diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 1784,
+  "global_step": 892,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,10745 +10,5375 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 4.9149,
+      "loss": 4.8006,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.0571,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 5.0552,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 5.153,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.8954,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.6888,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8986,
       "step": 4
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.7919,
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.7319,
       "step": 5
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.8255,
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.8418,
       "step": 6
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.683,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.8198,
       "step": 7
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.8379,
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.7216,
       "step": 8
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 5.0641,
+      "loss": 4.8858,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.7448,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.7542,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.7885,
+      "learning_rate": 1.8e-06,
+      "loss": 4.8323,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.8e-06,
-      "loss": 4.752,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.7913,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.9365,
+      "learning_rate": 2.2e-06,
+      "loss": 4.757,
       "step": 13
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-06,
-      "loss": 4.5473,
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.7019,
       "step": 14
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.6781,
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 4.6262,
       "step": 15
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.6e-06,
-      "loss": 4.7832,
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.7446,
       "step": 16
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.7768,
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 4.7368,
       "step": 17
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3e-06,
-      "loss": 4.6424,
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.8308,
       "step": 18
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.602,
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.5315,
       "step": 19
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.88,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-06,
+      "loss": 4.6872,
       "step": 20
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.6e-06,
-      "loss": 4.5707,
+      "epoch": 0.02,
+      "learning_rate": 3.8e-06,
+      "loss": 4.5977,
       "step": 21
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.8e-06,
-      "loss": 4.9183,
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.5822,
       "step": 22
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.7453,
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.5846,
       "step": 23
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.4636,
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 4.7091,
       "step": 24
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.4e-06,
-      "loss": 4.6252,
+      "epoch": 0.03,
+      "learning_rate": 4.6e-06,
+      "loss": 4.5738,
       "step": 25
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.6e-06,
-      "loss": 4.7267,
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.7385,
       "step": 26
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.5355,
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 4.454,
       "step": 27
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5e-06,
-      "loss": 4.2787,
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 4.3818,
       "step": 28
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.2e-06,
-      "loss": 4.6034,
+      "epoch": 0.03,
+      "learning_rate": 5.4e-06,
+      "loss": 4.3869,
       "step": 29
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.4e-06,
-      "loss": 4.4823,
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.5749,
       "step": 30
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.3662,
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5545,
       "step": 31
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.8e-06,
-      "loss": 4.7027,
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 4.3159,
       "step": 32
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6e-06,
-      "loss": 4.4031,
+      "epoch": 0.04,
+      "learning_rate": 6.2e-06,
+      "loss": 4.4299,
       "step": 33
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.2e-06,
-      "loss": 4.2832,
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.2923,
       "step": 34
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.5086,
+      "epoch": 0.04,
+      "learning_rate": 6.6e-06,
+      "loss": 4.4871,
       "step": 35
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.6e-06,
-      "loss": 4.7472,
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4389,
       "step": 36
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4181,
+      "epoch": 0.04,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.4767,
       "step": 37
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.5281,
+      "epoch": 0.04,
+      "learning_rate": 7.2e-06,
+      "loss": 4.484,
       "step": 38
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.2e-06,
-      "loss": 4.5992,
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 4.641,
       "step": 39
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.4e-06,
-      "loss": 4.6275,
+      "epoch": 0.04,
+      "learning_rate": 7.6e-06,
+      "loss": 4.4554,
       "step": 40
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.6e-06,
-      "loss": 4.4101,
+      "epoch": 0.05,
+      "learning_rate": 7.8e-06,
+      "loss": 4.3018,
       "step": 41
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.8e-06,
-      "loss": 4.2274,
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.4761,
       "step": 42
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.5139,
+      "epoch": 0.05,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.2683,
       "step": 43
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.3415,
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.426,
       "step": 44
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.6046,
+      "epoch": 0.05,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.4915,
       "step": 45
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.5936,
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 4.6235,
       "step": 46
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.8e-06,
-      "loss": 4.6534,
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 4.4703,
       "step": 47
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9e-06,
-      "loss": 4.9689,
+      "epoch": 0.05,
+      "learning_rate": 9.2e-06,
+      "loss": 4.8672,
       "step": 48
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.05,
       "learning_rate": 9.2e-06,
-      "loss": 4.9966,
+      "loss": 5.0521,
       "step": 49
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 9.4e-06,
-      "loss": 4.9436,
+      "loss": 4.6531,
       "step": 50
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2853,
+      "loss": 4.2156,
       "step": 51
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 9.800000000000001e-06,
-      "loss": 4.4186,
+      "loss": 4.348,
       "step": 52
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 1e-05,
-      "loss": 4.4083,
+      "loss": 4.2307,
       "step": 53
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 1.02e-05,
-      "loss": 4.2975,
+      "loss": 4.2949,
       "step": 54
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 1.04e-05,
-      "loss": 4.3418,
+      "loss": 4.28,
       "step": 55
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 1.06e-05,
-      "loss": 4.4606,
+      "loss": 4.3738,
       "step": 56
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 1.08e-05,
-      "loss": 4.3561,
+      "loss": 4.344,
       "step": 57
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.3226,
+      "loss": 4.5751,
       "step": 58
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.3613,
+      "loss": 4.3086,
       "step": 59
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.2703,
+      "loss": 4.1412,
       "step": 60
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 1.16e-05,
-      "loss": 4.3399,
+      "loss": 4.4096,
       "step": 61
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 1.18e-05,
-      "loss": 4.2631,
+      "loss": 4.3565,
       "step": 62
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 1.2e-05,
-      "loss": 4.6132,
+      "loss": 4.1857,
       "step": 63
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 1.22e-05,
-      "loss": 4.128,
+      "loss": 4.3458,
       "step": 64
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 1.24e-05,
-      "loss": 4.3217,
+      "loss": 4.1284,
       "step": 65
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.4146,
+      "loss": 4.4129,
       "step": 66
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.4274,
+      "loss": 4.1812,
       "step": 67
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.4433,
+      "loss": 4.2234,
       "step": 68
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.32e-05,
-      "loss": 4.3398,
+      "loss": 4.1651,
       "step": 69
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.2744,
+      "loss": 4.29,
       "step": 70
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.3691,
+      "loss": 4.4585,
       "step": 71
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.1392,
+      "loss": 4.3538,
       "step": 72
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.3788,
+      "loss": 4.281,
       "step": 73
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.42e-05,
-      "loss": 4.1495,
+      "loss": 4.3208,
       "step": 74
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 1.44e-05,
-      "loss": 4.6669,
+      "loss": 4.1282,
       "step": 75
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.4346,
+      "loss": 4.4093,
       "step": 76
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 1.48e-05,
-      "loss": 4.2604,
+      "loss": 4.3022,
       "step": 77
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 1.5e-05,
-      "loss": 4.2075,
+      "loss": 4.2166,
       "step": 78
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 1.52e-05,
-      "loss": 4.2025,
+      "loss": 4.0831,
       "step": 79
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.09,
       "learning_rate": 1.54e-05,
-      "loss": 4.5405,
+      "loss": 4.3279,
       "step": 80
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 1.56e-05,
-      "loss": 4.1777,
+      "loss": 4.3771,
       "step": 81
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 1.58e-05,
-      "loss": 4.3264,
+      "loss": 4.1824,
       "step": 82
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.1981,
+      "loss": 4.3989,
       "step": 83
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 1.62e-05,
-      "loss": 4.3238,
+      "loss": 4.3379,
       "step": 84
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.1848,
+      "loss": 4.2745,
       "step": 85
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.66e-05,
-      "loss": 4.1893,
+      "loss": 4.2694,
       "step": 86
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.5532,
+      "loss": 4.2729,
       "step": 87
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.1578,
+      "loss": 4.2621,
       "step": 88
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.5041,
+      "loss": 4.3116,
       "step": 89
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.74e-05,
-      "loss": 4.0696,
+      "loss": 4.2443,
       "step": 90
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.76e-05,
-      "loss": 4.4558,
+      "loss": 4.4649,
       "step": 91
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.78e-05,
-      "loss": 4.124,
+      "loss": 4.3735,
       "step": 92
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 1.8e-05,
-      "loss": 3.9798,
+      "loss": 4.0979,
       "step": 93
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.4077,
+      "loss": 4.3195,
       "step": 94
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 1.84e-05,
-      "loss": 4.6274,
+      "loss": 4.5964,
       "step": 95
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 1.86e-05,
-      "loss": 4.2473,
+      "loss": 4.5072,
       "step": 96
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 1.88e-05,
-      "loss": 4.4177,
+      "loss": 4.2899,
       "step": 97
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 1.9e-05,
-      "loss": 4.4741,
+      "loss": 4.5847,
       "step": 98
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.11,
       "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.7864,
+      "loss": 4.3735,
       "step": 99
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.6756,
+      "epoch": 0.11,
+      "learning_rate": 1.94e-05,
+      "loss": 4.3218,
       "step": 100
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.94e-05,
-      "loss": 4.342,
+      "epoch": 0.11,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.177,
       "step": 101
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.1304,
+      "epoch": 0.11,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.209,
       "step": 102
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.2758,
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 4.327,
       "step": 103
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2e-05,
-      "loss": 4.2114,
+      "epoch": 0.12,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.2536,
       "step": 104
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.2557,
+      "epoch": 0.12,
+      "learning_rate": 2.04e-05,
+      "loss": 4.0728,
       "step": 105
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.04e-05,
-      "loss": 4.2719,
+      "epoch": 0.12,
+      "learning_rate": 2.06e-05,
+      "loss": 4.2959,
       "step": 106
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.06e-05,
-      "loss": 4.3017,
+      "epoch": 0.12,
+      "learning_rate": 2.08e-05,
+      "loss": 4.2988,
       "step": 107
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.08e-05,
-      "loss": 4.3792,
+      "epoch": 0.12,
+      "learning_rate": 2.1e-05,
+      "loss": 4.4356,
       "step": 108
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.1e-05,
-      "loss": 4.2246,
+      "epoch": 0.12,
+      "learning_rate": 2.12e-05,
+      "loss": 4.0137,
       "step": 109
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.12e-05,
-      "loss": 4.2699,
+      "epoch": 0.12,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.1517,
       "step": 110
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.2453,
+      "epoch": 0.12,
+      "learning_rate": 2.16e-05,
+      "loss": 4.3351,
       "step": 111
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.16e-05,
-      "loss": 4.1408,
+      "epoch": 0.13,
+      "learning_rate": 2.18e-05,
+      "loss": 4.1899,
       "step": 112
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.18e-05,
-      "loss": 4.2765,
+      "epoch": 0.13,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.185,
       "step": 113
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.0103,
+      "epoch": 0.13,
+      "learning_rate": 2.22e-05,
+      "loss": 4.1203,
       "step": 114
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2.22e-05,
-      "loss": 4.0544,
+      "epoch": 0.13,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.1869,
       "step": 115
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.1861,
+      "epoch": 0.13,
+      "learning_rate": 2.26e-05,
+      "loss": 3.9657,
       "step": 116
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.26e-05,
-      "loss": 4.3476,
+      "epoch": 0.13,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.2749,
       "step": 117
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.1634,
+      "epoch": 0.13,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.1479,
       "step": 118
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.2868,
+      "epoch": 0.13,
+      "learning_rate": 2.32e-05,
+      "loss": 4.2752,
       "step": 119
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.32e-05,
-      "loss": 4.3688,
+      "epoch": 0.13,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.3304,
       "step": 120
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.3692,
+      "epoch": 0.14,
+      "learning_rate": 2.36e-05,
+      "loss": 4.2427,
       "step": 121
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.36e-05,
-      "loss": 4.2603,
+      "epoch": 0.14,
+      "learning_rate": 2.38e-05,
+      "loss": 4.2195,
       "step": 122
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.38e-05,
-      "loss": 4.366,
+      "epoch": 0.14,
+      "learning_rate": 2.4e-05,
+      "loss": 4.0881,
       "step": 123
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.4e-05,
-      "loss": 4.2786,
+      "epoch": 0.14,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.3312,
       "step": 124
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 3.9926,
+      "epoch": 0.14,
+      "learning_rate": 2.44e-05,
+      "loss": 4.2518,
       "step": 125
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.44e-05,
-      "loss": 4.5543,
+      "epoch": 0.14,
+      "learning_rate": 2.46e-05,
+      "loss": 4.2965,
       "step": 126
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.46e-05,
-      "loss": 4.3002,
+      "epoch": 0.14,
+      "learning_rate": 2.48e-05,
+      "loss": 4.3571,
       "step": 127
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.48e-05,
-      "loss": 4.2607,
+      "epoch": 0.14,
+      "learning_rate": 2.5e-05,
+      "loss": 4.1767,
       "step": 128
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.5e-05,
-      "loss": 4.1329,
+      "epoch": 0.14,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.1605,
       "step": 129
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.2557,
+      "epoch": 0.15,
+      "learning_rate": 2.54e-05,
+      "loss": 4.3313,
       "step": 130
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.54e-05,
-      "loss": 4.5012,
+      "epoch": 0.15,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.2277,
       "step": 131
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2243,
+      "epoch": 0.15,
+      "learning_rate": 2.58e-05,
+      "loss": 4.313,
       "step": 132
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.58e-05,
-      "loss": 4.2797,
+      "epoch": 0.15,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.2856,
       "step": 133
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.2891,
+      "epoch": 0.15,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.2828,
       "step": 134
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.1807,
+      "epoch": 0.15,
+      "learning_rate": 2.64e-05,
+      "loss": 4.1482,
       "step": 135
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.64e-05,
-      "loss": 4.3632,
+      "epoch": 0.15,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.2175,
       "step": 136
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.1102,
+      "epoch": 0.15,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.5134,
       "step": 137
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.3682,
+      "epoch": 0.15,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2704,
       "step": 138
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.4895,
+      "epoch": 0.16,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.3241,
       "step": 139
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.1464,
+      "epoch": 0.16,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.105,
       "step": 140
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.3417,
+      "epoch": 0.16,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.4102,
       "step": 141
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.5845,
+      "epoch": 0.16,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.4697,
       "step": 142
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.2798,
+      "epoch": 0.16,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.3797,
       "step": 143
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.3445,
+      "epoch": 0.16,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.2843,
       "step": 144
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.6849,
+      "epoch": 0.16,
+      "learning_rate": 2.84e-05,
+      "loss": 4.421,
       "step": 145
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.84e-05,
-      "loss": 4.4547,
+      "epoch": 0.16,
+      "learning_rate": 2.86e-05,
+      "loss": 4.3733,
       "step": 146
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.86e-05,
-      "loss": 4.426,
+      "epoch": 0.16,
+      "learning_rate": 2.88e-05,
+      "loss": 4.4609,
       "step": 147
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.88e-05,
-      "loss": 4.5393,
+      "epoch": 0.17,
+      "learning_rate": 2.9e-05,
+      "loss": 4.246,
       "step": 148
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.9e-05,
-      "loss": 4.7466,
+      "epoch": 0.17,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.5866,
       "step": 149
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.0924,
+      "epoch": 0.17,
+      "learning_rate": 2.94e-05,
+      "loss": 4.2249,
       "step": 150
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.94e-05,
-      "loss": 4.0084,
+      "epoch": 0.17,
+      "learning_rate": 2.96e-05,
+      "loss": 4.3224,
       "step": 151
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.96e-05,
-      "loss": 4.4036,
+      "epoch": 0.17,
+      "learning_rate": 2.98e-05,
+      "loss": 4.2143,
       "step": 152
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.98e-05,
-      "loss": 3.9931,
+      "epoch": 0.17,
+      "learning_rate": 3e-05,
+      "loss": 4.2868,
       "step": 153
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3e-05,
-      "loss": 4.4475,
+      "epoch": 0.17,
+      "learning_rate": 3.02e-05,
+      "loss": 4.1287,
       "step": 154
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.02e-05,
-      "loss": 4.2978,
+      "epoch": 0.17,
+      "learning_rate": 3.04e-05,
+      "loss": 4.0788,
       "step": 155
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.04e-05,
-      "loss": 4.535,
+      "epoch": 0.17,
+      "learning_rate": 3.06e-05,
+      "loss": 4.172,
       "step": 156
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.06e-05,
-      "loss": 4.334,
+      "epoch": 0.18,
+      "learning_rate": 3.08e-05,
+      "loss": 4.3126,
       "step": 157
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.08e-05,
-      "loss": 4.3887,
+      "epoch": 0.18,
+      "learning_rate": 3.1e-05,
+      "loss": 4.4143,
       "step": 158
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.1e-05,
-      "loss": 4.3107,
+      "epoch": 0.18,
+      "learning_rate": 3.12e-05,
+      "loss": 4.4585,
       "step": 159
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.12e-05,
-      "loss": 4.2977,
+      "epoch": 0.18,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.1789,
       "step": 160
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.3457,
+      "epoch": 0.18,
+      "learning_rate": 3.16e-05,
+      "loss": 4.1989,
       "step": 161
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.16e-05,
-      "loss": 4.4493,
+      "epoch": 0.18,
+      "learning_rate": 3.18e-05,
+      "loss": 4.2238,
       "step": 162
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.18e-05,
-      "loss": 4.2803,
+      "epoch": 0.18,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.2342,
       "step": 163
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.4809,
+      "epoch": 0.18,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.3552,
       "step": 164
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.2896,
+      "epoch": 0.18,
+      "learning_rate": 3.24e-05,
+      "loss": 4.257,
       "step": 165
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.24e-05,
-      "loss": 4.4736,
+      "epoch": 0.19,
+      "learning_rate": 3.26e-05,
+      "loss": 4.2857,
       "step": 166
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.26e-05,
-      "loss": 4.2675,
+      "epoch": 0.19,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.2484,
       "step": 167
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.0364,
+      "epoch": 0.19,
+      "learning_rate": 3.3e-05,
+      "loss": 4.0284,
       "step": 168
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.3e-05,
-      "loss": 4.138,
+      "epoch": 0.19,
+      "learning_rate": 3.32e-05,
+      "loss": 4.0501,
       "step": 169
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.32e-05,
-      "loss": 4.2559,
+      "epoch": 0.19,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.1841,
       "step": 170
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.3203,
+      "epoch": 0.19,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.4082,
       "step": 171
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.3158,
+      "epoch": 0.19,
+      "learning_rate": 3.38e-05,
+      "loss": 4.363,
       "step": 172
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.38e-05,
-      "loss": 4.2825,
+      "epoch": 0.19,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.2477,
       "step": 173
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.2205,
+      "epoch": 0.2,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.0851,
       "step": 174
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.4189,
+      "epoch": 0.2,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.2421,
       "step": 175
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.1599,
+      "epoch": 0.2,
+      "learning_rate": 3.46e-05,
+      "loss": 4.0494,
       "step": 176
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.46e-05,
-      "loss": 4.1716,
+      "epoch": 0.2,
+      "learning_rate": 3.48e-05,
+      "loss": 4.0945,
       "step": 177
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.48e-05,
-      "loss": 4.1818,
+      "epoch": 0.2,
+      "learning_rate": 3.5e-05,
+      "loss": 4.171,
       "step": 178
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.5e-05,
-      "loss": 3.9907,
+      "epoch": 0.2,
+      "learning_rate": 3.52e-05,
+      "loss": 4.2874,
       "step": 179
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.52e-05,
-      "loss": 4.4088,
+      "epoch": 0.2,
+      "learning_rate": 3.54e-05,
+      "loss": 4.2568,
       "step": 180
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.54e-05,
-      "loss": 4.2929,
+      "epoch": 0.2,
+      "learning_rate": 3.56e-05,
+      "loss": 4.2897,
       "step": 181
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.56e-05,
-      "loss": 4.2211,
+      "epoch": 0.2,
+      "learning_rate": 3.58e-05,
+      "loss": 4.1081,
       "step": 182
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.58e-05,
-      "loss": 4.4237,
+      "epoch": 0.21,
+      "learning_rate": 3.6e-05,
+      "loss": 4.313,
       "step": 183
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.6e-05,
-      "loss": 4.3746,
+      "epoch": 0.21,
+      "learning_rate": 3.62e-05,
+      "loss": 4.292,
       "step": 184
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.62e-05,
-      "loss": 4.3792,
+      "epoch": 0.21,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.2912,
       "step": 185
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.2139,
+      "epoch": 0.21,
+      "learning_rate": 3.66e-05,
+      "loss": 4.151,
       "step": 186
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.66e-05,
-      "loss": 4.2976,
+      "epoch": 0.21,
+      "learning_rate": 3.68e-05,
+      "loss": 4.3744,
       "step": 187
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.68e-05,
-      "loss": 4.4506,
+      "epoch": 0.21,
+      "learning_rate": 3.7e-05,
+      "loss": 4.0779,
       "step": 188
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.7e-05,
-      "loss": 4.167,
+      "epoch": 0.21,
+      "learning_rate": 3.72e-05,
+      "loss": 4.2057,
       "step": 189
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.72e-05,
-      "loss": 4.203,
+      "epoch": 0.21,
+      "learning_rate": 3.74e-05,
+      "loss": 4.3045,
       "step": 190
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.74e-05,
-      "loss": 4.4948,
+      "epoch": 0.21,
+      "learning_rate": 3.76e-05,
+      "loss": 4.3288,
       "step": 191
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.76e-05,
-      "loss": 4.1986,
+      "epoch": 0.22,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.2585,
       "step": 192
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.122,
+      "epoch": 0.22,
+      "learning_rate": 3.8e-05,
+      "loss": 4.491,
       "step": 193
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.8e-05,
-      "loss": 4.2625,
+      "epoch": 0.22,
+      "learning_rate": 3.82e-05,
+      "loss": 4.1982,
       "step": 194
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.82e-05,
-      "loss": 4.5575,
+      "epoch": 0.22,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.2791,
       "step": 195
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.3536,
+      "epoch": 0.22,
+      "learning_rate": 3.86e-05,
+      "loss": 4.4892,
       "step": 196
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.86e-05,
-      "loss": 4.3877,
+      "epoch": 0.22,
+      "learning_rate": 3.88e-05,
+      "loss": 4.4989,
       "step": 197
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.88e-05,
-      "loss": 4.3271,
+      "epoch": 0.22,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.3795,
       "step": 198
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.4041,
+      "epoch": 0.22,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.6247,
       "step": 199
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.3396,
+      "epoch": 0.22,
+      "learning_rate": 3.94e-05,
+      "loss": 4.6662,
       "step": 200
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.94e-05,
-      "loss": 4.3546,
+      "epoch": 0.23,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.0805,
       "step": 201
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 4.2722,
+      "epoch": 0.23,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.2222,
       "step": 202
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.395,
+      "epoch": 0.23,
+      "learning_rate": 4e-05,
+      "loss": 4.1281,
       "step": 203
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4e-05,
-      "loss": 4.18,
+      "epoch": 0.23,
+      "learning_rate": 4.02e-05,
+      "loss": 4.1377,
       "step": 204
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.02e-05,
-      "loss": 4.1993,
+      "epoch": 0.23,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.0093,
       "step": 205
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.3403,
+      "epoch": 0.23,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.0695,
       "step": 206
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.3792,
+      "epoch": 0.23,
+      "learning_rate": 4.08e-05,
+      "loss": 4.1305,
       "step": 207
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.08e-05,
-      "loss": 4.1122,
+      "epoch": 0.23,
+      "learning_rate": 4.1e-05,
+      "loss": 4.0018,
       "step": 208
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.1e-05,
-      "loss": 4.1175,
+      "epoch": 0.23,
+      "learning_rate": 4.12e-05,
+      "loss": 4.2204,
       "step": 209
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.12e-05,
-      "loss": 4.3285,
+      "epoch": 0.24,
+      "learning_rate": 4.14e-05,
+      "loss": 4.1703,
       "step": 210
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.14e-05,
-      "loss": 4.2507,
+      "epoch": 0.24,
+      "learning_rate": 4.16e-05,
+      "loss": 4.0995,
       "step": 211
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.16e-05,
-      "loss": 4.2593,
+      "epoch": 0.24,
+      "learning_rate": 4.18e-05,
+      "loss": 4.2342,
       "step": 212
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.18e-05,
-      "loss": 4.259,
+      "epoch": 0.24,
+      "learning_rate": 4.2e-05,
+      "loss": 4.1738,
       "step": 213
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.2e-05,
-      "loss": 4.1006,
+      "epoch": 0.24,
+      "learning_rate": 4.22e-05,
+      "loss": 4.2382,
       "step": 214
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.22e-05,
-      "loss": 4.4851,
+      "epoch": 0.24,
+      "learning_rate": 4.24e-05,
+      "loss": 4.1038,
       "step": 215
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.24e-05,
-      "loss": 3.8924,
+      "epoch": 0.24,
+      "learning_rate": 4.26e-05,
+      "loss": 4.2598,
       "step": 216
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.26e-05,
-      "loss": 4.1776,
+      "epoch": 0.24,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.1992,
       "step": 217
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.2651,
+      "epoch": 0.24,
+      "learning_rate": 4.3e-05,
+      "loss": 4.1448,
       "step": 218
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3e-05,
-      "loss": 4.2307,
+      "epoch": 0.25,
+      "learning_rate": 4.32e-05,
+      "loss": 4.2357,
       "step": 219
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.32e-05,
-      "loss": 4.5025,
+      "epoch": 0.25,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.1894,
       "step": 220
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.2906,
+      "epoch": 0.25,
+      "learning_rate": 4.36e-05,
+      "loss": 4.1205,
       "step": 221
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.36e-05,
-      "loss": 4.4008,
+      "epoch": 0.25,
+      "learning_rate": 4.38e-05,
+      "loss": 4.2463,
       "step": 222
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.38e-05,
-      "loss": 4.1553,
+      "epoch": 0.25,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.2651,
       "step": 223
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.2888,
+      "epoch": 0.25,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.1997,
       "step": 224
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.3341,
+      "epoch": 0.25,
+      "learning_rate": 4.44e-05,
+      "loss": 4.2946,
       "step": 225
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.44e-05,
-      "loss": 4.4576,
+      "epoch": 0.25,
+      "learning_rate": 4.46e-05,
+      "loss": 4.4428,
       "step": 226
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.46e-05,
-      "loss": 4.3345,
+      "epoch": 0.25,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 4.2206,
       "step": 227
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.0651,
+      "epoch": 0.26,
+      "learning_rate": 4.5e-05,
+      "loss": 4.1257,
       "step": 228
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.5e-05,
-      "loss": 4.2568,
+      "epoch": 0.26,
+      "learning_rate": 4.52e-05,
+      "loss": 4.1379,
       "step": 229
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.52e-05,
-      "loss": 4.2845,
+      "epoch": 0.26,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.1872,
       "step": 230
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.3083,
+      "epoch": 0.26,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.2277,
       "step": 231
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.2925,
+      "epoch": 0.26,
+      "learning_rate": 4.58e-05,
+      "loss": 4.3086,
       "step": 232
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.58e-05,
-      "loss": 4.4548,
+      "epoch": 0.26,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.1558,
       "step": 233
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 4.3042,
+      "epoch": 0.26,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.2961,
       "step": 234
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.1476,
+      "epoch": 0.26,
+      "learning_rate": 4.64e-05,
+      "loss": 4.3365,
       "step": 235
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.64e-05,
-      "loss": 4.7395,
+      "epoch": 0.26,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.1922,
       "step": 236
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 4.3636,
+      "epoch": 0.27,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.1784,
       "step": 237
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.3284,
+      "epoch": 0.27,
+      "learning_rate": 4.7e-05,
+      "loss": 4.4414,
       "step": 238
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.7e-05,
-      "loss": 4.3673,
+      "epoch": 0.27,
+      "learning_rate": 4.72e-05,
+      "loss": 4.2069,
       "step": 239
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.72e-05,
-      "loss": 4.2996,
+      "epoch": 0.27,
+      "learning_rate": 4.74e-05,
+      "loss": 4.3952,
       "step": 240
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.74e-05,
-      "loss": 4.5105,
+      "epoch": 0.27,
+      "learning_rate": 4.76e-05,
+      "loss": 4.295,
       "step": 241
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.76e-05,
-      "loss": 4.2509,
+      "epoch": 0.27,
+      "learning_rate": 4.78e-05,
+      "loss": 4.4594,
       "step": 242
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.78e-05,
-      "loss": 4.4027,
+      "epoch": 0.27,
+      "learning_rate": 4.8e-05,
+      "loss": 4.3772,
       "step": 243
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.8e-05,
-      "loss": 4.5563,
+      "epoch": 0.27,
+      "learning_rate": 4.82e-05,
+      "loss": 4.2769,
       "step": 244
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.82e-05,
-      "loss": 4.5825,
+      "epoch": 0.27,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.4663,
       "step": 245
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.5415,
+      "epoch": 0.28,
+      "learning_rate": 4.86e-05,
+      "loss": 4.3733,
       "step": 246
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.86e-05,
-      "loss": 4.7543,
+      "epoch": 0.28,
+      "learning_rate": 4.88e-05,
+      "loss": 4.7879,
       "step": 247
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.88e-05,
-      "loss": 4.5764,
+      "epoch": 0.28,
+      "learning_rate": 4.9e-05,
+      "loss": 4.7509,
       "step": 248
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.9e-05,
-      "loss": 4.3342,
+      "epoch": 0.28,
+      "learning_rate": 4.92e-05,
+      "loss": 4.7604,
       "step": 249
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.92e-05,
-      "loss": 4.8834,
+      "epoch": 0.28,
+      "learning_rate": 4.94e-05,
+      "loss": 4.1676,
       "step": 250
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.94e-05,
-      "loss": 4.2397,
+      "epoch": 0.28,
+      "learning_rate": 4.96e-05,
+      "loss": 4.4618,
       "step": 251
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.96e-05,
-      "loss": 4.6447,
+      "epoch": 0.28,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.1168,
       "step": 252
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.3595,
+      "epoch": 0.28,
+      "learning_rate": 5e-05,
+      "loss": 4.1508,
       "step": 253
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 5e-05,
-      "loss": 4.4588,
+      "epoch": 0.28,
+      "learning_rate": 5.02e-05,
+      "loss": 4.2563,
       "step": 254
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 5.02e-05,
-      "loss": 4.3061,
+      "epoch": 0.29,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.1567,
       "step": 255
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.2642,
+      "epoch": 0.29,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.1473,
       "step": 256
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.2662,
+      "epoch": 0.29,
+      "learning_rate": 5.08e-05,
+      "loss": 4.2933,
       "step": 257
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 5.08e-05,
-      "loss": 4.511,
+      "epoch": 0.29,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.1407,
       "step": 258
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.5593,
+      "epoch": 0.29,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 4.1175,
       "step": 259
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.0348,
+      "epoch": 0.29,
+      "learning_rate": 5.14e-05,
+      "loss": 4.3068,
       "step": 260
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.14e-05,
-      "loss": 4.2391,
+      "epoch": 0.29,
+      "learning_rate": 5.16e-05,
+      "loss": 4.2729,
       "step": 261
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.16e-05,
-      "loss": 4.4057,
+      "epoch": 0.29,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.1282,
       "step": 262
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.4139,
+      "epoch": 0.29,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.1259,
       "step": 263
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.2667,
+      "epoch": 0.3,
+      "learning_rate": 5.22e-05,
+      "loss": 4.1419,
       "step": 264
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.22e-05,
-      "loss": 4.2761,
+      "epoch": 0.3,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 4.2539,
       "step": 265
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.0799,
+      "epoch": 0.3,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.2848,
       "step": 266
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.4091,
+      "epoch": 0.3,
+      "learning_rate": 5.28e-05,
+      "loss": 4.1218,
       "step": 267
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.28e-05,
-      "loss": 4.3407,
+      "epoch": 0.3,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.4412,
       "step": 268
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 4.5031,
+      "epoch": 0.3,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.1862,
       "step": 269
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.0403,
-      "step": 270
-    },
-    {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.3373,
-      "step": 271
+      "loss": 4.3737,
+      "step": 270
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 5.360000000000001e-05,
-      "loss": 4.3094,
-      "step": 272
+      "loss": 3.9655,
+      "step": 271
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 5.380000000000001e-05,
-      "loss": 4.1104,
-      "step": 273
+      "loss": 4.1543,
+      "step": 272
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.5809,
-      "step": 274
+      "loss": 4.1888,
+      "step": 273
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 5.420000000000001e-05,
-      "loss": 4.3247,
-      "step": 275
+      "loss": 4.0681,
+      "step": 274
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 5.440000000000001e-05,
-      "loss": 4.5648,
-      "step": 276
+      "loss": 4.3354,
+      "step": 275
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 5.4600000000000006e-05,
       "loss": 4.2964,
-      "step": 277
+      "step": 276
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.4143,
-      "step": 278
+      "loss": 4.2131,
+      "step": 277
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 4.2007,
-      "step": 279
+      "loss": 4.1059,
+      "step": 278
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 5.520000000000001e-05,
-      "loss": 4.3295,
-      "step": 280
+      "loss": 4.1404,
+      "step": 279
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.5437,
-      "step": 281
+      "loss": 4.3474,
+      "step": 280
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.560000000000001e-05,
-      "loss": 4.481,
-      "step": 282
+      "loss": 4.0881,
+      "step": 281
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.580000000000001e-05,
-      "loss": 4.4487,
-      "step": 283
+      "loss": 4.2215,
+      "step": 282
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.6000000000000006e-05,
-      "loss": 4.1673,
-      "step": 284
+      "loss": 4.0005,
+      "step": 283
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.620000000000001e-05,
-      "loss": 4.433,
-      "step": 285
+      "loss": 4.224,
+      "step": 284
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.372,
-      "step": 286
+      "loss": 4.2709,
+      "step": 285
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.66e-05,
-      "loss": 4.3542,
-      "step": 287
+      "loss": 4.164,
+      "step": 286
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.68e-05,
-      "loss": 4.5186,
-      "step": 288
+      "loss": 4.3511,
+      "step": 287
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.2899,
-      "step": 289
+      "loss": 4.2992,
+      "step": 288
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 5.72e-05,
-      "loss": 4.626,
-      "step": 290
+      "loss": 4.3478,
+      "step": 289
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 5.74e-05,
-      "loss": 4.3431,
-      "step": 291
+      "loss": 4.3043,
+      "step": 290
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 5.76e-05,
-      "loss": 4.9951,
-      "step": 292
+      "loss": 4.3565,
+      "step": 291
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.49,
-      "step": 293
+      "loss": 4.2118,
+      "step": 292
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.33,
       "learning_rate": 5.8e-05,
-      "loss": 4.4871,
-      "step": 294
+      "loss": 4.5298,
+      "step": 293
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 5.82e-05,
-      "loss": 4.3942,
-      "step": 295
+      "loss": 4.2882,
+      "step": 294
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 5.8399999999999997e-05,
-      "loss": 5.0704,
-      "step": 296
+      "loss": 4.1672,
+      "step": 295
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 5.86e-05,
-      "loss": 4.541,
-      "step": 297
+      "loss": 4.5489,
+      "step": 296
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 5.88e-05,
-      "loss": 4.6234,
-      "step": 298
+      "loss": 4.5708,
+      "step": 297
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 5.9e-05,
-      "loss": 4.4043,
-      "step": 299
+      "loss": 4.3176,
+      "step": 298
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 5.92e-05,
-      "loss": 4.5694,
-      "step": 300
+      "loss": 4.7349,
+      "step": 299
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 5.94e-05,
-      "loss": 4.4918,
-      "step": 301
+      "loss": 4.7477,
+      "step": 300
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 5.96e-05,
-      "loss": 4.3884,
-      "step": 302
+      "loss": 4.1141,
+      "step": 301
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 5.9800000000000003e-05,
-      "loss": 4.4359,
-      "step": 303
+      "loss": 4.255,
+      "step": 302
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 6e-05,
-      "loss": 4.296,
-      "step": 304
+      "loss": 4.2093,
+      "step": 303
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 6.02e-05,
-      "loss": 4.6571,
-      "step": 305
+      "loss": 4.2512,
+      "step": 304
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 6.04e-05,
-      "loss": 4.2272,
-      "step": 306
+      "loss": 4.0745,
+      "step": 305
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 6.06e-05,
-      "loss": 4.5005,
-      "step": 307
+      "loss": 4.143,
+      "step": 306
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 6.08e-05,
-      "loss": 5.0023,
-      "step": 308
+      "loss": 4.1732,
+      "step": 307
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 6.1e-05,
-      "loss": 4.5169,
-      "step": 309
+      "loss": 4.3464,
+      "step": 308
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 6.12e-05,
-      "loss": 4.5637,
-      "step": 310
+      "loss": 4.2193,
+      "step": 309
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 6.14e-05,
-      "loss": 4.4496,
-      "step": 311
+      "loss": 4.3655,
+      "step": 310
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 6.16e-05,
-      "loss": 4.3185,
-      "step": 312
+      "loss": 4.4293,
+      "step": 311
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 6.18e-05,
-      "loss": 4.4565,
-      "step": 313
+      "loss": 4.3158,
+      "step": 312
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 6.2e-05,
-      "loss": 4.4566,
-      "step": 314
+      "loss": 4.2305,
+      "step": 313
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 6.220000000000001e-05,
-      "loss": 4.5122,
-      "step": 315
+      "loss": 4.2545,
+      "step": 314
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 6.24e-05,
-      "loss": 4.2506,
-      "step": 316
+      "loss": 4.092,
+      "step": 315
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.35,
       "learning_rate": 6.26e-05,
-      "loss": 4.3773,
-      "step": 317
+      "loss": 4.3718,
+      "step": 316
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.280000000000001e-05,
-      "loss": 4.1339,
-      "step": 318
+      "loss": 4.1923,
+      "step": 317
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.3e-05,
-      "loss": 4.3889,
-      "step": 319
+      "loss": 4.1044,
+      "step": 318
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.32e-05,
-      "loss": 4.5719,
-      "step": 320
+      "loss": 4.2423,
+      "step": 319
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.340000000000001e-05,
-      "loss": 4.5925,
-      "step": 321
+      "loss": 4.1932,
+      "step": 320
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.36e-05,
-      "loss": 4.5706,
-      "step": 322
+      "loss": 4.3054,
+      "step": 321
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.38e-05,
-      "loss": 4.688,
-      "step": 323
+      "loss": 4.204,
+      "step": 322
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.400000000000001e-05,
-      "loss": 4.2869,
-      "step": 324
+      "loss": 4.6242,
+      "step": 323
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.42e-05,
-      "loss": 4.4203,
-      "step": 325
+      "loss": 4.4789,
+      "step": 324
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 6.440000000000001e-05,
-      "loss": 4.3779,
-      "step": 326
+      "loss": 4.4935,
+      "step": 325
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 6.460000000000001e-05,
-      "loss": 4.4374,
-      "step": 327
+      "loss": 4.3337,
+      "step": 326
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 6.48e-05,
-      "loss": 4.6249,
-      "step": 328
+      "loss": 4.1572,
+      "step": 327
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 4.4099,
-      "step": 329
+      "loss": 4.1932,
+      "step": 328
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.37,
       "learning_rate": 6.52e-05,
-      "loss": 4.5451,
-      "step": 330
+      "loss": 4.4071,
+      "step": 329
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 6.54e-05,
-      "loss": 4.3933,
-      "step": 331
+      "loss": 4.4865,
+      "step": 330
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 6.560000000000001e-05,
-      "loss": 4.5052,
-      "step": 332
+      "loss": 4.1701,
+      "step": 331
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 6.58e-05,
-      "loss": 4.8762,
-      "step": 333
+      "loss": 4.2422,
+      "step": 332
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 6.6e-05,
-      "loss": 4.5531,
-      "step": 334
+      "loss": 4.2503,
+      "step": 333
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 6.620000000000001e-05,
-      "loss": 4.4358,
-      "step": 335
+      "loss": 4.2858,
+      "step": 334
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.64e-05,
-      "loss": 4.4014,
-      "step": 336
+      "loss": 4.3979,
+      "step": 335
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.66e-05,
-      "loss": 4.3377,
-      "step": 337
+      "loss": 4.1882,
+      "step": 336
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.680000000000001e-05,
-      "loss": 4.1627,
-      "step": 338
+      "loss": 4.1267,
+      "step": 337
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.7e-05,
-      "loss": 4.2922,
-      "step": 339
+      "loss": 4.3292,
+      "step": 338
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.720000000000001e-05,
-      "loss": 4.2985,
-      "step": 340
+      "loss": 4.2356,
+      "step": 339
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.740000000000001e-05,
-      "loss": 4.3682,
-      "step": 341
+      "loss": 4.2885,
+      "step": 340
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.76e-05,
-      "loss": 4.4173,
-      "step": 342
+      "loss": 4.5698,
+      "step": 341
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.780000000000001e-05,
-      "loss": 4.7219,
-      "step": 343
+      "loss": 4.4765,
+      "step": 342
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 4.1218,
-      "step": 344
+      "loss": 4.2685,
+      "step": 343
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 6.82e-05,
-      "loss": 5.1615,
-      "step": 345
+      "loss": 4.4023,
+      "step": 344
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 6.840000000000001e-05,
-      "loss": 4.2031,
-      "step": 346
+      "loss": 4.401,
+      "step": 345
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 6.860000000000001e-05,
-      "loss": 4.7484,
-      "step": 347
+      "loss": 4.6416,
+      "step": 346
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.879999999999999e-05,
-      "loss": 4.7245,
-      "step": 348
+      "loss": 4.6245,
+      "step": 347
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.9e-05,
-      "loss": 5.2333,
-      "step": 349
+      "loss": 4.5174,
+      "step": 348
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.92e-05,
-      "loss": 4.3798,
-      "step": 350
+      "loss": 4.5499,
+      "step": 349
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.939999999999999e-05,
-      "loss": 4.5493,
-      "step": 351
+      "loss": 4.4764,
+      "step": 350
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.96e-05,
-      "loss": 4.5503,
-      "step": 352
+      "loss": 4.4091,
+      "step": 351
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.39,
       "learning_rate": 6.98e-05,
-      "loss": 4.3416,
-      "step": 353
+      "loss": 4.2577,
+      "step": 352
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7e-05,
-      "loss": 4.2769,
-      "step": 354
+      "loss": 4.2241,
+      "step": 353
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.02e-05,
-      "loss": 4.046,
-      "step": 355
+      "loss": 4.2012,
+      "step": 354
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.04e-05,
-      "loss": 4.4378,
-      "step": 356
+      "loss": 4.4662,
+      "step": 355
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.06e-05,
-      "loss": 4.587,
-      "step": 357
+      "loss": 4.1124,
+      "step": 356
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.08e-05,
-      "loss": 4.3935,
-      "step": 358
+      "loss": 4.2003,
+      "step": 357
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.1e-05,
-      "loss": 5.1244,
-      "step": 359
+      "loss": 4.2004,
+      "step": 358
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.12e-05,
-      "loss": 4.3047,
-      "step": 360
+      "loss": 4.2059,
+      "step": 359
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.14e-05,
-      "loss": 4.2926,
-      "step": 361
+      "loss": 4.3448,
+      "step": 360
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 7.16e-05,
-      "loss": 4.1062,
-      "step": 362
+      "loss": 4.243,
+      "step": 361
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 7.18e-05,
-      "loss": 4.4389,
-      "step": 363
+      "loss": 4.2397,
+      "step": 362
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 7.2e-05,
-      "loss": 4.6112,
-      "step": 364
+      "loss": 4.1738,
+      "step": 363
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.41,
       "learning_rate": 7.22e-05,
-      "loss": 4.6197,
-      "step": 365
+      "loss": 4.4315,
+      "step": 364
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.24e-05,
-      "loss": 4.7279,
-      "step": 366
+      "loss": 4.0795,
+      "step": 365
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.26e-05,
-      "loss": 4.3998,
-      "step": 367
+      "loss": 4.2753,
+      "step": 366
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.280000000000001e-05,
-      "loss": 4.2711,
-      "step": 368
+      "loss": 4.181,
+      "step": 367
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.3e-05,
-      "loss": 4.5869,
-      "step": 369
+      "loss": 4.239,
+      "step": 368
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.32e-05,
-      "loss": 4.3734,
-      "step": 370
+      "loss": 4.3883,
+      "step": 369
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 7.340000000000001e-05,
-      "loss": 4.327,
-      "step": 371
+      "loss": 4.2789,
+      "step": 370
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.36e-05,
-      "loss": 4.9089,
-      "step": 372
+      "loss": 4.1056,
+      "step": 371
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.38e-05,
-      "loss": 4.3104,
-      "step": 373
+      "loss": 4.4694,
+      "step": 372
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.4e-05,
-      "loss": 4.6238,
-      "step": 374
+      "loss": 4.2561,
+      "step": 373
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.42e-05,
-      "loss": 4.1308,
-      "step": 375
+      "loss": 4.398,
+      "step": 374
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.44e-05,
-      "loss": 4.1997,
-      "step": 376
+      "loss": 4.1645,
+      "step": 375
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.46e-05,
-      "loss": 4.2768,
-      "step": 377
+      "loss": 4.2328,
+      "step": 376
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.48e-05,
-      "loss": 4.1571,
-      "step": 378
+      "loss": 4.2445,
+      "step": 377
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 4.3162,
-      "step": 379
+      "loss": 4.3664,
+      "step": 378
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 7.52e-05,
-      "loss": 4.8261,
-      "step": 380
+      "loss": 4.2296,
+      "step": 379
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 7.54e-05,
-      "loss": 4.0751,
-      "step": 381
+      "loss": 4.4382,
+      "step": 380
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 7.560000000000001e-05,
-      "loss": 4.3197,
-      "step": 382
+      "loss": 4.3286,
+      "step": 381
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 7.58e-05,
-      "loss": 4.351,
-      "step": 383
+      "loss": 4.2781,
+      "step": 382
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.6e-05,
-      "loss": 4.445,
-      "step": 384
+      "loss": 4.4275,
+      "step": 383
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.620000000000001e-05,
-      "loss": 4.4942,
-      "step": 385
+      "loss": 4.2383,
+      "step": 384
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.64e-05,
-      "loss": 4.5034,
-      "step": 386
+      "loss": 4.2161,
+      "step": 385
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.66e-05,
-      "loss": 5.08,
-      "step": 387
+      "loss": 4.2606,
+      "step": 386
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.680000000000001e-05,
-      "loss": 4.151,
-      "step": 388
+      "loss": 4.2192,
+      "step": 387
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.43,
       "learning_rate": 7.7e-05,
-      "loss": 4.7433,
-      "step": 389
+      "loss": 4.222,
+      "step": 388
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.72e-05,
-      "loss": 5.5514,
-      "step": 390
+      "loss": 4.2535,
+      "step": 389
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.740000000000001e-05,
-      "loss": 4.2534,
-      "step": 391
+      "loss": 4.323,
+      "step": 390
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.76e-05,
-      "loss": 4.875,
-      "step": 392
+      "loss": 4.3979,
+      "step": 391
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.780000000000001e-05,
-      "loss": 4.6321,
-      "step": 393
+      "loss": 4.1583,
+      "step": 392
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.800000000000001e-05,
-      "loss": 4.7679,
-      "step": 394
+      "loss": 4.4448,
+      "step": 393
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.82e-05,
-      "loss": 4.8455,
-      "step": 395
+      "loss": 4.4023,
+      "step": 394
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.840000000000001e-05,
-      "loss": 5.3884,
-      "step": 396
+      "loss": 4.6777,
+      "step": 395
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 7.860000000000001e-05,
-      "loss": 4.4752,
-      "step": 397
+      "loss": 4.2592,
+      "step": 396
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 7.88e-05,
-      "loss": 4.9622,
-      "step": 398
+      "loss": 4.2254,
+      "step": 397
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 7.900000000000001e-05,
-      "loss": 6.9666,
-      "step": 399
+      "loss": 4.6933,
+      "step": 398
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 7.920000000000001e-05,
-      "loss": 4.9461,
-      "step": 400
+      "loss": 4.4564,
+      "step": 399
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 7.94e-05,
-      "loss": 4.4964,
-      "step": 401
+      "loss": 4.8916,
+      "step": 400
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 7.960000000000001e-05,
-      "loss": 4.2518,
-      "step": 402
+      "loss": 4.4665,
+      "step": 401
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 7.98e-05,
-      "loss": 4.3012,
-      "step": 403
+      "loss": 4.3327,
+      "step": 402
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 8e-05,
-      "loss": 4.5439,
-      "step": 404
+      "loss": 4.4033,
+      "step": 403
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 8.020000000000001e-05,
-      "loss": 4.2933,
-      "step": 405
+      "loss": 4.3885,
+      "step": 404
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 8.04e-05,
-      "loss": 4.5368,
-      "step": 406
+      "loss": 4.4113,
+      "step": 405
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.060000000000001e-05,
-      "loss": 4.3831,
-      "step": 407
+      "loss": 4.3084,
+      "step": 406
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.080000000000001e-05,
-      "loss": 4.8358,
-      "step": 408
+      "loss": 4.2218,
+      "step": 407
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.1e-05,
-      "loss": 4.6065,
-      "step": 409
+      "loss": 4.202,
+      "step": 408
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.120000000000001e-05,
-      "loss": 4.4729,
-      "step": 410
+      "loss": 4.4395,
+      "step": 409
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.14e-05,
-      "loss": 4.4149,
-      "step": 411
+      "loss": 4.3359,
+      "step": 410
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.16e-05,
-      "loss": 4.5185,
-      "step": 412
+      "loss": 4.3585,
+      "step": 411
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.18e-05,
-      "loss": 4.5688,
-      "step": 413
+      "loss": 4.2204,
+      "step": 412
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.2e-05,
-      "loss": 4.4212,
-      "step": 414
+      "loss": 4.4704,
+      "step": 413
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.22e-05,
-      "loss": 4.6792,
-      "step": 415
+      "loss": 4.0948,
+      "step": 414
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 8.24e-05,
-      "loss": 4.2985,
-      "step": 416
+      "loss": 4.2758,
+      "step": 415
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 8.26e-05,
-      "loss": 4.7082,
-      "step": 417
+      "loss": 4.3346,
+      "step": 416
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 8.28e-05,
-      "loss": 4.3855,
-      "step": 418
+      "loss": 4.1562,
+      "step": 417
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.47,
       "learning_rate": 8.3e-05,
-      "loss": 4.4232,
-      "step": 419
+      "loss": 4.3247,
+      "step": 418
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 8.32e-05,
-      "loss": 4.5219,
-      "step": 420
+      "loss": 4.2338,
+      "step": 419
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 8.34e-05,
-      "loss": 4.2427,
-      "step": 421
+      "loss": 4.268,
+      "step": 420
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 8.36e-05,
-      "loss": 4.6261,
-      "step": 422
+      "loss": 4.242,
+      "step": 421
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 8.38e-05,
-      "loss": 4.2834,
-      "step": 423
+      "loss": 4.4886,
+      "step": 422
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 8.4e-05,
-      "loss": 5.0802,
-      "step": 424
+      "loss": 4.4959,
+      "step": 423
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.42e-05,
-      "loss": 4.5631,
-      "step": 425
+      "loss": 4.4986,
+      "step": 424
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.44e-05,
-      "loss": 4.3811,
-      "step": 426
+      "loss": 4.4402,
+      "step": 425
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.46e-05,
-      "loss": 4.5995,
-      "step": 427
+      "loss": 4.4681,
+      "step": 426
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.48e-05,
-      "loss": 4.8287,
-      "step": 428
+      "loss": 4.1911,
+      "step": 427
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.5e-05,
-      "loss": 4.3153,
-      "step": 429
+      "loss": 4.4122,
+      "step": 428
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.52e-05,
-      "loss": 4.4325,
-      "step": 430
+      "loss": 4.2988,
+      "step": 429
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.54e-05,
-      "loss": 4.5596,
-      "step": 431
+      "loss": 4.3522,
+      "step": 430
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.560000000000001e-05,
-      "loss": 4.3889,
-      "step": 432
+      "loss": 4.5413,
+      "step": 431
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 8.58e-05,
-      "loss": 4.3607,
-      "step": 433
+      "loss": 4.3583,
+      "step": 432
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 8.6e-05,
-      "loss": 4.4551,
-      "step": 434
+      "loss": 4.0939,
+      "step": 433
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 8.620000000000001e-05,
-      "loss": 4.4696,
-      "step": 435
+      "loss": 4.6722,
+      "step": 434
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 8.64e-05,
-      "loss": 4.091,
-      "step": 436
+      "loss": 4.2455,
+      "step": 435
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 8.66e-05,
-      "loss": 4.506,
-      "step": 437
+      "loss": 4.4837,
+      "step": 436
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 8.680000000000001e-05,
-      "loss": 4.5829,
-      "step": 438
+      "loss": 4.5514,
+      "step": 437
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 8.7e-05,
-      "loss": 4.2036,
-      "step": 439
+      "loss": 4.3723,
+      "step": 438
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 8.72e-05,
-      "loss": 4.5499,
-      "step": 440
+      "loss": 4.2979,
+      "step": 439
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 8.740000000000001e-05,
-      "loss": 4.5629,
-      "step": 441
+      "loss": 4.3634,
+      "step": 440
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.49,
       "learning_rate": 8.76e-05,
-      "loss": 4.9256,
-      "step": 442
+      "loss": 4.6395,
+      "step": 441
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.78e-05,
-      "loss": 4.4276,
-      "step": 443
+      "loss": 4.2519,
+      "step": 442
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.800000000000001e-05,
-      "loss": 4.8103,
-      "step": 444
+      "loss": 4.2913,
+      "step": 443
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.82e-05,
-      "loss": 4.4369,
-      "step": 445
+      "loss": 4.7257,
+      "step": 444
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.840000000000001e-05,
-      "loss": 5.8002,
-      "step": 446
-    },
+      "loss": 4.3425,
+      "step": 445
+    },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.86e-05,
-      "loss": 5.4151,
-      "step": 447
+      "loss": 4.1172,
+      "step": 446
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.88e-05,
-      "loss": 5.2083,
-      "step": 448
+      "loss": 4.3683,
+      "step": 447
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.900000000000001e-05,
-      "loss": 5.5146,
-      "step": 449
+      "loss": 4.5756,
+      "step": 448
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.92e-05,
-      "loss": 4.7396,
-      "step": 450
+      "loss": 4.9532,
+      "step": 449
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 8.94e-05,
-      "loss": 4.2002,
-      "step": 451
+      "loss": 5.6362,
+      "step": 450
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 8.960000000000001e-05,
-      "loss": 4.4459,
-      "step": 452
+      "loss": 4.3291,
+      "step": 451
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 8.98e-05,
-      "loss": 5.0119,
-      "step": 453
+      "loss": 4.4704,
+      "step": 452
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.51,
       "learning_rate": 9e-05,
-      "loss": 4.1807,
-      "step": 454
+      "loss": 4.5678,
+      "step": 453
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.020000000000001e-05,
-      "loss": 4.2995,
-      "step": 455
+      "loss": 4.2346,
+      "step": 454
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.04e-05,
-      "loss": 4.3197,
-      "step": 456
+      "loss": 4.5092,
+      "step": 455
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.06e-05,
-      "loss": 4.3958,
-      "step": 457
+      "loss": 4.3366,
+      "step": 456
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.080000000000001e-05,
-      "loss": 4.1961,
-      "step": 458
+      "loss": 4.3663,
+      "step": 457
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.1e-05,
-      "loss": 4.1307,
-      "step": 459
+      "loss": 4.3834,
+      "step": 458
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 9.120000000000001e-05,
-      "loss": 4.5432,
-      "step": 460
+      "loss": 4.3779,
+      "step": 459
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.140000000000001e-05,
-      "loss": 5.0377,
-      "step": 461
+      "loss": 4.5347,
+      "step": 460
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.16e-05,
-      "loss": 4.1762,
-      "step": 462
+      "loss": 4.2667,
+      "step": 461
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.180000000000001e-05,
-      "loss": 4.6035,
-      "step": 463
+      "loss": 4.4051,
+      "step": 462
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.200000000000001e-05,
-      "loss": 4.2628,
-      "step": 464
+      "loss": 4.3178,
+      "step": 463
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.22e-05,
-      "loss": 4.2222,
-      "step": 465
+      "loss": 4.1682,
+      "step": 464
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.240000000000001e-05,
-      "loss": 4.594,
-      "step": 466
+      "loss": 4.2923,
+      "step": 465
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.260000000000001e-05,
-      "loss": 4.3893,
-      "step": 467
+      "loss": 4.6037,
+      "step": 466
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.28e-05,
-      "loss": 4.775,
-      "step": 468
+      "loss": 4.2776,
+      "step": 467
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 9.300000000000001e-05,
-      "loss": 4.3595,
-      "step": 469
+      "loss": 4.2357,
+      "step": 468
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 9.320000000000002e-05,
-      "loss": 4.4091,
-      "step": 470
+      "loss": 4.3306,
+      "step": 469
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 9.340000000000001e-05,
-      "loss": 4.6583,
-      "step": 471
+      "loss": 4.2949,
+      "step": 470
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 9.360000000000001e-05,
-      "loss": 4.3583,
-      "step": 472
+      "loss": 4.1422,
+      "step": 471
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.38e-05,
-      "loss": 4.2611,
-      "step": 473
+      "loss": 4.2664,
+      "step": 472
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.4e-05,
-      "loss": 4.7104,
-      "step": 474
+      "loss": 5.0999,
+      "step": 473
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.42e-05,
-      "loss": 4.566,
-      "step": 475
+      "loss": 4.4737,
+      "step": 474
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.44e-05,
-      "loss": 4.5703,
-      "step": 476
+      "loss": 4.3051,
+      "step": 475
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.46e-05,
-      "loss": 4.7316,
-      "step": 477
+      "loss": 4.6362,
+      "step": 476
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 9.48e-05,
-      "loss": 4.4144,
-      "step": 478
+      "loss": 4.5125,
+      "step": 477
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.5e-05,
-      "loss": 4.5456,
-      "step": 479
+      "loss": 4.2002,
+      "step": 478
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.52e-05,
-      "loss": 4.7028,
-      "step": 480
+      "loss": 4.1203,
+      "step": 479
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.54e-05,
-      "loss": 4.2501,
-      "step": 481
+      "loss": 4.3217,
+      "step": 480
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.56e-05,
-      "loss": 4.7512,
-      "step": 482
+      "loss": 4.2459,
+      "step": 481
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.58e-05,
-      "loss": 4.6424,
-      "step": 483
+      "loss": 4.1917,
+      "step": 482
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.6e-05,
-      "loss": 4.3812,
-      "step": 484
+      "loss": 4.3971,
+      "step": 483
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.620000000000001e-05,
-      "loss": 4.9182,
-      "step": 485
+      "loss": 4.3083,
+      "step": 484
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.64e-05,
-      "loss": 5.1119,
-      "step": 486
+      "loss": 4.4234,
+      "step": 485
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 9.66e-05,
-      "loss": 4.5277,
-      "step": 487
+      "loss": 4.198,
+      "step": 486
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 9.680000000000001e-05,
-      "loss": 4.6888,
-      "step": 488
+      "loss": 4.3167,
+      "step": 487
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 9.7e-05,
-      "loss": 4.8445,
-      "step": 489
+      "loss": 4.1635,
+      "step": 488
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 9.72e-05,
-      "loss": 4.5293,
-      "step": 490
+      "loss": 4.3927,
+      "step": 489
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.74e-05,
-      "loss": 4.4639,
-      "step": 491
+      "loss": 4.3937,
+      "step": 490
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.76e-05,
-      "loss": 4.7523,
-      "step": 492
+      "loss": 4.2244,
+      "step": 491
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.78e-05,
-      "loss": 4.51,
-      "step": 493
+      "loss": 4.5039,
+      "step": 492
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.8e-05,
-      "loss": 5.2443,
-      "step": 494
+      "loss": 4.8541,
+      "step": 493
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.82e-05,
-      "loss": 5.097,
-      "step": 495
+      "loss": 4.2893,
+      "step": 494
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 9.84e-05,
-      "loss": 4.491,
-      "step": 496
+      "loss": 4.6025,
+      "step": 495
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 9.86e-05,
-      "loss": 4.2666,
-      "step": 497
+      "loss": 4.1377,
+      "step": 496
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 9.88e-05,
-      "loss": 4.7748,
-      "step": 498
+      "loss": 4.094,
+      "step": 497
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 9.900000000000001e-05,
-      "loss": 4.8976,
-      "step": 499
+      "loss": 4.8325,
+      "step": 498
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 9.92e-05,
-      "loss": 4.2675,
+      "loss": 4.6663,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.94e-05,
+      "loss": 4.1605,
       "step": 500
     },
     {
-      "epoch": 0.28,
-      "eval_loss": 4.673245429992676,
-      "eval_runtime": 910.6835,
-      "eval_samples_per_second": 2.901,
-      "eval_steps_per_second": 0.363,
-      "eval_wer": 1.7895676318921063,
+      "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": 0.28,
-      "learning_rate": 9.94e-05,
-      "loss": 4.9698,
+      "epoch": 0.56,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 4.6864,
       "step": 501
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 4.265,
+      "epoch": 0.56,
+      "learning_rate": 9.98e-05,
+      "loss": 4.2069,
       "step": 502
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.98e-05,
-      "loss": 4.439,
+      "epoch": 0.56,
+      "learning_rate": 0.0001,
+      "loss": 4.6738,
       "step": 503
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0001,
-      "loss": 4.5564,
+      "epoch": 0.57,
+      "learning_rate": 9.974489795918368e-05,
+      "loss": 4.4791,
       "step": 504
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.992211838006231e-05,
-      "loss": 4.8987,
+      "epoch": 0.57,
+      "learning_rate": 9.948979591836736e-05,
+      "loss": 4.1978,
       "step": 505
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.984423676012462e-05,
-      "loss": 4.4782,
+      "epoch": 0.57,
+      "learning_rate": 9.923469387755102e-05,
+      "loss": 4.3373,
       "step": 506
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.976635514018692e-05,
-      "loss": 4.2572,
+      "epoch": 0.57,
+      "learning_rate": 9.897959183673469e-05,
+      "loss": 4.4823,
       "step": 507
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.968847352024923e-05,
-      "loss": 4.3361,
+      "epoch": 0.57,
+      "learning_rate": 9.872448979591837e-05,
+      "loss": 4.6696,
       "step": 508
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.961059190031154e-05,
-      "loss": 5.0592,
+      "epoch": 0.57,
+      "learning_rate": 9.846938775510204e-05,
+      "loss": 4.3015,
       "step": 509
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.953271028037385e-05,
-      "loss": 4.7226,
+      "epoch": 0.57,
+      "learning_rate": 9.821428571428572e-05,
+      "loss": 4.6368,
       "step": 510
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.945482866043614e-05,
-      "loss": 4.5018,
+      "epoch": 0.57,
+      "learning_rate": 9.79591836734694e-05,
+      "loss": 4.3094,
       "step": 511
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.937694704049845e-05,
-      "loss": 4.8196,
+      "epoch": 0.57,
+      "learning_rate": 9.770408163265307e-05,
+      "loss": 4.3088,
       "step": 512
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.929906542056076e-05,
-      "loss": 4.6867,
+      "epoch": 0.58,
+      "learning_rate": 9.744897959183674e-05,
+      "loss": 5.0756,
       "step": 513
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.922118380062307e-05,
-      "loss": 4.2051,
+      "epoch": 0.58,
+      "learning_rate": 9.719387755102042e-05,
+      "loss": 4.4334,
       "step": 514
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.914330218068536e-05,
-      "loss": 4.2933,
+      "epoch": 0.58,
+      "learning_rate": 9.693877551020408e-05,
+      "loss": 4.5505,
       "step": 515
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.906542056074767e-05,
-      "loss": 5.0657,
+      "epoch": 0.58,
+      "learning_rate": 9.668367346938776e-05,
+      "loss": 4.2511,
       "step": 516
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.898753894080998e-05,
-      "loss": 4.3401,
+      "epoch": 0.58,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 4.4093,
       "step": 517
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.890965732087229e-05,
-      "loss": 4.8184,
+      "epoch": 0.58,
+      "learning_rate": 9.617346938775511e-05,
+      "loss": 4.3096,
       "step": 518
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.883177570093459e-05,
-      "loss": 4.5244,
+      "epoch": 0.58,
+      "learning_rate": 9.591836734693878e-05,
+      "loss": 4.5489,
       "step": 519
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.87538940809969e-05,
-      "loss": 4.3437,
+      "epoch": 0.58,
+      "learning_rate": 9.566326530612246e-05,
+      "loss": 4.4747,
       "step": 520
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.867601246105919e-05,
-      "loss": 4.1295,
+      "epoch": 0.58,
+      "learning_rate": 9.540816326530613e-05,
+      "loss": 4.274,
       "step": 521
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.85981308411215e-05,
-      "loss": 4.4826,
+      "epoch": 0.59,
+      "learning_rate": 9.515306122448981e-05,
+      "loss": 4.2253,
       "step": 522
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.852024922118381e-05,
-      "loss": 4.5952,
+      "epoch": 0.59,
+      "learning_rate": 9.489795918367348e-05,
+      "loss": 4.2597,
       "step": 523
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.84423676012461e-05,
-      "loss": 4.693,
+      "epoch": 0.59,
+      "learning_rate": 9.464285714285715e-05,
+      "loss": 4.5888,
       "step": 524
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.836448598130841e-05,
-      "loss": 6.0652,
+      "epoch": 0.59,
+      "learning_rate": 9.438775510204082e-05,
+      "loss": 4.4988,
       "step": 525
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.828660436137072e-05,
-      "loss": 4.5477,
+      "epoch": 0.59,
+      "learning_rate": 9.41326530612245e-05,
+      "loss": 4.2755,
       "step": 526
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.820872274143302e-05,
-      "loss": 4.7943,
+      "epoch": 0.59,
+      "learning_rate": 9.387755102040817e-05,
+      "loss": 4.4353,
       "step": 527
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.813084112149533e-05,
-      "loss": 4.1771,
+      "epoch": 0.59,
+      "learning_rate": 9.362244897959183e-05,
+      "loss": 4.3746,
       "step": 528
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.805295950155764e-05,
-      "loss": 4.5022,
+      "epoch": 0.59,
+      "learning_rate": 9.336734693877551e-05,
+      "loss": 4.3393,
       "step": 529
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.797507788161995e-05,
-      "loss": 5.028,
+      "epoch": 0.59,
+      "learning_rate": 9.311224489795918e-05,
+      "loss": 4.5366,
       "step": 530
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.789719626168224e-05,
-      "loss": 4.2462,
+      "epoch": 0.6,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 4.582,
       "step": 531
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.781931464174455e-05,
-      "loss": 4.1729,
+      "epoch": 0.6,
+      "learning_rate": 9.260204081632653e-05,
+      "loss": 4.1169,
       "step": 532
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.774143302180686e-05,
-      "loss": 4.143,
+      "epoch": 0.6,
+      "learning_rate": 9.234693877551021e-05,
+      "loss": 4.3229,
       "step": 533
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.766355140186917e-05,
-      "loss": 4.1541,
+      "epoch": 0.6,
+      "learning_rate": 9.209183673469388e-05,
+      "loss": 4.2116,
       "step": 534
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.758566978193146e-05,
-      "loss": 4.7219,
+      "epoch": 0.6,
+      "learning_rate": 9.183673469387756e-05,
+      "loss": 4.5244,
       "step": 535
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.750778816199377e-05,
-      "loss": 4.4498,
+      "epoch": 0.6,
+      "learning_rate": 9.158163265306124e-05,
+      "loss": 4.4066,
       "step": 536
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.742990654205608e-05,
-      "loss": 4.3454,
+      "epoch": 0.6,
+      "learning_rate": 9.13265306122449e-05,
+      "loss": 4.5668,
       "step": 537
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.735202492211839e-05,
-      "loss": 4.7287,
+      "epoch": 0.6,
+      "learning_rate": 9.107142857142857e-05,
+      "loss": 4.4726,
       "step": 538
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.727414330218069e-05,
-      "loss": 4.5,
+      "epoch": 0.6,
+      "learning_rate": 9.081632653061225e-05,
+      "loss": 4.2823,
       "step": 539
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.7196261682243e-05,
-      "loss": 4.7541,
+      "epoch": 0.61,
+      "learning_rate": 9.056122448979592e-05,
+      "loss": 4.3757,
       "step": 540
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.71183800623053e-05,
-      "loss": 4.9049,
+      "epoch": 0.61,
+      "learning_rate": 9.030612244897958e-05,
+      "loss": 4.3945,
       "step": 541
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.704049844236761e-05,
-      "loss": 4.5044,
+      "epoch": 0.61,
+      "learning_rate": 9.005102040816327e-05,
+      "loss": 4.4137,
       "step": 542
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.696261682242991e-05,
-      "loss": 6.4645,
+      "epoch": 0.61,
+      "learning_rate": 8.979591836734695e-05,
+      "loss": 4.5504,
       "step": 543
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.688473520249222e-05,
-      "loss": 4.5994,
+      "epoch": 0.61,
+      "learning_rate": 8.954081632653062e-05,
+      "loss": 4.6919,
       "step": 544
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.680685358255453e-05,
-      "loss": 4.2292,
+      "epoch": 0.61,
+      "learning_rate": 8.92857142857143e-05,
+      "loss": 4.677,
       "step": 545
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.672897196261684e-05,
-      "loss": 5.7364,
+      "epoch": 0.61,
+      "learning_rate": 8.903061224489796e-05,
+      "loss": 4.5456,
       "step": 546
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.665109034267913e-05,
-      "loss": 5.196,
+      "epoch": 0.61,
+      "learning_rate": 8.877551020408164e-05,
+      "loss": 5.0684,
       "step": 547
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.657320872274144e-05,
-      "loss": 5.3241,
+      "epoch": 0.61,
+      "learning_rate": 8.852040816326531e-05,
+      "loss": 6.0851,
       "step": 548
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.649532710280375e-05,
-      "loss": 5.1062,
+      "epoch": 0.62,
+      "learning_rate": 8.826530612244899e-05,
+      "loss": 4.369,
       "step": 549
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.641744548286606e-05,
-      "loss": 4.6928,
+      "epoch": 0.62,
+      "learning_rate": 8.801020408163265e-05,
+      "loss": 5.2507,
       "step": 550
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.633956386292835e-05,
-      "loss": 4.7138,
+      "epoch": 0.62,
+      "learning_rate": 8.775510204081632e-05,
+      "loss": 4.4075,
       "step": 551
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.626168224299066e-05,
-      "loss": 5.1959,
+      "epoch": 0.62,
+      "learning_rate": 8.75e-05,
+      "loss": 4.5339,
       "step": 552
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.618380062305297e-05,
-      "loss": 5.0342,
+      "epoch": 0.62,
+      "learning_rate": 8.724489795918367e-05,
+      "loss": 4.5553,
       "step": 553
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.610591900311527e-05,
-      "loss": 4.5409,
+      "epoch": 0.62,
+      "learning_rate": 8.698979591836735e-05,
+      "loss": 4.2438,
       "step": 554
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.602803738317758e-05,
-      "loss": 4.4415,
+      "epoch": 0.62,
+      "learning_rate": 8.673469387755102e-05,
+      "loss": 4.3284,
       "step": 555
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.595015576323989e-05,
-      "loss": 4.7392,
+      "epoch": 0.62,
+      "learning_rate": 8.64795918367347e-05,
+      "loss": 4.5701,
       "step": 556
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.587227414330218e-05,
-      "loss": 4.4739,
+      "epoch": 0.62,
+      "learning_rate": 8.622448979591838e-05,
+      "loss": 4.2063,
       "step": 557
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.579439252336449e-05,
-      "loss": 4.3078,
+      "epoch": 0.63,
+      "learning_rate": 8.596938775510205e-05,
+      "loss": 4.1729,
       "step": 558
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.571651090342679e-05,
-      "loss": 4.1923,
+      "epoch": 0.63,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 4.3359,
       "step": 559
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.56386292834891e-05,
-      "loss": 4.2617,
+      "epoch": 0.63,
+      "learning_rate": 8.545918367346939e-05,
+      "loss": 4.2974,
       "step": 560
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.55607476635514e-05,
-      "loss": 4.7006,
+      "epoch": 0.63,
+      "learning_rate": 8.520408163265306e-05,
+      "loss": 4.4944,
       "step": 561
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.548286604361371e-05,
-      "loss": 4.9497,
+      "epoch": 0.63,
+      "learning_rate": 8.494897959183674e-05,
+      "loss": 4.2742,
       "step": 562
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.540498442367601e-05,
-      "loss": 4.1691,
+      "epoch": 0.63,
+      "learning_rate": 8.469387755102041e-05,
+      "loss": 4.3104,
       "step": 563
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.532710280373832e-05,
-      "loss": 4.8095,
+      "epoch": 0.63,
+      "learning_rate": 8.443877551020409e-05,
+      "loss": 4.2998,
       "step": 564
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.524922118380063e-05,
-      "loss": 4.9433,
+      "epoch": 0.63,
+      "learning_rate": 8.418367346938776e-05,
+      "loss": 4.3185,
       "step": 565
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.517133956386294e-05,
-      "loss": 4.3185,
+      "epoch": 0.63,
+      "learning_rate": 8.392857142857144e-05,
+      "loss": 4.4354,
       "step": 566
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.509345794392523e-05,
-      "loss": 4.338,
+      "epoch": 0.64,
+      "learning_rate": 8.367346938775511e-05,
+      "loss": 4.3363,
       "step": 567
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.501557632398754e-05,
-      "loss": 4.7744,
+      "epoch": 0.64,
+      "learning_rate": 8.341836734693878e-05,
+      "loss": 4.2593,
       "step": 568
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.493769470404985e-05,
-      "loss": 4.3619,
+      "epoch": 0.64,
+      "learning_rate": 8.316326530612245e-05,
+      "loss": 4.3758,
       "step": 569
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.485981308411216e-05,
-      "loss": 4.8806,
+      "epoch": 0.64,
+      "learning_rate": 8.290816326530613e-05,
+      "loss": 4.3491,
       "step": 570
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.478193146417445e-05,
-      "loss": 4.4706,
+      "epoch": 0.64,
+      "learning_rate": 8.26530612244898e-05,
+      "loss": 4.2382,
       "step": 571
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.470404984423676e-05,
-      "loss": 4.6696,
+      "epoch": 0.64,
+      "learning_rate": 8.239795918367348e-05,
+      "loss": 4.3785,
       "step": 572
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.462616822429907e-05,
-      "loss": 5.0274,
+      "epoch": 0.64,
+      "learning_rate": 8.214285714285714e-05,
+      "loss": 4.2177,
       "step": 573
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.454828660436138e-05,
-      "loss": 4.1647,
+      "epoch": 0.64,
+      "learning_rate": 8.188775510204081e-05,
+      "loss": 4.1685,
       "step": 574
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.447040498442368e-05,
-      "loss": 4.2467,
+      "epoch": 0.64,
+      "learning_rate": 8.163265306122449e-05,
+      "loss": 4.6721,
       "step": 575
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.439252336448599e-05,
-      "loss": 4.5245,
+      "epoch": 0.65,
+      "learning_rate": 8.137755102040817e-05,
+      "loss": 4.2446,
       "step": 576
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.43146417445483e-05,
-      "loss": 4.4699,
+      "epoch": 0.65,
+      "learning_rate": 8.112244897959184e-05,
+      "loss": 4.3527,
       "step": 577
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.42367601246106e-05,
-      "loss": 4.3694,
+      "epoch": 0.65,
+      "learning_rate": 8.086734693877552e-05,
+      "loss": 4.585,
       "step": 578
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.41588785046729e-05,
-      "loss": 4.6329,
+      "epoch": 0.65,
+      "learning_rate": 8.061224489795919e-05,
+      "loss": 3.9455,
       "step": 579
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.408099688473521e-05,
-      "loss": 4.967,
+      "epoch": 0.65,
+      "learning_rate": 8.035714285714287e-05,
+      "loss": 4.9334,
       "step": 580
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.400311526479752e-05,
-      "loss": 4.2504,
+      "epoch": 0.65,
+      "learning_rate": 8.010204081632653e-05,
+      "loss": 4.4063,
       "step": 581
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.392523364485983e-05,
-      "loss": 4.716,
+      "epoch": 0.65,
+      "learning_rate": 7.98469387755102e-05,
+      "loss": 4.2461,
       "step": 582
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.384735202492212e-05,
-      "loss": 4.5056,
+      "epoch": 0.65,
+      "learning_rate": 7.959183673469388e-05,
+      "loss": 4.7348,
       "step": 583
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.376947040498443e-05,
-      "loss": 4.6983,
+      "epoch": 0.65,
+      "learning_rate": 7.933673469387755e-05,
+      "loss": 4.4816,
       "step": 584
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.369158878504674e-05,
-      "loss": 5.2671,
+      "epoch": 0.66,
+      "learning_rate": 7.908163265306123e-05,
+      "loss": 4.5978,
       "step": 585
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.361370716510905e-05,
-      "loss": 4.2643,
+      "epoch": 0.66,
+      "learning_rate": 7.882653061224489e-05,
+      "loss": 4.4656,
       "step": 586
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.353582554517135e-05,
-      "loss": 4.7544,
+      "epoch": 0.66,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 4.4803,
       "step": 587
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.345794392523365e-05,
-      "loss": 4.4757,
+      "epoch": 0.66,
+      "learning_rate": 7.831632653061226e-05,
+      "loss": 4.427,
       "step": 588
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.338006230529596e-05,
-      "loss": 4.8137,
+      "epoch": 0.66,
+      "learning_rate": 7.806122448979593e-05,
+      "loss": 4.1964,
       "step": 589
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.330218068535826e-05,
-      "loss": 4.5502,
+      "epoch": 0.66,
+      "learning_rate": 7.780612244897959e-05,
+      "loss": 4.2802,
       "step": 590
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.322429906542057e-05,
-      "loss": 4.4259,
+      "epoch": 0.66,
+      "learning_rate": 7.755102040816327e-05,
+      "loss": 4.3095,
       "step": 591
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.314641744548286e-05,
-      "loss": 4.8757,
+      "epoch": 0.66,
+      "learning_rate": 7.729591836734694e-05,
+      "loss": 4.5677,
       "step": 592
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.306853582554517e-05,
-      "loss": 4.674,
+      "epoch": 0.66,
+      "learning_rate": 7.704081632653062e-05,
+      "loss": 4.7527,
       "step": 593
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.299065420560748e-05,
-      "loss": 4.5856,
+      "epoch": 0.67,
+      "learning_rate": 7.67857142857143e-05,
+      "loss": 4.3602,
       "step": 594
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.291277258566978e-05,
-      "loss": 4.3733,
+      "epoch": 0.67,
+      "learning_rate": 7.653061224489796e-05,
+      "loss": 4.4132,
       "step": 595
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.283489096573209e-05,
-      "loss": 4.4966,
+      "epoch": 0.67,
+      "learning_rate": 7.627551020408163e-05,
+      "loss": 4.1445,
       "step": 596
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.27570093457944e-05,
-      "loss": 5.9475,
+      "epoch": 0.67,
+      "learning_rate": 7.60204081632653e-05,
+      "loss": 4.2382,
       "step": 597
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.26791277258567e-05,
-      "loss": 5.8244,
+      "epoch": 0.67,
+      "learning_rate": 7.576530612244898e-05,
+      "loss": 4.5613,
       "step": 598
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.2601246105919e-05,
-      "loss": 4.8917,
+      "epoch": 0.67,
+      "learning_rate": 7.551020408163266e-05,
+      "loss": 4.5066,
       "step": 599
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.252336448598131e-05,
-      "loss": 4.9375,
+      "epoch": 0.67,
+      "learning_rate": 7.525510204081633e-05,
+      "loss": 4.1874,
       "step": 600
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.244548286604362e-05,
-      "loss": 4.076,
+      "epoch": 0.67,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.3602,
       "step": 601
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.236760124610593e-05,
-      "loss": 4.6591,
+      "epoch": 0.67,
+      "learning_rate": 7.474489795918368e-05,
+      "loss": 4.3446,
       "step": 602
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.228971962616822e-05,
-      "loss": 4.2895,
+      "epoch": 0.68,
+      "learning_rate": 7.448979591836736e-05,
+      "loss": 4.4451,
       "step": 603
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.221183800623053e-05,
-      "loss": 4.3872,
+      "epoch": 0.68,
+      "learning_rate": 7.423469387755102e-05,
+      "loss": 4.3278,
       "step": 604
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.213395638629284e-05,
-      "loss": 4.3973,
+      "epoch": 0.68,
+      "learning_rate": 7.39795918367347e-05,
+      "loss": 4.2441,
       "step": 605
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.205607476635515e-05,
-      "loss": 4.1871,
+      "epoch": 0.68,
+      "learning_rate": 7.372448979591837e-05,
+      "loss": 4.3496,
       "step": 606
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.197819314641744e-05,
-      "loss": 4.8444,
+      "epoch": 0.68,
+      "learning_rate": 7.346938775510205e-05,
+      "loss": 4.425,
       "step": 607
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.190031152647975e-05,
-      "loss": 4.25,
+      "epoch": 0.68,
+      "learning_rate": 7.321428571428571e-05,
+      "loss": 4.2756,
       "step": 608
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.182242990654206e-05,
-      "loss": 4.7297,
+      "epoch": 0.68,
+      "learning_rate": 7.29591836734694e-05,
+      "loss": 4.2437,
       "step": 609
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.174454828660437e-05,
-      "loss": 4.2887,
+      "epoch": 0.68,
+      "learning_rate": 7.270408163265307e-05,
+      "loss": 4.2779,
       "step": 610
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.166666666666667e-05,
-      "loss": 4.7428,
+      "epoch": 0.68,
+      "learning_rate": 7.244897959183675e-05,
+      "loss": 4.2593,
       "step": 611
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.158878504672898e-05,
-      "loss": 4.3427,
+      "epoch": 0.69,
+      "learning_rate": 7.219387755102042e-05,
+      "loss": 4.4021,
       "step": 612
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.151090342679129e-05,
-      "loss": 4.4945,
+      "epoch": 0.69,
+      "learning_rate": 7.193877551020408e-05,
+      "loss": 4.3809,
       "step": 613
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.14330218068536e-05,
-      "loss": 4.6053,
+      "epoch": 0.69,
+      "learning_rate": 7.168367346938776e-05,
+      "loss": 4.4197,
       "step": 614
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.135514018691589e-05,
-      "loss": 4.5843,
+      "epoch": 0.69,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 4.2543,
       "step": 615
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.12772585669782e-05,
-      "loss": 4.189,
+      "epoch": 0.69,
+      "learning_rate": 7.117346938775511e-05,
+      "loss": 4.2681,
       "step": 616
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.119937694704051e-05,
-      "loss": 4.8201,
+      "epoch": 0.69,
+      "learning_rate": 7.091836734693877e-05,
+      "loss": 4.2117,
       "step": 617
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.112149532710282e-05,
-      "loss": 4.4516,
+      "epoch": 0.69,
+      "learning_rate": 7.066326530612245e-05,
+      "loss": 4.163,
       "step": 618
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.104361370716511e-05,
-      "loss": 4.255,
+      "epoch": 0.69,
+      "learning_rate": 7.040816326530612e-05,
+      "loss": 4.1755,
       "step": 619
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.096573208722742e-05,
-      "loss": 4.8702,
+      "epoch": 0.7,
+      "learning_rate": 7.01530612244898e-05,
+      "loss": 4.3606,
       "step": 620
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.088785046728973e-05,
-      "loss": 4.57,
+      "epoch": 0.7,
+      "learning_rate": 6.989795918367347e-05,
+      "loss": 4.2221,
       "step": 621
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.080996884735204e-05,
-      "loss": 4.8587,
+      "epoch": 0.7,
+      "learning_rate": 6.964285714285715e-05,
+      "loss": 4.1264,
       "step": 622
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.073208722741434e-05,
-      "loss": 4.5199,
+      "epoch": 0.7,
+      "learning_rate": 6.938775510204082e-05,
+      "loss": 4.3952,
       "step": 623
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.065420560747664e-05,
-      "loss": 5.3808,
+      "epoch": 0.7,
+      "learning_rate": 6.91326530612245e-05,
+      "loss": 4.2269,
       "step": 624
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.057632398753894e-05,
-      "loss": 4.6201,
+      "epoch": 0.7,
+      "learning_rate": 6.887755102040817e-05,
+      "loss": 4.608,
       "step": 625
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.049844236760125e-05,
-      "loss": 4.3285,
+      "epoch": 0.7,
+      "learning_rate": 6.862244897959184e-05,
+      "loss": 4.2958,
       "step": 626
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.042056074766356e-05,
-      "loss": 4.2365,
+      "epoch": 0.7,
+      "learning_rate": 6.836734693877551e-05,
+      "loss": 4.2362,
       "step": 627
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.034267912772585e-05,
-      "loss": 5.1293,
+      "epoch": 0.7,
+      "learning_rate": 6.811224489795919e-05,
+      "loss": 4.1795,
       "step": 628
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.026479750778816e-05,
-      "loss": 6.057,
+      "epoch": 0.71,
+      "learning_rate": 6.785714285714286e-05,
+      "loss": 4.1724,
       "step": 629
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.018691588785047e-05,
-      "loss": 4.2621,
+      "epoch": 0.71,
+      "learning_rate": 6.760204081632652e-05,
+      "loss": 4.5351,
       "step": 630
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.010903426791277e-05,
-      "loss": 4.5182,
+      "epoch": 0.71,
+      "learning_rate": 6.73469387755102e-05,
+      "loss": 4.3448,
       "step": 631
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.003115264797508e-05,
-      "loss": 4.8152,
+      "epoch": 0.71,
+      "learning_rate": 6.709183673469389e-05,
+      "loss": 4.2583,
       "step": 632
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.995327102803739e-05,
-      "loss": 4.5095,
+      "epoch": 0.71,
+      "learning_rate": 6.683673469387756e-05,
+      "loss": 4.266,
       "step": 633
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.98753894080997e-05,
-      "loss": 4.4015,
+      "epoch": 0.71,
+      "learning_rate": 6.658163265306124e-05,
+      "loss": 4.237,
       "step": 634
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.979750778816199e-05,
-      "loss": 4.3792,
+      "epoch": 0.71,
+      "learning_rate": 6.63265306122449e-05,
+      "loss": 4.4784,
       "step": 635
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.97196261682243e-05,
-      "loss": 4.2953,
+      "epoch": 0.71,
+      "learning_rate": 6.607142857142857e-05,
+      "loss": 4.3054,
       "step": 636
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.964174454828661e-05,
-      "loss": 4.4238,
+      "epoch": 0.71,
+      "learning_rate": 6.581632653061225e-05,
+      "loss": 4.2649,
       "step": 637
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.956386292834892e-05,
-      "loss": 4.1604,
+      "epoch": 0.72,
+      "learning_rate": 6.556122448979592e-05,
+      "loss": 4.1358,
       "step": 638
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.948598130841121e-05,
-      "loss": 4.5354,
+      "epoch": 0.72,
+      "learning_rate": 6.530612244897959e-05,
+      "loss": 4.3757,
       "step": 639
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.940809968847352e-05,
-      "loss": 5.3864,
+      "epoch": 0.72,
+      "learning_rate": 6.505102040816326e-05,
+      "loss": 4.0554,
       "step": 640
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.933021806853583e-05,
-      "loss": 4.9418,
+      "epoch": 0.72,
+      "learning_rate": 6.479591836734694e-05,
+      "loss": 4.2163,
       "step": 641
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.925233644859814e-05,
-      "loss": 4.69,
+      "epoch": 0.72,
+      "learning_rate": 6.454081632653061e-05,
+      "loss": 4.6037,
       "step": 642
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.917445482866044e-05,
-      "loss": 4.5578,
+      "epoch": 0.72,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 4.6982,
       "step": 643
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.909657320872274e-05,
-      "loss": 4.5504,
+      "epoch": 0.72,
+      "learning_rate": 6.403061224489796e-05,
+      "loss": 4.1454,
       "step": 644
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.901869158878505e-05,
-      "loss": 4.4431,
+      "epoch": 0.72,
+      "learning_rate": 6.377551020408164e-05,
+      "loss": 3.9796,
       "step": 645
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.894080996884736e-05,
-      "loss": 4.5409,
+      "epoch": 0.72,
+      "learning_rate": 6.352040816326531e-05,
+      "loss": 4.3531,
       "step": 646
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.886292834890966e-05,
-      "loss": 5.7369,
+      "epoch": 0.73,
+      "learning_rate": 6.326530612244899e-05,
+      "loss": 4.1176,
       "step": 647
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.878504672897197e-05,
-      "loss": 5.6865,
+      "epoch": 0.73,
+      "learning_rate": 6.301020408163265e-05,
+      "loss": 4.3527,
       "step": 648
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.870716510903428e-05,
-      "loss": 4.8577,
+      "epoch": 0.73,
+      "learning_rate": 6.275510204081633e-05,
+      "loss": 4.4182,
       "step": 649
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.862928348909659e-05,
-      "loss": 4.8717,
+      "epoch": 0.73,
+      "learning_rate": 6.25e-05,
+      "loss": 4.3515,
       "step": 650
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.855140186915888e-05,
-      "loss": 4.3227,
+      "epoch": 0.73,
+      "learning_rate": 6.224489795918368e-05,
+      "loss": 4.4742,
       "step": 651
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.847352024922119e-05,
-      "loss": 4.2694,
+      "epoch": 0.73,
+      "learning_rate": 6.198979591836735e-05,
+      "loss": 4.4929,
       "step": 652
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.83956386292835e-05,
-      "loss": 4.3063,
+      "epoch": 0.73,
+      "learning_rate": 6.173469387755101e-05,
+      "loss": 4.3374,
       "step": 653
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.831775700934581e-05,
-      "loss": 4.3785,
+      "epoch": 0.73,
+      "learning_rate": 6.14795918367347e-05,
+      "loss": 4.4383,
       "step": 654
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.82398753894081e-05,
-      "loss": 4.9765,
+      "epoch": 0.73,
+      "learning_rate": 6.122448979591838e-05,
+      "loss": 4.2959,
       "step": 655
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.816199376947041e-05,
-      "loss": 4.2823,
+      "epoch": 0.74,
+      "learning_rate": 6.0969387755102046e-05,
+      "loss": 4.4306,
       "step": 656
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.808411214953272e-05,
-      "loss": 4.2764,
+      "epoch": 0.74,
+      "learning_rate": 6.0714285714285715e-05,
+      "loss": 4.1845,
       "step": 657
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.800623052959502e-05,
-      "loss": 4.9974,
+      "epoch": 0.74,
+      "learning_rate": 6.045918367346939e-05,
+      "loss": 4.4121,
       "step": 658
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.792834890965733e-05,
-      "loss": 4.5588,
+      "epoch": 0.74,
+      "learning_rate": 6.0204081632653065e-05,
+      "loss": 4.1552,
       "step": 659
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.785046728971964e-05,
-      "loss": 4.3493,
+      "epoch": 0.74,
+      "learning_rate": 5.994897959183674e-05,
+      "loss": 4.4136,
       "step": 660
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.777258566978193e-05,
-      "loss": 5.0012,
+      "epoch": 0.74,
+      "learning_rate": 5.9693877551020416e-05,
+      "loss": 4.0537,
       "step": 661
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.769470404984424e-05,
-      "loss": 4.4923,
+      "epoch": 0.74,
+      "learning_rate": 5.9438775510204084e-05,
+      "loss": 4.0487,
       "step": 662
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.761682242990654e-05,
-      "loss": 4.396,
+      "epoch": 0.74,
+      "learning_rate": 5.918367346938776e-05,
+      "loss": 4.6006,
       "step": 663
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.753894080996884e-05,
-      "loss": 5.0483,
+      "epoch": 0.74,
+      "learning_rate": 5.8928571428571435e-05,
+      "loss": 4.1514,
       "step": 664
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.746105919003115e-05,
-      "loss": 4.4713,
+      "epoch": 0.75,
+      "learning_rate": 5.867346938775511e-05,
+      "loss": 4.0854,
       "step": 665
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.738317757009346e-05,
-      "loss": 5.0018,
+      "epoch": 0.75,
+      "learning_rate": 5.841836734693877e-05,
+      "loss": 4.3523,
       "step": 666
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.730529595015576e-05,
-      "loss": 4.401,
+      "epoch": 0.75,
+      "learning_rate": 5.816326530612245e-05,
+      "loss": 4.2261,
       "step": 667
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.722741433021807e-05,
-      "loss": 4.697,
+      "epoch": 0.75,
+      "learning_rate": 5.790816326530612e-05,
+      "loss": 4.3905,
       "step": 668
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.714953271028038e-05,
-      "loss": 4.2168,
+      "epoch": 0.75,
+      "learning_rate": 5.7653061224489805e-05,
+      "loss": 4.4202,
       "step": 669
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.707165109034269e-05,
-      "loss": 4.4951,
+      "epoch": 0.75,
+      "learning_rate": 5.739795918367348e-05,
+      "loss": 4.3253,
       "step": 670
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.699376947040498e-05,
-      "loss": 4.5363,
+      "epoch": 0.75,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 4.1013,
       "step": 671
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.691588785046729e-05,
-      "loss": 4.8743,
+      "epoch": 0.75,
+      "learning_rate": 5.688775510204082e-05,
+      "loss": 4.1175,
       "step": 672
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.68380062305296e-05,
-      "loss": 4.3071,
+      "epoch": 0.75,
+      "learning_rate": 5.663265306122449e-05,
+      "loss": 4.0785,
       "step": 673
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.676012461059191e-05,
-      "loss": 4.2854,
+      "epoch": 0.76,
+      "learning_rate": 5.637755102040817e-05,
+      "loss": 4.0308,
       "step": 674
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.66822429906542e-05,
-      "loss": 4.3462,
+      "epoch": 0.76,
+      "learning_rate": 5.6122448979591836e-05,
+      "loss": 4.0925,
       "step": 675
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.660436137071651e-05,
-      "loss": 4.3735,
+      "epoch": 0.76,
+      "learning_rate": 5.586734693877551e-05,
+      "loss": 4.323,
       "step": 676
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.652647975077882e-05,
-      "loss": 4.3374,
+      "epoch": 0.76,
+      "learning_rate": 5.561224489795919e-05,
+      "loss": 4.0947,
       "step": 677
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.644859813084113e-05,
-      "loss": 4.3014,
+      "epoch": 0.76,
+      "learning_rate": 5.535714285714286e-05,
+      "loss": 3.9544,
       "step": 678
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.637071651090343e-05,
-      "loss": 4.7161,
+      "epoch": 0.76,
+      "learning_rate": 5.510204081632653e-05,
+      "loss": 4.0468,
       "step": 679
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.629283489096574e-05,
-      "loss": 4.7852,
+      "epoch": 0.76,
+      "learning_rate": 5.4846938775510206e-05,
+      "loss": 4.1615,
       "step": 680
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.621495327102804e-05,
-      "loss": 4.4309,
+      "epoch": 0.76,
+      "learning_rate": 5.459183673469388e-05,
+      "loss": 4.8051,
       "step": 681
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.613707165109035e-05,
-      "loss": 4.4508,
+      "epoch": 0.76,
+      "learning_rate": 5.4336734693877556e-05,
+      "loss": 4.1085,
       "step": 682
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.605919003115265e-05,
-      "loss": 4.5658,
+      "epoch": 0.77,
+      "learning_rate": 5.408163265306123e-05,
+      "loss": 4.5086,
       "step": 683
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.598130841121496e-05,
-      "loss": 4.6858,
+      "epoch": 0.77,
+      "learning_rate": 5.382653061224489e-05,
+      "loss": 4.467,
       "step": 684
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.590342679127727e-05,
-      "loss": 4.901,
+      "epoch": 0.77,
+      "learning_rate": 5.3571428571428575e-05,
+      "loss": 4.4496,
       "step": 685
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.582554517133958e-05,
-      "loss": 4.7781,
+      "epoch": 0.77,
+      "learning_rate": 5.331632653061225e-05,
+      "loss": 4.2151,
       "step": 686
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.574766355140187e-05,
-      "loss": 4.3399,
+      "epoch": 0.77,
+      "learning_rate": 5.3061224489795926e-05,
+      "loss": 4.3281,
       "step": 687
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.566978193146418e-05,
-      "loss": 4.1825,
+      "epoch": 0.77,
+      "learning_rate": 5.280612244897959e-05,
+      "loss": 4.3739,
       "step": 688
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.559190031152649e-05,
-      "loss": 4.5355,
+      "epoch": 0.77,
+      "learning_rate": 5.255102040816326e-05,
+      "loss": 4.3452,
       "step": 689
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.55140186915888e-05,
-      "loss": 4.4293,
+      "epoch": 0.77,
+      "learning_rate": 5.229591836734694e-05,
+      "loss": 4.5481,
       "step": 690
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.54361370716511e-05,
-      "loss": 4.2052,
+      "epoch": 0.77,
+      "learning_rate": 5.2040816326530614e-05,
+      "loss": 4.084,
       "step": 691
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.53582554517134e-05,
-      "loss": 5.0693,
+      "epoch": 0.78,
+      "learning_rate": 5.1785714285714296e-05,
+      "loss": 4.304,
       "step": 692
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.528037383177571e-05,
-      "loss": 4.5691,
+      "epoch": 0.78,
+      "learning_rate": 5.153061224489796e-05,
+      "loss": 4.0724,
       "step": 693
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.520249221183801e-05,
-      "loss": 4.4294,
+      "epoch": 0.78,
+      "learning_rate": 5.127551020408163e-05,
+      "loss": 4.5427,
       "step": 694
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.512461059190032e-05,
-      "loss": 4.5869,
+      "epoch": 0.78,
+      "learning_rate": 5.102040816326531e-05,
+      "loss": 4.693,
       "step": 695
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.504672897196261e-05,
-      "loss": 4.5832,
+      "epoch": 0.78,
+      "learning_rate": 5.076530612244898e-05,
+      "loss": 4.3249,
       "step": 696
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.496884735202492e-05,
-      "loss": 4.0486,
+      "epoch": 0.78,
+      "learning_rate": 5.051020408163265e-05,
+      "loss": 4.4825,
       "step": 697
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.489096573208723e-05,
-      "loss": 4.4014,
+      "epoch": 0.78,
+      "learning_rate": 5.025510204081633e-05,
+      "loss": 4.2753,
       "step": 698
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.481308411214953e-05,
-      "loss": 5.2576,
+      "epoch": 0.78,
+      "learning_rate": 5e-05,
+      "loss": 4.5013,
       "step": 699
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.473520249221184e-05,
-      "loss": 4.313,
+      "epoch": 0.78,
+      "learning_rate": 4.974489795918368e-05,
+      "loss": 4.4463,
       "step": 700
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.465732087227414e-05,
-      "loss": 4.1705,
+      "epoch": 0.79,
+      "learning_rate": 4.9489795918367346e-05,
+      "loss": 4.0471,
       "step": 701
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.457943925233645e-05,
-      "loss": 4.3267,
+      "epoch": 0.79,
+      "learning_rate": 4.923469387755102e-05,
+      "loss": 4.2305,
       "step": 702
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.450155763239875e-05,
-      "loss": 4.5667,
+      "epoch": 0.79,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 4.1357,
       "step": 703
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 8.442367601246106e-05,
-      "loss": 5.0232,
+      "epoch": 0.79,
+      "learning_rate": 4.872448979591837e-05,
+      "loss": 4.0259,
       "step": 704
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.434579439252337e-05,
-      "loss": 5.0489,
+      "epoch": 0.79,
+      "learning_rate": 4.846938775510204e-05,
+      "loss": 4.2383,
       "step": 705
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.426791277258568e-05,
-      "loss": 4.8757,
+      "epoch": 0.79,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 4.3998,
       "step": 706
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.419003115264797e-05,
-      "loss": 4.8026,
+      "epoch": 0.79,
+      "learning_rate": 4.795918367346939e-05,
+      "loss": 4.4906,
       "step": 707
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.411214953271028e-05,
-      "loss": 4.8025,
+      "epoch": 0.79,
+      "learning_rate": 4.7704081632653066e-05,
+      "loss": 4.3952,
       "step": 708
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.403426791277259e-05,
-      "loss": 4.448,
+      "epoch": 0.79,
+      "learning_rate": 4.744897959183674e-05,
+      "loss": 4.2198,
       "step": 709
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.39563862928349e-05,
-      "loss": 4.4635,
+      "epoch": 0.8,
+      "learning_rate": 4.719387755102041e-05,
+      "loss": 4.2494,
       "step": 710
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.38785046728972e-05,
-      "loss": 4.6755,
+      "epoch": 0.8,
+      "learning_rate": 4.6938775510204086e-05,
+      "loss": 4.1282,
       "step": 711
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.38006230529595e-05,
-      "loss": 4.3179,
+      "epoch": 0.8,
+      "learning_rate": 4.6683673469387754e-05,
+      "loss": 4.1195,
       "step": 712
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.372274143302181e-05,
-      "loss": 4.334,
+      "epoch": 0.8,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 4.2118,
       "step": 713
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.364485981308412e-05,
-      "loss": 4.86,
+      "epoch": 0.8,
+      "learning_rate": 4.6173469387755105e-05,
+      "loss": 4.286,
       "step": 714
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.356697819314642e-05,
-      "loss": 4.3694,
+      "epoch": 0.8,
+      "learning_rate": 4.591836734693878e-05,
+      "loss": 4.351,
       "step": 715
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.348909657320873e-05,
-      "loss": 4.1123,
+      "epoch": 0.8,
+      "learning_rate": 4.566326530612245e-05,
+      "loss": 4.048,
       "step": 716
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.341121495327104e-05,
-      "loss": 4.319,
+      "epoch": 0.8,
+      "learning_rate": 4.5408163265306124e-05,
+      "loss": 4.4465,
       "step": 717
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.333333333333334e-05,
-      "loss": 4.5861,
+      "epoch": 0.8,
+      "learning_rate": 4.515306122448979e-05,
+      "loss": 4.1497,
       "step": 718
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.325545171339564e-05,
-      "loss": 4.3091,
+      "epoch": 0.81,
+      "learning_rate": 4.4897959183673474e-05,
+      "loss": 3.9604,
       "step": 719
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.317757009345795e-05,
-      "loss": 4.3475,
+      "epoch": 0.81,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 4.1792,
       "step": 720
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.309968847352026e-05,
-      "loss": 4.6787,
+      "epoch": 0.81,
+      "learning_rate": 4.438775510204082e-05,
+      "loss": 4.1827,
       "step": 721
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 8.302180685358257e-05,
-      "loss": 4.4693,
+      "epoch": 0.81,
+      "learning_rate": 4.4132653061224493e-05,
+      "loss": 4.0194,
       "step": 722
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.294392523364486e-05,
-      "loss": 4.3145,
+      "epoch": 0.81,
+      "learning_rate": 4.387755102040816e-05,
+      "loss": 4.2096,
       "step": 723
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.286604361370717e-05,
-      "loss": 4.8196,
+      "epoch": 0.81,
+      "learning_rate": 4.362244897959184e-05,
+      "loss": 4.2585,
       "step": 724
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.278816199376948e-05,
-      "loss": 4.4833,
+      "epoch": 0.81,
+      "learning_rate": 4.336734693877551e-05,
+      "loss": 4.077,
       "step": 725
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.271028037383179e-05,
-      "loss": 4.6014,
+      "epoch": 0.81,
+      "learning_rate": 4.311224489795919e-05,
+      "loss": 4.3033,
       "step": 726
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.263239875389408e-05,
-      "loss": 4.0677,
+      "epoch": 0.82,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 4.119,
       "step": 727
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.25545171339564e-05,
-      "loss": 4.6184,
+      "epoch": 0.82,
+      "learning_rate": 4.260204081632653e-05,
+      "loss": 4.1655,
       "step": 728
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.24766355140187e-05,
-      "loss": 4.336,
+      "epoch": 0.82,
+      "learning_rate": 4.234693877551021e-05,
+      "loss": 4.0441,
       "step": 729
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.2398753894081e-05,
-      "loss": 4.4639,
+      "epoch": 0.82,
+      "learning_rate": 4.209183673469388e-05,
+      "loss": 4.2595,
       "step": 730
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.232087227414331e-05,
-      "loss": 4.2889,
+      "epoch": 0.82,
+      "learning_rate": 4.183673469387756e-05,
+      "loss": 4.4831,
       "step": 731
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.22429906542056e-05,
-      "loss": 4.3153,
+      "epoch": 0.82,
+      "learning_rate": 4.1581632653061226e-05,
+      "loss": 4.3292,
       "step": 732
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.216510903426791e-05,
-      "loss": 5.0774,
+      "epoch": 0.82,
+      "learning_rate": 4.13265306122449e-05,
+      "loss": 4.0568,
       "step": 733
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.208722741433022e-05,
-      "loss": 4.2657,
+      "epoch": 0.82,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 4.2774,
       "step": 734
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.200934579439252e-05,
-      "loss": 4.5829,
+      "epoch": 0.82,
+      "learning_rate": 4.0816326530612245e-05,
+      "loss": 3.9564,
       "step": 735
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.193146417445483e-05,
-      "loss": 4.4103,
+      "epoch": 0.83,
+      "learning_rate": 4.056122448979592e-05,
+      "loss": 4.1067,
       "step": 736
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.185358255451713e-05,
-      "loss": 4.2791,
+      "epoch": 0.83,
+      "learning_rate": 4.0306122448979596e-05,
+      "loss": 4.2758,
       "step": 737
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.177570093457944e-05,
-      "loss": 4.2082,
+      "epoch": 0.83,
+      "learning_rate": 4.0051020408163264e-05,
+      "loss": 4.1247,
       "step": 738
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.169781931464174e-05,
-      "loss": 4.8782,
+      "epoch": 0.83,
+      "learning_rate": 3.979591836734694e-05,
+      "loss": 4.2443,
       "step": 739
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 8.161993769470405e-05,
-      "loss": 4.5045,
+      "epoch": 0.83,
+      "learning_rate": 3.9540816326530615e-05,
+      "loss": 4.2306,
       "step": 740
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.154205607476636e-05,
-      "loss": 4.8969,
+      "epoch": 0.83,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 4.4393,
       "step": 741
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.146417445482867e-05,
-      "loss": 4.2183,
+      "epoch": 0.83,
+      "learning_rate": 3.9030612244897965e-05,
+      "loss": 4.1833,
       "step": 742
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.138629283489096e-05,
-      "loss": 5.4842,
+      "epoch": 0.83,
+      "learning_rate": 3.8775510204081634e-05,
+      "loss": 3.9059,
       "step": 743
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.130841121495327e-05,
-      "loss": 4.5306,
+      "epoch": 0.83,
+      "learning_rate": 3.852040816326531e-05,
+      "loss": 4.3058,
       "step": 744
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.123052959501558e-05,
-      "loss": 5.2821,
+      "epoch": 0.84,
+      "learning_rate": 3.826530612244898e-05,
+      "loss": 4.2608,
       "step": 745
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.115264797507789e-05,
-      "loss": 4.4827,
+      "epoch": 0.84,
+      "learning_rate": 3.801020408163265e-05,
+      "loss": 4.4009,
       "step": 746
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.107476635514018e-05,
-      "loss": 4.5575,
+      "epoch": 0.84,
+      "learning_rate": 3.775510204081633e-05,
+      "loss": 4.2079,
       "step": 747
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.09968847352025e-05,
-      "loss": 4.7146,
+      "epoch": 0.84,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 4.3105,
       "step": 748
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.09190031152648e-05,
-      "loss": 4.305,
+      "epoch": 0.84,
+      "learning_rate": 3.724489795918368e-05,
+      "loss": 4.7069,
       "step": 749
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.084112149532711e-05,
-      "loss": 4.7422,
+      "epoch": 0.84,
+      "learning_rate": 3.698979591836735e-05,
+      "loss": 4.0475,
       "step": 750
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.076323987538941e-05,
-      "loss": 4.5634,
+      "epoch": 0.84,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 4.2485,
       "step": 751
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.068535825545172e-05,
-      "loss": 4.3077,
+      "epoch": 0.84,
+      "learning_rate": 3.64795918367347e-05,
+      "loss": 4.262,
       "step": 752
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.060747663551403e-05,
-      "loss": 4.5402,
+      "epoch": 0.84,
+      "learning_rate": 3.622448979591837e-05,
+      "loss": 4.2274,
       "step": 753
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.052959501557633e-05,
-      "loss": 4.4077,
+      "epoch": 0.85,
+      "learning_rate": 3.596938775510204e-05,
+      "loss": 4.1542,
       "step": 754
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.045171339563863e-05,
-      "loss": 4.322,
+      "epoch": 0.85,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 4.4089,
       "step": 755
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.037383177570094e-05,
-      "loss": 4.7583,
+      "epoch": 0.85,
+      "learning_rate": 3.5459183673469385e-05,
+      "loss": 4.2441,
       "step": 756
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.029595015576325e-05,
-      "loss": 4.3726,
+      "epoch": 0.85,
+      "learning_rate": 3.520408163265306e-05,
+      "loss": 4.198,
       "step": 757
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 8.021806853582556e-05,
-      "loss": 4.5185,
+      "epoch": 0.85,
+      "learning_rate": 3.4948979591836736e-05,
+      "loss": 4.2563,
       "step": 758
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 8.014018691588785e-05,
-      "loss": 4.3511,
+      "epoch": 0.85,
+      "learning_rate": 3.469387755102041e-05,
+      "loss": 4.1014,
       "step": 759
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 8.006230529595016e-05,
-      "loss": 4.3407,
+      "epoch": 0.85,
+      "learning_rate": 3.443877551020409e-05,
+      "loss": 4.2019,
       "step": 760
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.998442367601247e-05,
-      "loss": 4.59,
+      "epoch": 0.85,
+      "learning_rate": 3.4183673469387755e-05,
+      "loss": 4.1753,
       "step": 761
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.990654205607478e-05,
-      "loss": 4.1559,
+      "epoch": 0.85,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 4.2436,
       "step": 762
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.982866043613708e-05,
-      "loss": 4.3162,
+      "epoch": 0.86,
+      "learning_rate": 3.36734693877551e-05,
+      "loss": 4.107,
       "step": 763
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.975077881619938e-05,
-      "loss": 4.1151,
+      "epoch": 0.86,
+      "learning_rate": 3.341836734693878e-05,
+      "loss": 4.0647,
       "step": 764
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.967289719626168e-05,
-      "loss": 4.1855,
+      "epoch": 0.86,
+      "learning_rate": 3.316326530612245e-05,
+      "loss": 4.3057,
       "step": 765
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.959501557632399e-05,
-      "loss": 4.6788,
+      "epoch": 0.86,
+      "learning_rate": 3.2908163265306125e-05,
+      "loss": 4.109,
       "step": 766
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.95171339563863e-05,
-      "loss": 4.6394,
+      "epoch": 0.86,
+      "learning_rate": 3.265306122448979e-05,
+      "loss": 4.0356,
       "step": 767
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.94392523364486e-05,
-      "loss": 4.2721,
+      "epoch": 0.86,
+      "learning_rate": 3.239795918367347e-05,
+      "loss": 4.2105,
       "step": 768
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.93613707165109e-05,
-      "loss": 4.7748,
+      "epoch": 0.86,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 4.2356,
       "step": 769
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.928348909657321e-05,
-      "loss": 4.3394,
+      "epoch": 0.86,
+      "learning_rate": 3.188775510204082e-05,
+      "loss": 4.1832,
       "step": 770
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.920560747663551e-05,
-      "loss": 4.5556,
+      "epoch": 0.86,
+      "learning_rate": 3.1632653061224494e-05,
+      "loss": 4.0565,
       "step": 771
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.912772585669782e-05,
-      "loss": 4.2633,
+      "epoch": 0.87,
+      "learning_rate": 3.137755102040816e-05,
+      "loss": 4.2768,
       "step": 772
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.904984423676013e-05,
-      "loss": 4.1947,
+      "epoch": 0.87,
+      "learning_rate": 3.112244897959184e-05,
+      "loss": 4.369,
       "step": 773
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.897196261682243e-05,
-      "loss": 4.6906,
+      "epoch": 0.87,
+      "learning_rate": 3.086734693877551e-05,
+      "loss": 3.935,
       "step": 774
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.889408099688473e-05,
-      "loss": 4.2007,
+      "epoch": 0.87,
+      "learning_rate": 3.061224489795919e-05,
+      "loss": 3.9795,
       "step": 775
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.881619937694704e-05,
-      "loss": 4.4047,
+      "epoch": 0.87,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 4.3018,
       "step": 776
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.873831775700935e-05,
-      "loss": 4.0871,
+      "epoch": 0.87,
+      "learning_rate": 3.0102040816326533e-05,
+      "loss": 3.9756,
       "step": 777
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.866043613707166e-05,
-      "loss": 4.5159,
+      "epoch": 0.87,
+      "learning_rate": 2.9846938775510208e-05,
+      "loss": 4.3389,
       "step": 778
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.858255451713395e-05,
-      "loss": 4.5724,
+      "epoch": 0.87,
+      "learning_rate": 2.959183673469388e-05,
+      "loss": 4.0707,
       "step": 779
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.850467289719626e-05,
-      "loss": 4.4574,
+      "epoch": 0.87,
+      "learning_rate": 2.9336734693877555e-05,
+      "loss": 4.0017,
       "step": 780
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.842679127725857e-05,
-      "loss": 4.4397,
+      "epoch": 0.88,
+      "learning_rate": 2.9081632653061224e-05,
+      "loss": 3.9922,
       "step": 781
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.834890965732088e-05,
-      "loss": 4.556,
+      "epoch": 0.88,
+      "learning_rate": 2.8826530612244902e-05,
+      "loss": 4.1209,
       "step": 782
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.827102803738318e-05,
-      "loss": 4.2939,
+      "epoch": 0.88,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 4.1468,
       "step": 783
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.819314641744548e-05,
-      "loss": 4.3301,
+      "epoch": 0.88,
+      "learning_rate": 2.8316326530612246e-05,
+      "loss": 3.9952,
       "step": 784
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.81152647975078e-05,
-      "loss": 4.0398,
+      "epoch": 0.88,
+      "learning_rate": 2.8061224489795918e-05,
+      "loss": 4.1739,
       "step": 785
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.80373831775701e-05,
-      "loss": 4.7709,
+      "epoch": 0.88,
+      "learning_rate": 2.7806122448979593e-05,
+      "loss": 3.9362,
       "step": 786
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.79595015576324e-05,
-      "loss": 4.076,
+      "epoch": 0.88,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 4.2515,
       "step": 787
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.788161993769471e-05,
-      "loss": 4.7459,
+      "epoch": 0.88,
+      "learning_rate": 2.729591836734694e-05,
+      "loss": 4.1879,
       "step": 788
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.780373831775702e-05,
-      "loss": 4.5851,
+      "epoch": 0.88,
+      "learning_rate": 2.7040816326530616e-05,
+      "loss": 4.0157,
       "step": 789
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.772585669781933e-05,
-      "loss": 4.7151,
+      "epoch": 0.89,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 4.1654,
       "step": 790
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.764797507788162e-05,
-      "loss": 4.7314,
+      "epoch": 0.89,
+      "learning_rate": 2.6530612244897963e-05,
+      "loss": 4.1267,
       "step": 791
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.757009345794393e-05,
-      "loss": 4.3707,
+      "epoch": 0.89,
+      "learning_rate": 2.627551020408163e-05,
+      "loss": 4.1517,
       "step": 792
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.749221183800624e-05,
-      "loss": 4.6428,
+      "epoch": 0.89,
+      "learning_rate": 2.6020408163265307e-05,
+      "loss": 4.0912,
       "step": 793
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.741433021806855e-05,
-      "loss": 4.2851,
+      "epoch": 0.89,
+      "learning_rate": 2.576530612244898e-05,
+      "loss": 4.3466,
       "step": 794
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.733644859813084e-05,
-      "loss": 4.7223,
+      "epoch": 0.89,
+      "learning_rate": 2.5510204081632654e-05,
+      "loss": 4.2573,
       "step": 795
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.725856697819315e-05,
-      "loss": 5.4373,
+      "epoch": 0.89,
+      "learning_rate": 2.5255102040816326e-05,
+      "loss": 4.0715,
       "step": 796
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.718068535825546e-05,
-      "loss": 5.3637,
+      "epoch": 0.89,
+      "learning_rate": 2.5e-05,
+      "loss": 4.2313,
       "step": 797
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.710280373831776e-05,
-      "loss": 4.6564,
+      "epoch": 0.89,
+      "learning_rate": 2.4744897959183673e-05,
+      "loss": 4.2574,
       "step": 798
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.702492211838007e-05,
-      "loss": 4.7565,
+      "epoch": 0.9,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 4.8761,
       "step": 799
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.694704049844238e-05,
-      "loss": 4.5411,
+      "epoch": 0.9,
+      "learning_rate": 2.423469387755102e-05,
+      "loss": 4.3985,
       "step": 800
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.686915887850467e-05,
-      "loss": 4.7676,
+      "epoch": 0.9,
+      "learning_rate": 2.3979591836734696e-05,
+      "loss": 4.2022,
       "step": 801
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.679127725856698e-05,
-      "loss": 4.54,
+      "epoch": 0.9,
+      "learning_rate": 2.372448979591837e-05,
+      "loss": 4.0457,
       "step": 802
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.671339563862928e-05,
-      "loss": 4.809,
+      "epoch": 0.9,
+      "learning_rate": 2.3469387755102043e-05,
+      "loss": 4.1468,
       "step": 803
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.663551401869158e-05,
-      "loss": 4.4635,
+      "epoch": 0.9,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 4.1144,
       "step": 804
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.65576323987539e-05,
-      "loss": 4.4969,
+      "epoch": 0.9,
+      "learning_rate": 2.295918367346939e-05,
+      "loss": 3.9681,
       "step": 805
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.64797507788162e-05,
-      "loss": 4.4439,
+      "epoch": 0.9,
+      "learning_rate": 2.2704081632653062e-05,
+      "loss": 4.1506,
       "step": 806
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.64018691588785e-05,
-      "loss": 4.326,
+      "epoch": 0.9,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 4.1706,
       "step": 807
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.632398753894081e-05,
-      "loss": 4.0343,
+      "epoch": 0.91,
+      "learning_rate": 2.219387755102041e-05,
+      "loss": 4.0046,
       "step": 808
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.624610591900312e-05,
-      "loss": 4.4667,
+      "epoch": 0.91,
+      "learning_rate": 2.193877551020408e-05,
+      "loss": 4.2563,
       "step": 809
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.616822429906543e-05,
-      "loss": 4.6725,
+      "epoch": 0.91,
+      "learning_rate": 2.1683673469387756e-05,
+      "loss": 4.0046,
       "step": 810
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.609034267912772e-05,
-      "loss": 4.4931,
+      "epoch": 0.91,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 4.3614,
       "step": 811
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.601246105919003e-05,
-      "loss": 4.615,
+      "epoch": 0.91,
+      "learning_rate": 2.1173469387755103e-05,
+      "loss": 4.1648,
       "step": 812
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.593457943925234e-05,
-      "loss": 4.5447,
+      "epoch": 0.91,
+      "learning_rate": 2.091836734693878e-05,
+      "loss": 3.9555,
       "step": 813
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.585669781931465e-05,
-      "loss": 4.1368,
+      "epoch": 0.91,
+      "learning_rate": 2.066326530612245e-05,
+      "loss": 4.0926,
       "step": 814
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.577881619937694e-05,
-      "loss": 4.3032,
+      "epoch": 0.91,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 4.0064,
       "step": 815
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.570093457943925e-05,
-      "loss": 4.3253,
+      "epoch": 0.91,
+      "learning_rate": 2.0153061224489798e-05,
+      "loss": 4.0343,
       "step": 816
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.562305295950156e-05,
-      "loss": 4.5842,
+      "epoch": 0.92,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 4.1501,
       "step": 817
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.554517133956387e-05,
-      "loss": 4.4136,
+      "epoch": 0.92,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 4.1466,
       "step": 818
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.546728971962617e-05,
-      "loss": 4.3353,
+      "epoch": 0.92,
+      "learning_rate": 1.9387755102040817e-05,
+      "loss": 4.0237,
       "step": 819
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.538940809968848e-05,
-      "loss": 4.2845,
+      "epoch": 0.92,
+      "learning_rate": 1.913265306122449e-05,
+      "loss": 4.127,
       "step": 820
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.531152647975078e-05,
-      "loss": 4.4821,
+      "epoch": 0.92,
+      "learning_rate": 1.8877551020408164e-05,
+      "loss": 4.0763,
       "step": 821
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.52336448598131e-05,
-      "loss": 4.5264,
+      "epoch": 0.92,
+      "learning_rate": 1.862244897959184e-05,
+      "loss": 4.0363,
       "step": 822
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 7.515576323987539e-05,
-      "loss": 4.9346,
-      "step": 823
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.50778816199377e-05,
-      "loss": 4.455,
-      "step": 824
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.3142,
-      "step": 825
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.492211838006232e-05,
-      "loss": 4.3529,
-      "step": 826
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.484423676012463e-05,
-      "loss": 4.7274,
-      "step": 827
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.476635514018692e-05,
-      "loss": 4.3492,
-      "step": 828
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 7.468847352024923e-05,
-      "loss": 4.8814,
-      "step": 829
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.461059190031154e-05,
-      "loss": 4.3758,
-      "step": 830
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.453271028037383e-05,
-      "loss": 4.7222,
-      "step": 831
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.445482866043614e-05,
-      "loss": 4.9621,
-      "step": 832
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.437694704049845e-05,
-      "loss": 4.7864,
-      "step": 833
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.429906542056075e-05,
-      "loss": 4.0429,
-      "step": 834
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.422118380062306e-05,
-      "loss": 4.343,
-      "step": 835
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.414330218068535e-05,
-      "loss": 4.3791,
-      "step": 836
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.406542056074766e-05,
-      "loss": 4.537,
-      "step": 837
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.398753894080997e-05,
-      "loss": 4.429,
-      "step": 838
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.390965732087228e-05,
-      "loss": 4.1966,
-      "step": 839
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.383177570093458e-05,
-      "loss": 4.5571,
-      "step": 840
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.375389408099688e-05,
-      "loss": 5.0421,
-      "step": 841
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.36760124610592e-05,
-      "loss": 4.3453,
-      "step": 842
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.35981308411215e-05,
-      "loss": 4.3098,
-      "step": 843
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.35202492211838e-05,
-      "loss": 4.5596,
-      "step": 844
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.344236760124611e-05,
-      "loss": 4.0437,
-      "step": 845
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.336448598130842e-05,
-      "loss": 4.2405,
-      "step": 846
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.328660436137072e-05,
-      "loss": 4.149,
-      "step": 847
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.320872274143302e-05,
-      "loss": 4.545,
-      "step": 848
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.313084112149533e-05,
-      "loss": 4.6842,
-      "step": 849
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.305295950155764e-05,
-      "loss": 6.6157,
-      "step": 850
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.297507788161995e-05,
-      "loss": 4.4735,
-      "step": 851
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.289719626168224e-05,
-      "loss": 4.5122,
-      "step": 852
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.281931464174455e-05,
-      "loss": 4.7557,
-      "step": 853
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.274143302180686e-05,
-      "loss": 4.4691,
-      "step": 854
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.266355140186917e-05,
-      "loss": 4.3686,
-      "step": 855
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.258566978193147e-05,
-      "loss": 4.5868,
-      "step": 856
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.250778816199377e-05,
-      "loss": 4.3214,
-      "step": 857
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.242990654205608e-05,
-      "loss": 4.654,
-      "step": 858
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.235202492211839e-05,
-      "loss": 4.4606,
-      "step": 859
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.227414330218069e-05,
-      "loss": 4.8,
-      "step": 860
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.2196261682243e-05,
-      "loss": 4.2295,
-      "step": 861
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.21183800623053e-05,
-      "loss": 4.372,
-      "step": 862
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.204049844236762e-05,
-      "loss": 4.657,
-      "step": 863
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.196261682242991e-05,
-      "loss": 4.3324,
-      "step": 864
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.188473520249222e-05,
-      "loss": 4.7268,
-      "step": 865
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.180685358255453e-05,
-      "loss": 4.417,
-      "step": 866
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.172897196261682e-05,
-      "loss": 4.4625,
-      "step": 867
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.165109034267913e-05,
-      "loss": 4.5482,
-      "step": 868
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.157320872274143e-05,
-      "loss": 4.3628,
-      "step": 869
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.149532710280374e-05,
-      "loss": 4.2761,
-      "step": 870
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.141744548286605e-05,
-      "loss": 4.3612,
-      "step": 871
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.133956386292834e-05,
-      "loss": 4.5047,
-      "step": 872
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.126168224299065e-05,
-      "loss": 4.8103,
-      "step": 873
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.118380062305296e-05,
-      "loss": 4.3877,
-      "step": 874
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.110591900311527e-05,
-      "loss": 4.5575,
-      "step": 875
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.102803738317757e-05,
-      "loss": 4.6425,
-      "step": 876
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.095015576323987e-05,
-      "loss": 4.6706,
-      "step": 877
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.087227414330218e-05,
-      "loss": 4.5234,
-      "step": 878
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.079439252336449e-05,
-      "loss": 4.3465,
-      "step": 879
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.071651090342679e-05,
-      "loss": 4.4413,
-      "step": 880
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.06386292834891e-05,
-      "loss": 4.7193,
-      "step": 881
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.05607476635514e-05,
-      "loss": 4.3736,
-      "step": 882
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.048286604361372e-05,
-      "loss": 4.4847,
-      "step": 883
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.040498442367601e-05,
-      "loss": 4.8445,
-      "step": 884
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.032710280373832e-05,
-      "loss": 4.8584,
-      "step": 885
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.024922118380063e-05,
-      "loss": 4.5475,
-      "step": 886
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.017133956386294e-05,
-      "loss": 5.16,
-      "step": 887
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.009345794392523e-05,
-      "loss": 4.1336,
-      "step": 888
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.001557632398754e-05,
-      "loss": 4.648,
-      "step": 889
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.993769470404985e-05,
-      "loss": 4.746,
-      "step": 890
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.985981308411216e-05,
-      "loss": 4.5639,
-      "step": 891
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.978193146417446e-05,
-      "loss": 4.578,
-      "step": 892
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.970404984423677e-05,
-      "loss": 4.9982,
-      "step": 893
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.962616822429907e-05,
-      "loss": 4.4636,
-      "step": 894
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.954828660436138e-05,
-      "loss": 4.3529,
-      "step": 895
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.947040498442368e-05,
-      "loss": 4.3084,
-      "step": 896
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.939252336448599e-05,
-      "loss": 4.3474,
-      "step": 897
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.93146417445483e-05,
-      "loss": 5.3993,
-      "step": 898
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.92367601246106e-05,
-      "loss": 5.2447,
-      "step": 899
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 6.91588785046729e-05,
-      "loss": 5.7452,
-      "step": 900
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.908099688473521e-05,
-      "loss": 4.4296,
-      "step": 901
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.90031152647975e-05,
-      "loss": 4.3585,
-      "step": 902
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.892523364485982e-05,
-      "loss": 4.8067,
-      "step": 903
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.884735202492212e-05,
-      "loss": 4.4171,
-      "step": 904
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.876947040498442e-05,
-      "loss": 4.6432,
-      "step": 905
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.869158878504673e-05,
-      "loss": 4.7149,
-      "step": 906
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.861370716510904e-05,
-      "loss": 4.5827,
-      "step": 907
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.853582554517133e-05,
-      "loss": 4.2471,
-      "step": 908
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.845794392523364e-05,
-      "loss": 4.5014,
-      "step": 909
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.838006230529595e-05,
-      "loss": 4.2851,
-      "step": 910
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.830218068535826e-05,
-      "loss": 4.5608,
-      "step": 911
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.822429906542056e-05,
-      "loss": 4.267,
-      "step": 912
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.814641744548287e-05,
-      "loss": 4.554,
-      "step": 913
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.806853582554517e-05,
-      "loss": 4.3867,
-      "step": 914
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.799065420560748e-05,
-      "loss": 4.5623,
-      "step": 915
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.791277258566978e-05,
-      "loss": 4.5693,
-      "step": 916
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.783489096573209e-05,
-      "loss": 4.212,
-      "step": 917
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 6.77570093457944e-05,
-      "loss": 4.6835,
-      "step": 918
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.76791277258567e-05,
-      "loss": 4.3118,
-      "step": 919
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.7601246105919e-05,
-      "loss": 4.1267,
-      "step": 920
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.752336448598131e-05,
-      "loss": 4.2605,
-      "step": 921
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.744548286604362e-05,
-      "loss": 4.2969,
-      "step": 922
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.736760124610593e-05,
-      "loss": 4.2737,
-      "step": 923
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.728971962616822e-05,
-      "loss": 4.3347,
-      "step": 924
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.721183800623053e-05,
-      "loss": 4.5327,
-      "step": 925
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.713395638629284e-05,
-      "loss": 4.6206,
-      "step": 926
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.705607476635515e-05,
-      "loss": 4.271,
-      "step": 927
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.697819314641745e-05,
-      "loss": 4.2502,
-      "step": 928
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.690031152647976e-05,
-      "loss": 4.2066,
-      "step": 929
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.682242990654207e-05,
-      "loss": 4.5456,
-      "step": 930
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.674454828660437e-05,
-      "loss": 4.7881,
-      "step": 931
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.666666666666667e-05,
-      "loss": 4.5204,
-      "step": 932
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.658878504672898e-05,
-      "loss": 4.3622,
-      "step": 933
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.651090342679129e-05,
-      "loss": 4.4578,
-      "step": 934
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.64330218068536e-05,
-      "loss": 4.7832,
-      "step": 935
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 6.635514018691589e-05,
-      "loss": 4.4635,
-      "step": 936
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.62772585669782e-05,
-      "loss": 4.3835,
-      "step": 937
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.61993769470405e-05,
-      "loss": 4.0697,
-      "step": 938
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.61214953271028e-05,
-      "loss": 4.1073,
-      "step": 939
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.60436137071651e-05,
-      "loss": 4.6159,
-      "step": 940
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.596573208722741e-05,
-      "loss": 4.5158,
-      "step": 941
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.588785046728972e-05,
-      "loss": 4.3619,
-      "step": 942
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.580996884735203e-05,
-      "loss": 4.7987,
-      "step": 943
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.573208722741432e-05,
-      "loss": 4.4541,
-      "step": 944
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.565420560747663e-05,
-      "loss": 4.4287,
-      "step": 945
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.557632398753894e-05,
-      "loss": 4.5014,
-      "step": 946
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.549844236760125e-05,
-      "loss": 4.0671,
-      "step": 947
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.542056074766355e-05,
-      "loss": 4.4916,
-      "step": 948
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.534267912772586e-05,
-      "loss": 5.0872,
-      "step": 949
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.526479750778817e-05,
-      "loss": 3.9575,
-      "step": 950
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.518691588785047e-05,
-      "loss": 4.2716,
-      "step": 951
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.510903426791277e-05,
-      "loss": 4.5053,
-      "step": 952
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.503115264797508e-05,
-      "loss": 4.4414,
-      "step": 953
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 6.495327102803739e-05,
-      "loss": 4.0541,
-      "step": 954
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.48753894080997e-05,
-      "loss": 4.5223,
-      "step": 955
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.479750778816199e-05,
-      "loss": 4.3215,
-      "step": 956
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.47196261682243e-05,
-      "loss": 4.5157,
-      "step": 957
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.464174454828661e-05,
-      "loss": 4.2312,
-      "step": 958
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.456386292834892e-05,
-      "loss": 4.676,
-      "step": 959
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.448598130841122e-05,
-      "loss": 4.2813,
-      "step": 960
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.440809968847352e-05,
-      "loss": 4.2288,
-      "step": 961
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.433021806853583e-05,
-      "loss": 4.3602,
-      "step": 962
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.425233644859814e-05,
-      "loss": 4.6851,
-      "step": 963
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.417445482866044e-05,
-      "loss": 4.4708,
-      "step": 964
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.409657320872275e-05,
-      "loss": 4.2399,
-      "step": 965
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.401869158878506e-05,
-      "loss": 4.1976,
-      "step": 966
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.394080996884737e-05,
-      "loss": 4.6044,
-      "step": 967
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.386292834890966e-05,
-      "loss": 4.2295,
-      "step": 968
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.378504672897197e-05,
-      "loss": 4.4765,
-      "step": 969
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.370716510903428e-05,
-      "loss": 4.1369,
-      "step": 970
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.362928348909657e-05,
-      "loss": 4.189,
-      "step": 971
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 6.355140186915888e-05,
-      "loss": 4.2791,
-      "step": 972
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.347352024922119e-05,
-      "loss": 4.3525,
-      "step": 973
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.339563862928349e-05,
-      "loss": 5.2386,
-      "step": 974
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.33177570093458e-05,
-      "loss": 4.3434,
-      "step": 975
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.323987538940809e-05,
-      "loss": 4.4491,
-      "step": 976
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.31619937694704e-05,
-      "loss": 4.6706,
-      "step": 977
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.308411214953271e-05,
-      "loss": 4.3355,
-      "step": 978
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.300623052959502e-05,
-      "loss": 4.4488,
-      "step": 979
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.292834890965732e-05,
-      "loss": 4.2925,
-      "step": 980
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.285046728971962e-05,
-      "loss": 4.0622,
-      "step": 981
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.277258566978193e-05,
-      "loss": 4.013,
-      "step": 982
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.269470404984424e-05,
-      "loss": 4.5865,
-      "step": 983
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.261682242990654e-05,
-      "loss": 4.5794,
-      "step": 984
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.253894080996885e-05,
-      "loss": 4.14,
-      "step": 985
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.246105919003116e-05,
-      "loss": 4.3451,
-      "step": 986
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.238317757009346e-05,
-      "loss": 4.5025,
-      "step": 987
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.230529595015576e-05,
-      "loss": 4.3271,
-      "step": 988
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.222741433021807e-05,
-      "loss": 4.4626,
-      "step": 989
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 6.214953271028038e-05,
-      "loss": 4.1273,
-      "step": 990
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.207165109034269e-05,
-      "loss": 4.6504,
-      "step": 991
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.199376947040498e-05,
-      "loss": 4.6726,
-      "step": 992
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.191588785046729e-05,
-      "loss": 4.6183,
-      "step": 993
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.18380062305296e-05,
-      "loss": 4.4209,
-      "step": 994
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.176012461059191e-05,
-      "loss": 4.2627,
-      "step": 995
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.16822429906542e-05,
-      "loss": 3.9652,
-      "step": 996
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.160436137071651e-05,
-      "loss": 4.4874,
-      "step": 997
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.152647975077882e-05,
-      "loss": 5.2848,
-      "step": 998
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.144859813084113e-05,
-      "loss": 4.7767,
-      "step": 999
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.137071651090343e-05,
-      "loss": 4.3772,
-      "step": 1000
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.529038429260254,
-      "eval_runtime": 914.6987,
-      "eval_samples_per_second": 2.888,
-      "eval_steps_per_second": 0.362,
-      "eval_wer": 1.945854819516065,
-      "step": 1000
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.129283489096574e-05,
-      "loss": 4.6015,
-      "step": 1001
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.121495327102805e-05,
-      "loss": 4.2687,
-      "step": 1002
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.113707165109036e-05,
-      "loss": 4.5858,
-      "step": 1003
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.105919003115265e-05,
-      "loss": 4.5042,
-      "step": 1004
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.0981308411214953e-05,
-      "loss": 4.3579,
-      "step": 1005
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.090342679127726e-05,
-      "loss": 4.319,
-      "step": 1006
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 6.082554517133957e-05,
-      "loss": 4.4899,
-      "step": 1007
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.074766355140187e-05,
-      "loss": 4.5053,
-      "step": 1008
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.0669781931464176e-05,
-      "loss": 4.5659,
-      "step": 1009
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.0591900311526485e-05,
-      "loss": 4.1679,
-      "step": 1010
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.0514018691588794e-05,
-      "loss": 4.4757,
-      "step": 1011
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.043613707165109e-05,
-      "loss": 4.3362,
-      "step": 1012
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.03582554517134e-05,
-      "loss": 4.5333,
-      "step": 1013
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.028037383177571e-05,
-      "loss": 4.4481,
-      "step": 1014
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.020249221183801e-05,
-      "loss": 5.0417,
-      "step": 1015
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.012461059190031e-05,
-      "loss": 4.2914,
-      "step": 1016
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 6.0046728971962615e-05,
-      "loss": 4.3677,
-      "step": 1017
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.9968847352024924e-05,
-      "loss": 4.3617,
-      "step": 1018
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.989096573208723e-05,
-      "loss": 4.7534,
-      "step": 1019
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.981308411214953e-05,
-      "loss": 4.662,
-      "step": 1020
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.973520249221184e-05,
-      "loss": 4.5215,
-      "step": 1021
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.9657320872274146e-05,
-      "loss": 4.523,
-      "step": 1022
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.9579439252336456e-05,
-      "loss": 4.4842,
-      "step": 1023
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.950155763239875e-05,
-      "loss": 4.3534,
-      "step": 1024
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 5.942367601246106e-05,
-      "loss": 4.4577,
-      "step": 1025
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.934579439252337e-05,
-      "loss": 4.2901,
-      "step": 1026
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.926791277258568e-05,
-      "loss": 4.414,
-      "step": 1027
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.9190031152647974e-05,
-      "loss": 4.4763,
-      "step": 1028
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.911214953271028e-05,
-      "loss": 4.308,
-      "step": 1029
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.903426791277259e-05,
-      "loss": 4.4972,
-      "step": 1030
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.89563862928349e-05,
-      "loss": 4.6251,
-      "step": 1031
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.8878504672897196e-05,
-      "loss": 4.3092,
-      "step": 1032
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.8800623052959505e-05,
-      "loss": 4.6766,
-      "step": 1033
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.872274143302181e-05,
-      "loss": 4.123,
-      "step": 1034
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.864485981308412e-05,
-      "loss": 4.4027,
-      "step": 1035
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.856697819314641e-05,
-      "loss": 4.3853,
-      "step": 1036
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.848909657320872e-05,
-      "loss": 4.7567,
-      "step": 1037
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.841121495327103e-05,
-      "loss": 4.4436,
-      "step": 1038
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.833333333333334e-05,
-      "loss": 4.3448,
-      "step": 1039
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.8255451713395635e-05,
-      "loss": 4.9213,
-      "step": 1040
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.8177570093457944e-05,
-      "loss": 4.6438,
-      "step": 1041
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.809968847352025e-05,
-      "loss": 4.4548,
-      "step": 1042
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 5.802180685358256e-05,
-      "loss": 4.774,
-      "step": 1043
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.794392523364486e-05,
-      "loss": 4.6717,
-      "step": 1044
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.786604361370717e-05,
-      "loss": 4.8856,
-      "step": 1045
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7788161993769476e-05,
-      "loss": 4.5642,
-      "step": 1046
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7710280373831785e-05,
-      "loss": 4.3194,
-      "step": 1047
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.763239875389408e-05,
-      "loss": 4.2188,
-      "step": 1048
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.755451713395639e-05,
-      "loss": 4.4623,
-      "step": 1049
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.74766355140187e-05,
-      "loss": 5.315,
-      "step": 1050
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7398753894081e-05,
-      "loss": 4.9169,
-      "step": 1051
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.73208722741433e-05,
-      "loss": 4.2001,
-      "step": 1052
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7242990654205605e-05,
-      "loss": 4.3098,
-      "step": 1053
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7165109034267914e-05,
-      "loss": 4.116,
-      "step": 1054
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.7087227414330223e-05,
-      "loss": 4.3818,
-      "step": 1055
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.700934579439252e-05,
-      "loss": 4.2545,
-      "step": 1056
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.693146417445483e-05,
-      "loss": 4.6167,
-      "step": 1057
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.685358255451714e-05,
-      "loss": 4.2949,
-      "step": 1058
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.6775700934579446e-05,
-      "loss": 4.8033,
-      "step": 1059
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.669781931464174e-05,
-      "loss": 4.1379,
-      "step": 1060
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 5.661993769470405e-05,
-      "loss": 4.3968,
-      "step": 1061
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.654205607476636e-05,
-      "loss": 4.3155,
-      "step": 1062
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.646417445482867e-05,
-      "loss": 4.7487,
-      "step": 1063
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.6386292834890964e-05,
-      "loss": 3.9359,
-      "step": 1064
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.6308411214953273e-05,
-      "loss": 4.4449,
-      "step": 1065
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.623052959501558e-05,
-      "loss": 4.1155,
-      "step": 1066
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.6152647975077885e-05,
-      "loss": 4.3763,
-      "step": 1067
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.607476635514019e-05,
-      "loss": 4.2415,
-      "step": 1068
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.5996884735202496e-05,
-      "loss": 4.5161,
-      "step": 1069
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.59190031152648e-05,
-      "loss": 4.0975,
-      "step": 1070
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.584112149532711e-05,
-      "loss": 4.2463,
-      "step": 1071
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.57632398753894e-05,
-      "loss": 3.992,
-      "step": 1072
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.568535825545171e-05,
-      "loss": 4.096,
-      "step": 1073
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.560747663551402e-05,
-      "loss": 4.6875,
-      "step": 1074
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.552959501557633e-05,
-      "loss": 4.5092,
-      "step": 1075
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.5451713395638626e-05,
-      "loss": 4.5192,
-      "step": 1076
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.5373831775700935e-05,
-      "loss": 4.2677,
-      "step": 1077
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.5295950155763244e-05,
-      "loss": 4.3105,
-      "step": 1078
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 5.521806853582555e-05,
-      "loss": 4.2753,
-      "step": 1079
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.514018691588785e-05,
-      "loss": 4.5332,
-      "step": 1080
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.506230529595016e-05,
-      "loss": 4.9779,
-      "step": 1081
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4984423676012466e-05,
-      "loss": 4.4892,
-      "step": 1082
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4906542056074776e-05,
-      "loss": 4.021,
-      "step": 1083
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.482866043613707e-05,
-      "loss": 4.3186,
-      "step": 1084
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.475077881619938e-05,
-      "loss": 4.3034,
-      "step": 1085
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.467289719626168e-05,
-      "loss": 4.4385,
-      "step": 1086
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.459501557632399e-05,
-      "loss": 4.4888,
-      "step": 1087
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4517133956386294e-05,
-      "loss": 4.3995,
-      "step": 1088
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4439252336448596e-05,
-      "loss": 4.6545,
-      "step": 1089
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4361370716510905e-05,
-      "loss": 4.0838,
-      "step": 1090
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.4283489096573214e-05,
-      "loss": 4.2485,
-      "step": 1091
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.420560747663551e-05,
-      "loss": 4.4484,
-      "step": 1092
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.412772585669782e-05,
-      "loss": 4.4574,
-      "step": 1093
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.404984423676013e-05,
-      "loss": 4.6459,
-      "step": 1094
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.397196261682244e-05,
-      "loss": 4.4064,
-      "step": 1095
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.389408099688473e-05,
-      "loss": 4.68,
-      "step": 1096
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 5.381619937694704e-05,
-      "loss": 4.7088,
-      "step": 1097
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.373831775700935e-05,
-      "loss": 5.191,
-      "step": 1098
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.366043613707166e-05,
-      "loss": 4.4289,
-      "step": 1099
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.3582554517133955e-05,
-      "loss": 4.7703,
-      "step": 1100
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.3504672897196264e-05,
-      "loss": 4.1973,
-      "step": 1101
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.342679127725857e-05,
-      "loss": 4.4327,
-      "step": 1102
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.3348909657320875e-05,
-      "loss": 4.3634,
-      "step": 1103
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.327102803738318e-05,
-      "loss": 4.0646,
-      "step": 1104
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.319314641744548e-05,
-      "loss": 4.1736,
-      "step": 1105
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.311526479750779e-05,
-      "loss": 4.1289,
-      "step": 1106
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.30373831775701e-05,
-      "loss": 4.1247,
-      "step": 1107
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.2959501557632394e-05,
-      "loss": 4.0423,
-      "step": 1108
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.28816199376947e-05,
-      "loss": 4.4604,
-      "step": 1109
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.280373831775701e-05,
-      "loss": 4.3894,
-      "step": 1110
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.272585669781932e-05,
-      "loss": 4.1065,
-      "step": 1111
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.2647975077881616e-05,
-      "loss": 4.4884,
-      "step": 1112
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.2570093457943925e-05,
-      "loss": 4.0924,
-      "step": 1113
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.2492211838006234e-05,
-      "loss": 4.4073,
-      "step": 1114
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 5.2414330218068543e-05,
-      "loss": 4.6868,
-      "step": 1115
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.233644859813084e-05,
-      "loss": 4.3109,
-      "step": 1116
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.225856697819315e-05,
-      "loss": 4.33,
-      "step": 1117
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.218068535825546e-05,
-      "loss": 4.2482,
-      "step": 1118
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.210280373831776e-05,
-      "loss": 4.2026,
-      "step": 1119
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.202492211838006e-05,
-      "loss": 4.1258,
-      "step": 1120
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.194704049844237e-05,
-      "loss": 4.6109,
-      "step": 1121
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.186915887850467e-05,
-      "loss": 4.3238,
-      "step": 1122
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.179127725856698e-05,
-      "loss": 4.1622,
-      "step": 1123
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.171339563862928e-05,
-      "loss": 4.0847,
-      "step": 1124
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.163551401869159e-05,
-      "loss": 4.668,
-      "step": 1125
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.1557632398753896e-05,
-      "loss": 4.2466,
-      "step": 1126
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.1479750778816205e-05,
-      "loss": 4.1221,
-      "step": 1127
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.14018691588785e-05,
-      "loss": 4.8746,
-      "step": 1128
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.132398753894081e-05,
-      "loss": 3.9023,
-      "step": 1129
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.124610591900312e-05,
-      "loss": 4.7075,
-      "step": 1130
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.116822429906543e-05,
-      "loss": 4.4182,
-      "step": 1131
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 5.109034267912772e-05,
-      "loss": 4.2424,
-      "step": 1132
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.101246105919003e-05,
-      "loss": 4.348,
-      "step": 1133
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.093457943925234e-05,
-      "loss": 4.2908,
-      "step": 1134
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.085669781931465e-05,
-      "loss": 4.5152,
-      "step": 1135
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.0778816199376946e-05,
-      "loss": 4.2475,
-      "step": 1136
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.0700934579439255e-05,
-      "loss": 4.35,
-      "step": 1137
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.062305295950156e-05,
-      "loss": 4.396,
-      "step": 1138
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.0545171339563866e-05,
-      "loss": 4.118,
-      "step": 1139
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.046728971962617e-05,
-      "loss": 4.3626,
-      "step": 1140
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.038940809968847e-05,
-      "loss": 4.2797,
-      "step": 1141
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.031152647975078e-05,
-      "loss": 4.5896,
-      "step": 1142
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.023364485981309e-05,
-      "loss": 4.4953,
-      "step": 1143
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.0155763239875384e-05,
-      "loss": 4.668,
-      "step": 1144
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5.007788161993769e-05,
-      "loss": 4.2498,
-      "step": 1145
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 5e-05,
-      "loss": 4.4041,
-      "step": 1146
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 4.992211838006231e-05,
-      "loss": 4.3739,
-      "step": 1147
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 4.9844236760124614e-05,
-      "loss": 4.4117,
-      "step": 1148
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 4.976635514018692e-05,
-      "loss": 4.6354,
-      "step": 1149
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 4.9688473520249225e-05,
-      "loss": 4.2319,
-      "step": 1150
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.9610591900311534e-05,
-      "loss": 4.2423,
-      "step": 1151
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.9532710280373836e-05,
-      "loss": 4.4923,
-      "step": 1152
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.9454828660436145e-05,
-      "loss": 4.2873,
-      "step": 1153
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.937694704049845e-05,
-      "loss": 4.4257,
-      "step": 1154
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.929906542056075e-05,
-      "loss": 4.3293,
-      "step": 1155
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.922118380062305e-05,
-      "loss": 4.4787,
-      "step": 1156
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.914330218068536e-05,
-      "loss": 4.1489,
-      "step": 1157
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.9065420560747664e-05,
-      "loss": 4.3449,
-      "step": 1158
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.898753894080997e-05,
-      "loss": 4.1195,
-      "step": 1159
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.8909657320872275e-05,
-      "loss": 4.2664,
-      "step": 1160
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.8831775700934584e-05,
-      "loss": 4.3815,
-      "step": 1161
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.8753894080996886e-05,
-      "loss": 4.3555,
-      "step": 1162
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.8676012461059195e-05,
-      "loss": 4.367,
-      "step": 1163
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.85981308411215e-05,
-      "loss": 4.0633,
-      "step": 1164
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.852024922118381e-05,
-      "loss": 4.1204,
-      "step": 1165
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.844236760124611e-05,
-      "loss": 4.415,
-      "step": 1166
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.836448598130842e-05,
-      "loss": 4.1851,
-      "step": 1167
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 4.828660436137072e-05,
-      "loss": 4.0629,
-      "step": 1168
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.820872274143303e-05,
-      "loss": 4.3468,
-      "step": 1169
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.813084112149533e-05,
-      "loss": 4.436,
-      "step": 1170
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.8052959501557634e-05,
-      "loss": 4.2286,
-      "step": 1171
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.797507788161994e-05,
-      "loss": 4.0717,
-      "step": 1172
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.7897196261682245e-05,
-      "loss": 4.5554,
-      "step": 1173
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.781931464174455e-05,
-      "loss": 4.1033,
-      "step": 1174
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.774143302180686e-05,
-      "loss": 4.2497,
-      "step": 1175
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.766355140186916e-05,
-      "loss": 4.5333,
-      "step": 1176
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.758566978193147e-05,
-      "loss": 4.4551,
-      "step": 1177
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.750778816199377e-05,
-      "loss": 4.5887,
-      "step": 1178
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.742990654205608e-05,
-      "loss": 3.9683,
-      "step": 1179
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.735202492211838e-05,
-      "loss": 4.2246,
-      "step": 1180
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.727414330218069e-05,
-      "loss": 4.0693,
-      "step": 1181
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.719626168224299e-05,
-      "loss": 4.5678,
-      "step": 1182
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.71183800623053e-05,
-      "loss": 4.099,
-      "step": 1183
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.7040498442367604e-05,
-      "loss": 4.0786,
-      "step": 1184
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.6962616822429913e-05,
-      "loss": 4.9068,
-      "step": 1185
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 4.6884735202492216e-05,
-      "loss": 4.0923,
-      "step": 1186
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.6806853582554525e-05,
-      "loss": 4.6307,
-      "step": 1187
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.672897196261683e-05,
-      "loss": 4.2289,
-      "step": 1188
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.665109034267913e-05,
-      "loss": 4.2041,
-      "step": 1189
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.657320872274143e-05,
-      "loss": 4.1228,
-      "step": 1190
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.649532710280374e-05,
-      "loss": 4.3065,
-      "step": 1191
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.641744548286604e-05,
-      "loss": 4.6191,
-      "step": 1192
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.633956386292835e-05,
-      "loss": 4.6268,
-      "step": 1193
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.6261682242990654e-05,
-      "loss": 4.2466,
-      "step": 1194
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.6183800623052963e-05,
-      "loss": 4.8836,
-      "step": 1195
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.6105919003115266e-05,
-      "loss": 4.2361,
-      "step": 1196
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.6028037383177575e-05,
-      "loss": 4.2494,
-      "step": 1197
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.595015576323988e-05,
-      "loss": 4.509,
-      "step": 1198
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.5872274143302186e-05,
-      "loss": 4.4354,
-      "step": 1199
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.579439252336449e-05,
-      "loss": 3.9719,
-      "step": 1200
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.57165109034268e-05,
-      "loss": 4.3422,
-      "step": 1201
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.56386292834891e-05,
-      "loss": 4.2472,
-      "step": 1202
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.556074766355141e-05,
-      "loss": 4.246,
-      "step": 1203
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 4.548286604361371e-05,
-      "loss": 4.0319,
-      "step": 1204
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.540498442367602e-05,
-      "loss": 3.9719,
-      "step": 1205
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.532710280373832e-05,
-      "loss": 4.3047,
-      "step": 1206
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.5249221183800625e-05,
-      "loss": 4.4972,
-      "step": 1207
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.517133956386293e-05,
-      "loss": 4.2166,
-      "step": 1208
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.5093457943925236e-05,
-      "loss": 4.2779,
-      "step": 1209
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.501557632398754e-05,
-      "loss": 4.2135,
-      "step": 1210
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.493769470404985e-05,
-      "loss": 4.439,
-      "step": 1211
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.485981308411215e-05,
-      "loss": 4.5173,
-      "step": 1212
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.478193146417446e-05,
-      "loss": 4.3147,
-      "step": 1213
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.470404984423676e-05,
-      "loss": 4.1052,
-      "step": 1214
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.462616822429907e-05,
-      "loss": 4.1973,
-      "step": 1215
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.454828660436137e-05,
-      "loss": 4.2971,
-      "step": 1216
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.447040498442368e-05,
-      "loss": 4.1608,
-      "step": 1217
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.4392523364485984e-05,
-      "loss": 3.9726,
-      "step": 1218
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.431464174454829e-05,
-      "loss": 3.9738,
-      "step": 1219
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.4236760124610595e-05,
-      "loss": 4.4955,
-      "step": 1220
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.4158878504672904e-05,
-      "loss": 4.2195,
-      "step": 1221
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 4.4080996884735206e-05,
-      "loss": 3.9199,
-      "step": 1222
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.400311526479751e-05,
-      "loss": 4.3446,
-      "step": 1223
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.392523364485982e-05,
-      "loss": 4.4153,
-      "step": 1224
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.384735202492212e-05,
-      "loss": 4.7262,
-      "step": 1225
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.376947040498442e-05,
-      "loss": 4.2929,
-      "step": 1226
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.369158878504673e-05,
-      "loss": 4.0987,
-      "step": 1227
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.3613707165109034e-05,
-      "loss": 4.0939,
-      "step": 1228
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.353582554517134e-05,
-      "loss": 4.1691,
-      "step": 1229
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.3457943925233645e-05,
-      "loss": 4.4065,
-      "step": 1230
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.3380062305295954e-05,
-      "loss": 4.3661,
-      "step": 1231
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.3302180685358256e-05,
-      "loss": 4.2481,
-      "step": 1232
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.3224299065420565e-05,
-      "loss": 4.2708,
-      "step": 1233
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.314641744548287e-05,
-      "loss": 4.0901,
-      "step": 1234
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.306853582554518e-05,
-      "loss": 4.5713,
-      "step": 1235
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.299065420560748e-05,
-      "loss": 4.1857,
-      "step": 1236
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.291277258566979e-05,
-      "loss": 4.3475,
-      "step": 1237
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.283489096573209e-05,
-      "loss": 4.0865,
-      "step": 1238
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 4.27570093457944e-05,
-      "loss": 4.4001,
-      "step": 1239
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.26791277258567e-05,
-      "loss": 4.5291,
-      "step": 1240
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.2601246105919004e-05,
-      "loss": 4.2773,
-      "step": 1241
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.2523364485981306e-05,
-      "loss": 4.6681,
-      "step": 1242
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.2445482866043615e-05,
-      "loss": 4.3623,
-      "step": 1243
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.236760124610592e-05,
-      "loss": 4.2711,
-      "step": 1244
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.228971962616823e-05,
-      "loss": 3.9714,
-      "step": 1245
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.221183800623053e-05,
-      "loss": 4.4319,
-      "step": 1246
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.213395638629284e-05,
-      "loss": 4.4446,
-      "step": 1247
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.205607476635514e-05,
-      "loss": 4.2391,
-      "step": 1248
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.197819314641745e-05,
-      "loss": 4.437,
-      "step": 1249
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.190031152647975e-05,
-      "loss": 4.799,
-      "step": 1250
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.182242990654206e-05,
-      "loss": 4.4402,
-      "step": 1251
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.174454828660436e-05,
-      "loss": 4.4421,
-      "step": 1252
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 4.3281,
-      "step": 1253
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.1588785046728974e-05,
-      "loss": 4.3467,
-      "step": 1254
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.1510903426791283e-05,
-      "loss": 4.5584,
-      "step": 1255
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.1433021806853586e-05,
-      "loss": 4.0999,
-      "step": 1256
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 4.1355140186915895e-05,
-      "loss": 4.0705,
-      "step": 1257
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.12772585669782e-05,
-      "loss": 3.967,
-      "step": 1258
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.11993769470405e-05,
-      "loss": 4.3453,
-      "step": 1259
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.11214953271028e-05,
-      "loss": 4.5452,
-      "step": 1260
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.104361370716511e-05,
-      "loss": 4.0491,
-      "step": 1261
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.096573208722741e-05,
-      "loss": 4.3795,
-      "step": 1262
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.088785046728972e-05,
-      "loss": 4.3197,
-      "step": 1263
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.0809968847352024e-05,
-      "loss": 4.2348,
-      "step": 1264
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.073208722741433e-05,
-      "loss": 4.8607,
-      "step": 1265
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.0654205607476636e-05,
-      "loss": 4.3463,
-      "step": 1266
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.0576323987538945e-05,
-      "loss": 4.3886,
-      "step": 1267
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.049844236760125e-05,
-      "loss": 4.23,
-      "step": 1268
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.0420560747663556e-05,
-      "loss": 4.3879,
-      "step": 1269
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.034267912772586e-05,
-      "loss": 4.2232,
-      "step": 1270
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.026479750778817e-05,
-      "loss": 4.3146,
-      "step": 1271
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.018691588785047e-05,
-      "loss": 4.2428,
-      "step": 1272
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.010903426791278e-05,
-      "loss": 4.4687,
-      "step": 1273
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.003115264797508e-05,
-      "loss": 4.138,
-      "step": 1274
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 3.995327102803739e-05,
-      "loss": 4.1916,
-      "step": 1275
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.987538940809969e-05,
-      "loss": 4.6077,
-      "step": 1276
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.9797507788161995e-05,
-      "loss": 4.3284,
-      "step": 1277
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.97196261682243e-05,
-      "loss": 4.3357,
-      "step": 1278
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.9641744548286606e-05,
-      "loss": 4.4382,
-      "step": 1279
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.956386292834891e-05,
-      "loss": 4.3991,
-      "step": 1280
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.948598130841122e-05,
-      "loss": 4.2626,
-      "step": 1281
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.940809968847352e-05,
-      "loss": 4.2226,
-      "step": 1282
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.933021806853583e-05,
-      "loss": 4.4283,
-      "step": 1283
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.925233644859813e-05,
-      "loss": 4.1638,
-      "step": 1284
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.917445482866044e-05,
-      "loss": 4.2195,
-      "step": 1285
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.909657320872274e-05,
-      "loss": 4.2536,
-      "step": 1286
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.901869158878505e-05,
-      "loss": 4.2326,
-      "step": 1287
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.8940809968847354e-05,
-      "loss": 4.1957,
-      "step": 1288
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.886292834890966e-05,
-      "loss": 4.2866,
-      "step": 1289
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.8785046728971965e-05,
-      "loss": 4.0611,
-      "step": 1290
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.8707165109034274e-05,
-      "loss": 4.4948,
-      "step": 1291
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.8629283489096576e-05,
-      "loss": 4.4958,
-      "step": 1292
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.855140186915888e-05,
-      "loss": 4.865,
-      "step": 1293
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.847352024922119e-05,
-      "loss": 4.0426,
-      "step": 1294
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.839563862928349e-05,
-      "loss": 4.052,
-      "step": 1295
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.831775700934579e-05,
-      "loss": 3.93,
-      "step": 1296
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.82398753894081e-05,
-      "loss": 4.0005,
-      "step": 1297
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.8161993769470404e-05,
-      "loss": 3.8967,
-      "step": 1298
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.808411214953271e-05,
-      "loss": 4.9155,
-      "step": 1299
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.8006230529595015e-05,
-      "loss": 3.9032,
-      "step": 1300
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.7928348909657324e-05,
-      "loss": 4.3688,
-      "step": 1301
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.7850467289719626e-05,
-      "loss": 4.3931,
-      "step": 1302
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.7772585669781935e-05,
-      "loss": 4.373,
-      "step": 1303
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.769470404984424e-05,
-      "loss": 4.2493,
-      "step": 1304
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.761682242990655e-05,
-      "loss": 4.1779,
-      "step": 1305
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.753894080996885e-05,
-      "loss": 4.1601,
-      "step": 1306
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.746105919003116e-05,
-      "loss": 4.1706,
-      "step": 1307
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.738317757009346e-05,
-      "loss": 3.9523,
-      "step": 1308
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.730529595015577e-05,
-      "loss": 4.1229,
-      "step": 1309
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.722741433021807e-05,
-      "loss": 4.0223,
-      "step": 1310
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.7149532710280374e-05,
-      "loss": 4.1257,
-      "step": 1311
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.7071651090342676e-05,
-      "loss": 4.3985,
-      "step": 1312
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.6993769470404985e-05,
-      "loss": 4.1552,
-      "step": 1313
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.691588785046729e-05,
-      "loss": 4.0948,
-      "step": 1314
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.68380062305296e-05,
-      "loss": 4.4219,
-      "step": 1315
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.67601246105919e-05,
-      "loss": 4.2809,
-      "step": 1316
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.668224299065421e-05,
-      "loss": 4.0397,
-      "step": 1317
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.660436137071651e-05,
-      "loss": 4.7476,
-      "step": 1318
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.652647975077882e-05,
-      "loss": 4.1158,
-      "step": 1319
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.644859813084112e-05,
-      "loss": 3.8913,
-      "step": 1320
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.637071651090343e-05,
-      "loss": 3.8661,
-      "step": 1321
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.629283489096573e-05,
-      "loss": 4.1589,
-      "step": 1322
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.621495327102804e-05,
-      "loss": 4.1017,
-      "step": 1323
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.6137071651090344e-05,
-      "loss": 4.1095,
-      "step": 1324
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.605919003115265e-05,
-      "loss": 4.1554,
-      "step": 1325
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.5981308411214956e-05,
-      "loss": 4.0112,
-      "step": 1326
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.5903426791277265e-05,
-      "loss": 3.9658,
-      "step": 1327
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.582554517133957e-05,
-      "loss": 4.2842,
-      "step": 1328
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 3.574766355140187e-05,
-      "loss": 4.391,
-      "step": 1329
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.566978193146417e-05,
-      "loss": 4.466,
-      "step": 1330
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.559190031152648e-05,
-      "loss": 4.0828,
-      "step": 1331
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.551401869158878e-05,
-      "loss": 4.8758,
-      "step": 1332
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.543613707165109e-05,
-      "loss": 4.2157,
-      "step": 1333
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.5358255451713394e-05,
-      "loss": 4.2764,
-      "step": 1334
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.52803738317757e-05,
-      "loss": 4.0116,
-      "step": 1335
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.5202492211838006e-05,
-      "loss": 4.324,
-      "step": 1336
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.5124610591900315e-05,
-      "loss": 4.2538,
-      "step": 1337
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.504672897196262e-05,
-      "loss": 4.376,
-      "step": 1338
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.4968847352024926e-05,
-      "loss": 4.4872,
-      "step": 1339
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.489096573208723e-05,
-      "loss": 4.557,
-      "step": 1340
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.481308411214954e-05,
-      "loss": 4.4586,
-      "step": 1341
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.473520249221184e-05,
-      "loss": 4.3354,
-      "step": 1342
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.465732087227415e-05,
-      "loss": 4.4036,
-      "step": 1343
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.457943925233645e-05,
-      "loss": 4.7248,
-      "step": 1344
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.450155763239875e-05,
-      "loss": 4.4862,
-      "step": 1345
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.442367601246106e-05,
-      "loss": 4.3865,
-      "step": 1346
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.4345794392523365e-05,
-      "loss": 4.8887,
-      "step": 1347
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.426791277258567e-05,
-      "loss": 4.0969,
-      "step": 1348
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.4190031152647976e-05,
-      "loss": 4.2285,
-      "step": 1349
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.411214953271028e-05,
-      "loss": 4.4396,
-      "step": 1350
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.403426791277259e-05,
-      "loss": 4.3677,
-      "step": 1351
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.395638629283489e-05,
-      "loss": 4.3754,
-      "step": 1352
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.38785046728972e-05,
-      "loss": 4.3757,
-      "step": 1353
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.38006230529595e-05,
-      "loss": 4.373,
-      "step": 1354
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.372274143302181e-05,
-      "loss": 4.1201,
-      "step": 1355
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.364485981308411e-05,
-      "loss": 4.495,
-      "step": 1356
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.356697819314642e-05,
-      "loss": 3.9332,
-      "step": 1357
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.3489096573208724e-05,
-      "loss": 4.3883,
-      "step": 1358
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.341121495327103e-05,
-      "loss": 4.515,
-      "step": 1359
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 4.6292,
-      "step": 1360
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.3255451713395644e-05,
-      "loss": 4.2248,
-      "step": 1361
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.3177570093457946e-05,
-      "loss": 4.0747,
-      "step": 1362
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.309968847352025e-05,
-      "loss": 4.0186,
-      "step": 1363
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 3.302180685358255e-05,
-      "loss": 4.544,
-      "step": 1364
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.294392523364486e-05,
-      "loss": 4.1644,
-      "step": 1365
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.286604361370716e-05,
-      "loss": 3.9352,
-      "step": 1366
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.278816199376947e-05,
-      "loss": 4.4515,
-      "step": 1367
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.2710280373831774e-05,
-      "loss": 4.0316,
-      "step": 1368
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.263239875389408e-05,
-      "loss": 4.398,
-      "step": 1369
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.2554517133956385e-05,
-      "loss": 4.3897,
-      "step": 1370
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.2476635514018694e-05,
-      "loss": 4.4292,
-      "step": 1371
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.2398753894080996e-05,
-      "loss": 3.9847,
-      "step": 1372
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.2320872274143305e-05,
-      "loss": 4.2985,
-      "step": 1373
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.224299065420561e-05,
-      "loss": 4.3266,
-      "step": 1374
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.216510903426792e-05,
-      "loss": 4.0702,
-      "step": 1375
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.208722741433022e-05,
-      "loss": 3.9473,
-      "step": 1376
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.200934579439253e-05,
-      "loss": 4.2026,
-      "step": 1377
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.193146417445483e-05,
-      "loss": 4.146,
-      "step": 1378
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.185358255451714e-05,
-      "loss": 4.1708,
-      "step": 1379
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.177570093457944e-05,
-      "loss": 4.2791,
-      "step": 1380
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.1697819314641744e-05,
-      "loss": 3.8482,
-      "step": 1381
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.1619937694704046e-05,
-      "loss": 4.3895,
-      "step": 1382
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.1542056074766355e-05,
-      "loss": 4.1216,
-      "step": 1383
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.146417445482866e-05,
-      "loss": 4.3121,
-      "step": 1384
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.1386292834890967e-05,
-      "loss": 4.422,
-      "step": 1385
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.130841121495327e-05,
-      "loss": 4.321,
-      "step": 1386
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.123052959501558e-05,
-      "loss": 4.3969,
-      "step": 1387
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.115264797507788e-05,
-      "loss": 4.3011,
-      "step": 1388
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.107476635514019e-05,
-      "loss": 4.1768,
-      "step": 1389
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.099688473520249e-05,
-      "loss": 4.3812,
-      "step": 1390
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.09190031152648e-05,
-      "loss": 4.1121,
-      "step": 1391
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.08411214953271e-05,
-      "loss": 4.0353,
-      "step": 1392
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.076323987538941e-05,
-      "loss": 4.2014,
-      "step": 1393
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.0685358255451714e-05,
-      "loss": 4.4223,
-      "step": 1394
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.060747663551402e-05,
-      "loss": 5.0309,
-      "step": 1395
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.0529595015576326e-05,
-      "loss": 4.1576,
-      "step": 1396
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.045171339563863e-05,
-      "loss": 4.4707,
-      "step": 1397
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.0373831775700934e-05,
-      "loss": 4.1187,
-      "step": 1398
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.0295950155763243e-05,
-      "loss": 4.2561,
-      "step": 1399
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.0218068535825545e-05,
-      "loss": 4.3858,
-      "step": 1400
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.0140186915887854e-05,
-      "loss": 3.9948,
-      "step": 1401
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.0062305295950156e-05,
-      "loss": 4.3427,
-      "step": 1402
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9984423676012462e-05,
-      "loss": 4.0433,
-      "step": 1403
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9906542056074764e-05,
-      "loss": 4.0749,
-      "step": 1404
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9828660436137073e-05,
-      "loss": 3.9265,
-      "step": 1405
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9750778816199376e-05,
-      "loss": 4.7849,
-      "step": 1406
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9672897196261685e-05,
-      "loss": 4.3203,
-      "step": 1407
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9595015576323987e-05,
-      "loss": 4.0094,
-      "step": 1408
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9517133956386296e-05,
-      "loss": 4.0389,
-      "step": 1409
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9439252336448598e-05,
-      "loss": 4.036,
-      "step": 1410
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9361370716510904e-05,
-      "loss": 3.9625,
-      "step": 1411
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9283489096573206e-05,
-      "loss": 4.0096,
-      "step": 1412
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9205607476635515e-05,
-      "loss": 4.3164,
-      "step": 1413
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9127725856697818e-05,
-      "loss": 4.4349,
-      "step": 1414
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.9049844236760127e-05,
-      "loss": 4.1878,
-      "step": 1415
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.897196261682243e-05,
-      "loss": 4.3408,
-      "step": 1416
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.8894080996884738e-05,
-      "loss": 4.2447,
-      "step": 1417
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 2.881619937694704e-05,
-      "loss": 4.2827,
-      "step": 1418
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.873831775700935e-05,
-      "loss": 4.3,
-      "step": 1419
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.866043613707165e-05,
-      "loss": 4.0295,
-      "step": 1420
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.8582554517133957e-05,
-      "loss": 4.1759,
-      "step": 1421
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.850467289719626e-05,
-      "loss": 4.2377,
-      "step": 1422
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.842679127725857e-05,
-      "loss": 4.203,
-      "step": 1423
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.834890965732087e-05,
-      "loss": 4.2818,
-      "step": 1424
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.827102803738318e-05,
-      "loss": 4.4467,
-      "step": 1425
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.8193146417445482e-05,
-      "loss": 4.3404,
-      "step": 1426
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.811526479750779e-05,
-      "loss": 4.0546,
-      "step": 1427
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.8037383177570094e-05,
-      "loss": 4.0267,
-      "step": 1428
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.79595015576324e-05,
-      "loss": 4.2553,
-      "step": 1429
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.78816199376947e-05,
-      "loss": 4.5516,
-      "step": 1430
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.780373831775701e-05,
-      "loss": 4.4551,
-      "step": 1431
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.7725856697819313e-05,
-      "loss": 4.3037,
-      "step": 1432
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.7647975077881622e-05,
-      "loss": 4.6152,
-      "step": 1433
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.7570093457943924e-05,
-      "loss": 4.0649,
-      "step": 1434
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.7492211838006233e-05,
-      "loss": 3.9965,
-      "step": 1435
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 2.7414330218068536e-05,
-      "loss": 4.0527,
-      "step": 1436
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.733644859813084e-05,
-      "loss": 4.3788,
-      "step": 1437
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.7258566978193147e-05,
-      "loss": 4.0924,
-      "step": 1438
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.7180685358255453e-05,
-      "loss": 4.2613,
-      "step": 1439
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.7102803738317755e-05,
-      "loss": 4.211,
-      "step": 1440
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.7024922118380064e-05,
-      "loss": 4.5365,
-      "step": 1441
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6947040498442366e-05,
-      "loss": 4.0323,
-      "step": 1442
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6869158878504675e-05,
-      "loss": 3.9741,
-      "step": 1443
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6791277258566978e-05,
-      "loss": 4.1318,
-      "step": 1444
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6713395638629287e-05,
-      "loss": 4.1568,
-      "step": 1445
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.663551401869159e-05,
-      "loss": 4.6068,
-      "step": 1446
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6557632398753895e-05,
-      "loss": 4.2433,
-      "step": 1447
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6479750778816197e-05,
-      "loss": 4.4372,
-      "step": 1448
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6401869158878506e-05,
-      "loss": 4.4818,
-      "step": 1449
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6323987538940808e-05,
-      "loss": 4.0195,
-      "step": 1450
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.6246105919003117e-05,
-      "loss": 4.1391,
-      "step": 1451
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.616822429906542e-05,
-      "loss": 4.2419,
-      "step": 1452
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 2.609034267912773e-05,
-      "loss": 4.081,
-      "step": 1453
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.601246105919003e-05,
-      "loss": 4.0568,
-      "step": 1454
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5934579439252337e-05,
-      "loss": 4.4954,
-      "step": 1455
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.585669781931464e-05,
-      "loss": 4.1372,
-      "step": 1456
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5778816199376948e-05,
-      "loss": 4.5402,
-      "step": 1457
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.570093457943925e-05,
-      "loss": 4.2944,
-      "step": 1458
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.562305295950156e-05,
-      "loss": 4.1349,
-      "step": 1459
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.554517133956386e-05,
-      "loss": 4.4732,
-      "step": 1460
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.546728971962617e-05,
-      "loss": 4.2236,
-      "step": 1461
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5389408099688473e-05,
-      "loss": 4.3118,
-      "step": 1462
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.531152647975078e-05,
-      "loss": 4.1629,
-      "step": 1463
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5233644859813084e-05,
-      "loss": 4.3452,
-      "step": 1464
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.515576323987539e-05,
-      "loss": 4.5502,
-      "step": 1465
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5077881619937692e-05,
-      "loss": 3.98,
-      "step": 1466
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.5e-05,
-      "loss": 4.1132,
-      "step": 1467
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.4922118380062307e-05,
-      "loss": 4.2159,
-      "step": 1468
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.4844236760124613e-05,
-      "loss": 3.8988,
-      "step": 1469
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.4766355140186918e-05,
-      "loss": 3.9275,
-      "step": 1470
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 2.4688473520249224e-05,
-      "loss": 4.2077,
-      "step": 1471
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4610591900311526e-05,
-      "loss": 3.9493,
-      "step": 1472
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4532710280373832e-05,
-      "loss": 4.0551,
-      "step": 1473
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4454828660436138e-05,
-      "loss": 4.0572,
-      "step": 1474
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4376947040498443e-05,
-      "loss": 4.2404,
-      "step": 1475
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.429906542056075e-05,
-      "loss": 4.3425,
-      "step": 1476
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4221183800623055e-05,
-      "loss": 4.2847,
-      "step": 1477
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.414330218068536e-05,
-      "loss": 3.8309,
-      "step": 1478
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.4065420560747666e-05,
-      "loss": 4.2097,
-      "step": 1479
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.398753894080997e-05,
-      "loss": 4.011,
-      "step": 1480
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3909657320872274e-05,
-      "loss": 4.0915,
-      "step": 1481
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.383177570093458e-05,
-      "loss": 4.6853,
-      "step": 1482
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3753894080996885e-05,
-      "loss": 4.224,
-      "step": 1483
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.367601246105919e-05,
-      "loss": 3.7889,
-      "step": 1484
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3598130841121497e-05,
-      "loss": 4.1848,
-      "step": 1485
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3520249221183802e-05,
-      "loss": 4.3708,
-      "step": 1486
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3442367601246108e-05,
-      "loss": 4.278,
-      "step": 1487
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3364485981308414e-05,
-      "loss": 4.2989,
-      "step": 1488
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.3286604361370716e-05,
-      "loss": 4.2445,
-      "step": 1489
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.320872274143302e-05,
-      "loss": 3.9499,
-      "step": 1490
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.3130841121495327e-05,
-      "loss": 4.5518,
-      "step": 1491
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.3052959501557633e-05,
-      "loss": 4.1498,
-      "step": 1492
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.297507788161994e-05,
-      "loss": 3.7795,
-      "step": 1493
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2897196261682244e-05,
-      "loss": 4.0954,
-      "step": 1494
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.281931464174455e-05,
-      "loss": 4.1334,
-      "step": 1495
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2741433021806856e-05,
-      "loss": 5.0529,
-      "step": 1496
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.266355140186916e-05,
-      "loss": 4.2893,
-      "step": 1497
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2585669781931463e-05,
-      "loss": 4.3939,
-      "step": 1498
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.250778816199377e-05,
-      "loss": 3.9323,
-      "step": 1499
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2429906542056075e-05,
-      "loss": 4.1345,
-      "step": 1500
-    },
-    {
-      "epoch": 0.84,
-      "eval_loss": 4.305478096008301,
-      "eval_runtime": 911.7255,
-      "eval_samples_per_second": 2.898,
-      "eval_steps_per_second": 0.363,
-      "eval_wer": 1.9280642602142006,
-      "step": 1500
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.235202492211838e-05,
-      "loss": 4.1455,
-      "step": 1501
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2274143302180686e-05,
-      "loss": 4.193,
-      "step": 1502
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2196261682242992e-05,
-      "loss": 4.3586,
-      "step": 1503
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2118380062305298e-05,
-      "loss": 4.1367,
-      "step": 1504
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.2040498442367603e-05,
-      "loss": 4.3236,
-      "step": 1505
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.196261682242991e-05,
-      "loss": 4.1133,
-      "step": 1506
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.188473520249221e-05,
-      "loss": 4.0626,
-      "step": 1507
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1806853582554517e-05,
-      "loss": 4.1023,
-      "step": 1508
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1728971962616822e-05,
-      "loss": 4.1873,
-      "step": 1509
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1651090342679128e-05,
-      "loss": 4.1271,
-      "step": 1510
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1573208722741434e-05,
-      "loss": 3.9457,
-      "step": 1511
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.149532710280374e-05,
-      "loss": 4.2579,
-      "step": 1512
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1417445482866045e-05,
-      "loss": 3.8794,
-      "step": 1513
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.133956386292835e-05,
-      "loss": 4.0571,
-      "step": 1514
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1261682242990653e-05,
-      "loss": 4.1546,
-      "step": 1515
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.118380062305296e-05,
-      "loss": 4.4866,
-      "step": 1516
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.1105919003115264e-05,
-      "loss": 4.0309,
-      "step": 1517
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.102803738317757e-05,
-      "loss": 4.2736,
-      "step": 1518
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.0950155763239876e-05,
-      "loss": 4.3651,
-      "step": 1519
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.087227414330218e-05,
-      "loss": 4.1066,
-      "step": 1520
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.0794392523364487e-05,
-      "loss": 4.2309,
-      "step": 1521
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.0716510903426793e-05,
-      "loss": 4.1243,
-      "step": 1522
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.06386292834891e-05,
-      "loss": 4.36,
-      "step": 1523
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.05607476635514e-05,
-      "loss": 3.9723,
-      "step": 1524
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.0482866043613706e-05,
-      "loss": 4.0935,
-      "step": 1525
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.0404984423676012e-05,
-      "loss": 3.9896,
-      "step": 1526
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.0327102803738318e-05,
-      "loss": 4.0823,
-      "step": 1527
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.0249221183800623e-05,
-      "loss": 4.1029,
-      "step": 1528
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.017133956386293e-05,
-      "loss": 3.6848,
-      "step": 1529
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.0093457943925235e-05,
-      "loss": 3.9678,
-      "step": 1530
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.001557632398754e-05,
-      "loss": 3.9976,
-      "step": 1531
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9937694704049846e-05,
-      "loss": 3.9839,
-      "step": 1532
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.985981308411215e-05,
-      "loss": 4.146,
-      "step": 1533
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9781931464174454e-05,
-      "loss": 4.0741,
-      "step": 1534
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.970404984423676e-05,
-      "loss": 4.1238,
-      "step": 1535
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9626168224299065e-05,
-      "loss": 3.8155,
-      "step": 1536
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.954828660436137e-05,
-      "loss": 4.2027,
-      "step": 1537
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9470404984423677e-05,
-      "loss": 4.3859,
-      "step": 1538
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9392523364485982e-05,
-      "loss": 4.0954,
-      "step": 1539
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9314641744548288e-05,
-      "loss": 4.2721,
-      "step": 1540
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9236760124610594e-05,
-      "loss": 4.3526,
-      "step": 1541
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9158878504672896e-05,
-      "loss": 3.9384,
-      "step": 1542
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 1.9080996884735202e-05,
-      "loss": 4.3128,
-      "step": 1543
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.9003115264797507e-05,
-      "loss": 4.1997,
-      "step": 1544
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8925233644859813e-05,
-      "loss": 4.6874,
-      "step": 1545
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.884735202492212e-05,
-      "loss": 3.8704,
-      "step": 1546
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8769470404984424e-05,
-      "loss": 4.1862,
-      "step": 1547
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.869158878504673e-05,
-      "loss": 4.1832,
-      "step": 1548
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8613707165109036e-05,
-      "loss": 4.6238,
-      "step": 1549
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8535825545171338e-05,
-      "loss": 3.9044,
-      "step": 1550
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8457943925233644e-05,
-      "loss": 4.4059,
-      "step": 1551
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.838006230529595e-05,
-      "loss": 4.1995,
-      "step": 1552
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8302180685358255e-05,
-      "loss": 4.051,
-      "step": 1553
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.822429906542056e-05,
-      "loss": 4.0623,
-      "step": 1554
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8146417445482866e-05,
-      "loss": 4.4478,
-      "step": 1555
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.8068535825545172e-05,
-      "loss": 4.5505,
-      "step": 1556
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.7990654205607478e-05,
-      "loss": 4.211,
-      "step": 1557
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.7912772585669783e-05,
-      "loss": 4.3894,
-      "step": 1558
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.7834890965732086e-05,
-      "loss": 4.079,
-      "step": 1559
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 1.775700934579439e-05,
-      "loss": 4.3031,
-      "step": 1560
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.7679127725856697e-05,
-      "loss": 4.2886,
-      "step": 1561
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.7601246105919003e-05,
-      "loss": 4.1584,
-      "step": 1562
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.752336448598131e-05,
-      "loss": 4.2858,
-      "step": 1563
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.7445482866043614e-05,
-      "loss": 4.0312,
-      "step": 1564
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.736760124610592e-05,
-      "loss": 4.2351,
-      "step": 1565
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.7289719626168225e-05,
-      "loss": 3.9863,
-      "step": 1566
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.721183800623053e-05,
-      "loss": 3.9364,
-      "step": 1567
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.7133956386292833e-05,
-      "loss": 4.2124,
-      "step": 1568
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.705607476635514e-05,
-      "loss": 4.2369,
-      "step": 1569
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6978193146417445e-05,
-      "loss": 4.0026,
-      "step": 1570
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.690031152647975e-05,
-      "loss": 4.1091,
-      "step": 1571
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6822429906542056e-05,
-      "loss": 4.3581,
-      "step": 1572
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6744548286604362e-05,
-      "loss": 4.2328,
-      "step": 1573
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6666666666666667e-05,
-      "loss": 3.9551,
-      "step": 1574
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6588785046728973e-05,
-      "loss": 3.9855,
-      "step": 1575
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6510903426791275e-05,
-      "loss": 4.2361,
-      "step": 1576
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.643302180685358e-05,
-      "loss": 3.997,
-      "step": 1577
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 1.6355140186915887e-05,
-      "loss": 4.2827,
-      "step": 1578
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.6277258566978192e-05,
-      "loss": 4.249,
-      "step": 1579
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.6199376947040498e-05,
-      "loss": 4.4398,
-      "step": 1580
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.6121495327102804e-05,
-      "loss": 4.2462,
-      "step": 1581
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.604361370716511e-05,
-      "loss": 4.2239,
-      "step": 1582
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5965732087227415e-05,
-      "loss": 4.0932,
-      "step": 1583
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.588785046728972e-05,
-      "loss": 4.0289,
-      "step": 1584
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5809968847352023e-05,
-      "loss": 3.9372,
-      "step": 1585
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.573208722741433e-05,
-      "loss": 4.1108,
-      "step": 1586
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5654205607476634e-05,
-      "loss": 4.0599,
-      "step": 1587
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.557632398753894e-05,
-      "loss": 4.2325,
-      "step": 1588
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5498442367601246e-05,
-      "loss": 4.2809,
-      "step": 1589
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.542056074766355e-05,
-      "loss": 3.9384,
-      "step": 1590
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5342679127725857e-05,
-      "loss": 4.0852,
-      "step": 1591
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5264797507788163e-05,
-      "loss": 3.9348,
-      "step": 1592
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5186915887850467e-05,
-      "loss": 4.0039,
-      "step": 1593
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5109034267912772e-05,
-      "loss": 4.0862,
-      "step": 1594
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.5031152647975078e-05,
-      "loss": 4.0421,
-      "step": 1595
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.4953271028037382e-05,
-      "loss": 4.2836,
-      "step": 1596
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4875389408099688e-05,
-      "loss": 4.2177,
-      "step": 1597
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4797507788161993e-05,
-      "loss": 4.3741,
-      "step": 1598
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4719626168224299e-05,
-      "loss": 5.0492,
-      "step": 1599
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4641744548286603e-05,
-      "loss": 4.5488,
-      "step": 1600
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4563862928348909e-05,
-      "loss": 4.216,
-      "step": 1601
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4485981308411214e-05,
-      "loss": 3.9483,
-      "step": 1602
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.440809968847352e-05,
-      "loss": 4.1343,
-      "step": 1603
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4330218068535826e-05,
-      "loss": 4.1724,
-      "step": 1604
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.425233644859813e-05,
-      "loss": 3.9089,
-      "step": 1605
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4174454828660435e-05,
-      "loss": 4.1766,
-      "step": 1606
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4096573208722741e-05,
-      "loss": 4.075,
-      "step": 1607
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.4018691588785047e-05,
-      "loss": 4.1685,
-      "step": 1608
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.394080996884735e-05,
-      "loss": 4.2934,
-      "step": 1609
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.3862928348909656e-05,
-      "loss": 4.0326,
-      "step": 1610
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.3785046728971962e-05,
-      "loss": 4.3429,
-      "step": 1611
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.3707165109034268e-05,
-      "loss": 4.0983,
-      "step": 1612
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.3629283489096573e-05,
-      "loss": 4.2354,
-      "step": 1613
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.3551401869158877e-05,
-      "loss": 3.9244,
-      "step": 1614
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3473520249221183e-05,
-      "loss": 4.158,
-      "step": 1615
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3395638629283489e-05,
-      "loss": 4.0944,
-      "step": 1616
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3317757009345794e-05,
-      "loss": 4.2497,
-      "step": 1617
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3239875389408098e-05,
-      "loss": 4.0857,
-      "step": 1618
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3161993769470404e-05,
-      "loss": 4.158,
-      "step": 1619
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.308411214953271e-05,
-      "loss": 4.1413,
-      "step": 1620
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.3006230529595015e-05,
-      "loss": 4.375,
-      "step": 1621
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.292834890965732e-05,
-      "loss": 3.8563,
-      "step": 1622
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2850467289719625e-05,
-      "loss": 4.0692,
-      "step": 1623
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.277258566978193e-05,
-      "loss": 3.9633,
-      "step": 1624
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2694704049844236e-05,
-      "loss": 4.0887,
-      "step": 1625
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2616822429906542e-05,
-      "loss": 4.3703,
-      "step": 1626
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2538940809968846e-05,
-      "loss": 4.1546,
-      "step": 1627
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2461059190031153e-05,
-      "loss": 4.0143,
-      "step": 1628
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2383177570093459e-05,
-      "loss": 4.219,
-      "step": 1629
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2305295950155763e-05,
-      "loss": 4.0389,
-      "step": 1630
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2227414330218069e-05,
-      "loss": 4.1118,
-      "step": 1631
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.2149532710280374e-05,
-      "loss": 4.1591,
-      "step": 1632
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.207165109034268e-05,
-      "loss": 4.2756,
-      "step": 1633
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1993769470404986e-05,
-      "loss": 4.0372,
-      "step": 1634
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.191588785046729e-05,
-      "loss": 4.1718,
-      "step": 1635
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1838006230529595e-05,
-      "loss": 4.0724,
-      "step": 1636
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1760124610591901e-05,
-      "loss": 4.0551,
-      "step": 1637
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1682242990654207e-05,
-      "loss": 4.3195,
-      "step": 1638
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.160436137071651e-05,
-      "loss": 4.0259,
-      "step": 1639
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1526479750778816e-05,
-      "loss": 4.3292,
-      "step": 1640
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1448598130841122e-05,
-      "loss": 4.2155,
-      "step": 1641
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1370716510903428e-05,
-      "loss": 4.0932,
-      "step": 1642
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1292834890965732e-05,
-      "loss": 4.0212,
-      "step": 1643
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1214953271028037e-05,
-      "loss": 4.1187,
-      "step": 1644
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1137071651090343e-05,
-      "loss": 4.3971,
-      "step": 1645
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.1059190031152649e-05,
-      "loss": 4.425,
-      "step": 1646
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.0981308411214954e-05,
-      "loss": 4.1025,
-      "step": 1647
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.0903426791277258e-05,
-      "loss": 4.8495,
-      "step": 1648
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.0825545171339564e-05,
-      "loss": 4.3046,
-      "step": 1649
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.074766355140187e-05,
-      "loss": 3.9236,
-      "step": 1650
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0669781931464175e-05,
-      "loss": 4.1615,
-      "step": 1651
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.059190031152648e-05,
-      "loss": 4.0827,
-      "step": 1652
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0514018691588785e-05,
-      "loss": 4.0824,
-      "step": 1653
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.043613707165109e-05,
-      "loss": 4.2182,
-      "step": 1654
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0358255451713396e-05,
-      "loss": 3.9192,
-      "step": 1655
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.02803738317757e-05,
-      "loss": 4.0556,
-      "step": 1656
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0202492211838006e-05,
-      "loss": 4.3625,
-      "step": 1657
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0124610591900312e-05,
-      "loss": 4.1543,
-      "step": 1658
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.0046728971962617e-05,
-      "loss": 3.9912,
-      "step": 1659
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.968847352024923e-06,
-      "loss": 4.0971,
-      "step": 1660
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.890965732087227e-06,
-      "loss": 4.2835,
-      "step": 1661
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.813084112149533e-06,
-      "loss": 4.1699,
-      "step": 1662
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.735202492211838e-06,
-      "loss": 3.8204,
-      "step": 1663
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.657320872274144e-06,
-      "loss": 4.2402,
-      "step": 1664
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.579439252336448e-06,
-      "loss": 3.9306,
-      "step": 1665
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.501557632398754e-06,
-      "loss": 4.1673,
-      "step": 1666
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.42367601246106e-06,
-      "loss": 4.03,
-      "step": 1667
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.345794392523365e-06,
-      "loss": 4.0042,
-      "step": 1668
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.267912772585669e-06,
-      "loss": 3.9968,
-      "step": 1669
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.190031152647975e-06,
-      "loss": 4.0478,
-      "step": 1670
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.11214953271028e-06,
-      "loss": 4.0592,
-      "step": 1671
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.034267912772586e-06,
-      "loss": 4.3227,
-      "step": 1672
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 8.956386292834892e-06,
-      "loss": 4.0577,
-      "step": 1673
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 8.878504672897196e-06,
-      "loss": 4.2165,
-      "step": 1674
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 8.800623052959501e-06,
-      "loss": 4.4118,
-      "step": 1675
+      "epoch": 0.92,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 4.1314,
+      "step": 823
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.722741433021807e-06,
-      "loss": 3.9917,
-      "step": 1676
+      "epoch": 0.92,
+      "learning_rate": 1.8112244897959187e-05,
+      "loss": 4.1815,
+      "step": 824
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.644859813084113e-06,
-      "loss": 3.8111,
-      "step": 1677
+      "epoch": 0.92,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 4.3073,
+      "step": 825
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.566978193146417e-06,
-      "loss": 4.0834,
-      "step": 1678
+      "epoch": 0.93,
+      "learning_rate": 1.760204081632653e-05,
+      "loss": 4.2567,
+      "step": 826
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.489096573208722e-06,
-      "loss": 4.2499,
-      "step": 1679
+      "epoch": 0.93,
+      "learning_rate": 1.7346938775510206e-05,
+      "loss": 4.1142,
+      "step": 827
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.411214953271028e-06,
-      "loss": 4.3969,
-      "step": 1680
+      "epoch": 0.93,
+      "learning_rate": 1.7091836734693878e-05,
+      "loss": 3.9881,
+      "step": 828
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.333333333333334e-06,
-      "loss": 3.7865,
-      "step": 1681
+      "epoch": 0.93,
+      "learning_rate": 1.683673469387755e-05,
+      "loss": 3.9615,
+      "step": 829
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.255451713395638e-06,
-      "loss": 3.8812,
-      "step": 1682
+      "epoch": 0.93,
+      "learning_rate": 1.6581632653061225e-05,
+      "loss": 4.2256,
+      "step": 830
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.177570093457943e-06,
-      "loss": 4.3668,
-      "step": 1683
+      "epoch": 0.93,
+      "learning_rate": 1.6326530612244897e-05,
+      "loss": 3.9554,
+      "step": 831
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.099688473520249e-06,
-      "loss": 3.9009,
-      "step": 1684
+      "epoch": 0.93,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 4.0022,
+      "step": 832
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.021806853582555e-06,
-      "loss": 4.2569,
-      "step": 1685
+      "epoch": 0.93,
+      "learning_rate": 1.5816326530612247e-05,
+      "loss": 4.1985,
+      "step": 833
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.94392523364486e-06,
-      "loss": 4.1395,
-      "step": 1686
+      "epoch": 0.93,
+      "learning_rate": 1.556122448979592e-05,
+      "loss": 4.0862,
+      "step": 834
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.866043613707164e-06,
-      "loss": 4.0136,
-      "step": 1687
+      "epoch": 0.94,
+      "learning_rate": 1.5306122448979594e-05,
+      "loss": 4.065,
+      "step": 835
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.78816199376947e-06,
-      "loss": 4.331,
-      "step": 1688
+      "epoch": 0.94,
+      "learning_rate": 1.5051020408163266e-05,
+      "loss": 4.0925,
+      "step": 836
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.710280373831776e-06,
-      "loss": 4.1365,
-      "step": 1689
+      "epoch": 0.94,
+      "learning_rate": 1.479591836734694e-05,
+      "loss": 4.0593,
+      "step": 837
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.632398753894081e-06,
-      "loss": 3.9473,
-      "step": 1690
+      "epoch": 0.94,
+      "learning_rate": 1.4540816326530612e-05,
+      "loss": 4.3442,
+      "step": 838
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.554517133956386e-06,
-      "loss": 4.3261,
-      "step": 1691
+      "epoch": 0.94,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 3.9675,
+      "step": 839
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.476635514018691e-06,
-      "loss": 4.2032,
-      "step": 1692
+      "epoch": 0.94,
+      "learning_rate": 1.4030612244897959e-05,
+      "loss": 4.1694,
+      "step": 840
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.398753894080997e-06,
-      "loss": 4.3535,
-      "step": 1693
+      "epoch": 0.94,
+      "learning_rate": 1.3775510204081633e-05,
+      "loss": 4.2668,
+      "step": 841
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 7.3208722741433015e-06,
-      "loss": 4.4135,
-      "step": 1694
+      "epoch": 0.94,
+      "learning_rate": 1.3520408163265308e-05,
+      "loss": 4.1724,
+      "step": 842
     },
     {
       "epoch": 0.95,
-      "learning_rate": 7.242990654205607e-06,
-      "loss": 4.5697,
-      "step": 1695
+      "learning_rate": 1.3265306122448982e-05,
+      "loss": 4.2712,
+      "step": 843
     },
     {
       "epoch": 0.95,
-      "learning_rate": 7.165109034267913e-06,
-      "loss": 5.4692,
-      "step": 1696
+      "learning_rate": 1.3010204081632653e-05,
+      "loss": 4.2255,
+      "step": 844
     },
     {
       "epoch": 0.95,
-      "learning_rate": 7.087227414330218e-06,
-      "loss": 4.18,
-      "step": 1697
+      "learning_rate": 1.2755102040816327e-05,
+      "loss": 4.3171,
+      "step": 845
     },
     {
       "epoch": 0.95,
-      "learning_rate": 7.009345794392523e-06,
-      "loss": 4.3532,
-      "step": 1698
+      "learning_rate": 1.25e-05,
+      "loss": 4.6056,
+      "step": 846
     },
     {
       "epoch": 0.95,
-      "learning_rate": 6.931464174454828e-06,
-      "loss": 4.4678,
-      "step": 1699
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 4.0734,
+      "step": 847
     },
     {
       "epoch": 0.95,
-      "learning_rate": 6.853582554517134e-06,
-      "loss": 4.264,
-      "step": 1700
+      "learning_rate": 1.1989795918367348e-05,
+      "loss": 4.725,
+      "step": 848
     },
     {
       "epoch": 0.95,
-      "learning_rate": 6.775700934579439e-06,
-      "loss": 4.7274,
-      "step": 1701
+      "learning_rate": 1.1734693877551021e-05,
+      "loss": 4.3384,
+      "step": 849
     },
     {
       "epoch": 0.95,
-      "learning_rate": 6.697819314641744e-06,
-      "loss": 4.1766,
-      "step": 1702
+      "learning_rate": 1.1479591836734695e-05,
+      "loss": 4.1118,
+      "step": 850
     },
     {
       "epoch": 0.95,
-      "learning_rate": 6.619937694704049e-06,
-      "loss": 4.0105,
-      "step": 1703
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.542056074766355e-06,
-      "loss": 4.1038,
-      "step": 1704
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.46417445482866e-06,
-      "loss": 4.1229,
-      "step": 1705
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.386292834890965e-06,
-      "loss": 3.9882,
-      "step": 1706
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.308411214953271e-06,
-      "loss": 3.9628,
-      "step": 1707
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.230529595015577e-06,
-      "loss": 4.2692,
-      "step": 1708
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.1526479750778815e-06,
-      "loss": 4.3997,
-      "step": 1709
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 6.074766355140187e-06,
-      "loss": 4.3127,
-      "step": 1710
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.996884735202493e-06,
-      "loss": 3.9961,
-      "step": 1711
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.919003115264798e-06,
-      "loss": 4.2177,
-      "step": 1712
+      "learning_rate": 1.1224489795918369e-05,
+      "loss": 4.1718,
+      "step": 851
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.841121495327103e-06,
-      "loss": 4.5403,
-      "step": 1713
+      "learning_rate": 1.096938775510204e-05,
+      "loss": 4.1493,
+      "step": 852
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.763239875389408e-06,
-      "loss": 4.0421,
-      "step": 1714
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 4.0594,
+      "step": 853
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.685358255451714e-06,
-      "loss": 4.1472,
-      "step": 1715
+      "learning_rate": 1.045918367346939e-05,
+      "loss": 4.0984,
+      "step": 854
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.607476635514019e-06,
-      "loss": 4.1848,
-      "step": 1716
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 4.0627,
+      "step": 855
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.529595015576324e-06,
-      "loss": 4.4559,
-      "step": 1717
+      "learning_rate": 9.948979591836735e-06,
+      "loss": 4.0885,
+      "step": 856
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.451713395638629e-06,
-      "loss": 4.5943,
-      "step": 1718
+      "learning_rate": 9.693877551020408e-06,
+      "loss": 4.1545,
+      "step": 857
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.373831775700935e-06,
-      "loss": 3.9997,
-      "step": 1719
+      "learning_rate": 9.438775510204082e-06,
+      "loss": 4.1281,
+      "step": 858
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.29595015576324e-06,
-      "loss": 3.9963,
-      "step": 1720
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 3.9807,
+      "step": 859
     },
     {
       "epoch": 0.96,
-      "learning_rate": 5.218068535825545e-06,
-      "loss": 4.2278,
-      "step": 1721
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.14018691588785e-06,
-      "loss": 4.3779,
-      "step": 1722
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.062305295950156e-06,
-      "loss": 4.2929,
-      "step": 1723
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.9844236760124615e-06,
-      "loss": 4.1245,
-      "step": 1724
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.906542056074766e-06,
-      "loss": 4.0958,
-      "step": 1725
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.828660436137072e-06,
-      "loss": 4.0344,
-      "step": 1726
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.750778816199377e-06,
-      "loss": 4.3233,
-      "step": 1727
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.6728971962616825e-06,
-      "loss": 4.2354,
-      "step": 1728
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.595015576323987e-06,
-      "loss": 4.4561,
-      "step": 1729
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.517133956386293e-06,
-      "loss": 4.192,
-      "step": 1730
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 4.1072,
+      "step": 860
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.439252336448598e-06,
-      "loss": 3.8557,
-      "step": 1731
+      "learning_rate": 8.673469387755103e-06,
+      "loss": 4.2454,
+      "step": 861
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.3613707165109035e-06,
-      "loss": 4.0138,
-      "step": 1732
+      "learning_rate": 8.418367346938775e-06,
+      "loss": 4.0128,
+      "step": 862
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.283489096573208e-06,
-      "loss": 4.044,
-      "step": 1733
+      "learning_rate": 8.163265306122448e-06,
+      "loss": 4.1705,
+      "step": 863
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.205607476635514e-06,
-      "loss": 4.2168,
-      "step": 1734
+      "learning_rate": 7.908163265306124e-06,
+      "loss": 4.3512,
+      "step": 864
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.127725856697819e-06,
-      "loss": 4.4955,
-      "step": 1735
+      "learning_rate": 7.653061224489797e-06,
+      "loss": 4.264,
+      "step": 865
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.0498442367601245e-06,
-      "loss": 4.5085,
-      "step": 1736
+      "learning_rate": 7.39795918367347e-06,
+      "loss": 3.9359,
+      "step": 866
     },
     {
       "epoch": 0.97,
-      "learning_rate": 3.97196261682243e-06,
-      "loss": 3.8179,
-      "step": 1737
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 4.1472,
+      "step": 867
     },
     {
       "epoch": 0.97,
-      "learning_rate": 3.894080996884735e-06,
-      "loss": 4.181,
-      "step": 1738
+      "learning_rate": 6.887755102040816e-06,
+      "loss": 3.9617,
+      "step": 868
     },
     {
       "epoch": 0.97,
-      "learning_rate": 3.816199376947041e-06,
-      "loss": 4.0023,
-      "step": 1739
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.7383177570093455e-06,
-      "loss": 4.0269,
-      "step": 1740
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.6604361370716508e-06,
-      "loss": 4.1436,
-      "step": 1741
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.5825545171339564e-06,
-      "loss": 4.227,
-      "step": 1742
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.5046728971962617e-06,
-      "loss": 4.0525,
-      "step": 1743
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.426791277258567e-06,
-      "loss": 4.0061,
-      "step": 1744
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.348909657320872e-06,
-      "loss": 4.0849,
-      "step": 1745
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.2710280373831774e-06,
-      "loss": 4.2392,
-      "step": 1746
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.1931464174454827e-06,
-      "loss": 4.2125,
-      "step": 1747
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 3.1152647975077884e-06,
-      "loss": 4.0288,
-      "step": 1748
+      "learning_rate": 6.632653061224491e-06,
+      "loss": 4.1297,
+      "step": 869
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.0373831775700936e-06,
-      "loss": 4.0494,
-      "step": 1749
+      "learning_rate": 6.3775510204081635e-06,
+      "loss": 4.2363,
+      "step": 870
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.959501557632399e-06,
-      "loss": 4.0498,
-      "step": 1750
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 3.966,
+      "step": 871
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.881619937694704e-06,
-      "loss": 3.9222,
-      "step": 1751
+      "learning_rate": 5.867346938775511e-06,
+      "loss": 4.1766,
+      "step": 872
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.8037383177570094e-06,
-      "loss": 4.0954,
-      "step": 1752
+      "learning_rate": 5.612244897959184e-06,
+      "loss": 4.1703,
+      "step": 873
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.7258566978193146e-06,
-      "loss": 4.2722,
-      "step": 1753
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 4.138,
+      "step": 874
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.64797507788162e-06,
-      "loss": 4.3409,
-      "step": 1754
+      "learning_rate": 5.102040816326531e-06,
+      "loss": 4.2855,
+      "step": 875
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.570093457943925e-06,
-      "loss": 4.3169,
-      "step": 1755
+      "learning_rate": 4.846938775510204e-06,
+      "loss": 4.1395,
+      "step": 876
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.4922118380062308e-06,
-      "loss": 3.9821,
-      "step": 1756
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 4.0096,
+      "step": 877
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.414330218068536e-06,
-      "loss": 4.0028,
-      "step": 1757
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.3364485981308413e-06,
-      "loss": 4.0223,
-      "step": 1758
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.2585669781931465e-06,
-      "loss": 4.3552,
-      "step": 1759
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.1806853582554518e-06,
-      "loss": 4.1026,
-      "step": 1760
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.102803738317757e-06,
-      "loss": 4.3499,
-      "step": 1761
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.0249221183800623e-06,
-      "loss": 3.8653,
-      "step": 1762
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.9470404984423675e-06,
-      "loss": 4.0525,
-      "step": 1763
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.8691588785046728e-06,
-      "loss": 4.2736,
-      "step": 1764
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.7912772585669782e-06,
-      "loss": 3.9191,
-      "step": 1765
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.7133956386292835e-06,
-      "loss": 4.0542,
-      "step": 1766
+      "learning_rate": 4.336734693877551e-06,
+      "loss": 4.0787,
+      "step": 878
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.6355140186915887e-06,
-      "loss": 3.9638,
-      "step": 1767
+      "learning_rate": 4.081632653061224e-06,
+      "loss": 4.1907,
+      "step": 879
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.5576323987538942e-06,
-      "loss": 4.0942,
-      "step": 1768
+      "learning_rate": 3.826530612244899e-06,
+      "loss": 4.348,
+      "step": 880
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.4797507788161994e-06,
-      "loss": 4.0524,
-      "step": 1769
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 3.9686,
+      "step": 881
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.4018691588785047e-06,
-      "loss": 3.9628,
-      "step": 1770
+      "learning_rate": 3.3163265306122454e-06,
+      "loss": 4.0974,
+      "step": 882
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.32398753894081e-06,
-      "loss": 4.2197,
-      "step": 1771
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 4.0149,
+      "step": 883
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.2461059190031154e-06,
-      "loss": 4.0186,
-      "step": 1772
+      "learning_rate": 2.806122448979592e-06,
+      "loss": 4.0868,
+      "step": 884
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.1682242990654206e-06,
-      "loss": 4.0753,
-      "step": 1773
+      "learning_rate": 2.5510204081632653e-06,
+      "loss": 4.0935,
+      "step": 885
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.0903426791277259e-06,
-      "loss": 4.2658,
-      "step": 1774
+      "learning_rate": 2.295918367346939e-06,
+      "loss": 4.2536,
+      "step": 886
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.0124610591900311e-06,
-      "loss": 4.0253,
-      "step": 1775
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.345794392523364e-07,
-      "loss": 4.1123,
-      "step": 1776
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 8.566978193146417e-07,
-      "loss": 4.2103,
-      "step": 1777
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 7.788161993769471e-07,
-      "loss": 4.1655,
-      "step": 1778
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 7.009345794392523e-07,
-      "loss": 4.5377,
-      "step": 1779
+      "learning_rate": 2.040816326530612e-06,
+      "loss": 3.9868,
+      "step": 887
     },
     {
       "epoch": 1.0,
-      "learning_rate": 6.230529595015577e-07,
-      "loss": 3.9874,
-      "step": 1780
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 4.2264,
+      "step": 888
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.451713395638629e-07,
-      "loss": 4.3995,
-      "step": 1781
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 4.2771,
+      "step": 889
     },
     {
       "epoch": 1.0,
-      "learning_rate": 4.672897196261682e-07,
-      "loss": 3.9944,
-      "step": 1782
+      "learning_rate": 1.2755102040816327e-06,
+      "loss": 4.0625,
+      "step": 890
     },
     {
       "epoch": 1.0,
-      "learning_rate": 3.8940809968847354e-07,
-      "loss": 4.5299,
-      "step": 1783
+      "learning_rate": 1.020408163265306e-06,
+      "loss": 4.2802,
+      "step": 891
     },
     {
       "epoch": 1.0,
-      "learning_rate": 3.1152647975077885e-07,
-      "loss": 3.4028,
-      "step": 1784
+      "learning_rate": 7.653061224489796e-07,
+      "loss": 3.737,
+      "step": 892
     },
     {
       "epoch": 1.0,
-      "step": 1784,
+      "step": 892,
       "total_flos": 0.0,
-      "train_loss": 4.395840967850835,
-      "train_runtime": 8240.9748,
-      "train_samples_per_second": 3.463,
-      "train_steps_per_second": 0.216
+      "train_loss": 4.3097993645967385,
+      "train_runtime": 5935.828,
+      "train_samples_per_second": 4.808,
+      "train_steps_per_second": 0.15
     }
   ],
-  "max_steps": 1784,
+  "max_steps": 892,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,