diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,5383 +2,2698 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 4.8006,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.7719,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 5.0552,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9065,
       "step": 2
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.8954,
+      "loss": 4.885,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 6.000000000000001e-07,
-      "loss": 4.8986,
+      "loss": 4.8861,
       "step": 4
     },
     {
       "epoch": 0.01,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 4.7319,
+      "loss": 4.8113,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.8418,
+      "loss": 4.8648,
       "step": 6
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.8198,
+      "loss": 4.8398,
       "step": 7
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.7216,
+      "loss": 4.8766,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.8858,
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8603,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.7542,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 4.7241,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-06,
-      "loss": 4.8323,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.8192,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.7913,
+      "epoch": 0.03,
+      "learning_rate": 2.2e-06,
+      "loss": 4.7752,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-06,
-      "loss": 4.757,
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.7171,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.7019,
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 4.707,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-06,
-      "loss": 4.6262,
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.5768,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.7446,
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 4.7214,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3e-06,
-      "loss": 4.7368,
+      "epoch": 0.04,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.6155,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.8308,
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.6982,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.5315,
+      "epoch": 0.04,
+      "learning_rate": 3.6e-06,
+      "loss": 4.5043,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-06,
-      "loss": 4.6872,
+      "epoch": 0.04,
+      "learning_rate": 3.8e-06,
+      "loss": 4.6563,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-06,
-      "loss": 4.5977,
+      "epoch": 0.05,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.6315,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.5822,
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.5689,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.5846,
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 4.6517,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-06,
-      "loss": 4.7091,
+      "epoch": 0.05,
+      "learning_rate": 4.6e-06,
+      "loss": 4.626,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6e-06,
-      "loss": 4.5738,
+      "epoch": 0.06,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.4167,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.7385,
+      "epoch": 0.06,
+      "learning_rate": 5e-06,
+      "loss": 4.6779,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5e-06,
-      "loss": 4.454,
+      "epoch": 0.06,
+      "learning_rate": 5.2e-06,
+      "loss": 4.5106,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-06,
-      "loss": 4.3818,
+      "epoch": 0.06,
+      "learning_rate": 5.4e-06,
+      "loss": 4.3687,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-06,
-      "loss": 4.3869,
+      "epoch": 0.07,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.3321,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.5749,
+      "epoch": 0.07,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5401,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.8e-06,
-      "loss": 4.5545,
+      "epoch": 0.07,
+      "learning_rate": 6e-06,
+      "loss": 4.5029,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6e-06,
-      "loss": 4.3159,
+      "epoch": 0.07,
+      "learning_rate": 6.2e-06,
+      "loss": 4.3607,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.2e-06,
-      "loss": 4.4299,
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.4283,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.2923,
+      "epoch": 0.08,
+      "learning_rate": 6.6e-06,
+      "loss": 4.3708,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.6e-06,
-      "loss": 4.4871,
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4411,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4389,
+      "epoch": 0.08,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.408,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.4767,
+      "epoch": 0.08,
+      "learning_rate": 7.2e-06,
+      "loss": 4.4535,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-06,
-      "loss": 4.484,
+      "epoch": 0.09,
+      "learning_rate": 7.4e-06,
+      "loss": 4.3956,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.4e-06,
-      "loss": 4.641,
+      "epoch": 0.09,
+      "learning_rate": 7.6e-06,
+      "loss": 4.5053,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.6e-06,
-      "loss": 4.4554,
+      "epoch": 0.09,
+      "learning_rate": 7.8e-06,
+      "loss": 4.3493,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.8e-06,
-      "loss": 4.3018,
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.3782,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.4761,
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.4784,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.2683,
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.2425,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.426,
+      "epoch": 0.1,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.4174,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.4915,
+      "epoch": 0.1,
+      "learning_rate": 8.8e-06,
+      "loss": 4.5094,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.8e-06,
-      "loss": 4.6235,
+      "epoch": 0.1,
+      "learning_rate": 9e-06,
+      "loss": 4.5613,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9e-06,
-      "loss": 4.4703,
+      "epoch": 0.11,
+      "learning_rate": 9.2e-06,
+      "loss": 4.4531,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.2e-06,
-      "loss": 4.8672,
+      "epoch": 0.11,
+      "learning_rate": 9.4e-06,
+      "loss": 4.7066,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.2e-06,
-      "loss": 5.0521,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.8332,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.4e-06,
-      "loss": 4.6531,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.5146,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2156,
+      "epoch": 0.11,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.3416,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.348,
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 4.214,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1e-05,
-      "loss": 4.2307,
+      "epoch": 0.12,
+      "learning_rate": 1.02e-05,
+      "loss": 4.3766,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.02e-05,
-      "loss": 4.2949,
+      "epoch": 0.12,
+      "learning_rate": 1.04e-05,
+      "loss": 4.2895,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.04e-05,
-      "loss": 4.28,
+      "epoch": 0.12,
+      "learning_rate": 1.06e-05,
+      "loss": 4.1125,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.06e-05,
-      "loss": 4.3738,
+      "epoch": 0.13,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3304,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.08e-05,
-      "loss": 4.344,
+      "epoch": 0.13,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.3424,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.5751,
+      "epoch": 0.13,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.4227,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.3086,
+      "epoch": 0.13,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.3562,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.1412,
+      "epoch": 0.13,
+      "learning_rate": 1.16e-05,
+      "loss": 4.217,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.16e-05,
-      "loss": 4.4096,
+      "epoch": 0.14,
+      "learning_rate": 1.18e-05,
+      "loss": 4.2907,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.18e-05,
-      "loss": 4.3565,
+      "epoch": 0.14,
+      "learning_rate": 1.2e-05,
+      "loss": 4.246,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2e-05,
-      "loss": 4.1857,
+      "epoch": 0.14,
+      "learning_rate": 1.22e-05,
+      "loss": 4.2075,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.22e-05,
-      "loss": 4.3458,
+      "epoch": 0.14,
+      "learning_rate": 1.24e-05,
+      "loss": 4.3065,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.24e-05,
-      "loss": 4.1284,
+      "epoch": 0.15,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.285,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.4129,
+      "epoch": 0.15,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.1584,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.1812,
-      "step": 67
+      "epoch": 0.15,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.2986,
+      "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.2234,
+      "epoch": 0.15,
+      "learning_rate": 1.32e-05,
+      "loss": 4.1812,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.32e-05,
-      "loss": 4.1651,
+      "epoch": 0.15,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.229,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.29,
+      "epoch": 0.16,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.1952,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.4585,
+      "epoch": 0.16,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.3216,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.3538,
+      "epoch": 0.16,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.3481,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.281,
+      "epoch": 0.16,
+      "learning_rate": 1.42e-05,
+      "loss": 4.1716,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.42e-05,
-      "loss": 4.3208,
+      "epoch": 0.17,
+      "learning_rate": 1.44e-05,
+      "loss": 4.2743,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.44e-05,
-      "loss": 4.1282,
+      "epoch": 0.17,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.2567,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.4093,
+      "epoch": 0.17,
+      "learning_rate": 1.48e-05,
+      "loss": 4.2005,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.48e-05,
-      "loss": 4.3022,
+      "epoch": 0.17,
+      "learning_rate": 1.5e-05,
+      "loss": 4.2588,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2166,
+      "epoch": 0.17,
+      "learning_rate": 1.52e-05,
+      "loss": 4.2041,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.52e-05,
-      "loss": 4.0831,
+      "epoch": 0.18,
+      "learning_rate": 1.54e-05,
+      "loss": 4.1828,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.54e-05,
-      "loss": 4.3279,
+      "epoch": 0.18,
+      "learning_rate": 1.56e-05,
+      "loss": 4.3437,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.56e-05,
-      "loss": 4.3771,
+      "epoch": 0.18,
+      "learning_rate": 1.58e-05,
+      "loss": 4.2745,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.58e-05,
-      "loss": 4.1824,
+      "epoch": 0.18,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.2209,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.3989,
+      "epoch": 0.19,
+      "learning_rate": 1.62e-05,
+      "loss": 4.339,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.62e-05,
-      "loss": 4.3379,
+      "epoch": 0.19,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.2074,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.2745,
+      "epoch": 0.19,
+      "learning_rate": 1.66e-05,
+      "loss": 4.287,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.66e-05,
-      "loss": 4.2694,
+      "epoch": 0.19,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.2516,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.2729,
+      "epoch": 0.2,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.3282,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.2621,
+      "epoch": 0.2,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.2117,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.3116,
+      "epoch": 0.2,
+      "learning_rate": 1.74e-05,
+      "loss": 4.3481,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.74e-05,
-      "loss": 4.2443,
+      "epoch": 0.2,
+      "learning_rate": 1.76e-05,
+      "loss": 4.3362,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.76e-05,
-      "loss": 4.4649,
+      "epoch": 0.2,
+      "learning_rate": 1.78e-05,
+      "loss": 4.2163,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.78e-05,
-      "loss": 4.3735,
+      "epoch": 0.21,
+      "learning_rate": 1.8e-05,
+      "loss": 4.4685,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.8e-05,
-      "loss": 4.0979,
+      "epoch": 0.21,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.3722,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.3195,
+      "epoch": 0.21,
+      "learning_rate": 1.84e-05,
+      "loss": 4.3702,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.84e-05,
-      "loss": 4.5964,
+      "epoch": 0.21,
+      "learning_rate": 1.86e-05,
+      "loss": 4.2807,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.86e-05,
-      "loss": 4.5072,
+      "epoch": 0.22,
+      "learning_rate": 1.88e-05,
+      "loss": 4.5093,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.88e-05,
-      "loss": 4.2899,
+      "epoch": 0.22,
+      "learning_rate": 1.9e-05,
+      "loss": 4.4255,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9e-05,
-      "loss": 4.5847,
+      "epoch": 0.22,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.2955,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.3735,
+      "epoch": 0.22,
+      "learning_rate": 1.94e-05,
+      "loss": 4.631,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.94e-05,
-      "loss": 4.3218,
+      "epoch": 0.22,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.4001,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.177,
+      "epoch": 0.23,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.2505,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.209,
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 4.1911,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2e-05,
-      "loss": 4.327,
+      "epoch": 0.23,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.0639,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.2536,
+      "epoch": 0.23,
+      "learning_rate": 2.04e-05,
+      "loss": 4.1361,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.04e-05,
-      "loss": 4.0728,
+      "epoch": 0.24,
+      "learning_rate": 2.06e-05,
+      "loss": 4.0851,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.06e-05,
-      "loss": 4.2959,
+      "epoch": 0.24,
+      "learning_rate": 2.08e-05,
+      "loss": 4.1384,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.08e-05,
-      "loss": 4.2988,
+      "epoch": 0.24,
+      "learning_rate": 2.1e-05,
+      "loss": 4.2131,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1e-05,
-      "loss": 4.4356,
+      "epoch": 0.24,
+      "learning_rate": 2.12e-05,
+      "loss": 4.1188,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.12e-05,
-      "loss": 4.0137,
+      "epoch": 0.24,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.1356,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.1517,
+      "epoch": 0.25,
+      "learning_rate": 2.16e-05,
+      "loss": 4.1708,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.16e-05,
-      "loss": 4.3351,
+      "epoch": 0.25,
+      "learning_rate": 2.18e-05,
+      "loss": 4.2002,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.18e-05,
-      "loss": 4.1899,
+      "epoch": 0.25,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.1928,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.185,
+      "epoch": 0.25,
+      "learning_rate": 2.22e-05,
+      "loss": 4.1811,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.22e-05,
-      "loss": 4.1203,
+      "epoch": 0.26,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.1331,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.1869,
+      "epoch": 0.26,
+      "learning_rate": 2.26e-05,
+      "loss": 4.184,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.26e-05,
-      "loss": 3.9657,
+      "epoch": 0.26,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.2902,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.2749,
+      "epoch": 0.26,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.1586,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.1479,
+      "epoch": 0.26,
+      "learning_rate": 2.32e-05,
+      "loss": 4.3134,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.32e-05,
-      "loss": 4.2752,
+      "epoch": 0.27,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.2146,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.3304,
+      "epoch": 0.27,
+      "learning_rate": 2.36e-05,
+      "loss": 4.239,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.36e-05,
-      "loss": 4.2427,
+      "epoch": 0.27,
+      "learning_rate": 2.38e-05,
+      "loss": 4.0619,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.38e-05,
-      "loss": 4.2195,
+      "epoch": 0.27,
+      "learning_rate": 2.4e-05,
+      "loss": 4.1774,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4e-05,
-      "loss": 4.0881,
+      "epoch": 0.28,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.1803,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 4.3312,
+      "epoch": 0.28,
+      "learning_rate": 2.44e-05,
+      "loss": 4.1746,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.44e-05,
-      "loss": 4.2518,
+      "epoch": 0.28,
+      "learning_rate": 2.46e-05,
+      "loss": 4.2787,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.46e-05,
-      "loss": 4.2965,
+      "epoch": 0.28,
+      "learning_rate": 2.48e-05,
+      "loss": 4.2883,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.48e-05,
-      "loss": 4.3571,
+      "epoch": 0.28,
+      "learning_rate": 2.5e-05,
+      "loss": 4.2007,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.5e-05,
-      "loss": 4.1767,
+      "epoch": 0.29,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.1726,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.1605,
+      "epoch": 0.29,
+      "learning_rate": 2.54e-05,
+      "loss": 4.0857,
       "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.54e-05,
-      "loss": 4.3313,
+      "epoch": 0.29,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.236,
       "step": 130
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2277,
+      "epoch": 0.29,
+      "learning_rate": 2.58e-05,
+      "loss": 4.1492,
       "step": 131
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.58e-05,
-      "loss": 4.313,
+      "epoch": 0.3,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.2103,
       "step": 132
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.2856,
+      "epoch": 0.3,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.1179,
       "step": 133
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.2828,
+      "epoch": 0.3,
+      "learning_rate": 2.64e-05,
+      "loss": 4.1823,
       "step": 134
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.64e-05,
-      "loss": 4.1482,
+      "epoch": 0.3,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.3251,
       "step": 135
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.2175,
+      "epoch": 0.3,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.1815,
       "step": 136
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.5134,
+      "epoch": 0.31,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2615,
       "step": 137
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.2704,
+      "epoch": 0.31,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.3035,
       "step": 138
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.3241,
+      "epoch": 0.31,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.3086,
       "step": 139
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.105,
+      "epoch": 0.31,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.2835,
       "step": 140
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.4102,
+      "epoch": 0.32,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.3894,
       "step": 141
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.4697,
+      "epoch": 0.32,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.3306,
       "step": 142
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.3797,
+      "epoch": 0.32,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.2895,
       "step": 143
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.2843,
+      "epoch": 0.32,
+      "learning_rate": 2.84e-05,
+      "loss": 4.3263,
       "step": 144
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.84e-05,
-      "loss": 4.421,
+      "epoch": 0.33,
+      "learning_rate": 2.86e-05,
+      "loss": 4.2518,
       "step": 145
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.86e-05,
-      "loss": 4.3733,
+      "epoch": 0.33,
+      "learning_rate": 2.88e-05,
+      "loss": 4.3215,
       "step": 146
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.88e-05,
-      "loss": 4.4609,
+      "epoch": 0.33,
+      "learning_rate": 2.9e-05,
+      "loss": 4.598,
       "step": 147
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9e-05,
-      "loss": 4.246,
+      "epoch": 0.33,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.7036,
       "step": 148
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.5866,
+      "epoch": 0.33,
+      "learning_rate": 2.94e-05,
+      "loss": 4.4583,
       "step": 149
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.94e-05,
-      "loss": 4.2249,
+      "epoch": 0.34,
+      "learning_rate": 2.96e-05,
+      "loss": 4.5292,
       "step": 150
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.96e-05,
-      "loss": 4.3224,
+      "epoch": 0.34,
+      "learning_rate": 2.98e-05,
+      "loss": 4.1027,
       "step": 151
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.98e-05,
-      "loss": 4.2143,
+      "epoch": 0.34,
+      "learning_rate": 3e-05,
+      "loss": 4.1962,
       "step": 152
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3e-05,
-      "loss": 4.2868,
+      "epoch": 0.34,
+      "learning_rate": 3.02e-05,
+      "loss": 4.2274,
       "step": 153
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.02e-05,
-      "loss": 4.1287,
+      "epoch": 0.35,
+      "learning_rate": 3.04e-05,
+      "loss": 4.0786,
       "step": 154
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.04e-05,
-      "loss": 4.0788,
+      "epoch": 0.35,
+      "learning_rate": 3.06e-05,
+      "loss": 4.2101,
       "step": 155
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.06e-05,
-      "loss": 4.172,
+      "epoch": 0.35,
+      "learning_rate": 3.08e-05,
+      "loss": 4.0632,
       "step": 156
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.08e-05,
-      "loss": 4.3126,
+      "epoch": 0.35,
+      "learning_rate": 3.1e-05,
+      "loss": 4.0863,
       "step": 157
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1e-05,
-      "loss": 4.4143,
+      "epoch": 0.35,
+      "learning_rate": 3.12e-05,
+      "loss": 4.184,
       "step": 158
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.12e-05,
-      "loss": 4.4585,
+      "epoch": 0.36,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.1929,
       "step": 159
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.1789,
+      "epoch": 0.36,
+      "learning_rate": 3.16e-05,
+      "loss": 4.2031,
       "step": 160
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.16e-05,
-      "loss": 4.1989,
+      "epoch": 0.36,
+      "learning_rate": 3.18e-05,
+      "loss": 4.2709,
       "step": 161
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.18e-05,
-      "loss": 4.2238,
+      "epoch": 0.36,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.212,
       "step": 162
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.2342,
+      "epoch": 0.37,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.1633,
       "step": 163
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.3552,
+      "epoch": 0.37,
+      "learning_rate": 3.24e-05,
+      "loss": 4.2215,
       "step": 164
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.24e-05,
-      "loss": 4.257,
+      "epoch": 0.37,
+      "learning_rate": 3.26e-05,
+      "loss": 4.0926,
       "step": 165
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.26e-05,
-      "loss": 4.2857,
+      "epoch": 0.37,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.1606,
       "step": 166
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.2484,
+      "epoch": 0.37,
+      "learning_rate": 3.3e-05,
+      "loss": 4.0721,
       "step": 167
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3e-05,
-      "loss": 4.0284,
+      "epoch": 0.38,
+      "learning_rate": 3.32e-05,
+      "loss": 4.1272,
       "step": 168
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.32e-05,
-      "loss": 4.0501,
+      "epoch": 0.38,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.2054,
       "step": 169
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.1841,
+      "epoch": 0.38,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.2183,
       "step": 170
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.4082,
+      "epoch": 0.38,
+      "learning_rate": 3.38e-05,
+      "loss": 4.2042,
       "step": 171
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.38e-05,
-      "loss": 4.363,
+      "epoch": 0.39,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.275,
       "step": 172
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.2477,
+      "epoch": 0.39,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.0782,
       "step": 173
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.0851,
+      "epoch": 0.39,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.2265,
       "step": 174
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.2421,
+      "epoch": 0.39,
+      "learning_rate": 3.46e-05,
+      "loss": 4.2685,
       "step": 175
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.46e-05,
-      "loss": 4.0494,
+      "epoch": 0.39,
+      "learning_rate": 3.48e-05,
+      "loss": 4.1473,
       "step": 176
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.48e-05,
-      "loss": 4.0945,
+      "epoch": 0.4,
+      "learning_rate": 3.5e-05,
+      "loss": 4.0942,
       "step": 177
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.5e-05,
-      "loss": 4.171,
+      "epoch": 0.4,
+      "learning_rate": 3.52e-05,
+      "loss": 4.2087,
       "step": 178
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.52e-05,
-      "loss": 4.2874,
+      "epoch": 0.4,
+      "learning_rate": 3.54e-05,
+      "loss": 4.2813,
       "step": 179
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.54e-05,
-      "loss": 4.2568,
+      "epoch": 0.4,
+      "learning_rate": 3.56e-05,
+      "loss": 4.2491,
       "step": 180
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.56e-05,
-      "loss": 4.2897,
+      "epoch": 0.41,
+      "learning_rate": 3.58e-05,
+      "loss": 4.1937,
       "step": 181
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.58e-05,
-      "loss": 4.1081,
+      "epoch": 0.41,
+      "learning_rate": 3.6e-05,
+      "loss": 4.2598,
       "step": 182
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6e-05,
-      "loss": 4.313,
+      "epoch": 0.41,
+      "learning_rate": 3.62e-05,
+      "loss": 4.2181,
       "step": 183
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.62e-05,
-      "loss": 4.292,
+      "epoch": 0.41,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.236,
       "step": 184
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.2912,
+      "epoch": 0.41,
+      "learning_rate": 3.66e-05,
+      "loss": 4.2193,
       "step": 185
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.66e-05,
-      "loss": 4.151,
-      "step": 186
+      "epoch": 0.42,
+      "learning_rate": 3.68e-05,
+      "loss": 4.0767,
+      "step": 186
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.68e-05,
-      "loss": 4.3744,
+      "epoch": 0.42,
+      "learning_rate": 3.7e-05,
+      "loss": 4.112,
       "step": 187
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7e-05,
-      "loss": 4.0779,
+      "epoch": 0.42,
+      "learning_rate": 3.72e-05,
+      "loss": 4.0809,
       "step": 188
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.72e-05,
-      "loss": 4.2057,
+      "epoch": 0.42,
+      "learning_rate": 3.74e-05,
+      "loss": 4.2022,
       "step": 189
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.74e-05,
-      "loss": 4.3045,
+      "epoch": 0.43,
+      "learning_rate": 3.76e-05,
+      "loss": 4.3143,
       "step": 190
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.76e-05,
-      "loss": 4.3288,
+      "epoch": 0.43,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.2295,
       "step": 191
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.2585,
+      "epoch": 0.43,
+      "learning_rate": 3.8e-05,
+      "loss": 4.1771,
       "step": 192
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8e-05,
-      "loss": 4.491,
+      "epoch": 0.43,
+      "learning_rate": 3.82e-05,
+      "loss": 4.157,
       "step": 193
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.82e-05,
-      "loss": 4.1982,
+      "epoch": 0.43,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.2961,
       "step": 194
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.2791,
+      "epoch": 0.44,
+      "learning_rate": 3.86e-05,
+      "loss": 4.42,
       "step": 195
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.86e-05,
-      "loss": 4.4892,
+      "epoch": 0.44,
+      "learning_rate": 3.88e-05,
+      "loss": 4.3345,
       "step": 196
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.88e-05,
-      "loss": 4.4989,
+      "epoch": 0.44,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.41,
       "step": 197
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.3795,
+      "epoch": 0.44,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.5266,
       "step": 198
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.6247,
+      "epoch": 0.45,
+      "learning_rate": 3.94e-05,
+      "loss": 4.4871,
       "step": 199
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.94e-05,
-      "loss": 4.6662,
+      "epoch": 0.45,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.4572,
       "step": 200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 4.0805,
+      "epoch": 0.45,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.2354,
       "step": 201
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.2222,
+      "epoch": 0.45,
+      "learning_rate": 4e-05,
+      "loss": 4.1817,
       "step": 202
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4e-05,
-      "loss": 4.1281,
+      "epoch": 0.46,
+      "learning_rate": 4.02e-05,
+      "loss": 4.214,
       "step": 203
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.02e-05,
-      "loss": 4.1377,
+      "epoch": 0.46,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.1809,
       "step": 204
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.0093,
+      "epoch": 0.46,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.2551,
       "step": 205
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.0695,
+      "epoch": 0.46,
+      "learning_rate": 4.08e-05,
+      "loss": 4.2098,
       "step": 206
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.08e-05,
-      "loss": 4.1305,
+      "epoch": 0.46,
+      "learning_rate": 4.1e-05,
+      "loss": 4.0845,
       "step": 207
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.1e-05,
-      "loss": 4.0018,
+      "epoch": 0.47,
+      "learning_rate": 4.12e-05,
+      "loss": 4.2135,
       "step": 208
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.12e-05,
-      "loss": 4.2204,
+      "epoch": 0.47,
+      "learning_rate": 4.14e-05,
+      "loss": 4.2722,
       "step": 209
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.14e-05,
-      "loss": 4.1703,
+      "epoch": 0.47,
+      "learning_rate": 4.16e-05,
+      "loss": 4.2086,
       "step": 210
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.16e-05,
-      "loss": 4.0995,
+      "epoch": 0.47,
+      "learning_rate": 4.18e-05,
+      "loss": 4.0768,
       "step": 211
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.18e-05,
-      "loss": 4.2342,
+      "epoch": 0.48,
+      "learning_rate": 4.2e-05,
+      "loss": 4.1328,
       "step": 212
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2e-05,
-      "loss": 4.1738,
+      "epoch": 0.48,
+      "learning_rate": 4.22e-05,
+      "loss": 4.0752,
       "step": 213
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.22e-05,
-      "loss": 4.2382,
+      "epoch": 0.48,
+      "learning_rate": 4.24e-05,
+      "loss": 4.1268,
       "step": 214
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.24e-05,
-      "loss": 4.1038,
+      "epoch": 0.48,
+      "learning_rate": 4.26e-05,
+      "loss": 4.1781,
       "step": 215
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.26e-05,
-      "loss": 4.2598,
+      "epoch": 0.48,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.163,
       "step": 216
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.1992,
+      "epoch": 0.49,
+      "learning_rate": 4.3e-05,
+      "loss": 4.193,
       "step": 217
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.3e-05,
-      "loss": 4.1448,
+      "epoch": 0.49,
+      "learning_rate": 4.32e-05,
+      "loss": 4.1417,
       "step": 218
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.32e-05,
-      "loss": 4.2357,
+      "epoch": 0.49,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.1006,
       "step": 219
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.1894,
+      "epoch": 0.49,
+      "learning_rate": 4.36e-05,
+      "loss": 4.183,
       "step": 220
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.36e-05,
-      "loss": 4.1205,
+      "epoch": 0.5,
+      "learning_rate": 4.38e-05,
+      "loss": 4.1586,
       "step": 221
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.38e-05,
-      "loss": 4.2463,
+      "epoch": 0.5,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.1583,
       "step": 222
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.2651,
+      "epoch": 0.5,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.1894,
       "step": 223
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.1997,
+      "epoch": 0.5,
+      "learning_rate": 4.44e-05,
+      "loss": 4.1255,
       "step": 224
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.44e-05,
-      "loss": 4.2946,
+      "epoch": 0.5,
+      "learning_rate": 4.46e-05,
+      "loss": 4.2794,
       "step": 225
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.46e-05,
-      "loss": 4.4428,
+      "epoch": 0.51,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 4.1867,
       "step": 226
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.2206,
+      "epoch": 0.51,
+      "learning_rate": 4.5e-05,
+      "loss": 4.2804,
       "step": 227
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5e-05,
-      "loss": 4.1257,
+      "epoch": 0.51,
+      "learning_rate": 4.52e-05,
+      "loss": 4.2764,
       "step": 228
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.52e-05,
-      "loss": 4.1379,
+      "epoch": 0.51,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.1211,
       "step": 229
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.1872,
+      "epoch": 0.52,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.2534,
       "step": 230
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.2277,
+      "epoch": 0.52,
+      "learning_rate": 4.58e-05,
+      "loss": 4.2681,
       "step": 231
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.58e-05,
-      "loss": 4.3086,
+      "epoch": 0.52,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.0795,
       "step": 232
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 4.1558,
+      "epoch": 0.52,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.1336,
       "step": 233
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.2961,
+      "epoch": 0.52,
+      "learning_rate": 4.64e-05,
+      "loss": 4.1979,
       "step": 234
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.64e-05,
-      "loss": 4.3365,
+      "epoch": 0.53,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.2561,
       "step": 235
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 4.1922,
+      "epoch": 0.53,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.2117,
       "step": 236
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.1784,
+      "epoch": 0.53,
+      "learning_rate": 4.7e-05,
+      "loss": 4.192,
       "step": 237
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.7e-05,
-      "loss": 4.4414,
+      "epoch": 0.53,
+      "learning_rate": 4.72e-05,
+      "loss": 4.1534,
       "step": 238
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.72e-05,
-      "loss": 4.2069,
+      "epoch": 0.54,
+      "learning_rate": 4.74e-05,
+      "loss": 4.1088,
       "step": 239
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.74e-05,
-      "loss": 4.3952,
+      "epoch": 0.54,
+      "learning_rate": 4.76e-05,
+      "loss": 4.1981,
       "step": 240
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.76e-05,
-      "loss": 4.295,
+      "epoch": 0.54,
+      "learning_rate": 4.78e-05,
+      "loss": 4.2105,
       "step": 241
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.78e-05,
-      "loss": 4.4594,
+      "epoch": 0.54,
+      "learning_rate": 4.8e-05,
+      "loss": 4.2749,
       "step": 242
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.8e-05,
-      "loss": 4.3772,
+      "epoch": 0.54,
+      "learning_rate": 4.82e-05,
+      "loss": 4.4264,
       "step": 243
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.82e-05,
-      "loss": 4.2769,
+      "epoch": 0.55,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.2585,
       "step": 244
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.4663,
+      "epoch": 0.55,
+      "learning_rate": 4.86e-05,
+      "loss": 4.2482,
       "step": 245
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.86e-05,
-      "loss": 4.3733,
+      "epoch": 0.55,
+      "learning_rate": 4.88e-05,
+      "loss": 4.1165,
       "step": 246
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.88e-05,
-      "loss": 4.7879,
+      "epoch": 0.55,
+      "learning_rate": 4.9e-05,
+      "loss": 4.0864,
       "step": 247
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.9e-05,
-      "loss": 4.7509,
+      "epoch": 0.56,
+      "learning_rate": 4.92e-05,
+      "loss": 4.5551,
       "step": 248
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.92e-05,
-      "loss": 4.7604,
+      "epoch": 0.56,
+      "learning_rate": 4.94e-05,
+      "loss": 4.5483,
       "step": 249
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.94e-05,
-      "loss": 4.1676,
+      "epoch": 0.56,
+      "learning_rate": 4.96e-05,
+      "loss": 4.5265,
       "step": 250
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.96e-05,
-      "loss": 4.4618,
+      "epoch": 0.56,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.2203,
       "step": 251
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.1168,
+      "epoch": 0.57,
+      "learning_rate": 5e-05,
+      "loss": 4.2102,
       "step": 252
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5e-05,
-      "loss": 4.1508,
+      "epoch": 0.57,
+      "learning_rate": 5.02e-05,
+      "loss": 4.2521,
       "step": 253
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5.02e-05,
-      "loss": 4.2563,
+      "epoch": 0.57,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.1037,
       "step": 254
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.1567,
+      "epoch": 0.57,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.1975,
       "step": 255
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.1473,
+      "epoch": 0.57,
+      "learning_rate": 5.08e-05,
+      "loss": 4.1548,
       "step": 256
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.08e-05,
-      "loss": 4.2933,
+      "epoch": 0.58,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.1562,
       "step": 257
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.1407,
+      "epoch": 0.58,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 4.1191,
       "step": 258
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.1175,
+      "epoch": 0.58,
+      "learning_rate": 5.14e-05,
+      "loss": 4.1629,
       "step": 259
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.14e-05,
-      "loss": 4.3068,
+      "epoch": 0.58,
+      "learning_rate": 5.16e-05,
+      "loss": 4.1329,
       "step": 260
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.16e-05,
-      "loss": 4.2729,
+      "epoch": 0.59,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.2366,
       "step": 261
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.1282,
+      "epoch": 0.59,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.1222,
       "step": 262
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.1259,
+      "epoch": 0.59,
+      "learning_rate": 5.22e-05,
+      "loss": 4.2949,
       "step": 263
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.22e-05,
-      "loss": 4.1419,
+      "epoch": 0.59,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 4.1441,
       "step": 264
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.2539,
+      "epoch": 0.59,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.2206,
       "step": 265
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.2848,
+      "epoch": 0.6,
+      "learning_rate": 5.28e-05,
+      "loss": 4.1359,
       "step": 266
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.28e-05,
-      "loss": 4.1218,
+      "epoch": 0.6,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.18,
       "step": 267
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 4.4412,
+      "epoch": 0.6,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.0921,
       "step": 268
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.1862,
+      "epoch": 0.6,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 4.2223,
       "step": 269
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.3737,
+      "epoch": 0.61,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 4.1064,
       "step": 270
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 3.9655,
+      "epoch": 0.61,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 4.0658,
       "step": 271
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 4.1543,
+      "epoch": 0.61,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.2146,
       "step": 272
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.1888,
+      "epoch": 0.61,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 4.0509,
       "step": 273
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 4.0681,
+      "epoch": 0.61,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 4.1473,
       "step": 274
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 4.3354,
+      "epoch": 0.62,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 4.2196,
       "step": 275
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 4.2964,
+      "epoch": 0.62,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 4.1378,
       "step": 276
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.2131,
+      "epoch": 0.62,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.0931,
       "step": 277
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 4.1059,
+      "epoch": 0.62,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 4.1675,
       "step": 278
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 4.1404,
+      "epoch": 0.63,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 4.1488,
       "step": 279
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.3474,
+      "epoch": 0.63,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 4.2942,
       "step": 280
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 4.0881,
+      "epoch": 0.63,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 4.1823,
       "step": 281
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 4.2215,
+      "epoch": 0.63,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.0571,
       "step": 282
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 4.0005,
+      "epoch": 0.63,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 4.0695,
       "step": 283
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 4.224,
+      "epoch": 0.64,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.2051,
       "step": 284
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.2709,
+      "epoch": 0.64,
+      "learning_rate": 5.66e-05,
+      "loss": 4.2795,
       "step": 285
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.66e-05,
-      "loss": 4.164,
+      "epoch": 0.64,
+      "learning_rate": 5.68e-05,
+      "loss": 4.1569,
       "step": 286
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.68e-05,
-      "loss": 4.3511,
+      "epoch": 0.64,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.2607,
       "step": 287
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.2992,
+      "epoch": 0.65,
+      "learning_rate": 5.72e-05,
+      "loss": 4.2057,
       "step": 288
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.72e-05,
-      "loss": 4.3478,
+      "epoch": 0.65,
+      "learning_rate": 5.74e-05,
+      "loss": 4.1146,
       "step": 289
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.74e-05,
-      "loss": 4.3043,
+      "epoch": 0.65,
+      "learning_rate": 5.76e-05,
+      "loss": 4.1695,
       "step": 290
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.76e-05,
-      "loss": 4.3565,
+      "epoch": 0.65,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 4.4908,
       "step": 291
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.2118,
+      "epoch": 0.65,
+      "learning_rate": 5.8e-05,
+      "loss": 4.2949,
       "step": 292
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8e-05,
-      "loss": 4.5298,
+      "epoch": 0.66,
+      "learning_rate": 5.82e-05,
+      "loss": 4.3671,
       "step": 293
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.82e-05,
-      "loss": 4.2882,
+      "epoch": 0.66,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 4.307,
       "step": 294
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 4.1672,
+      "epoch": 0.66,
+      "learning_rate": 5.86e-05,
+      "loss": 4.3258,
       "step": 295
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.86e-05,
-      "loss": 4.5489,
+      "epoch": 0.66,
+      "learning_rate": 5.88e-05,
+      "loss": 4.3597,
       "step": 296
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.88e-05,
-      "loss": 4.5708,
+      "epoch": 0.67,
+      "learning_rate": 5.9e-05,
+      "loss": 4.2197,
       "step": 297
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.9e-05,
-      "loss": 4.3176,
+      "epoch": 0.67,
+      "learning_rate": 5.92e-05,
+      "loss": 4.6226,
       "step": 298
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.92e-05,
-      "loss": 4.7349,
+      "epoch": 0.67,
+      "learning_rate": 5.94e-05,
+      "loss": 4.2705,
       "step": 299
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.94e-05,
-      "loss": 4.7477,
+      "epoch": 0.67,
+      "learning_rate": 5.96e-05,
+      "loss": 4.714,
       "step": 300
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.96e-05,
-      "loss": 4.1141,
+      "epoch": 0.67,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 4.2312,
       "step": 301
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 4.255,
+      "epoch": 0.68,
+      "learning_rate": 6e-05,
+      "loss": 4.2223,
       "step": 302
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6e-05,
-      "loss": 4.2093,
+      "epoch": 0.68,
+      "learning_rate": 6.02e-05,
+      "loss": 4.1857,
       "step": 303
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.02e-05,
-      "loss": 4.2512,
+      "epoch": 0.68,
+      "learning_rate": 6.04e-05,
+      "loss": 4.215,
       "step": 304
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.04e-05,
-      "loss": 4.0745,
+      "epoch": 0.68,
+      "learning_rate": 6.06e-05,
+      "loss": 4.1659,
       "step": 305
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.06e-05,
-      "loss": 4.143,
+      "epoch": 0.69,
+      "learning_rate": 6.08e-05,
+      "loss": 4.1269,
       "step": 306
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.08e-05,
-      "loss": 4.1732,
+      "epoch": 0.69,
+      "learning_rate": 6.1e-05,
+      "loss": 4.1849,
       "step": 307
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.1e-05,
-      "loss": 4.3464,
+      "epoch": 0.69,
+      "learning_rate": 6.12e-05,
+      "loss": 4.1578,
       "step": 308
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.12e-05,
-      "loss": 4.2193,
+      "epoch": 0.69,
+      "learning_rate": 6.14e-05,
+      "loss": 4.1545,
       "step": 309
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.14e-05,
-      "loss": 4.3655,
+      "epoch": 0.7,
+      "learning_rate": 6.16e-05,
+      "loss": 4.2542,
       "step": 310
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.16e-05,
-      "loss": 4.4293,
+      "epoch": 0.7,
+      "learning_rate": 6.18e-05,
+      "loss": 3.9987,
       "step": 311
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.18e-05,
-      "loss": 4.3158,
+      "epoch": 0.7,
+      "learning_rate": 6.2e-05,
+      "loss": 4.2024,
       "step": 312
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.2e-05,
-      "loss": 4.2305,
+      "epoch": 0.7,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 4.2768,
       "step": 313
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 4.2545,
+      "epoch": 0.7,
+      "learning_rate": 6.24e-05,
+      "loss": 4.128,
       "step": 314
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.24e-05,
-      "loss": 4.092,
+      "epoch": 0.71,
+      "learning_rate": 6.26e-05,
+      "loss": 4.1512,
       "step": 315
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.26e-05,
-      "loss": 4.3718,
+      "epoch": 0.71,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 4.1232,
       "step": 316
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 4.1923,
+      "epoch": 0.71,
+      "learning_rate": 6.3e-05,
+      "loss": 4.1407,
       "step": 317
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.3e-05,
-      "loss": 4.1044,
+      "epoch": 0.71,
+      "learning_rate": 6.32e-05,
+      "loss": 4.1325,
       "step": 318
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.32e-05,
-      "loss": 4.2423,
+      "epoch": 0.72,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 4.1598,
       "step": 319
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 4.1932,
+      "epoch": 0.72,
+      "learning_rate": 6.36e-05,
+      "loss": 4.2247,
       "step": 320
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.36e-05,
-      "loss": 4.3054,
+      "epoch": 0.72,
+      "learning_rate": 6.38e-05,
+      "loss": 4.1314,
       "step": 321
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.38e-05,
-      "loss": 4.204,
+      "epoch": 0.72,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 3.981,
       "step": 322
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 4.6242,
+      "epoch": 0.72,
+      "learning_rate": 6.42e-05,
+      "loss": 4.1958,
       "step": 323
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.42e-05,
-      "loss": 4.4789,
+      "epoch": 0.73,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.0383,
       "step": 324
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 4.4935,
+      "epoch": 0.73,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 4.2174,
       "step": 325
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 4.3337,
+      "epoch": 0.73,
+      "learning_rate": 6.48e-05,
+      "loss": 4.1948,
       "step": 326
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.48e-05,
-      "loss": 4.1572,
+      "epoch": 0.73,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.1112,
       "step": 327
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 4.1932,
+      "epoch": 0.74,
+      "learning_rate": 6.52e-05,
+      "loss": 4.0492,
       "step": 328
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.52e-05,
-      "loss": 4.4071,
+      "epoch": 0.74,
+      "learning_rate": 6.54e-05,
+      "loss": 4.079,
       "step": 329
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.54e-05,
-      "loss": 4.4865,
+      "epoch": 0.74,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 4.2257,
       "step": 330
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 4.1701,
+      "epoch": 0.74,
+      "learning_rate": 6.58e-05,
+      "loss": 4.2159,
       "step": 331
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.58e-05,
-      "loss": 4.2422,
+      "epoch": 0.74,
+      "learning_rate": 6.6e-05,
+      "loss": 4.0892,
       "step": 332
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.6e-05,
-      "loss": 4.2503,
+      "epoch": 0.75,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 4.2605,
       "step": 333
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 4.2858,
+      "epoch": 0.75,
+      "learning_rate": 6.64e-05,
+      "loss": 4.2019,
       "step": 334
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.64e-05,
-      "loss": 4.3979,
+      "epoch": 0.75,
+      "learning_rate": 6.66e-05,
+      "loss": 4.3192,
       "step": 335
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.66e-05,
-      "loss": 4.1882,
+      "epoch": 0.75,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 4.2142,
       "step": 336
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 4.1267,
+      "epoch": 0.76,
+      "learning_rate": 6.7e-05,
+      "loss": 4.1008,
       "step": 337
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.7e-05,
-      "loss": 4.3292,
+      "epoch": 0.76,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 4.1991,
       "step": 338
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 4.2356,
+      "epoch": 0.76,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 4.2477,
       "step": 339
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 4.2885,
+      "epoch": 0.76,
+      "learning_rate": 6.76e-05,
+      "loss": 4.277,
       "step": 340
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.76e-05,
-      "loss": 4.5698,
+      "epoch": 0.76,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 4.1477,
       "step": 341
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 4.4765,
+      "epoch": 0.77,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.3392,
       "step": 342
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 4.2685,
+      "epoch": 0.77,
+      "learning_rate": 6.82e-05,
+      "loss": 4.2372,
       "step": 343
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.82e-05,
-      "loss": 4.4023,
+      "epoch": 0.77,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 4.355,
       "step": 344
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 4.401,
+      "epoch": 0.77,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 4.0971,
       "step": 345
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 4.6416,
+      "epoch": 0.78,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 4.2799,
       "step": 346
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 4.6245,
+      "epoch": 0.78,
+      "learning_rate": 6.9e-05,
+      "loss": 4.3116,
       "step": 347
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.9e-05,
-      "loss": 4.5174,
+      "epoch": 0.78,
+      "learning_rate": 6.92e-05,
+      "loss": 4.2765,
       "step": 348
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.92e-05,
-      "loss": 4.5499,
+      "epoch": 0.78,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 4.3885,
       "step": 349
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 4.4764,
+      "epoch": 0.78,
+      "learning_rate": 6.96e-05,
+      "loss": 4.3474,
       "step": 350
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.96e-05,
-      "loss": 4.4091,
+      "epoch": 0.79,
+      "learning_rate": 6.98e-05,
+      "loss": 4.2769,
       "step": 351
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.98e-05,
-      "loss": 4.2577,
+      "epoch": 0.79,
+      "learning_rate": 7e-05,
+      "loss": 4.1673,
       "step": 352
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7e-05,
-      "loss": 4.2241,
+      "epoch": 0.79,
+      "learning_rate": 7.02e-05,
+      "loss": 4.1257,
       "step": 353
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.02e-05,
-      "loss": 4.2012,
+      "epoch": 0.79,
+      "learning_rate": 7.04e-05,
+      "loss": 4.0416,
       "step": 354
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.04e-05,
-      "loss": 4.4662,
+      "epoch": 0.8,
+      "learning_rate": 7.06e-05,
+      "loss": 4.1925,
       "step": 355
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.06e-05,
-      "loss": 4.1124,
+      "epoch": 0.8,
+      "learning_rate": 7.08e-05,
+      "loss": 4.3139,
       "step": 356
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.08e-05,
-      "loss": 4.2003,
+      "epoch": 0.8,
+      "learning_rate": 7.1e-05,
+      "loss": 4.1854,
       "step": 357
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.1e-05,
-      "loss": 4.2004,
+      "epoch": 0.8,
+      "learning_rate": 7.12e-05,
+      "loss": 4.2114,
       "step": 358
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.12e-05,
-      "loss": 4.2059,
+      "epoch": 0.8,
+      "learning_rate": 7.14e-05,
+      "loss": 4.0914,
       "step": 359
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.14e-05,
-      "loss": 4.3448,
+      "epoch": 0.81,
+      "learning_rate": 7.16e-05,
+      "loss": 4.1907,
       "step": 360
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.16e-05,
-      "loss": 4.243,
+      "epoch": 0.81,
+      "learning_rate": 7.18e-05,
+      "loss": 4.0951,
       "step": 361
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.18e-05,
-      "loss": 4.2397,
+      "epoch": 0.81,
+      "learning_rate": 7.2e-05,
+      "loss": 4.1182,
       "step": 362
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.2e-05,
-      "loss": 4.1738,
+      "epoch": 0.81,
+      "learning_rate": 7.22e-05,
+      "loss": 4.0534,
       "step": 363
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.22e-05,
-      "loss": 4.4315,
+      "epoch": 0.82,
+      "learning_rate": 7.24e-05,
+      "loss": 4.1806,
       "step": 364
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.24e-05,
-      "loss": 4.0795,
+      "epoch": 0.82,
+      "learning_rate": 7.26e-05,
+      "loss": 4.2851,
       "step": 365
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.26e-05,
-      "loss": 4.2753,
+      "epoch": 0.82,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 4.1213,
       "step": 366
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 4.181,
+      "epoch": 0.82,
+      "learning_rate": 7.3e-05,
+      "loss": 4.205,
       "step": 367
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.3e-05,
-      "loss": 4.239,
+      "epoch": 0.83,
+      "learning_rate": 7.32e-05,
+      "loss": 4.1709,
       "step": 368
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.32e-05,
-      "loss": 4.3883,
+      "epoch": 0.83,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 4.064,
       "step": 369
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 4.2789,
+      "epoch": 0.83,
+      "learning_rate": 7.36e-05,
+      "loss": 4.0942,
       "step": 370
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.36e-05,
-      "loss": 4.1056,
+      "epoch": 0.83,
+      "learning_rate": 7.38e-05,
+      "loss": 4.0847,
       "step": 371
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.38e-05,
-      "loss": 4.4694,
+      "epoch": 0.83,
+      "learning_rate": 7.4e-05,
+      "loss": 4.1675,
       "step": 372
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.4e-05,
-      "loss": 4.2561,
+      "epoch": 0.84,
+      "learning_rate": 7.42e-05,
+      "loss": 4.1618,
       "step": 373
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.42e-05,
-      "loss": 4.398,
+      "epoch": 0.84,
+      "learning_rate": 7.44e-05,
+      "loss": 3.9479,
       "step": 374
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.44e-05,
-      "loss": 4.1645,
+      "epoch": 0.84,
+      "learning_rate": 7.46e-05,
+      "loss": 4.1896,
       "step": 375
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.46e-05,
-      "loss": 4.2328,
+      "epoch": 0.84,
+      "learning_rate": 7.48e-05,
+      "loss": 4.0284,
       "step": 376
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.48e-05,
-      "loss": 4.2445,
+      "epoch": 0.85,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.1114,
       "step": 377
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.3664,
+      "epoch": 0.85,
+      "learning_rate": 7.52e-05,
+      "loss": 4.2131,
       "step": 378
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.52e-05,
-      "loss": 4.2296,
+      "epoch": 0.85,
+      "learning_rate": 7.54e-05,
+      "loss": 4.1098,
       "step": 379
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.54e-05,
-      "loss": 4.4382,
+      "epoch": 0.85,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 4.0568,
       "step": 380
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 4.3286,
+      "epoch": 0.85,
+      "learning_rate": 7.58e-05,
+      "loss": 4.1287,
       "step": 381
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.58e-05,
-      "loss": 4.2781,
+      "epoch": 0.86,
+      "learning_rate": 7.6e-05,
+      "loss": 4.2378,
       "step": 382
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.6e-05,
-      "loss": 4.4275,
+      "epoch": 0.86,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 4.1744,
       "step": 383
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 4.2383,
+      "epoch": 0.86,
+      "learning_rate": 7.64e-05,
+      "loss": 4.0495,
       "step": 384
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.64e-05,
-      "loss": 4.2161,
+      "epoch": 0.86,
+      "learning_rate": 7.66e-05,
+      "loss": 4.2233,
       "step": 385
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.66e-05,
-      "loss": 4.2606,
+      "epoch": 0.87,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 4.0447,
       "step": 386
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 4.2192,
+      "epoch": 0.87,
+      "learning_rate": 7.7e-05,
+      "loss": 4.1888,
       "step": 387
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.7e-05,
-      "loss": 4.222,
+      "epoch": 0.87,
+      "learning_rate": 7.72e-05,
+      "loss": 4.2306,
       "step": 388
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.72e-05,
-      "loss": 4.2535,
+      "epoch": 0.87,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 4.1236,
       "step": 389
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 4.323,
+      "epoch": 0.87,
+      "learning_rate": 7.76e-05,
+      "loss": 4.1625,
       "step": 390
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.76e-05,
-      "loss": 4.3979,
+      "epoch": 0.88,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 4.2749,
       "step": 391
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 4.1583,
+      "epoch": 0.88,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 4.0711,
       "step": 392
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 4.4448,
+      "epoch": 0.88,
+      "learning_rate": 7.82e-05,
+      "loss": 4.0423,
       "step": 393
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.82e-05,
-      "loss": 4.4023,
+      "epoch": 0.88,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 4.1891,
       "step": 394
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 4.6777,
+      "epoch": 0.89,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 4.353,
       "step": 395
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 4.2592,
+      "epoch": 0.89,
+      "learning_rate": 7.88e-05,
+      "loss": 4.2261,
       "step": 396
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.88e-05,
-      "loss": 4.2254,
+      "epoch": 0.89,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 4.5605,
       "step": 397
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 4.6933,
+      "epoch": 0.89,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 4.3021,
       "step": 398
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 4.4564,
+      "epoch": 0.89,
+      "learning_rate": 7.94e-05,
+      "loss": 4.6879,
       "step": 399
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.94e-05,
-      "loss": 4.8916,
+      "epoch": 0.9,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 4.4051,
       "step": 400
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 4.4665,
+      "epoch": 0.9,
+      "learning_rate": 7.98e-05,
+      "loss": 4.2403,
       "step": 401
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.98e-05,
-      "loss": 4.3327,
+      "epoch": 0.9,
+      "learning_rate": 8e-05,
+      "loss": 4.1457,
       "step": 402
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8e-05,
-      "loss": 4.4033,
+      "epoch": 0.9,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 4.2249,
       "step": 403
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 4.3885,
+      "epoch": 0.91,
+      "learning_rate": 8.04e-05,
+      "loss": 4.1852,
       "step": 404
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.04e-05,
-      "loss": 4.4113,
+      "epoch": 0.91,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 4.0872,
       "step": 405
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 4.3084,
+      "epoch": 0.91,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 4.2534,
       "step": 406
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 4.2218,
+      "epoch": 0.91,
+      "learning_rate": 8.1e-05,
+      "loss": 4.128,
       "step": 407
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.1e-05,
-      "loss": 4.202,
+      "epoch": 0.91,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 4.2292,
       "step": 408
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 4.4395,
+      "epoch": 0.92,
+      "learning_rate": 8.14e-05,
+      "loss": 4.1299,
       "step": 409
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.14e-05,
-      "loss": 4.3359,
+      "epoch": 0.92,
+      "learning_rate": 8.16e-05,
+      "loss": 4.0981,
       "step": 410
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.16e-05,
-      "loss": 4.3585,
+      "epoch": 0.92,
+      "learning_rate": 8.18e-05,
+      "loss": 4.353,
       "step": 411
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.18e-05,
-      "loss": 4.2204,
+      "epoch": 0.92,
+      "learning_rate": 8.2e-05,
+      "loss": 4.0653,
       "step": 412
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.2e-05,
-      "loss": 4.4704,
+      "epoch": 0.93,
+      "learning_rate": 8.22e-05,
+      "loss": 4.0245,
       "step": 413
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.22e-05,
-      "loss": 4.0948,
+      "epoch": 0.93,
+      "learning_rate": 8.24e-05,
+      "loss": 4.2077,
       "step": 414
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.24e-05,
-      "loss": 4.2758,
+      "epoch": 0.93,
+      "learning_rate": 8.26e-05,
+      "loss": 4.3732,
       "step": 415
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.26e-05,
-      "loss": 4.3346,
+      "epoch": 0.93,
+      "learning_rate": 8.28e-05,
+      "loss": 4.1227,
       "step": 416
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.28e-05,
-      "loss": 4.1562,
+      "epoch": 0.93,
+      "learning_rate": 8.3e-05,
+      "loss": 4.1922,
       "step": 417
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.3e-05,
-      "loss": 4.3247,
+      "epoch": 0.94,
+      "learning_rate": 8.32e-05,
+      "loss": 4.0332,
       "step": 418
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.32e-05,
-      "loss": 4.2338,
+      "epoch": 0.94,
+      "learning_rate": 8.34e-05,
+      "loss": 4.1886,
       "step": 419
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.34e-05,
-      "loss": 4.268,
+      "epoch": 0.94,
+      "learning_rate": 8.36e-05,
+      "loss": 4.0436,
       "step": 420
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.36e-05,
-      "loss": 4.242,
+      "epoch": 0.94,
+      "learning_rate": 8.38e-05,
+      "loss": 4.2622,
       "step": 421
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.38e-05,
-      "loss": 4.4886,
+      "epoch": 0.95,
+      "learning_rate": 8.4e-05,
+      "loss": 4.0481,
       "step": 422
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.4e-05,
-      "loss": 4.4959,
+      "epoch": 0.95,
+      "learning_rate": 8.42e-05,
+      "loss": 4.1584,
       "step": 423
     },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.42e-05,
-      "loss": 4.4986,
-      "step": 424
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.44e-05,
-      "loss": 4.4402,
-      "step": 425
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.46e-05,
-      "loss": 4.4681,
-      "step": 426
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.48e-05,
-      "loss": 4.1911,
-      "step": 427
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.5e-05,
-      "loss": 4.4122,
-      "step": 428
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.52e-05,
-      "loss": 4.2988,
-      "step": 429
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.54e-05,
-      "loss": 4.3522,
-      "step": 430
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 4.5413,
-      "step": 431
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.58e-05,
-      "loss": 4.3583,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.6e-05,
-      "loss": 4.0939,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 4.6722,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.64e-05,
-      "loss": 4.2455,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.66e-05,
-      "loss": 4.4837,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 4.5514,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.7e-05,
-      "loss": 4.3723,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.72e-05,
-      "loss": 4.2979,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 4.3634,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.76e-05,
-      "loss": 4.6395,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.78e-05,
-      "loss": 4.2519,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 4.2913,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-05,
-      "loss": 4.7257,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 4.3425,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.86e-05,
-      "loss": 4.1172,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.88e-05,
-      "loss": 4.3683,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 4.5756,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.92e-05,
-      "loss": 4.9532,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.94e-05,
-      "loss": 5.6362,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 4.3291,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.98e-05,
-      "loss": 4.4704,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9e-05,
-      "loss": 4.5678,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 4.2346,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.04e-05,
-      "loss": 4.5092,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.06e-05,
-      "loss": 4.3366,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 4.3663,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.1e-05,
-      "loss": 4.3834,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 4.3779,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 4.5347,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.16e-05,
-      "loss": 4.2667,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 4.4051,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 4.3178,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.22e-05,
-      "loss": 4.1682,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 4.2923,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.260000000000001e-05,
-      "loss": 4.6037,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.28e-05,
-      "loss": 4.2776,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 4.2357,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.320000000000002e-05,
-      "loss": 4.3306,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.340000000000001e-05,
-      "loss": 4.2949,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.360000000000001e-05,
-      "loss": 4.1422,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.38e-05,
-      "loss": 4.2664,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.4e-05,
-      "loss": 5.0999,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-05,
-      "loss": 4.4737,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.44e-05,
-      "loss": 4.3051,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.46e-05,
-      "loss": 4.6362,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.48e-05,
-      "loss": 4.5125,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.5e-05,
-      "loss": 4.2002,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.52e-05,
-      "loss": 4.1203,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-05,
-      "loss": 4.3217,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.56e-05,
-      "loss": 4.2459,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.58e-05,
-      "loss": 4.1917,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.6e-05,
-      "loss": 4.3971,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.620000000000001e-05,
-      "loss": 4.3083,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.64e-05,
-      "loss": 4.4234,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.66e-05,
-      "loss": 4.198,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.680000000000001e-05,
-      "loss": 4.3167,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.7e-05,
-      "loss": 4.1635,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-05,
-      "loss": 4.3927,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.74e-05,
-      "loss": 4.3937,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.76e-05,
-      "loss": 4.2244,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.78e-05,
-      "loss": 4.5039,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.8e-05,
-      "loss": 4.8541,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.82e-05,
-      "loss": 4.2893,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.84e-05,
-      "loss": 4.6025,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.86e-05,
-      "loss": 4.1377,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.88e-05,
-      "loss": 4.094,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 4.8325,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.92e-05,
-      "loss": 4.6663,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.94e-05,
-      "loss": 4.1605,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.537853240966797,
-      "eval_runtime": 894.8393,
-      "eval_samples_per_second": 2.952,
-      "eval_steps_per_second": 0.37,
-      "eval_wer": 1.929650932169774,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 4.6864,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.98e-05,
-      "loss": 4.2069,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0001,
-      "loss": 4.6738,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.974489795918368e-05,
-      "loss": 4.4791,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.948979591836736e-05,
-      "loss": 4.1978,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.923469387755102e-05,
-      "loss": 4.3373,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.897959183673469e-05,
-      "loss": 4.4823,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.872448979591837e-05,
-      "loss": 4.6696,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.846938775510204e-05,
-      "loss": 4.3015,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.821428571428572e-05,
-      "loss": 4.6368,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.79591836734694e-05,
-      "loss": 4.3094,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.770408163265307e-05,
-      "loss": 4.3088,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.744897959183674e-05,
-      "loss": 5.0756,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.719387755102042e-05,
-      "loss": 4.4334,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.693877551020408e-05,
-      "loss": 4.5505,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.668367346938776e-05,
-      "loss": 4.2511,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.642857142857143e-05,
-      "loss": 4.4093,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.617346938775511e-05,
-      "loss": 4.3096,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.591836734693878e-05,
-      "loss": 4.5489,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.566326530612246e-05,
-      "loss": 4.4747,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.540816326530613e-05,
-      "loss": 4.274,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.515306122448981e-05,
-      "loss": 4.2253,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.489795918367348e-05,
-      "loss": 4.2597,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.464285714285715e-05,
-      "loss": 4.5888,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.438775510204082e-05,
-      "loss": 4.4988,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.41326530612245e-05,
-      "loss": 4.2755,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.387755102040817e-05,
-      "loss": 4.4353,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.362244897959183e-05,
-      "loss": 4.3746,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.336734693877551e-05,
-      "loss": 4.3393,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.311224489795918e-05,
-      "loss": 4.5366,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.285714285714286e-05,
-      "loss": 4.582,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.260204081632653e-05,
-      "loss": 4.1169,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.234693877551021e-05,
-      "loss": 4.3229,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.209183673469388e-05,
-      "loss": 4.2116,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.183673469387756e-05,
-      "loss": 4.5244,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.158163265306124e-05,
-      "loss": 4.4066,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.13265306122449e-05,
-      "loss": 4.5668,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.107142857142857e-05,
-      "loss": 4.4726,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.081632653061225e-05,
-      "loss": 4.2823,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.056122448979592e-05,
-      "loss": 4.3757,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.030612244897958e-05,
-      "loss": 4.3945,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.005102040816327e-05,
-      "loss": 4.4137,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.979591836734695e-05,
-      "loss": 4.5504,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.954081632653062e-05,
-      "loss": 4.6919,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.92857142857143e-05,
-      "loss": 4.677,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.903061224489796e-05,
-      "loss": 4.5456,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.877551020408164e-05,
-      "loss": 5.0684,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.852040816326531e-05,
-      "loss": 6.0851,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.826530612244899e-05,
-      "loss": 4.369,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.801020408163265e-05,
-      "loss": 5.2507,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.775510204081632e-05,
-      "loss": 4.4075,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.75e-05,
-      "loss": 4.5339,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.724489795918367e-05,
-      "loss": 4.5553,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.698979591836735e-05,
-      "loss": 4.2438,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.673469387755102e-05,
-      "loss": 4.3284,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.64795918367347e-05,
-      "loss": 4.5701,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.622448979591838e-05,
-      "loss": 4.2063,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.596938775510205e-05,
-      "loss": 4.1729,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.571428571428571e-05,
-      "loss": 4.3359,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.545918367346939e-05,
-      "loss": 4.2974,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.520408163265306e-05,
-      "loss": 4.4944,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.494897959183674e-05,
-      "loss": 4.2742,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.469387755102041e-05,
-      "loss": 4.3104,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.443877551020409e-05,
-      "loss": 4.2998,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.418367346938776e-05,
-      "loss": 4.3185,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.392857142857144e-05,
-      "loss": 4.4354,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.367346938775511e-05,
-      "loss": 4.3363,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.341836734693878e-05,
-      "loss": 4.2593,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.316326530612245e-05,
-      "loss": 4.3758,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.290816326530613e-05,
-      "loss": 4.3491,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.26530612244898e-05,
-      "loss": 4.2382,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.239795918367348e-05,
-      "loss": 4.3785,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.214285714285714e-05,
-      "loss": 4.2177,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.188775510204081e-05,
-      "loss": 4.1685,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.163265306122449e-05,
-      "loss": 4.6721,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.137755102040817e-05,
-      "loss": 4.2446,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.112244897959184e-05,
-      "loss": 4.3527,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.086734693877552e-05,
-      "loss": 4.585,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.061224489795919e-05,
-      "loss": 3.9455,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.035714285714287e-05,
-      "loss": 4.9334,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.010204081632653e-05,
-      "loss": 4.4063,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.98469387755102e-05,
-      "loss": 4.2461,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.959183673469388e-05,
-      "loss": 4.7348,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.933673469387755e-05,
-      "loss": 4.4816,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.908163265306123e-05,
-      "loss": 4.5978,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.882653061224489e-05,
-      "loss": 4.4656,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.857142857142858e-05,
-      "loss": 4.4803,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.831632653061226e-05,
-      "loss": 4.427,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.806122448979593e-05,
-      "loss": 4.1964,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.780612244897959e-05,
-      "loss": 4.2802,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.755102040816327e-05,
-      "loss": 4.3095,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.729591836734694e-05,
-      "loss": 4.5677,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.704081632653062e-05,
-      "loss": 4.7527,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.67857142857143e-05,
-      "loss": 4.3602,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.653061224489796e-05,
-      "loss": 4.4132,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.627551020408163e-05,
-      "loss": 4.1445,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.60204081632653e-05,
-      "loss": 4.2382,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.576530612244898e-05,
-      "loss": 4.5613,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.551020408163266e-05,
-      "loss": 4.5066,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.525510204081633e-05,
-      "loss": 4.1874,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.3602,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.474489795918368e-05,
-      "loss": 4.3446,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.448979591836736e-05,
-      "loss": 4.4451,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.423469387755102e-05,
-      "loss": 4.3278,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.39795918367347e-05,
-      "loss": 4.2441,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.372448979591837e-05,
-      "loss": 4.3496,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.346938775510205e-05,
-      "loss": 4.425,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.321428571428571e-05,
-      "loss": 4.2756,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.29591836734694e-05,
-      "loss": 4.2437,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.270408163265307e-05,
-      "loss": 4.2779,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.244897959183675e-05,
-      "loss": 4.2593,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.219387755102042e-05,
-      "loss": 4.4021,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.193877551020408e-05,
-      "loss": 4.3809,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.168367346938776e-05,
-      "loss": 4.4197,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.142857142857143e-05,
-      "loss": 4.2543,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.117346938775511e-05,
-      "loss": 4.2681,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.091836734693877e-05,
-      "loss": 4.2117,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.066326530612245e-05,
-      "loss": 4.163,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.040816326530612e-05,
-      "loss": 4.1755,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.01530612244898e-05,
-      "loss": 4.3606,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.989795918367347e-05,
-      "loss": 4.2221,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.964285714285715e-05,
-      "loss": 4.1264,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.938775510204082e-05,
-      "loss": 4.3952,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.91326530612245e-05,
-      "loss": 4.2269,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.887755102040817e-05,
-      "loss": 4.608,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.862244897959184e-05,
-      "loss": 4.2958,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.836734693877551e-05,
-      "loss": 4.2362,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.811224489795919e-05,
-      "loss": 4.1795,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.785714285714286e-05,
-      "loss": 4.1724,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.760204081632652e-05,
-      "loss": 4.5351,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.73469387755102e-05,
-      "loss": 4.3448,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.709183673469389e-05,
-      "loss": 4.2583,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.683673469387756e-05,
-      "loss": 4.266,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.658163265306124e-05,
-      "loss": 4.237,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.63265306122449e-05,
-      "loss": 4.4784,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.607142857142857e-05,
-      "loss": 4.3054,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.581632653061225e-05,
-      "loss": 4.2649,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.556122448979592e-05,
-      "loss": 4.1358,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.530612244897959e-05,
-      "loss": 4.3757,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.505102040816326e-05,
-      "loss": 4.0554,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.479591836734694e-05,
-      "loss": 4.2163,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.454081632653061e-05,
-      "loss": 4.6037,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.428571428571429e-05,
-      "loss": 4.6982,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.403061224489796e-05,
-      "loss": 4.1454,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.377551020408164e-05,
-      "loss": 3.9796,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.352040816326531e-05,
-      "loss": 4.3531,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.326530612244899e-05,
-      "loss": 4.1176,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.301020408163265e-05,
-      "loss": 4.3527,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.275510204081633e-05,
-      "loss": 4.4182,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.25e-05,
-      "loss": 4.3515,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.224489795918368e-05,
-      "loss": 4.4742,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.198979591836735e-05,
-      "loss": 4.4929,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.173469387755101e-05,
-      "loss": 4.3374,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.14795918367347e-05,
-      "loss": 4.4383,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.122448979591838e-05,
-      "loss": 4.2959,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0969387755102046e-05,
-      "loss": 4.4306,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0714285714285715e-05,
-      "loss": 4.1845,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.045918367346939e-05,
-      "loss": 4.4121,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.0204081632653065e-05,
-      "loss": 4.1552,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.994897959183674e-05,
-      "loss": 4.4136,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.9693877551020416e-05,
-      "loss": 4.0537,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.9438775510204084e-05,
-      "loss": 4.0487,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.918367346938776e-05,
-      "loss": 4.6006,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.8928571428571435e-05,
-      "loss": 4.1514,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.867346938775511e-05,
-      "loss": 4.0854,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.841836734693877e-05,
-      "loss": 4.3523,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.816326530612245e-05,
-      "loss": 4.2261,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.790816326530612e-05,
-      "loss": 4.3905,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7653061224489805e-05,
-      "loss": 4.4202,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.739795918367348e-05,
-      "loss": 4.3253,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.714285714285714e-05,
-      "loss": 4.1013,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.688775510204082e-05,
-      "loss": 4.1175,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.663265306122449e-05,
-      "loss": 4.0785,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.637755102040817e-05,
-      "loss": 4.0308,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.6122448979591836e-05,
-      "loss": 4.0925,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.586734693877551e-05,
-      "loss": 4.323,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.561224489795919e-05,
-      "loss": 4.0947,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.535714285714286e-05,
-      "loss": 3.9544,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.510204081632653e-05,
-      "loss": 4.0468,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.4846938775510206e-05,
-      "loss": 4.1615,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.459183673469388e-05,
-      "loss": 4.8051,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.4336734693877556e-05,
-      "loss": 4.1085,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.408163265306123e-05,
-      "loss": 4.5086,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.382653061224489e-05,
-      "loss": 4.467,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.3571428571428575e-05,
-      "loss": 4.4496,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.331632653061225e-05,
-      "loss": 4.2151,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.3061224489795926e-05,
-      "loss": 4.3281,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.280612244897959e-05,
-      "loss": 4.3739,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.255102040816326e-05,
-      "loss": 4.3452,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.229591836734694e-05,
-      "loss": 4.5481,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.2040816326530614e-05,
-      "loss": 4.084,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1785714285714296e-05,
-      "loss": 4.304,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.153061224489796e-05,
-      "loss": 4.0724,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.127551020408163e-05,
-      "loss": 4.5427,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.102040816326531e-05,
-      "loss": 4.693,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.076530612244898e-05,
-      "loss": 4.3249,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.051020408163265e-05,
-      "loss": 4.4825,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.025510204081633e-05,
-      "loss": 4.2753,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5e-05,
-      "loss": 4.5013,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 4.974489795918368e-05,
-      "loss": 4.4463,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.9489795918367346e-05,
-      "loss": 4.0471,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.923469387755102e-05,
-      "loss": 4.2305,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.89795918367347e-05,
-      "loss": 4.1357,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.872448979591837e-05,
-      "loss": 4.0259,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.846938775510204e-05,
-      "loss": 4.2383,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.8214285714285716e-05,
-      "loss": 4.3998,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.795918367346939e-05,
-      "loss": 4.4906,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.7704081632653066e-05,
-      "loss": 4.3952,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.744897959183674e-05,
-      "loss": 4.2198,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.719387755102041e-05,
-      "loss": 4.2494,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6938775510204086e-05,
-      "loss": 4.1282,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6683673469387754e-05,
-      "loss": 4.1195,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.642857142857143e-05,
-      "loss": 4.2118,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.6173469387755105e-05,
-      "loss": 4.286,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.591836734693878e-05,
-      "loss": 4.351,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.566326530612245e-05,
-      "loss": 4.048,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.5408163265306124e-05,
-      "loss": 4.4465,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.515306122448979e-05,
-      "loss": 4.1497,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4897959183673474e-05,
-      "loss": 3.9604,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.464285714285715e-05,
-      "loss": 4.1792,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.438775510204082e-05,
-      "loss": 4.1827,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4132653061224493e-05,
-      "loss": 4.0194,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.387755102040816e-05,
-      "loss": 4.2096,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.362244897959184e-05,
-      "loss": 4.2585,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.336734693877551e-05,
-      "loss": 4.077,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.311224489795919e-05,
-      "loss": 4.3033,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.2857142857142856e-05,
-      "loss": 4.119,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.260204081632653e-05,
-      "loss": 4.1655,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.234693877551021e-05,
-      "loss": 4.0441,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.209183673469388e-05,
-      "loss": 4.2595,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.183673469387756e-05,
-      "loss": 4.4831,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.1581632653061226e-05,
-      "loss": 4.3292,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.13265306122449e-05,
-      "loss": 4.0568,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.107142857142857e-05,
-      "loss": 4.2774,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.0816326530612245e-05,
-      "loss": 3.9564,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.056122448979592e-05,
-      "loss": 4.1067,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.0306122448979596e-05,
-      "loss": 4.2758,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.0051020408163264e-05,
-      "loss": 4.1247,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.979591836734694e-05,
-      "loss": 4.2443,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.9540816326530615e-05,
-      "loss": 4.2306,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.928571428571429e-05,
-      "loss": 4.4393,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.9030612244897965e-05,
-      "loss": 4.1833,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.8775510204081634e-05,
-      "loss": 3.9059,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.852040816326531e-05,
-      "loss": 4.3058,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.826530612244898e-05,
-      "loss": 4.2608,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.801020408163265e-05,
-      "loss": 4.4009,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.775510204081633e-05,
-      "loss": 4.2079,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 4.3105,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.724489795918368e-05,
-      "loss": 4.7069,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.698979591836735e-05,
-      "loss": 4.0475,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.673469387755102e-05,
-      "loss": 4.2485,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.64795918367347e-05,
-      "loss": 4.262,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.622448979591837e-05,
-      "loss": 4.2274,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.596938775510204e-05,
-      "loss": 4.1542,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.571428571428572e-05,
-      "loss": 4.4089,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5459183673469385e-05,
-      "loss": 4.2441,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.520408163265306e-05,
-      "loss": 4.198,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4948979591836736e-05,
-      "loss": 4.2563,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.469387755102041e-05,
-      "loss": 4.1014,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.443877551020409e-05,
-      "loss": 4.2019,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4183673469387755e-05,
-      "loss": 4.1753,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.392857142857143e-05,
-      "loss": 4.2436,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.36734693877551e-05,
-      "loss": 4.107,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.341836734693878e-05,
-      "loss": 4.0647,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.316326530612245e-05,
-      "loss": 4.3057,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2908163265306125e-05,
-      "loss": 4.109,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.265306122448979e-05,
-      "loss": 4.0356,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.239795918367347e-05,
-      "loss": 4.2105,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2142857142857144e-05,
-      "loss": 4.2356,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.188775510204082e-05,
-      "loss": 4.1832,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1632653061224494e-05,
-      "loss": 4.0565,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.137755102040816e-05,
-      "loss": 4.2768,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.112244897959184e-05,
-      "loss": 4.369,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.086734693877551e-05,
-      "loss": 3.935,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.061224489795919e-05,
-      "loss": 3.9795,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0357142857142857e-05,
-      "loss": 4.3018,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0102040816326533e-05,
-      "loss": 3.9756,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9846938775510208e-05,
-      "loss": 4.3389,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.959183673469388e-05,
-      "loss": 4.0707,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9336734693877555e-05,
-      "loss": 4.0017,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.9081632653061224e-05,
-      "loss": 3.9922,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8826530612244902e-05,
-      "loss": 4.1209,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.857142857142857e-05,
-      "loss": 4.1468,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8316326530612246e-05,
-      "loss": 3.9952,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8061224489795918e-05,
-      "loss": 4.1739,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7806122448979593e-05,
-      "loss": 3.9362,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7551020408163265e-05,
-      "loss": 4.2515,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.729591836734694e-05,
-      "loss": 4.1879,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7040816326530616e-05,
-      "loss": 4.0157,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6785714285714288e-05,
-      "loss": 4.1654,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6530612244897963e-05,
-      "loss": 4.1267,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.627551020408163e-05,
-      "loss": 4.1517,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6020408163265307e-05,
-      "loss": 4.0912,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.576530612244898e-05,
-      "loss": 4.3466,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5510204081632654e-05,
-      "loss": 4.2573,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5255102040816326e-05,
-      "loss": 4.0715,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5e-05,
-      "loss": 4.2313,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.4744897959183673e-05,
-      "loss": 4.2574,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.448979591836735e-05,
-      "loss": 4.8761,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.423469387755102e-05,
-      "loss": 4.3985,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3979591836734696e-05,
-      "loss": 4.2022,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.372448979591837e-05,
-      "loss": 4.0457,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3469387755102043e-05,
-      "loss": 4.1468,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3214285714285715e-05,
-      "loss": 4.1144,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.295918367346939e-05,
-      "loss": 3.9681,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2704081632653062e-05,
-      "loss": 4.1506,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2448979591836737e-05,
-      "loss": 4.1706,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.219387755102041e-05,
-      "loss": 4.0046,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.193877551020408e-05,
-      "loss": 4.2563,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1683673469387756e-05,
-      "loss": 4.0046,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1428571428571428e-05,
-      "loss": 4.3614,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1173469387755103e-05,
-      "loss": 4.1648,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.091836734693878e-05,
-      "loss": 3.9555,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.066326530612245e-05,
-      "loss": 4.0926,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0408163265306123e-05,
-      "loss": 4.0064,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0153061224489798e-05,
-      "loss": 4.0343,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.989795918367347e-05,
-      "loss": 4.1501,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9642857142857145e-05,
-      "loss": 4.1466,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9387755102040817e-05,
-      "loss": 4.0237,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.913265306122449e-05,
-      "loss": 4.127,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8877551020408164e-05,
-      "loss": 4.0763,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.862244897959184e-05,
-      "loss": 4.0363,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.836734693877551e-05,
-      "loss": 4.1314,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8112244897959187e-05,
-      "loss": 4.1815,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.785714285714286e-05,
-      "loss": 4.3073,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.760204081632653e-05,
-      "loss": 4.2567,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7346938775510206e-05,
-      "loss": 4.1142,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7091836734693878e-05,
-      "loss": 3.9881,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.683673469387755e-05,
-      "loss": 3.9615,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6581632653061225e-05,
-      "loss": 4.2256,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6326530612244897e-05,
-      "loss": 3.9554,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6071428571428572e-05,
-      "loss": 4.0022,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.5816326530612247e-05,
-      "loss": 4.1985,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.556122448979592e-05,
-      "loss": 4.0862,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5306122448979594e-05,
-      "loss": 4.065,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5051020408163266e-05,
-      "loss": 4.0925,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.479591836734694e-05,
-      "loss": 4.0593,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4540816326530612e-05,
-      "loss": 4.3442,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4285714285714285e-05,
-      "loss": 3.9675,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4030612244897959e-05,
-      "loss": 4.1694,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3775510204081633e-05,
-      "loss": 4.2668,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3520408163265308e-05,
-      "loss": 4.1724,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3265306122448982e-05,
-      "loss": 4.2712,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3010204081632653e-05,
-      "loss": 4.2255,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2755102040816327e-05,
-      "loss": 4.3171,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.25e-05,
-      "loss": 4.6056,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2244897959183674e-05,
-      "loss": 4.0734,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1989795918367348e-05,
-      "loss": 4.725,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1734693877551021e-05,
-      "loss": 4.3384,
-      "step": 849
-    },
     {
       "epoch": 0.95,
-      "learning_rate": 1.1479591836734695e-05,
-      "loss": 4.1118,
-      "step": 850
+      "learning_rate": 8.44e-05,
+      "loss": 4.2609,
+      "step": 424
     },
     {
       "epoch": 0.95,
-      "learning_rate": 1.1224489795918369e-05,
-      "loss": 4.1718,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.096938775510204e-05,
-      "loss": 4.1493,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0714285714285714e-05,
-      "loss": 4.0594,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.045918367346939e-05,
-      "loss": 4.0984,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0204081632653061e-05,
-      "loss": 4.0627,
-      "step": 855
+      "learning_rate": 8.46e-05,
+      "loss": 4.153,
+      "step": 425
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.948979591836735e-06,
-      "loss": 4.0885,
-      "step": 856
+      "learning_rate": 8.48e-05,
+      "loss": 4.1653,
+      "step": 426
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.693877551020408e-06,
-      "loss": 4.1545,
-      "step": 857
+      "learning_rate": 8.5e-05,
+      "loss": 4.1399,
+      "step": 427
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.438775510204082e-06,
-      "loss": 4.1281,
-      "step": 858
+      "learning_rate": 8.52e-05,
+      "loss": 4.1541,
+      "step": 428
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 3.9807,
-      "step": 859
+      "learning_rate": 8.54e-05,
+      "loss": 4.0411,
+      "step": 429
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.92857142857143e-06,
-      "loss": 4.1072,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755103e-06,
-      "loss": 4.2454,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938775e-06,
-      "loss": 4.0128,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.163265306122448e-06,
-      "loss": 4.1705,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306124e-06,
-      "loss": 4.3512,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.653061224489797e-06,
-      "loss": 4.264,
-      "step": 865
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.2159,
+      "step": 430
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-06,
-      "loss": 3.9359,
-      "step": 866
+      "learning_rate": 8.58e-05,
+      "loss": 4.203,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.142857142857143e-06,
-      "loss": 4.1472,
-      "step": 867
+      "learning_rate": 8.6e-05,
+      "loss": 4.3561,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.887755102040816e-06,
-      "loss": 3.9617,
-      "step": 868
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.147,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.632653061224491e-06,
-      "loss": 4.1297,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.3775510204081635e-06,
-      "loss": 4.2363,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 3.966,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.867346938775511e-06,
-      "loss": 4.1766,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.612244897959184e-06,
-      "loss": 4.1703,
-      "step": 873
+      "learning_rate": 8.64e-05,
+      "loss": 4.0882,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 4.138,
-      "step": 874
+      "learning_rate": 8.66e-05,
+      "loss": 4.239,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-06,
-      "loss": 4.2855,
-      "step": 875
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.047,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.846938775510204e-06,
-      "loss": 4.1395,
-      "step": 876
+      "learning_rate": 8.7e-05,
+      "loss": 4.0671,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 4.0096,
-      "step": 877
+      "learning_rate": 8.72e-05,
+      "loss": 4.3211,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.336734693877551e-06,
-      "loss": 4.0787,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.081632653061224e-06,
-      "loss": 4.1907,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.826530612244899e-06,
-      "loss": 4.348,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.5714285714285714e-06,
-      "loss": 3.9686,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3163265306122454e-06,
-      "loss": 4.0974,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 4.0149,
-      "step": 883
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 4.2255,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.806122448979592e-06,
-      "loss": 4.0868,
-      "step": 884
+      "learning_rate": 8.76e-05,
+      "loss": 4.2234,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.5510204081632653e-06,
-      "loss": 4.0935,
-      "step": 885
+      "learning_rate": 8.78e-05,
+      "loss": 4.1753,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 4.2536,
-      "step": 886
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.4779,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.040816326530612e-06,
-      "loss": 3.9868,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142857e-06,
-      "loss": 4.2264,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 4.2771,
-      "step": 889
+      "learning_rate": 8.82e-05,
+      "loss": 4.4142,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.2755102040816327e-06,
-      "loss": 4.0625,
-      "step": 890
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.3129,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.020408163265306e-06,
-      "loss": 4.2802,
-      "step": 891
+      "learning_rate": 8.86e-05,
+      "loss": 4.4601,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 3.737,
-      "step": 892
+      "learning_rate": 8.88e-05,
+      "loss": 4.2186,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.3097993645967385,
-      "train_runtime": 5935.828,
-      "train_samples_per_second": 4.808,
-      "train_steps_per_second": 0.15
+      "train_loss": 4.259838603537179,
+      "train_runtime": 4900.3549,
+      "train_samples_per_second": 5.824,
+      "train_steps_per_second": 0.091
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,