diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9995796553173603,
-  "global_step": 1189,
+  "epoch": 0.9987389659520807,
+  "global_step": 594,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,7166 +10,3587 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 4.8613,
+      "loss": 4.8209,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.0539,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.9975,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 5.0438,
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.8762,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 4.8735,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.8257,
       "step": 4
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 4.6018,
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.7808,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 4.9754,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.795,
       "step": 6
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 4.721,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.8669,
       "step": 7
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.7079,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.7709,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 4.9124,
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.8758,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 4.7564,
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.7759,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 4.8114,
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 4.7793,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-06,
-      "loss": 4.6734,
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.8164,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 4.914,
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 4.7334,
       "step": 13
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.2e-06,
-      "loss": 4.713,
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.6811,
       "step": 14
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 4.5812,
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 4.5624,
       "step": 15
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.6e-06,
-      "loss": 4.7971,
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.7251,
       "step": 16
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 4.7246,
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 4.7181,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3e-06,
-      "loss": 4.8549,
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.7752,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 4.5943,
+      "epoch": 0.03,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.5284,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 4.7956,
+      "epoch": 0.03,
+      "learning_rate": 3.6e-06,
+      "loss": 4.7388,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-06,
-      "loss": 4.7439,
+      "epoch": 0.04,
+      "learning_rate": 3.8e-06,
+      "loss": 4.679,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.8e-06,
-      "loss": 4.5445,
+      "epoch": 0.04,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.6217,
       "step": 22
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 4.6856,
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.6101,
       "step": 23
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 4.7182,
+      "epoch": 0.04,
+      "learning_rate": 4.4e-06,
+      "loss": 4.6706,
       "step": 24
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.4e-06,
-      "loss": 4.5307,
+      "epoch": 0.04,
+      "learning_rate": 4.6e-06,
+      "loss": 4.4344,
       "step": 25
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.6e-06,
-      "loss": 4.6767,
+      "epoch": 0.04,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.8303,
       "step": 26
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 4.5058,
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 4.4799,
       "step": 27
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5e-06,
-      "loss": 4.3801,
+      "epoch": 0.05,
+      "learning_rate": 5.2e-06,
+      "loss": 4.3854,
       "step": 28
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.2e-06,
-      "loss": 4.4467,
+      "epoch": 0.05,
+      "learning_rate": 5.4e-06,
+      "loss": 4.3895,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-06,
-      "loss": 4.4919,
+      "epoch": 0.05,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.5382,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 4.6095,
+      "epoch": 0.05,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5426,
       "step": 31
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.8e-06,
-      "loss": 4.3976,
+      "epoch": 0.05,
+      "learning_rate": 6e-06,
+      "loss": 4.4101,
       "step": 32
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6e-06,
-      "loss": 4.4978,
+      "epoch": 0.06,
+      "learning_rate": 6.2e-06,
+      "loss": 4.3962,
       "step": 33
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.2e-06,
-      "loss": 4.2034,
+      "epoch": 0.06,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.3527,
       "step": 34
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 4.3845,
+      "epoch": 0.06,
+      "learning_rate": 6.6e-06,
+      "loss": 4.3979,
       "step": 35
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.6e-06,
-      "loss": 4.6754,
+      "epoch": 0.06,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.5459,
       "step": 36
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 4.5147,
+      "epoch": 0.06,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.3364,
       "step": 37
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 4.3481,
+      "epoch": 0.06,
+      "learning_rate": 7.2e-06,
+      "loss": 4.4963,
       "step": 38
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.2e-06,
-      "loss": 4.8411,
+      "epoch": 0.07,
+      "learning_rate": 7.4e-06,
+      "loss": 4.6584,
       "step": 39
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.4e-06,
-      "loss": 4.502,
+      "epoch": 0.07,
+      "learning_rate": 7.6e-06,
+      "loss": 4.3229,
       "step": 40
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.6e-06,
-      "loss": 4.3018,
+      "epoch": 0.07,
+      "learning_rate": 7.8e-06,
+      "loss": 4.3119,
       "step": 41
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.8e-06,
-      "loss": 4.368,
+      "epoch": 0.07,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.5691,
       "step": 42
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 4.3542,
+      "epoch": 0.07,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.298,
       "step": 43
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 4.4008,
+      "epoch": 0.07,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.4279,
       "step": 44
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 4.5618,
+      "epoch": 0.08,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.5293,
       "step": 45
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 4.5316,
+      "epoch": 0.08,
+      "learning_rate": 8.8e-06,
+      "loss": 4.564,
       "step": 46
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.8e-06,
-      "loss": 4.6923,
+      "epoch": 0.08,
+      "learning_rate": 9e-06,
+      "loss": 4.4752,
       "step": 47
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9e-06,
-      "loss": 4.909,
+      "epoch": 0.08,
+      "learning_rate": 9.2e-06,
+      "loss": 4.8316,
       "step": 48
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.2e-06,
-      "loss": 5.1014,
+      "epoch": 0.08,
+      "learning_rate": 9.4e-06,
+      "loss": 4.9229,
       "step": 49
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.2e-06,
-      "loss": 4.8069,
+      "epoch": 0.08,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.5942,
       "step": 50
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.4e-06,
-      "loss": 4.3221,
+      "epoch": 0.09,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.2587,
       "step": 51
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 4.3242,
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 4.2727,
       "step": 52
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 4.1941,
+      "epoch": 0.09,
+      "learning_rate": 1.02e-05,
+      "loss": 4.2882,
       "step": 53
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1e-05,
-      "loss": 4.2626,
+      "epoch": 0.09,
+      "learning_rate": 1.04e-05,
+      "loss": 4.3738,
       "step": 54
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.02e-05,
-      "loss": 4.4186,
+      "epoch": 0.09,
+      "learning_rate": 1.06e-05,
+      "loss": 4.2194,
       "step": 55
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.04e-05,
-      "loss": 4.2228,
+      "epoch": 0.09,
+      "learning_rate": 1.08e-05,
+      "loss": 4.4177,
       "step": 56
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.06e-05,
-      "loss": 4.4831,
+      "epoch": 0.1,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.2308,
       "step": 57
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.08e-05,
-      "loss": 4.3735,
+      "epoch": 0.1,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.5297,
       "step": 58
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 4.2564,
+      "epoch": 0.1,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.2934,
       "step": 59
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 4.4065,
+      "epoch": 0.1,
+      "learning_rate": 1.16e-05,
+      "loss": 4.072,
       "step": 60
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 4.2041,
+      "epoch": 0.1,
+      "learning_rate": 1.18e-05,
+      "loss": 4.3357,
       "step": 61
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.16e-05,
-      "loss": 4.382,
+      "epoch": 0.1,
+      "learning_rate": 1.2e-05,
+      "loss": 4.3799,
       "step": 62
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.18e-05,
-      "loss": 4.2571,
+      "epoch": 0.11,
+      "learning_rate": 1.22e-05,
+      "loss": 4.2646,
       "step": 63
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.2e-05,
-      "loss": 4.1404,
+      "epoch": 0.11,
+      "learning_rate": 1.24e-05,
+      "loss": 4.2315,
       "step": 64
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.22e-05,
-      "loss": 4.1314,
+      "epoch": 0.11,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.2819,
       "step": 65
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.24e-05,
-      "loss": 4.2845,
+      "epoch": 0.11,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.2281,
       "step": 66
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 4.3532,
+      "epoch": 0.11,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.1835,
       "step": 67
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 4.2654,
+      "epoch": 0.11,
+      "learning_rate": 1.32e-05,
+      "loss": 4.1648,
       "step": 68
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 4.241,
+      "epoch": 0.12,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.2733,
       "step": 69
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.32e-05,
-      "loss": 4.2837,
+      "epoch": 0.12,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.2692,
       "step": 70
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 4.4424,
+      "epoch": 0.12,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.347,
       "step": 71
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 4.3367,
+      "epoch": 0.12,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.3248,
       "step": 72
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 4.2615,
+      "epoch": 0.12,
+      "learning_rate": 1.42e-05,
+      "loss": 4.1086,
       "step": 73
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 4.3435,
+      "epoch": 0.12,
+      "learning_rate": 1.44e-05,
+      "loss": 4.323,
       "step": 74
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.42e-05,
-      "loss": 4.1861,
+      "epoch": 0.13,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.2992,
       "step": 75
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.44e-05,
-      "loss": 4.5718,
+      "epoch": 0.13,
+      "learning_rate": 1.48e-05,
+      "loss": 4.2775,
       "step": 76
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 4.3487,
+      "epoch": 0.13,
+      "learning_rate": 1.5e-05,
+      "loss": 4.2624,
       "step": 77
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.48e-05,
-      "loss": 4.2016,
+      "epoch": 0.13,
+      "learning_rate": 1.52e-05,
+      "loss": 4.2317,
       "step": 78
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.5e-05,
-      "loss": 4.2025,
+      "epoch": 0.13,
+      "learning_rate": 1.54e-05,
+      "loss": 4.1565,
       "step": 79
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.52e-05,
-      "loss": 4.3344,
+      "epoch": 0.13,
+      "learning_rate": 1.56e-05,
+      "loss": 4.3455,
       "step": 80
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.54e-05,
-      "loss": 4.3465,
+      "epoch": 0.14,
+      "learning_rate": 1.58e-05,
+      "loss": 4.2909,
       "step": 81
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.56e-05,
-      "loss": 4.2836,
+      "epoch": 0.14,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.2639,
       "step": 82
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.58e-05,
-      "loss": 4.1951,
+      "epoch": 0.14,
+      "learning_rate": 1.62e-05,
+      "loss": 4.3352,
       "step": 83
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 4.3508,
+      "epoch": 0.14,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.3157,
       "step": 84
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.62e-05,
-      "loss": 4.2344,
+      "epoch": 0.14,
+      "learning_rate": 1.66e-05,
+      "loss": 4.2394,
       "step": 85
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 4.2478,
+      "epoch": 0.14,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.1759,
       "step": 86
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.66e-05,
-      "loss": 4.1424,
+      "epoch": 0.15,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.4326,
       "step": 87
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 4.3444,
+      "epoch": 0.15,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.2897,
       "step": 88
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 4.3999,
+      "epoch": 0.15,
+      "learning_rate": 1.74e-05,
+      "loss": 4.2653,
       "step": 89
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 4.2999,
+      "epoch": 0.15,
+      "learning_rate": 1.76e-05,
+      "loss": 4.3216,
       "step": 90
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.74e-05,
-      "loss": 4.3622,
+      "epoch": 0.15,
+      "learning_rate": 1.78e-05,
+      "loss": 4.2713,
       "step": 91
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.76e-05,
-      "loss": 4.4123,
+      "epoch": 0.15,
+      "learning_rate": 1.8e-05,
+      "loss": 4.3816,
       "step": 92
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.78e-05,
-      "loss": 4.025,
+      "epoch": 0.16,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.2577,
       "step": 93
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8e-05,
-      "loss": 4.408,
+      "epoch": 0.16,
+      "learning_rate": 1.84e-05,
+      "loss": 4.4847,
       "step": 94
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 4.5952,
+      "epoch": 0.16,
+      "learning_rate": 1.86e-05,
+      "loss": 4.3693,
       "step": 95
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.84e-05,
-      "loss": 4.407,
+      "epoch": 0.16,
+      "learning_rate": 1.88e-05,
+      "loss": 4.4311,
       "step": 96
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.86e-05,
-      "loss": 4.3912,
+      "epoch": 0.16,
+      "learning_rate": 1.9e-05,
+      "loss": 4.4079,
       "step": 97
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.88e-05,
-      "loss": 4.5389,
+      "epoch": 0.16,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.2798,
       "step": 98
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9e-05,
-      "loss": 4.5746,
+      "epoch": 0.17,
+      "learning_rate": 1.94e-05,
+      "loss": 4.5225,
       "step": 99
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 4.289,
+      "epoch": 0.17,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.2098,
       "step": 100
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.94e-05,
-      "loss": 4.033,
+      "epoch": 0.17,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.3498,
       "step": 101
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 4.3666,
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 4.1786,
       "step": 102
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 4.2106,
+      "epoch": 0.17,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.1771,
       "step": 103
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2e-05,
-      "loss": 4.2842,
+      "epoch": 0.17,
+      "learning_rate": 2.04e-05,
+      "loss": 4.1286,
       "step": 104
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 4.2092,
+      "epoch": 0.18,
+      "learning_rate": 2.06e-05,
+      "loss": 4.1734,
       "step": 105
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.04e-05,
-      "loss": 4.3428,
+      "epoch": 0.18,
+      "learning_rate": 2.08e-05,
+      "loss": 4.1577,
       "step": 106
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.06e-05,
-      "loss": 4.4373,
+      "epoch": 0.18,
+      "learning_rate": 2.1e-05,
+      "loss": 4.2961,
       "step": 107
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.08e-05,
-      "loss": 4.3963,
+      "epoch": 0.18,
+      "learning_rate": 2.12e-05,
+      "loss": 4.3128,
       "step": 108
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1e-05,
-      "loss": 4.3324,
+      "epoch": 0.18,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.395,
       "step": 109
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.12e-05,
-      "loss": 4.0756,
+      "epoch": 0.18,
+      "learning_rate": 2.16e-05,
+      "loss": 4.207,
       "step": 110
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 4.353,
+      "epoch": 0.19,
+      "learning_rate": 2.18e-05,
+      "loss": 4.199,
       "step": 111
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.16e-05,
-      "loss": 4.4118,
+      "epoch": 0.19,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.2547,
       "step": 112
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.18e-05,
-      "loss": 4.208,
+      "epoch": 0.19,
+      "learning_rate": 2.22e-05,
+      "loss": 4.2414,
       "step": 113
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 4.354,
+      "epoch": 0.19,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.2794,
       "step": 114
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.22e-05,
-      "loss": 4.2994,
+      "epoch": 0.19,
+      "learning_rate": 2.26e-05,
+      "loss": 4.2513,
       "step": 115
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 4.4316,
+      "epoch": 0.2,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.2407,
       "step": 116
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.26e-05,
-      "loss": 4.1913,
+      "epoch": 0.2,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.3321,
       "step": 117
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 4.059,
+      "epoch": 0.2,
+      "learning_rate": 2.32e-05,
+      "loss": 4.1167,
       "step": 118
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 4.1587,
+      "epoch": 0.2,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.108,
       "step": 119
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.32e-05,
-      "loss": 4.2109,
+      "epoch": 0.2,
+      "learning_rate": 2.36e-05,
+      "loss": 4.102,
       "step": 120
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 4.5004,
+      "epoch": 0.2,
+      "learning_rate": 2.38e-05,
+      "loss": 4.3165,
       "step": 121
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.36e-05,
-      "loss": 4.2103,
+      "epoch": 0.21,
+      "learning_rate": 2.4e-05,
+      "loss": 4.4299,
       "step": 122
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.38e-05,
-      "loss": 4.3288,
+      "epoch": 0.21,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.1823,
       "step": 123
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.4e-05,
-      "loss": 4.3284,
+      "epoch": 0.21,
+      "learning_rate": 2.44e-05,
+      "loss": 4.2409,
       "step": 124
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 4.0892,
+      "epoch": 0.21,
+      "learning_rate": 2.46e-05,
+      "loss": 4.2589,
       "step": 125
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.44e-05,
-      "loss": 4.268,
+      "epoch": 0.21,
+      "learning_rate": 2.48e-05,
+      "loss": 4.0746,
       "step": 126
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.46e-05,
-      "loss": 4.458,
+      "epoch": 0.21,
+      "learning_rate": 2.5e-05,
+      "loss": 4.1253,
       "step": 127
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.48e-05,
-      "loss": 4.113,
+      "epoch": 0.22,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.1079,
       "step": 128
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5e-05,
-      "loss": 4.0363,
+      "epoch": 0.22,
+      "learning_rate": 2.54e-05,
+      "loss": 4.237,
       "step": 129
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 4.3122,
+      "epoch": 0.22,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.2734,
       "step": 130
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.54e-05,
-      "loss": 4.2714,
+      "epoch": 0.22,
+      "learning_rate": 2.58e-05,
+      "loss": 4.2196,
       "step": 131
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 4.2019,
+      "epoch": 0.22,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.1854,
       "step": 132
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.58e-05,
-      "loss": 4.4254,
+      "epoch": 0.22,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.253,
       "step": 133
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 4.3042,
+      "epoch": 0.23,
+      "learning_rate": 2.64e-05,
+      "loss": 4.3073,
       "step": 134
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 4.4085,
+      "epoch": 0.23,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.3313,
       "step": 135
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.64e-05,
-      "loss": 4.1049,
+      "epoch": 0.23,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.1854,
       "step": 136
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 4.3965,
+      "epoch": 0.23,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.1861,
       "step": 137
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 4.3541,
+      "epoch": 0.23,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.2034,
       "step": 138
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 4.1298,
+      "epoch": 0.23,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.217,
       "step": 139
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 4.3547,
+      "epoch": 0.24,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.3816,
       "step": 140
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 4.3903,
+      "epoch": 0.24,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.2944,
       "step": 141
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 4.1986,
+      "epoch": 0.24,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.3615,
       "step": 142
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 4.1182,
+      "epoch": 0.24,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.4018,
       "step": 143
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 4.494,
+      "epoch": 0.24,
+      "learning_rate": 2.84e-05,
+      "loss": 4.2217,
       "step": 144
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.2511,
+      "epoch": 0.24,
+      "learning_rate": 2.86e-05,
+      "loss": 4.2796,
       "step": 145
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.84e-05,
-      "loss": 4.5495,
+      "epoch": 0.25,
+      "learning_rate": 2.88e-05,
+      "loss": 4.612,
       "step": 146
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.86e-05,
-      "loss": 4.3014,
+      "epoch": 0.25,
+      "learning_rate": 2.9e-05,
+      "loss": 4.5694,
       "step": 147
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.88e-05,
-      "loss": 4.1714,
+      "epoch": 0.25,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.5402,
       "step": 148
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.9e-05,
-      "loss": 4.4194,
+      "epoch": 0.25,
+      "learning_rate": 2.94e-05,
+      "loss": 4.5132,
       "step": 149
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 4.1878,
+      "epoch": 0.25,
+      "learning_rate": 2.96e-05,
+      "loss": 4.564,
       "step": 150
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.94e-05,
-      "loss": 4.2826,
+      "epoch": 0.25,
+      "learning_rate": 2.98e-05,
+      "loss": 4.217,
       "step": 151
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.96e-05,
-      "loss": 4.1653,
+      "epoch": 0.26,
+      "learning_rate": 3e-05,
+      "loss": 4.1436,
       "step": 152
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.98e-05,
-      "loss": 4.2249,
+      "epoch": 0.26,
+      "learning_rate": 3.02e-05,
+      "loss": 4.1186,
       "step": 153
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3e-05,
-      "loss": 4.3612,
+      "epoch": 0.26,
+      "learning_rate": 3.04e-05,
+      "loss": 4.0641,
       "step": 154
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.02e-05,
-      "loss": 4.2347,
+      "epoch": 0.26,
+      "learning_rate": 3.06e-05,
+      "loss": 4.0898,
       "step": 155
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.04e-05,
-      "loss": 4.0843,
+      "epoch": 0.26,
+      "learning_rate": 3.08e-05,
+      "loss": 4.1023,
       "step": 156
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.06e-05,
-      "loss": 4.2469,
+      "epoch": 0.26,
+      "learning_rate": 3.1e-05,
+      "loss": 4.2444,
       "step": 157
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.08e-05,
-      "loss": 4.4524,
+      "epoch": 0.27,
+      "learning_rate": 3.12e-05,
+      "loss": 4.0765,
       "step": 158
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.1e-05,
-      "loss": 4.1758,
+      "epoch": 0.27,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.0442,
       "step": 159
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.12e-05,
-      "loss": 3.9497,
+      "epoch": 0.27,
+      "learning_rate": 3.16e-05,
+      "loss": 4.1485,
       "step": 160
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 4.2799,
+      "epoch": 0.27,
+      "learning_rate": 3.18e-05,
+      "loss": 4.1854,
       "step": 161
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.16e-05,
-      "loss": 4.3005,
+      "epoch": 0.27,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.0798,
       "step": 162
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.18e-05,
-      "loss": 4.2474,
+      "epoch": 0.27,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.1013,
       "step": 163
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 4.0124,
+      "epoch": 0.28,
+      "learning_rate": 3.24e-05,
+      "loss": 4.1308,
       "step": 164
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 4.2493,
+      "epoch": 0.28,
+      "learning_rate": 3.26e-05,
+      "loss": 4.1663,
       "step": 165
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.24e-05,
-      "loss": 3.8951,
+      "epoch": 0.28,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.2826,
       "step": 166
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.26e-05,
-      "loss": 4.2486,
+      "epoch": 0.28,
+      "learning_rate": 3.3e-05,
+      "loss": 4.12,
       "step": 167
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 4.1281,
+      "epoch": 0.28,
+      "learning_rate": 3.32e-05,
+      "loss": 4.3403,
       "step": 168
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3e-05,
-      "loss": 4.3356,
+      "epoch": 0.28,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.1641,
       "step": 169
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.32e-05,
-      "loss": 4.2058,
+      "epoch": 0.29,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.2228,
       "step": 170
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 4.2742,
+      "epoch": 0.29,
+      "learning_rate": 3.38e-05,
+      "loss": 4.0751,
       "step": 171
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 4.1186,
+      "epoch": 0.29,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.1022,
       "step": 172
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.38e-05,
-      "loss": 4.0249,
+      "epoch": 0.29,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.1089,
       "step": 173
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.3921,
+      "epoch": 0.29,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.1186,
       "step": 174
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 4.2444,
+      "epoch": 0.29,
+      "learning_rate": 3.46e-05,
+      "loss": 4.239,
       "step": 175
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 4.2484,
+      "epoch": 0.3,
+      "learning_rate": 3.48e-05,
+      "loss": 4.2725,
       "step": 176
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.46e-05,
-      "loss": 4.3264,
+      "epoch": 0.3,
+      "learning_rate": 3.5e-05,
+      "loss": 4.2117,
       "step": 177
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.48e-05,
-      "loss": 4.1729,
+      "epoch": 0.3,
+      "learning_rate": 3.52e-05,
+      "loss": 4.1026,
       "step": 178
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.5e-05,
-      "loss": 4.2341,
+      "epoch": 0.3,
+      "learning_rate": 3.54e-05,
+      "loss": 4.2162,
       "step": 179
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.52e-05,
-      "loss": 4.3517,
+      "epoch": 0.3,
+      "learning_rate": 3.56e-05,
+      "loss": 4.094,
       "step": 180
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.54e-05,
-      "loss": 4.3434,
+      "epoch": 0.3,
+      "learning_rate": 3.58e-05,
+      "loss": 4.1863,
       "step": 181
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.56e-05,
-      "loss": 4.3427,
+      "epoch": 0.31,
+      "learning_rate": 3.6e-05,
+      "loss": 4.1617,
       "step": 182
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.58e-05,
-      "loss": 4.1891,
+      "epoch": 0.31,
+      "learning_rate": 3.62e-05,
+      "loss": 4.1582,
       "step": 183
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.6e-05,
-      "loss": 4.2859,
+      "epoch": 0.31,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.0647,
       "step": 184
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.62e-05,
-      "loss": 4.0825,
+      "epoch": 0.31,
+      "learning_rate": 3.66e-05,
+      "loss": 4.4047,
       "step": 185
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 4.198,
+      "epoch": 0.31,
+      "learning_rate": 3.68e-05,
+      "loss": 4.1967,
       "step": 186
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.66e-05,
-      "loss": 4.4674,
+      "epoch": 0.31,
+      "learning_rate": 3.7e-05,
+      "loss": 4.2453,
       "step": 187
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.68e-05,
-      "loss": 4.2519,
+      "epoch": 0.32,
+      "learning_rate": 3.72e-05,
+      "loss": 4.2804,
       "step": 188
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7e-05,
-      "loss": 4.4296,
+      "epoch": 0.32,
+      "learning_rate": 3.74e-05,
+      "loss": 4.426,
       "step": 189
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.72e-05,
-      "loss": 4.0456,
+      "epoch": 0.32,
+      "learning_rate": 3.76e-05,
+      "loss": 4.1321,
       "step": 190
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.74e-05,
-      "loss": 4.3583,
+      "epoch": 0.32,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.4128,
       "step": 191
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.76e-05,
-      "loss": 4.546,
+      "epoch": 0.32,
+      "learning_rate": 3.8e-05,
+      "loss": 4.3332,
       "step": 192
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 4.4776,
+      "epoch": 0.32,
+      "learning_rate": 3.82e-05,
+      "loss": 4.329,
       "step": 193
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.8e-05,
-      "loss": 4.2351,
+      "epoch": 0.33,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.3248,
       "step": 194
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.82e-05,
-      "loss": 4.452,
+      "epoch": 0.33,
+      "learning_rate": 3.86e-05,
+      "loss": 4.2567,
       "step": 195
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 4.3113,
+      "epoch": 0.33,
+      "learning_rate": 3.88e-05,
+      "loss": 4.1483,
       "step": 196
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.86e-05,
-      "loss": 4.4017,
+      "epoch": 0.33,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.5865,
       "step": 197
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.88e-05,
-      "loss": 4.5343,
+      "epoch": 0.33,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.5307,
       "step": 198
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 4.5779,
+      "epoch": 0.33,
+      "learning_rate": 3.94e-05,
+      "loss": 4.5087,
       "step": 199
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 4.3443,
+      "epoch": 0.34,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.5711,
       "step": 200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.94e-05,
-      "loss": 4.3128,
+      "epoch": 0.34,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.0395,
       "step": 201
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 4.2181,
+      "epoch": 0.34,
+      "learning_rate": 4e-05,
+      "loss": 4.1998,
       "step": 202
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 4.3269,
+      "epoch": 0.34,
+      "learning_rate": 4.02e-05,
+      "loss": 4.2598,
       "step": 203
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4e-05,
-      "loss": 4.1791,
+      "epoch": 0.34,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.0994,
       "step": 204
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.02e-05,
-      "loss": 4.1472,
+      "epoch": 0.34,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.1703,
       "step": 205
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 4.2105,
+      "epoch": 0.35,
+      "learning_rate": 4.08e-05,
+      "loss": 4.1169,
       "step": 206
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 4.2759,
+      "epoch": 0.35,
+      "learning_rate": 4.1e-05,
+      "loss": 4.1565,
       "step": 207
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.08e-05,
-      "loss": 4.5245,
+      "epoch": 0.35,
+      "learning_rate": 4.12e-05,
+      "loss": 4.1541,
       "step": 208
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1e-05,
-      "loss": 4.2966,
+      "epoch": 0.35,
+      "learning_rate": 4.14e-05,
+      "loss": 4.1809,
       "step": 209
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.12e-05,
-      "loss": 4.4305,
+      "epoch": 0.35,
+      "learning_rate": 4.16e-05,
+      "loss": 4.2621,
       "step": 210
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.14e-05,
-      "loss": 4.272,
+      "epoch": 0.35,
+      "learning_rate": 4.18e-05,
+      "loss": 4.3034,
       "step": 211
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.16e-05,
-      "loss": 4.3508,
+      "epoch": 0.36,
+      "learning_rate": 4.2e-05,
+      "loss": 4.2205,
       "step": 212
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.18e-05,
-      "loss": 4.2033,
+      "epoch": 0.36,
+      "learning_rate": 4.22e-05,
+      "loss": 4.1836,
       "step": 213
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.2e-05,
-      "loss": 4.2488,
+      "epoch": 0.36,
+      "learning_rate": 4.24e-05,
+      "loss": 4.2609,
       "step": 214
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.22e-05,
-      "loss": 4.3297,
+      "epoch": 0.36,
+      "learning_rate": 4.26e-05,
+      "loss": 4.1626,
       "step": 215
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.24e-05,
-      "loss": 4.2388,
+      "epoch": 0.36,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.1637,
       "step": 216
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.26e-05,
-      "loss": 4.3794,
+      "epoch": 0.36,
+      "learning_rate": 4.3e-05,
+      "loss": 4.0815,
       "step": 217
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 4.2471,
+      "epoch": 0.37,
+      "learning_rate": 4.32e-05,
+      "loss": 4.1643,
       "step": 218
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.3e-05,
-      "loss": 3.9721,
+      "epoch": 0.37,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.1492,
       "step": 219
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.32e-05,
-      "loss": 4.2809,
+      "epoch": 0.37,
+      "learning_rate": 4.36e-05,
+      "loss": 4.2274,
       "step": 220
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 4.198,
+      "epoch": 0.37,
+      "learning_rate": 4.38e-05,
+      "loss": 4.2597,
       "step": 221
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.36e-05,
-      "loss": 4.5828,
+      "epoch": 0.37,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.2746,
       "step": 222
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.38e-05,
-      "loss": 4.36,
+      "epoch": 0.37,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.0996,
       "step": 223
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 4.3545,
+      "epoch": 0.38,
+      "learning_rate": 4.44e-05,
+      "loss": 4.2349,
       "step": 224
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 4.1133,
+      "epoch": 0.38,
+      "learning_rate": 4.46e-05,
+      "loss": 4.4081,
       "step": 225
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.44e-05,
-      "loss": 4.0773,
+      "epoch": 0.38,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 4.1651,
       "step": 226
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.46e-05,
-      "loss": 4.2645,
+      "epoch": 0.38,
+      "learning_rate": 4.5e-05,
+      "loss": 4.097,
       "step": 227
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 4.2319,
+      "epoch": 0.38,
+      "learning_rate": 4.52e-05,
+      "loss": 4.1481,
       "step": 228
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.5e-05,
-      "loss": 4.1472,
+      "epoch": 0.39,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.3052,
       "step": 229
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.52e-05,
-      "loss": 4.3393,
+      "epoch": 0.39,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.2498,
       "step": 230
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 4.3586,
+      "epoch": 0.39,
+      "learning_rate": 4.58e-05,
+      "loss": 4.1329,
       "step": 231
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 4.2482,
+      "epoch": 0.39,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.2091,
       "step": 232
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.58e-05,
-      "loss": 4.2654,
+      "epoch": 0.39,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.3565,
       "step": 233
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 4.1693,
+      "epoch": 0.39,
+      "learning_rate": 4.64e-05,
+      "loss": 4.1954,
       "step": 234
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 4.2974,
+      "epoch": 0.4,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.3483,
       "step": 235
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.64e-05,
-      "loss": 4.4114,
+      "epoch": 0.4,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.1283,
       "step": 236
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 4.1475,
+      "epoch": 0.4,
+      "learning_rate": 4.7e-05,
+      "loss": 4.0929,
       "step": 237
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 4.2295,
+      "epoch": 0.4,
+      "learning_rate": 4.72e-05,
+      "loss": 4.1404,
       "step": 238
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.7e-05,
-      "loss": 4.2803,
+      "epoch": 0.4,
+      "learning_rate": 4.74e-05,
+      "loss": 4.0962,
       "step": 239
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.72e-05,
-      "loss": 4.3166,
+      "epoch": 0.4,
+      "learning_rate": 4.76e-05,
+      "loss": 4.2646,
       "step": 240
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.74e-05,
-      "loss": 4.4425,
+      "epoch": 0.41,
+      "learning_rate": 4.78e-05,
+      "loss": 4.186,
       "step": 241
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.76e-05,
-      "loss": 4.1783,
+      "epoch": 0.41,
+      "learning_rate": 4.8e-05,
+      "loss": 4.2541,
       "step": 242
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.78e-05,
-      "loss": 4.4353,
-      "step": 243
-    },
+      "epoch": 0.41,
+      "learning_rate": 4.82e-05,
+      "loss": 4.1949,
+      "step": 243
+    },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.8e-05,
-      "loss": 4.2845,
+      "epoch": 0.41,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.3293,
       "step": 244
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.82e-05,
-      "loss": 4.424,
+      "epoch": 0.41,
+      "learning_rate": 4.86e-05,
+      "loss": 4.477,
       "step": 245
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 4.1741,
+      "epoch": 0.41,
+      "learning_rate": 4.88e-05,
+      "loss": 4.4003,
       "step": 246
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.86e-05,
-      "loss": 4.9749,
+      "epoch": 0.42,
+      "learning_rate": 4.9e-05,
+      "loss": 4.5343,
       "step": 247
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.88e-05,
-      "loss": 4.4119,
+      "epoch": 0.42,
+      "learning_rate": 4.92e-05,
+      "loss": 4.5222,
       "step": 248
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.9e-05,
-      "loss": 4.5577,
+      "epoch": 0.42,
+      "learning_rate": 4.94e-05,
+      "loss": 4.4392,
       "step": 249
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.92e-05,
-      "loss": 4.4571,
+      "epoch": 0.42,
+      "learning_rate": 4.96e-05,
+      "loss": 4.5092,
       "step": 250
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.94e-05,
-      "loss": 4.4127,
+      "epoch": 0.42,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.2232,
       "step": 251
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.96e-05,
-      "loss": 4.0991,
+      "epoch": 0.42,
+      "learning_rate": 5e-05,
+      "loss": 4.2222,
       "step": 252
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 4.4255,
+      "epoch": 0.43,
+      "learning_rate": 5.02e-05,
+      "loss": 4.2069,
       "step": 253
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5e-05,
-      "loss": 4.0798,
+      "epoch": 0.43,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.1118,
       "step": 254
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.02e-05,
-      "loss": 4.0992,
+      "epoch": 0.43,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.2737,
       "step": 255
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 4.2249,
+      "epoch": 0.43,
+      "learning_rate": 5.08e-05,
+      "loss": 4.1708,
       "step": 256
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 4.0997,
+      "epoch": 0.43,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.1226,
       "step": 257
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.08e-05,
-      "loss": 4.5592,
+      "epoch": 0.43,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 4.1544,
       "step": 258
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 4.2932,
+      "epoch": 0.44,
+      "learning_rate": 5.14e-05,
+      "loss": 4.2259,
       "step": 259
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 4.1998,
+      "epoch": 0.44,
+      "learning_rate": 5.16e-05,
+      "loss": 4.2689,
       "step": 260
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.14e-05,
-      "loss": 4.2676,
+      "epoch": 0.44,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.172,
       "step": 261
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.16e-05,
-      "loss": 4.2827,
+      "epoch": 0.44,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.1095,
       "step": 262
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 4.4494,
+      "epoch": 0.44,
+      "learning_rate": 5.22e-05,
+      "loss": 4.2901,
       "step": 263
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 4.3267,
+      "epoch": 0.44,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 3.9607,
       "step": 264
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.22e-05,
-      "loss": 4.253,
+      "epoch": 0.45,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.0989,
       "step": 265
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 4.3077,
+      "epoch": 0.45,
+      "learning_rate": 5.28e-05,
+      "loss": 4.1631,
       "step": 266
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 4.1922,
+      "epoch": 0.45,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.1817,
       "step": 267
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.28e-05,
-      "loss": 4.2516,
+      "epoch": 0.45,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.0973,
       "step": 268
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 4.239,
+      "epoch": 0.45,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 4.2258,
       "step": 269
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 4.1309,
+      "epoch": 0.45,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 4.2652,
       "step": 270
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 4.4683,
+      "epoch": 0.46,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 4.114,
       "step": 271
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 4.1969,
+      "epoch": 0.46,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.2871,
       "step": 272
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 4.217,
+      "epoch": 0.46,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 4.1757,
       "step": 273
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 4.2796,
+      "epoch": 0.46,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 4.1485,
       "step": 274
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 4.3009,
+      "epoch": 0.46,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 4.2358,
       "step": 275
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 4.079,
+      "epoch": 0.46,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 4.1717,
       "step": 276
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 4.109,
+      "epoch": 0.47,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.281,
       "step": 277
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 4.2593,
+      "epoch": 0.47,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 4.1168,
       "step": 278
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 4.3454,
+      "epoch": 0.47,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 4.301,
       "step": 279
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 4.2799,
+      "epoch": 0.47,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 4.3667,
       "step": 280
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 4.2545,
+      "epoch": 0.47,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 4.3479,
       "step": 281
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 4.3556,
+      "epoch": 0.47,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.2431,
       "step": 282
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 4.4132,
+      "epoch": 0.48,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 4.1512,
       "step": 283
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 4.1269,
+      "epoch": 0.48,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.1295,
       "step": 284
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 4.3881,
+      "epoch": 0.48,
+      "learning_rate": 5.66e-05,
+      "loss": 4.3472,
       "step": 285
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.1672,
+      "epoch": 0.48,
+      "learning_rate": 5.68e-05,
+      "loss": 4.0996,
       "step": 286
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.66e-05,
-      "loss": 4.1611,
+      "epoch": 0.48,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.5036,
       "step": 287
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.68e-05,
-      "loss": 4.2614,
+      "epoch": 0.48,
+      "learning_rate": 5.72e-05,
+      "loss": 3.9666,
       "step": 288
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.443,
+      "epoch": 0.49,
+      "learning_rate": 5.74e-05,
+      "loss": 4.2155,
       "step": 289
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.72e-05,
-      "loss": 4.3057,
+      "epoch": 0.49,
+      "learning_rate": 5.76e-05,
+      "loss": 4.2013,
       "step": 290
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 5.74e-05,
-      "loss": 4.3724,
+      "epoch": 0.49,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 4.3886,
       "step": 291
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.76e-05,
-      "loss": 4.6575,
+      "epoch": 0.49,
+      "learning_rate": 5.8e-05,
+      "loss": 4.0919,
       "step": 292
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.1862,
+      "epoch": 0.49,
+      "learning_rate": 5.82e-05,
+      "loss": 4.2544,
       "step": 293
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.8e-05,
-      "loss": 4.5073,
+      "epoch": 0.49,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 4.3885,
       "step": 294
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.82e-05,
-      "loss": 4.419,
+      "epoch": 0.5,
+      "learning_rate": 5.86e-05,
+      "loss": 4.3546,
       "step": 295
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 5.0644,
+      "epoch": 0.5,
+      "learning_rate": 5.88e-05,
+      "loss": 4.1183,
       "step": 296
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.86e-05,
-      "loss": 4.3461,
+      "epoch": 0.5,
+      "learning_rate": 5.9e-05,
+      "loss": 4.4789,
       "step": 297
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.88e-05,
-      "loss": 4.8568,
+      "epoch": 0.5,
+      "learning_rate": 5.92e-05,
+      "loss": 4.6137,
       "step": 298
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.9e-05,
-      "loss": 5.1008,
+      "epoch": 0.5,
+      "learning_rate": 5.94e-05,
+      "loss": 4.5089,
       "step": 299
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.92e-05,
-      "loss": 4.7251,
+      "epoch": 0.5,
+      "learning_rate": 5.96e-05,
+      "loss": 4.6724,
       "step": 300
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.94e-05,
-      "loss": 4.1274,
+      "epoch": 0.51,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 4.3125,
       "step": 301
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.96e-05,
-      "loss": 4.2699,
+      "epoch": 0.51,
+      "learning_rate": 6e-05,
+      "loss": 4.2696,
       "step": 302
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 4.3602,
+      "epoch": 0.51,
+      "learning_rate": 6.02e-05,
+      "loss": 4.3593,
       "step": 303
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6e-05,
-      "loss": 3.9879,
+      "epoch": 0.51,
+      "learning_rate": 6.04e-05,
+      "loss": 4.0906,
       "step": 304
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.02e-05,
-      "loss": 4.1636,
+      "epoch": 0.51,
+      "learning_rate": 6.06e-05,
+      "loss": 4.2907,
       "step": 305
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.04e-05,
-      "loss": 4.1518,
+      "epoch": 0.51,
+      "learning_rate": 6.08e-05,
+      "loss": 4.3188,
       "step": 306
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.06e-05,
-      "loss": 4.226,
+      "epoch": 0.52,
+      "learning_rate": 6.1e-05,
+      "loss": 4.0912,
       "step": 307
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.08e-05,
-      "loss": 4.0587,
+      "epoch": 0.52,
+      "learning_rate": 6.12e-05,
+      "loss": 4.284,
       "step": 308
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.1e-05,
-      "loss": 3.9997,
+      "epoch": 0.52,
+      "learning_rate": 6.14e-05,
+      "loss": 4.2691,
       "step": 309
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.12e-05,
-      "loss": 4.2045,
+      "epoch": 0.52,
+      "learning_rate": 6.16e-05,
+      "loss": 4.2903,
       "step": 310
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.14e-05,
-      "loss": 4.3838,
+      "epoch": 0.52,
+      "learning_rate": 6.18e-05,
+      "loss": 4.1323,
       "step": 311
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.16e-05,
-      "loss": 4.1267,
+      "epoch": 0.52,
+      "learning_rate": 6.2e-05,
+      "loss": 4.1224,
       "step": 312
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.18e-05,
-      "loss": 4.089,
+      "epoch": 0.53,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 4.2004,
       "step": 313
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.2e-05,
-      "loss": 4.3035,
+      "epoch": 0.53,
+      "learning_rate": 6.24e-05,
+      "loss": 4.2083,
       "step": 314
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 4.0432,
+      "epoch": 0.53,
+      "learning_rate": 6.26e-05,
+      "loss": 4.1996,
       "step": 315
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.24e-05,
-      "loss": 4.3994,
+      "epoch": 0.53,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 4.2615,
       "step": 316
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.26e-05,
-      "loss": 4.4245,
+      "epoch": 0.53,
+      "learning_rate": 6.3e-05,
+      "loss": 4.1322,
       "step": 317
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 4.3869,
+      "epoch": 0.53,
+      "learning_rate": 6.32e-05,
+      "loss": 4.2435,
       "step": 318
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.3e-05,
-      "loss": 4.1928,
+      "epoch": 0.54,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 4.1421,
       "step": 319
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.32e-05,
-      "loss": 4.3573,
+      "epoch": 0.54,
+      "learning_rate": 6.36e-05,
+      "loss": 4.2892,
       "step": 320
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 4.1723,
+      "epoch": 0.54,
+      "learning_rate": 6.38e-05,
+      "loss": 4.0954,
       "step": 321
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.36e-05,
-      "loss": 4.1477,
+      "epoch": 0.54,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 4.1498,
       "step": 322
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.38e-05,
-      "loss": 4.1159,
+      "epoch": 0.54,
+      "learning_rate": 6.42e-05,
+      "loss": 4.2609,
       "step": 323
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 4.3985,
+      "epoch": 0.54,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.2246,
       "step": 324
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.42e-05,
-      "loss": 4.2315,
+      "epoch": 0.55,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 4.1842,
       "step": 325
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 4.6484,
+      "epoch": 0.55,
+      "learning_rate": 6.48e-05,
+      "loss": 4.1417,
       "step": 326
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 4.5978,
+      "epoch": 0.55,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.4157,
       "step": 327
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.48e-05,
-      "loss": 4.4536,
+      "epoch": 0.55,
+      "learning_rate": 6.52e-05,
+      "loss": 4.1353,
       "step": 328
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 4.261,
+      "epoch": 0.55,
+      "learning_rate": 6.54e-05,
+      "loss": 4.0712,
       "step": 329
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.52e-05,
-      "loss": 4.4895,
+      "epoch": 0.55,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 4.2765,
       "step": 330
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.54e-05,
-      "loss": 4.1584,
+      "epoch": 0.56,
+      "learning_rate": 6.58e-05,
+      "loss": 4.263,
       "step": 331
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 4.5309,
+      "epoch": 0.56,
+      "learning_rate": 6.6e-05,
+      "loss": 4.0798,
       "step": 332
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.58e-05,
-      "loss": 4.3519,
+      "epoch": 0.56,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 4.328,
       "step": 333
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.6e-05,
-      "loss": 4.1383,
+      "epoch": 0.56,
+      "learning_rate": 6.64e-05,
+      "loss": 4.2533,
       "step": 334
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 4.6769,
+      "epoch": 0.56,
+      "learning_rate": 6.66e-05,
+      "loss": 4.2726,
       "step": 335
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.64e-05,
-      "loss": 4.5392,
+      "epoch": 0.56,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 4.1403,
       "step": 336
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.66e-05,
-      "loss": 4.2656,
+      "epoch": 0.57,
+      "learning_rate": 6.7e-05,
+      "loss": 4.1282,
       "step": 337
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 4.4932,
+      "epoch": 0.57,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 4.2455,
       "step": 338
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 6.7e-05,
-      "loss": 4.4263,
+      "epoch": 0.57,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 4.1952,
       "step": 339
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 4.3649,
+      "epoch": 0.57,
+      "learning_rate": 6.76e-05,
+      "loss": 4.3327,
       "step": 340
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 4.4397,
+      "epoch": 0.57,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 4.2905,
       "step": 341
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.76e-05,
-      "loss": 4.4268,
+      "epoch": 0.58,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.374,
       "step": 342
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 4.3656,
+      "epoch": 0.58,
+      "learning_rate": 6.82e-05,
+      "loss": 4.487,
       "step": 343
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 4.6366,
+      "epoch": 0.58,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 4.3049,
       "step": 344
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.82e-05,
-      "loss": 4.3914,
+      "epoch": 0.58,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 4.4144,
       "step": 345
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 4.2631,
+      "epoch": 0.58,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 4.1528,
       "step": 346
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 4.8014,
+      "epoch": 0.58,
+      "learning_rate": 6.9e-05,
+      "loss": 4.4872,
       "step": 347
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 4.8968,
+      "epoch": 0.59,
+      "learning_rate": 6.92e-05,
+      "loss": 4.5556,
       "step": 348
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.9e-05,
-      "loss": 4.8371,
+      "epoch": 0.59,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 4.5082,
       "step": 349
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 6.92e-05,
-      "loss": 4.3916,
+      "epoch": 0.59,
+      "learning_rate": 6.96e-05,
+      "loss": 4.6211,
       "step": 350
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 4.6358,
+      "epoch": 0.59,
+      "learning_rate": 6.98e-05,
+      "loss": 4.3283,
       "step": 351
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.96e-05,
-      "loss": 4.2495,
+      "epoch": 0.59,
+      "learning_rate": 7e-05,
+      "loss": 4.2621,
       "step": 352
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 6.98e-05,
-      "loss": 4.3738,
+      "epoch": 0.59,
+      "learning_rate": 7.02e-05,
+      "loss": 4.3229,
       "step": 353
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7e-05,
-      "loss": 4.3225,
+      "epoch": 0.6,
+      "learning_rate": 7.04e-05,
+      "loss": 4.1606,
       "step": 354
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.02e-05,
-      "loss": 4.2455,
+      "epoch": 0.6,
+      "learning_rate": 7.06e-05,
+      "loss": 4.224,
       "step": 355
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.04e-05,
-      "loss": 4.314,
+      "epoch": 0.6,
+      "learning_rate": 7.08e-05,
+      "loss": 4.2125,
       "step": 356
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.06e-05,
-      "loss": 4.3146,
+      "epoch": 0.6,
+      "learning_rate": 7.1e-05,
+      "loss": 4.2523,
       "step": 357
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.08e-05,
-      "loss": 4.113,
+      "epoch": 0.6,
+      "learning_rate": 7.12e-05,
+      "loss": 4.1371,
       "step": 358
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.1e-05,
-      "loss": 4.2994,
+      "epoch": 0.6,
+      "learning_rate": 7.14e-05,
+      "loss": 4.1853,
       "step": 359
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.12e-05,
-      "loss": 4.3986,
+      "epoch": 0.61,
+      "learning_rate": 7.16e-05,
+      "loss": 4.1967,
       "step": 360
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.14e-05,
-      "loss": 4.2533,
+      "epoch": 0.61,
+      "learning_rate": 7.18e-05,
+      "loss": 4.3175,
       "step": 361
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.16e-05,
-      "loss": 4.3614,
+      "epoch": 0.61,
+      "learning_rate": 7.2e-05,
+      "loss": 4.1987,
       "step": 362
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.18e-05,
-      "loss": 4.2621,
+      "epoch": 0.61,
+      "learning_rate": 7.22e-05,
+      "loss": 4.375,
       "step": 363
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.2e-05,
-      "loss": 4.3092,
+      "epoch": 0.61,
+      "learning_rate": 7.24e-05,
+      "loss": 4.1402,
       "step": 364
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.22e-05,
-      "loss": 4.44,
+      "epoch": 0.61,
+      "learning_rate": 7.26e-05,
+      "loss": 4.2573,
       "step": 365
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.24e-05,
-      "loss": 4.5031,
+      "epoch": 0.62,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 4.1376,
       "step": 366
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.26e-05,
-      "loss": 4.1445,
+      "epoch": 0.62,
+      "learning_rate": 7.3e-05,
+      "loss": 4.2101,
       "step": 367
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 4.5989,
+      "epoch": 0.62,
+      "learning_rate": 7.32e-05,
+      "loss": 4.1182,
       "step": 368
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.3e-05,
-      "loss": 4.5223,
+      "epoch": 0.62,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 4.2608,
       "step": 369
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.32e-05,
-      "loss": 4.3656,
+      "epoch": 0.62,
+      "learning_rate": 7.36e-05,
+      "loss": 4.2736,
       "step": 370
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 4.3882,
+      "epoch": 0.62,
+      "learning_rate": 7.38e-05,
+      "loss": 4.1054,
       "step": 371
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.36e-05,
-      "loss": 4.3717,
+      "epoch": 0.63,
+      "learning_rate": 7.4e-05,
+      "loss": 4.2215,
       "step": 372
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.38e-05,
-      "loss": 4.1942,
+      "epoch": 0.63,
+      "learning_rate": 7.42e-05,
+      "loss": 4.0778,
       "step": 373
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.4e-05,
-      "loss": 4.1795,
+      "epoch": 0.63,
+      "learning_rate": 7.44e-05,
+      "loss": 4.1373,
       "step": 374
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.42e-05,
-      "loss": 4.2802,
+      "epoch": 0.63,
+      "learning_rate": 7.46e-05,
+      "loss": 4.3252,
       "step": 375
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.44e-05,
-      "loss": 4.4486,
+      "epoch": 0.63,
+      "learning_rate": 7.48e-05,
+      "loss": 4.316,
       "step": 376
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.46e-05,
-      "loss": 4.258,
+      "epoch": 0.63,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.2559,
       "step": 377
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.48e-05,
-      "loss": 4.3141,
+      "epoch": 0.64,
+      "learning_rate": 7.52e-05,
+      "loss": 4.2918,
       "step": 378
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 4.205,
+      "epoch": 0.64,
+      "learning_rate": 7.54e-05,
+      "loss": 4.0799,
       "step": 379
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.52e-05,
-      "loss": 4.4629,
+      "epoch": 0.64,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 4.3441,
       "step": 380
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.54e-05,
-      "loss": 4.1435,
+      "epoch": 0.64,
+      "learning_rate": 7.58e-05,
+      "loss": 4.2877,
       "step": 381
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 4.2804,
+      "epoch": 0.64,
+      "learning_rate": 7.6e-05,
+      "loss": 4.1236,
       "step": 382
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.58e-05,
-      "loss": 4.0632,
+      "epoch": 0.64,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 4.1902,
       "step": 383
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.6e-05,
-      "loss": 4.4205,
+      "epoch": 0.65,
+      "learning_rate": 7.64e-05,
+      "loss": 4.2293,
       "step": 384
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 4.4089,
+      "epoch": 0.65,
+      "learning_rate": 7.66e-05,
+      "loss": 4.5028,
       "step": 385
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 7.64e-05,
-      "loss": 4.2022,
+      "epoch": 0.65,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 4.2033,
       "step": 386
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.66e-05,
-      "loss": 4.4224,
+      "epoch": 0.65,
+      "learning_rate": 7.7e-05,
+      "loss": 4.3762,
       "step": 387
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 4.2963,
+      "epoch": 0.65,
+      "learning_rate": 7.72e-05,
+      "loss": 4.318,
       "step": 388
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.7e-05,
-      "loss": 4.4236,
+      "epoch": 0.65,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 4.2075,
       "step": 389
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.72e-05,
-      "loss": 4.3769,
+      "epoch": 0.66,
+      "learning_rate": 7.76e-05,
+      "loss": 4.1818,
       "step": 390
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 4.5785,
+      "epoch": 0.66,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 4.3408,
       "step": 391
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.76e-05,
-      "loss": 4.4797,
+      "epoch": 0.66,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 4.3702,
       "step": 392
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 4.4384,
+      "epoch": 0.66,
+      "learning_rate": 7.82e-05,
+      "loss": 4.5948,
       "step": 393
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 4.6547,
+      "epoch": 0.66,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 4.2956,
       "step": 394
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.82e-05,
-      "loss": 4.2065,
+      "epoch": 0.66,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 4.3763,
       "step": 395
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 4.9664,
+      "epoch": 0.67,
+      "learning_rate": 7.88e-05,
+      "loss": 4.3044,
       "step": 396
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 4.8446,
+      "epoch": 0.67,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 4.3955,
       "step": 397
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 7.88e-05,
-      "loss": 4.9567,
+      "epoch": 0.67,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 4.7559,
       "step": 398
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 5.0118,
+      "epoch": 0.67,
+      "learning_rate": 7.94e-05,
+      "loss": 4.4716,
       "step": 399
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 4.7075,
+      "epoch": 0.67,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 4.7723,
       "step": 400
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 7.94e-05,
-      "loss": 4.2394,
+      "epoch": 0.67,
+      "learning_rate": 7.98e-05,
+      "loss": 4.3635,
       "step": 401
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 4.5547,
+      "epoch": 0.68,
+      "learning_rate": 8e-05,
+      "loss": 4.2703,
       "step": 402
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 7.98e-05,
-      "loss": 4.416,
+      "epoch": 0.68,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 4.3672,
       "step": 403
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8e-05,
-      "loss": 4.4052,
+      "epoch": 0.68,
+      "learning_rate": 8.04e-05,
+      "loss": 4.3438,
       "step": 404
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 4.2415,
+      "epoch": 0.68,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 4.1643,
       "step": 405
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.04e-05,
-      "loss": 4.2665,
+      "epoch": 0.68,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 4.2507,
       "step": 406
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 4.1584,
+      "epoch": 0.68,
+      "learning_rate": 8.1e-05,
+      "loss": 4.3051,
       "step": 407
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 4.6275,
+      "epoch": 0.69,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 4.1477,
       "step": 408
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.1e-05,
-      "loss": 4.4738,
+      "epoch": 0.69,
+      "learning_rate": 8.14e-05,
+      "loss": 4.1829,
       "step": 409
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 4.4424,
+      "epoch": 0.69,
+      "learning_rate": 8.16e-05,
+      "loss": 4.1823,
       "step": 410
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.14e-05,
-      "loss": 4.6931,
+      "epoch": 0.69,
+      "learning_rate": 8.18e-05,
+      "loss": 4.1373,
       "step": 411
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.16e-05,
-      "loss": 4.3011,
+      "epoch": 0.69,
+      "learning_rate": 8.2e-05,
+      "loss": 4.3487,
       "step": 412
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.18e-05,
-      "loss": 4.5431,
+      "epoch": 0.69,
+      "learning_rate": 8.22e-05,
+      "loss": 4.2404,
       "step": 413
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.2e-05,
-      "loss": 4.1608,
+      "epoch": 0.7,
+      "learning_rate": 8.24e-05,
+      "loss": 4.3134,
       "step": 414
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.22e-05,
-      "loss": 4.4028,
+      "epoch": 0.7,
+      "learning_rate": 8.26e-05,
+      "loss": 4.2718,
       "step": 415
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.24e-05,
-      "loss": 4.4587,
+      "epoch": 0.7,
+      "learning_rate": 8.28e-05,
+      "loss": 4.1852,
       "step": 416
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.26e-05,
-      "loss": 3.9879,
+      "epoch": 0.7,
+      "learning_rate": 8.3e-05,
+      "loss": 4.2251,
       "step": 417
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.28e-05,
-      "loss": 4.2723,
+      "epoch": 0.7,
+      "learning_rate": 8.32e-05,
+      "loss": 4.0132,
       "step": 418
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.3e-05,
-      "loss": 4.5346,
+      "epoch": 0.7,
+      "learning_rate": 8.34e-05,
+      "loss": 4.4129,
       "step": 419
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.32e-05,
-      "loss": 4.2373,
+      "epoch": 0.71,
+      "learning_rate": 8.36e-05,
+      "loss": 4.2629,
       "step": 420
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.34e-05,
-      "loss": 4.5587,
+      "epoch": 0.71,
+      "learning_rate": 8.38e-05,
+      "loss": 4.0219,
       "step": 421
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 8.36e-05,
-      "loss": 4.3363,
+      "epoch": 0.71,
+      "learning_rate": 8.4e-05,
+      "loss": 4.1473,
       "step": 422
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.38e-05,
-      "loss": 4.4376,
+      "epoch": 0.71,
+      "learning_rate": 8.42e-05,
+      "loss": 4.3076,
       "step": 423
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.4e-05,
-      "loss": 4.2965,
+      "epoch": 0.71,
+      "learning_rate": 8.44e-05,
+      "loss": 4.2272,
       "step": 424
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.42e-05,
-      "loss": 4.6611,
+      "epoch": 0.71,
+      "learning_rate": 8.46e-05,
+      "loss": 4.2579,
       "step": 425
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.44e-05,
-      "loss": 4.3474,
+      "epoch": 0.72,
+      "learning_rate": 8.48e-05,
+      "loss": 4.3549,
       "step": 426
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.46e-05,
-      "loss": 4.1519,
+      "epoch": 0.72,
+      "learning_rate": 8.5e-05,
+      "loss": 4.1541,
       "step": 427
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.48e-05,
-      "loss": 4.172,
+      "epoch": 0.72,
+      "learning_rate": 8.52e-05,
+      "loss": 4.1218,
       "step": 428
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.5e-05,
-      "loss": 4.701,
+      "epoch": 0.72,
+      "learning_rate": 8.54e-05,
+      "loss": 4.2926,
       "step": 429
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.52e-05,
-      "loss": 4.7364,
+      "epoch": 0.72,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.2178,
       "step": 430
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.54e-05,
-      "loss": 4.3116,
+      "epoch": 0.72,
+      "learning_rate": 8.58e-05,
+      "loss": 4.341,
       "step": 431
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 4.282,
+      "epoch": 0.73,
+      "learning_rate": 8.6e-05,
+      "loss": 4.0324,
       "step": 432
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.58e-05,
-      "loss": 4.6155,
+      "epoch": 0.73,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.295,
       "step": 433
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 8.6e-05,
-      "loss": 4.2399,
+      "epoch": 0.73,
+      "learning_rate": 8.64e-05,
+      "loss": 4.3282,
       "step": 434
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 4.4376,
+      "epoch": 0.73,
+      "learning_rate": 8.66e-05,
+      "loss": 4.5126,
       "step": 435
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.64e-05,
-      "loss": 4.215,
+      "epoch": 0.73,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.2555,
       "step": 436
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.66e-05,
-      "loss": 4.1807,
+      "epoch": 0.73,
+      "learning_rate": 8.7e-05,
+      "loss": 4.1247,
       "step": 437
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 4.1934,
+      "epoch": 0.74,
+      "learning_rate": 8.72e-05,
+      "loss": 4.2901,
       "step": 438
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.7e-05,
-      "loss": 4.4575,
+      "epoch": 0.74,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 4.2579,
       "step": 439
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.72e-05,
-      "loss": 4.364,
+      "epoch": 0.74,
+      "learning_rate": 8.76e-05,
+      "loss": 4.4895,
       "step": 440
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 5.0437,
+      "epoch": 0.74,
+      "learning_rate": 8.78e-05,
+      "loss": 4.373,
       "step": 441
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.76e-05,
-      "loss": 4.4172,
+      "epoch": 0.74,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.6938,
       "step": 442
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.78e-05,
-      "loss": 4.266,
+      "epoch": 0.74,
+      "learning_rate": 8.82e-05,
+      "loss": 4.5459,
       "step": 443
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 4.362,
+      "epoch": 0.75,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.333,
       "step": 444
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.82e-05,
-      "loss": 4.3234,
+      "epoch": 0.75,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.2817,
       "step": 445
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 4.5212,
+      "epoch": 0.75,
+      "learning_rate": 8.86e-05,
+      "loss": 4.4313,
       "step": 446
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.86e-05,
-      "loss": 5.3969,
+      "epoch": 0.75,
+      "learning_rate": 8.88e-05,
+      "loss": 4.4515,
       "step": 447
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.88e-05,
-      "loss": 4.5906,
+      "epoch": 0.75,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 4.3566,
       "step": 448
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 4.5637,
+      "epoch": 0.75,
+      "learning_rate": 8.92e-05,
+      "loss": 4.583,
       "step": 449
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.92e-05,
-      "loss": 4.6229,
+      "epoch": 0.76,
+      "learning_rate": 8.94e-05,
+      "loss": 4.3883,
       "step": 450
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.94e-05,
-      "loss": 4.1351,
+      "epoch": 0.76,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 4.308,
       "step": 451
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 4.1728,
+      "epoch": 0.76,
+      "learning_rate": 8.98e-05,
+      "loss": 4.4201,
       "step": 452
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 8.98e-05,
-      "loss": 4.3363,
+      "epoch": 0.76,
+      "learning_rate": 9e-05,
+      "loss": 4.2154,
       "step": 453
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9e-05,
-      "loss": 4.2729,
+      "epoch": 0.76,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 4.1942,
       "step": 454
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 4.5894,
+      "epoch": 0.77,
+      "learning_rate": 9.04e-05,
+      "loss": 4.2917,
       "step": 455
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.04e-05,
-      "loss": 4.7324,
+      "epoch": 0.77,
+      "learning_rate": 9.06e-05,
+      "loss": 4.2434,
       "step": 456
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.06e-05,
-      "loss": 4.4036,
+      "epoch": 0.77,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 4.4964,
       "step": 457
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 4.8129,
+      "epoch": 0.77,
+      "learning_rate": 9.1e-05,
+      "loss": 4.3874,
       "step": 458
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.1e-05,
-      "loss": 4.2551,
+      "epoch": 0.77,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 4.3414,
       "step": 459
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 4.3308,
+      "epoch": 0.77,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 4.3112,
       "step": 460
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 4.4474,
+      "epoch": 0.78,
+      "learning_rate": 9.16e-05,
+      "loss": 4.2885,
       "step": 461
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.16e-05,
-      "loss": 4.4537,
+      "epoch": 0.78,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 4.1744,
       "step": 462
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 4.2873,
+      "epoch": 0.78,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 4.174,
       "step": 463
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 4.5693,
+      "epoch": 0.78,
+      "learning_rate": 9.22e-05,
+      "loss": 4.3217,
       "step": 464
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.22e-05,
-      "loss": 4.4454,
+      "loss": 4.3614,
       "step": 465
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 9.240000000000001e-05,
-      "loss": 4.2398,
+      "loss": 4.2587,
       "step": 466
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 9.260000000000001e-05,
-      "loss": 4.392,
+      "loss": 4.2522,
       "step": 467
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 9.28e-05,
-      "loss": 4.3057,
+      "loss": 4.2726,
       "step": 468
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.79,
       "learning_rate": 9.300000000000001e-05,
-      "loss": 4.1792,
+      "loss": 4.2295,
       "step": 469
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.320000000000002e-05,
-      "loss": 4.4736,
+      "loss": 4.194,
       "step": 470
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.340000000000001e-05,
-      "loss": 4.2611,
+      "loss": 4.1531,
       "step": 471
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 9.360000000000001e-05,
-      "loss": 4.4917,
+      "loss": 4.2138,
       "step": 472
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.38e-05,
-      "loss": 4.0621,
+      "loss": 4.2742,
       "step": 473
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.4e-05,
-      "loss": 4.3977,
+      "loss": 4.1743,
       "step": 474
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.42e-05,
-      "loss": 4.547,
+      "loss": 4.1392,
       "step": 475
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.44e-05,
-      "loss": 4.4353,
+      "loss": 4.3719,
       "step": 476
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.46e-05,
-      "loss": 4.0947,
+      "loss": 4.2289,
       "step": 477
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 9.48e-05,
-      "loss": 4.3555,
+      "loss": 4.2521,
       "step": 478
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 9.5e-05,
-      "loss": 4.3024,
+      "loss": 4.2699,
       "step": 479
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 9.52e-05,
-      "loss": 4.2973,
+      "loss": 4.0737,
       "step": 480
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 9.54e-05,
-      "loss": 4.1742,
+      "loss": 4.488,
       "step": 481
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.56e-05,
-      "loss": 4.2801,
+      "loss": 4.3975,
       "step": 482
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.58e-05,
-      "loss": 4.6342,
+      "loss": 4.4541,
       "step": 483
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.81,
       "learning_rate": 9.6e-05,
-      "loss": 4.3777,
+      "loss": 4.392,
       "step": 484
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.620000000000001e-05,
-      "loss": 4.2769,
+      "loss": 4.258,
       "step": 485
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.64e-05,
-      "loss": 4.3859,
+      "loss": 4.2439,
       "step": 486
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.66e-05,
-      "loss": 4.1238,
+      "loss": 4.3562,
       "step": 487
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.680000000000001e-05,
-      "loss": 4.2757,
+      "loss": 4.3051,
       "step": 488
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.7e-05,
-      "loss": 4.4325,
+      "loss": 4.3883,
       "step": 489
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 9.72e-05,
-      "loss": 4.2617,
+      "loss": 4.1472,
       "step": 490
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 9.74e-05,
-      "loss": 4.496,
+      "loss": 4.4604,
       "step": 491
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 9.76e-05,
-      "loss": 4.5499,
+      "loss": 4.351,
       "step": 492
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 9.78e-05,
-      "loss": 4.7684,
+      "loss": 4.0272,
       "step": 493
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 9.8e-05,
-      "loss": 4.4239,
+      "loss": 4.2968,
       "step": 494
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 9.82e-05,
-      "loss": 5.2452,
+      "loss": 4.5583,
       "step": 495
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 9.84e-05,
-      "loss": 4.7005,
+      "loss": 4.5452,
       "step": 496
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.86e-05,
-      "loss": 4.4708,
+      "loss": 4.3781,
       "step": 497
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.88e-05,
-      "loss": 4.785,
+      "loss": 4.3979,
       "step": 498
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.900000000000001e-05,
-      "loss": 4.4005,
+      "loss": 4.6125,
       "step": 499
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.92e-05,
-      "loss": 4.7833,
+      "loss": 4.3519,
       "step": 500
     },
     {
-      "epoch": 0.42,
-      "eval_loss": 4.556314468383789,
-      "eval_runtime": 794.1323,
-      "eval_samples_per_second": 3.327,
-      "eval_steps_per_second": 0.278,
-      "eval_wer": 1.8866521221737407,
+      "epoch": 0.84,
+      "eval_loss": 4.385700225830078,
+      "eval_runtime": 755.5833,
+      "eval_samples_per_second": 3.497,
+      "eval_steps_per_second": 0.292,
+      "eval_wer": 1.9208647362157873,
       "step": 500
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.94e-05,
-      "loss": 4.3792,
+      "loss": 4.331,
       "step": 501
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 9.960000000000001e-05,
-      "loss": 4.5542,
+      "loss": 4.1535,
       "step": 502
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 9.98e-05,
-      "loss": 4.5263,
+      "loss": 4.2661,
       "step": 503
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.85,
       "learning_rate": 0.0001,
-      "loss": 4.3444,
+      "loss": 4.3242,
       "step": 504
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 9.985486211901307e-05,
-      "loss": 4.4147,
+      "epoch": 0.85,
+      "learning_rate": 9.893617021276596e-05,
+      "loss": 4.3194,
       "step": 505
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.970972423802612e-05,
-      "loss": 4.7056,
+      "epoch": 0.85,
+      "learning_rate": 9.787234042553192e-05,
+      "loss": 4.4679,
       "step": 506
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.956458635703919e-05,
-      "loss": 4.7912,
+      "epoch": 0.85,
+      "learning_rate": 9.680851063829788e-05,
+      "loss": 4.2307,
       "step": 507
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.941944847605225e-05,
-      "loss": 4.4993,
+      "epoch": 0.85,
+      "learning_rate": 9.574468085106384e-05,
+      "loss": 4.2865,
       "step": 508
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.927431059506532e-05,
-      "loss": 4.2958,
+      "epoch": 0.86,
+      "learning_rate": 9.468085106382978e-05,
+      "loss": 4.0963,
       "step": 509
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.912917271407838e-05,
-      "loss": 4.7891,
+      "epoch": 0.86,
+      "learning_rate": 9.361702127659576e-05,
+      "loss": 4.2769,
       "step": 510
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.898403483309145e-05,
-      "loss": 4.6263,
+      "epoch": 0.86,
+      "learning_rate": 9.25531914893617e-05,
+      "loss": 4.3236,
       "step": 511
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.883889695210451e-05,
-      "loss": 4.2116,
+      "epoch": 0.86,
+      "learning_rate": 9.148936170212766e-05,
+      "loss": 4.1472,
       "step": 512
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.869375907111757e-05,
-      "loss": 4.6641,
+      "epoch": 0.86,
+      "learning_rate": 9.042553191489363e-05,
+      "loss": 4.1778,
       "step": 513
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.854862119013063e-05,
-      "loss": 4.1309,
+      "epoch": 0.86,
+      "learning_rate": 8.936170212765958e-05,
+      "loss": 4.2354,
       "step": 514
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.84034833091437e-05,
-      "loss": 4.1667,
+      "epoch": 0.87,
+      "learning_rate": 8.829787234042553e-05,
+      "loss": 4.253,
       "step": 515
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.825834542815675e-05,
-      "loss": 4.4538,
+      "epoch": 0.87,
+      "learning_rate": 8.723404255319149e-05,
+      "loss": 4.2079,
       "step": 516
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.811320754716981e-05,
-      "loss": 4.6166,
+      "epoch": 0.87,
+      "learning_rate": 8.617021276595745e-05,
+      "loss": 4.3176,
       "step": 517
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.796806966618288e-05,
-      "loss": 4.1742,
+      "epoch": 0.87,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 4.3198,
       "step": 518
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.782293178519594e-05,
-      "loss": 4.805,
+      "epoch": 0.87,
+      "learning_rate": 8.404255319148937e-05,
+      "loss": 4.121,
       "step": 519
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.767779390420901e-05,
-      "loss": 4.318,
+      "epoch": 0.87,
+      "learning_rate": 8.297872340425533e-05,
+      "loss": 4.1787,
       "step": 520
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.753265602322207e-05,
-      "loss": 4.3766,
+      "epoch": 0.88,
+      "learning_rate": 8.191489361702128e-05,
+      "loss": 4.265,
       "step": 521
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.738751814223513e-05,
-      "loss": 4.32,
+      "epoch": 0.88,
+      "learning_rate": 8.085106382978723e-05,
+      "loss": 4.4652,
       "step": 522
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.724238026124819e-05,
-      "loss": 4.8161,
+      "epoch": 0.88,
+      "learning_rate": 7.978723404255319e-05,
+      "loss": 4.1396,
       "step": 523
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.709724238026126e-05,
-      "loss": 4.3812,
+      "epoch": 0.88,
+      "learning_rate": 7.872340425531916e-05,
+      "loss": 3.9964,
       "step": 524
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.695210449927431e-05,
-      "loss": 4.323,
+      "epoch": 0.88,
+      "learning_rate": 7.76595744680851e-05,
+      "loss": 4.2279,
       "step": 525
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.680696661828737e-05,
-      "loss": 4.0547,
+      "epoch": 0.88,
+      "learning_rate": 7.659574468085106e-05,
+      "loss": 4.1907,
       "step": 526
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.666182873730044e-05,
+      "epoch": 0.89,
+      "learning_rate": 7.553191489361703e-05,
       "loss": 4.3998,
       "step": 527
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.65166908563135e-05,
-      "loss": 4.5361,
+      "epoch": 0.89,
+      "learning_rate": 7.446808510638298e-05,
+      "loss": 4.1426,
       "step": 528
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.637155297532656e-05,
-      "loss": 4.4911,
+      "epoch": 0.89,
+      "learning_rate": 7.340425531914894e-05,
+      "loss": 4.1125,
       "step": 529
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.622641509433963e-05,
-      "loss": 4.4619,
+      "epoch": 0.89,
+      "learning_rate": 7.23404255319149e-05,
+      "loss": 4.1092,
       "step": 530
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.60812772133527e-05,
-      "loss": 4.4594,
-      "step": 531
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.593613933236575e-05,
-      "loss": 4.7005,
-      "step": 532
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.579100145137882e-05,
-      "loss": 4.3492,
-      "step": 533
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.564586357039188e-05,
-      "loss": 4.571,
-      "step": 534
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.550072568940493e-05,
-      "loss": 4.2786,
-      "step": 535
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.5355587808418e-05,
-      "loss": 4.1727,
-      "step": 536
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.521044992743106e-05,
-      "loss": 4.6178,
-      "step": 537
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.506531204644412e-05,
-      "loss": 4.3364,
-      "step": 538
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.492017416545718e-05,
-      "loss": 4.4373,
-      "step": 539
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.477503628447025e-05,
-      "loss": 4.2645,
-      "step": 540
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.462989840348333e-05,
-      "loss": 4.5577,
-      "step": 541
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.448476052249638e-05,
-      "loss": 4.4293,
-      "step": 542
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.433962264150944e-05,
-      "loss": 4.2291,
-      "step": 543
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.419448476052251e-05,
-      "loss": 4.4036,
-      "step": 544
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.404934687953556e-05,
-      "loss": 4.7339,
-      "step": 545
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.390420899854863e-05,
-      "loss": 4.3083,
-      "step": 546
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.375907111756169e-05,
-      "loss": 4.9033,
-      "step": 547
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.361393323657474e-05,
-      "loss": 4.8686,
-      "step": 548
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.346879535558781e-05,
-      "loss": 4.7346,
-      "step": 549
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.332365747460087e-05,
-      "loss": 4.9567,
-      "step": 550
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.317851959361394e-05,
-      "loss": 4.4727,
-      "step": 551
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.3033381712627e-05,
-      "loss": 4.3568,
-      "step": 552
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.288824383164007e-05,
-      "loss": 4.43,
-      "step": 553
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.274310595065312e-05,
-      "loss": 4.4838,
-      "step": 554
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.259796806966619e-05,
-      "loss": 4.4556,
-      "step": 555
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.245283018867925e-05,
-      "loss": 4.3291,
-      "step": 556
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.230769230769232e-05,
-      "loss": 4.2059,
-      "step": 557
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.216255442670537e-05,
-      "loss": 4.2412,
-      "step": 558
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.201741654571843e-05,
-      "loss": 4.4715,
-      "step": 559
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.18722786647315e-05,
-      "loss": 4.3221,
-      "step": 560
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.172714078374456e-05,
-      "loss": 4.4779,
-      "step": 561
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.158200290275763e-05,
-      "loss": 4.1747,
-      "step": 562
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.14368650217707e-05,
-      "loss": 4.641,
-      "step": 563
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.129172714078375e-05,
-      "loss": 4.0936,
-      "step": 564
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.114658925979681e-05,
-      "loss": 4.2934,
-      "step": 565
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.100145137880988e-05,
-      "loss": 4.3757,
-      "step": 566
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.085631349782293e-05,
-      "loss": 4.3595,
-      "step": 567
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.0711175616836e-05,
-      "loss": 4.4591,
-      "step": 568
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.056603773584906e-05,
-      "loss": 4.1337,
-      "step": 569
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.042089985486212e-05,
-      "loss": 4.2946,
-      "step": 570
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.027576197387519e-05,
-      "loss": 4.2776,
-      "step": 571
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.013062409288826e-05,
-      "loss": 4.5828,
-      "step": 572
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.998548621190132e-05,
-      "loss": 4.3009,
-      "step": 573
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.984034833091437e-05,
-      "loss": 4.7324,
-      "step": 574
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.969521044992744e-05,
-      "loss": 4.4953,
-      "step": 575
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 8.95500725689405e-05,
-      "loss": 4.6216,
-      "step": 576
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.940493468795355e-05,
-      "loss": 4.4785,
-      "step": 577
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.925979680696662e-05,
-      "loss": 4.2908,
-      "step": 578
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.911465892597968e-05,
-      "loss": 4.3157,
-      "step": 579
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.896952104499274e-05,
-      "loss": 4.4017,
-      "step": 580
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.882438316400582e-05,
-      "loss": 4.5322,
-      "step": 581
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.867924528301888e-05,
-      "loss": 4.2026,
-      "step": 582
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.853410740203193e-05,
-      "loss": 4.2356,
-      "step": 583
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.8388969521045e-05,
-      "loss": 4.5635,
-      "step": 584
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.824383164005806e-05,
-      "loss": 4.5617,
-      "step": 585
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.809869375907113e-05,
-      "loss": 4.4101,
-      "step": 586
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.795355587808418e-05,
-      "loss": 4.6033,
-      "step": 587
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 8.780841799709725e-05,
-      "loss": 4.3074,
-      "step": 588
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.766328011611031e-05,
-      "loss": 4.5875,
-      "step": 589
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.751814223512336e-05,
-      "loss": 4.482,
-      "step": 590
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.737300435413643e-05,
-      "loss": 4.3303,
-      "step": 591
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.722786647314949e-05,
-      "loss": 4.1894,
-      "step": 592
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.708272859216256e-05,
-      "loss": 5.1543,
-      "step": 593
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.693759071117562e-05,
-      "loss": 4.3172,
-      "step": 594
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.679245283018869e-05,
-      "loss": 4.4337,
-      "step": 595
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.664731494920174e-05,
-      "loss": 4.077,
-      "step": 596
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.65021770682148e-05,
-      "loss": 4.3498,
-      "step": 597
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.635703918722787e-05,
-      "loss": 5.0575,
-      "step": 598
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.621190130624092e-05,
-      "loss": 4.9551,
-      "step": 599
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 8.606676342525399e-05,
-      "loss": 5.3085,
-      "step": 600
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.592162554426705e-05,
-      "loss": 4.3878,
-      "step": 601
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.577648766328012e-05,
-      "loss": 4.7158,
-      "step": 602
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.563134978229318e-05,
-      "loss": 4.7155,
-      "step": 603
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.548621190130625e-05,
-      "loss": 4.1384,
-      "step": 604
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.534107402031931e-05,
-      "loss": 4.4012,
-      "step": 605
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.519593613933237e-05,
-      "loss": 4.5157,
-      "step": 606
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.505079825834543e-05,
-      "loss": 4.3355,
-      "step": 607
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.49056603773585e-05,
-      "loss": 4.5176,
-      "step": 608
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.476052249637155e-05,
-      "loss": 4.3601,
-      "step": 609
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.461538461538461e-05,
-      "loss": 4.4276,
-      "step": 610
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.447024673439768e-05,
-      "loss": 4.4225,
-      "step": 611
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 8.432510885341074e-05,
-      "loss": 4.2086,
-      "step": 612
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.417997097242381e-05,
-      "loss": 4.7467,
-      "step": 613
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.403483309143688e-05,
-      "loss": 4.1817,
-      "step": 614
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.388969521044994e-05,
-      "loss": 4.3207,
-      "step": 615
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.374455732946299e-05,
-      "loss": 4.5621,
-      "step": 616
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.359941944847606e-05,
-      "loss": 4.4927,
-      "step": 617
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.345428156748912e-05,
-      "loss": 4.6222,
-      "step": 618
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.330914368650217e-05,
-      "loss": 4.373,
-      "step": 619
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.316400580551524e-05,
-      "loss": 4.3561,
-      "step": 620
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.30188679245283e-05,
-      "loss": 4.0413,
-      "step": 621
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.287373004354137e-05,
-      "loss": 4.2229,
-      "step": 622
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.272859216255444e-05,
-      "loss": 4.9314,
-      "step": 623
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 8.25834542815675e-05,
-      "loss": 4.468,
-      "step": 624
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.243831640058055e-05,
-      "loss": 4.602,
-      "step": 625
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.229317851959362e-05,
-      "loss": 4.7789,
-      "step": 626
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.214804063860668e-05,
-      "loss": 4.5341,
-      "step": 627
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.200290275761974e-05,
-      "loss": 4.1949,
-      "step": 628
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.18577648766328e-05,
-      "loss": 4.4245,
-      "step": 629
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.171262699564587e-05,
-      "loss": 4.3408,
-      "step": 630
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.156748911465893e-05,
-      "loss": 4.4314,
-      "step": 631
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.142235123367198e-05,
-      "loss": 4.2712,
-      "step": 632
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.127721335268506e-05,
-      "loss": 4.4992,
-      "step": 633
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.113207547169813e-05,
-      "loss": 4.3726,
-      "step": 634
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.098693759071118e-05,
-      "loss": 4.607,
-      "step": 635
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 8.084179970972424e-05,
-      "loss": 4.2791,
-      "step": 636
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.069666182873731e-05,
-      "loss": 3.9959,
-      "step": 637
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.055152394775036e-05,
-      "loss": 4.1824,
-      "step": 638
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.040638606676343e-05,
-      "loss": 4.4283,
-      "step": 639
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.026124818577649e-05,
-      "loss": 4.338,
-      "step": 640
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 8.011611030478954e-05,
-      "loss": 4.3196,
-      "step": 641
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.997097242380261e-05,
-      "loss": 4.5253,
-      "step": 642
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.982583454281567e-05,
-      "loss": 4.7036,
-      "step": 643
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.968069666182875e-05,
-      "loss": 4.2378,
-      "step": 644
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.95355587808418e-05,
-      "loss": 4.3364,
-      "step": 645
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.939042089985487e-05,
-      "loss": 4.2319,
-      "step": 646
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.924528301886794e-05,
-      "loss": 4.1085,
-      "step": 647
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.910014513788099e-05,
-      "loss": 4.8411,
-      "step": 648
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.895500725689405e-05,
-      "loss": 4.7293,
-      "step": 649
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.880986937590712e-05,
-      "loss": 4.3295,
-      "step": 650
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.866473149492017e-05,
-      "loss": 4.4942,
-      "step": 651
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.851959361393323e-05,
-      "loss": 4.2249,
-      "step": 652
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.83744557329463e-05,
-      "loss": 4.5424,
-      "step": 653
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.822931785195937e-05,
-      "loss": 4.4698,
-      "step": 654
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.808417997097243e-05,
-      "loss": 4.3183,
-      "step": 655
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.79390420899855e-05,
-      "loss": 4.4584,
-      "step": 656
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.779390420899855e-05,
-      "loss": 4.3474,
-      "step": 657
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.764876632801161e-05,
-      "loss": 4.3766,
-      "step": 658
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.750362844702468e-05,
-      "loss": 4.4664,
-      "step": 659
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.735849056603774e-05,
-      "loss": 4.244,
-      "step": 660
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.72133526850508e-05,
-      "loss": 4.3386,
-      "step": 661
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.706821480406386e-05,
-      "loss": 4.1731,
-      "step": 662
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.692307692307693e-05,
-      "loss": 4.4649,
-      "step": 663
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.677793904208999e-05,
-      "loss": 4.2032,
-      "step": 664
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.663280116110306e-05,
-      "loss": 4.7186,
-      "step": 665
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.648766328011612e-05,
-      "loss": 4.4221,
-      "step": 666
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.634252539912917e-05,
-      "loss": 4.6093,
-      "step": 667
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.619738751814224e-05,
-      "loss": 4.2385,
-      "step": 668
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.60522496371553e-05,
-      "loss": 4.4157,
-      "step": 669
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.590711175616836e-05,
-      "loss": 4.3038,
-      "step": 670
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.576197387518142e-05,
-      "loss": 4.3337,
-      "step": 671
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.561683599419449e-05,
-      "loss": 4.4547,
-      "step": 672
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.547169811320755e-05,
-      "loss": 4.5339,
-      "step": 673
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.532656023222062e-05,
-      "loss": 4.4666,
-      "step": 674
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.518142235123368e-05,
-      "loss": 4.2231,
-      "step": 675
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.503628447024675e-05,
-      "loss": 4.2687,
-      "step": 676
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.48911465892598e-05,
-      "loss": 4.4993,
-      "step": 677
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.474600870827286e-05,
-      "loss": 4.4445,
-      "step": 678
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.460087082728593e-05,
-      "loss": 4.4061,
-      "step": 679
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.445573294629898e-05,
-      "loss": 4.4366,
-      "step": 680
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.431059506531205e-05,
-      "loss": 4.3157,
-      "step": 681
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.416545718432511e-05,
-      "loss": 4.2019,
-      "step": 682
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.402031930333816e-05,
-      "loss": 4.3612,
-      "step": 683
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.387518142235124e-05,
-      "loss": 4.1437,
-      "step": 684
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.373004354136431e-05,
-      "loss": 4.4437,
-      "step": 685
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.358490566037736e-05,
-      "loss": 4.4095,
-      "step": 686
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.343976777939043e-05,
-      "loss": 4.2448,
-      "step": 687
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.329462989840349e-05,
-      "loss": 4.2367,
-      "step": 688
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.314949201741656e-05,
-      "loss": 4.4103,
-      "step": 689
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.300435413642961e-05,
-      "loss": 4.6344,
-      "step": 690
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.285921625544267e-05,
-      "loss": 4.5954,
-      "step": 691
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.271407837445574e-05,
-      "loss": 4.4383,
-      "step": 692
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.256894049346879e-05,
-      "loss": 4.5851,
-      "step": 693
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.242380261248185e-05,
-      "loss": 4.8426,
-      "step": 694
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.227866473149493e-05,
-      "loss": 4.2986,
-      "step": 695
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.213352685050799e-05,
-      "loss": 4.4581,
-      "step": 696
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.198838896952105e-05,
-      "loss": 4.4466,
-      "step": 697
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.184325108853412e-05,
-      "loss": 4.9081,
-      "step": 698
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.169811320754717e-05,
-      "loss": 4.1446,
-      "step": 699
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.155297532656023e-05,
-      "loss": 5.1299,
-      "step": 700
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.14078374455733e-05,
-      "loss": 4.5274,
-      "step": 701
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.126269956458636e-05,
-      "loss": 4.1469,
-      "step": 702
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.111756168359942e-05,
-      "loss": 4.3366,
-      "step": 703
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.097242380261248e-05,
-      "loss": 4.2239,
-      "step": 704
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.082728592162555e-05,
-      "loss": 4.1561,
-      "step": 705
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.068214804063861e-05,
-      "loss": 4.3857,
-      "step": 706
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.053701015965168e-05,
-      "loss": 4.2455,
-      "step": 707
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.039187227866474e-05,
-      "loss": 4.4733,
-      "step": 708
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.02467343976778e-05,
-      "loss": 4.2625,
-      "step": 709
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.010159651669086e-05,
-      "loss": 4.3452,
-      "step": 710
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.995645863570392e-05,
-      "loss": 4.0508,
-      "step": 711
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.981132075471698e-05,
-      "loss": 4.451,
-      "step": 712
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.966618287373004e-05,
-      "loss": 4.515,
-      "step": 713
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.95210449927431e-05,
-      "loss": 4.1929,
-      "step": 714
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.937590711175617e-05,
-      "loss": 4.5161,
-      "step": 715
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.923076923076924e-05,
-      "loss": 4.2423,
-      "step": 716
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.90856313497823e-05,
-      "loss": 4.139,
-      "step": 717
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.894049346879537e-05,
-      "loss": 4.4982,
-      "step": 718
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 6.879535558780842e-05,
-      "loss": 4.2417,
-      "step": 719
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.865021770682148e-05,
-      "loss": 4.1495,
-      "step": 720
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.850507982583455e-05,
-      "loss": 4.2026,
-      "step": 721
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.83599419448476e-05,
-      "loss": 4.0793,
-      "step": 722
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.821480406386067e-05,
-      "loss": 4.4962,
-      "step": 723
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.806966618287373e-05,
-      "loss": 4.2095,
-      "step": 724
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.79245283018868e-05,
-      "loss": 4.3029,
-      "step": 725
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.777939042089986e-05,
-      "loss": 4.5572,
-      "step": 726
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.763425253991293e-05,
-      "loss": 4.3373,
-      "step": 727
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.748911465892598e-05,
-      "loss": 4.6486,
-      "step": 728
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.734397677793905e-05,
-      "loss": 4.1032,
-      "step": 729
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.719883889695211e-05,
-      "loss": 4.3897,
-      "step": 730
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 6.705370101596516e-05,
-      "loss": 4.7747,
-      "step": 731
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.690856313497823e-05,
-      "loss": 4.5149,
-      "step": 732
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.676342525399129e-05,
-      "loss": 4.1189,
-      "step": 733
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.661828737300436e-05,
-      "loss": 4.3142,
-      "step": 734
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.647314949201742e-05,
-      "loss": 4.3112,
-      "step": 735
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.632801161103049e-05,
-      "loss": 4.2533,
-      "step": 736
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.618287373004355e-05,
-      "loss": 4.3274,
-      "step": 737
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.60377358490566e-05,
-      "loss": 4.6027,
-      "step": 738
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.589259796806967e-05,
-      "loss": 4.3918,
-      "step": 739
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.574746008708274e-05,
-      "loss": 4.4124,
-      "step": 740
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.560232220609579e-05,
-      "loss": 4.3783,
-      "step": 741
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.545718432510885e-05,
-      "loss": 4.4281,
-      "step": 742
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 6.531204644412192e-05,
-      "loss": 4.2891,
-      "step": 743
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.516690856313497e-05,
-      "loss": 4.3645,
-      "step": 744
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.502177068214804e-05,
-      "loss": 4.6455,
-      "step": 745
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.487663280116111e-05,
-      "loss": 4.5791,
-      "step": 746
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.473149492017418e-05,
-      "loss": 4.5406,
-      "step": 747
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.458635703918723e-05,
-      "loss": 5.0932,
-      "step": 748
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.44412191582003e-05,
-      "loss": 4.2563,
-      "step": 749
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.429608127721336e-05,
-      "loss": 4.501,
-      "step": 750
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.415094339622641e-05,
-      "loss": 4.3047,
-      "step": 751
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.400580551523948e-05,
-      "loss": 4.5601,
-      "step": 752
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.386066763425254e-05,
-      "loss": 4.3934,
-      "step": 753
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.37155297532656e-05,
-      "loss": 4.1254,
-      "step": 754
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 6.357039187227866e-05,
-      "loss": 4.4322,
-      "step": 755
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.342525399129173e-05,
-      "loss": 4.2348,
-      "step": 756
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.328011611030479e-05,
-      "loss": 4.1855,
-      "step": 757
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.313497822931786e-05,
-      "loss": 4.03,
-      "step": 758
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.298984034833092e-05,
-      "loss": 4.2221,
-      "step": 759
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.284470246734397e-05,
-      "loss": 4.5104,
-      "step": 760
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.269956458635704e-05,
-      "loss": 4.1147,
-      "step": 761
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.25544267053701e-05,
-      "loss": 4.3114,
-      "step": 762
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.240928882438317e-05,
-      "loss": 4.2124,
-      "step": 763
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.226415094339622e-05,
-      "loss": 4.2235,
-      "step": 764
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.211901306240929e-05,
-      "loss": 4.5241,
-      "step": 765
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.197387518142235e-05,
-      "loss": 4.2109,
-      "step": 766
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 6.182873730043542e-05,
-      "loss": 4.2899,
-      "step": 767
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.168359941944848e-05,
-      "loss": 4.0363,
-      "step": 768
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.153846153846155e-05,
-      "loss": 4.3609,
-      "step": 769
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.13933236574746e-05,
-      "loss": 4.5502,
-      "step": 770
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.124818577648767e-05,
-      "loss": 4.3071,
-      "step": 771
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.110304789550073e-05,
-      "loss": 4.3386,
-      "step": 772
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.095791001451378e-05,
-      "loss": 4.1573,
-      "step": 773
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.0812772133526855e-05,
-      "loss": 4.1109,
-      "step": 774
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.066763425253992e-05,
-      "loss": 4.2752,
-      "step": 775
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.0522496371552985e-05,
-      "loss": 4.1387,
-      "step": 776
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.037735849056604e-05,
-      "loss": 4.2608,
-      "step": 777
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.02322206095791e-05,
-      "loss": 4.5297,
-      "step": 778
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 6.008708272859217e-05,
-      "loss": 4.257,
-      "step": 779
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.9941944847605226e-05,
-      "loss": 4.4121,
-      "step": 780
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.979680696661829e-05,
-      "loss": 4.1318,
-      "step": 781
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.965166908563136e-05,
-      "loss": 4.2932,
-      "step": 782
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.950653120464441e-05,
-      "loss": 4.207,
-      "step": 783
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.9361393323657474e-05,
-      "loss": 4.4313,
-      "step": 784
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.9216255442670546e-05,
-      "loss": 4.434,
-      "step": 785
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.90711175616836e-05,
-      "loss": 4.243,
-      "step": 786
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.892597968069666e-05,
-      "loss": 4.6414,
-      "step": 787
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.878084179970973e-05,
-      "loss": 4.231,
-      "step": 788
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.863570391872279e-05,
-      "loss": 4.0864,
-      "step": 789
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.849056603773585e-05,
-      "loss": 4.1057,
-      "step": 790
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 5.834542815674892e-05,
-      "loss": 4.4971,
-      "step": 791
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.820029027576198e-05,
-      "loss": 4.5999,
-      "step": 792
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.8055152394775034e-05,
-      "loss": 4.6644,
-      "step": 793
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.79100145137881e-05,
-      "loss": 4.141,
-      "step": 794
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.7764876632801165e-05,
-      "loss": 4.4483,
-      "step": 795
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.7619738751814224e-05,
-      "loss": 3.9794,
-      "step": 796
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.747460087082729e-05,
-      "loss": 4.1051,
-      "step": 797
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.7329462989840354e-05,
-      "loss": 4.4866,
-      "step": 798
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.718432510885341e-05,
-      "loss": 4.8904,
-      "step": 799
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.703918722786648e-05,
-      "loss": 4.1762,
-      "step": 800
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.689404934687954e-05,
-      "loss": 4.149,
-      "step": 801
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 5.6748911465892595e-05,
-      "loss": 4.1887,
-      "step": 802
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.660377358490566e-05,
-      "loss": 4.1483,
-      "step": 803
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.6458635703918726e-05,
-      "loss": 4.2156,
-      "step": 804
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.631349782293179e-05,
-      "loss": 4.0261,
-      "step": 805
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.616835994194485e-05,
-      "loss": 4.2398,
-      "step": 806
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.6023222060957915e-05,
-      "loss": 4.3819,
-      "step": 807
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.587808417997098e-05,
-      "loss": 4.116,
-      "step": 808
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.573294629898403e-05,
-      "loss": 4.2239,
-      "step": 809
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.55878084179971e-05,
-      "loss": 4.1741,
-      "step": 810
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.544267053701017e-05,
-      "loss": 4.2842,
-      "step": 811
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.529753265602322e-05,
-      "loss": 4.4011,
-      "step": 812
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.5152394775036286e-05,
-      "loss": 4.1783,
-      "step": 813
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 5.500725689404935e-05,
-      "loss": 4.668,
-      "step": 814
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.486211901306241e-05,
-      "loss": 4.2243,
-      "step": 815
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.4716981132075475e-05,
-      "loss": 4.2685,
-      "step": 816
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.457184325108854e-05,
-      "loss": 4.2393,
-      "step": 817
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.442670537010159e-05,
-      "loss": 4.1064,
-      "step": 818
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.428156748911466e-05,
-      "loss": 4.1142,
-      "step": 819
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.413642960812772e-05,
-      "loss": 4.2538,
-      "step": 820
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.399129172714079e-05,
-      "loss": 4.3754,
-      "step": 821
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.384615384615385e-05,
-      "loss": 4.0257,
-      "step": 822
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.370101596516691e-05,
-      "loss": 4.4128,
-      "step": 823
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.355587808417998e-05,
-      "loss": 4.7139,
-      "step": 824
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.3410740203193036e-05,
-      "loss": 4.4549,
-      "step": 825
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 5.32656023222061e-05,
-      "loss": 4.6,
-      "step": 826
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.3120464441219166e-05,
-      "loss": 4.1839,
-      "step": 827
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.297532656023222e-05,
-      "loss": 4.2969,
-      "step": 828
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.283018867924528e-05,
-      "loss": 4.0749,
-      "step": 829
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.268505079825835e-05,
-      "loss": 4.1758,
-      "step": 830
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.253991291727141e-05,
-      "loss": 4.2415,
-      "step": 831
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.239477503628447e-05,
-      "loss": 4.2942,
-      "step": 832
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.224963715529754e-05,
-      "loss": 4.098,
-      "step": 833
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.21044992743106e-05,
-      "loss": 4.2875,
-      "step": 834
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.1959361393323655e-05,
-      "loss": 4.3308,
-      "step": 835
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.181422351233673e-05,
-      "loss": 4.4749,
-      "step": 836
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.166908563134979e-05,
-      "loss": 4.209,
-      "step": 837
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 5.1523947750362844e-05,
-      "loss": 3.9856,
-      "step": 838
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.137880986937591e-05,
-      "loss": 4.2604,
-      "step": 839
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.1233671988388975e-05,
-      "loss": 4.0022,
-      "step": 840
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.108853410740203e-05,
-      "loss": 4.3227,
-      "step": 841
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.09433962264151e-05,
-      "loss": 4.4233,
-      "step": 842
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.0798258345428164e-05,
-      "loss": 4.6508,
-      "step": 843
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.0653120464441215e-05,
-      "loss": 4.6199,
-      "step": 844
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.050798258345428e-05,
-      "loss": 4.1975,
-      "step": 845
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.0362844702467346e-05,
-      "loss": 4.0901,
-      "step": 846
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.0217706821480404e-05,
-      "loss": 4.3676,
-      "step": 847
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 5.007256894049347e-05,
-      "loss": 4.0069,
-      "step": 848
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.9927431059506535e-05,
-      "loss": 4.3912,
-      "step": 849
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 4.9782293178519594e-05,
-      "loss": 4.3724,
-      "step": 850
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.963715529753266e-05,
-      "loss": 4.5021,
-      "step": 851
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.9492017416545724e-05,
-      "loss": 4.3342,
-      "step": 852
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.934687953555878e-05,
-      "loss": 4.4982,
-      "step": 853
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.920174165457185e-05,
-      "loss": 4.2877,
-      "step": 854
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.9056603773584906e-05,
-      "loss": 4.1374,
-      "step": 855
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.891146589259797e-05,
-      "loss": 4.2488,
-      "step": 856
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.876632801161104e-05,
-      "loss": 4.2235,
-      "step": 857
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.8621190130624096e-05,
-      "loss": 4.0085,
-      "step": 858
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.8476052249637154e-05,
-      "loss": 4.1544,
-      "step": 859
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.833091436865022e-05,
-      "loss": 4.3112,
-      "step": 860
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.818577648766328e-05,
-      "loss": 3.9839,
-      "step": 861
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 4.804063860667635e-05,
-      "loss": 4.2978,
-      "step": 862
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.789550072568941e-05,
-      "loss": 4.258,
-      "step": 863
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.775036284470247e-05,
-      "loss": 4.1525,
-      "step": 864
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.760522496371553e-05,
-      "loss": 4.1409,
-      "step": 865
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.746008708272859e-05,
-      "loss": 4.1352,
-      "step": 866
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.731494920174166e-05,
-      "loss": 4.0375,
-      "step": 867
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.716981132075472e-05,
-      "loss": 4.1772,
-      "step": 868
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.702467343976778e-05,
-      "loss": 4.4641,
-      "step": 869
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.6879535558780845e-05,
-      "loss": 3.9396,
-      "step": 870
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.6734397677793904e-05,
-      "loss": 3.8723,
-      "step": 871
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.658925979680697e-05,
-      "loss": 4.0104,
-      "step": 872
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.6444121915820034e-05,
-      "loss": 4.2314,
-      "step": 873
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 4.629898403483309e-05,
-      "loss": 4.0193,
-      "step": 874
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.615384615384616e-05,
-      "loss": 4.3003,
-      "step": 875
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.600870827285922e-05,
-      "loss": 4.0089,
-      "step": 876
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.586357039187228e-05,
-      "loss": 4.0866,
-      "step": 877
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.571843251088535e-05,
-      "loss": 4.6334,
-      "step": 878
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.5573294629898406e-05,
-      "loss": 4.1313,
-      "step": 879
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.5428156748911464e-05,
-      "loss": 4.3888,
-      "step": 880
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.528301886792453e-05,
-      "loss": 3.9973,
-      "step": 881
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.5137880986937595e-05,
-      "loss": 4.3439,
-      "step": 882
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.499274310595066e-05,
-      "loss": 4.1157,
-      "step": 883
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.484760522496372e-05,
-      "loss": 4.1588,
-      "step": 884
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.470246734397678e-05,
-      "loss": 4.1945,
-      "step": 885
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 4.455732946298984e-05,
-      "loss": 4.2185,
-      "step": 886
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.441219158200291e-05,
-      "loss": 4.3485,
-      "step": 887
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.4267053701015966e-05,
-      "loss": 4.3356,
-      "step": 888
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.412191582002903e-05,
-      "loss": 4.415,
-      "step": 889
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.397677793904209e-05,
-      "loss": 4.3724,
-      "step": 890
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.3831640058055155e-05,
-      "loss": 4.4802,
-      "step": 891
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.3686502177068214e-05,
-      "loss": 4.1653,
-      "step": 892
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.354136429608128e-05,
-      "loss": 4.1748,
-      "step": 893
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.3396226415094345e-05,
-      "loss": 4.0551,
-      "step": 894
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.32510885341074e-05,
-      "loss": 4.2498,
-      "step": 895
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.310595065312046e-05,
-      "loss": 4.3364,
-      "step": 896
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.296081277213353e-05,
-      "loss": 4.2678,
-      "step": 897
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 4.281567489114659e-05,
-      "loss": 4.2637,
-      "step": 898
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.267053701015966e-05,
-      "loss": 4.4373,
-      "step": 899
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.2525399129172716e-05,
-      "loss": 4.3848,
-      "step": 900
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.2380261248185774e-05,
-      "loss": 4.1722,
-      "step": 901
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.223512336719884e-05,
-      "loss": 4.2707,
-      "step": 902
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.2089985486211905e-05,
-      "loss": 4.2313,
-      "step": 903
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.194484760522497e-05,
-      "loss": 4.4407,
-      "step": 904
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.179970972423803e-05,
-      "loss": 4.1798,
-      "step": 905
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.165457184325109e-05,
-      "loss": 4.2912,
-      "step": 906
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.150943396226415e-05,
-      "loss": 4.2129,
-      "step": 907
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.136429608127722e-05,
-      "loss": 4.6821,
-      "step": 908
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.1219158200290276e-05,
-      "loss": 4.2339,
-      "step": 909
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.107402031930334e-05,
-      "loss": 4.4074,
-      "step": 910
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.09288824383164e-05,
-      "loss": 4.0823,
-      "step": 911
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.0783744557329466e-05,
-      "loss": 4.0311,
-      "step": 912
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.063860667634253e-05,
-      "loss": 4.1563,
-      "step": 913
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.049346879535559e-05,
-      "loss": 4.1927,
-      "step": 914
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.0348330914368655e-05,
-      "loss": 4.4148,
-      "step": 915
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.020319303338171e-05,
-      "loss": 4.1228,
-      "step": 916
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 4.005805515239477e-05,
-      "loss": 4.103,
-      "step": 917
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.991291727140784e-05,
-      "loss": 4.1851,
-      "step": 918
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.97677793904209e-05,
-      "loss": 4.6208,
-      "step": 919
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.962264150943397e-05,
-      "loss": 4.2571,
-      "step": 920
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.9477503628447026e-05,
-      "loss": 4.0951,
-      "step": 921
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.9332365747460085e-05,
-      "loss": 4.2407,
-      "step": 922
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.918722786647315e-05,
-      "loss": 4.0125,
-      "step": 923
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.9042089985486215e-05,
-      "loss": 4.1566,
-      "step": 924
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.8896952104499274e-05,
-      "loss": 3.9781,
-      "step": 925
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.875181422351234e-05,
-      "loss": 4.0215,
-      "step": 926
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.86066763425254e-05,
-      "loss": 4.2516,
-      "step": 927
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.846153846153846e-05,
-      "loss": 4.1838,
-      "step": 928
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.831640058055153e-05,
-      "loss": 4.0272,
-      "step": 929
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.817126269956459e-05,
-      "loss": 4.0641,
-      "step": 930
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.802612481857765e-05,
-      "loss": 4.3323,
-      "step": 931
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.788098693759071e-05,
-      "loss": 4.3029,
-      "step": 932
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.7735849056603776e-05,
-      "loss": 4.1865,
-      "step": 933
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.759071117561684e-05,
-      "loss": 4.1979,
-      "step": 934
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.74455732946299e-05,
-      "loss": 4.3242,
-      "step": 935
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.7300435413642965e-05,
-      "loss": 4.2908,
-      "step": 936
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.7155297532656023e-05,
-      "loss": 4.2043,
-      "step": 937
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.701015965166908e-05,
-      "loss": 4.3978,
-      "step": 938
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6865021770682154e-05,
-      "loss": 4.1612,
-      "step": 939
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.671988388969521e-05,
-      "loss": 4.3126,
-      "step": 940
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.657474600870828e-05,
-      "loss": 4.0399,
-      "step": 941
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6429608127721336e-05,
-      "loss": 4.249,
-      "step": 942
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.6284470246734395e-05,
-      "loss": 3.9184,
-      "step": 943
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.613933236574747e-05,
-      "loss": 4.1561,
-      "step": 944
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.5994194484760525e-05,
-      "loss": 4.5341,
-      "step": 945
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.5849056603773584e-05,
-      "loss": 3.9604,
-      "step": 946
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.570391872278665e-05,
-      "loss": 4.2536,
-      "step": 947
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.555878084179971e-05,
-      "loss": 4.3736,
-      "step": 948
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.541364296081277e-05,
-      "loss": 4.1409,
-      "step": 949
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.526850507982584e-05,
-      "loss": 4.0764,
-      "step": 950
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.51233671988389e-05,
-      "loss": 4.0417,
-      "step": 951
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.497822931785196e-05,
-      "loss": 4.3548,
-      "step": 952
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.483309143686502e-05,
-      "loss": 3.993,
-      "step": 953
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4687953555878086e-05,
-      "loss": 4.0538,
-      "step": 954
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.454281567489115e-05,
-      "loss": 4.1509,
-      "step": 955
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.439767779390421e-05,
-      "loss": 4.302,
-      "step": 956
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 3.4252539912917275e-05,
-      "loss": 4.3541,
-      "step": 957
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.4107402031930334e-05,
-      "loss": 4.1235,
-      "step": 958
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.39622641509434e-05,
-      "loss": 4.2222,
-      "step": 959
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3817126269956464e-05,
-      "loss": 4.1887,
-      "step": 960
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.367198838896952e-05,
-      "loss": 4.09,
-      "step": 961
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.352685050798258e-05,
-      "loss": 4.2085,
-      "step": 962
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.3381712626995646e-05,
-      "loss": 3.9025,
-      "step": 963
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.323657474600871e-05,
-      "loss": 4.5243,
-      "step": 964
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.309143686502178e-05,
-      "loss": 4.2681,
-      "step": 965
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.2946298984034836e-05,
-      "loss": 4.0479,
-      "step": 966
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.2801161103047894e-05,
-      "loss": 4.2015,
-      "step": 967
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.265602322206096e-05,
-      "loss": 4.2428,
-      "step": 968
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.251088534107402e-05,
-      "loss": 3.7816,
-      "step": 969
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.236574746008709e-05,
-      "loss": 4.0991,
-      "step": 970
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.222060957910015e-05,
-      "loss": 3.987,
-      "step": 971
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.207547169811321e-05,
-      "loss": 4.0026,
-      "step": 972
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.193033381712627e-05,
-      "loss": 4.281,
-      "step": 973
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.178519593613933e-05,
-      "loss": 4.0748,
-      "step": 974
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1640058055152396e-05,
-      "loss": 4.1662,
-      "step": 975
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.149492017416546e-05,
-      "loss": 4.2899,
-      "step": 976
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.134978229317852e-05,
-      "loss": 4.1242,
-      "step": 977
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1204644412191585e-05,
-      "loss": 4.1137,
-      "step": 978
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.1059506531204644e-05,
-      "loss": 4.0633,
-      "step": 979
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.091436865021771e-05,
-      "loss": 4.1539,
-      "step": 980
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 3.0769230769230774e-05,
-      "loss": 4.3347,
-      "step": 981
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.062409288824383e-05,
-      "loss": 4.3015,
-      "step": 982
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.047895500725689e-05,
-      "loss": 3.9847,
-      "step": 983
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.033381712626996e-05,
-      "loss": 4.2116,
-      "step": 984
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.018867924528302e-05,
-      "loss": 3.9691,
-      "step": 985
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 3.0043541364296084e-05,
-      "loss": 4.0735,
-      "step": 986
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9898403483309146e-05,
-      "loss": 4.2754,
-      "step": 987
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9753265602322204e-05,
-      "loss": 4.1031,
-      "step": 988
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9608127721335273e-05,
-      "loss": 4.0847,
-      "step": 989
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.946298984034833e-05,
-      "loss": 4.1152,
-      "step": 990
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9317851959361393e-05,
-      "loss": 4.3425,
-      "step": 991
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.917271407837446e-05,
-      "loss": 4.199,
-      "step": 992
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 2.9027576197387517e-05,
-      "loss": 3.9366,
-      "step": 993
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8882438316400583e-05,
-      "loss": 4.143,
-      "step": 994
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8737300435413644e-05,
-      "loss": 4.2442,
-      "step": 995
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8592162554426706e-05,
-      "loss": 4.8587,
-      "step": 996
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.844702467343977e-05,
-      "loss": 4.2903,
-      "step": 997
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.830188679245283e-05,
-      "loss": 4.242,
-      "step": 998
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8156748911465895e-05,
-      "loss": 4.3118,
-      "step": 999
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.8011611030478957e-05,
-      "loss": 4.311,
-      "step": 1000
-    },
-    {
-      "epoch": 0.84,
-      "eval_loss": 4.258539199829102,
-      "eval_runtime": 770.0216,
-      "eval_samples_per_second": 3.431,
-      "eval_steps_per_second": 0.287,
-      "eval_wer": 1.9291550971836573,
-      "step": 1000
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.7866473149492016e-05,
-      "loss": 4.1991,
-      "step": 1001
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.7721335268505085e-05,
-      "loss": 4.1808,
-      "step": 1002
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.7576197387518143e-05,
-      "loss": 4.0856,
-      "step": 1003
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.7431059506531205e-05,
-      "loss": 4.2277,
-      "step": 1004
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 2.728592162554427e-05,
-      "loss": 4.4143,
-      "step": 1005
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.714078374455733e-05,
-      "loss": 4.075,
-      "step": 1006
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6995645863570394e-05,
-      "loss": 4.1104,
-      "step": 1007
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6850507982583456e-05,
-      "loss": 4.075,
-      "step": 1008
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6705370101596518e-05,
-      "loss": 4.118,
-      "step": 1009
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6560232220609583e-05,
-      "loss": 4.0828,
-      "step": 1010
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.641509433962264e-05,
-      "loss": 4.0465,
-      "step": 1011
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.6269956458635704e-05,
-      "loss": 4.199,
-      "step": 1012
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.612481857764877e-05,
-      "loss": 4.0022,
-      "step": 1013
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.5979680696661827e-05,
-      "loss": 4.0851,
-      "step": 1014
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.5834542815674896e-05,
-      "loss": 4.1083,
-      "step": 1015
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.5689404934687955e-05,
-      "loss": 4.2556,
-      "step": 1016
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 2.5544267053701017e-05,
-      "loss": 4.0008,
-      "step": 1017
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.5399129172714082e-05,
-      "loss": 4.1897,
-      "step": 1018
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.525399129172714e-05,
-      "loss": 4.2987,
-      "step": 1019
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.5108853410740202e-05,
-      "loss": 4.0189,
-      "step": 1020
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.4963715529753268e-05,
-      "loss": 4.0965,
-      "step": 1021
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.481857764876633e-05,
-      "loss": 4.3147,
-      "step": 1022
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.467343976777939e-05,
-      "loss": 4.4302,
-      "step": 1023
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.4528301886792453e-05,
-      "loss": 3.9109,
-      "step": 1024
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.438316400580552e-05,
-      "loss": 4.1563,
-      "step": 1025
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.4238026124818577e-05,
-      "loss": 3.9886,
-      "step": 1026
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.409288824383164e-05,
-      "loss": 4.1377,
-      "step": 1027
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.3947750362844704e-05,
-      "loss": 4.027,
-      "step": 1028
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.3802612481857766e-05,
-      "loss": 4.034,
-      "step": 1029
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.365747460087083e-05,
-      "loss": 3.87,
-      "step": 1030
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.351233671988389e-05,
-      "loss": 4.0746,
-      "step": 1031
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.3367198838896952e-05,
-      "loss": 4.1965,
-      "step": 1032
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.3222060957910017e-05,
-      "loss": 4.3258,
-      "step": 1033
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.307692307692308e-05,
-      "loss": 3.9464,
-      "step": 1034
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.293178519593614e-05,
-      "loss": 4.0203,
-      "step": 1035
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.2786647314949203e-05,
-      "loss": 3.9579,
-      "step": 1036
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.2641509433962265e-05,
-      "loss": 4.1346,
-      "step": 1037
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.249637155297533e-05,
-      "loss": 4.3289,
-      "step": 1038
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.235123367198839e-05,
-      "loss": 4.115,
-      "step": 1039
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.2206095791001454e-05,
-      "loss": 4.042,
-      "step": 1040
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.2060957910014516e-05,
-      "loss": 4.1967,
-      "step": 1041
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1915820029027578e-05,
-      "loss": 4.1753,
-      "step": 1042
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.177068214804064e-05,
-      "loss": 4.07,
-      "step": 1043
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.16255442670537e-05,
-      "loss": 4.1848,
-      "step": 1044
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1480406386066763e-05,
-      "loss": 4.3217,
-      "step": 1045
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.133526850507983e-05,
-      "loss": 3.9443,
-      "step": 1046
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1190130624092887e-05,
-      "loss": 4.0229,
-      "step": 1047
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.1044992743105953e-05,
-      "loss": 4.2855,
-      "step": 1048
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.0899854862119014e-05,
-      "loss": 4.5764,
-      "step": 1049
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.0754716981132076e-05,
-      "loss": 4.2982,
-      "step": 1050
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.0609579100145138e-05,
-      "loss": 4.2772,
-      "step": 1051
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.04644412191582e-05,
-      "loss": 4.0519,
-      "step": 1052
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.0319303338171265e-05,
-      "loss": 3.9874,
-      "step": 1053
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.0174165457184327e-05,
-      "loss": 4.2141,
-      "step": 1054
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 2.0029027576197386e-05,
-      "loss": 4.2264,
-      "step": 1055
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.988388969521045e-05,
-      "loss": 4.2353,
-      "step": 1056
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9738751814223513e-05,
-      "loss": 4.2388,
-      "step": 1057
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9593613933236575e-05,
-      "loss": 3.9108,
-      "step": 1058
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9448476052249637e-05,
-      "loss": 4.2573,
-      "step": 1059
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.93033381712627e-05,
-      "loss": 4.196,
-      "step": 1060
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9158200290275764e-05,
-      "loss": 4.2198,
-      "step": 1061
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.9013062409288826e-05,
-      "loss": 4.0222,
-      "step": 1062
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.8867924528301888e-05,
-      "loss": 4.2861,
-      "step": 1063
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.872278664731495e-05,
-      "loss": 4.0791,
-      "step": 1064
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8577648766328012e-05,
-      "loss": 3.9227,
-      "step": 1065
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8432510885341077e-05,
-      "loss": 4.0212,
-      "step": 1066
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.828737300435414e-05,
-      "loss": 4.1144,
-      "step": 1067
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.8142235123367197e-05,
-      "loss": 4.0819,
-      "step": 1068
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.7997097242380263e-05,
-      "loss": 4.0857,
-      "step": 1069
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.7851959361393325e-05,
-      "loss": 4.2806,
-      "step": 1070
+      "epoch": 0.89,
+      "learning_rate": 7.127659574468085e-05,
+      "loss": 4.1359,
+      "step": 531
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.7706821480406387e-05,
-      "loss": 4.0845,
-      "step": 1071
+      "epoch": 0.89,
+      "learning_rate": 7.021276595744681e-05,
+      "loss": 4.4409,
+      "step": 532
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.756168359941945e-05,
-      "loss": 4.1884,
-      "step": 1072
+      "learning_rate": 6.914893617021277e-05,
+      "loss": 4.2359,
+      "step": 533
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.741654571843251e-05,
-      "loss": 3.7855,
-      "step": 1073
+      "learning_rate": 6.808510638297873e-05,
+      "loss": 4.2104,
+      "step": 534
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.7271407837445576e-05,
-      "loss": 4.0682,
-      "step": 1074
+      "learning_rate": 6.702127659574469e-05,
+      "loss": 4.1657,
+      "step": 535
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.7126269956458638e-05,
-      "loss": 4.2037,
-      "step": 1075
+      "learning_rate": 6.595744680851063e-05,
+      "loss": 4.1756,
+      "step": 536
     },
     {
       "epoch": 0.9,
-      "learning_rate": 1.69811320754717e-05,
-      "loss": 4.0765,
-      "step": 1076
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.683599419448476e-05,
-      "loss": 4.4026,
-      "step": 1077
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6690856313497823e-05,
-      "loss": 4.1575,
-      "step": 1078
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.654571843251089e-05,
-      "loss": 4.1035,
-      "step": 1079
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.6400580551523947e-05,
-      "loss": 3.9163,
-      "step": 1080
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.625544267053701e-05,
-      "loss": 4.2456,
-      "step": 1081
+      "learning_rate": 6.489361702127659e-05,
+      "loss": 4.237,
+      "step": 537
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.6110304789550074e-05,
-      "loss": 4.1341,
-      "step": 1082
+      "epoch": 0.9,
+      "learning_rate": 6.382978723404256e-05,
+      "loss": 4.2265,
+      "step": 538
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5965166908563136e-05,
-      "loss": 4.0596,
-      "step": 1083
+      "learning_rate": 6.276595744680851e-05,
+      "loss": 4.1348,
+      "step": 539
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5820029027576198e-05,
-      "loss": 4.0834,
-      "step": 1084
+      "learning_rate": 6.170212765957447e-05,
+      "loss": 4.1533,
+      "step": 540
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.567489114658926e-05,
-      "loss": 4.0972,
-      "step": 1085
+      "learning_rate": 6.063829787234043e-05,
+      "loss": 4.1785,
+      "step": 541
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5529753265602322e-05,
-      "loss": 3.9966,
-      "step": 1086
+      "learning_rate": 5.9574468085106384e-05,
+      "loss": 4.2902,
+      "step": 542
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5384615384615387e-05,
-      "loss": 4.0749,
-      "step": 1087
+      "learning_rate": 5.851063829787234e-05,
+      "loss": 4.2662,
+      "step": 543
     },
     {
       "epoch": 0.91,
-      "learning_rate": 1.5239477503628446e-05,
-      "loss": 4.2641,
-      "step": 1088
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.509433962264151e-05,
-      "loss": 4.0122,
-      "step": 1089
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4949201741654573e-05,
-      "loss": 4.0295,
-      "step": 1090
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4804063860667636e-05,
-      "loss": 4.1092,
-      "step": 1091
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4658925979680697e-05,
-      "loss": 3.9983,
-      "step": 1092
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4513788098693759e-05,
-      "loss": 4.0365,
-      "step": 1093
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 1.4368650217706822e-05,
-      "loss": 4.182,
-      "step": 1094
+      "learning_rate": 5.744680851063831e-05,
+      "loss": 4.2221,
+      "step": 544
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.4223512336719886e-05,
-      "loss": 4.1117,
-      "step": 1095
+      "learning_rate": 5.638297872340426e-05,
+      "loss": 4.3467,
+      "step": 545
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.4078374455732948e-05,
-      "loss": 4.5541,
-      "step": 1096
+      "learning_rate": 5.531914893617022e-05,
+      "loss": 4.5082,
+      "step": 546
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3933236574746008e-05,
-      "loss": 4.1907,
-      "step": 1097
+      "learning_rate": 5.425531914893617e-05,
+      "loss": 4.2303,
+      "step": 547
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3788098693759072e-05,
-      "loss": 4.597,
-      "step": 1098
+      "learning_rate": 5.319148936170213e-05,
+      "loss": 4.4299,
+      "step": 548
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3642960812772135e-05,
-      "loss": 4.4064,
-      "step": 1099
+      "learning_rate": 5.212765957446809e-05,
+      "loss": 4.9461,
+      "step": 549
     },
     {
       "epoch": 0.92,
-      "learning_rate": 1.3497822931785197e-05,
-      "loss": 4.0361,
-      "step": 1100
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.3352685050798259e-05,
-      "loss": 4.5052,
-      "step": 1101
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.320754716981132e-05,
-      "loss": 4.099,
-      "step": 1102
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.3062409288824384e-05,
-      "loss": 4.0892,
-      "step": 1103
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.2917271407837448e-05,
-      "loss": 3.9891,
-      "step": 1104
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.2772133526850508e-05,
-      "loss": 4.0789,
-      "step": 1105
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 1.262699564586357e-05,
-      "loss": 3.9073,
-      "step": 1106
+      "learning_rate": 5.1063829787234044e-05,
+      "loss": 4.3165,
+      "step": 550
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.2481857764876634e-05,
-      "loss": 4.2941,
-      "step": 1107
+      "learning_rate": 5e-05,
+      "loss": 4.2883,
+      "step": 551
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.2336719883889696e-05,
-      "loss": 4.2011,
-      "step": 1108
+      "learning_rate": 4.893617021276596e-05,
+      "loss": 4.2367,
+      "step": 552
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.219158200290276e-05,
-      "loss": 4.1511,
-      "step": 1109
+      "learning_rate": 4.787234042553192e-05,
+      "loss": 4.1961,
+      "step": 553
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.204644412191582e-05,
-      "loss": 3.9941,
-      "step": 1110
+      "learning_rate": 4.680851063829788e-05,
+      "loss": 4.1235,
+      "step": 554
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.1901306240928883e-05,
-      "loss": 4.0902,
-      "step": 1111
+      "learning_rate": 4.574468085106383e-05,
+      "loss": 4.1092,
+      "step": 555
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.1756168359941945e-05,
-      "loss": 4.4598,
-      "step": 1112
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1611030478955009e-05,
-      "loss": 4.1022,
-      "step": 1113
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.146589259796807e-05,
-      "loss": 3.8271,
-      "step": 1114
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1320754716981132e-05,
-      "loss": 4.2166,
-      "step": 1115
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1175616835994194e-05,
-      "loss": 4.2164,
-      "step": 1116
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.1030478955007258e-05,
-      "loss": 4.2495,
-      "step": 1117
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 1.088534107402032e-05,
-      "loss": 3.9463,
-      "step": 1118
+      "learning_rate": 4.468085106382979e-05,
+      "loss": 4.216,
+      "step": 556
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0740203193033382e-05,
-      "loss": 3.7749,
-      "step": 1119
+      "learning_rate": 4.3617021276595746e-05,
+      "loss": 4.3598,
+      "step": 557
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0595065312046444e-05,
-      "loss": 4.1315,
-      "step": 1120
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 4.2637,
+      "step": 558
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0449927431059507e-05,
-      "loss": 4.1492,
-      "step": 1121
+      "learning_rate": 4.148936170212766e-05,
+      "loss": 4.1133,
+      "step": 559
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0304789550072569e-05,
-      "loss": 4.2606,
-      "step": 1122
+      "learning_rate": 4.0425531914893614e-05,
+      "loss": 4.1171,
+      "step": 560
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0159651669085633e-05,
-      "loss": 3.9415,
-      "step": 1123
+      "learning_rate": 3.936170212765958e-05,
+      "loss": 4.4514,
+      "step": 561
     },
     {
       "epoch": 0.94,
-      "learning_rate": 1.0014513788098693e-05,
-      "loss": 4.1395,
-      "step": 1124
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.869375907111757e-06,
-      "loss": 4.3784,
-      "step": 1125
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.724238026124818e-06,
-      "loss": 4.1208,
-      "step": 1126
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.579100145137882e-06,
-      "loss": 4.2163,
-      "step": 1127
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.433962264150944e-06,
-      "loss": 4.0158,
-      "step": 1128
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.288824383164006e-06,
-      "loss": 4.0983,
-      "step": 1129
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.14368650217707e-06,
-      "loss": 4.2249,
-      "step": 1130
+      "learning_rate": 3.829787234042553e-05,
+      "loss": 4.0989,
+      "step": 562
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.998548621190131e-06,
-      "loss": 3.8736,
-      "step": 1131
+      "learning_rate": 3.723404255319149e-05,
+      "loss": 4.1127,
+      "step": 563
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.853410740203193e-06,
-      "loss": 3.8999,
-      "step": 1132
+      "learning_rate": 3.617021276595745e-05,
+      "loss": 4.402,
+      "step": 564
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.708272859216255e-06,
-      "loss": 4.1821,
-      "step": 1133
+      "learning_rate": 3.5106382978723407e-05,
+      "loss": 4.2184,
+      "step": 565
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.563134978229319e-06,
-      "loss": 4.0493,
-      "step": 1134
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 4.2015,
+      "step": 566
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.41799709724238e-06,
-      "loss": 4.2937,
-      "step": 1135
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.272859216255444e-06,
-      "loss": 3.9883,
-      "step": 1136
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 8.127721335268504e-06,
-      "loss": 4.1129,
-      "step": 1137
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.982583454281568e-06,
-      "loss": 4.0921,
-      "step": 1138
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.83744557329463e-06,
-      "loss": 4.0106,
-      "step": 1139
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.692307692307694e-06,
-      "loss": 3.9612,
-      "step": 1140
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 7.547169811320755e-06,
-      "loss": 4.3304,
-      "step": 1141
+      "learning_rate": 3.2978723404255317e-05,
+      "loss": 4.2155,
+      "step": 567
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.402031930333818e-06,
-      "loss": 4.0442,
-      "step": 1142
+      "learning_rate": 3.191489361702128e-05,
+      "loss": 4.0767,
+      "step": 568
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.256894049346879e-06,
-      "loss": 4.2674,
-      "step": 1143
+      "learning_rate": 3.085106382978723e-05,
+      "loss": 4.0548,
+      "step": 569
     },
     {
       "epoch": 0.96,
-      "learning_rate": 7.111756168359943e-06,
-      "loss": 4.3341,
-      "step": 1144
+      "learning_rate": 2.9787234042553192e-05,
+      "loss": 4.2154,
+      "step": 570
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.966618287373004e-06,
-      "loss": 4.0317,
-      "step": 1145
+      "learning_rate": 2.8723404255319154e-05,
+      "loss": 4.1975,
+      "step": 571
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.8214804063860676e-06,
-      "loss": 4.4574,
-      "step": 1146
+      "learning_rate": 2.765957446808511e-05,
+      "loss": 4.2138,
+      "step": 572
     },
     {
       "epoch": 0.96,
-      "learning_rate": 6.6763425253991295e-06,
-      "loss": 4.7992,
-      "step": 1147
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.531204644412192e-06,
-      "loss": 4.256,
-      "step": 1148
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.386066763425254e-06,
-      "loss": 4.4737,
-      "step": 1149
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.240928882438317e-06,
-      "loss": 4.0619,
-      "step": 1150
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 6.09579100145138e-06,
-      "loss": 4.012,
-      "step": 1151
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.9506531204644415e-06,
-      "loss": 3.8931,
-      "step": 1152
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.805515239477504e-06,
-      "loss": 4.0239,
-      "step": 1153
+      "learning_rate": 2.6595744680851064e-05,
+      "loss": 4.3058,
+      "step": 573
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.660377358490566e-06,
-      "loss": 3.9503,
-      "step": 1154
+      "learning_rate": 2.5531914893617022e-05,
+      "loss": 4.1456,
+      "step": 574
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.515239477503629e-06,
-      "loss": 4.2771,
-      "step": 1155
+      "learning_rate": 2.446808510638298e-05,
+      "loss": 4.2585,
+      "step": 575
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.370101596516691e-06,
-      "loss": 4.0878,
-      "step": 1156
+      "learning_rate": 2.340425531914894e-05,
+      "loss": 4.1975,
+      "step": 576
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.224963715529754e-06,
-      "loss": 4.1873,
-      "step": 1157
+      "learning_rate": 2.2340425531914894e-05,
+      "loss": 4.2476,
+      "step": 577
     },
     {
       "epoch": 0.97,
-      "learning_rate": 5.079825834542816e-06,
-      "loss": 4.057,
-      "step": 1158
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 4.0191,
+      "step": 578
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.934687953555878e-06,
-      "loss": 3.9503,
-      "step": 1159
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.789550072568941e-06,
-      "loss": 4.1809,
-      "step": 1160
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.644412191582003e-06,
-      "loss": 4.0986,
-      "step": 1161
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.499274310595066e-06,
-      "loss": 4.2512,
-      "step": 1162
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.354136429608128e-06,
-      "loss": 4.2879,
-      "step": 1163
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.20899854862119e-06,
-      "loss": 3.94,
-      "step": 1164
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.063860667634252e-06,
-      "loss": 4.1304,
-      "step": 1165
+      "learning_rate": 2.0212765957446807e-05,
+      "loss": 4.0424,
+      "step": 579
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.918722786647315e-06,
-      "loss": 4.252,
-      "step": 1166
+      "learning_rate": 1.9148936170212766e-05,
+      "loss": 4.2502,
+      "step": 580
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.7735849056603773e-06,
-      "loss": 3.8809,
-      "step": 1167
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 4.4058,
+      "step": 581
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.6284470246734397e-06,
-      "loss": 3.992,
-      "step": 1168
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 4.269,
+      "step": 582
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.483309143686502e-06,
-      "loss": 4.0648,
-      "step": 1169
+      "learning_rate": 1.595744680851064e-05,
+      "loss": 4.0582,
+      "step": 583
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.3381712626995647e-06,
-      "loss": 4.0951,
-      "step": 1170
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 4.1872,
+      "step": 584
     },
     {
       "epoch": 0.98,
-      "learning_rate": 3.193033381712627e-06,
-      "loss": 4.133,
-      "step": 1171
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.04789550072569e-06,
-      "loss": 3.9248,
-      "step": 1172
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.902757619738752e-06,
-      "loss": 4.1076,
-      "step": 1173
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.7576197387518145e-06,
-      "loss": 4.0541,
-      "step": 1174
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.612481857764877e-06,
-      "loss": 3.9535,
-      "step": 1175
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.467343976777939e-06,
-      "loss": 3.9946,
-      "step": 1176
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 2.3222060957910015e-06,
-      "loss": 4.1118,
-      "step": 1177
+      "learning_rate": 1.3829787234042554e-05,
+      "loss": 4.156,
+      "step": 585
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.177068214804064e-06,
-      "loss": 4.432,
-      "step": 1178
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 4.0976,
+      "step": 586
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.031930333817126e-06,
-      "loss": 3.9261,
-      "step": 1179
+      "learning_rate": 1.170212765957447e-05,
+      "loss": 4.2902,
+      "step": 587
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.8867924528301887e-06,
-      "loss": 4.1944,
-      "step": 1180
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 4.2762,
+      "step": 588
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.741654571843251e-06,
-      "loss": 4.074,
-      "step": 1181
+      "learning_rate": 9.574468085106383e-06,
+      "loss": 4.2949,
+      "step": 589
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.5965166908563135e-06,
-      "loss": 4.2008,
-      "step": 1182
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 4.2574,
+      "step": 590
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.451378809869376e-06,
-      "loss": 4.0974,
-      "step": 1183
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.3062409288824384e-06,
-      "loss": 4.1208,
-      "step": 1184
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.1611030478955007e-06,
-      "loss": 4.0579,
-      "step": 1185
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.015965166908563e-06,
-      "loss": 4.2338,
-      "step": 1186
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 4.4847,
+      "step": 591
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.708272859216255e-07,
-      "loss": 4.0405,
-      "step": 1187
+      "learning_rate": 6.3829787234042555e-06,
+      "loss": 4.579,
+      "step": 592
     },
     {
       "epoch": 1.0,
-      "learning_rate": 7.25689404934688e-07,
-      "loss": 4.0618,
-      "step": 1188
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 4.6057,
+      "step": 593
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.805515239477504e-07,
-      "loss": 4.2369,
-      "step": 1189
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 4.4414,
+      "step": 594
     },
     {
       "epoch": 1.0,
-      "step": 1189,
+      "step": 594,
       "total_flos": 0.0,
-      "train_loss": 4.318590786599431,
-      "train_runtime": 6722.7288,
-      "train_samples_per_second": 4.245,
-      "train_steps_per_second": 0.177
+      "train_loss": 4.292364594912288,
+      "train_runtime": 5438.2563,
+      "train_samples_per_second": 5.248,
+      "train_steps_per_second": 0.109
     }
   ],
-  "max_steps": 1189,
+  "max_steps": 594,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,