 4.5636,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.406542056074766e-05,
+      "loss": 4.1809,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3831775700934576e-05,
+      "loss": 4.4536,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3598130841121493e-05,
+      "loss": 4.3799,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3364485981308407e-05,
+      "loss": 4.2094,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3130841121495324e-05,
+      "loss": 4.472,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.289719626168224e-05,
+      "loss": 4.4078,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2663551401869154e-05,
+      "loss": 4.26,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.242990654205607e-05,
+      "loss": 4.6382,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.219626168224299e-05,
+      "loss": 4.4575,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1962616822429902e-05,
+      "loss": 4.6798,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.172897196261682e-05,
+      "loss": 4.7406,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1495327102803736e-05,
+      "loss": 4.695,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126168224299065e-05,
+      "loss": 7.1009,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1028037383177567e-05,
+      "loss": 4.4314,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0794392523364484e-05,
+      "loss": 4.8768,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0560747663551397e-05,
+      "loss": 4.6938,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0327102803738314e-05,
+      "loss": 4.7121,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.009345794392523e-05,
+      "loss": 5.2045,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9859813084112145e-05,
+      "loss": 4.4894,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9626168224299062e-05,
+      "loss": 4.1926,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.939252336448598e-05,
+      "loss": 4.3933,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9158878504672893e-05,
+      "loss": 4.382,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.892523364485981e-05,
+      "loss": 4.2146,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8691588785046727e-05,
+      "loss": 4.1402,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8457943925233644e-05,
+      "loss": 4.6311,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8224299065420557e-05,
+      "loss": 4.8325,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7990654205607474e-05,
+      "loss": 4.5718,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775700934579439e-05,
+      "loss": 4.2275,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7523364485981305e-05,
+      "loss": 4.4351,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7289719626168222e-05,
+      "loss": 4.9079,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.705607476635514e-05,
+      "loss": 4.3129,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6822429906542053e-05,
+      "loss": 4.3238,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.658878504672897e-05,
+      "loss": 4.6123,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 4.749,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.61214953271028e-05,
+      "loss": 4.8785,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5887850467289717e-05,
+      "loss": 4.2084,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5654205607476634e-05,
+      "loss": 4.2968,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5420560747663548e-05,
+      "loss": 4.5638,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5186915887850465e-05,
+      "loss": 4.5775,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4953271028037382e-05,
+      "loss": 4.7003,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4719626168224297e-05,
+      "loss": 4.3962,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4485981308411213e-05,
+      "loss": 4.3603,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.425233644859813e-05,
+      "loss": 4.31,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4018691588785045e-05,
+      "loss": 4.7106,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.378504672897196e-05,
+      "loss": 4.4806,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3551401869158877e-05,
+      "loss": 4.6771,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3317757009345793e-05,
+      "loss": 4.443,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3084112149532708e-05,
+      "loss": 4.0589,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2850467289719625e-05,
+      "loss": 4.2644,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.261682242990654e-05,
+      "loss": 4.3072,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2383177570093456e-05,
+      "loss": 4.4704,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2149532710280373e-05,
+      "loss": 4.9955,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1915887850467288e-05,
+      "loss": 4.892,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1682242990654203e-05,
+      "loss": 4.0713,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.144859813084112e-05,
+      "loss": 4.4093,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1214953271028036e-05,
+      "loss": 4.3263,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0981308411214951e-05,
+      "loss": 4.2525,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0747663551401868e-05,
+      "loss": 4.4634,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0514018691588783e-05,
+      "loss": 4.4893,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0280373831775699e-05,
+      "loss": 4.218,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0046728971962616e-05,
+      "loss": 4.2345,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.813084112149531e-06,
+      "loss": 4.4428,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.579439252336446e-06,
+      "loss": 4.5344,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.345794392523363e-06,
+      "loss": 4.4899,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.112149532710279e-06,
+      "loss": 4.2383,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.878504672897196e-06,
+      "loss": 4.4987,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.644859813084111e-06,
+      "loss": 4.5145,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.411214953271026e-06,
+      "loss": 4.0823,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 4.2956,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.943925233644859e-06,
+      "loss": 4.5371,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.710280373831774e-06,
+      "loss": 4.7801,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.476635514018691e-06,
+      "loss": 4.5643,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.242990654205606e-06,
+      "loss": 4.3877,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.0093457943925225e-06,
+      "loss": 4.2976,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.775700934579439e-06,
+      "loss": 4.2241,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.542056074766354e-06,
+      "loss": 4.6761,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.30841121495327e-06,
+      "loss": 4.3976,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.074766355140186e-06,
+      "loss": 4.8187,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.841121495327102e-06,
+      "loss": 4.203,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.607476635514018e-06,
+      "loss": 4.2129,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.373831775700934e-06,
+      "loss": 4.661,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.140186915887849e-06,
+      "loss": 4.1663,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9065420560747655e-06,
+      "loss": 4.3171,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.672897196261682e-06,
+      "loss": 4.3728,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.439252336448598e-06,
+      "loss": 4.2334,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.205607476635513e-06,
+      "loss": 4.537,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.971962616822429e-06,
+      "loss": 4.1671,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7383177570093455e-06,
+      "loss": 4.2857,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5046728971962613e-06,
+      "loss": 4.2345,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.271028037383177e-06,
+      "loss": 4.3355,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.037383177570093e-06,
+      "loss": 4.3151,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.803738317757009e-06,
+      "loss": 4.285,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5700934579439247e-06,
+      "loss": 4.3268,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.336448598130841e-06,
+      "loss": 4.5804,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1028037383177566e-06,
+      "loss": 4.3359,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8691588785046728e-06,
+      "loss": 4.8936,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6355140186915885e-06,
+      "loss": 4.5387,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4018691588785045e-06,
+      "loss": 4.8575,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1682242990654204e-06,
+      "loss": 4.3431,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.345794392523364e-07,
+      "loss": 4.9246,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.009345794392522e-07,
+      "loss": 3.7122,
+      "step": 1784
+    },
     {
       "epoch": 1.0,
-      "step": 297,
+      "step": 1784,
       "total_flos": 0.0,
-      "train_loss": 4.260829443883414,
-      "train_runtime": 4532.5114,
-      "train_samples_per_second": 6.296,
-      "train_steps_per_second": 0.066
+      "train_loss": 4.771062652893665,
+      "train_runtime": 8121.4369,
+      "train_samples_per_second": 3.514,
+      "train_steps_per_second": 0.22
     }
   ],
-  "max_steps": 297,
+  "max_steps": 1784,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,