diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2699 +1,7175 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2e-08,
-      "loss": 4.7633,
+      "learning_rate": 0.0,
+      "loss": 4.8701,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4e-08,
-      "loss": 4.9111,
+      "learning_rate": 0.0,
+      "loss": 5.0707,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4e-08,
-      "loss": 4.8796,
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 5.0731,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 4.8776,
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 4.8923,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8e-08,
-      "loss": 4.8182,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.6258,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 4.8851,
+      "learning_rate": 8e-08,
+      "loss": 4.9952,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.8575,
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.7569,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.4e-07,
-      "loss": 4.9088,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.7364,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.6e-07,
-      "loss": 4.9109,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.9657,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.8e-07,
-      "loss": 4.7878,
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 4.8063,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.9099,
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 4.8652,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.2e-07,
-      "loss": 4.8879,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 4.7189,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8415,
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 5.0142,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.6e-07,
-      "loss": 4.8503,
+      "epoch": 0.01,
+      "learning_rate": 2.2e-07,
+      "loss": 4.8025,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 2.8e-07,
-      "loss": 4.7343,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.697,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.8936,
+      "epoch": 0.01,
+      "learning_rate": 2.6e-07,
+      "loss": 4.941,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.2e-07,
-      "loss": 4.8205,
+      "epoch": 0.01,
+      "learning_rate": 2.8e-07,
+      "loss": 4.8771,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 4.9299,
+      "epoch": 0.02,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 5.0554,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.6e-07,
-      "loss": 4.7331,
+      "epoch": 0.02,
+      "learning_rate": 3.2e-07,
+      "loss": 4.7867,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 3.8e-07,
-      "loss": 4.9161,
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 5.0221,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.91,
+      "epoch": 0.02,
+      "learning_rate": 3.6e-07,
+      "loss": 4.991,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 4.8262,
+      "epoch": 0.02,
+      "learning_rate": 3.8e-07,
+      "loss": 4.7568,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.4e-07,
-      "loss": 4.9479,
+      "epoch": 0.02,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9541,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 4.942,
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 5.0154,
       "step": 24
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 4.7356,
+      "epoch": 0.02,
+      "learning_rate": 4.4e-07,
+      "loss": 4.8169,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 5.0038,
+      "epoch": 0.02,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 4.9814,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.2e-07,
-      "loss": 4.8565,
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.8629,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 5.4e-07,
-      "loss": 4.7324,
+      "epoch": 0.02,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.6962,
       "step": 28
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.6e-07,
-      "loss": 4.7126,
+      "epoch": 0.02,
+      "learning_rate": 5.2e-07,
+      "loss": 4.7963,
       "step": 29
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 5.800000000000001e-07,
-      "loss": 4.9015,
+      "epoch": 0.03,
+      "learning_rate": 5.4e-07,
+      "loss": 4.8153,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.869,
+      "epoch": 0.03,
+      "learning_rate": 5.6e-07,
+      "loss": 4.9438,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.200000000000001e-07,
-      "loss": 4.7292,
+      "epoch": 0.03,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 4.7618,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 6.4e-07,
-      "loss": 4.8362,
+      "epoch": 0.03,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.9067,
       "step": 33
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.6e-07,
-      "loss": 4.75,
+      "epoch": 0.03,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 4.57,
       "step": 34
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 6.800000000000001e-07,
-      "loss": 4.8524,
+      "epoch": 0.03,
+      "learning_rate": 6.4e-07,
+      "loss": 4.746,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 4.7962,
+      "epoch": 0.03,
+      "learning_rate": 6.6e-07,
+      "loss": 5.094,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 7.2e-07,
-      "loss": 4.859,
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 4.8869,
       "step": 37
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.4e-07,
-      "loss": 4.7945,
+      "epoch": 0.03,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.7572,
       "step": 38
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.6e-07,
-      "loss": 4.9263,
+      "epoch": 0.03,
+      "learning_rate": 7.2e-07,
+      "loss": 5.2519,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 7.8e-07,
-      "loss": 4.7248,
+      "epoch": 0.03,
+      "learning_rate": 7.4e-07,
+      "loss": 4.849,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.803,
+      "epoch": 0.03,
+      "learning_rate": 7.6e-07,
+      "loss": 4.7037,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 4.9282,
+      "epoch": 0.04,
+      "learning_rate": 7.8e-07,
+      "loss": 4.7759,
       "step": 42
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 4.7007,
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.7871,
       "step": 43
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.6e-07,
-      "loss": 4.8718,
+      "epoch": 0.04,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 4.8614,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.8e-07,
-      "loss": 5.0068,
+      "epoch": 0.04,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 5.0933,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 5.1512,
+      "epoch": 0.04,
+      "learning_rate": 8.6e-07,
+      "loss": 5.087,
       "step": 46
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.0423,
+      "epoch": 0.04,
+      "learning_rate": 8.8e-07,
+      "loss": 5.3174,
       "step": 47
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 5.3227,
+      "epoch": 0.04,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 5.5401,
       "step": 48
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 5.4297,
+      "epoch": 0.04,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.6871,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 5.339,
+      "epoch": 0.04,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.6735,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.800000000000001e-07,
-      "loss": 4.657,
+      "epoch": 0.04,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 4.616,
       "step": 51
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.5155,
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.6751,
       "step": 52
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.02e-06,
-      "loss": 4.6886,
+      "epoch": 0.04,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 4.4852,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.04e-06,
-      "loss": 4.5949,
+      "epoch": 0.05,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.5643,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.06e-06,
-      "loss": 4.4149,
+      "epoch": 0.05,
+      "learning_rate": 1.02e-06,
+      "loss": 4.7176,
       "step": 55
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.08e-06,
-      "loss": 4.6266,
+      "epoch": 0.05,
+      "learning_rate": 1.04e-06,
+      "loss": 4.4899,
       "step": 56
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.1e-06,
-      "loss": 4.6458,
+      "epoch": 0.05,
+      "learning_rate": 1.06e-06,
+      "loss": 4.737,
       "step": 57
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.12e-06,
-      "loss": 4.7385,
+      "epoch": 0.05,
+      "learning_rate": 1.08e-06,
+      "loss": 4.6653,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.14e-06,
-      "loss": 4.6687,
+      "epoch": 0.05,
+      "learning_rate": 1.1e-06,
+      "loss": 4.5618,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.1600000000000001e-06,
-      "loss": 4.5012,
+      "epoch": 0.05,
+      "learning_rate": 1.12e-06,
+      "loss": 4.7332,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.1800000000000001e-06,
-      "loss": 4.6079,
+      "epoch": 0.05,
+      "learning_rate": 1.14e-06,
+      "loss": 4.4809,
       "step": 61
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.5349,
+      "epoch": 0.05,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 4.6674,
       "step": 62
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 4.5074,
+      "epoch": 0.05,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 4.5984,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 4.611,
+      "epoch": 0.05,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.4619,
       "step": 64
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.26e-06,
-      "loss": 4.565,
+      "epoch": 0.05,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 4.4137,
       "step": 65
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.28e-06,
-      "loss": 4.4438,
+      "epoch": 0.06,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 4.568,
       "step": 66
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.3e-06,
-      "loss": 4.6036,
+      "epoch": 0.06,
+      "learning_rate": 1.26e-06,
+      "loss": 4.7204,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.32e-06,
-      "loss": 4.4427,
+      "epoch": 0.06,
+      "learning_rate": 1.28e-06,
+      "loss": 4.5221,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.34e-06,
-      "loss": 4.5095,
+      "epoch": 0.06,
+      "learning_rate": 1.3e-06,
+      "loss": 4.497,
       "step": 69
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.3600000000000001e-06,
-      "loss": 4.4858,
+      "epoch": 0.06,
+      "learning_rate": 1.32e-06,
+      "loss": 4.5477,
       "step": 70
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 4.618,
+      "epoch": 0.06,
+      "learning_rate": 1.34e-06,
+      "loss": 4.7146,
       "step": 71
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.6007,
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 4.5997,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.42e-06,
-      "loss": 4.453,
+      "epoch": 0.06,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 4.5564,
       "step": 73
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.44e-06,
-      "loss": 4.519,
+      "epoch": 0.06,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.6454,
       "step": 74
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.46e-06,
-      "loss": 4.5242,
+      "epoch": 0.06,
+      "learning_rate": 1.42e-06,
+      "loss": 4.4289,
       "step": 75
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.48e-06,
-      "loss": 4.4702,
+      "epoch": 0.06,
+      "learning_rate": 1.44e-06,
+      "loss": 4.9177,
       "step": 76
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.5e-06,
-      "loss": 4.5255,
+      "epoch": 0.06,
+      "learning_rate": 1.46e-06,
+      "loss": 4.5739,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.52e-06,
-      "loss": 4.4652,
+      "epoch": 0.07,
+      "learning_rate": 1.48e-06,
+      "loss": 4.4393,
       "step": 78
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.54e-06,
-      "loss": 4.4498,
+      "epoch": 0.07,
+      "learning_rate": 1.5e-06,
+      "loss": 4.4889,
       "step": 79
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.56e-06,
-      "loss": 4.5705,
+      "epoch": 0.07,
+      "learning_rate": 1.52e-06,
+      "loss": 4.585,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.5800000000000001e-06,
-      "loss": 4.5202,
+      "epoch": 0.07,
+      "learning_rate": 1.54e-06,
+      "loss": 4.6074,
       "step": 81
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.4593,
+      "epoch": 0.07,
+      "learning_rate": 1.56e-06,
+      "loss": 4.5827,
       "step": 82
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6200000000000002e-06,
-      "loss": 4.5733,
+      "epoch": 0.07,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 4.4494,
       "step": 83
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 4.45,
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.5706,
       "step": 84
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 4.4868,
+      "epoch": 0.07,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 4.4794,
       "step": 85
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.6800000000000002e-06,
-      "loss": 4.48,
+      "epoch": 0.07,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 4.4845,
       "step": 86
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 4.5859,
+      "epoch": 0.07,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 4.4077,
       "step": 87
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.72e-06,
-      "loss": 4.4624,
+      "epoch": 0.07,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 4.5921,
       "step": 88
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.74e-06,
-      "loss": 4.6002,
+      "epoch": 0.07,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.625,
       "step": 89
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.76e-06,
-      "loss": 4.5497,
+      "epoch": 0.08,
+      "learning_rate": 1.72e-06,
+      "loss": 4.5165,
       "step": 90
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 4.4896,
+      "epoch": 0.08,
+      "learning_rate": 1.74e-06,
+      "loss": 4.6137,
       "step": 91
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8000000000000001e-06,
-      "loss": 4.7272,
+      "epoch": 0.08,
+      "learning_rate": 1.76e-06,
+      "loss": 4.663,
       "step": 92
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 4.5785,
+      "epoch": 0.08,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 4.2875,
       "step": 93
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 4.5895,
+      "epoch": 0.08,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.6684,
       "step": 94
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8600000000000002e-06,
-      "loss": 4.5834,
+      "epoch": 0.08,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 4.8948,
       "step": 95
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 4.7481,
+      "epoch": 0.08,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 4.7073,
       "step": 96
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9000000000000002e-06,
-      "loss": 4.7947,
+      "epoch": 0.08,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 4.6509,
       "step": 97
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.567,
+      "epoch": 0.08,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 4.8975,
       "step": 98
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.94e-06,
-      "loss": 4.8742,
+      "epoch": 0.08,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.9052,
       "step": 99
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 4.6728,
+      "epoch": 0.08,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.7903,
       "step": 100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.98e-06,
-      "loss": 4.4148,
+      "epoch": 0.08,
+      "learning_rate": 1.94e-06,
+      "loss": 4.1743,
       "step": 101
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.3493,
+      "epoch": 0.09,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 4.4867,
       "step": 102
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.02e-06,
-      "loss": 4.2165,
+      "epoch": 0.09,
+      "learning_rate": 1.98e-06,
+      "loss": 4.3706,
       "step": 103
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.04e-06,
-      "loss": 4.285,
+      "epoch": 0.09,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.4715,
       "step": 104
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.06e-06,
-      "loss": 4.2356,
+      "epoch": 0.09,
+      "learning_rate": 2.02e-06,
+      "loss": 4.3478,
       "step": 105
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.08e-06,
-      "loss": 4.278,
+      "epoch": 0.09,
+      "learning_rate": 2.04e-06,
+      "loss": 4.4912,
       "step": 106
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.38,
+      "epoch": 0.09,
+      "learning_rate": 2.06e-06,
+      "loss": 4.6269,
       "step": 107
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.12e-06,
-      "loss": 4.2958,
+      "epoch": 0.09,
+      "learning_rate": 2.08e-06,
+      "loss": 4.5301,
       "step": 108
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.1400000000000003e-06,
-      "loss": 4.2908,
+      "epoch": 0.09,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.4822,
       "step": 109
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.16e-06,
-      "loss": 4.3311,
+      "epoch": 0.09,
+      "learning_rate": 2.12e-06,
+      "loss": 4.1702,
       "step": 110
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 4.3749,
+      "epoch": 0.09,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 4.4782,
       "step": 111
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.2e-06,
-      "loss": 4.3236,
+      "epoch": 0.09,
+      "learning_rate": 2.16e-06,
+      "loss": 4.5534,
       "step": 112
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 4.348,
+      "epoch": 0.09,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 4.3559,
       "step": 113
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.24e-06,
-      "loss": 4.3211,
+      "epoch": 0.1,
+      "learning_rate": 2.2e-06,
+      "loss": 4.5086,
       "step": 114
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.2600000000000004e-06,
-      "loss": 4.3697,
+      "epoch": 0.1,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 4.4195,
       "step": 115
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.28e-06,
-      "loss": 4.4212,
+      "epoch": 0.1,
+      "learning_rate": 2.24e-06,
+      "loss": 4.6085,
       "step": 116
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.3000000000000004e-06,
-      "loss": 4.3025,
+      "epoch": 0.1,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 4.3308,
       "step": 117
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 4.4696,
+      "epoch": 0.1,
+      "learning_rate": 2.28e-06,
+      "loss": 4.1663,
       "step": 118
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 4.3551,
+      "epoch": 0.1,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 4.271,
       "step": 119
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.3600000000000003e-06,
-      "loss": 4.3851,
+      "epoch": 0.1,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 4.3932,
       "step": 120
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.38e-06,
-      "loss": 4.2232,
+      "epoch": 0.1,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 4.6553,
       "step": 121
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.3045,
+      "epoch": 0.1,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 4.3989,
       "step": 122
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.42e-06,
-      "loss": 4.33,
+      "epoch": 0.1,
+      "learning_rate": 2.38e-06,
+      "loss": 4.4578,
       "step": 123
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 4.3141,
+      "epoch": 0.1,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.4331,
       "step": 124
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.46e-06,
-      "loss": 4.4295,
+      "epoch": 0.11,
+      "learning_rate": 2.42e-06,
+      "loss": 4.2289,
       "step": 125
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.4800000000000004e-06,
-      "loss": 4.44,
+      "epoch": 0.11,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 4.3991,
       "step": 126
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3421,
+      "epoch": 0.11,
+      "learning_rate": 2.46e-06,
+      "loss": 4.634,
       "step": 127
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.52e-06,
-      "loss": 4.3164,
+      "epoch": 0.11,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 4.2652,
       "step": 128
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.5400000000000002e-06,
-      "loss": 4.2223,
+      "epoch": 0.11,
+      "learning_rate": 2.5e-06,
+      "loss": 4.1415,
       "step": 129
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.56e-06,
-      "loss": 4.3852,
+      "epoch": 0.11,
+      "learning_rate": 2.52e-06,
+      "loss": 4.438,
       "step": 130
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 4.2854,
+      "epoch": 0.11,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 4.3727,
       "step": 131
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6e-06,
-      "loss": 4.3269,
+      "epoch": 0.11,
+      "learning_rate": 2.56e-06,
+      "loss": 4.3326,
       "step": 132
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 4.2724,
+      "epoch": 0.11,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 4.5097,
       "step": 133
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.64e-06,
-      "loss": 4.2822,
+      "epoch": 0.11,
+      "learning_rate": 2.6e-06,
+      "loss": 4.4034,
       "step": 134
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.6600000000000004e-06,
-      "loss": 4.4477,
+      "epoch": 0.11,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 4.5444,
       "step": 135
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 2.68e-06,
-      "loss": 4.3023,
+      "epoch": 0.11,
+      "learning_rate": 2.64e-06,
+      "loss": 4.2092,
       "step": 136
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 4.4109,
+      "epoch": 0.12,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 4.4904,
       "step": 137
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7200000000000002e-06,
-      "loss": 4.4397,
+      "epoch": 0.12,
+      "learning_rate": 2.68e-06,
+      "loss": 4.4909,
       "step": 138
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 4.4502,
+      "epoch": 0.12,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 4.2405,
       "step": 139
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 4.4306,
+      "epoch": 0.12,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 4.4832,
       "step": 140
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.7800000000000005e-06,
-      "loss": 4.5436,
+      "epoch": 0.12,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 4.5207,
       "step": 141
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.4715,
+      "epoch": 0.12,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 4.3791,
       "step": 142
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.82e-06,
-      "loss": 4.3932,
+      "epoch": 0.12,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 4.2597,
       "step": 143
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 2.84e-06,
-      "loss": 4.4735,
+      "epoch": 0.12,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.6319,
       "step": 144
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.86e-06,
-      "loss": 4.3691,
+      "epoch": 0.12,
+      "learning_rate": 2.82e-06,
+      "loss": 4.3822,
       "step": 145
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.88e-06,
-      "loss": 4.4599,
+      "epoch": 0.12,
+      "learning_rate": 2.84e-06,
+      "loss": 4.6909,
       "step": 146
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.9e-06,
-      "loss": 4.7192,
+      "epoch": 0.12,
+      "learning_rate": 2.86e-06,
+      "loss": 4.4473,
       "step": 147
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.92e-06,
-      "loss": 4.8634,
+      "epoch": 0.12,
+      "learning_rate": 2.88e-06,
+      "loss": 4.2824,
       "step": 148
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.5975,
+      "epoch": 0.13,
+      "learning_rate": 2.9e-06,
+      "loss": 4.5017,
       "step": 149
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.96e-06,
-      "loss": 4.7077,
+      "epoch": 0.13,
+      "learning_rate": 2.92e-06,
+      "loss": 4.3623,
       "step": 150
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 4.2145,
+      "epoch": 0.13,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.3672,
       "step": 151
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 3e-06,
-      "loss": 4.2845,
+      "epoch": 0.13,
+      "learning_rate": 2.96e-06,
+      "loss": 4.2128,
       "step": 152
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 4.2969,
+      "epoch": 0.13,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 4.3112,
       "step": 153
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.04e-06,
-      "loss": 4.2126,
+      "epoch": 0.13,
+      "learning_rate": 3e-06,
+      "loss": 4.4011,
       "step": 154
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.0600000000000003e-06,
-      "loss": 4.3056,
+      "epoch": 0.13,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 4.3079,
       "step": 155
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.08e-06,
-      "loss": 4.172,
+      "epoch": 0.13,
+      "learning_rate": 3.04e-06,
+      "loss": 4.1596,
       "step": 156
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.1000000000000004e-06,
-      "loss": 4.2029,
+      "epoch": 0.13,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 4.3133,
       "step": 157
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.12e-06,
-      "loss": 4.2939,
+      "epoch": 0.13,
+      "learning_rate": 3.08e-06,
+      "loss": 4.4966,
       "step": 158
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.1400000000000004e-06,
-      "loss": 4.3015,
+      "epoch": 0.13,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.2545,
       "step": 159
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.1600000000000002e-06,
-      "loss": 4.3041,
+      "epoch": 0.13,
+      "learning_rate": 3.12e-06,
+      "loss": 4.0574,
       "step": 160
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 4.3576,
+      "epoch": 0.14,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 4.3679,
       "step": 161
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.3136,
+      "epoch": 0.14,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 4.379,
       "step": 162
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2200000000000005e-06,
-      "loss": 4.2696,
+      "epoch": 0.14,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 4.2736,
       "step": 163
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 4.3336,
+      "epoch": 0.14,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.0694,
       "step": 164
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2600000000000006e-06,
-      "loss": 4.1924,
+      "epoch": 0.14,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 4.3049,
       "step": 165
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 4.2559,
+      "epoch": 0.14,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 4.0068,
       "step": 166
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.3000000000000006e-06,
-      "loss": 4.1719,
+      "epoch": 0.14,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 4.3278,
       "step": 167
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 4.2586,
+      "epoch": 0.14,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 4.174,
       "step": 168
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 4.3229,
+      "epoch": 0.14,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 4.4477,
       "step": 169
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3600000000000004e-06,
-      "loss": 4.3361,
+      "epoch": 0.14,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 4.2804,
       "step": 170
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 4.3238,
+      "epoch": 0.14,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 4.3643,
       "step": 171
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.3695,
+      "epoch": 0.14,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 4.1522,
       "step": 172
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 4.1802,
+      "epoch": 0.15,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 4.112,
       "step": 173
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.44e-06,
-      "loss": 4.3182,
+      "epoch": 0.15,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.4288,
       "step": 174
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.46e-06,
-      "loss": 4.3777,
+      "epoch": 0.15,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 4.3384,
       "step": 175
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.48e-06,
-      "loss": 4.2564,
+      "epoch": 0.15,
+      "learning_rate": 3.44e-06,
+      "loss": 4.3094,
       "step": 176
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.5e-06,
-      "loss": 4.2225,
+      "epoch": 0.15,
+      "learning_rate": 3.46e-06,
+      "loss": 4.4643,
       "step": 177
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.52e-06,
-      "loss": 4.3079,
+      "epoch": 0.15,
+      "learning_rate": 3.48e-06,
+      "loss": 4.2445,
       "step": 178
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.54e-06,
-      "loss": 4.3581,
+      "epoch": 0.15,
+      "learning_rate": 3.5e-06,
+      "loss": 4.2844,
       "step": 179
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 4.3614,
+      "epoch": 0.15,
+      "learning_rate": 3.52e-06,
+      "loss": 4.4112,
       "step": 180
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.58e-06,
-      "loss": 4.2853,
+      "epoch": 0.15,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3695,
       "step": 181
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 4.3422,
+      "epoch": 0.15,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 4.4264,
       "step": 182
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.62e-06,
-      "loss": 4.348,
+      "epoch": 0.15,
+      "learning_rate": 3.58e-06,
+      "loss": 4.2508,
       "step": 183
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 4.3491,
+      "epoch": 0.15,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 4.3513,
       "step": 184
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.66e-06,
-      "loss": 4.3238,
+      "epoch": 0.16,
+      "learning_rate": 3.62e-06,
+      "loss": 4.117,
       "step": 185
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.6800000000000003e-06,
-      "loss": 4.1929,
-      "step": 186
+      "epoch": 0.16,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 4.2583,
+      "step": 186
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.7e-06,
-      "loss": 4.21,
+      "epoch": 0.16,
+      "learning_rate": 3.66e-06,
+      "loss": 4.6049,
       "step": 187
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.7200000000000004e-06,
-      "loss": 4.164,
+      "epoch": 0.16,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 4.3709,
       "step": 188
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.74e-06,
-      "loss": 4.2805,
+      "epoch": 0.16,
+      "learning_rate": 3.7e-06,
+      "loss": 4.4913,
       "step": 189
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 4.405,
+      "epoch": 0.16,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 4.1205,
       "step": 190
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.3344,
+      "epoch": 0.16,
+      "learning_rate": 3.74e-06,
+      "loss": 4.3753,
       "step": 191
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.8000000000000005e-06,
-      "loss": 4.2669,
+      "epoch": 0.16,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 4.5688,
       "step": 192
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.820000000000001e-06,
-      "loss": 4.2899,
+      "epoch": 0.16,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.5949,
       "step": 193
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.3873,
+      "epoch": 0.16,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.3615,
       "step": 194
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.86e-06,
-      "loss": 4.4804,
+      "epoch": 0.16,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 4.5328,
       "step": 195
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.88e-06,
-      "loss": 4.4503,
+      "epoch": 0.16,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.3987,
       "step": 196
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.900000000000001e-06,
-      "loss": 4.3873,
+      "epoch": 0.17,
+      "learning_rate": 3.86e-06,
+      "loss": 4.563,
       "step": 197
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 4.6998,
+      "epoch": 0.17,
+      "learning_rate": 3.88e-06,
+      "loss": 4.6521,
       "step": 198
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.94e-06,
-      "loss": 4.6039,
+      "epoch": 0.17,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 4.5894,
       "step": 199
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.96e-06,
-      "loss": 4.5695,
+      "epoch": 0.17,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 4.3758,
       "step": 200
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 4.3181,
+      "epoch": 0.17,
+      "learning_rate": 3.94e-06,
+      "loss": 4.3884,
       "step": 201
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.2565,
+      "epoch": 0.17,
+      "learning_rate": 3.96e-06,
+      "loss": 4.212,
       "step": 202
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.2566,
+      "epoch": 0.17,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 4.2784,
       "step": 203
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.04e-06,
-      "loss": 4.2488,
+      "epoch": 0.17,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.2339,
       "step": 204
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.060000000000001e-06,
-      "loss": 4.3213,
+      "epoch": 0.17,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.1346,
       "step": 205
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.08e-06,
-      "loss": 4.3263,
+      "epoch": 0.17,
+      "learning_rate": 4.04e-06,
+      "loss": 4.2499,
       "step": 206
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.1e-06,
-      "loss": 4.1808,
+      "epoch": 0.17,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 4.3453,
       "step": 207
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.12e-06,
-      "loss": 4.2721,
+      "epoch": 0.17,
+      "learning_rate": 4.08e-06,
+      "loss": 4.529,
       "step": 208
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.14e-06,
-      "loss": 4.3141,
+      "epoch": 0.18,
+      "learning_rate": 4.1e-06,
+      "loss": 4.4129,
       "step": 209
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.16e-06,
-      "loss": 4.285,
+      "epoch": 0.18,
+      "learning_rate": 4.12e-06,
+      "loss": 4.4088,
       "step": 210
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.18e-06,
-      "loss": 4.1726,
+      "epoch": 0.18,
+      "learning_rate": 4.14e-06,
+      "loss": 4.2574,
       "step": 211
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.2216,
+      "epoch": 0.18,
+      "learning_rate": 4.16e-06,
+      "loss": 4.3576,
       "step": 212
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.22e-06,
-      "loss": 4.1285,
+      "epoch": 0.18,
+      "learning_rate": 4.18e-06,
+      "loss": 4.2746,
       "step": 213
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.24e-06,
-      "loss": 4.2122,
+      "epoch": 0.18,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.2906,
       "step": 214
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.26e-06,
-      "loss": 4.2578,
+      "epoch": 0.18,
+      "learning_rate": 4.22e-06,
+      "loss": 4.3553,
       "step": 215
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.2800000000000005e-06,
-      "loss": 4.2336,
+      "epoch": 0.18,
+      "learning_rate": 4.24e-06,
+      "loss": 4.324,
       "step": 216
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.3e-06,
-      "loss": 4.2851,
+      "epoch": 0.18,
+      "learning_rate": 4.26e-06,
+      "loss": 4.3934,
       "step": 217
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.32e-06,
-      "loss": 4.2353,
+      "epoch": 0.18,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 4.2407,
       "step": 218
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.34e-06,
-      "loss": 4.1564,
+      "epoch": 0.18,
+      "learning_rate": 4.3e-06,
+      "loss": 3.9572,
       "step": 219
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 4.2741,
+      "epoch": 0.18,
+      "learning_rate": 4.32e-06,
+      "loss": 4.291,
       "step": 220
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.38e-06,
-      "loss": 4.2666,
+      "epoch": 0.19,
+      "learning_rate": 4.34e-06,
+      "loss": 4.2514,
       "step": 221
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.4e-06,
-      "loss": 4.2329,
+      "epoch": 0.19,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 4.5583,
       "step": 222
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.42e-06,
-      "loss": 4.2691,
+      "epoch": 0.19,
+      "learning_rate": 4.38e-06,
+      "loss": 4.3251,
       "step": 223
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 4.1904,
+      "epoch": 0.19,
+      "learning_rate": 4.4e-06,
+      "loss": 4.394,
       "step": 224
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 4.3606,
+      "epoch": 0.19,
+      "learning_rate": 4.42e-06,
+      "loss": 4.1718,
       "step": 225
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.48e-06,
-      "loss": 4.2548,
+      "epoch": 0.19,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 4.1136,
       "step": 226
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.5e-06,
-      "loss": 4.3776,
+      "epoch": 0.19,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 4.2544,
       "step": 227
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.520000000000001e-06,
-      "loss": 4.3096,
+      "epoch": 0.19,
+      "learning_rate": 4.48e-06,
+      "loss": 4.3033,
       "step": 228
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 4.1924,
+      "epoch": 0.19,
+      "learning_rate": 4.5e-06,
+      "loss": 4.1742,
       "step": 229
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.56e-06,
-      "loss": 4.2802,
+      "epoch": 0.19,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 4.3733,
       "step": 230
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.58e-06,
-      "loss": 4.3578,
+      "epoch": 0.19,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 4.3445,
       "step": 231
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 4.1489,
+      "epoch": 0.2,
+      "learning_rate": 4.56e-06,
+      "loss": 4.311,
       "step": 232
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.620000000000001e-06,
-      "loss": 4.1684,
+      "epoch": 0.2,
+      "learning_rate": 4.58e-06,
+      "loss": 4.2491,
       "step": 233
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.6400000000000005e-06,
-      "loss": 4.2505,
+      "epoch": 0.2,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.2232,
       "step": 234
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.66e-06,
-      "loss": 4.3324,
+      "epoch": 0.2,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 4.2887,
       "step": 235
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.680000000000001e-06,
-      "loss": 4.2627,
+      "epoch": 0.2,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 4.4628,
       "step": 236
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.7e-06,
-      "loss": 4.304,
+      "epoch": 0.2,
+      "learning_rate": 4.66e-06,
+      "loss": 4.183,
       "step": 237
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.7200000000000005e-06,
-      "loss": 4.2457,
+      "epoch": 0.2,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 4.3189,
       "step": 238
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.74e-06,
-      "loss": 4.1868,
+      "epoch": 0.2,
+      "learning_rate": 4.7e-06,
+      "loss": 4.2823,
       "step": 239
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.76e-06,
-      "loss": 4.3028,
+      "epoch": 0.2,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 4.3368,
       "step": 240
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.78e-06,
-      "loss": 4.2704,
+      "epoch": 0.2,
+      "learning_rate": 4.74e-06,
+      "loss": 4.4337,
       "step": 241
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3031,
+      "epoch": 0.2,
+      "learning_rate": 4.76e-06,
+      "loss": 4.2657,
       "step": 242
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 4.5193,
+      "epoch": 0.2,
+      "learning_rate": 4.78e-06,
+      "loss": 4.4958,
       "step": 243
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.84e-06,
-      "loss": 4.3524,
+      "epoch": 0.21,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.3085,
       "step": 244
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.86e-06,
-      "loss": 4.3437,
+      "epoch": 0.21,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 4.3229,
       "step": 245
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 4.2623,
+      "epoch": 0.21,
+      "learning_rate": 4.84e-06,
+      "loss": 4.2645,
       "step": 246
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 4.1807,
+      "epoch": 0.21,
+      "learning_rate": 4.86e-06,
+      "loss": 4.821,
       "step": 247
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.92e-06,
-      "loss": 4.5587,
+      "epoch": 0.21,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 4.4648,
       "step": 248
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.94e-06,
-      "loss": 4.6662,
+      "epoch": 0.21,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.3857,
       "step": 249
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 4.6559,
+      "epoch": 0.21,
+      "learning_rate": 4.92e-06,
+      "loss": 4.4901,
       "step": 250
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.2338,
+      "epoch": 0.21,
+      "learning_rate": 4.94e-06,
+      "loss": 4.3904,
       "step": 251
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5e-06,
-      "loss": 4.2654,
+      "epoch": 0.21,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 4.0535,
       "step": 252
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.02e-06,
-      "loss": 4.311,
+      "epoch": 0.21,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.3428,
       "step": 253
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.04e-06,
-      "loss": 4.1525,
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 4.0766,
       "step": 254
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.060000000000001e-06,
-      "loss": 4.2531,
+      "epoch": 0.21,
+      "learning_rate": 5.02e-06,
+      "loss": 4.082,
       "step": 255
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 5.0800000000000005e-06,
-      "loss": 4.223,
+      "epoch": 0.22,
+      "learning_rate": 5.04e-06,
+      "loss": 4.2663,
       "step": 256
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.1e-06,
-      "loss": 4.1962,
+      "epoch": 0.22,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 4.0657,
       "step": 257
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.12e-06,
-      "loss": 4.2138,
+      "epoch": 0.22,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 4.5064,
       "step": 258
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 4.2155,
+      "epoch": 0.22,
+      "learning_rate": 5.1e-06,
+      "loss": 4.3159,
       "step": 259
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 4.1582,
+      "epoch": 0.22,
+      "learning_rate": 5.12e-06,
+      "loss": 4.2698,
       "step": 260
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.18e-06,
-      "loss": 4.2736,
+      "epoch": 0.22,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 4.2277,
       "step": 261
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.2e-06,
-      "loss": 4.2103,
+      "epoch": 0.22,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 4.2356,
       "step": 262
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 4.3473,
+      "epoch": 0.22,
+      "learning_rate": 5.18e-06,
+      "loss": 4.3581,
       "step": 263
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 4.2133,
+      "epoch": 0.22,
+      "learning_rate": 5.2e-06,
+      "loss": 4.3402,
       "step": 264
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 4.2476,
+      "epoch": 0.22,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 4.2444,
       "step": 265
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.28e-06,
-      "loss": 4.2244,
+      "epoch": 0.22,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 4.3126,
       "step": 266
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.300000000000001e-06,
-      "loss": 4.2517,
+      "epoch": 0.22,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 4.1779,
       "step": 267
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 4.2084,
+      "epoch": 0.23,
+      "learning_rate": 5.28e-06,
+      "loss": 4.2128,
       "step": 268
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 4.2887,
+      "epoch": 0.23,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 4.2191,
       "step": 269
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.36e-06,
-      "loss": 4.1665,
+      "epoch": 0.23,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 4.1468,
       "step": 270
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 4.0903,
+      "epoch": 0.23,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 4.4825,
       "step": 271
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 4.2757,
+      "epoch": 0.23,
+      "learning_rate": 5.36e-06,
+      "loss": 4.1264,
       "step": 272
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 4.0794,
+      "epoch": 0.23,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 4.2394,
       "step": 273
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 5.4400000000000004e-06,
-      "loss": 4.1791,
+      "epoch": 0.23,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.2723,
       "step": 274
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 4.2729,
+      "epoch": 0.23,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 4.3256,
       "step": 275
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 4.2107,
+      "epoch": 0.23,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 4.0625,
       "step": 276
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 4.1623,
+      "epoch": 0.23,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 4.0314,
       "step": 277
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 4.2513,
+      "epoch": 0.23,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 4.2099,
       "step": 278
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.540000000000001e-06,
-      "loss": 4.235,
+      "epoch": 0.23,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.3404,
       "step": 279
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.560000000000001e-06,
-      "loss": 4.3647,
+      "epoch": 0.24,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 4.1768,
       "step": 280
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 4.2633,
+      "epoch": 0.24,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 4.2114,
       "step": 281
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.12,
+      "epoch": 0.24,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 4.3666,
       "step": 282
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 5.620000000000001e-06,
-      "loss": 4.1451,
+      "epoch": 0.24,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 4.3276,
       "step": 283
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.64e-06,
-      "loss": 4.2252,
+      "epoch": 0.24,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.1825,
       "step": 284
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.66e-06,
-      "loss": 4.3205,
+      "epoch": 0.24,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 4.3904,
       "step": 285
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.68e-06,
-      "loss": 4.2208,
+      "epoch": 0.24,
+      "learning_rate": 5.64e-06,
+      "loss": 4.2196,
       "step": 286
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.7e-06,
-      "loss": 4.3254,
+      "epoch": 0.24,
+      "learning_rate": 5.66e-06,
+      "loss": 4.1458,
       "step": 287
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.72e-06,
-      "loss": 4.2817,
+      "epoch": 0.24,
+      "learning_rate": 5.68e-06,
+      "loss": 4.2283,
       "step": 288
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.74e-06,
-      "loss": 4.192,
+      "epoch": 0.24,
+      "learning_rate": 5.7e-06,
+      "loss": 4.4968,
       "step": 289
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.76e-06,
-      "loss": 4.2462,
+      "epoch": 0.24,
+      "learning_rate": 5.72e-06,
+      "loss": 4.3905,
       "step": 290
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.78e-06,
-      "loss": 4.4973,
+      "epoch": 0.24,
+      "learning_rate": 5.74e-06,
+      "loss": 4.3475,
       "step": 291
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.8e-06,
-      "loss": 4.3751,
+      "epoch": 0.25,
+      "learning_rate": 5.76e-06,
+      "loss": 4.6025,
       "step": 292
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.82e-06,
-      "loss": 4.441,
+      "epoch": 0.25,
+      "learning_rate": 5.78e-06,
+      "loss": 4.1237,
       "step": 293
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.84e-06,
-      "loss": 4.4067,
+      "epoch": 0.25,
+      "learning_rate": 5.8e-06,
+      "loss": 4.4909,
       "step": 294
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.86e-06,
-      "loss": 4.3823,
+      "epoch": 0.25,
+      "learning_rate": 5.82e-06,
+      "loss": 4.4225,
       "step": 295
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.3868,
+      "epoch": 0.25,
+      "learning_rate": 5.84e-06,
+      "loss": 4.9821,
       "step": 296
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.9e-06,
-      "loss": 4.2941,
+      "epoch": 0.25,
+      "learning_rate": 5.86e-06,
+      "loss": 4.3145,
       "step": 297
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.92e-06,
-      "loss": 4.6034,
+      "epoch": 0.25,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.9149,
       "step": 298
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.94e-06,
-      "loss": 4.37,
+      "epoch": 0.25,
+      "learning_rate": 5.9e-06,
+      "loss": 4.5643,
       "step": 299
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 4.7808,
+      "epoch": 0.25,
+      "learning_rate": 5.92e-06,
+      "loss": 4.7415,
       "step": 300
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.98e-06,
-      "loss": 4.268,
+      "epoch": 0.25,
+      "learning_rate": 5.94e-06,
+      "loss": 4.0378,
       "step": 301
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6e-06,
-      "loss": 4.2326,
+      "epoch": 0.25,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 4.1696,
       "step": 302
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.02e-06,
-      "loss": 4.2418,
+      "epoch": 0.25,
+      "learning_rate": 5.98e-06,
+      "loss": 4.1995,
       "step": 303
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.040000000000001e-06,
-      "loss": 4.2376,
-      "step": 304
+      "epoch": 0.26,
+      "learning_rate": 6e-06,
+      "loss": 3.8702,
+      "step": 304
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.2091,
+      "epoch": 0.26,
+      "learning_rate": 6.02e-06,
+      "loss": 4.0612,
       "step": 305
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.08e-06,
-      "loss": 4.2111,
+      "epoch": 0.26,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 4.061,
       "step": 306
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.1e-06,
-      "loss": 4.2004,
+      "epoch": 0.26,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.263,
       "step": 307
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.120000000000001e-06,
-      "loss": 4.1965,
+      "epoch": 0.26,
+      "learning_rate": 6.08e-06,
+      "loss": 3.9791,
       "step": 308
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 6.1400000000000005e-06,
-      "loss": 4.1974,
+      "epoch": 0.26,
+      "learning_rate": 6.1e-06,
+      "loss": 3.9582,
       "step": 309
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.16e-06,
-      "loss": 4.2748,
+      "epoch": 0.26,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 4.1761,
       "step": 310
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.18e-06,
-      "loss": 4.0933,
+      "epoch": 0.26,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 4.095,
       "step": 311
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.200000000000001e-06,
-      "loss": 4.2521,
+      "epoch": 0.26,
+      "learning_rate": 6.16e-06,
+      "loss": 4.1012,
       "step": 312
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.220000000000001e-06,
-      "loss": 4.2651,
+      "epoch": 0.26,
+      "learning_rate": 6.18e-06,
+      "loss": 4.0731,
       "step": 313
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 6.24e-06,
-      "loss": 4.182,
+      "epoch": 0.26,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.1352,
       "step": 314
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.26e-06,
-      "loss": 4.2096,
+      "epoch": 0.26,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 4.0554,
       "step": 315
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.280000000000001e-06,
-      "loss": 4.1797,
+      "epoch": 0.27,
+      "learning_rate": 6.24e-06,
+      "loss": 4.4064,
       "step": 316
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 4.1667,
+      "epoch": 0.27,
+      "learning_rate": 6.26e-06,
+      "loss": 4.365,
       "step": 317
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 6.3200000000000005e-06,
-      "loss": 4.2119,
+      "epoch": 0.27,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 4.288,
       "step": 318
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.34e-06,
-      "loss": 4.1995,
+      "epoch": 0.27,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.1091,
       "step": 319
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.360000000000001e-06,
-      "loss": 4.2797,
+      "epoch": 0.27,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 4.2798,
       "step": 320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.380000000000001e-06,
-      "loss": 4.1701,
+      "epoch": 0.27,
+      "learning_rate": 6.34e-06,
+      "loss": 4.1208,
       "step": 321
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.0619,
+      "epoch": 0.27,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 4.1396,
       "step": 322
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 6.42e-06,
-      "loss": 4.2404,
+      "epoch": 0.27,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 4.0769,
       "step": 323
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.440000000000001e-06,
-      "loss": 4.1039,
+      "epoch": 0.27,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.2889,
       "step": 324
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 4.2368,
+      "epoch": 0.27,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2089,
       "step": 325
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.480000000000001e-06,
-      "loss": 4.1967,
+      "epoch": 0.27,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 4.3849,
       "step": 326
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 4.1744,
+      "epoch": 0.27,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 4.3868,
       "step": 327
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.520000000000001e-06,
-      "loss": 4.1501,
+      "epoch": 0.28,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 4.2376,
       "step": 328
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.540000000000001e-06,
-      "loss": 4.1861,
+      "epoch": 0.28,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.1523,
       "step": 329
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.560000000000001e-06,
-      "loss": 4.2375,
+      "epoch": 0.28,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 4.3864,
       "step": 330
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.5800000000000005e-06,
-      "loss": 4.2348,
+      "epoch": 0.28,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 4.0835,
       "step": 331
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 6.600000000000001e-06,
-      "loss": 4.133,
+      "epoch": 0.28,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 4.2166,
       "step": 332
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.620000000000001e-06,
-      "loss": 4.2921,
+      "epoch": 0.28,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 4.2806,
       "step": 333
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.640000000000001e-06,
-      "loss": 4.1948,
+      "epoch": 0.28,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.0441,
       "step": 334
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.3708,
+      "epoch": 0.28,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 4.292,
       "step": 335
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 6.680000000000001e-06,
-      "loss": 4.1533,
+      "epoch": 0.28,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 4.3229,
       "step": 336
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 4.1618,
+      "epoch": 0.28,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.2457,
       "step": 337
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.720000000000001e-06,
-      "loss": 4.267,
+      "epoch": 0.28,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 4.4293,
       "step": 338
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 4.2544,
+      "epoch": 0.28,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.4176,
       "step": 339
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.760000000000001e-06,
-      "loss": 4.3039,
+      "epoch": 0.29,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 4.2563,
       "step": 340
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 6.780000000000001e-06,
-      "loss": 4.2502,
+      "epoch": 0.29,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 4.3147,
       "step": 341
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.4135,
+      "epoch": 0.29,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 4.4248,
       "step": 342
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.820000000000001e-06,
-      "loss": 4.3116,
+      "epoch": 0.29,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 4.3332,
       "step": 343
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.8400000000000014e-06,
-      "loss": 4.3977,
+      "epoch": 0.29,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4504,
       "step": 344
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 6.860000000000001e-06,
-      "loss": 4.135,
+      "epoch": 0.29,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 4.3014,
       "step": 345
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.88e-06,
-      "loss": 4.3846,
+      "epoch": 0.29,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 4.2469,
       "step": 346
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.9e-06,
-      "loss": 4.3741,
+      "epoch": 0.29,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 4.4505,
       "step": 347
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.92e-06,
-      "loss": 4.3161,
+      "epoch": 0.29,
+      "learning_rate": 6.88e-06,
+      "loss": 4.6718,
       "step": 348
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.9400000000000005e-06,
-      "loss": 4.5413,
+      "epoch": 0.29,
+      "learning_rate": 6.9e-06,
+      "loss": 4.8438,
       "step": 349
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 6.96e-06,
-      "loss": 4.5532,
+      "epoch": 0.29,
+      "learning_rate": 6.92e-06,
+      "loss": 4.4597,
       "step": 350
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 6.98e-06,
-      "loss": 4.275,
+      "epoch": 0.3,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 4.451,
       "step": 351
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7e-06,
-      "loss": 4.2207,
+      "epoch": 0.3,
+      "learning_rate": 6.96e-06,
+      "loss": 4.1534,
       "step": 352
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.1401,
+      "epoch": 0.3,
+      "learning_rate": 6.98e-06,
+      "loss": 4.1522,
       "step": 353
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 7.04e-06,
-      "loss": 4.0229,
+      "epoch": 0.3,
+      "learning_rate": 7e-06,
+      "loss": 4.3362,
       "step": 354
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.06e-06,
-      "loss": 4.2197,
+      "epoch": 0.3,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.1482,
       "step": 355
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.08e-06,
-      "loss": 4.3189,
+      "epoch": 0.3,
+      "learning_rate": 7.04e-06,
+      "loss": 4.1704,
       "step": 356
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 4.1967,
+      "epoch": 0.3,
+      "learning_rate": 7.06e-06,
+      "loss": 4.3393,
       "step": 357
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.1200000000000004e-06,
-      "loss": 4.2343,
+      "epoch": 0.3,
+      "learning_rate": 7.08e-06,
+      "loss": 4.0683,
       "step": 358
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 7.14e-06,
-      "loss": 4.1164,
+      "epoch": 0.3,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 4.2408,
       "step": 359
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.16e-06,
-      "loss": 4.1943,
+      "epoch": 0.3,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 4.3694,
       "step": 360
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.180000000000001e-06,
-      "loss": 4.1379,
+      "epoch": 0.3,
+      "learning_rate": 7.14e-06,
+      "loss": 4.2325,
       "step": 361
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.2000000000000005e-06,
-      "loss": 4.1926,
+      "epoch": 0.3,
+      "learning_rate": 7.16e-06,
+      "loss": 4.0877,
       "step": 362
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 7.22e-06,
-      "loss": 4.1157,
+      "epoch": 0.31,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 4.0998,
       "step": 363
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.24e-06,
-      "loss": 4.2101,
+      "epoch": 0.31,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 4.2975,
       "step": 364
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.260000000000001e-06,
-      "loss": 4.3345,
+      "epoch": 0.31,
+      "learning_rate": 7.22e-06,
+      "loss": 4.2319,
       "step": 365
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 4.1492,
+      "epoch": 0.31,
+      "learning_rate": 7.24e-06,
+      "loss": 4.1875,
       "step": 366
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 7.3e-06,
-      "loss": 4.2292,
+      "epoch": 0.31,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 4.0969,
       "step": 367
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.32e-06,
-      "loss": 4.212,
+      "epoch": 0.31,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 4.5461,
       "step": 368
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.340000000000001e-06,
-      "loss": 4.1305,
+      "epoch": 0.31,
+      "learning_rate": 7.3e-06,
+      "loss": 4.3578,
       "step": 369
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.360000000000001e-06,
-      "loss": 4.141,
+      "epoch": 0.31,
+      "learning_rate": 7.32e-06,
+      "loss": 4.2193,
       "step": 370
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.1366,
+      "epoch": 0.31,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 4.1472,
       "step": 371
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 7.4e-06,
-      "loss": 4.2152,
+      "epoch": 0.31,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 4.0958,
       "step": 372
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.420000000000001e-06,
-      "loss": 4.2039,
+      "epoch": 0.31,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.1036,
       "step": 373
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.440000000000001e-06,
-      "loss": 4.0068,
+      "epoch": 0.31,
+      "learning_rate": 7.4e-06,
+      "loss": 4.1224,
       "step": 374
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.4600000000000006e-06,
-      "loss": 4.2379,
+      "epoch": 0.32,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 4.1717,
       "step": 375
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.48e-06,
-      "loss": 4.0893,
+      "epoch": 0.32,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 4.2178,
       "step": 376
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 4.2137,
+      "epoch": 0.32,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 4.1724,
       "step": 377
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.520000000000001e-06,
-      "loss": 4.1977,
+      "epoch": 0.32,
+      "learning_rate": 7.48e-06,
+      "loss": 4.1578,
       "step": 378
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.540000000000001e-06,
-      "loss": 4.1133,
+      "epoch": 0.32,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 4.1216,
       "step": 379
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.1347,
+      "epoch": 0.32,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 4.2636,
       "step": 380
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 7.58e-06,
-      "loss": 4.1941,
+      "epoch": 0.32,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 4.0921,
       "step": 381
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.600000000000001e-06,
-      "loss": 4.2719,
+      "epoch": 0.32,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.2838,
       "step": 382
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.620000000000001e-06,
-      "loss": 4.1831,
+      "epoch": 0.32,
+      "learning_rate": 7.58e-06,
+      "loss": 4.0517,
       "step": 383
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.640000000000001e-06,
-      "loss": 4.1165,
+      "epoch": 0.32,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 4.3327,
       "step": 384
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.660000000000001e-06,
-      "loss": 4.2029,
+      "epoch": 0.32,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 4.3687,
       "step": 385
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 4.0781,
+      "epoch": 0.32,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 4.0989,
       "step": 386
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.7e-06,
-      "loss": 4.2651,
+      "epoch": 0.33,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 4.3151,
       "step": 387
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.72e-06,
-      "loss": 4.2909,
+      "epoch": 0.33,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.3546,
       "step": 388
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.74e-06,
-      "loss": 4.1941,
+      "epoch": 0.33,
+      "learning_rate": 7.7e-06,
+      "loss": 4.4011,
       "step": 389
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.76e-06,
-      "loss": 4.172,
+      "epoch": 0.33,
+      "learning_rate": 7.72e-06,
+      "loss": 4.3619,
       "step": 390
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.78e-06,
-      "loss": 4.2985,
+      "epoch": 0.33,
+      "learning_rate": 7.74e-06,
+      "loss": 4.4163,
       "step": 391
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.800000000000002e-06,
-      "loss": 4.1451,
+      "epoch": 0.33,
+      "learning_rate": 7.76e-06,
+      "loss": 4.2793,
       "step": 392
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.820000000000001e-06,
-      "loss": 4.1401,
+      "epoch": 0.33,
+      "learning_rate": 7.78e-06,
+      "loss": 4.1889,
       "step": 393
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.840000000000001e-06,
-      "loss": 4.2258,
+      "epoch": 0.33,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 4.4287,
       "step": 394
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 4.3488,
+      "epoch": 0.33,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 4.1601,
       "step": 395
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.88e-06,
-      "loss": 4.3106,
+      "epoch": 0.33,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 4.4287,
       "step": 396
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.9e-06,
-      "loss": 4.4439,
+      "epoch": 0.33,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.353,
       "step": 397
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.92e-06,
-      "loss": 4.4548,
+      "epoch": 0.33,
+      "learning_rate": 7.88e-06,
+      "loss": 4.4626,
       "step": 398
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 7.94e-06,
-      "loss": 4.5917,
+      "epoch": 0.34,
+      "learning_rate": 7.9e-06,
+      "loss": 4.744,
       "step": 399
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 7.960000000000002e-06,
-      "loss": 4.5111,
+      "epoch": 0.34,
+      "learning_rate": 7.92e-06,
+      "loss": 4.6632,
       "step": 400
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 7.980000000000002e-06,
-      "loss": 4.2176,
+      "epoch": 0.34,
+      "learning_rate": 7.94e-06,
+      "loss": 4.0428,
       "step": 401
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.0665,
+      "epoch": 0.34,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 4.1774,
       "step": 402
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.020000000000001e-06,
-      "loss": 4.1243,
+      "epoch": 0.34,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 4.1853,
       "step": 403
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 4.1244,
+      "epoch": 0.34,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.2826,
       "step": 404
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.06e-06,
-      "loss": 4.0896,
+      "epoch": 0.34,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 4.1573,
       "step": 405
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.08e-06,
-      "loss": 4.1743,
+      "epoch": 0.34,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.1539,
       "step": 406
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.1e-06,
-      "loss": 4.1469,
+      "epoch": 0.34,
+      "learning_rate": 8.06e-06,
+      "loss": 4.0891,
       "step": 407
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 8.120000000000002e-06,
-      "loss": 4.2224,
+      "epoch": 0.34,
+      "learning_rate": 8.08e-06,
+      "loss": 4.1684,
       "step": 408
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.14e-06,
-      "loss": 4.1593,
+      "epoch": 0.34,
+      "learning_rate": 8.1e-06,
+      "loss": 4.1047,
       "step": 409
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.16e-06,
-      "loss": 4.1418,
+      "epoch": 0.34,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 4.3541,
       "step": 410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.18e-06,
-      "loss": 4.3698,
+      "epoch": 0.35,
+      "learning_rate": 8.14e-06,
+      "loss": 4.3095,
       "step": 411
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.2e-06,
-      "loss": 4.1025,
+      "epoch": 0.35,
+      "learning_rate": 8.16e-06,
+      "loss": 4.2742,
       "step": 412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 4.0313,
+      "epoch": 0.35,
+      "learning_rate": 8.18e-06,
+      "loss": 4.2365,
       "step": 413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.24e-06,
-      "loss": 4.192,
+      "epoch": 0.35,
+      "learning_rate": 8.2e-06,
+      "loss": 4.0894,
       "step": 414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.26e-06,
-      "loss": 4.2583,
+      "epoch": 0.35,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.1794,
       "step": 415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.28e-06,
-      "loss": 4.1274,
+      "epoch": 0.35,
+      "learning_rate": 8.24e-06,
+      "loss": 4.281,
       "step": 416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 8.3e-06,
-      "loss": 4.152,
+      "epoch": 0.35,
+      "learning_rate": 8.26e-06,
+      "loss": 3.9741,
       "step": 417
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.32e-06,
-      "loss": 4.0524,
+      "epoch": 0.35,
+      "learning_rate": 8.28e-06,
+      "loss": 4.0795,
       "step": 418
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.34e-06,
-      "loss": 4.1577,
+      "epoch": 0.35,
+      "learning_rate": 8.3e-06,
+      "loss": 4.2742,
       "step": 419
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.36e-06,
-      "loss": 4.0665,
+      "epoch": 0.35,
+      "learning_rate": 8.32e-06,
+      "loss": 4.2141,
       "step": 420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.380000000000001e-06,
-      "loss": 4.2451,
+      "epoch": 0.35,
+      "learning_rate": 8.34e-06,
+      "loss": 4.1605,
       "step": 421
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.0875,
-      "step": 422
-    },
+      "epoch": 0.35,
+      "learning_rate": 8.36e-06,
+      "loss": 4.2531,
+      "step": 422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 4.2343,
+      "step": 423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.1663,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.42e-06,
+      "loss": 4.4545,
+      "step": 425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.44e-06,
+      "loss": 4.2354,
+      "step": 426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.46e-06,
+      "loss": 4.1138,
+      "step": 427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.48e-06,
+      "loss": 4.153,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.5e-06,
+      "loss": 4.4106,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.52e-06,
+      "loss": 4.3526,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 4.2253,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 4.1793,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 4.2763,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.6e-06,
+      "loss": 4.1914,
+      "step": 434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.62e-06,
+      "loss": 4.4306,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.64e-06,
+      "loss": 4.1213,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.66e-06,
+      "loss": 4.1672,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.68e-06,
+      "loss": 4.089,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 4.3148,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 4.2049,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 4.1278,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.76e-06,
+      "loss": 4.2936,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.78e-06,
+      "loss": 4.2392,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.8e-06,
+      "loss": 4.3485,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.82e-06,
+      "loss": 4.2356,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.84e-06,
+      "loss": 4.5869,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 4.5158,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 4.4858,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 4.5241,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 4.5914,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.94e-06,
+      "loss": 4.1306,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.96e-06,
+      "loss": 4.0371,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.98e-06,
+      "loss": 4.1242,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9e-06,
+      "loss": 4.2175,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 4.092,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 4.1632,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 4.0976,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 4.3298,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 4.1006,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.12e-06,
+      "loss": 4.2458,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.14e-06,
+      "loss": 4.3319,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.16e-06,
+      "loss": 4.2077,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 4.2044,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 4.2696,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 4.289,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 4.0717,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 4.2096,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 4.3025,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.3e-06,
+      "loss": 4.0658,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.32e-06,
+      "loss": 4.4158,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 4.2958,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 4.3186,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.38e-06,
+      "loss": 3.985,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.4e-06,
+      "loss": 4.3012,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.42e-06,
+      "loss": 4.3727,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 4.2418,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 4.0469,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.48e-06,
+      "loss": 4.1727,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.5e-06,
+      "loss": 4.2292,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.52e-06,
+      "loss": 4.2552,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.54e-06,
+      "loss": 4.1526,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.56e-06,
+      "loss": 4.2476,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.58e-06,
+      "loss": 4.2338,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.2995,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 4.2562,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 4.2279,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.66e-06,
+      "loss": 4.0867,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.68e-06,
+      "loss": 4.1969,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.7e-06,
+      "loss": 3.9597,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.72e-06,
+      "loss": 4.2366,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.74e-06,
+      "loss": 4.2481,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 4.1417,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.3447,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.3314,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 4.5135,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.84e-06,
+      "loss": 4.3053,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.86e-06,
+      "loss": 4.3441,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.88e-06,
+      "loss": 4.5401,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.9e-06,
+      "loss": 4.4404,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 4.4352,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 4.231882095336914,
+      "eval_runtime": 764.4732,
+      "eval_samples_per_second": 3.456,
+      "eval_steps_per_second": 0.289,
+      "eval_wer": 1.9545418484728283,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 4.1748,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.3075,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 4.1517,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 3.972,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.985486211901307e-06,
+      "loss": 4.198,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970972423802613e-06,
+      "loss": 4.3154,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.95645863570392e-06,
+      "loss": 4.058,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941944847605226e-06,
+      "loss": 4.115,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.927431059506532e-06,
+      "loss": 4.2455,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.912917271407838e-06,
+      "loss": 4.4064,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.898403483309144e-06,
+      "loss": 4.2228,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.883889695210452e-06,
+      "loss": 4.0367,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.869375907111757e-06,
+      "loss": 4.2327,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.854862119013063e-06,
+      "loss": 4.0423,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.84034833091437e-06,
+      "loss": 4.0366,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.825834542815675e-06,
+      "loss": 4.1171,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 4.1935,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.79680696661829e-06,
+      "loss": 4.0721,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.782293178519594e-06,
+      "loss": 4.5655,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.7677793904209e-06,
+      "loss": 4.1439,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.753265602322208e-06,
+      "loss": 4.2295,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.738751814223512e-06,
+      "loss": 4.2278,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.724238026124818e-06,
+      "loss": 4.2069,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.709724238026126e-06,
+      "loss": 4.1885,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.69521044992743e-06,
+      "loss": 4.2169,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.680696661828739e-06,
+      "loss": 4.0353,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.666182873730045e-06,
+      "loss": 4.1873,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.651669085631351e-06,
+      "loss": 4.1126,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.637155297532657e-06,
+      "loss": 4.3759,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 4.3769,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.60812772133527e-06,
+      "loss": 4.3706,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.593613933236576e-06,
+      "loss": 4.3061,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 4.226,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.564586357039188e-06,
+      "loss": 4.2568,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.550072568940494e-06,
+      "loss": 4.1621,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.5355587808418e-06,
+      "loss": 4.1059,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.521044992743107e-06,
+      "loss": 4.2401,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.506531204644413e-06,
+      "loss": 3.8417,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.49201741654572e-06,
+      "loss": 4.3224,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.477503628447025e-06,
+      "loss": 4.2402,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.462989840348332e-06,
+      "loss": 4.3781,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.448476052249638e-06,
+      "loss": 4.2109,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 4.1765,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.41944847605225e-06,
+      "loss": 4.2998,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.404934687953556e-06,
+      "loss": 4.4192,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.390420899854863e-06,
+      "loss": 4.3058,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.375907111756169e-06,
+      "loss": 4.1501,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.361393323657475e-06,
+      "loss": 4.3889,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.346879535558781e-06,
+      "loss": 4.6156,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.332365747460089e-06,
+      "loss": 4.5578,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.317851959361393e-06,
+      "loss": 4.3208,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.3033381712627e-06,
+      "loss": 4.1234,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.288824383164008e-06,
+      "loss": 4.1842,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.274310595065312e-06,
+      "loss": 4.2397,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.259796806966618e-06,
+      "loss": 4.2865,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 4.1397,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 4.1075,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.216255442670537e-06,
+      "loss": 4.1896,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.201741654571845e-06,
+      "loss": 4.3666,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.187227866473151e-06,
+      "loss": 4.0635,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.172714078374455e-06,
+      "loss": 4.1871,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.158200290275763e-06,
+      "loss": 4.0503,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.14368650217707e-06,
+      "loss": 4.2526,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.129172714078376e-06,
+      "loss": 3.9915,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.114658925979682e-06,
+      "loss": 4.0582,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.100145137880988e-06,
+      "loss": 4.2332,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.085631349782294e-06,
+      "loss": 4.1486,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.0711175616836e-06,
+      "loss": 4.1691,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 4.0449,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.042089985486213e-06,
+      "loss": 4.1586,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.027576197387519e-06,
+      "loss": 4.1773,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.013062409288825e-06,
+      "loss": 4.309,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.998548621190131e-06,
+      "loss": 4.1865,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.984034833091438e-06,
+      "loss": 4.1503,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.969521044992744e-06,
+      "loss": 4.2702,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.95500725689405e-06,
+      "loss": 4.3293,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.940493468795356e-06,
+      "loss": 4.3696,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.925979680696662e-06,
+      "loss": 4.1541,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.911465892597968e-06,
+      "loss": 4.2035,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.896952104499275e-06,
+      "loss": 4.2706,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.882438316400581e-06,
+      "loss": 4.4201,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 3.9767,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.853410740203193e-06,
+      "loss": 4.1309,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.8388969521045e-06,
+      "loss": 4.135,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.824383164005806e-06,
+      "loss": 4.3354,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.809869375907114e-06,
+      "loss": 4.3084,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.795355587808418e-06,
+      "loss": 4.3638,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.780841799709724e-06,
+      "loss": 4.1925,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.766328011611032e-06,
+      "loss": 4.1138,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.751814223512337e-06,
+      "loss": 4.1781,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.737300435413644e-06,
+      "loss": 4.3113,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.72278664731495e-06,
+      "loss": 4.105,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 4.536,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.693759071117563e-06,
+      "loss": 4.209,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 4.322,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.664731494920174e-06,
+      "loss": 4.0547,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.650217706821482e-06,
+      "loss": 4.3088,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.635703918722788e-06,
+      "loss": 4.3375,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.621190130624092e-06,
+      "loss": 4.2899,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.6066763425254e-06,
+      "loss": 4.7879,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.592162554426706e-06,
+      "loss": 4.1294,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.577648766328013e-06,
+      "loss": 4.2565,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.563134978229319e-06,
+      "loss": 4.4176,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.548621190130625e-06,
+      "loss": 3.9883,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.534107402031931e-06,
+      "loss": 4.2052,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.519593613933237e-06,
+      "loss": 4.2851,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.505079825834544e-06,
+      "loss": 4.2083,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 4.2721,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.476052249637156e-06,
+      "loss": 4.2145,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 4.2757,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.447024673439768e-06,
+      "loss": 4.2753,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.432510885341074e-06,
+      "loss": 4.0202,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.41799709724238e-06,
+      "loss": 4.1608,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.403483309143687e-06,
+      "loss": 4.0613,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.388969521044995e-06,
+      "loss": 4.2349,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.3744557329463e-06,
+      "loss": 4.1795,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.359941944847605e-06,
+      "loss": 4.1027,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.345428156748913e-06,
+      "loss": 4.1338,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.330914368650218e-06,
+      "loss": 4.1729,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.316400580551524e-06,
+      "loss": 4.2097,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.301886792452832e-06,
+      "loss": 3.9941,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.287373004354136e-06,
+      "loss": 3.9089,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.272859216255443e-06,
+      "loss": 4.1862,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.25834542815675e-06,
+      "loss": 4.2405,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.243831640058055e-06,
+      "loss": 4.1887,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.229317851959361e-06,
+      "loss": 4.2669,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.214804063860669e-06,
+      "loss": 4.3261,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.200290275761974e-06,
+      "loss": 3.9848,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.185776487663281e-06,
+      "loss": 3.9823,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.171262699564588e-06,
+      "loss": 4.2373,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.156748911465894e-06,
+      "loss": 4.2744,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.1422351233672e-06,
+      "loss": 4.1337,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.127721335268506e-06,
+      "loss": 4.1879,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 4.1397,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.098693759071119e-06,
+      "loss": 4.3102,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.084179970972425e-06,
+      "loss": 4.1755,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.069666182873731e-06,
+      "loss": 3.9267,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.055152394775037e-06,
+      "loss": 4.0985,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.040638606676343e-06,
+      "loss": 4.1355,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.02612481857765e-06,
+      "loss": 4.2305,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.011611030478956e-06,
+      "loss": 4.1717,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.997097242380262e-06,
+      "loss": 4.135,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.982583454281568e-06,
+      "loss": 4.4083,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.968069666182874e-06,
+      "loss": 4.2115,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.95355587808418e-06,
+      "loss": 4.2313,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.939042089985487e-06,
+      "loss": 4.2436,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 4.0744,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.910014513788099e-06,
+      "loss": 4.591,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.895500725689405e-06,
+      "loss": 4.7783,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.880986937590711e-06,
+      "loss": 4.1967,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.866473149492018e-06,
+      "loss": 4.3649,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.851959361393324e-06,
+      "loss": 4.1432,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.837445573294632e-06,
+      "loss": 4.3148,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.822931785195936e-06,
+      "loss": 4.2245,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.808417997097242e-06,
+      "loss": 4.217,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.79390420899855e-06,
+      "loss": 4.1993,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.779390420899855e-06,
+      "loss": 4.0829,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.764876632801161e-06,
+      "loss": 4.1662,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.750362844702469e-06,
+      "loss": 4.1472,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.735849056603775e-06,
+      "loss": 4.1191,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.72133526850508e-06,
+      "loss": 4.1532,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.706821480406387e-06,
+      "loss": 4.0615,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 4.2343,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.677793904208998e-06,
+      "loss": 4.1796,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.663280116110306e-06,
+      "loss": 4.3712,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.648766328011612e-06,
+      "loss": 4.2091,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.634252539912917e-06,
+      "loss": 4.2797,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.619738751814224e-06,
+      "loss": 4.1666,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.605224963715531e-06,
+      "loss": 4.0812,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.590711175616836e-06,
+      "loss": 4.0626,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.576197387518143e-06,
+      "loss": 4.1902,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.561683599419449e-06,
+      "loss": 4.119,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.5471698113207555e-06,
+      "loss": 4.302,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.532656023222062e-06,
+      "loss": 4.2364,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.518142235123368e-06,
+      "loss": 4.1363,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.503628447024675e-06,
+      "loss": 3.8786,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.48911465892598e-06,
+      "loss": 4.3422,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.4746008708272864e-06,
+      "loss": 4.1787,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.4600870827285935e-06,
+      "loss": 4.0855,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.445573294629899e-06,
+      "loss": 4.3822,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.431059506531205e-06,
+      "loss": 3.894,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.416545718432512e-06,
+      "loss": 4.0226,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.402031930333817e-06,
+      "loss": 4.3122,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387518142235124e-06,
+      "loss": 4.0051,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373004354136431e-06,
+      "loss": 4.1807,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 4.2128,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.343976777939042e-06,
+      "loss": 4.157,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.329462989840349e-06,
+      "loss": 4.1944,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.314949201741655e-06,
+      "loss": 4.3527,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.300435413642961e-06,
+      "loss": 4.2305,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.285921625544268e-06,
+      "loss": 4.4885,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.271407837445574e-06,
+      "loss": 4.395,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.25689404934688e-06,
+      "loss": 4.4228,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.242380261248186e-06,
+      "loss": 4.385,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.227866473149493e-06,
+      "loss": 4.1925,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213352685050799e-06,
+      "loss": 4.0685,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.198838896952105e-06,
+      "loss": 4.3157,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184325108853412e-06,
+      "loss": 4.5706,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 4.1089,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1552975326560235e-06,
+      "loss": 4.9165,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1407837445573305e-06,
+      "loss": 4.2557,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126269956458637e-06,
+      "loss": 4.0924,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.111756168359942e-06,
+      "loss": 4.1665,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097242380261249e-06,
+      "loss": 3.8733,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.082728592162555e-06,
+      "loss": 4.1113,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.068214804063861e-06,
+      "loss": 4.1563,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.053701015965168e-06,
+      "loss": 4.1824,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039187227866474e-06,
+      "loss": 4.0821,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.024673439767779e-06,
+      "loss": 4.1771,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010159651669086e-06,
+      "loss": 4.1424,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9956458635703924e-06,
+      "loss": 4.0633,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 4.265,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.966618287373005e-06,
+      "loss": 4.092,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.952104499274311e-06,
+      "loss": 4.0288,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937590711175617e-06,
+      "loss": 4.1539,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 4.0319,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90856313497823e-06,
+      "loss": 4.1057,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894049346879537e-06,
+      "loss": 4.2959,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.879535558780842e-06,
+      "loss": 4.1332,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865021770682149e-06,
+      "loss": 4.0295,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850507982583455e-06,
+      "loss": 4.1275,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8359941944847605e-06,
+      "loss": 4.0037,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8214804063860676e-06,
+      "loss": 4.0735,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.806966618287374e-06,
+      "loss": 4.0577,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 4.2495,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777939042089986e-06,
+      "loss": 4.2403,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763425253991292e-06,
+      "loss": 4.1391,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.748911465892598e-06,
+      "loss": 4.3035,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734397677793905e-06,
+      "loss": 3.9943,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719883889695211e-06,
+      "loss": 4.3323,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705370101596517e-06,
+      "loss": 4.2674,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690856313497823e-06,
+      "loss": 4.2782,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6763425253991295e-06,
+      "loss": 3.9513,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6618287373004365e-06,
+      "loss": 4.2667,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647314949201742e-06,
+      "loss": 4.1659,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632801161103049e-06,
+      "loss": 4.1086,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618287373004355e-06,
+      "loss": 4.1891,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 4.3583,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5892597968069675e-06,
+      "loss": 4.2854,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574746008708274e-06,
+      "loss": 4.0883,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560232220609579e-06,
+      "loss": 4.4145,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545718432510886e-06,
+      "loss": 4.1187,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531204644412192e-06,
+      "loss": 4.2233,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5166908563134976e-06,
+      "loss": 4.3805,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502177068214805e-06,
+      "loss": 4.3494,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487663280116111e-06,
+      "loss": 4.3733,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473149492017418e-06,
+      "loss": 4.5544,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458635703918723e-06,
+      "loss": 4.7237,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444121915820029e-06,
+      "loss": 4.2943,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429608127721336e-06,
+      "loss": 4.5182,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 4.0224,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400580551523948e-06,
+      "loss": 4.2205,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386066763425255e-06,
+      "loss": 4.2447,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37155297532656e-06,
+      "loss": 4.0992,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357039187227867e-06,
+      "loss": 4.2706,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.3425253991291735e-06,
+      "loss": 4.1102,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.328011611030479e-06,
+      "loss": 4.1502,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.313497822931786e-06,
+      "loss": 4.0218,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.298984034833092e-06,
+      "loss": 4.0999,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2844702467343975e-06,
+      "loss": 4.298,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2699564586357045e-06,
+      "loss": 4.0177,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.255442670537011e-06,
+      "loss": 4.1421,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.240928882438318e-06,
+      "loss": 4.1744,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 4.0544,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211901306240929e-06,
+      "loss": 4.305,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197387518142236e-06,
+      "loss": 4.1454,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182873730043542e-06,
+      "loss": 4.1641,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.168359941944848e-06,
+      "loss": 3.964,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 4.1901,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.13933236574746e-06,
+      "loss": 4.328,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.124818577648766e-06,
+      "loss": 3.8464,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.1103047895500734e-06,
+      "loss": 4.2541,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095791001451379e-06,
+      "loss": 4.0349,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.081277213352685e-06,
+      "loss": 4.0205,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.066763425253992e-06,
+      "loss": 4.0569,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.052249637155298e-06,
+      "loss": 4.1137,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 4.1656,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.023222060957911e-06,
+      "loss": 4.1021,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.008708272859218e-06,
+      "loss": 4.0497,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.994194484760523e-06,
+      "loss": 4.2799,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.979680696661829e-06,
+      "loss": 4.0747,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.965166908563136e-06,
+      "loss": 4.0694,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9506531204644415e-06,
+      "loss": 3.8638,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.936139332365748e-06,
+      "loss": 4.3295,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.921625544267055e-06,
+      "loss": 4.3837,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.90711175616836e-06,
+      "loss": 4.0586,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.892597968069666e-06,
+      "loss": 4.3525,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.878084179970973e-06,
+      "loss": 4.0167,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.863570391872279e-06,
+      "loss": 4.0685,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.849056603773585e-06,
+      "loss": 3.911,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.834542815674892e-06,
+      "loss": 4.4236,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.820029027576198e-06,
+      "loss": 4.422,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8055152394775034e-06,
+      "loss": 4.6496,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7910014513788105e-06,
+      "loss": 4.1691,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.776487663280117e-06,
+      "loss": 4.2223,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.761973875181423e-06,
+      "loss": 3.9652,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.747460087082729e-06,
+      "loss": 4.1935,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.732946298984035e-06,
+      "loss": 4.378,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7184325108853414e-06,
+      "loss": 4.375,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.703918722786648e-06,
+      "loss": 4.2844,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.689404934687955e-06,
+      "loss": 4.0941,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.67489114658926e-06,
+      "loss": 4.0862,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 4.1142,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.645863570391873e-06,
+      "loss": 4.1178,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6313497822931794e-06,
+      "loss": 4.0416,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.616835994194485e-06,
+      "loss": 4.1524,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.602322206095792e-06,
+      "loss": 4.19,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587808417997098e-06,
+      "loss": 4.0566,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.573294629898403e-06,
+      "loss": 4.0622,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.55878084179971e-06,
+      "loss": 4.0385,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5442670537010166e-06,
+      "loss": 4.1658,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.529753265602322e-06,
+      "loss": 4.1893,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.515239477503629e-06,
+      "loss": 4.1749,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.500725689404935e-06,
+      "loss": 4.2915,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.486211901306241e-06,
+      "loss": 4.1206,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4716981132075475e-06,
+      "loss": 4.1914,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.457184325108854e-06,
+      "loss": 4.0393,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.44267053701016e-06,
+      "loss": 4.0259,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.428156748911466e-06,
+      "loss": 4.0525,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.413642960812773e-06,
+      "loss": 4.2178,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.399129172714079e-06,
+      "loss": 4.1946,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 3.9999,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.370101596516692e-06,
+      "loss": 4.1559,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.355587808417998e-06,
+      "loss": 4.3444,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.341074020319303e-06,
+      "loss": 4.0414,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.32656023222061e-06,
+      "loss": 4.4226,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.3120464441219165e-06,
+      "loss": 4.1593,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.297532656023222e-06,
+      "loss": 4.29,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 4.0673,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.268505079825835e-06,
+      "loss": 4.1496,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.25399129172714e-06,
+      "loss": 4.1466,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.239477503628447e-06,
+      "loss": 4.0823,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 3.9969,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.210449927431061e-06,
+      "loss": 4.2214,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.195936139332366e-06,
+      "loss": 4.2331,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.181422351233672e-06,
+      "loss": 4.2961,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.166908563134979e-06,
+      "loss": 4.0543,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1523947750362846e-06,
+      "loss": 3.8357,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.137880986937592e-06,
+      "loss": 4.2441,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.123367198838898e-06,
+      "loss": 3.9684,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108853410740203e-06,
+      "loss": 4.2472,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.09433962264151e-06,
+      "loss": 4.2463,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.079825834542816e-06,
+      "loss": 4.4879,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.065312046444122e-06,
+      "loss": 4.2717,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.050798258345429e-06,
+      "loss": 4.178,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.036284470246735e-06,
+      "loss": 4.124,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.02177068214804e-06,
+      "loss": 4.432,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.007256894049347e-06,
+      "loss": 3.9594,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9927431059506535e-06,
+      "loss": 4.5705,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.97822931785196e-06,
+      "loss": 4.6286,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.963715529753266e-06,
+      "loss": 4.2794,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.949201741654572e-06,
+      "loss": 4.127,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.934687953555878e-06,
+      "loss": 4.2281,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.920174165457185e-06,
+      "loss": 4.1554,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 4.088,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.891146589259797e-06,
+      "loss": 4.0942,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.876632801161104e-06,
+      "loss": 4.1263,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.862119013062409e-06,
+      "loss": 4.0371,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.847605224963715e-06,
+      "loss": 4.1453,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8330914368650224e-06,
+      "loss": 4.1848,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.818577648766329e-06,
+      "loss": 3.9399,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.804063860667635e-06,
+      "loss": 4.2367,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 4.2325,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.775036284470247e-06,
+      "loss": 4.0618,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.760522496371553e-06,
+      "loss": 4.1201,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.74600870827286e-06,
+      "loss": 4.0901,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.731494920174166e-06,
+      "loss": 4.0507,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 4.202,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.702467343976778e-06,
+      "loss": 4.2683,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.687953555878084e-06,
+      "loss": 3.8477,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6734397677793905e-06,
+      "loss": 3.8841,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.658925979680697e-06,
+      "loss": 4.0321,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644412191582004e-06,
+      "loss": 4.1771,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.629898403483309e-06,
+      "loss": 3.9922,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 4.1326,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.600870827285922e-06,
+      "loss": 4.0226,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.586357039187228e-06,
+      "loss": 3.99,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.571843251088535e-06,
+      "loss": 4.2206,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.557329462989841e-06,
+      "loss": 4.125,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.542815674891147e-06,
+      "loss": 4.3072,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 4.002,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5137880986937595e-06,
+      "loss": 4.2045,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.499274310595066e-06,
+      "loss": 4.1146,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.484760522496372e-06,
+      "loss": 4.1946,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.470246734397678e-06,
+      "loss": 4.0079,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.455732946298984e-06,
+      "loss": 4.1412,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.4412191582002904e-06,
+      "loss": 4.368,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.426705370101597e-06,
+      "loss": 4.27,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.412191582002903e-06,
+      "loss": 4.2894,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.397677793904209e-06,
+      "loss": 4.2795,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.383164005805516e-06,
+      "loss": 4.2595,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.368650217706822e-06,
+      "loss": 4.2167,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 4.168,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 4.0369,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.325108853410741e-06,
+      "loss": 4.2328,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310595065312046e-06,
+      "loss": 4.2767,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.296081277213353e-06,
+      "loss": 4.3103,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.281567489114659e-06,
+      "loss": 4.1354,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2670537010159656e-06,
+      "loss": 4.1333,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252539912917272e-06,
+      "loss": 4.4595,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.238026124818578e-06,
+      "loss": 4.1815,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.223512336719884e-06,
+      "loss": 4.2437,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.20899854862119e-06,
+      "loss": 4.1927,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.194484760522497e-06,
+      "loss": 4.1797,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.179970972423803e-06,
+      "loss": 4.0734,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.165457184325109e-06,
+      "loss": 4.2315,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 4.1391,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.136429608127721e-06,
+      "loss": 4.4274,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1219158200290275e-06,
+      "loss": 4.2076,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1074020319303345e-06,
+      "loss": 4.2602,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.092888243831641e-06,
+      "loss": 4.0593,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.078374455732947e-06,
+      "loss": 4.0798,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063860667634253e-06,
+      "loss": 4.1426,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.049346879535559e-06,
+      "loss": 4.0929,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0348330914368655e-06,
+      "loss": 4.1844,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.020319303338172e-06,
+      "loss": 4.1298,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.005805515239478e-06,
+      "loss": 4.1434,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.991291727140784e-06,
+      "loss": 4.0783,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.97677793904209e-06,
+      "loss": 4.315,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 4.1936,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.947750362844703e-06,
+      "loss": 4.1137,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.933236574746009e-06,
+      "loss": 4.2065,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.918722786647316e-06,
+      "loss": 4.0515,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.904208998548621e-06,
+      "loss": 4.1109,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.889695210449927e-06,
+      "loss": 4.0143,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.875181422351234e-06,
+      "loss": 4.0165,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.86066763425254e-06,
+      "loss": 4.1842,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 4.2189,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.831640058055153e-06,
+      "loss": 4.0395,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.817126269956458e-06,
+      "loss": 3.9832,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8026124818577654e-06,
+      "loss": 4.187,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7880986937590716e-06,
+      "loss": 4.2237,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 4.172,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.759071117561684e-06,
+      "loss": 4.2099,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.74455732946299e-06,
+      "loss": 4.2794,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7300435413642967e-06,
+      "loss": 4.2194,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7155297532656025e-06,
+      "loss": 4.182,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7010159651669087e-06,
+      "loss": 4.3308,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6865021770682153e-06,
+      "loss": 4.165,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.671988388969521e-06,
+      "loss": 4.1604,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6574746008708277e-06,
+      "loss": 4.0493,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.642960812772134e-06,
+      "loss": 4.2262,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.62844702467344e-06,
+      "loss": 4.0553,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6139332365747467e-06,
+      "loss": 4.2358,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.5994194484760525e-06,
+      "loss": 4.2829,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 4.0405,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5703918722786653e-06,
+      "loss": 4.272,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.555878084179971e-06,
+      "loss": 4.2367,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5413642960812776e-06,
+      "loss": 4.3242,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.526850507982584e-06,
+      "loss": 4.1556,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5123367198838896e-06,
+      "loss": 4.0573,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4978229317851962e-06,
+      "loss": 4.2132,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4833091436865024e-06,
+      "loss": 3.9992,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4687953555878086e-06,
+      "loss": 3.8626,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.454281567489115e-06,
+      "loss": 4.0842,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.439767779390421e-06,
+      "loss": 4.1929,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4252539912917276e-06,
+      "loss": 4.1345,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4107402031930338e-06,
+      "loss": 4.0344,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 4.0894,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.381712626995646e-06,
+      "loss": 4.0573,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3671988388969524e-06,
+      "loss": 4.0034,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3526850507982585e-06,
+      "loss": 4.2028,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3381712626995647e-06,
+      "loss": 3.994,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.323657474600871e-06,
+      "loss": 4.1946,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3091436865021775e-06,
+      "loss": 4.3112,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2946298984034837e-06,
+      "loss": 3.9698,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2801161103047895e-06,
+      "loss": 4.192,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.265602322206096e-06,
+      "loss": 4.2462,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2510885341074023e-06,
+      "loss": 3.8119,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.236574746008709e-06,
+      "loss": 4.1274,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2220609579100147e-06,
+      "loss": 3.9745,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 4.0464,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1930333817126275e-06,
+      "loss": 4.1265,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1785195936139337e-06,
+      "loss": 4.0401,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1640058055152394e-06,
+      "loss": 4.1446,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.149492017416546e-06,
+      "loss": 4.1552,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1349782293178522e-06,
+      "loss": 3.9802,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.120464441219159e-06,
+      "loss": 4.1522,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1059506531204646e-06,
+      "loss": 4.0127,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.091436865021771e-06,
+      "loss": 4.1173,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 4.2185,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.062409288824383e-06,
+      "loss": 4.1847,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0478955007256894e-06,
+      "loss": 4.0421,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033381712626996e-06,
+      "loss": 4.1865,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 3.9954,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.004354136429609e-06,
+      "loss": 4.1291,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9898403483309146e-06,
+      "loss": 4.2463,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9753265602322208e-06,
+      "loss": 4.0991,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9608127721335274e-06,
+      "loss": 4.1562,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.946298984034833e-06,
+      "loss": 4.1903,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9317851959361393e-06,
+      "loss": 4.2105,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.917271407837446e-06,
+      "loss": 4.1059,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9027576197387517e-06,
+      "loss": 3.9467,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8882438316400583e-06,
+      "loss": 4.1693,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8737300435413645e-06,
+      "loss": 4.2874,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8592162554426707e-06,
+      "loss": 4.6259,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8447024673439773e-06,
+      "loss": 4.2964,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 4.401,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8156748911465897e-06,
+      "loss": 4.3838,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.801161103047896e-06,
+      "loss": 4.5355,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.157743453979492,
+      "eval_runtime": 781.6139,
+      "eval_samples_per_second": 3.38,
+      "eval_steps_per_second": 0.283,
+      "eval_wer": 1.9685640618802063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7866473149492017e-06,
+      "loss": 4.254,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7721335268505083e-06,
+      "loss": 4.1922,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7576197387518145e-06,
+      "loss": 4.0897,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7431059506531207e-06,
+      "loss": 4.0624,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.728592162554427e-06,
+      "loss": 4.2441,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.714078374455733e-06,
+      "loss": 4.0828,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6995645863570397e-06,
+      "loss": 4.0739,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.685050798258346e-06,
+      "loss": 4.0826,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6705370101596516e-06,
+      "loss": 4.1005,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6560232220609582e-06,
+      "loss": 4.123,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 4.0644,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.62699564586357e-06,
+      "loss": 4.242,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 4.0057,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.597968069666183e-06,
+      "loss": 4.1296,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5834542815674896e-06,
+      "loss": 4.1317,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.568940493468796e-06,
+      "loss": 4.2157,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5544267053701016e-06,
+      "loss": 4.0638,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.539912917271408e-06,
+      "loss": 4.1692,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5253991291727144e-06,
+      "loss": 4.2603,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.51088534107402e-06,
+      "loss": 3.9801,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4963715529753268e-06,
+      "loss": 4.0573,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.481857764876633e-06,
+      "loss": 4.2687,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.467343976777939e-06,
+      "loss": 4.211,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 3.9504,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.438316400580552e-06,
+      "loss": 4.1229,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4238026124818577e-06,
+      "loss": 4.051,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4092888243831643e-06,
+      "loss": 4.1387,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3947750362844705e-06,
+      "loss": 4.0488,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3802612481857767e-06,
+      "loss": 3.9618,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.365747460087083e-06,
+      "loss": 3.9251,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.351233671988389e-06,
+      "loss": 4.1369,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3367198838896953e-06,
+      "loss": 4.0354,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.322206095791002e-06,
+      "loss": 4.1082,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 4.0638,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.293178519593614e-06,
+      "loss": 3.9715,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2786647314949205e-06,
+      "loss": 3.9709,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 3.935,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.249637155297533e-06,
+      "loss": 4.3424,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.235123367198839e-06,
+      "loss": 4.1501,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2206095791001452e-06,
+      "loss": 4.1094,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2060957910014514e-06,
+      "loss": 4.2221,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.191582002902758e-06,
+      "loss": 4.0662,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 4.0988,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1625544267053704e-06,
+      "loss": 4.2404,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1480406386066766e-06,
+      "loss": 4.2359,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1335268505079828e-06,
+      "loss": 4.1153,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.119013062409289e-06,
+      "loss": 4.133,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.104499274310595e-06,
+      "loss": 4.3974,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0899854862119014e-06,
+      "loss": 4.6807,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.075471698113208e-06,
+      "loss": 4.3677,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0609579100145137e-06,
+      "loss": 4.0933,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0464441219158204e-06,
+      "loss": 4.0579,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0319303338171265e-06,
+      "loss": 4.0224,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0174165457184327e-06,
+      "loss": 4.1403,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.002902757619739e-06,
+      "loss": 4.2036,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.988388969521045e-06,
+      "loss": 4.2683,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9738751814223513e-06,
+      "loss": 4.2294,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.959361393323658e-06,
+      "loss": 3.9501,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9448476052249637e-06,
+      "loss": 4.2079,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.93033381712627e-06,
+      "loss": 4.0999,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9158200290275765e-06,
+      "loss": 4.2559,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9013062409288827e-06,
+      "loss": 4.0269,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 4.1607,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.872278664731495e-06,
+      "loss": 4.0338,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8577648766328013e-06,
+      "loss": 3.9439,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8432510885341077e-06,
+      "loss": 4.0357,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8287373004354138e-06,
+      "loss": 4.097,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.81422351233672e-06,
+      "loss": 4.0727,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7997097242380262e-06,
+      "loss": 4.1095,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7851959361393326e-06,
+      "loss": 4.2551,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7706821480406388e-06,
+      "loss": 4.147,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7561683599419448e-06,
+      "loss": 4.099,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7416545718432512e-06,
+      "loss": 3.8733,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7271407837445576e-06,
+      "loss": 4.006,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7126269956458638e-06,
+      "loss": 3.9563,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 4.0179,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6835994194484762e-06,
+      "loss": 4.2796,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6690856313497824e-06,
+      "loss": 4.1347,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6545718432510888e-06,
+      "loss": 4.1331,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6400580551523947e-06,
+      "loss": 3.9432,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6255442670537012e-06,
+      "loss": 4.2474,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6110304789550073e-06,
+      "loss": 4.2026,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5965166908563137e-06,
+      "loss": 4.0464,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5820029027576197e-06,
+      "loss": 4.1152,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5674891146589261e-06,
+      "loss": 4.1948,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5529753265602323e-06,
+      "loss": 4.0272,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 4.2128,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5239477503628447e-06,
+      "loss": 3.9826,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 4.061,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4949201741654573e-06,
+      "loss": 4.1095,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4804063860667637e-06,
+      "loss": 4.1081,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4658925979680697e-06,
+      "loss": 4.1331,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4513788098693759e-06,
+      "loss": 4.1291,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4368650217706823e-06,
+      "loss": 4.2512,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4223512336719887e-06,
+      "loss": 4.2021,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4078374455732949e-06,
+      "loss": 4.6266,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3933236574746008e-06,
+      "loss": 4.4112,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3788098693759072e-06,
+      "loss": 4.4073,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3642960812772134e-06,
+      "loss": 4.5972,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3497822931785198e-06,
+      "loss": 4.3284,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3352685050798258e-06,
+      "loss": 4.3021,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 4.114,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 4.0569,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2917271407837448e-06,
+      "loss": 3.9537,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2772133526850508e-06,
+      "loss": 4.0962,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2626995645863572e-06,
+      "loss": 3.9386,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2481857764876634e-06,
+      "loss": 4.3002,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336719883889696e-06,
+      "loss": 4.2419,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219158200290276e-06,
+      "loss": 4.1903,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2046444121915822e-06,
+      "loss": 4.0671,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1901306240928884e-06,
+      "loss": 4.0627,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1756168359941945e-06,
+      "loss": 4.407,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.161103047895501e-06,
+      "loss": 4.1788,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146589259796807e-06,
+      "loss": 3.8379,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 4.1517,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1175616835994195e-06,
+      "loss": 4.2039,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1030478955007257e-06,
+      "loss": 4.1615,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.088534107402032e-06,
+      "loss": 3.9953,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0740203193033383e-06,
+      "loss": 3.8276,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0595065312046445e-06,
+      "loss": 4.1608,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0449927431059507e-06,
+      "loss": 4.1507,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0304789550072569e-06,
+      "loss": 4.1814,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0159651669085633e-06,
+      "loss": 4.0179,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0014513788098695e-06,
+      "loss": 4.1728,
+      "step": 1124
+    },
     {
       "epoch": 0.95,
-      "learning_rate": 8.42e-06,
-      "loss": 4.1906,
-      "step": 423
+      "learning_rate": 9.869375907111757e-07,
+      "loss": 4.3532,
+      "step": 1125
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.44e-06,
-      "loss": 4.2461,
-      "step": 424
+      "learning_rate": 9.724238026124818e-07,
+      "loss": 4.116,
+      "step": 1126
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.46e-06,
-      "loss": 4.1716,
-      "step": 425
+      "learning_rate": 9.579100145137882e-07,
+      "loss": 4.2546,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433962264150944e-07,
+      "loss": 4.0645,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.288824383164006e-07,
+      "loss": 4.1469,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.143686502177069e-07,
+      "loss": 4.2182,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.998548621190131e-07,
+      "loss": 3.9178,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.853410740203194e-07,
+      "loss": 3.9556,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.708272859216256e-07,
+      "loss": 4.2795,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.563134978229319e-07,
+      "loss": 4.1759,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.417997097242381e-07,
+      "loss": 4.3057,
+      "step": 1135
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.48e-06,
-      "loss": 4.2216,
-      "step": 426
+      "learning_rate": 8.272859216255444e-07,
+      "loss": 4.0645,
+      "step": 1136
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.5e-06,
-      "loss": 4.18,
-      "step": 427
+      "learning_rate": 8.127721335268506e-07,
+      "loss": 4.0856,
+      "step": 1137
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.52e-06,
-      "loss": 4.1531,
-      "step": 428
+      "learning_rate": 7.982583454281569e-07,
+      "loss": 4.0946,
+      "step": 1138
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.540000000000001e-06,
-      "loss": 4.104,
-      "step": 429
+      "learning_rate": 7.837445573294631e-07,
+      "loss": 4.0796,
+      "step": 1139
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.560000000000001e-06,
-      "loss": 4.1649,
-      "step": 430
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 3.9725,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 4.366,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402031930333818e-07,
+      "loss": 4.1314,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.256894049346879e-07,
+      "loss": 4.3359,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111756168359943e-07,
+      "loss": 4.3711,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966618287373004e-07,
+      "loss": 4.1217,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.821480406386067e-07,
+      "loss": 4.4173,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.676342525399129e-07,
+      "loss": 4.3912,
+      "step": 1147
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.580000000000001e-06,
-      "loss": 4.2599,
-      "step": 431
+      "learning_rate": 6.531204644412192e-07,
+      "loss": 4.2627,
+      "step": 1148
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.6e-06,
-      "loss": 4.3766,
-      "step": 432
+      "learning_rate": 6.386066763425254e-07,
+      "loss": 4.6431,
+      "step": 1149
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.62e-06,
-      "loss": 4.1677,
-      "step": 433
+      "learning_rate": 6.240928882438317e-07,
+      "loss": 4.3746,
+      "step": 1150
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.64e-06,
-      "loss": 4.1718,
-      "step": 434
+      "learning_rate": 6.09579100145138e-07,
+      "loss": 4.013,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.950653120464442e-07,
+      "loss": 3.9813,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805515239477505e-07,
+      "loss": 4.035,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 3.9639,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.515239477503629e-07,
+      "loss": 4.1003,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370101596516691e-07,
+      "loss": 4.1378,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224963715529753e-07,
+      "loss": 3.9805,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.079825834542816e-07,
+      "loss": 4.1325,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934687953555878e-07,
+      "loss": 3.993,
+      "step": 1159
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.66e-06,
-      "loss": 4.2196,
-      "step": 435
+      "learning_rate": 4.789550072568941e-07,
+      "loss": 4.2318,
+      "step": 1160
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.68e-06,
-      "loss": 4.1337,
-      "step": 436
+      "learning_rate": 4.644412191582003e-07,
+      "loss": 3.9432,
+      "step": 1161
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.700000000000001e-06,
-      "loss": 4.1335,
-      "step": 437
+      "learning_rate": 4.4992743105950656e-07,
+      "loss": 4.2337,
+      "step": 1162
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.720000000000001e-06,
-      "loss": 4.3058,
-      "step": 438
+      "learning_rate": 4.354136429608128e-07,
+      "loss": 4.1767,
+      "step": 1163
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.740000000000001e-06,
-      "loss": 4.2738,
-      "step": 439
+      "learning_rate": 4.2089985486211904e-07,
+      "loss": 4.0068,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.063860667634253e-07,
+      "loss": 3.9833,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9187227866473153e-07,
+      "loss": 4.2234,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 3.9178,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6284470246734397e-07,
+      "loss": 4.0042,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483309143686502e-07,
+      "loss": 4.0426,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3381712626995645e-07,
+      "loss": 4.1249,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193033381712627e-07,
+      "loss": 4.1383,
+      "step": 1171
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.76e-06,
-      "loss": 4.3081,
-      "step": 440
+      "learning_rate": 3.04789550072569e-07,
+      "loss": 3.9936,
+      "step": 1172
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.78e-06,
-      "loss": 4.2582,
-      "step": 441
+      "learning_rate": 2.9027576197387524e-07,
+      "loss": 4.2004,
+      "step": 1173
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.8e-06,
-      "loss": 4.4641,
-      "step": 442
+      "learning_rate": 2.757619738751814e-07,
+      "loss": 4.0239,
+      "step": 1174
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.82e-06,
-      "loss": 4.3681,
-      "step": 443
+      "learning_rate": 2.6124818577648767e-07,
+      "loss": 3.9843,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.467343976777939e-07,
+      "loss": 4.0407,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3222060957910016e-07,
+      "loss": 4.2081,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.177068214804064e-07,
+      "loss": 4.4838,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0319303338171264e-07,
+      "loss": 4.0287,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.886792452830189e-07,
+      "loss": 4.2293,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.741654571843251e-07,
+      "loss": 4.1361,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5965166908563135e-07,
+      "loss": 4.1122,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4513788098693762e-07,
+      "loss": 4.1374,
+      "step": 1183
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.84e-06,
-      "loss": 4.3675,
-      "step": 444
+      "learning_rate": 1.3062409288824383e-07,
+      "loss": 4.109,
+      "step": 1184
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.860000000000002e-06,
-      "loss": 4.6441,
-      "step": 445
+      "learning_rate": 1.1611030478955008e-07,
+      "loss": 4.1723,
+      "step": 1185
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.860000000000002e-06,
-      "loss": 4.3047,
-      "step": 446
+      "learning_rate": 1.0159651669085632e-07,
+      "loss": 4.371,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.708272859216255e-08,
+      "loss": 4.2052,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256894049346881e-08,
+      "loss": 4.2771,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.805515239477504e-08,
+      "loss": 4.391,
+      "step": 1189
     },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 4.377411050112258,
-      "train_runtime": 4925.13,
-      "train_samples_per_second": 5.794,
-      "train_steps_per_second": 0.091
+      "train_loss": 4.264789203719195,
+      "train_runtime": 6594.5505,
+      "train_samples_per_second": 4.328,
+      "train_steps_per_second": 0.18
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,