diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,5383 +2,2698 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 4.8242,
+      "learning_rate": 2e-08,
+      "loss": 4.7633,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2e-08,
-      "loss": 5.0694,
+      "learning_rate": 4e-08,
+      "loss": 4.9111,
       "step": 2
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 4e-08,
-      "loss": 4.9273,
+      "loss": 4.8796,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 6.000000000000001e-08,
-      "loss": 4.9226,
+      "loss": 4.8776,
       "step": 4
     },
     {
       "epoch": 0.01,
       "learning_rate": 8e-08,
-      "loss": 4.755,
+      "loss": 4.8182,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8e-08,
-      "loss": 4.8633,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.8851,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 4.8616,
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.8575,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 4.7782,
+      "epoch": 0.02,
+      "learning_rate": 1.4e-07,
+      "loss": 4.9088,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-07,
-      "loss": 4.9333,
+      "epoch": 0.02,
+      "learning_rate": 1.6e-07,
+      "loss": 4.9109,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6e-07,
-      "loss": 4.8321,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-07,
+      "loss": 4.7878,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 4.8966,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.9099,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 4.8762,
+      "epoch": 0.03,
+      "learning_rate": 2.2e-07,
+      "loss": 4.8879,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-07,
-      "loss": 4.8738,
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.8415,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 4.8319,
+      "epoch": 0.03,
+      "learning_rate": 2.6e-07,
+      "loss": 4.8503,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-07,
-      "loss": 4.7762,
+      "epoch": 0.03,
+      "learning_rate": 2.8e-07,
+      "loss": 4.7343,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8e-07,
-      "loss": 4.9082,
+      "epoch": 0.04,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.8936,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 4.9178,
+      "epoch": 0.04,
+      "learning_rate": 3.2e-07,
+      "loss": 4.8205,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2e-07,
-      "loss": 5.055,
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 4.9299,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 4.7725,
+      "epoch": 0.04,
+      "learning_rate": 3.6e-07,
+      "loss": 4.7331,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 4.9495,
+      "epoch": 0.04,
+      "learning_rate": 3.8e-07,
+      "loss": 4.9161,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-07,
-      "loss": 4.8793,
+      "epoch": 0.05,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.91,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.8091,
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 4.8262,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 4.8869,
+      "epoch": 0.05,
+      "learning_rate": 4.4e-07,
+      "loss": 4.9479,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-07,
-      "loss": 5.031,
+      "epoch": 0.05,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 4.942,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 4.9099,
+      "epoch": 0.06,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.7356,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 5.0558,
+      "epoch": 0.06,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 5.0038,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 4.8086,
+      "epoch": 0.06,
+      "learning_rate": 5.2e-07,
+      "loss": 4.8565,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-07,
-      "loss": 4.7263,
+      "epoch": 0.06,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7324,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-07,
-      "loss": 4.7646,
+      "epoch": 0.07,
+      "learning_rate": 5.6e-07,
+      "loss": 4.7126,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.6e-07,
-      "loss": 4.9278,
+      "epoch": 0.07,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 4.9015,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.800000000000001e-07,
-      "loss": 4.9052,
+      "epoch": 0.07,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.869,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.6965,
+      "epoch": 0.07,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 4.7292,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.200000000000001e-07,
-      "loss": 4.7982,
+      "epoch": 0.07,
+      "learning_rate": 6.4e-07,
+      "loss": 4.8362,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.4e-07,
-      "loss": 4.666,
+      "epoch": 0.08,
+      "learning_rate": 6.6e-07,
+      "loss": 4.75,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.6e-07,
-      "loss": 4.8597,
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 4.8524,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.800000000000001e-07,
-      "loss": 4.8249,
+      "epoch": 0.08,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.7962,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 4.8771,
+      "epoch": 0.08,
+      "learning_rate": 7.2e-07,
+      "loss": 4.859,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-07,
-      "loss": 4.8633,
+      "epoch": 0.09,
+      "learning_rate": 7.4e-07,
+      "loss": 4.7945,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.4e-07,
-      "loss": 5.0942,
+      "epoch": 0.09,
+      "learning_rate": 7.6e-07,
+      "loss": 4.9263,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.6e-07,
-      "loss": 4.8164,
+      "epoch": 0.09,
+      "learning_rate": 7.8e-07,
+      "loss": 4.7248,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.8e-07,
-      "loss": 4.7004,
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.803,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.9064,
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 4.9282,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 4.7133,
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 4.7007,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 4.9141,
+      "epoch": 0.1,
+      "learning_rate": 8.6e-07,
+      "loss": 4.8718,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.6e-07,
-      "loss": 4.9973,
+      "epoch": 0.1,
+      "learning_rate": 8.8e-07,
+      "loss": 5.0068,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.8e-07,
-      "loss": 5.2097,
+      "epoch": 0.1,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 5.1512,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 5.1196,
+      "epoch": 0.11,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.0423,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.4716,
+      "epoch": 0.11,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 5.3227,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.6522,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 5.4297,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 5.4503,
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 5.339,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 4.5297,
+      "epoch": 0.11,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 4.657,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.800000000000001e-07,
-      "loss": 4.673,
+      "epoch": 0.12,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.5155,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.5282,
+      "epoch": 0.12,
+      "learning_rate": 1.02e-06,
+      "loss": 4.6886,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.02e-06,
-      "loss": 4.6237,
+      "epoch": 0.12,
+      "learning_rate": 1.04e-06,
+      "loss": 4.5949,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.04e-06,
-      "loss": 4.6055,
+      "epoch": 0.12,
+      "learning_rate": 1.06e-06,
+      "loss": 4.4149,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.06e-06,
-      "loss": 4.6426,
+      "epoch": 0.13,
+      "learning_rate": 1.08e-06,
+      "loss": 4.6266,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.08e-06,
-      "loss": 4.6344,
+      "epoch": 0.13,
+      "learning_rate": 1.1e-06,
+      "loss": 4.6458,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1e-06,
-      "loss": 4.8693,
+      "epoch": 0.13,
+      "learning_rate": 1.12e-06,
+      "loss": 4.7385,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.12e-06,
-      "loss": 4.6156,
+      "epoch": 0.13,
+      "learning_rate": 1.14e-06,
+      "loss": 4.6687,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.14e-06,
-      "loss": 4.4609,
+      "epoch": 0.13,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 4.5012,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1600000000000001e-06,
-      "loss": 4.6899,
+      "epoch": 0.14,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 4.6079,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1800000000000001e-06,
-      "loss": 4.6055,
+      "epoch": 0.14,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.5349,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.5051,
+      "epoch": 0.14,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 4.5074,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 4.6539,
+      "epoch": 0.14,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 4.611,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 4.393,
+      "epoch": 0.15,
+      "learning_rate": 1.26e-06,
+      "loss": 4.565,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.26e-06,
-      "loss": 4.6817,
+      "epoch": 0.15,
+      "learning_rate": 1.28e-06,
+      "loss": 4.4438,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.28e-06,
-      "loss": 4.4762,
+      "epoch": 0.15,
+      "learning_rate": 1.3e-06,
+      "loss": 4.6036,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3e-06,
-      "loss": 4.5006,
+      "epoch": 0.15,
+      "learning_rate": 1.32e-06,
+      "loss": 4.4427,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.32e-06,
-      "loss": 4.4399,
+      "epoch": 0.15,
+      "learning_rate": 1.34e-06,
+      "loss": 4.5095,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.34e-06,
-      "loss": 4.5411,
+      "epoch": 0.16,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 4.4858,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3600000000000001e-06,
-      "loss": 4.7146,
+      "epoch": 0.16,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 4.618,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 4.6193,
+      "epoch": 0.16,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.6007,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.5544,
+      "epoch": 0.16,
+      "learning_rate": 1.42e-06,
+      "loss": 4.453,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.42e-06,
-      "loss": 4.6302,
+      "epoch": 0.17,
+      "learning_rate": 1.44e-06,
+      "loss": 4.519,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.44e-06,
-      "loss": 4.3761,
+      "epoch": 0.17,
+      "learning_rate": 1.46e-06,
+      "loss": 4.5242,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.46e-06,
-      "loss": 4.6907,
+      "epoch": 0.17,
+      "learning_rate": 1.48e-06,
+      "loss": 4.4702,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.48e-06,
-      "loss": 4.5436,
+      "epoch": 0.17,
+      "learning_rate": 1.5e-06,
+      "loss": 4.5255,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5e-06,
-      "loss": 4.4896,
+      "epoch": 0.17,
+      "learning_rate": 1.52e-06,
+      "loss": 4.4652,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.52e-06,
-      "loss": 4.3514,
+      "epoch": 0.18,
+      "learning_rate": 1.54e-06,
+      "loss": 4.4498,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.54e-06,
-      "loss": 4.5454,
+      "epoch": 0.18,
+      "learning_rate": 1.56e-06,
+      "loss": 4.5705,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.56e-06,
-      "loss": 4.6405,
+      "epoch": 0.18,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 4.5202,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.5800000000000001e-06,
-      "loss": 4.4412,
+      "epoch": 0.18,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.4593,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.6026,
+      "epoch": 0.19,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 4.5733,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.6200000000000002e-06,
-      "loss": 4.5577,
+      "epoch": 0.19,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 4.45,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 4.5492,
+      "epoch": 0.19,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 4.4868,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 4.4596,
+      "epoch": 0.19,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 4.48,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6800000000000002e-06,
-      "loss": 4.4947,
+      "epoch": 0.2,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.5859,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 4.5046,
+      "epoch": 0.2,
+      "learning_rate": 1.72e-06,
+      "loss": 4.4624,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.72e-06,
-      "loss": 4.5577,
+      "epoch": 0.2,
+      "learning_rate": 1.74e-06,
+      "loss": 4.6002,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.74e-06,
-      "loss": 4.4095,
+      "epoch": 0.2,
+      "learning_rate": 1.76e-06,
+      "loss": 4.5497,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.76e-06,
-      "loss": 4.6886,
+      "epoch": 0.2,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 4.4896,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 4.6398,
+      "epoch": 0.21,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.7272,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.8000000000000001e-06,
-      "loss": 4.3426,
+      "epoch": 0.21,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 4.5785,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 4.5306,
+      "epoch": 0.21,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 4.5895,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 4.803,
+      "epoch": 0.21,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 4.5834,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8600000000000002e-06,
-      "loss": 4.7855,
+      "epoch": 0.22,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 4.7481,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 4.5809,
+      "epoch": 0.22,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.7947,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9000000000000002e-06,
-      "loss": 4.8796,
+      "epoch": 0.22,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 4.567,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.6557,
+      "epoch": 0.22,
+      "learning_rate": 1.94e-06,
+      "loss": 4.8742,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.94e-06,
-      "loss": 4.7918,
+      "epoch": 0.22,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 4.6728,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 4.3464,
+      "epoch": 0.23,
+      "learning_rate": 1.98e-06,
+      "loss": 4.4148,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.98e-06,
-      "loss": 4.3529,
+      "epoch": 0.23,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.3493,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.5084,
+      "epoch": 0.23,
+      "learning_rate": 2.02e-06,
+      "loss": 4.2165,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.02e-06,
-      "loss": 4.4042,
+      "epoch": 0.23,
+      "learning_rate": 2.04e-06,
+      "loss": 4.285,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.04e-06,
-      "loss": 4.2093,
+      "epoch": 0.24,
+      "learning_rate": 2.06e-06,
+      "loss": 4.2356,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.06e-06,
-      "loss": 4.4784,
+      "epoch": 0.24,
+      "learning_rate": 2.08e-06,
+      "loss": 4.278,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.08e-06,
-      "loss": 4.4191,
+      "epoch": 0.24,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.38,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 4.5951,
+      "epoch": 0.24,
+      "learning_rate": 2.12e-06,
+      "loss": 4.2958,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.12e-06,
-      "loss": 4.1595,
+      "epoch": 0.24,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 4.2908,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.1400000000000003e-06,
-      "loss": 4.3082,
+      "epoch": 0.25,
+      "learning_rate": 2.16e-06,
+      "loss": 4.3311,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.16e-06,
-      "loss": 4.5168,
+      "epoch": 0.25,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 4.3749,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 4.362,
+      "epoch": 0.25,
+      "learning_rate": 2.2e-06,
+      "loss": 4.3236,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2e-06,
-      "loss": 4.3284,
+      "epoch": 0.25,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 4.348,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 4.2495,
+      "epoch": 0.26,
+      "learning_rate": 2.24e-06,
+      "loss": 4.3211,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.24e-06,
-      "loss": 4.329,
+      "epoch": 0.26,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 4.3697,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.2600000000000004e-06,
-      "loss": 4.1321,
+      "epoch": 0.26,
+      "learning_rate": 2.28e-06,
+      "loss": 4.4212,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.28e-06,
-      "loss": 4.4528,
+      "epoch": 0.26,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 4.3025,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3000000000000004e-06,
-      "loss": 4.2502,
+      "epoch": 0.26,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 4.4696,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 4.4529,
+      "epoch": 0.27,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 4.3551,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 4.4849,
+      "epoch": 0.27,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 4.3851,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.3600000000000003e-06,
-      "loss": 4.4343,
+      "epoch": 0.27,
+      "learning_rate": 2.38e-06,
+      "loss": 4.2232,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.38e-06,
-      "loss": 4.3086,
+      "epoch": 0.27,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.3045,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.2446,
+      "epoch": 0.28,
+      "learning_rate": 2.42e-06,
+      "loss": 4.33,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.42e-06,
-      "loss": 4.4509,
+      "epoch": 0.28,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 4.3141,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 4.4062,
+      "epoch": 0.28,
+      "learning_rate": 2.46e-06,
+      "loss": 4.4295,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.46e-06,
-      "loss": 4.4681,
+      "epoch": 0.28,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 4.44,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.4800000000000004e-06,
-      "loss": 4.5481,
+      "epoch": 0.28,
+      "learning_rate": 2.5e-06,
+      "loss": 4.3421,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3067,
+      "epoch": 0.29,
+      "learning_rate": 2.52e-06,
+      "loss": 4.3164,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.52e-06,
-      "loss": 4.2584,
+      "epoch": 0.29,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 4.2223,
       "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5400000000000002e-06,
-      "loss": 4.4602,
+      "epoch": 0.29,
+      "learning_rate": 2.56e-06,
+      "loss": 4.3852,
       "step": 130
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.56e-06,
-      "loss": 4.3518,
+      "epoch": 0.29,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 4.2854,
       "step": 131
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 4.4616,
+      "epoch": 0.3,
+      "learning_rate": 2.6e-06,
+      "loss": 4.3269,
       "step": 132
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6e-06,
-      "loss": 4.3817,
+      "epoch": 0.3,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 4.2724,
       "step": 133
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 4.4235,
+      "epoch": 0.3,
+      "learning_rate": 2.64e-06,
+      "loss": 4.2822,
       "step": 134
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.64e-06,
-      "loss": 4.2847,
+      "epoch": 0.3,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 4.4477,
       "step": 135
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.6600000000000004e-06,
-      "loss": 4.3116,
+      "epoch": 0.3,
+      "learning_rate": 2.68e-06,
+      "loss": 4.3023,
       "step": 136
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.68e-06,
-      "loss": 4.6905,
+      "epoch": 0.31,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 4.4109,
       "step": 137
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.7000000000000004e-06,
-      "loss": 4.4184,
+      "epoch": 0.31,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 4.4397,
       "step": 138
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7200000000000002e-06,
-      "loss": 4.463,
+      "epoch": 0.31,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 4.4502,
       "step": 139
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 4.2542,
+      "epoch": 0.31,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 4.4306,
       "step": 140
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 4.5367,
+      "epoch": 0.32,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 4.5436,
       "step": 141
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.7800000000000005e-06,
-      "loss": 4.6165,
+      "epoch": 0.32,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.4715,
       "step": 142
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.5384,
+      "epoch": 0.32,
+      "learning_rate": 2.82e-06,
+      "loss": 4.3932,
       "step": 143
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.82e-06,
-      "loss": 4.4601,
+      "epoch": 0.32,
+      "learning_rate": 2.84e-06,
+      "loss": 4.4735,
       "step": 144
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.84e-06,
-      "loss": 4.5431,
+      "epoch": 0.33,
+      "learning_rate": 2.86e-06,
+      "loss": 4.3691,
       "step": 145
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.86e-06,
-      "loss": 4.5828,
+      "epoch": 0.33,
+      "learning_rate": 2.88e-06,
+      "loss": 4.4599,
       "step": 146
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.88e-06,
-      "loss": 4.6763,
+      "epoch": 0.33,
+      "learning_rate": 2.9e-06,
+      "loss": 4.7192,
       "step": 147
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9e-06,
-      "loss": 4.3709,
+      "epoch": 0.33,
+      "learning_rate": 2.92e-06,
+      "loss": 4.8634,
       "step": 148
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.92e-06,
-      "loss": 4.6528,
+      "epoch": 0.33,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.5975,
       "step": 149
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 4.2986,
+      "epoch": 0.34,
+      "learning_rate": 2.96e-06,
+      "loss": 4.7077,
       "step": 150
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.96e-06,
-      "loss": 4.4293,
+      "epoch": 0.34,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 4.2145,
       "step": 151
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 4.2866,
+      "epoch": 0.34,
+      "learning_rate": 3e-06,
+      "loss": 4.2845,
       "step": 152
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3e-06,
-      "loss": 4.3511,
+      "epoch": 0.34,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 4.2969,
       "step": 153
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 4.2369,
+      "epoch": 0.35,
+      "learning_rate": 3.04e-06,
+      "loss": 4.2126,
       "step": 154
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.04e-06,
-      "loss": 4.1249,
+      "epoch": 0.35,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 4.3056,
       "step": 155
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.0600000000000003e-06,
-      "loss": 4.2803,
+      "epoch": 0.35,
+      "learning_rate": 3.08e-06,
+      "loss": 4.172,
       "step": 156
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.08e-06,
-      "loss": 4.4222,
+      "epoch": 0.35,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.2029,
       "step": 157
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1000000000000004e-06,
-      "loss": 4.5166,
+      "epoch": 0.35,
+      "learning_rate": 3.12e-06,
+      "loss": 4.2939,
       "step": 158
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.12e-06,
-      "loss": 4.5553,
+      "epoch": 0.36,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 4.3015,
       "step": 159
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1400000000000004e-06,
-      "loss": 4.2464,
+      "epoch": 0.36,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 4.3041,
       "step": 160
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1600000000000002e-06,
-      "loss": 4.2719,
+      "epoch": 0.36,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 4.3576,
       "step": 161
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 4.3566,
+      "epoch": 0.36,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.3136,
       "step": 162
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.2978,
+      "epoch": 0.37,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 4.2696,
       "step": 163
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2200000000000005e-06,
-      "loss": 4.434,
+      "epoch": 0.37,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 4.3336,
       "step": 164
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 4.3524,
+      "epoch": 0.37,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 4.1924,
       "step": 165
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.2600000000000006e-06,
-      "loss": 4.3696,
+      "epoch": 0.37,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 4.2559,
       "step": 166
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 4.3229,
+      "epoch": 0.37,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 4.1719,
       "step": 167
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3000000000000006e-06,
-      "loss": 4.1114,
+      "epoch": 0.38,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 4.2586,
       "step": 168
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 4.0973,
+      "epoch": 0.38,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 4.3229,
       "step": 169
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 4.2897,
+      "epoch": 0.38,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 4.3361,
       "step": 170
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3600000000000004e-06,
-      "loss": 4.4917,
+      "epoch": 0.38,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 4.3238,
       "step": 171
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 4.396,
+      "epoch": 0.39,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.3695,
       "step": 172
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.3476,
+      "epoch": 0.39,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 4.1802,
       "step": 173
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 4.1774,
+      "epoch": 0.39,
+      "learning_rate": 3.44e-06,
+      "loss": 4.3182,
       "step": 174
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.44e-06,
-      "loss": 4.3004,
+      "epoch": 0.39,
+      "learning_rate": 3.46e-06,
+      "loss": 4.3777,
       "step": 175
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.46e-06,
-      "loss": 4.1297,
+      "epoch": 0.39,
+      "learning_rate": 3.48e-06,
+      "loss": 4.2564,
       "step": 176
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.48e-06,
-      "loss": 4.1742,
+      "epoch": 0.4,
+      "learning_rate": 3.5e-06,
+      "loss": 4.2225,
       "step": 177
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.5e-06,
-      "loss": 4.2256,
+      "epoch": 0.4,
+      "learning_rate": 3.52e-06,
+      "loss": 4.3079,
       "step": 178
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.52e-06,
-      "loss": 4.3416,
+      "epoch": 0.4,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3581,
       "step": 179
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.54e-06,
-      "loss": 4.3123,
+      "epoch": 0.4,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 4.3614,
       "step": 180
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 4.3697,
+      "epoch": 0.41,
+      "learning_rate": 3.58e-06,
+      "loss": 4.2853,
       "step": 181
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.58e-06,
-      "loss": 4.1602,
+      "epoch": 0.41,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 4.3422,
       "step": 182
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 4.3987,
+      "epoch": 0.41,
+      "learning_rate": 3.62e-06,
+      "loss": 4.348,
       "step": 183
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.62e-06,
-      "loss": 4.3355,
+      "epoch": 0.41,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 4.3491,
       "step": 184
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 4.3648,
+      "epoch": 0.41,
+      "learning_rate": 3.66e-06,
+      "loss": 4.3238,
       "step": 185
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.66e-06,
-      "loss": 4.258,
+      "epoch": 0.42,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 4.1929,
       "step": 186
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.6800000000000003e-06,
-      "loss": 4.4511,
+      "epoch": 0.42,
+      "learning_rate": 3.7e-06,
+      "loss": 4.21,
       "step": 187
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7e-06,
-      "loss": 4.1707,
+      "epoch": 0.42,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 4.164,
       "step": 188
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7200000000000004e-06,
-      "loss": 4.2588,
+      "epoch": 0.42,
+      "learning_rate": 3.74e-06,
+      "loss": 4.2805,
       "step": 189
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.74e-06,
-      "loss": 4.4404,
+      "epoch": 0.43,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 4.405,
       "step": 190
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 4.4161,
+      "epoch": 0.43,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.3344,
       "step": 191
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 4.3722,
+      "epoch": 0.43,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.2669,
       "step": 192
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8000000000000005e-06,
-      "loss": 4.5612,
+      "epoch": 0.43,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 4.2899,
       "step": 193
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.820000000000001e-06,
-      "loss": 4.2952,
+      "epoch": 0.43,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.3873,
       "step": 194
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 4.3148,
+      "epoch": 0.44,
+      "learning_rate": 3.86e-06,
+      "loss": 4.4804,
       "step": 195
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.86e-06,
-      "loss": 4.5931,
+      "epoch": 0.44,
+      "learning_rate": 3.88e-06,
+      "loss": 4.4503,
       "step": 196
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.88e-06,
-      "loss": 4.5598,
+      "epoch": 0.44,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 4.3873,
       "step": 197
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.900000000000001e-06,
-      "loss": 4.536,
+      "epoch": 0.44,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 4.6998,
       "step": 198
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 4.6728,
+      "epoch": 0.45,
+      "learning_rate": 3.94e-06,
+      "loss": 4.6039,
       "step": 199
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.94e-06,
-      "loss": 4.7729,
+      "epoch": 0.45,
+      "learning_rate": 3.96e-06,
+      "loss": 4.5695,
       "step": 200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.96e-06,
-      "loss": 4.1568,
+      "epoch": 0.45,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 4.3181,
       "step": 201
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 4.2509,
+      "epoch": 0.45,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.2565,
       "step": 202
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.1336,
+      "epoch": 0.46,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.2566,
       "step": 203
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 4.1179,
+      "epoch": 0.46,
+      "learning_rate": 4.04e-06,
+      "loss": 4.2488,
       "step": 204
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.04e-06,
-      "loss": 4.0396,
+      "epoch": 0.46,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 4.3213,
       "step": 205
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.060000000000001e-06,
-      "loss": 4.1159,
+      "epoch": 0.46,
+      "learning_rate": 4.08e-06,
+      "loss": 4.3263,
       "step": 206
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.08e-06,
-      "loss": 4.2318,
+      "epoch": 0.46,
+      "learning_rate": 4.1e-06,
+      "loss": 4.1808,
       "step": 207
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.1e-06,
-      "loss": 4.0496,
+      "epoch": 0.47,
+      "learning_rate": 4.12e-06,
+      "loss": 4.2721,
       "step": 208
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.12e-06,
-      "loss": 4.2799,
+      "epoch": 0.47,
+      "learning_rate": 4.14e-06,
+      "loss": 4.3141,
       "step": 209
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.14e-06,
-      "loss": 4.23,
+      "epoch": 0.47,
+      "learning_rate": 4.16e-06,
+      "loss": 4.285,
       "step": 210
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.16e-06,
-      "loss": 4.1737,
+      "epoch": 0.47,
+      "learning_rate": 4.18e-06,
+      "loss": 4.1726,
       "step": 211
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.18e-06,
-      "loss": 4.2731,
+      "epoch": 0.48,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.2216,
       "step": 212
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.2519,
+      "epoch": 0.48,
+      "learning_rate": 4.22e-06,
+      "loss": 4.1285,
       "step": 213
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.22e-06,
-      "loss": 4.3022,
+      "epoch": 0.48,
+      "learning_rate": 4.24e-06,
+      "loss": 4.2122,
       "step": 214
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.24e-06,
-      "loss": 4.1748,
+      "epoch": 0.48,
+      "learning_rate": 4.26e-06,
+      "loss": 4.2578,
       "step": 215
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.26e-06,
-      "loss": 4.307,
+      "epoch": 0.48,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 4.2336,
       "step": 216
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2800000000000005e-06,
-      "loss": 4.2685,
+      "epoch": 0.49,
+      "learning_rate": 4.3e-06,
+      "loss": 4.2851,
       "step": 217
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.3e-06,
-      "loss": 4.1908,
+      "epoch": 0.49,
+      "learning_rate": 4.32e-06,
+      "loss": 4.2353,
       "step": 218
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.32e-06,
-      "loss": 4.279,
+      "epoch": 0.49,
+      "learning_rate": 4.34e-06,
+      "loss": 4.1564,
       "step": 219
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.34e-06,
-      "loss": 4.2546,
+      "epoch": 0.49,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 4.2741,
       "step": 220
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 4.2365,
+      "epoch": 0.5,
+      "learning_rate": 4.38e-06,
+      "loss": 4.2666,
       "step": 221
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.38e-06,
-      "loss": 4.2913,
+      "epoch": 0.5,
+      "learning_rate": 4.4e-06,
+      "loss": 4.2329,
       "step": 222
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4e-06,
-      "loss": 4.3505,
+      "epoch": 0.5,
+      "learning_rate": 4.42e-06,
+      "loss": 4.2691,
       "step": 223
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.42e-06,
-      "loss": 4.2653,
+      "epoch": 0.5,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 4.1904,
       "step": 224
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 4.34,
+      "epoch": 0.5,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 4.3606,
       "step": 225
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 4.509,
+      "epoch": 0.51,
+      "learning_rate": 4.48e-06,
+      "loss": 4.2548,
       "step": 226
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.48e-06,
-      "loss": 4.2533,
+      "epoch": 0.51,
+      "learning_rate": 4.5e-06,
+      "loss": 4.3776,
       "step": 227
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.5e-06,
-      "loss": 4.1684,
+      "epoch": 0.51,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 4.3096,
       "step": 228
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.520000000000001e-06,
-      "loss": 4.1599,
+      "epoch": 0.51,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 4.1924,
       "step": 229
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 4.253,
+      "epoch": 0.52,
+      "learning_rate": 4.56e-06,
+      "loss": 4.2802,
       "step": 230
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.56e-06,
-      "loss": 4.2055,
+      "epoch": 0.52,
+      "learning_rate": 4.58e-06,
+      "loss": 4.3578,
       "step": 231
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.58e-06,
-      "loss": 4.347,
+      "epoch": 0.52,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.1489,
       "step": 232
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 4.2403,
+      "epoch": 0.52,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 4.1684,
       "step": 233
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.620000000000001e-06,
-      "loss": 4.3107,
+      "epoch": 0.52,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 4.2505,
       "step": 234
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.6400000000000005e-06,
-      "loss": 4.3986,
+      "epoch": 0.53,
+      "learning_rate": 4.66e-06,
+      "loss": 4.3324,
       "step": 235
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.66e-06,
-      "loss": 4.2434,
+      "epoch": 0.53,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 4.2627,
       "step": 236
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.680000000000001e-06,
-      "loss": 4.282,
+      "epoch": 0.53,
+      "learning_rate": 4.7e-06,
+      "loss": 4.304,
       "step": 237
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.7e-06,
-      "loss": 4.4659,
+      "epoch": 0.53,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 4.2457,
       "step": 238
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.7200000000000005e-06,
-      "loss": 4.2934,
+      "epoch": 0.54,
+      "learning_rate": 4.74e-06,
+      "loss": 4.1868,
       "step": 239
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.74e-06,
-      "loss": 4.3816,
+      "epoch": 0.54,
+      "learning_rate": 4.76e-06,
+      "loss": 4.3028,
       "step": 240
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.76e-06,
-      "loss": 4.3785,
+      "epoch": 0.54,
+      "learning_rate": 4.78e-06,
+      "loss": 4.2704,
       "step": 241
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.78e-06,
-      "loss": 4.5443,
+      "epoch": 0.54,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.3031,
       "step": 242
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.3825,
+      "epoch": 0.54,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 4.5193,
       "step": 243
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 4.3406,
+      "epoch": 0.55,
+      "learning_rate": 4.84e-06,
+      "loss": 4.3524,
       "step": 244
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.84e-06,
-      "loss": 4.5317,
+      "epoch": 0.55,
+      "learning_rate": 4.86e-06,
+      "loss": 4.3437,
       "step": 245
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.86e-06,
-      "loss": 4.3991,
+      "epoch": 0.55,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 4.2623,
       "step": 246
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 4.7736,
+      "epoch": 0.55,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.1807,
       "step": 247
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 4.7937,
+      "epoch": 0.56,
+      "learning_rate": 4.92e-06,
+      "loss": 4.5587,
       "step": 248
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.92e-06,
-      "loss": 4.617,
+      "epoch": 0.56,
+      "learning_rate": 4.94e-06,
+      "loss": 4.6662,
       "step": 249
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.94e-06,
-      "loss": 4.2682,
+      "epoch": 0.56,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 4.6559,
       "step": 250
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 4.4206,
+      "epoch": 0.56,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2338,
       "step": 251
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 4.1331,
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 4.2654,
       "step": 252
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5e-06,
-      "loss": 4.1561,
+      "epoch": 0.57,
+      "learning_rate": 5.02e-06,
+      "loss": 4.311,
       "step": 253
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 5.02e-06,
-      "loss": 4.3146,
+      "epoch": 0.57,
+      "learning_rate": 5.04e-06,
+      "loss": 4.1525,
       "step": 254
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.04e-06,
-      "loss": 4.1541,
+      "epoch": 0.57,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 4.2531,
       "step": 255
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.060000000000001e-06,
-      "loss": 4.1555,
+      "epoch": 0.57,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 4.223,
       "step": 256
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.0800000000000005e-06,
-      "loss": 4.3562,
+      "epoch": 0.58,
+      "learning_rate": 5.1e-06,
+      "loss": 4.1962,
       "step": 257
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1e-06,
-      "loss": 4.1731,
+      "epoch": 0.58,
+      "learning_rate": 5.12e-06,
+      "loss": 4.2138,
       "step": 258
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.12e-06,
-      "loss": 4.1383,
+      "epoch": 0.58,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 4.2155,
       "step": 259
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 4.2964,
+      "epoch": 0.58,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 4.1582,
       "step": 260
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 4.3322,
+      "epoch": 0.59,
+      "learning_rate": 5.18e-06,
+      "loss": 4.2736,
       "step": 261
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.18e-06,
-      "loss": 4.0483,
+      "epoch": 0.59,
+      "learning_rate": 5.2e-06,
+      "loss": 4.2103,
       "step": 262
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 5.2e-06,
-      "loss": 4.1846,
+      "epoch": 0.59,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 4.3473,
       "step": 263
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 4.1761,
+      "epoch": 0.59,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 4.2133,
       "step": 264
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 4.3097,
+      "epoch": 0.59,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 4.2476,
       "step": 265
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 4.3513,
+      "epoch": 0.6,
+      "learning_rate": 5.28e-06,
+      "loss": 4.2244,
       "step": 266
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.28e-06,
-      "loss": 4.1831,
+      "epoch": 0.6,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 4.2517,
       "step": 267
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.300000000000001e-06,
-      "loss": 4.516,
+      "epoch": 0.6,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 4.2084,
       "step": 268
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 4.2319,
+      "epoch": 0.6,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 4.2887,
       "step": 269
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 4.386,
+      "epoch": 0.61,
+      "learning_rate": 5.36e-06,
+      "loss": 4.1665,
       "step": 270
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.36e-06,
-      "loss": 4.0012,
+      "epoch": 0.61,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 4.0903,
       "step": 271
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 4.1407,
+      "epoch": 0.61,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.2757,
       "step": 272
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.400000000000001e-06,
-      "loss": 4.1697,
+      "epoch": 0.61,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 4.0794,
       "step": 273
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 4.096,
+      "epoch": 0.61,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 4.1791,
       "step": 274
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.4400000000000004e-06,
-      "loss": 4.3577,
+      "epoch": 0.62,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 4.2729,
       "step": 275
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 4.2639,
+      "epoch": 0.62,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 4.2107,
       "step": 276
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 4.2025,
+      "epoch": 0.62,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.1623,
       "step": 277
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 4.1926,
+      "epoch": 0.62,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 4.2513,
       "step": 278
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 4.1823,
+      "epoch": 0.63,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 4.235,
       "step": 279
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 5.540000000000001e-06,
-      "loss": 4.3278,
+      "epoch": 0.63,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 4.3647,
       "step": 280
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.560000000000001e-06,
-      "loss": 4.1204,
+      "epoch": 0.63,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 4.2633,
       "step": 281
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 4.2377,
+      "epoch": 0.63,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.12,
       "step": 282
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.0139,
+      "epoch": 0.63,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 4.1451,
       "step": 283
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.620000000000001e-06,
-      "loss": 4.2235,
+      "epoch": 0.64,
+      "learning_rate": 5.64e-06,
+      "loss": 4.2252,
       "step": 284
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.64e-06,
-      "loss": 4.3209,
+      "epoch": 0.64,
+      "learning_rate": 5.66e-06,
+      "loss": 4.3205,
       "step": 285
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.66e-06,
-      "loss": 4.2323,
+      "epoch": 0.64,
+      "learning_rate": 5.68e-06,
+      "loss": 4.2208,
       "step": 286
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.68e-06,
-      "loss": 4.3746,
+      "epoch": 0.64,
+      "learning_rate": 5.7e-06,
+      "loss": 4.3254,
       "step": 287
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.7e-06,
-      "loss": 4.332,
+      "epoch": 0.65,
+      "learning_rate": 5.72e-06,
+      "loss": 4.2817,
       "step": 288
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 5.72e-06,
-      "loss": 4.3801,
+      "epoch": 0.65,
+      "learning_rate": 5.74e-06,
+      "loss": 4.192,
       "step": 289
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.74e-06,
-      "loss": 4.3752,
+      "epoch": 0.65,
+      "learning_rate": 5.76e-06,
+      "loss": 4.2462,
       "step": 290
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.76e-06,
-      "loss": 4.4016,
+      "epoch": 0.65,
+      "learning_rate": 5.78e-06,
+      "loss": 4.4973,
       "step": 291
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.78e-06,
-      "loss": 4.2473,
+      "epoch": 0.65,
+      "learning_rate": 5.8e-06,
+      "loss": 4.3751,
       "step": 292
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8e-06,
-      "loss": 4.3151,
+      "epoch": 0.66,
+      "learning_rate": 5.82e-06,
+      "loss": 4.441,
       "step": 293
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.82e-06,
-      "loss": 4.3135,
+      "epoch": 0.66,
+      "learning_rate": 5.84e-06,
+      "loss": 4.4067,
       "step": 294
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.84e-06,
-      "loss": 4.2031,
+      "epoch": 0.66,
+      "learning_rate": 5.86e-06,
+      "loss": 4.3823,
       "step": 295
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.86e-06,
-      "loss": 4.4624,
+      "epoch": 0.66,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.3868,
       "step": 296
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 4.6031,
+      "epoch": 0.67,
+      "learning_rate": 5.9e-06,
+      "loss": 4.2941,
       "step": 297
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 5.9e-06,
-      "loss": 4.3948,
+      "epoch": 0.67,
+      "learning_rate": 5.92e-06,
+      "loss": 4.6034,
       "step": 298
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.92e-06,
-      "loss": 4.8168,
+      "epoch": 0.67,
+      "learning_rate": 5.94e-06,
+      "loss": 4.37,
       "step": 299
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.94e-06,
-      "loss": 4.816,
+      "epoch": 0.67,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 4.7808,
       "step": 300
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 4.0983,
+      "epoch": 0.67,
+      "learning_rate": 5.98e-06,
+      "loss": 4.268,
       "step": 301
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 5.98e-06,
-      "loss": 4.2092,
+      "epoch": 0.68,
+      "learning_rate": 6e-06,
+      "loss": 4.2326,
       "step": 302
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6e-06,
-      "loss": 4.0984,
+      "epoch": 0.68,
+      "learning_rate": 6.02e-06,
+      "loss": 4.2418,
       "step": 303
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.02e-06,
-      "loss": 4.2683,
+      "epoch": 0.68,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 4.2376,
       "step": 304
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.040000000000001e-06,
-      "loss": 4.0981,
+      "epoch": 0.68,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.2091,
       "step": 305
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 4.1561,
+      "epoch": 0.69,
+      "learning_rate": 6.08e-06,
+      "loss": 4.2111,
       "step": 306
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.08e-06,
-      "loss": 4.1834,
+      "epoch": 0.69,
+      "learning_rate": 6.1e-06,
+      "loss": 4.2004,
       "step": 307
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.1e-06,
-      "loss": 4.2697,
+      "epoch": 0.69,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 4.1965,
       "step": 308
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.120000000000001e-06,
-      "loss": 4.1863,
+      "epoch": 0.69,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 4.1974,
       "step": 309
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.1400000000000005e-06,
-      "loss": 4.3187,
+      "epoch": 0.7,
+      "learning_rate": 6.16e-06,
+      "loss": 4.2748,
       "step": 310
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.16e-06,
-      "loss": 4.3841,
-      "step": 311
+      "epoch": 0.7,
+      "learning_rate": 6.18e-06,
+      "loss": 4.0933,
+      "step": 311
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.18e-06,
-      "loss": 4.3403,
+      "epoch": 0.7,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.2521,
       "step": 312
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.200000000000001e-06,
-      "loss": 4.2338,
+      "epoch": 0.7,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 4.2651,
       "step": 313
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.220000000000001e-06,
-      "loss": 4.2755,
+      "epoch": 0.7,
+      "learning_rate": 6.24e-06,
+      "loss": 4.182,
       "step": 314
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.24e-06,
-      "loss": 4.1207,
+      "epoch": 0.71,
+      "learning_rate": 6.26e-06,
+      "loss": 4.2096,
       "step": 315
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 6.26e-06,
-      "loss": 4.2767,
+      "epoch": 0.71,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 4.1797,
       "step": 316
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.280000000000001e-06,
-      "loss": 4.127,
+      "epoch": 0.71,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.1667,
       "step": 317
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 4.1406,
+      "epoch": 0.71,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 4.2119,
       "step": 318
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.3200000000000005e-06,
-      "loss": 4.2472,
+      "epoch": 0.72,
+      "learning_rate": 6.34e-06,
+      "loss": 4.1995,
       "step": 319
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.34e-06,
-      "loss": 4.1778,
+      "epoch": 0.72,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 4.2797,
       "step": 320
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.360000000000001e-06,
-      "loss": 4.2838,
+      "epoch": 0.72,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 4.1701,
       "step": 321
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.380000000000001e-06,
-      "loss": 4.1906,
+      "epoch": 0.72,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.0619,
       "step": 322
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.1239,
+      "epoch": 0.72,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2404,
       "step": 323
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.42e-06,
-      "loss": 4.3128,
+      "epoch": 0.73,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 4.1039,
       "step": 324
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.440000000000001e-06,
-      "loss": 4.3291,
+      "epoch": 0.73,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 4.2368,
       "step": 325
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 4.3103,
+      "epoch": 0.73,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 4.1967,
       "step": 326
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.480000000000001e-06,
-      "loss": 4.1972,
+      "epoch": 0.73,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.1744,
       "step": 327
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 4.1749,
+      "epoch": 0.74,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 4.1501,
       "step": 328
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.520000000000001e-06,
-      "loss": 4.3494,
+      "epoch": 0.74,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 4.1861,
       "step": 329
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.540000000000001e-06,
-      "loss": 4.3741,
+      "epoch": 0.74,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 4.2375,
       "step": 330
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.560000000000001e-06,
-      "loss": 4.2197,
+      "epoch": 0.74,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 4.2348,
       "step": 331
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.5800000000000005e-06,
-      "loss": 4.2233,
+      "epoch": 0.74,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.133,
       "step": 332
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.600000000000001e-06,
-      "loss": 4.265,
+      "epoch": 0.75,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 4.2921,
       "step": 333
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.620000000000001e-06,
-      "loss": 4.2857,
+      "epoch": 0.75,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 4.1948,
       "step": 334
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.640000000000001e-06,
-      "loss": 4.4136,
+      "epoch": 0.75,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.3708,
       "step": 335
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 4.1856,
+      "epoch": 0.75,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 4.1533,
       "step": 336
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.680000000000001e-06,
-      "loss": 4.0796,
+      "epoch": 0.76,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.1618,
       "step": 337
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 4.3553,
+      "epoch": 0.76,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 4.267,
       "step": 338
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.720000000000001e-06,
-      "loss": 4.1237,
+      "epoch": 0.76,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 4.2544,
       "step": 339
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 4.2866,
+      "epoch": 0.76,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 4.3039,
       "step": 340
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.760000000000001e-06,
-      "loss": 4.1974,
+      "epoch": 0.76,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 4.2502,
       "step": 341
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.780000000000001e-06,
-      "loss": 4.3292,
+      "epoch": 0.77,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.4135,
       "step": 342
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.2909,
+      "epoch": 0.77,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 4.3116,
       "step": 343
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.820000000000001e-06,
-      "loss": 4.3585,
+      "epoch": 0.77,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 4.3977,
       "step": 344
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.8400000000000014e-06,
-      "loss": 4.3761,
+      "epoch": 0.77,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 4.135,
       "step": 345
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.860000000000001e-06,
-      "loss": 4.5745,
+      "epoch": 0.78,
+      "learning_rate": 6.88e-06,
+      "loss": 4.3846,
       "step": 346
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.88e-06,
-      "loss": 4.3797,
+      "epoch": 0.78,
+      "learning_rate": 6.9e-06,
+      "loss": 4.3741,
       "step": 347
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.9e-06,
-      "loss": 4.45,
+      "epoch": 0.78,
+      "learning_rate": 6.92e-06,
+      "loss": 4.3161,
       "step": 348
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.92e-06,
-      "loss": 4.6471,
+      "epoch": 0.78,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 4.5413,
       "step": 349
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.9400000000000005e-06,
-      "loss": 4.4856,
+      "epoch": 0.78,
+      "learning_rate": 6.96e-06,
+      "loss": 4.5532,
       "step": 350
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.96e-06,
-      "loss": 4.2085,
+      "epoch": 0.79,
+      "learning_rate": 6.98e-06,
+      "loss": 4.275,
       "step": 351
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.98e-06,
-      "loss": 4.213,
+      "epoch": 0.79,
+      "learning_rate": 7e-06,
+      "loss": 4.2207,
       "step": 352
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7e-06,
-      "loss": 4.1744,
+      "epoch": 0.79,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.1401,
       "step": 353
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 4.1688,
+      "epoch": 0.79,
+      "learning_rate": 7.04e-06,
+      "loss": 4.0229,
       "step": 354
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.04e-06,
-      "loss": 4.2941,
+      "epoch": 0.8,
+      "learning_rate": 7.06e-06,
+      "loss": 4.2197,
       "step": 355
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.06e-06,
-      "loss": 4.0359,
+      "epoch": 0.8,
+      "learning_rate": 7.08e-06,
+      "loss": 4.3189,
       "step": 356
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.08e-06,
-      "loss": 4.1188,
+      "epoch": 0.8,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 4.1967,
       "step": 357
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 4.1802,
+      "epoch": 0.8,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 4.2343,
       "step": 358
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.1200000000000004e-06,
-      "loss": 4.119,
+      "epoch": 0.8,
+      "learning_rate": 7.14e-06,
+      "loss": 4.1164,
       "step": 359
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.14e-06,
-      "loss": 4.3071,
+      "epoch": 0.81,
+      "learning_rate": 7.16e-06,
+      "loss": 4.1943,
       "step": 360
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.16e-06,
-      "loss": 4.1673,
+      "epoch": 0.81,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 4.1379,
       "step": 361
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.180000000000001e-06,
-      "loss": 4.1629,
+      "epoch": 0.81,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 4.1926,
       "step": 362
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.2000000000000005e-06,
-      "loss": 4.0868,
+      "epoch": 0.81,
+      "learning_rate": 7.22e-06,
+      "loss": 4.1157,
       "step": 363
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.22e-06,
-      "loss": 4.3206,
+      "epoch": 0.82,
+      "learning_rate": 7.24e-06,
+      "loss": 4.2101,
       "step": 364
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.24e-06,
-      "loss": 4.0581,
+      "epoch": 0.82,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 4.3345,
       "step": 365
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.260000000000001e-06,
-      "loss": 4.1781,
+      "epoch": 0.82,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 4.1492,
       "step": 366
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 4.1172,
+      "epoch": 0.82,
+      "learning_rate": 7.3e-06,
+      "loss": 4.2292,
       "step": 367
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.3e-06,
-      "loss": 4.2449,
+      "epoch": 0.83,
+      "learning_rate": 7.32e-06,
+      "loss": 4.212,
       "step": 368
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.32e-06,
-      "loss": 4.2815,
+      "epoch": 0.83,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 4.1305,
       "step": 369
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 7.340000000000001e-06,
-      "loss": 4.3224,
+      "epoch": 0.83,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 4.141,
       "step": 370
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.360000000000001e-06,
-      "loss": 4.1183,
+      "epoch": 0.83,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.1366,
       "step": 371
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 4.4127,
+      "epoch": 0.83,
+      "learning_rate": 7.4e-06,
+      "loss": 4.2152,
       "step": 372
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.4e-06,
-      "loss": 4.1838,
+      "epoch": 0.84,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 4.2039,
       "step": 373
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.420000000000001e-06,
-      "loss": 4.2526,
+      "epoch": 0.84,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 4.0068,
       "step": 374
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.440000000000001e-06,
-      "loss": 4.1397,
+      "epoch": 0.84,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 4.2379,
       "step": 375
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.4600000000000006e-06,
-      "loss": 4.2553,
+      "epoch": 0.84,
+      "learning_rate": 7.48e-06,
+      "loss": 4.0893,
       "step": 376
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.48e-06,
-      "loss": 3.9823,
+      "epoch": 0.85,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 4.2137,
       "step": 377
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 4.3109,
+      "epoch": 0.85,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 4.1977,
       "step": 378
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 7.520000000000001e-06,
-      "loss": 4.2333,
+      "epoch": 0.85,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 4.1133,
       "step": 379
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.540000000000001e-06,
-      "loss": 4.3185,
+      "epoch": 0.85,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.1347,
       "step": 380
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 4.2813,
+      "epoch": 0.85,
+      "learning_rate": 7.58e-06,
+      "loss": 4.1941,
       "step": 381
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.58e-06,
-      "loss": 4.2275,
+      "epoch": 0.86,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 4.2719,
       "step": 382
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.600000000000001e-06,
-      "loss": 4.263,
+      "epoch": 0.86,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 4.1831,
       "step": 383
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.620000000000001e-06,
-      "loss": 4.2492,
+      "epoch": 0.86,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 4.1165,
       "step": 384
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.640000000000001e-06,
-      "loss": 4.1748,
+      "epoch": 0.86,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 4.2029,
       "step": 385
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.660000000000001e-06,
-      "loss": 4.2262,
+      "epoch": 0.87,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.0781,
       "step": 386
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 4.1372,
+      "epoch": 0.87,
+      "learning_rate": 7.7e-06,
+      "loss": 4.2651,
       "step": 387
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 7.7e-06,
-      "loss": 4.0275,
+      "epoch": 0.87,
+      "learning_rate": 7.72e-06,
+      "loss": 4.2909,
       "step": 388
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.72e-06,
-      "loss": 4.199,
+      "epoch": 0.87,
+      "learning_rate": 7.74e-06,
+      "loss": 4.1941,
       "step": 389
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.74e-06,
-      "loss": 4.3031,
+      "epoch": 0.87,
+      "learning_rate": 7.76e-06,
+      "loss": 4.172,
       "step": 390
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.76e-06,
-      "loss": 4.383,
+      "epoch": 0.88,
+      "learning_rate": 7.78e-06,
+      "loss": 4.2985,
       "step": 391
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.78e-06,
-      "loss": 4.0266,
+      "epoch": 0.88,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 4.1451,
       "step": 392
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.800000000000002e-06,
-      "loss": 4.2173,
+      "epoch": 0.88,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 4.1401,
       "step": 393
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.820000000000001e-06,
-      "loss": 4.353,
+      "epoch": 0.88,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 4.2258,
       "step": 394
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.840000000000001e-06,
-      "loss": 4.4657,
+      "epoch": 0.89,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.3488,
       "step": 395
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 4.2373,
+      "epoch": 0.89,
+      "learning_rate": 7.88e-06,
+      "loss": 4.3106,
       "step": 396
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.88e-06,
-      "loss": 4.2282,
+      "epoch": 0.89,
+      "learning_rate": 7.9e-06,
+      "loss": 4.4439,
       "step": 397
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.9e-06,
-      "loss": 4.7703,
+      "epoch": 0.89,
+      "learning_rate": 7.92e-06,
+      "loss": 4.4548,
       "step": 398
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.92e-06,
-      "loss": 4.4753,
+      "epoch": 0.89,
+      "learning_rate": 7.94e-06,
+      "loss": 4.5917,
       "step": 399
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.94e-06,
-      "loss": 4.5751,
+      "epoch": 0.9,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 4.5111,
       "step": 400
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.960000000000002e-06,
-      "loss": 4.2925,
+      "epoch": 0.9,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 4.2176,
       "step": 401
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 7.980000000000002e-06,
-      "loss": 4.146,
+      "epoch": 0.9,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.0665,
       "step": 402
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.2272,
+      "epoch": 0.9,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 4.1243,
       "step": 403
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.020000000000001e-06,
-      "loss": 4.1373,
+      "epoch": 0.91,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.1244,
       "step": 404
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 4.313,
+      "epoch": 0.91,
+      "learning_rate": 8.06e-06,
+      "loss": 4.0896,
       "step": 405
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.06e-06,
-      "loss": 4.1998,
+      "epoch": 0.91,
+      "learning_rate": 8.08e-06,
+      "loss": 4.1743,
       "step": 406
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.08e-06,
-      "loss": 4.2171,
+      "epoch": 0.91,
+      "learning_rate": 8.1e-06,
+      "loss": 4.1469,
       "step": 407
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.1e-06,
-      "loss": 4.1043,
+      "epoch": 0.91,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 4.2224,
       "step": 408
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.120000000000002e-06,
-      "loss": 4.2893,
+      "epoch": 0.92,
+      "learning_rate": 8.14e-06,
+      "loss": 4.1593,
       "step": 409
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.14e-06,
-      "loss": 4.2159,
+      "epoch": 0.92,
+      "learning_rate": 8.16e-06,
+      "loss": 4.1418,
       "step": 410
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.16e-06,
-      "loss": 4.0993,
+      "epoch": 0.92,
+      "learning_rate": 8.18e-06,
+      "loss": 4.3698,
       "step": 411
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.18e-06,
-      "loss": 4.2023,
+      "epoch": 0.92,
+      "learning_rate": 8.2e-06,
+      "loss": 4.1025,
       "step": 412
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.2e-06,
-      "loss": 4.205,
+      "epoch": 0.93,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.0313,
       "step": 413
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 4.0813,
+      "epoch": 0.93,
+      "learning_rate": 8.24e-06,
+      "loss": 4.192,
       "step": 414
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.24e-06,
-      "loss": 4.0771,
+      "epoch": 0.93,
+      "learning_rate": 8.26e-06,
+      "loss": 4.2583,
       "step": 415
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.26e-06,
-      "loss": 4.3072,
+      "epoch": 0.93,
+      "learning_rate": 8.28e-06,
+      "loss": 4.1274,
       "step": 416
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.28e-06,
-      "loss": 4.1095,
+      "epoch": 0.93,
+      "learning_rate": 8.3e-06,
+      "loss": 4.152,
       "step": 417
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.3e-06,
-      "loss": 4.241,
+      "epoch": 0.94,
+      "learning_rate": 8.32e-06,
+      "loss": 4.0524,
       "step": 418
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.32e-06,
-      "loss": 4.1763,
+      "epoch": 0.94,
+      "learning_rate": 8.34e-06,
+      "loss": 4.1577,
       "step": 419
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.34e-06,
-      "loss": 4.2173,
+      "epoch": 0.94,
+      "learning_rate": 8.36e-06,
+      "loss": 4.0665,
       "step": 420
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.36e-06,
-      "loss": 4.2836,
+      "epoch": 0.94,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 4.2451,
       "step": 421
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.380000000000001e-06,
-      "loss": 4.2942,
+      "epoch": 0.95,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.0875,
       "step": 422
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.2181,
+      "epoch": 0.95,
+      "learning_rate": 8.42e-06,
+      "loss": 4.1906,
       "step": 423
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.42e-06,
-      "loss": 4.219,
+      "epoch": 0.95,
+      "learning_rate": 8.44e-06,
+      "loss": 4.2461,
       "step": 424
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.44e-06,
-      "loss": 4.2753,
+      "epoch": 0.95,
+      "learning_rate": 8.46e-06,
+      "loss": 4.1716,
       "step": 425
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.46e-06,
-      "loss": 4.3047,
+      "epoch": 0.96,
+      "learning_rate": 8.48e-06,
+      "loss": 4.2216,
       "step": 426
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.48e-06,
-      "loss": 4.1395,
+      "epoch": 0.96,
+      "learning_rate": 8.5e-06,
+      "loss": 4.18,
       "step": 427
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.5e-06,
-      "loss": 4.3815,
+      "epoch": 0.96,
+      "learning_rate": 8.52e-06,
+      "loss": 4.1531,
       "step": 428
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.52e-06,
-      "loss": 4.1654,
+      "epoch": 0.96,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 4.104,
       "step": 429
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.540000000000001e-06,
-      "loss": 4.3267,
+      "epoch": 0.96,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 4.1649,
       "step": 430
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.560000000000001e-06,
-      "loss": 4.414,
+      "epoch": 0.97,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 4.2599,
       "step": 431
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 8.580000000000001e-06,
-      "loss": 4.26,
+      "epoch": 0.97,
+      "learning_rate": 8.6e-06,
+      "loss": 4.3766,
       "step": 432
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.6e-06,
-      "loss": 3.976,
+      "epoch": 0.97,
+      "learning_rate": 8.62e-06,
+      "loss": 4.1677,
       "step": 433
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.62e-06,
-      "loss": 4.206,
+      "epoch": 0.97,
+      "learning_rate": 8.64e-06,
+      "loss": 4.1718,
       "step": 434
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.64e-06,
-      "loss": 4.1705,
+      "epoch": 0.98,
+      "learning_rate": 8.66e-06,
+      "loss": 4.2196,
       "step": 435
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.66e-06,
-      "loss": 4.4165,
+      "epoch": 0.98,
+      "learning_rate": 8.68e-06,
+      "loss": 4.1337,
       "step": 436
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.68e-06,
-      "loss": 4.4192,
+      "epoch": 0.98,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 4.1335,
       "step": 437
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.700000000000001e-06,
-      "loss": 4.3003,
+      "epoch": 0.98,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 4.3058,
       "step": 438
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.720000000000001e-06,
-      "loss": 4.1106,
+      "epoch": 0.98,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 4.2738,
       "step": 439
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.740000000000001e-06,
-      "loss": 4.1738,
+      "epoch": 0.99,
+      "learning_rate": 8.76e-06,
+      "loss": 4.3081,
       "step": 440
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 8.76e-06,
-      "loss": 4.355,
+      "epoch": 0.99,
+      "learning_rate": 8.78e-06,
+      "loss": 4.2582,
       "step": 441
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.78e-06,
-      "loss": 4.2148,
+      "epoch": 0.99,
+      "learning_rate": 8.8e-06,
+      "loss": 4.4641,
       "step": 442
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.8e-06,
-      "loss": 4.2945,
+      "epoch": 0.99,
+      "learning_rate": 8.82e-06,
+      "loss": 4.3681,
       "step": 443
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.82e-06,
-      "loss": 4.4645,
+      "epoch": 1.0,
+      "learning_rate": 8.84e-06,
+      "loss": 4.3675,
       "step": 444
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 8.84e-06,
-      "loss": 4.2696,
+      "epoch": 1.0,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 4.6441,
       "step": 445
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 8.860000000000002e-06,
-      "loss": 4.1095,
+      "loss": 4.3047,
       "step": 446
     },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.880000000000001e-06,
-      "loss": 4.3117,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.900000000000001e-06,
-      "loss": 4.3888,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.920000000000001e-06,
-      "loss": 4.3431,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.94e-06,
-      "loss": 4.8657,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.96e-06,
-      "loss": 4.2217,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.98e-06,
-      "loss": 4.2118,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9e-06,
-      "loss": 4.358,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.020000000000002e-06,
-      "loss": 4.0613,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.040000000000002e-06,
-      "loss": 4.3277,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.060000000000001e-06,
-      "loss": 4.2623,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.080000000000001e-06,
-      "loss": 4.2232,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 4.1554,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.12e-06,
-      "loss": 4.1916,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.14e-06,
-      "loss": 4.3454,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.16e-06,
-      "loss": 4.1008,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.180000000000002e-06,
-      "loss": 4.1496,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.200000000000002e-06,
-      "loss": 4.1758,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.220000000000002e-06,
-      "loss": 4.1112,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.240000000000001e-06,
-      "loss": 4.1816,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.260000000000001e-06,
-      "loss": 4.2129,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.280000000000001e-06,
-      "loss": 4.1762,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.3e-06,
-      "loss": 4.0683,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.32e-06,
-      "loss": 4.1022,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.340000000000002e-06,
-      "loss": 4.2871,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.360000000000002e-06,
-      "loss": 4.0804,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.38e-06,
-      "loss": 4.0755,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.4e-06,
-      "loss": 4.1634,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.42e-06,
-      "loss": 4.2345,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.440000000000001e-06,
-      "loss": 4.191,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.460000000000001e-06,
-      "loss": 4.2504,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.48e-06,
-      "loss": 4.4041,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.5e-06,
-      "loss": 4.1215,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.52e-06,
-      "loss": 3.9798,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.54e-06,
-      "loss": 4.2905,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.56e-06,
-      "loss": 4.1178,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.58e-06,
-      "loss": 4.0777,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.2803,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.620000000000001e-06,
-      "loss": 4.1497,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.640000000000001e-06,
-      "loss": 4.3243,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.66e-06,
-      "loss": 4.1365,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.68e-06,
-      "loss": 4.0382,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.7e-06,
-      "loss": 4.1827,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.72e-06,
-      "loss": 4.2169,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.74e-06,
-      "loss": 4.2879,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.760000000000001e-06,
-      "loss": 4.1137,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 4.3924,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.4231,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.820000000000001e-06,
-      "loss": 4.2154,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.84e-06,
-      "loss": 4.3733,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.86e-06,
-      "loss": 4.1577,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.88e-06,
-      "loss": 4.1023,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.9e-06,
-      "loss": 4.7654,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 4.7262,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.940000000000001e-06,
-      "loss": 4.3268,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 4.228239059448242,
-      "eval_runtime": 920.7538,
-      "eval_samples_per_second": 2.869,
-      "eval_steps_per_second": 0.359,
-      "eval_wer": 1.9680880602935342,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.4602,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 4.1627,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 1e-05,
-      "loss": 4.2864,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.974489795918369e-06,
-      "loss": 4.1179,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.948979591836737e-06,
-      "loss": 4.1012,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.923469387755103e-06,
-      "loss": 4.2587,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.89795918367347e-06,
-      "loss": 4.2037,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.872448979591838e-06,
-      "loss": 4.2532,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.846938775510205e-06,
-      "loss": 4.1469,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.821428571428573e-06,
-      "loss": 4.2007,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.795918367346939e-06,
-      "loss": 4.1991,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.770408163265307e-06,
-      "loss": 4.164,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.744897959183674e-06,
-      "loss": 4.4233,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.719387755102042e-06,
-      "loss": 4.1772,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.693877551020408e-06,
-      "loss": 4.1559,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.668367346938776e-06,
-      "loss": 4.0684,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.642857142857144e-06,
-      "loss": 4.2825,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.617346938775512e-06,
-      "loss": 4.2294,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.591836734693878e-06,
-      "loss": 4.2505,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.566326530612246e-06,
-      "loss": 4.144,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.540816326530612e-06,
-      "loss": 4.1488,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.51530612244898e-06,
-      "loss": 4.1249,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.489795918367348e-06,
-      "loss": 4.0448,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.464285714285714e-06,
-      "loss": 4.2133,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.438775510204082e-06,
-      "loss": 4.2394,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.41326530612245e-06,
-      "loss": 4.0197,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.387755102040818e-06,
-      "loss": 4.2476,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.362244897959184e-06,
-      "loss": 4.2209,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.336734693877552e-06,
-      "loss": 4.1473,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.31122448979592e-06,
-      "loss": 4.4424,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.285714285714288e-06,
-      "loss": 4.2181,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.260204081632654e-06,
-      "loss": 4.0263,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.234693877551022e-06,
-      "loss": 4.2009,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.209183673469388e-06,
-      "loss": 4.0569,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.183673469387756e-06,
-      "loss": 4.1976,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.158163265306124e-06,
-      "loss": 4.2564,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.13265306122449e-06,
-      "loss": 4.3148,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.107142857142858e-06,
-      "loss": 4.3963,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.081632653061225e-06,
-      "loss": 4.2059,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.056122448979593e-06,
-      "loss": 4.3102,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.03061224489796e-06,
-      "loss": 4.3175,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.005102040816327e-06,
-      "loss": 4.3354,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.979591836734695e-06,
-      "loss": 4.5025,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.954081632653061e-06,
-      "loss": 4.3214,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.92857142857143e-06,
-      "loss": 4.4099,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.903061224489795e-06,
-      "loss": 4.4373,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.877551020408163e-06,
-      "loss": 4.3766,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 8.852040816326531e-06,
-      "loss": 4.7145,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.826530612244899e-06,
-      "loss": 4.1757,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.801020408163265e-06,
-      "loss": 5.1677,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.775510204081633e-06,
-      "loss": 3.9488,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.750000000000001e-06,
-      "loss": 4.3083,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.724489795918369e-06,
-      "loss": 4.2125,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.698979591836737e-06,
-      "loss": 4.1317,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.673469387755103e-06,
-      "loss": 4.1393,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.64795918367347e-06,
-      "loss": 4.1868,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 8.622448979591837e-06,
-      "loss": 4.056,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.596938775510205e-06,
-      "loss": 4.1679,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.571428571428571e-06,
-      "loss": 4.1314,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.545918367346939e-06,
-      "loss": 4.0573,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.520408163265307e-06,
-      "loss": 4.2746,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.494897959183675e-06,
-      "loss": 4.1251,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.469387755102042e-06,
-      "loss": 4.1467,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.443877551020409e-06,
-      "loss": 4.1499,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.418367346938776e-06,
-      "loss": 4.2789,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 8.392857142857144e-06,
-      "loss": 4.2684,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.36734693877551e-06,
-      "loss": 4.127,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.341836734693878e-06,
-      "loss": 4.1092,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.316326530612246e-06,
-      "loss": 4.2655,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.290816326530612e-06,
-      "loss": 4.0523,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.26530612244898e-06,
-      "loss": 4.0347,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.239795918367348e-06,
-      "loss": 4.2677,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.214285714285714e-06,
-      "loss": 4.0504,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.188775510204082e-06,
-      "loss": 4.0672,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 8.16326530612245e-06,
-      "loss": 4.228,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.137755102040818e-06,
-      "loss": 4.07,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.112244897959184e-06,
-      "loss": 4.1979,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.086734693877552e-06,
-      "loss": 4.2986,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.06122448979592e-06,
-      "loss": 3.8692,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.035714285714286e-06,
-      "loss": 4.3322,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 8.010204081632654e-06,
-      "loss": 4.31,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.98469387755102e-06,
-      "loss": 4.0711,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.959183673469388e-06,
-      "loss": 4.0197,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.933673469387756e-06,
-      "loss": 4.2065,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.908163265306124e-06,
-      "loss": 4.3477,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.88265306122449e-06,
-      "loss": 4.2166,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.857142857142858e-06,
-      "loss": 4.2325,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.831632653061226e-06,
-      "loss": 4.2987,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.806122448979593e-06,
-      "loss": 4.0,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.78061224489796e-06,
-      "loss": 4.1028,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.755102040816327e-06,
-      "loss": 4.2336,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.729591836734695e-06,
-      "loss": 4.3247,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.704081632653061e-06,
-      "loss": 4.5599,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.67857142857143e-06,
-      "loss": 4.3741,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.653061224489796e-06,
-      "loss": 4.3032,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.627551020408163e-06,
-      "loss": 4.1155,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.602040816326531e-06,
-      "loss": 4.2505,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.576530612244899e-06,
-      "loss": 4.3653,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.551020408163265e-06,
-      "loss": 4.3459,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.525510204081633e-06,
-      "loss": 4.2851,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 4.2087,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 7.474489795918368e-06,
-      "loss": 4.0383,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.448979591836736e-06,
-      "loss": 4.1835,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.423469387755102e-06,
-      "loss": 4.1534,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.39795918367347e-06,
-      "loss": 4.1189,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.372448979591838e-06,
-      "loss": 4.1494,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.346938775510205e-06,
-      "loss": 4.1877,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.321428571428572e-06,
-      "loss": 4.1186,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.295918367346939e-06,
-      "loss": 4.1042,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.270408163265307e-06,
-      "loss": 4.1356,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 7.244897959183675e-06,
-      "loss": 4.0916,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.219387755102042e-06,
-      "loss": 4.2797,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.193877551020409e-06,
-      "loss": 4.2055,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.168367346938776e-06,
-      "loss": 4.2332,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1428571428571436e-06,
-      "loss": 4.1259,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.1173469387755114e-06,
-      "loss": 4.1521,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.091836734693878e-06,
-      "loss": 4.0654,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.0663265306122455e-06,
-      "loss": 4.1303,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 7.0408163265306125e-06,
-      "loss": 4.0865,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 7.01530612244898e-06,
-      "loss": 4.1392,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.989795918367348e-06,
-      "loss": 4.1308,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.964285714285714e-06,
-      "loss": 4.0555,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.938775510204082e-06,
-      "loss": 4.2929,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.91326530612245e-06,
-      "loss": 4.1076,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.887755102040817e-06,
-      "loss": 4.2519,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.862244897959184e-06,
-      "loss": 4.2675,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.836734693877551e-06,
-      "loss": 4.2299,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.811224489795919e-06,
-      "loss": 4.1654,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.785714285714287e-06,
-      "loss": 4.1816,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.760204081632653e-06,
-      "loss": 4.2973,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.734693877551021e-06,
-      "loss": 4.2669,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.709183673469388e-06,
-      "loss": 4.1018,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.683673469387756e-06,
-      "loss": 4.2136,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.658163265306124e-06,
-      "loss": 4.1718,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.63265306122449e-06,
-      "loss": 4.2685,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.607142857142858e-06,
-      "loss": 4.2419,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.581632653061225e-06,
-      "loss": 4.1071,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.556122448979593e-06,
-      "loss": 4.0533,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.530612244897959e-06,
-      "loss": 4.2629,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.505102040816327e-06,
-      "loss": 4.0422,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4795918367346946e-06,
-      "loss": 4.223,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.454081632653062e-06,
-      "loss": 4.5604,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.4285714285714295e-06,
-      "loss": 4.3802,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.403061224489796e-06,
-      "loss": 4.1618,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.3775510204081635e-06,
-      "loss": 4.011,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.352040816326531e-06,
-      "loss": 4.3477,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.326530612244899e-06,
-      "loss": 4.2592,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.301020408163265e-06,
-      "loss": 4.1053,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.275510204081633e-06,
-      "loss": 4.5911,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.25e-06,
-      "loss": 4.4,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.224489795918368e-06,
-      "loss": 4.2669,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.198979591836736e-06,
-      "loss": 4.2794,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.173469387755102e-06,
-      "loss": 4.2399,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.14795918367347e-06,
-      "loss": 4.2069,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.122448979591837e-06,
-      "loss": 4.1954,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.096938775510205e-06,
-      "loss": 4.1589,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.071428571428571e-06,
-      "loss": 4.0785,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.045918367346939e-06,
-      "loss": 4.3309,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.020408163265307e-06,
-      "loss": 4.1584,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.994897959183674e-06,
-      "loss": 4.2442,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.969387755102042e-06,
-      "loss": 4.0299,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.943877551020408e-06,
-      "loss": 4.0324,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.918367346938776e-06,
-      "loss": 4.2622,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 5.892857142857144e-06,
-      "loss": 4.0929,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.867346938775511e-06,
-      "loss": 4.0442,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.841836734693878e-06,
-      "loss": 4.3026,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.816326530612246e-06,
-      "loss": 4.0586,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.790816326530613e-06,
-      "loss": 4.2993,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7653061224489805e-06,
-      "loss": 4.2909,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.739795918367348e-06,
-      "loss": 4.2345,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.7142857142857145e-06,
-      "loss": 4.1124,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.688775510204082e-06,
-      "loss": 3.9775,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 5.663265306122449e-06,
-      "loss": 4.0079,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.637755102040817e-06,
-      "loss": 4.0691,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.6122448979591834e-06,
-      "loss": 4.0466,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.586734693877551e-06,
-      "loss": 4.166,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.561224489795919e-06,
-      "loss": 4.0409,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.535714285714286e-06,
-      "loss": 3.9967,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.510204081632653e-06,
-      "loss": 4.0257,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.48469387755102e-06,
-      "loss": 4.0889,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.459183673469388e-06,
-      "loss": 4.3042,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 5.433673469387756e-06,
-      "loss": 4.0469,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.408163265306123e-06,
-      "loss": 4.1806,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.38265306122449e-06,
-      "loss": 4.1818,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.357142857142857e-06,
-      "loss": 4.2281,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.331632653061225e-06,
-      "loss": 4.1242,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.306122448979593e-06,
-      "loss": 4.2261,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.280612244897959e-06,
-      "loss": 4.2931,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.255102040816327e-06,
-      "loss": 4.2491,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.229591836734695e-06,
-      "loss": 4.3815,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 5.204081632653062e-06,
-      "loss": 4.1046,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1785714285714296e-06,
-      "loss": 4.2037,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.153061224489796e-06,
-      "loss": 4.1736,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.127551020408164e-06,
-      "loss": 4.4844,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.1020408163265315e-06,
-      "loss": 4.2122,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0765306122448985e-06,
-      "loss": 4.3424,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0510204081632655e-06,
-      "loss": 4.3975,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5.0255102040816325e-06,
-      "loss": 4.3493,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 5e-06,
-      "loss": 4.3533,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 4.974489795918368e-06,
-      "loss": 4.55,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.948979591836735e-06,
-      "loss": 4.1124,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.923469387755102e-06,
-      "loss": 4.1696,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.897959183673469e-06,
-      "loss": 4.0838,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.872448979591837e-06,
-      "loss": 3.9315,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.846938775510204e-06,
-      "loss": 4.0708,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.821428571428572e-06,
-      "loss": 4.2055,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.795918367346939e-06,
-      "loss": 4.1657,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.770408163265306e-06,
-      "loss": 4.1916,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 4.744897959183674e-06,
-      "loss": 4.0709,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.719387755102041e-06,
-      "loss": 4.1493,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.693877551020409e-06,
-      "loss": 4.0559,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.668367346938776e-06,
-      "loss": 4.1486,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.642857142857144e-06,
-      "loss": 4.0489,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.617346938775511e-06,
-      "loss": 4.2475,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.591836734693878e-06,
-      "loss": 4.2727,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.566326530612245e-06,
-      "loss": 4.0665,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.540816326530613e-06,
-      "loss": 4.3117,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.51530612244898e-06,
-      "loss": 4.1286,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.489795918367348e-06,
-      "loss": 3.8977,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.464285714285715e-06,
-      "loss": 4.186,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.438775510204082e-06,
-      "loss": 3.9928,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.4132653061224495e-06,
-      "loss": 4.0517,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.3877551020408165e-06,
-      "loss": 4.1138,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.362244897959184e-06,
-      "loss": 4.2047,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.336734693877551e-06,
-      "loss": 4.1134,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 4.3112244897959184e-06,
-      "loss": 4.1869,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.2857142857142855e-06,
-      "loss": 4.1596,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.260204081632653e-06,
-      "loss": 4.1462,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.234693877551021e-06,
-      "loss": 3.995,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.209183673469388e-06,
-      "loss": 4.2226,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.183673469387755e-06,
-      "loss": 4.2854,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.158163265306123e-06,
-      "loss": 4.1906,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.13265306122449e-06,
-      "loss": 4.1204,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.107142857142857e-06,
-      "loss": 4.2558,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.081632653061225e-06,
-      "loss": 4.009,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.056122448979592e-06,
-      "loss": 4.1341,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.03061224489796e-06,
-      "loss": 4.3323,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.005102040816327e-06,
-      "loss": 4.1847,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.979591836734694e-06,
-      "loss": 4.0944,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.954081632653062e-06,
-      "loss": 4.2967,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.928571428571429e-06,
-      "loss": 4.2602,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.903061224489797e-06,
-      "loss": 4.1164,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.877551020408164e-06,
-      "loss": 4.0042,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.852040816326531e-06,
-      "loss": 4.146,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.826530612244898e-06,
-      "loss": 4.241,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.8010204081632656e-06,
-      "loss": 4.4222,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7755102040816327e-06,
-      "loss": 4.2069,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.7500000000000005e-06,
-      "loss": 4.4836,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.724489795918368e-06,
-      "loss": 4.4231,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.698979591836735e-06,
-      "loss": 4.3741,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6734693877551024e-06,
-      "loss": 4.2393,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6479591836734694e-06,
-      "loss": 4.2289,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.6224489795918373e-06,
-      "loss": 4.1617,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5969387755102043e-06,
-      "loss": 4.0349,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5714285714285718e-06,
-      "loss": 4.2334,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.545918367346939e-06,
-      "loss": 4.2539,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.5204081632653062e-06,
-      "loss": 4.1684,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.494897959183674e-06,
-      "loss": 4.2257,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.469387755102041e-06,
-      "loss": 4.0784,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4438775510204086e-06,
-      "loss": 4.1651,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.4183673469387756e-06,
-      "loss": 4.0865,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.3928571428571435e-06,
-      "loss": 4.1835,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.3673469387755105e-06,
-      "loss": 4.0762,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.341836734693878e-06,
-      "loss": 4.1157,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.316326530612245e-06,
-      "loss": 4.1859,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2908163265306124e-06,
-      "loss": 4.1285,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2653061224489794e-06,
-      "loss": 4.1179,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2397959183673473e-06,
-      "loss": 4.1538,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.2142857142857147e-06,
-      "loss": 4.1982,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1887755102040818e-06,
-      "loss": 4.0693,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 3.1632653061224496e-06,
-      "loss": 4.1364,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.1377551020408166e-06,
-      "loss": 4.264,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.112244897959184e-06,
-      "loss": 4.2401,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.086734693877551e-06,
-      "loss": 3.9516,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0612244897959185e-06,
-      "loss": 4.0435,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0357142857142856e-06,
-      "loss": 4.1435,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 3.0102040816326534e-06,
-      "loss": 3.9758,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.984693877551021e-06,
-      "loss": 4.1771,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.959183673469388e-06,
-      "loss": 4.028,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.9336734693877553e-06,
-      "loss": 4.0337,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.908163265306123e-06,
-      "loss": 4.0541,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8826530612244902e-06,
-      "loss": 4.1735,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 4.1014,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8316326530612247e-06,
-      "loss": 4.1063,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.8061224489795917e-06,
-      "loss": 4.0727,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7806122448979596e-06,
-      "loss": 3.9978,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7551020408163266e-06,
-      "loss": 4.1545,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.729591836734694e-06,
-      "loss": 4.1971,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.7040816326530615e-06,
-      "loss": 4.0807,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6785714285714285e-06,
-      "loss": 4.1616,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6530612244897964e-06,
-      "loss": 4.1776,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.6275510204081634e-06,
-      "loss": 3.9967,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.602040816326531e-06,
-      "loss": 4.1897,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.576530612244898e-06,
-      "loss": 4.2856,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5510204081632657e-06,
-      "loss": 4.3239,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5255102040816328e-06,
-      "loss": 4.2239,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.5e-06,
-      "loss": 4.3134,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.4744897959183676e-06,
-      "loss": 4.4016,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.4489795918367347e-06,
-      "loss": 4.7601,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.423469387755102e-06,
-      "loss": 4.5584,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3979591836734696e-06,
-      "loss": 4.173,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.372448979591837e-06,
-      "loss": 4.0763,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.3469387755102044e-06,
-      "loss": 4.0927,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.321428571428572e-06,
-      "loss": 4.1416,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 3.9634,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.2704081632653064e-06,
-      "loss": 4.1256,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 2.244897959183674e-06,
-      "loss": 4.2027,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.219387755102041e-06,
-      "loss": 4.0409,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1938775510204083e-06,
-      "loss": 4.2505,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1683673469387757e-06,
-      "loss": 4.0677,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1428571428571427e-06,
-      "loss": 4.3241,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.1173469387755106e-06,
-      "loss": 4.1454,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0918367346938776e-06,
-      "loss": 3.9713,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.066326530612245e-06,
-      "loss": 4.0555,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.0408163265306125e-06,
-      "loss": 4.0508,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 2.01530612244898e-06,
-      "loss": 4.0331,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.989795918367347e-06,
-      "loss": 4.0781,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.9642857142857144e-06,
-      "loss": 4.1002,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.938775510204082e-06,
-      "loss": 4.0934,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.913265306122449e-06,
-      "loss": 4.1358,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8877551020408163e-06,
-      "loss": 4.0446,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.862244897959184e-06,
-      "loss": 4.0584,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8367346938775512e-06,
-      "loss": 3.9872,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.8112244897959187e-06,
-      "loss": 4.0894,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.7857142857142859e-06,
-      "loss": 4.3481,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7602040816326531e-06,
-      "loss": 4.1779,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7346938775510206e-06,
-      "loss": 4.1513,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.7091836734693878e-06,
-      "loss": 4.0619,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6836734693877552e-06,
-      "loss": 4.0468,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6581632653061225e-06,
-      "loss": 4.2047,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6326530612244897e-06,
-      "loss": 4.0403,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.6071428571428574e-06,
-      "loss": 3.9643,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.5816326530612248e-06,
-      "loss": 4.2552,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.556122448979592e-06,
-      "loss": 4.1078,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 4.1998,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.5051020408163267e-06,
-      "loss": 4.1312,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.479591836734694e-06,
-      "loss": 4.129,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4540816326530614e-06,
-      "loss": 4.2123,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 4.0576,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.4030612244897959e-06,
-      "loss": 4.1398,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3775510204081633e-06,
-      "loss": 4.341,
-      "step": 841
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.3520408163265307e-06,
-      "loss": 4.2066,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3265306122448982e-06,
-      "loss": 4.341,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.3010204081632654e-06,
-      "loss": 4.3768,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2755102040816329e-06,
-      "loss": 4.4031,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.25e-06,
-      "loss": 4.4808,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.2244897959183673e-06,
-      "loss": 4.2938,
-      "step": 847
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1989795918367348e-06,
-      "loss": 4.5986,
-      "step": 848
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1734693877551022e-06,
-      "loss": 4.566,
-      "step": 849
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.1479591836734695e-06,
-      "loss": 4.3436,
-      "step": 850
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 1.122448979591837e-06,
-      "loss": 4.1048,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0969387755102041e-06,
-      "loss": 4.0466,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 4.1017,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0459183673469388e-06,
-      "loss": 4.0296,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 1.0204081632653063e-06,
-      "loss": 4.0512,
-      "step": 855
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.948979591836735e-07,
-      "loss": 4.0673,
-      "step": 856
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.69387755102041e-07,
-      "loss": 4.1707,
-      "step": 857
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.438775510204082e-07,
-      "loss": 4.1872,
-      "step": 858
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.183673469387756e-07,
-      "loss": 4.0315,
-      "step": 859
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.928571428571429e-07,
-      "loss": 4.1158,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755103e-07,
-      "loss": 4.3055,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938776e-07,
-      "loss": 4.026,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.163265306122449e-07,
-      "loss": 4.1558,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306124e-07,
-      "loss": 4.2992,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 4.1812,
-      "step": 865
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-07,
-      "loss": 3.914,
-      "step": 866
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.142857142857143e-07,
-      "loss": 4.1402,
-      "step": 867
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.887755102040817e-07,
-      "loss": 4.0025,
-      "step": 868
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.632653061224491e-07,
-      "loss": 4.0408,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.377551020408164e-07,
-      "loss": 4.2121,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591837e-07,
-      "loss": 4.0377,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.867346938775511e-07,
-      "loss": 4.2279,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.612244897959184e-07,
-      "loss": 4.1868,
-      "step": 873
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-07,
-      "loss": 4.2097,
-      "step": 874
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-07,
-      "loss": 4.2013,
-      "step": 875
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.846938775510205e-07,
-      "loss": 4.0373,
-      "step": 876
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.591836734693878e-07,
-      "loss": 4.1152,
-      "step": 877
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.3367346938775514e-07,
-      "loss": 4.1555,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.0816326530612243e-07,
-      "loss": 4.3356,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.826530612244898e-07,
-      "loss": 4.4388,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.5714285714285716e-07,
-      "loss": 4.0538,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.3163265306122455e-07,
-      "loss": 4.1586,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.0612244897959183e-07,
-      "loss": 4.0861,
-      "step": 883
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.806122448979592e-07,
-      "loss": 4.0793,
-      "step": 884
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.5510204081632656e-07,
-      "loss": 4.1621,
-      "step": 885
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.295918367346939e-07,
-      "loss": 4.1331,
-      "step": 886
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.0408163265306121e-07,
-      "loss": 4.0988,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142858e-07,
-      "loss": 4.3043,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.5306122448979592e-07,
-      "loss": 4.3478,
-      "step": 889
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.2755102040816328e-07,
-      "loss": 4.1803,
-      "step": 890
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.0204081632653061e-07,
-      "loss": 4.4327,
-      "step": 891
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 7.653061224489796e-08,
-      "loss": 4.0019,
-      "step": 892
-    },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 4.296594612534271,
-      "train_runtime": 6075.7342,
-      "train_samples_per_second": 4.697,
-      "train_steps_per_second": 0.147
+      "train_loss": 4.377411050112258,
+      "train_runtime": 4925.13,
+      "train_samples_per_second": 5.794,
+      "train_steps_per_second": 0.091
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,