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": 1.0,
-  "global_step": 446,
+  "epoch": 0.9995796553173603,
+  "global_step": 1189,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,2690 +10,7166 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.0661,
+      "loss": 9.2645,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.5613,
+      "loss": 8.7817,
       "step": 2
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.8408,
+      "loss": 9.3407,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-08,
-      "loss": 8.8901,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.1586,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-07,
-      "loss": 8.6498,
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 9.0122,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.8e-07,
-      "loss": 8.8394,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 9.1074,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4000000000000003e-07,
-      "loss": 8.7049,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.7889,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 8.5633,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.9644,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-07,
-      "loss": 8.5671,
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 9.2456,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.2e-07,
-      "loss": 8.666,
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 9.0984,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.800000000000001e-07,
-      "loss": 8.5021,
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 8.8478,
       "step": 11
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-07,
-      "loss": 8.4256,
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 9.243,
       "step": 12
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 8.3533,
+      "epoch": 0.01,
+      "learning_rate": 4.2e-07,
+      "loss": 8.9251,
       "step": 13
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 8.5756,
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 8.604,
       "step": 14
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 6.599999999999999e-07,
-      "loss": 8.268,
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 8.8964,
       "step": 15
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-07,
-      "loss": 8.5085,
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 8.5091,
       "step": 16
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.799999999999999e-07,
-      "loss": 8.2511,
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 8.8942,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 8.4e-07,
-      "loss": 8.2268,
+      "epoch": 0.02,
+      "learning_rate": 7.2e-07,
+      "loss": 9.0732,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9e-07,
-      "loss": 8.2555,
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 8.4959,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 8.2484,
+      "epoch": 0.02,
+      "learning_rate": 8.4e-07,
+      "loss": 8.6661,
       "step": 20
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 8.4371,
+      "epoch": 0.02,
+      "learning_rate": 9e-07,
+      "loss": 8.3645,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.08e-06,
-      "loss": 8.2564,
+      "epoch": 0.02,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 8.268,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.14e-06,
-      "loss": 8.1644,
+      "epoch": 0.02,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 8.4268,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 8.202,
+      "epoch": 0.02,
+      "learning_rate": 1.08e-06,
+      "loss": 8.5528,
       "step": 24
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.26e-06,
-      "loss": 8.0417,
+      "epoch": 0.02,
+      "learning_rate": 1.14e-06,
+      "loss": 8.3211,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.3199999999999999e-06,
-      "loss": 8.0118,
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 8.2367,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.38e-06,
-      "loss": 7.8806,
+      "epoch": 0.02,
+      "learning_rate": 1.26e-06,
+      "loss": 8.2729,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.44e-06,
-      "loss": 7.8138,
+      "epoch": 0.02,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 8.0739,
       "step": 28
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.5e-06,
-      "loss": 7.9453,
+      "epoch": 0.02,
+      "learning_rate": 1.38e-06,
+      "loss": 8.0769,
       "step": 29
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.5599999999999999e-06,
-      "loss": 7.7751,
+      "epoch": 0.03,
+      "learning_rate": 1.44e-06,
+      "loss": 8.1845,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.62e-06,
-      "loss": 7.7891,
+      "epoch": 0.03,
+      "learning_rate": 1.5e-06,
+      "loss": 8.0313,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.68e-06,
-      "loss": 7.8177,
+      "epoch": 0.03,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 7.8045,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.74e-06,
-      "loss": 7.726,
+      "epoch": 0.03,
+      "learning_rate": 1.62e-06,
+      "loss": 7.9822,
       "step": 33
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.8e-06,
-      "loss": 7.7002,
+      "epoch": 0.03,
+      "learning_rate": 1.68e-06,
+      "loss": 7.6708,
       "step": 34
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.86e-06,
-      "loss": 7.7217,
+      "epoch": 0.03,
+      "learning_rate": 1.74e-06,
+      "loss": 8.2451,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 7.7386,
+      "epoch": 0.03,
+      "learning_rate": 1.8e-06,
+      "loss": 7.9231,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.98e-06,
-      "loss": 7.6806,
+      "epoch": 0.03,
+      "learning_rate": 1.86e-06,
+      "loss": 7.8853,
       "step": 37
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 7.6063,
+      "epoch": 0.03,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 7.7427,
       "step": 38
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 7.6285,
+      "epoch": 0.03,
+      "learning_rate": 1.98e-06,
+      "loss": 7.9188,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.16e-06,
-      "loss": 7.5746,
+      "epoch": 0.03,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 7.7559,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.22e-06,
-      "loss": 7.6043,
+      "epoch": 0.03,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 7.6086,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.28e-06,
-      "loss": 7.6903,
+      "epoch": 0.04,
+      "learning_rate": 2.16e-06,
+      "loss": 7.7548,
       "step": 42
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.34e-06,
-      "loss": 7.4152,
+      "epoch": 0.04,
+      "learning_rate": 2.22e-06,
+      "loss": 7.6587,
       "step": 43
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 7.6104,
+      "epoch": 0.04,
+      "learning_rate": 2.28e-06,
+      "loss": 7.6761,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.46e-06,
-      "loss": 7.6849,
+      "epoch": 0.04,
+      "learning_rate": 2.34e-06,
+      "loss": 7.8011,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.52e-06,
-      "loss": 7.6706,
+      "epoch": 0.04,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 7.8535,
       "step": 46
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.58e-06,
-      "loss": 7.6569,
+      "epoch": 0.04,
+      "learning_rate": 2.46e-06,
+      "loss": 7.8245,
       "step": 47
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.6399999999999997e-06,
-      "loss": 7.7951,
+      "epoch": 0.04,
+      "learning_rate": 2.52e-06,
+      "loss": 7.9046,
       "step": 48
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.7e-06,
-      "loss": 7.6621,
+      "epoch": 0.04,
+      "learning_rate": 2.58e-06,
+      "loss": 7.801,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.76e-06,
-      "loss": 7.7049,
+      "epoch": 0.04,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 7.9401,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 2.82e-06,
-      "loss": 7.4315,
+      "epoch": 0.04,
+      "learning_rate": 2.7e-06,
+      "loss": 7.5953,
       "step": 51
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.88e-06,
-      "loss": 7.3429,
+      "epoch": 0.04,
+      "learning_rate": 2.76e-06,
+      "loss": 7.4365,
       "step": 52
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 7.4178,
+      "epoch": 0.04,
+      "learning_rate": 2.82e-06,
+      "loss": 7.32,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3e-06,
-      "loss": 7.3289,
+      "epoch": 0.05,
+      "learning_rate": 2.88e-06,
+      "loss": 7.389,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.06e-06,
-      "loss": 7.2022,
+      "epoch": 0.05,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 7.488,
       "step": 55
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.1199999999999998e-06,
-      "loss": 7.3158,
+      "epoch": 0.05,
+      "learning_rate": 3e-06,
+      "loss": 7.2294,
       "step": 56
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.18e-06,
-      "loss": 7.3564,
+      "epoch": 0.05,
+      "learning_rate": 3.06e-06,
+      "loss": 7.6729,
       "step": 57
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.24e-06,
-      "loss": 7.4193,
+      "epoch": 0.05,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 7.2778,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.3e-06,
-      "loss": 7.2878,
+      "epoch": 0.05,
+      "learning_rate": 3.18e-06,
+      "loss": 7.2242,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.36e-06,
-      "loss": 7.2734,
+      "epoch": 0.05,
+      "learning_rate": 3.24e-06,
+      "loss": 7.3987,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 7.2506,
+      "epoch": 0.05,
+      "learning_rate": 3.3e-06,
+      "loss": 7.2189,
       "step": 61
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.48e-06,
-      "loss": 7.2649,
+      "epoch": 0.05,
+      "learning_rate": 3.36e-06,
+      "loss": 7.2479,
       "step": 62
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.54e-06,
-      "loss": 7.2061,
+      "epoch": 0.05,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 7.3781,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 3.6e-06,
-      "loss": 7.2911,
+      "epoch": 0.05,
+      "learning_rate": 3.48e-06,
+      "loss": 7.3324,
       "step": 64
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.66e-06,
-      "loss": 7.2255,
+      "epoch": 0.05,
+      "learning_rate": 3.54e-06,
+      "loss": 7.2571,
       "step": 65
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.72e-06,
-      "loss": 7.1992,
+      "epoch": 0.06,
+      "learning_rate": 3.6e-06,
+      "loss": 7.3767,
       "step": 66
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 7.2259,
+      "epoch": 0.06,
+      "learning_rate": 3.66e-06,
+      "loss": 7.3103,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 7.2138,
+      "epoch": 0.06,
+      "learning_rate": 3.72e-06,
+      "loss": 7.2252,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.9e-06,
-      "loss": 7.2534,
+      "epoch": 0.06,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 7.1928,
       "step": 69
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 3.96e-06,
-      "loss": 7.1484,
+      "epoch": 0.06,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 7.22,
       "step": 70
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 7.2939,
+      "epoch": 0.06,
+      "learning_rate": 3.9e-06,
+      "loss": 7.3639,
       "step": 71
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 7.2613,
+      "epoch": 0.06,
+      "learning_rate": 3.96e-06,
+      "loss": 7.4193,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.14e-06,
-      "loss": 7.2353,
+      "epoch": 0.06,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 7.2587,
       "step": 73
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 7.2668,
+      "epoch": 0.06,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 7.3456,
       "step": 74
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.26e-06,
-      "loss": 7.2238,
+      "epoch": 0.06,
+      "learning_rate": 4.14e-06,
+      "loss": 7.1123,
       "step": 75
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.32e-06,
-      "loss": 7.2507,
+      "epoch": 0.06,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 7.4739,
       "step": 76
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.3799999999999996e-06,
-      "loss": 7.2616,
+      "epoch": 0.06,
+      "learning_rate": 4.26e-06,
+      "loss": 7.224,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.44e-06,
-      "loss": 7.2783,
+      "epoch": 0.07,
+      "learning_rate": 4.32e-06,
+      "loss": 7.078,
       "step": 78
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.5e-06,
-      "loss": 7.1368,
+      "epoch": 0.07,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 7.1178,
       "step": 79
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.56e-06,
-      "loss": 7.2477,
+      "epoch": 0.07,
+      "learning_rate": 4.44e-06,
+      "loss": 7.1862,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.62e-06,
-      "loss": 7.2056,
+      "epoch": 0.07,
+      "learning_rate": 4.5e-06,
+      "loss": 7.3286,
       "step": 81
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.68e-06,
-      "loss": 7.2238,
+      "epoch": 0.07,
+      "learning_rate": 4.56e-06,
+      "loss": 7.3456,
       "step": 82
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.74e-06,
-      "loss": 7.2219,
+      "epoch": 0.07,
+      "learning_rate": 4.62e-06,
+      "loss": 7.1114,
       "step": 83
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 7.2008,
+      "epoch": 0.07,
+      "learning_rate": 4.68e-06,
+      "loss": 7.2386,
       "step": 84
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.86e-06,
-      "loss": 7.1812,
+      "epoch": 0.07,
+      "learning_rate": 4.74e-06,
+      "loss": 7.1793,
       "step": 85
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.92e-06,
-      "loss": 7.2346,
+      "epoch": 0.07,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 7.2678,
       "step": 86
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 7.1994,
+      "epoch": 0.07,
+      "learning_rate": 4.86e-06,
+      "loss": 7.1509,
       "step": 87
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.04e-06,
-      "loss": 7.1441,
+      "epoch": 0.07,
+      "learning_rate": 4.92e-06,
+      "loss": 7.2565,
       "step": 88
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.1e-06,
-      "loss": 7.1018,
+      "epoch": 0.07,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 7.3592,
       "step": 89
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.16e-06,
-      "loss": 7.2426,
+      "epoch": 0.08,
+      "learning_rate": 5.04e-06,
+      "loss": 7.1701,
       "step": 90
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 5.22e-06,
-      "loss": 7.1448,
+      "epoch": 0.08,
+      "learning_rate": 5.1e-06,
+      "loss": 7.4446,
       "step": 91
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.279999999999999e-06,
-      "loss": 7.3725,
+      "epoch": 0.08,
+      "learning_rate": 5.16e-06,
+      "loss": 7.289,
       "step": 92
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.34e-06,
-      "loss": 7.2283,
+      "epoch": 0.08,
+      "learning_rate": 5.22e-06,
+      "loss": 6.9803,
       "step": 93
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.4e-06,
-      "loss": 7.0783,
+      "epoch": 0.08,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 7.2983,
       "step": 94
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.46e-06,
-      "loss": 6.9872,
+      "epoch": 0.08,
+      "learning_rate": 5.34e-06,
+      "loss": 7.4766,
       "step": 95
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.52e-06,
-      "loss": 7.219,
+      "epoch": 0.08,
+      "learning_rate": 5.4e-06,
+      "loss": 7.1428,
       "step": 96
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.58e-06,
-      "loss": 7.1056,
+      "epoch": 0.08,
+      "learning_rate": 5.46e-06,
+      "loss": 7.3311,
       "step": 97
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.64e-06,
-      "loss": 6.8522,
+      "epoch": 0.08,
+      "learning_rate": 5.52e-06,
+      "loss": 7.2306,
       "step": 98
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 6.9854,
+      "epoch": 0.08,
+      "learning_rate": 5.58e-06,
+      "loss": 6.9861,
       "step": 99
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.76e-06,
-      "loss": 6.6152,
+      "epoch": 0.08,
+      "learning_rate": 5.64e-06,
+      "loss": 6.9041,
       "step": 100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.82e-06,
-      "loss": 7.2604,
+      "epoch": 0.08,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 7.1725,
       "step": 101
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 7.2338,
+      "epoch": 0.09,
+      "learning_rate": 5.76e-06,
+      "loss": 7.0943,
       "step": 102
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 7.1187,
+      "epoch": 0.09,
+      "learning_rate": 5.82e-06,
+      "loss": 7.1549,
       "step": 103
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 6e-06,
-      "loss": 7.1481,
+      "epoch": 0.09,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 7.1566,
       "step": 104
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 7.157,
+      "epoch": 0.09,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 7.0838,
       "step": 105
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.12e-06,
-      "loss": 7.1806,
+      "epoch": 0.09,
+      "learning_rate": 6e-06,
+      "loss": 7.1625,
       "step": 106
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.18e-06,
-      "loss": 7.1208,
+      "epoch": 0.09,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 7.1241,
       "step": 107
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.2399999999999995e-06,
-      "loss": 7.1334,
+      "epoch": 0.09,
+      "learning_rate": 6.12e-06,
+      "loss": 7.1856,
       "step": 108
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.3e-06,
-      "loss": 7.0156,
+      "epoch": 0.09,
+      "learning_rate": 6.18e-06,
+      "loss": 7.3023,
       "step": 109
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.36e-06,
-      "loss": 7.1107,
+      "epoch": 0.09,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 7.1465,
       "step": 110
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.42e-06,
-      "loss": 7.0161,
+      "epoch": 0.09,
+      "learning_rate": 6.3e-06,
+      "loss": 7.2196,
       "step": 111
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.48e-06,
-      "loss": 7.0205,
+      "epoch": 0.09,
+      "learning_rate": 6.36e-06,
+      "loss": 7.1786,
       "step": 112
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 6.54e-06,
-      "loss": 7.0943,
+      "epoch": 0.09,
+      "learning_rate": 6.42e-06,
+      "loss": 7.1026,
       "step": 113
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.6e-06,
-      "loss": 7.1342,
+      "epoch": 0.1,
+      "learning_rate": 6.48e-06,
+      "loss": 7.0959,
       "step": 114
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 7.066,
+      "epoch": 0.1,
+      "learning_rate": 6.54e-06,
+      "loss": 7.1844,
       "step": 115
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.72e-06,
-      "loss": 7.1983,
+      "epoch": 0.1,
+      "learning_rate": 6.6e-06,
+      "loss": 7.3039,
       "step": 116
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.78e-06,
-      "loss": 7.1548,
+      "epoch": 0.1,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 7.113,
       "step": 117
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 7.1237,
+      "epoch": 0.1,
+      "learning_rate": 6.72e-06,
+      "loss": 6.9747,
       "step": 118
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 7.0314,
+      "epoch": 0.1,
+      "learning_rate": 6.78e-06,
+      "loss": 6.9265,
       "step": 119
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 6.96e-06,
-      "loss": 7.1185,
+      "epoch": 0.1,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 7.1765,
       "step": 120
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 7.059,
+      "epoch": 0.1,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 7.2618,
       "step": 121
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.08e-06,
-      "loss": 7.0087,
+      "epoch": 0.1,
+      "learning_rate": 6.96e-06,
+      "loss": 7.0998,
       "step": 122
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.14e-06,
-      "loss": 7.0584,
+      "epoch": 0.1,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 7.1757,
       "step": 123
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.2e-06,
-      "loss": 7.0146,
+      "epoch": 0.1,
+      "learning_rate": 7.08e-06,
+      "loss": 7.1495,
       "step": 124
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.26e-06,
-      "loss": 7.1049,
+      "epoch": 0.11,
+      "learning_rate": 7.14e-06,
+      "loss": 7.103,
       "step": 125
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.32e-06,
-      "loss": 7.1239,
+      "epoch": 0.11,
+      "learning_rate": 7.2e-06,
+      "loss": 6.9668,
       "step": 126
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 7.1031,
+      "epoch": 0.11,
+      "learning_rate": 7.26e-06,
+      "loss": 7.3312,
       "step": 127
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.44e-06,
-      "loss": 7.0613,
+      "epoch": 0.11,
+      "learning_rate": 7.32e-06,
+      "loss": 7.0505,
       "step": 128
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.5e-06,
-      "loss": 7.127,
-      "step": 129
+      "epoch": 0.11,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 7.0606,
+      "step": 129
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 7.1652,
+      "epoch": 0.11,
+      "learning_rate": 7.44e-06,
+      "loss": 7.0443,
       "step": 130
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.62e-06,
-      "loss": 6.9731,
+      "epoch": 0.11,
+      "learning_rate": 7.5e-06,
+      "loss": 6.9168,
       "step": 131
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.680000000000001e-06,
-      "loss": 7.1539,
+      "epoch": 0.11,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 7.0832,
       "step": 132
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.74e-06,
-      "loss": 7.0214,
+      "epoch": 0.11,
+      "learning_rate": 7.62e-06,
+      "loss": 7.2462,
       "step": 133
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.8e-06,
-      "loss": 7.0548,
+      "epoch": 0.11,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 7.0327,
       "step": 134
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 7.1714,
+      "epoch": 0.11,
+      "learning_rate": 7.74e-06,
+      "loss": 7.0976,
       "step": 135
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.92e-06,
-      "loss": 6.9496,
+      "epoch": 0.11,
+      "learning_rate": 7.8e-06,
+      "loss": 7.0195,
       "step": 136
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.98e-06,
-      "loss": 7.1553,
+      "epoch": 0.12,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 7.0406,
       "step": 137
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 7.0663,
+      "epoch": 0.12,
+      "learning_rate": 7.92e-06,
+      "loss": 7.1136,
       "step": 138
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.1e-06,
-      "loss": 7.0589,
+      "epoch": 0.12,
+      "learning_rate": 7.98e-06,
+      "loss": 7.039,
       "step": 139
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 7.1641,
+      "epoch": 0.12,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 7.1659,
       "step": 140
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.220000000000001e-06,
-      "loss": 7.1462,
+      "epoch": 0.12,
+      "learning_rate": 8.1e-06,
+      "loss": 7.093,
       "step": 141
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.28e-06,
-      "loss": 7.1284,
+      "epoch": 0.12,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 7.0387,
       "step": 142
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.340000000000001e-06,
-      "loss": 6.9173,
+      "epoch": 0.12,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 6.9572,
       "step": 143
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 6.9755,
+      "epoch": 0.12,
+      "learning_rate": 8.28e-06,
+      "loss": 7.0803,
       "step": 144
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.459999999999999e-06,
-      "loss": 6.9108,
+      "epoch": 0.12,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 6.9381,
       "step": 145
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.52e-06,
-      "loss": 6.9447,
+      "epoch": 0.12,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 7.1332,
       "step": 146
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.58e-06,
-      "loss": 7.0189,
+      "epoch": 0.12,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 6.8416,
       "step": 147
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.64e-06,
-      "loss": 7.0141,
+      "epoch": 0.12,
+      "learning_rate": 8.52e-06,
+      "loss": 6.6269,
       "step": 148
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 8.7e-06,
-      "loss": 6.795,
+      "epoch": 0.13,
+      "learning_rate": 8.58e-06,
+      "loss": 6.7934,
       "step": 149
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.759999999999999e-06,
-      "loss": 6.657,
+      "epoch": 0.13,
+      "learning_rate": 8.64e-06,
+      "loss": 6.66,
       "step": 150
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.82e-06,
-      "loss": 7.1582,
+      "epoch": 0.13,
+      "learning_rate": 8.7e-06,
+      "loss": 7.2848,
       "step": 151
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.88e-06,
-      "loss": 7.1828,
+      "epoch": 0.13,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 7.3452,
       "step": 152
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 7.1247,
+      "epoch": 0.13,
+      "learning_rate": 8.82e-06,
+      "loss": 7.1858,
       "step": 153
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9e-06,
-      "loss": 7.0606,
+      "epoch": 0.13,
+      "learning_rate": 8.88e-06,
+      "loss": 7.2224,
       "step": 154
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.06e-06,
-      "loss": 7.0153,
+      "epoch": 0.13,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 7.0784,
       "step": 155
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.12e-06,
-      "loss": 7.0451,
+      "epoch": 0.13,
+      "learning_rate": 9e-06,
+      "loss": 7.0554,
       "step": 156
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.18e-06,
-      "loss": 7.1135,
+      "epoch": 0.13,
+      "learning_rate": 9.06e-06,
+      "loss": 7.0581,
       "step": 157
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.24e-06,
-      "loss": 7.0684,
+      "epoch": 0.13,
+      "learning_rate": 9.12e-06,
+      "loss": 7.2136,
       "step": 158
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.3e-06,
-      "loss": 7.0523,
+      "epoch": 0.13,
+      "learning_rate": 9.18e-06,
+      "loss": 6.9891,
       "step": 159
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.36e-06,
-      "loss": 7.0882,
+      "epoch": 0.13,
+      "learning_rate": 9.24e-06,
+      "loss": 6.9661,
       "step": 160
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.42e-06,
-      "loss": 7.0388,
+      "epoch": 0.14,
+      "learning_rate": 9.3e-06,
+      "loss": 7.0525,
       "step": 161
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.48e-06,
-      "loss": 7.0899,
+      "epoch": 0.14,
+      "learning_rate": 9.36e-06,
+      "loss": 7.1545,
       "step": 162
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.54e-06,
-      "loss": 7.0785,
+      "epoch": 0.14,
+      "learning_rate": 9.42e-06,
+      "loss": 7.2081,
       "step": 163
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 7.069,
+      "epoch": 0.14,
+      "learning_rate": 9.48e-06,
+      "loss": 7.0543,
       "step": 164
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.66e-06,
-      "loss": 7.0036,
+      "epoch": 0.14,
+      "learning_rate": 9.54e-06,
+      "loss": 7.028,
       "step": 165
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.72e-06,
-      "loss": 7.0883,
+      "epoch": 0.14,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 6.9173,
       "step": 166
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 6.9732,
+      "epoch": 0.14,
+      "learning_rate": 9.66e-06,
+      "loss": 7.0702,
       "step": 167
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.84e-06,
-      "loss": 6.9378,
+      "epoch": 0.14,
+      "learning_rate": 9.72e-06,
+      "loss": 7.0819,
       "step": 168
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.9e-06,
-      "loss": 6.9888,
+      "epoch": 0.14,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 7.1053,
       "step": 169
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 7.1208,
+      "epoch": 0.14,
+      "learning_rate": 9.84e-06,
+      "loss": 6.9988,
       "step": 170
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.002e-05,
-      "loss": 7.0391,
+      "epoch": 0.14,
+      "learning_rate": 9.9e-06,
+      "loss": 7.0302,
       "step": 171
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.008e-05,
-      "loss": 7.0709,
+      "epoch": 0.14,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 7.0264,
       "step": 172
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 7.0227,
+      "epoch": 0.15,
+      "learning_rate": 1.002e-05,
+      "loss": 6.8528,
       "step": 173
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.02e-05,
-      "loss": 7.081,
+      "epoch": 0.15,
+      "learning_rate": 1.008e-05,
+      "loss": 7.0871,
       "step": 174
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.0260000000000002e-05,
-      "loss": 7.0427,
+      "epoch": 0.15,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 7.119,
       "step": 175
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.032e-05,
-      "loss": 6.9713,
+      "epoch": 0.15,
+      "learning_rate": 1.02e-05,
+      "loss": 7.2414,
       "step": 176
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.0379999999999999e-05,
-      "loss": 6.9645,
+      "epoch": 0.15,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 7.1961,
       "step": 177
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.044e-05,
-      "loss": 7.0354,
+      "epoch": 0.15,
+      "learning_rate": 1.032e-05,
+      "loss": 7.1284,
       "step": 178
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.05e-05,
-      "loss": 7.0338,
+      "epoch": 0.15,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 7.0386,
       "step": 179
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.0559999999999999e-05,
-      "loss": 7.095,
+      "epoch": 0.15,
+      "learning_rate": 1.044e-05,
+      "loss": 7.1003,
       "step": 180
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.062e-05,
-      "loss": 7.0756,
+      "epoch": 0.15,
+      "learning_rate": 1.05e-05,
+      "loss": 7.1393,
       "step": 181
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.068e-05,
-      "loss": 6.9755,
+      "epoch": 0.15,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 7.2247,
       "step": 182
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.074e-05,
-      "loss": 7.0655,
+      "epoch": 0.15,
+      "learning_rate": 1.062e-05,
+      "loss": 7.0304,
       "step": 183
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.08e-05,
-      "loss": 7.1194,
+      "epoch": 0.15,
+      "learning_rate": 1.068e-05,
+      "loss": 7.0428,
       "step": 184
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.086e-05,
-      "loss": 7.0337,
+      "epoch": 0.16,
+      "learning_rate": 1.074e-05,
+      "loss": 6.9073,
       "step": 185
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.092e-05,
-      "loss": 6.9139,
+      "epoch": 0.16,
+      "learning_rate": 1.08e-05,
+      "loss": 6.9143,
       "step": 186
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.098e-05,
-      "loss": 7.0081,
+      "epoch": 0.16,
+      "learning_rate": 1.086e-05,
+      "loss": 7.1989,
       "step": 187
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.104e-05,
-      "loss": 6.9888,
+      "epoch": 0.16,
+      "learning_rate": 1.092e-05,
+      "loss": 7.0031,
       "step": 188
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.11e-05,
-      "loss": 7.0655,
+      "epoch": 0.16,
+      "learning_rate": 1.098e-05,
+      "loss": 7.0327,
       "step": 189
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.116e-05,
-      "loss": 7.1192,
-      "step": 190
+      "epoch": 0.16,
+      "learning_rate": 1.104e-05,
+      "loss": 6.9772,
+      "step": 190
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.1220000000000001e-05,
-      "loss": 7.1267,
+      "epoch": 0.16,
+      "learning_rate": 1.11e-05,
+      "loss": 7.2237,
       "step": 191
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.128e-05,
-      "loss": 6.9655,
+      "epoch": 0.16,
+      "learning_rate": 1.116e-05,
+      "loss": 7.2198,
       "step": 192
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.134e-05,
-      "loss": 7.0361,
+      "epoch": 0.16,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 7.0253,
       "step": 193
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 6.9628,
+      "epoch": 0.16,
+      "learning_rate": 1.128e-05,
+      "loss": 7.0092,
       "step": 194
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.146e-05,
-      "loss": 6.9358,
+      "epoch": 0.16,
+      "learning_rate": 1.134e-05,
+      "loss": 6.895,
       "step": 195
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.152e-05,
-      "loss": 6.8373,
+      "epoch": 0.16,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 6.8361,
       "step": 196
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 6.7758,
+      "epoch": 0.17,
+      "learning_rate": 1.146e-05,
+      "loss": 7.1704,
       "step": 197
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.164e-05,
-      "loss": 6.9332,
+      "epoch": 0.17,
+      "learning_rate": 1.152e-05,
+      "loss": 6.8643,
       "step": 198
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 6.7629,
+      "epoch": 0.17,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 6.7912,
       "step": 199
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 6.5633,
+      "epoch": 0.17,
+      "learning_rate": 1.164e-05,
+      "loss": 6.3712,
       "step": 200
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.182e-05,
-      "loss": 7.2408,
+      "epoch": 0.17,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 7.115,
       "step": 201
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 7.2488,
+      "epoch": 0.17,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 7.0692,
       "step": 202
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 7.2284,
+      "epoch": 0.17,
+      "learning_rate": 1.182e-05,
+      "loss": 7.0783,
       "step": 203
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2e-05,
-      "loss": 7.1491,
+      "epoch": 0.17,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 6.9704,
       "step": 204
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2060000000000001e-05,
-      "loss": 7.1344,
+      "epoch": 0.17,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 6.9688,
       "step": 205
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 7.1017,
+      "epoch": 0.17,
+      "learning_rate": 1.2e-05,
+      "loss": 7.0884,
       "step": 206
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 6.9784,
+      "epoch": 0.17,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 7.0586,
       "step": 207
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.224e-05,
-      "loss": 7.1243,
+      "epoch": 0.17,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 7.2809,
       "step": 208
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.2299999999999999e-05,
-      "loss": 7.0496,
+      "epoch": 0.18,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 7.1043,
       "step": 209
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.236e-05,
-      "loss": 7.0962,
+      "epoch": 0.18,
+      "learning_rate": 1.224e-05,
+      "loss": 7.2292,
       "step": 210
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.242e-05,
-      "loss": 7.0022,
+      "epoch": 0.18,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 7.0086,
       "step": 211
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.2479999999999999e-05,
-      "loss": 7.0542,
+      "epoch": 0.18,
+      "learning_rate": 1.236e-05,
+      "loss": 6.9504,
       "step": 212
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.254e-05,
-      "loss": 6.9093,
+      "epoch": 0.18,
+      "learning_rate": 1.242e-05,
+      "loss": 7.041,
       "step": 213
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.26e-05,
-      "loss": 7.009,
+      "epoch": 0.18,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 7.0411,
       "step": 214
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 6.9863,
+      "epoch": 0.18,
+      "learning_rate": 1.254e-05,
+      "loss": 7.0403,
       "step": 215
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.272e-05,
-      "loss": 7.0156,
+      "epoch": 0.18,
+      "learning_rate": 1.26e-05,
+      "loss": 7.009,
       "step": 216
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.278e-05,
-      "loss": 6.9996,
+      "epoch": 0.18,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 7.0629,
       "step": 217
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.284e-05,
-      "loss": 7.0404,
+      "epoch": 0.18,
+      "learning_rate": 1.272e-05,
+      "loss": 7.0728,
       "step": 218
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.29e-05,
-      "loss": 7.0057,
+      "epoch": 0.18,
+      "learning_rate": 1.278e-05,
+      "loss": 6.9722,
       "step": 219
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.296e-05,
-      "loss": 7.0278,
+      "epoch": 0.18,
+      "learning_rate": 1.284e-05,
+      "loss": 7.1592,
       "step": 220
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.302e-05,
-      "loss": 7.0697,
+      "epoch": 0.19,
+      "learning_rate": 1.29e-05,
+      "loss": 6.9572,
       "step": 221
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.308e-05,
-      "loss": 7.0499,
+      "epoch": 0.19,
+      "learning_rate": 1.296e-05,
+      "loss": 7.2919,
       "step": 222
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.314e-05,
-      "loss": 6.9859,
+      "epoch": 0.19,
+      "learning_rate": 1.302e-05,
+      "loss": 7.1749,
       "step": 223
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.32e-05,
-      "loss": 6.9388,
+      "epoch": 0.19,
+      "learning_rate": 1.308e-05,
+      "loss": 7.1748,
       "step": 224
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.326e-05,
-      "loss": 7.1033,
+      "epoch": 0.19,
+      "learning_rate": 1.314e-05,
+      "loss": 6.9979,
       "step": 225
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 7.0169,
+      "epoch": 0.19,
+      "learning_rate": 1.32e-05,
+      "loss": 6.8968,
       "step": 226
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.338e-05,
-      "loss": 7.1146,
+      "epoch": 0.19,
+      "learning_rate": 1.326e-05,
+      "loss": 7.1294,
       "step": 227
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.344e-05,
-      "loss": 7.0342,
+      "epoch": 0.19,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 7.1157,
       "step": 228
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 6.9501,
+      "epoch": 0.19,
+      "learning_rate": 1.338e-05,
+      "loss": 6.9978,
       "step": 229
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.356e-05,
-      "loss": 7.0036,
+      "epoch": 0.19,
+      "learning_rate": 1.344e-05,
+      "loss": 7.0651,
       "step": 230
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.362e-05,
-      "loss": 7.0218,
+      "epoch": 0.19,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 7.1366,
       "step": 231
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 6.8725,
+      "epoch": 0.2,
+      "learning_rate": 1.356e-05,
+      "loss": 7.0621,
       "step": 232
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.374e-05,
-      "loss": 6.9275,
+      "epoch": 0.2,
+      "learning_rate": 1.362e-05,
+      "loss": 7.0283,
       "step": 233
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 6.9323,
+      "epoch": 0.2,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 6.9114,
       "step": 234
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 7.0464,
+      "epoch": 0.2,
+      "learning_rate": 1.374e-05,
+      "loss": 7.1182,
       "step": 235
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.392e-05,
-      "loss": 6.9281,
+      "epoch": 0.2,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 7.1036,
       "step": 236
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 7.0814,
+      "epoch": 0.2,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 7.0777,
       "step": 237
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 6.9639,
+      "epoch": 0.2,
+      "learning_rate": 1.392e-05,
+      "loss": 6.9912,
       "step": 238
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 6.8773,
+      "epoch": 0.2,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 7.0432,
       "step": 239
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.416e-05,
-      "loss": 7.1066,
+      "epoch": 0.2,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 6.8894,
       "step": 240
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.422e-05,
-      "loss": 6.9166,
+      "epoch": 0.2,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 7.0478,
       "step": 241
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.428e-05,
-      "loss": 6.9288,
+      "epoch": 0.2,
+      "learning_rate": 1.416e-05,
+      "loss": 7.0257,
       "step": 242
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.434e-05,
-      "loss": 7.0485,
+      "epoch": 0.2,
+      "learning_rate": 1.422e-05,
+      "loss": 7.1119,
       "step": 243
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.44e-05,
-      "loss": 6.8353,
+      "epoch": 0.21,
+      "learning_rate": 1.428e-05,
+      "loss": 7.1485,
       "step": 244
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.446e-05,
-      "loss": 6.8549,
+      "epoch": 0.21,
+      "learning_rate": 1.434e-05,
+      "loss": 6.8617,
       "step": 245
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.452e-05,
-      "loss": 6.8157,
+      "epoch": 0.21,
+      "learning_rate": 1.44e-05,
+      "loss": 6.8215,
       "step": 246
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.458e-05,
-      "loss": 6.5884,
+      "epoch": 0.21,
+      "learning_rate": 1.446e-05,
+      "loss": 7.3806,
       "step": 247
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.464e-05,
-      "loss": 6.8573,
+      "epoch": 0.21,
+      "learning_rate": 1.452e-05,
+      "loss": 6.7644,
       "step": 248
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.47e-05,
-      "loss": 6.7128,
+      "epoch": 0.21,
+      "learning_rate": 1.458e-05,
+      "loss": 6.6447,
       "step": 249
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 6.4373,
+      "epoch": 0.21,
+      "learning_rate": 1.464e-05,
+      "loss": 6.4345,
       "step": 250
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.482e-05,
-      "loss": 7.2887,
-      "step": 251
+      "epoch": 0.21,
+      "learning_rate": 1.47e-05,
+      "loss": 7.2032,
+      "step": 251
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.488e-05,
-      "loss": 7.2864,
+      "epoch": 0.21,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 6.9527,
       "step": 252
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.4940000000000001e-05,
-      "loss": 7.2313,
+      "epoch": 0.21,
+      "learning_rate": 1.482e-05,
+      "loss": 7.1685,
       "step": 253
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.5e-05,
-      "loss": 7.1999,
+      "epoch": 0.21,
+      "learning_rate": 1.488e-05,
+      "loss": 6.9349,
       "step": 254
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.506e-05,
-      "loss": 7.1424,
+      "epoch": 0.21,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 6.9983,
       "step": 255
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.5120000000000001e-05,
-      "loss": 7.0273,
+      "epoch": 0.22,
+      "learning_rate": 1.5e-05,
+      "loss": 7.0868,
       "step": 256
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.518e-05,
-      "loss": 6.9975,
+      "epoch": 0.22,
+      "learning_rate": 1.506e-05,
+      "loss": 6.9868,
       "step": 257
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.524e-05,
-      "loss": 7.0017,
+      "epoch": 0.22,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 7.1154,
       "step": 258
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.53e-05,
-      "loss": 6.9275,
+      "epoch": 0.22,
+      "learning_rate": 1.518e-05,
+      "loss": 6.9745,
       "step": 259
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.5360000000000002e-05,
-      "loss": 6.965,
+      "epoch": 0.22,
+      "learning_rate": 1.524e-05,
+      "loss": 6.9459,
       "step": 260
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.542e-05,
-      "loss": 6.9751,
+      "epoch": 0.22,
+      "learning_rate": 1.53e-05,
+      "loss": 6.999,
       "step": 261
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.548e-05,
-      "loss": 6.944,
+      "epoch": 0.22,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 6.9631,
       "step": 262
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.554e-05,
-      "loss": 7.0152,
+      "epoch": 0.22,
+      "learning_rate": 1.542e-05,
+      "loss": 6.9308,
       "step": 263
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.56e-05,
-      "loss": 7.002,
+      "epoch": 0.22,
+      "learning_rate": 1.548e-05,
+      "loss": 7.047,
       "step": 264
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.5660000000000003e-05,
-      "loss": 7.0258,
+      "epoch": 0.22,
+      "learning_rate": 1.554e-05,
+      "loss": 6.9798,
       "step": 265
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.5720000000000002e-05,
-      "loss": 6.8102,
+      "epoch": 0.22,
+      "learning_rate": 1.56e-05,
+      "loss": 7.029,
       "step": 266
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.578e-05,
-      "loss": 6.9953,
+      "epoch": 0.22,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 6.9405,
       "step": 267
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.584e-05,
-      "loss": 6.8559,
+      "epoch": 0.23,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 6.9293,
       "step": 268
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.59e-05,
-      "loss": 6.9861,
+      "epoch": 0.23,
+      "learning_rate": 1.578e-05,
+      "loss": 7.0008,
       "step": 269
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.596e-05,
-      "loss": 7.0232,
+      "epoch": 0.23,
+      "learning_rate": 1.584e-05,
+      "loss": 7.0297,
       "step": 270
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6020000000000002e-05,
-      "loss": 6.8916,
+      "epoch": 0.23,
+      "learning_rate": 1.59e-05,
+      "loss": 7.0667,
       "step": 271
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6080000000000002e-05,
-      "loss": 6.9062,
+      "epoch": 0.23,
+      "learning_rate": 1.596e-05,
+      "loss": 6.9658,
       "step": 272
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.614e-05,
-      "loss": 6.8069,
+      "epoch": 0.23,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 7.0879,
       "step": 273
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.62e-05,
-      "loss": 6.7746,
+      "epoch": 0.23,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 7.0809,
       "step": 274
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.626e-05,
-      "loss": 7.0022,
+      "epoch": 0.23,
+      "learning_rate": 1.614e-05,
+      "loss": 7.089,
       "step": 275
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6320000000000003e-05,
-      "loss": 6.8936,
+      "epoch": 0.23,
+      "learning_rate": 1.62e-05,
+      "loss": 6.8732,
       "step": 276
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6380000000000002e-05,
-      "loss": 6.8545,
+      "epoch": 0.23,
+      "learning_rate": 1.626e-05,
+      "loss": 6.9728,
       "step": 277
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6440000000000002e-05,
-      "loss": 6.9683,
+      "epoch": 0.23,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 6.9928,
       "step": 278
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.65e-05,
-      "loss": 7.0178,
+      "epoch": 0.23,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 7.1316,
       "step": 279
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.656e-05,
-      "loss": 7.0096,
+      "epoch": 0.24,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 6.8699,
       "step": 280
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6620000000000004e-05,
-      "loss": 6.9511,
+      "epoch": 0.24,
+      "learning_rate": 1.65e-05,
+      "loss": 6.8749,
       "step": 281
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6680000000000003e-05,
-      "loss": 6.9008,
+      "epoch": 0.24,
+      "learning_rate": 1.656e-05,
+      "loss": 6.9679,
       "step": 282
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6740000000000002e-05,
-      "loss": 6.8291,
+      "epoch": 0.24,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 7.1804,
       "step": 283
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 6.9077,
+      "epoch": 0.24,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 7.1059,
       "step": 284
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.686e-05,
-      "loss": 6.9171,
+      "epoch": 0.24,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 6.9953,
       "step": 285
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.6919999999999997e-05,
-      "loss": 6.9774,
+      "epoch": 0.24,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 7.0,
       "step": 286
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.698e-05,
-      "loss": 6.9438,
+      "epoch": 0.24,
+      "learning_rate": 1.686e-05,
+      "loss": 6.9042,
       "step": 287
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.704e-05,
-      "loss": 6.9102,
+      "epoch": 0.24,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 6.9412,
       "step": 288
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.71e-05,
-      "loss": 6.8897,
+      "epoch": 0.24,
+      "learning_rate": 1.698e-05,
+      "loss": 7.0456,
       "step": 289
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.716e-05,
-      "loss": 6.9512,
+      "epoch": 0.24,
+      "learning_rate": 1.704e-05,
+      "loss": 6.9086,
       "step": 290
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.7219999999999998e-05,
-      "loss": 6.9808,
+      "epoch": 0.24,
+      "learning_rate": 1.71e-05,
+      "loss": 7.1757,
       "step": 291
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.728e-05,
-      "loss": 6.9348,
+      "epoch": 0.25,
+      "learning_rate": 1.716e-05,
+      "loss": 7.0248,
       "step": 292
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.734e-05,
-      "loss": 6.8064,
+      "epoch": 0.25,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 6.7068,
       "step": 293
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.74e-05,
-      "loss": 6.7494,
+      "epoch": 0.25,
+      "learning_rate": 1.728e-05,
+      "loss": 6.8182,
       "step": 294
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.746e-05,
-      "loss": 6.735,
+      "epoch": 0.25,
+      "learning_rate": 1.734e-05,
+      "loss": 6.9292,
       "step": 295
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.7519999999999998e-05,
-      "loss": 6.7118,
+      "epoch": 0.25,
+      "learning_rate": 1.74e-05,
+      "loss": 7.0076,
       "step": 296
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.758e-05,
-      "loss": 6.7464,
+      "epoch": 0.25,
+      "learning_rate": 1.746e-05,
+      "loss": 6.7041,
       "step": 297
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.764e-05,
-      "loss": 6.7746,
+      "epoch": 0.25,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 6.7559,
       "step": 298
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.77e-05,
-      "loss": 6.3516,
+      "epoch": 0.25,
+      "learning_rate": 1.758e-05,
+      "loss": 6.8882,
       "step": 299
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.776e-05,
-      "loss": 6.3981,
+      "epoch": 0.25,
+      "learning_rate": 1.764e-05,
+      "loss": 6.5027,
       "step": 300
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.782e-05,
-      "loss": 7.3073,
+      "epoch": 0.25,
+      "learning_rate": 1.77e-05,
+      "loss": 7.0319,
       "step": 301
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.7879999999999998e-05,
-      "loss": 7.2329,
+      "epoch": 0.25,
+      "learning_rate": 1.776e-05,
+      "loss": 7.1752,
       "step": 302
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.794e-05,
-      "loss": 7.1772,
+      "epoch": 0.25,
+      "learning_rate": 1.782e-05,
+      "loss": 7.2225,
       "step": 303
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.8e-05,
-      "loss": 7.0102,
+      "epoch": 0.26,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 6.9304,
       "step": 304
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.806e-05,
-      "loss": 7.0688,
+      "epoch": 0.26,
+      "learning_rate": 1.794e-05,
+      "loss": 6.9973,
       "step": 305
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.812e-05,
-      "loss": 6.8743,
+      "epoch": 0.26,
+      "learning_rate": 1.8e-05,
+      "loss": 6.9159,
       "step": 306
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.818e-05,
-      "loss": 6.8592,
+      "epoch": 0.26,
+      "learning_rate": 1.806e-05,
+      "loss": 6.9768,
       "step": 307
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.824e-05,
-      "loss": 7.0024,
+      "epoch": 0.26,
+      "learning_rate": 1.812e-05,
+      "loss": 6.9426,
       "step": 308
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.83e-05,
-      "loss": 6.8502,
+      "epoch": 0.26,
+      "learning_rate": 1.818e-05,
+      "loss": 6.8278,
       "step": 309
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.836e-05,
-      "loss": 6.9345,
+      "epoch": 0.26,
+      "learning_rate": 1.824e-05,
+      "loss": 7.0905,
       "step": 310
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.842e-05,
-      "loss": 6.8982,
+      "epoch": 0.26,
+      "learning_rate": 1.83e-05,
+      "loss": 6.7491,
       "step": 311
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.848e-05,
-      "loss": 6.9162,
+      "epoch": 0.26,
+      "learning_rate": 1.836e-05,
+      "loss": 6.7998,
       "step": 312
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.854e-05,
-      "loss": 6.8384,
+      "epoch": 0.26,
+      "learning_rate": 1.842e-05,
+      "loss": 6.8631,
       "step": 313
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.86e-05,
-      "loss": 6.8191,
+      "epoch": 0.26,
+      "learning_rate": 1.848e-05,
+      "loss": 6.9686,
       "step": 314
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.866e-05,
-      "loss": 6.9335,
+      "epoch": 0.26,
+      "learning_rate": 1.854e-05,
+      "loss": 6.8764,
       "step": 315
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.872e-05,
-      "loss": 6.8588,
+      "epoch": 0.27,
+      "learning_rate": 1.86e-05,
+      "loss": 7.1163,
       "step": 316
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.878e-05,
-      "loss": 6.8222,
+      "epoch": 0.27,
+      "learning_rate": 1.866e-05,
+      "loss": 6.9662,
       "step": 317
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.884e-05,
-      "loss": 6.7389,
+      "epoch": 0.27,
+      "learning_rate": 1.872e-05,
+      "loss": 7.0056,
       "step": 318
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 6.8438,
+      "epoch": 0.27,
+      "learning_rate": 1.878e-05,
+      "loss": 6.8595,
       "step": 319
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.896e-05,
-      "loss": 6.8296,
+      "epoch": 0.27,
+      "learning_rate": 1.884e-05,
+      "loss": 6.8479,
       "step": 320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.902e-05,
-      "loss": 6.8216,
+      "epoch": 0.27,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 6.8847,
       "step": 321
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.908e-05,
-      "loss": 6.7952,
+      "epoch": 0.27,
+      "learning_rate": 1.896e-05,
+      "loss": 6.8581,
       "step": 322
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.914e-05,
-      "loss": 6.8608,
+      "epoch": 0.27,
+      "learning_rate": 1.902e-05,
+      "loss": 6.8487,
       "step": 323
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 6.7743,
+      "epoch": 0.27,
+      "learning_rate": 1.908e-05,
+      "loss": 6.9216,
       "step": 324
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.9260000000000002e-05,
-      "loss": 6.8434,
+      "epoch": 0.27,
+      "learning_rate": 1.914e-05,
+      "loss": 6.7967,
       "step": 325
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.932e-05,
-      "loss": 6.8534,
+      "epoch": 0.27,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 6.9835,
       "step": 326
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.938e-05,
-      "loss": 6.8232,
+      "epoch": 0.27,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 6.9141,
       "step": 327
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.944e-05,
-      "loss": 6.7599,
+      "epoch": 0.28,
+      "learning_rate": 1.932e-05,
+      "loss": 6.9604,
       "step": 328
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.95e-05,
-      "loss": 6.8421,
+      "epoch": 0.28,
+      "learning_rate": 1.938e-05,
+      "loss": 6.8948,
       "step": 329
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.9560000000000002e-05,
-      "loss": 6.8585,
+      "epoch": 0.28,
+      "learning_rate": 1.944e-05,
+      "loss": 7.0328,
       "step": 330
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.9620000000000002e-05,
-      "loss": 6.8067,
+      "epoch": 0.28,
+      "learning_rate": 1.95e-05,
+      "loss": 6.9079,
       "step": 331
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.968e-05,
-      "loss": 6.7269,
+      "epoch": 0.28,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 7.0494,
       "step": 332
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.974e-05,
-      "loss": 6.8948,
+      "epoch": 0.28,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 7.0634,
       "step": 333
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.98e-05,
-      "loss": 6.7869,
+      "epoch": 0.28,
+      "learning_rate": 1.968e-05,
+      "loss": 6.8763,
       "step": 334
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.9860000000000003e-05,
-      "loss": 6.8859,
+      "epoch": 0.28,
+      "learning_rate": 1.974e-05,
+      "loss": 6.9562,
       "step": 335
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.9920000000000002e-05,
-      "loss": 6.7629,
+      "epoch": 0.28,
+      "learning_rate": 1.98e-05,
+      "loss": 7.0171,
       "step": 336
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.9980000000000002e-05,
-      "loss": 6.6553,
+      "epoch": 0.28,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 6.8003,
       "step": 337
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.004e-05,
-      "loss": 6.7713,
+      "epoch": 0.28,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 7.075,
       "step": 338
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.01e-05,
-      "loss": 6.7695,
+      "epoch": 0.28,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 6.8714,
       "step": 339
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.016e-05,
-      "loss": 6.753,
+      "epoch": 0.29,
+      "learning_rate": 2.004e-05,
+      "loss": 6.9402,
       "step": 340
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.0220000000000003e-05,
-      "loss": 6.6573,
+      "epoch": 0.29,
+      "learning_rate": 2.01e-05,
+      "loss": 6.8398,
       "step": 341
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.0280000000000002e-05,
-      "loss": 6.7765,
+      "epoch": 0.29,
+      "learning_rate": 2.016e-05,
+      "loss": 7.0171,
       "step": 342
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.0340000000000002e-05,
-      "loss": 6.5781,
+      "epoch": 0.29,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 6.779,
       "step": 343
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.04e-05,
-      "loss": 6.6206,
+      "epoch": 0.29,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 6.9852,
       "step": 344
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.046e-05,
-      "loss": 6.5597,
+      "epoch": 0.29,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 6.7591,
       "step": 345
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.0520000000000003e-05,
-      "loss": 6.6449,
+      "epoch": 0.29,
+      "learning_rate": 2.04e-05,
+      "loss": 6.7046,
       "step": 346
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.0580000000000003e-05,
-      "loss": 6.4841,
+      "epoch": 0.29,
+      "learning_rate": 2.046e-05,
+      "loss": 6.8935,
       "step": 347
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.064e-05,
-      "loss": 6.3366,
+      "epoch": 0.29,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 6.9409,
       "step": 348
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.07e-05,
-      "loss": 6.4165,
+      "epoch": 0.29,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 6.5184,
       "step": 349
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.0759999999999998e-05,
-      "loss": 6.1109,
+      "epoch": 0.29,
+      "learning_rate": 2.064e-05,
+      "loss": 6.6459,
       "step": 350
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.082e-05,
-      "loss": 7.1829,
+      "epoch": 0.3,
+      "learning_rate": 2.07e-05,
+      "loss": 7.2592,
       "step": 351
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.088e-05,
-      "loss": 7.0255,
+      "epoch": 0.3,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 7.1277,
       "step": 352
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.094e-05,
-      "loss": 6.7856,
+      "epoch": 0.3,
+      "learning_rate": 2.082e-05,
+      "loss": 7.0065,
       "step": 353
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1e-05,
-      "loss": 6.6913,
+      "epoch": 0.3,
+      "learning_rate": 2.088e-05,
+      "loss": 7.0755,
       "step": 354
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.1059999999999998e-05,
-      "loss": 6.7244,
+      "epoch": 0.3,
+      "learning_rate": 2.094e-05,
+      "loss": 6.9331,
       "step": 355
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.1119999999999998e-05,
-      "loss": 6.8081,
+      "epoch": 0.3,
+      "learning_rate": 2.1e-05,
+      "loss": 7.0235,
       "step": 356
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.118e-05,
-      "loss": 6.6442,
+      "epoch": 0.3,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 7.0566,
       "step": 357
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.124e-05,
-      "loss": 6.6805,
+      "epoch": 0.3,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 6.8507,
       "step": 358
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.13e-05,
-      "loss": 6.5563,
+      "epoch": 0.3,
+      "learning_rate": 2.118e-05,
+      "loss": 6.9441,
       "step": 359
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.136e-05,
-      "loss": 6.5178,
+      "epoch": 0.3,
+      "learning_rate": 2.124e-05,
+      "loss": 7.0454,
       "step": 360
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.1419999999999998e-05,
-      "loss": 6.5198,
+      "epoch": 0.3,
+      "learning_rate": 2.13e-05,
+      "loss": 6.8663,
       "step": 361
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.148e-05,
-      "loss": 6.4337,
+      "epoch": 0.3,
+      "learning_rate": 2.136e-05,
+      "loss": 6.8307,
       "step": 362
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.154e-05,
-      "loss": 6.3207,
+      "epoch": 0.31,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 7.0394,
       "step": 363
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.16e-05,
-      "loss": 6.3906,
+      "epoch": 0.31,
+      "learning_rate": 2.148e-05,
+      "loss": 6.9532,
       "step": 364
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.166e-05,
-      "loss": 6.4776,
+      "epoch": 0.31,
+      "learning_rate": 2.154e-05,
+      "loss": 6.8977,
       "step": 365
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.172e-05,
-      "loss": 6.3665,
+      "epoch": 0.31,
+      "learning_rate": 2.16e-05,
+      "loss": 6.995,
       "step": 366
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.178e-05,
-      "loss": 6.4642,
+      "epoch": 0.31,
+      "learning_rate": 2.166e-05,
+      "loss": 6.963,
       "step": 367
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.184e-05,
-      "loss": 6.3793,
+      "epoch": 0.31,
+      "learning_rate": 2.172e-05,
+      "loss": 7.0942,
       "step": 368
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.19e-05,
-      "loss": 6.2186,
+      "epoch": 0.31,
+      "learning_rate": 2.178e-05,
+      "loss": 6.9486,
       "step": 369
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.196e-05,
-      "loss": 6.2923,
+      "epoch": 0.31,
+      "learning_rate": 2.184e-05,
+      "loss": 7.0199,
       "step": 370
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.202e-05,
-      "loss": 6.2903,
+      "epoch": 0.31,
+      "learning_rate": 2.19e-05,
+      "loss": 6.8288,
       "step": 371
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.208e-05,
-      "loss": 6.2592,
+      "epoch": 0.31,
+      "learning_rate": 2.196e-05,
+      "loss": 6.8878,
       "step": 372
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.214e-05,
-      "loss": 6.173,
+      "epoch": 0.31,
+      "learning_rate": 2.202e-05,
+      "loss": 6.7846,
       "step": 373
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.22e-05,
-      "loss": 6.0022,
+      "epoch": 0.31,
+      "learning_rate": 2.208e-05,
+      "loss": 6.8995,
       "step": 374
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.226e-05,
-      "loss": 6.1701,
+      "epoch": 0.32,
+      "learning_rate": 2.214e-05,
+      "loss": 6.8338,
       "step": 375
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.232e-05,
-      "loss": 6.0781,
+      "epoch": 0.32,
+      "learning_rate": 2.22e-05,
+      "loss": 7.0227,
       "step": 376
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.238e-05,
-      "loss": 6.1111,
+      "epoch": 0.32,
+      "learning_rate": 2.226e-05,
+      "loss": 6.859,
       "step": 377
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.2440000000000002e-05,
-      "loss": 6.0296,
+      "epoch": 0.32,
+      "learning_rate": 2.232e-05,
+      "loss": 6.9013,
       "step": 378
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.25e-05,
-      "loss": 6.0029,
+      "epoch": 0.32,
+      "learning_rate": 2.238e-05,
+      "loss": 6.929,
       "step": 379
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.256e-05,
-      "loss": 6.0384,
+      "epoch": 0.32,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 6.8989,
       "step": 380
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.262e-05,
-      "loss": 6.1197,
+      "epoch": 0.32,
+      "learning_rate": 2.25e-05,
+      "loss": 6.8926,
       "step": 381
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.268e-05,
-      "loss": 6.0461,
+      "epoch": 0.32,
+      "learning_rate": 2.256e-05,
+      "loss": 6.8915,
       "step": 382
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.274e-05,
-      "loss": 5.9918,
+      "epoch": 0.32,
+      "learning_rate": 2.262e-05,
+      "loss": 6.7224,
       "step": 383
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 5.9277,
+      "epoch": 0.32,
+      "learning_rate": 2.268e-05,
+      "loss": 7.0929,
       "step": 384
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.286e-05,
-      "loss": 5.9975,
+      "epoch": 0.32,
+      "learning_rate": 2.274e-05,
+      "loss": 6.8961,
       "step": 385
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.292e-05,
-      "loss": 5.9381,
+      "epoch": 0.32,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 6.7993,
       "step": 386
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.298e-05,
-      "loss": 5.9793,
+      "epoch": 0.33,
+      "learning_rate": 2.286e-05,
+      "loss": 6.8447,
       "step": 387
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.304e-05,
-      "loss": 5.9764,
+      "epoch": 0.33,
+      "learning_rate": 2.292e-05,
+      "loss": 6.914,
       "step": 388
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.3100000000000002e-05,
-      "loss": 5.955,
+      "epoch": 0.33,
+      "learning_rate": 2.298e-05,
+      "loss": 7.0236,
       "step": 389
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.3160000000000002e-05,
-      "loss": 5.8213,
+      "epoch": 0.33,
+      "learning_rate": 2.304e-05,
+      "loss": 7.1832,
       "step": 390
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.322e-05,
-      "loss": 5.9625,
+      "epoch": 0.33,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 7.0389,
       "step": 391
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.328e-05,
-      "loss": 5.951,
+      "epoch": 0.33,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 6.8648,
       "step": 392
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.334e-05,
-      "loss": 5.7552,
+      "epoch": 0.33,
+      "learning_rate": 2.322e-05,
+      "loss": 6.7297,
       "step": 393
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 5.8177,
+      "epoch": 0.33,
+      "learning_rate": 2.328e-05,
+      "loss": 6.858,
       "step": 394
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.3460000000000002e-05,
-      "loss": 5.8311,
+      "epoch": 0.33,
+      "learning_rate": 2.334e-05,
+      "loss": 6.677,
       "step": 395
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.3520000000000002e-05,
-      "loss": 5.8945,
+      "epoch": 0.33,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 6.7588,
       "step": 396
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.358e-05,
-      "loss": 5.8766,
+      "epoch": 0.33,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 6.4795,
       "step": 397
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.364e-05,
-      "loss": 5.8002,
+      "epoch": 0.33,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 6.5628,
       "step": 398
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.37e-05,
-      "loss": 5.9109,
+      "epoch": 0.34,
+      "learning_rate": 2.358e-05,
+      "loss": 6.4478,
       "step": 399
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.3760000000000003e-05,
-      "loss": 5.6506,
+      "epoch": 0.34,
+      "learning_rate": 2.364e-05,
+      "loss": 6.2483,
       "step": 400
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.3820000000000002e-05,
-      "loss": 6.0025,
+      "epoch": 0.34,
+      "learning_rate": 2.37e-05,
+      "loss": 7.0719,
       "step": 401
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.3880000000000002e-05,
-      "loss": 5.8234,
+      "epoch": 0.34,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 7.0878,
       "step": 402
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.394e-05,
-      "loss": 5.9229,
+      "epoch": 0.34,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 7.1169,
       "step": 403
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4e-05,
-      "loss": 5.7209,
+      "epoch": 0.34,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 7.1507,
       "step": 404
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4060000000000003e-05,
-      "loss": 5.6651,
+      "epoch": 0.34,
+      "learning_rate": 2.394e-05,
+      "loss": 6.8746,
       "step": 405
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4120000000000003e-05,
-      "loss": 5.824,
+      "epoch": 0.34,
+      "learning_rate": 2.4e-05,
+      "loss": 6.9829,
       "step": 406
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4180000000000002e-05,
-      "loss": 5.7593,
+      "epoch": 0.34,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 6.9614,
       "step": 407
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.4240000000000002e-05,
-      "loss": 5.6796,
+      "epoch": 0.34,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 6.9648,
       "step": 408
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.43e-05,
-      "loss": 5.6031,
+      "epoch": 0.34,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 6.8028,
       "step": 409
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.4360000000000004e-05,
-      "loss": 5.6286,
+      "epoch": 0.34,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 7.0344,
       "step": 410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.442e-05,
-      "loss": 5.7918,
+      "epoch": 0.35,
+      "learning_rate": 2.43e-05,
+      "loss": 6.9742,
       "step": 411
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.448e-05,
-      "loss": 5.6609,
+      "epoch": 0.35,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 6.9888,
       "step": 412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.454e-05,
-      "loss": 5.6679,
+      "epoch": 0.35,
+      "learning_rate": 2.442e-05,
+      "loss": 6.8872,
       "step": 413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 5.7014,
+      "epoch": 0.35,
+      "learning_rate": 2.448e-05,
+      "loss": 6.8158,
       "step": 414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.4659999999999998e-05,
-      "loss": 5.7206,
+      "epoch": 0.35,
+      "learning_rate": 2.454e-05,
+      "loss": 6.8687,
       "step": 415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.472e-05,
-      "loss": 5.6453,
+      "epoch": 0.35,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 7.0971,
       "step": 416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.478e-05,
-      "loss": 5.6255,
+      "epoch": 0.35,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 6.8733,
       "step": 417
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.484e-05,
-      "loss": 5.5376,
+      "epoch": 0.35,
+      "learning_rate": 2.472e-05,
+      "loss": 6.7277,
       "step": 418
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.49e-05,
-      "loss": 5.5823,
+      "epoch": 0.35,
+      "learning_rate": 2.478e-05,
+      "loss": 6.8349,
       "step": 419
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.4959999999999998e-05,
-      "loss": 5.5957,
+      "epoch": 0.35,
+      "learning_rate": 2.484e-05,
+      "loss": 7.0487,
       "step": 420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.502e-05,
-      "loss": 5.6762,
+      "epoch": 0.35,
+      "learning_rate": 2.49e-05,
+      "loss": 6.9402,
       "step": 421
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.508e-05,
-      "loss": 5.4908,
+      "epoch": 0.35,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 6.8249,
       "step": 422
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.514e-05,
-      "loss": 5.5229,
+      "epoch": 0.36,
+      "learning_rate": 2.502e-05,
+      "loss": 6.9416,
       "step": 423
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.52e-05,
-      "loss": 5.6172,
+      "epoch": 0.36,
+      "learning_rate": 2.508e-05,
+      "loss": 6.8635,
       "step": 424
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.526e-05,
-      "loss": 5.4369,
+      "epoch": 0.36,
+      "learning_rate": 2.514e-05,
+      "loss": 7.0993,
       "step": 425
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.5319999999999998e-05,
-      "loss": 5.5638,
+      "epoch": 0.36,
+      "learning_rate": 2.52e-05,
+      "loss": 6.9045,
       "step": 426
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.538e-05,
-      "loss": 5.4714,
+      "epoch": 0.36,
+      "learning_rate": 2.526e-05,
+      "loss": 6.8628,
       "step": 427
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.544e-05,
-      "loss": 5.441,
+      "epoch": 0.36,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 6.94,
       "step": 428
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.55e-05,
-      "loss": 5.3765,
+      "epoch": 0.36,
+      "learning_rate": 2.538e-05,
+      "loss": 7.0008,
       "step": 429
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.556e-05,
-      "loss": 5.4495,
+      "epoch": 0.36,
+      "learning_rate": 2.544e-05,
+      "loss": 6.928,
       "step": 430
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.562e-05,
-      "loss": 5.4951,
+      "epoch": 0.36,
+      "learning_rate": 2.55e-05,
+      "loss": 7.0018,
       "step": 431
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.568e-05,
-      "loss": 5.6691,
+      "epoch": 0.36,
+      "learning_rate": 2.556e-05,
+      "loss": 6.6915,
       "step": 432
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.574e-05,
-      "loss": 5.4317,
+      "epoch": 0.36,
+      "learning_rate": 2.562e-05,
+      "loss": 6.9509,
       "step": 433
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.58e-05,
-      "loss": 5.49,
+      "epoch": 0.36,
+      "learning_rate": 2.568e-05,
+      "loss": 6.894,
       "step": 434
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.586e-05,
-      "loss": 5.4139,
+      "epoch": 0.37,
+      "learning_rate": 2.574e-05,
+      "loss": 7.0125,
       "step": 435
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.592e-05,
-      "loss": 5.4088,
+      "epoch": 0.37,
+      "learning_rate": 2.58e-05,
+      "loss": 6.743,
       "step": 436
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.5980000000000002e-05,
-      "loss": 5.3708,
+      "epoch": 0.37,
+      "learning_rate": 2.586e-05,
+      "loss": 6.8109,
       "step": 437
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.604e-05,
-      "loss": 5.4885,
-      "step": 438
+      "epoch": 0.37,
+      "learning_rate": 2.592e-05,
+      "loss": 6.7761,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 6.9957,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.604e-05,
+      "loss": 6.8469,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.61e-05,
+      "loss": 6.9463,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.616e-05,
+      "loss": 6.7968,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.622e-05,
+      "loss": 6.8395,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.628e-05,
+      "loss": 6.7478,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 6.5671,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.64e-05,
+      "loss": 6.766,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.646e-05,
+      "loss": 6.6719,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.652e-05,
+      "loss": 6.8373,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.658e-05,
+      "loss": 6.5786,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 6.3753,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 6.9095,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.676e-05,
+      "loss": 6.8914,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.682e-05,
+      "loss": 6.9704,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.688e-05,
+      "loss": 6.9067,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 6.9692,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 6.9132,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 6.945,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.712e-05,
+      "loss": 7.0245,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.718e-05,
+      "loss": 7.0023,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.724e-05,
+      "loss": 6.981,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 6.8999,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 6.8976,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 6.9406,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.748e-05,
+      "loss": 6.9898,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.754e-05,
+      "loss": 6.9237,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 6.9051,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 6.7976,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 6.86,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.778e-05,
+      "loss": 6.6787,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.784e-05,
+      "loss": 7.101,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.79e-05,
+      "loss": 6.8838,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 7.0176,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 6.9806,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 7.0431,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 6.8727,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 6.7498,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.826e-05,
+      "loss": 6.6904,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.832e-05,
+      "loss": 6.8535,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.838e-05,
+      "loss": 6.842,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.844e-05,
+      "loss": 7.0386,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 6.8259,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.856e-05,
+      "loss": 6.8453,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.862e-05,
+      "loss": 6.8384,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.868e-05,
+      "loss": 6.9339,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.874e-05,
+      "loss": 6.9688,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.88e-05,
+      "loss": 6.7787,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 6.8239,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.892e-05,
+      "loss": 6.9513,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.898e-05,
+      "loss": 6.9371,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.904e-05,
+      "loss": 6.8265,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.91e-05,
+      "loss": 6.973,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.916e-05,
+      "loss": 6.9751,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.922e-05,
+      "loss": 7.0204,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.928e-05,
+      "loss": 6.9157,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.934e-05,
+      "loss": 6.8475,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.94e-05,
+      "loss": 6.7319,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.946e-05,
+      "loss": 6.7064,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 6.5884,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.958e-05,
+      "loss": 6.547,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.964e-05,
+      "loss": 6.5033,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 6.841328144073486,
+      "eval_runtime": 400.1835,
+      "eval_samples_per_second": 6.602,
+      "eval_steps_per_second": 0.552,
+      "eval_wer": 0.9372383276782348,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.97e-05,
+      "loss": 7.0164,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.976e-05,
+      "loss": 7.0229,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.982e-05,
+      "loss": 7.0215,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 6.9711,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.994e-05,
+      "loss": 6.971,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3e-05,
+      "loss": 7.077,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.995645863570392e-05,
+      "loss": 6.9226,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9912917271407838e-05,
+      "loss": 6.9394,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9869375907111758e-05,
+      "loss": 6.8403,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9825834542815675e-05,
+      "loss": 6.9181,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9782293178519592e-05,
+      "loss": 6.7456,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9738751814223512e-05,
+      "loss": 6.9487,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9695210449927432e-05,
+      "loss": 6.9014,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651669085631353e-05,
+      "loss": 6.8337,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.960812772133527e-05,
+      "loss": 6.9608,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.956458635703919e-05,
+      "loss": 6.7855,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9521044992743107e-05,
+      "loss": 6.9306,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9477503628447024e-05,
+      "loss": 6.8124,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9433962264150944e-05,
+      "loss": 7.0741,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9390420899854864e-05,
+      "loss": 6.8125,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.934687953555878e-05,
+      "loss": 6.9402,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.93033381712627e-05,
+      "loss": 6.9112,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.925979680696662e-05,
+      "loss": 6.9402,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9216255442670535e-05,
+      "loss": 6.8678,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9172714078374455e-05,
+      "loss": 6.8617,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9129172714078376e-05,
+      "loss": 6.7642,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9085631349782292e-05,
+      "loss": 6.9013,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9042089985486213e-05,
+      "loss": 6.9558,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8998548621190133e-05,
+      "loss": 6.8787,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8955007256894053e-05,
+      "loss": 6.9184,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8911465892597967e-05,
+      "loss": 6.8576,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8867924528301887e-05,
+      "loss": 6.9178,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8824383164005807e-05,
+      "loss": 6.7771,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8780841799709724e-05,
+      "loss": 6.813,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8737300435413644e-05,
+      "loss": 6.9827,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8693759071117565e-05,
+      "loss": 6.7773,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.865021770682148e-05,
+      "loss": 6.8303,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.86066763425254e-05,
+      "loss": 6.7075,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.856313497822932e-05,
+      "loss": 6.9048,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8519593613933236e-05,
+      "loss": 6.9648,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8476052249637156e-05,
+      "loss": 6.9927,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8432510885341076e-05,
+      "loss": 6.7953,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8388969521044996e-05,
+      "loss": 6.6269,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.834542815674891e-05,
+      "loss": 6.9967,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 6.8509,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.825834542815675e-05,
+      "loss": 6.7416,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8214804063860667e-05,
+      "loss": 6.4946,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8171262699564588e-05,
+      "loss": 6.6493,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8127721335268508e-05,
+      "loss": 6.5604,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8084179970972425e-05,
+      "loss": 6.447,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.804063860667634e-05,
+      "loss": 7.1064,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7997097242380262e-05,
+      "loss": 6.9909,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.795355587808418e-05,
+      "loss": 7.0095,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.79100145137881e-05,
+      "loss": 6.9143,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.786647314949202e-05,
+      "loss": 7.0488,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7822931785195936e-05,
+      "loss": 6.9259,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7779390420899856e-05,
+      "loss": 6.9317,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7735849056603773e-05,
+      "loss": 6.8489,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 6.9947,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.764876632801161e-05,
+      "loss": 6.8184,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.760522496371553e-05,
+      "loss": 7.1007,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.756168359941945e-05,
+      "loss": 6.7257,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7518142235123368e-05,
+      "loss": 6.9874,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7474600870827288e-05,
+      "loss": 6.6412,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7431059506531205e-05,
+      "loss": 6.8179,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7387518142235122e-05,
+      "loss": 6.8657,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7343976777939042e-05,
+      "loss": 6.7908,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7300435413642962e-05,
+      "loss": 6.8527,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.725689404934688e-05,
+      "loss": 6.8393,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.72133526850508e-05,
+      "loss": 6.8598,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.716981132075472e-05,
+      "loss": 6.932,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7126269956458637e-05,
+      "loss": 6.8856,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7082728592162554e-05,
+      "loss": 6.8599,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7039187227866474e-05,
+      "loss": 6.9485,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6995645863570394e-05,
+      "loss": 7.0858,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.695210449927431e-05,
+      "loss": 6.9847,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.690856313497823e-05,
+      "loss": 6.9401,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.686502177068215e-05,
+      "loss": 6.8819,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6821480406386065e-05,
+      "loss": 6.833,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6777939042089985e-05,
+      "loss": 7.0195,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6734397677793905e-05,
+      "loss": 7.0122,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6690856313497822e-05,
+      "loss": 6.7956,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6647314949201743e-05,
+      "loss": 6.8513,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6603773584905663e-05,
+      "loss": 6.7966,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656023222060958e-05,
+      "loss": 7.043,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6516690856313497e-05,
+      "loss": 6.8917,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6473149492017417e-05,
+      "loss": 6.9744,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6429608127721337e-05,
+      "loss": 6.8343,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6386066763425254e-05,
+      "loss": 6.7409,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6342525399129174e-05,
+      "loss": 6.8391,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6298984034833095e-05,
+      "loss": 6.9164,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.625544267053701e-05,
+      "loss": 6.7872,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.621190130624093e-05,
+      "loss": 6.9429,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.616835994194485e-05,
+      "loss": 6.7511,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6124818577648765e-05,
+      "loss": 6.6429,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6081277213352686e-05,
+      "loss": 6.6894,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6037735849056606e-05,
+      "loss": 6.586,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5994194484760523e-05,
+      "loss": 6.6416,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5950653120464443e-05,
+      "loss": 6.47,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.590711175616836e-05,
+      "loss": 6.3261,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5863570391872277e-05,
+      "loss": 6.8392,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5820029027576197e-05,
+      "loss": 7.0316,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5776487663280117e-05,
+      "loss": 7.05,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5732946298984038e-05,
+      "loss": 6.9018,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5689404934687955e-05,
+      "loss": 7.0057,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.564586357039187e-05,
+      "loss": 7.0843,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5602322206095792e-05,
+      "loss": 6.7689,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.555878084179971e-05,
+      "loss": 6.9561,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.551523947750363e-05,
+      "loss": 6.8926,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 6.932,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5428156748911466e-05,
+      "loss": 6.9758,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 6.8192,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5341074020319303e-05,
+      "loss": 6.8481,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.529753265602322e-05,
+      "loss": 6.8006,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.525399129172714e-05,
+      "loss": 6.9059,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.521044992743106e-05,
+      "loss": 6.7516,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.516690856313498e-05,
+      "loss": 6.8489,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5123367198838898e-05,
+      "loss": 6.8529,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5079825834542818e-05,
+      "loss": 6.902,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5036284470246735e-05,
+      "loss": 7.0009,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4992743105950652e-05,
+      "loss": 6.6571,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4949201741654572e-05,
+      "loss": 6.7878,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4905660377358492e-05,
+      "loss": 6.8093,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.486211901306241e-05,
+      "loss": 6.8768,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.481857764876633e-05,
+      "loss": 6.9355,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.477503628447025e-05,
+      "loss": 6.9823,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4731494920174163e-05,
+      "loss": 6.9445,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4687953555878083e-05,
+      "loss": 6.6985,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4644412191582004e-05,
+      "loss": 6.7972,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.460087082728592e-05,
+      "loss": 6.787,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.455732946298984e-05,
+      "loss": 6.8726,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.451378809869376e-05,
+      "loss": 6.9417,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.447024673439768e-05,
+      "loss": 6.8121,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4426705370101595e-05,
+      "loss": 6.9578,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4383164005805515e-05,
+      "loss": 6.8855,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4339622641509435e-05,
+      "loss": 6.9087,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4296081277213352e-05,
+      "loss": 6.7884,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4252539912917273e-05,
+      "loss": 6.7684,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4208998548621193e-05,
+      "loss": 6.8416,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.416545718432511e-05,
+      "loss": 7.0718,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4121915820029027e-05,
+      "loss": 6.7635,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4078374455732947e-05,
+      "loss": 6.7437,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4034833091436864e-05,
+      "loss": 6.9673,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3991291727140784e-05,
+      "loss": 6.7992,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3947750362844704e-05,
+      "loss": 6.7811,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3904208998548625e-05,
+      "loss": 6.7006,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.386066763425254e-05,
+      "loss": 6.2627,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3817126269956458e-05,
+      "loss": 6.7778,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.377358490566038e-05,
+      "loss": 6.7382,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3730043541364295e-05,
+      "loss": 6.2313,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3686502177068216e-05,
+      "loss": 7.1546,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3642960812772136e-05,
+      "loss": 6.9485,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3599419448476053e-05,
+      "loss": 7.0624,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3555878084179973e-05,
+      "loss": 7.0845,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.351233671988389e-05,
+      "loss": 7.0455,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3468795355587807e-05,
+      "loss": 6.9887,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3425253991291727e-05,
+      "loss": 6.8703,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3381712626995647e-05,
+      "loss": 6.9818,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3338171262699564e-05,
+      "loss": 6.9037,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3294629898403485e-05,
+      "loss": 6.8921,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3251088534107405e-05,
+      "loss": 6.809,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.320754716981132e-05,
+      "loss": 6.9329,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.316400580551524e-05,
+      "loss": 6.7156,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.312046444121916e-05,
+      "loss": 6.9578,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 6.9316,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3033381712626996e-05,
+      "loss": 6.7878,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2989840348330916e-05,
+      "loss": 6.9459,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2946298984034833e-05,
+      "loss": 6.8437,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.290275761973875e-05,
+      "loss": 7.1687,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.285921625544267e-05,
+      "loss": 6.9882,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.281567489114659e-05,
+      "loss": 6.8498,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2772133526850507e-05,
+      "loss": 6.8296,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2728592162554428e-05,
+      "loss": 6.8387,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2685050798258348e-05,
+      "loss": 6.8033,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 6.9487,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.259796806966618e-05,
+      "loss": 6.8176,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2554426705370102e-05,
+      "loss": 7.0318,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2510885341074022e-05,
+      "loss": 7.0025,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.246734397677794e-05,
+      "loss": 6.868,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.242380261248186e-05,
+      "loss": 7.0214,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.238026124818578e-05,
+      "loss": 6.6984,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2336719883889693e-05,
+      "loss": 6.7508,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2293178519593613e-05,
+      "loss": 6.8234,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2249637155297534e-05,
+      "loss": 6.7897,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.220609579100145e-05,
+      "loss": 6.8332,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.216255442670537e-05,
+      "loss": 6.8198,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.211901306240929e-05,
+      "loss": 6.9087,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2075471698113208e-05,
+      "loss": 6.854,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2031930333817125e-05,
+      "loss": 7.0296,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1988388969521045e-05,
+      "loss": 6.8097,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1944847605224965e-05,
+      "loss": 7.042,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1901306240928882e-05,
+      "loss": 6.756,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1857764876632802e-05,
+      "loss": 6.7333,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1814223512336723e-05,
+      "loss": 6.8924,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.177068214804064e-05,
+      "loss": 6.7768,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1727140783744556e-05,
+      "loss": 6.5173,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1683599419448477e-05,
+      "loss": 6.868,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1640058055152394e-05,
+      "loss": 6.8194,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1596516690856314e-05,
+      "loss": 6.4428,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1552975326560234e-05,
+      "loss": 6.6834,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.150943396226415e-05,
+      "loss": 7.0682,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.146589259796807e-05,
+      "loss": 6.8321,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1422351233671988e-05,
+      "loss": 7.0101,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.137880986937591e-05,
+      "loss": 6.8518,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1335268505079825e-05,
+      "loss": 6.9182,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1291727140783746e-05,
+      "loss": 6.9742,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1248185776487666e-05,
+      "loss": 6.9318,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1204644412191583e-05,
+      "loss": 6.8179,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1161103047895503e-05,
+      "loss": 6.9117,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.111756168359942e-05,
+      "loss": 6.978,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1074020319303337e-05,
+      "loss": 6.7896,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1030478955007257e-05,
+      "loss": 7.0812,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0986937590711177e-05,
+      "loss": 6.9035,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0943396226415094e-05,
+      "loss": 6.9047,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0899854862119014e-05,
+      "loss": 6.8181,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0856313497822935e-05,
+      "loss": 6.8149,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0812772133526848e-05,
+      "loss": 6.8505,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.076923076923077e-05,
+      "loss": 6.9516,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.072568940493469e-05,
+      "loss": 6.9214,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.068214804063861e-05,
+      "loss": 6.8467,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0638606676342526e-05,
+      "loss": 6.7056,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0595065312046446e-05,
+      "loss": 6.7862,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0551523947750363e-05,
+      "loss": 6.7903,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.050798258345428e-05,
+      "loss": 6.7591,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.04644412191582e-05,
+      "loss": 6.8493,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.042089985486212e-05,
+      "loss": 6.8346,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0377358490566037e-05,
+      "loss": 6.7957,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0333817126269958e-05,
+      "loss": 6.9421,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0290275761973878e-05,
+      "loss": 6.8396,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.024673439767779e-05,
+      "loss": 6.8857,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.020319303338171e-05,
+      "loss": 6.8232,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0159651669085632e-05,
+      "loss": 6.8886,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.011611030478955e-05,
+      "loss": 6.9045,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.007256894049347e-05,
+      "loss": 6.7822,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.002902757619739e-05,
+      "loss": 6.8733,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.998548621190131e-05,
+      "loss": 6.8417,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9941944847605223e-05,
+      "loss": 7.0115,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9898403483309143e-05,
+      "loss": 6.9775,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9854862119013064e-05,
+      "loss": 6.9662,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.981132075471698e-05,
+      "loss": 6.8292,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97677793904209e-05,
+      "loss": 6.9662,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.972423802612482e-05,
+      "loss": 6.9709,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9680696661828738e-05,
+      "loss": 6.9464,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9637155297532655e-05,
+      "loss": 6.7093,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9593613933236575e-05,
+      "loss": 6.7003,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9550072568940492e-05,
+      "loss": 6.6034,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9506531204644412e-05,
+      "loss": 6.7655,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9462989840348332e-05,
+      "loss": 6.9473,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9419448476052253e-05,
+      "loss": 6.5144,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.937590711175617e-05,
+      "loss": 6.1647,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9332365747460086e-05,
+      "loss": 6.9554,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9288824383164007e-05,
+      "loss": 7.0884,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9245283018867924e-05,
+      "loss": 6.9059,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9201741654571844e-05,
+      "loss": 6.8753,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9158200290275764e-05,
+      "loss": 7.0667,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.911465892597968e-05,
+      "loss": 6.9856,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.90711175616836e-05,
+      "loss": 6.9196,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9027576197387518e-05,
+      "loss": 7.0209,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8984034833091435e-05,
+      "loss": 6.7748,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8940493468795355e-05,
+      "loss": 6.9459,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8896952104499276e-05,
+      "loss": 6.8833,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8853410740203192e-05,
+      "loss": 6.7457,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8809869375907113e-05,
+      "loss": 6.8224,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8766328011611033e-05,
+      "loss": 6.7349,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.872278664731495e-05,
+      "loss": 7.1265,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8679245283018867e-05,
+      "loss": 6.8845,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8635703918722787e-05,
+      "loss": 6.8941,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8592162554426707e-05,
+      "loss": 6.7019,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8548621190130624e-05,
+      "loss": 6.8317,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8505079825834544e-05,
+      "loss": 6.9009,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 6.8667,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8417997097242378e-05,
+      "loss": 7.018,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.83744557329463e-05,
+      "loss": 6.8595,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.833091436865022e-05,
+      "loss": 6.7181,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8287373004354136e-05,
+      "loss": 6.9806,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243831640058056e-05,
+      "loss": 6.8173,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200290275761976e-05,
+      "loss": 6.7969,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8156748911465896e-05,
+      "loss": 6.8389,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.811320754716981e-05,
+      "loss": 6.8899,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.806966618287373e-05,
+      "loss": 7.0669,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.802612481857765e-05,
+      "loss": 6.9312,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7982583454281567e-05,
+      "loss": 6.784,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7939042089985487e-05,
+      "loss": 6.6587,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7895500725689408e-05,
+      "loss": 6.9062,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7851959361393325e-05,
+      "loss": 6.9787,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.780841799709724e-05,
+      "loss": 6.6987,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7764876632801162e-05,
+      "loss": 6.9932,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.772133526850508e-05,
+      "loss": 6.7969,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7677793904209e-05,
+      "loss": 6.8405,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.763425253991292e-05,
+      "loss": 6.7624,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7590711175616836e-05,
+      "loss": 6.941,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7547169811320753e-05,
+      "loss": 6.7807,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7503628447024673e-05,
+      "loss": 6.7793,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7460087082728593e-05,
+      "loss": 6.6572,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 6.6306,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.737300435413643e-05,
+      "loss": 6.5386,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.732946298984035e-05,
+      "loss": 6.5509,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7285921625544268e-05,
+      "loss": 6.5294,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7242380261248185e-05,
+      "loss": 6.4724,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7198838896952105e-05,
+      "loss": 6.2469,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7155297532656022e-05,
+      "loss": 6.902,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7111756168359942e-05,
+      "loss": 6.8033,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7068214804063862e-05,
+      "loss": 6.8523,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.702467343976778e-05,
+      "loss": 6.8993,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 6.9154,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6937590711175616e-05,
+      "loss": 6.8057,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6894049346879537e-05,
+      "loss": 6.8261,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6850507982583453e-05,
+      "loss": 6.8919,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6806966618287374e-05,
+      "loss": 6.8706,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6763425253991294e-05,
+      "loss": 6.8791,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.671988388969521e-05,
+      "loss": 6.7978,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.667634252539913e-05,
+      "loss": 6.8467,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6632801161103048e-05,
+      "loss": 6.7455,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6589259796806965e-05,
+      "loss": 7.0637,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6545718432510885e-05,
+      "loss": 7.0001,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6502177068214805e-05,
+      "loss": 6.7307,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6458635703918722e-05,
+      "loss": 6.8449,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 6.6097,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6371552975326563e-05,
+      "loss": 6.8532,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6328011611030476e-05,
+      "loss": 6.8925,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6284470246734397e-05,
+      "loss": 6.8499,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6240928882438317e-05,
+      "loss": 6.7906,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6197387518142237e-05,
+      "loss": 6.8578,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 6.9541,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6110304789550074e-05,
+      "loss": 6.8053,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6066763425253995e-05,
+      "loss": 6.8078,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6023222060957908e-05,
+      "loss": 6.9305,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5979680696661828e-05,
+      "loss": 6.9147,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.593613933236575e-05,
+      "loss": 6.8646,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5892597968069665e-05,
+      "loss": 7.0318,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 6.8768,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5805515239477506e-05,
+      "loss": 6.6636,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5761973875181423e-05,
+      "loss": 6.7513,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.571843251088534e-05,
+      "loss": 6.9217,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.567489114658926e-05,
+      "loss": 6.8903,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.563134978229318e-05,
+      "loss": 7.0025,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5587808417997097e-05,
+      "loss": 6.926,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5544267053701017e-05,
+      "loss": 6.5696,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5500725689404938e-05,
+      "loss": 7.0113,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5457184325108855e-05,
+      "loss": 6.793,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.541364296081277e-05,
+      "loss": 6.9198,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.537010159651669e-05,
+      "loss": 6.761,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.532656023222061e-05,
+      "loss": 7.0276,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.528301886792453e-05,
+      "loss": 6.6037,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5239477503628449e-05,
+      "loss": 6.7909,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5195936139332364e-05,
+      "loss": 6.7064,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5152394775036285e-05,
+      "loss": 6.6922,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5108853410740205e-05,
+      "loss": 6.4188,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.506531204644412e-05,
+      "loss": 6.769,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.502177068214804e-05,
+      "loss": 6.6032,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.497822931785196e-05,
+      "loss": 7.065,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4934687953555879e-05,
+      "loss": 6.877,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4891146589259796e-05,
+      "loss": 6.9379,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4847605224963716e-05,
+      "loss": 6.9983,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4804063860667635e-05,
+      "loss": 6.8895,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4760522496371553e-05,
+      "loss": 6.9192,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 6.7546,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.467343976777939e-05,
+      "loss": 6.9329,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.462989840348331e-05,
+      "loss": 6.7429,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4586357039187228e-05,
+      "loss": 6.8161,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4542815674891146e-05,
+      "loss": 6.8705,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4499274310595066e-05,
+      "loss": 6.896,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4455732946298983e-05,
+      "loss": 6.6946,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4412191582002904e-05,
+      "loss": 6.8785,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4368650217706822e-05,
+      "loss": 6.91,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.432510885341074e-05,
+      "loss": 6.7578,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.428156748911466e-05,
+      "loss": 6.7131,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4238026124818578e-05,
+      "loss": 6.7858,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4194484760522498e-05,
+      "loss": 6.7311,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 6.6836,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4107402031930334e-05,
+      "loss": 6.7571,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4063860667634254e-05,
+      "loss": 6.6392,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.402031930333817e-05,
+      "loss": 6.9785,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.397677793904209e-05,
+      "loss": 6.9024,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.393323657474601e-05,
+      "loss": 6.9245,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3889695210449928e-05,
+      "loss": 6.7119,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 6.7674,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3802612481857765e-05,
+      "loss": 6.9421,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3759071117561684e-05,
+      "loss": 6.7645,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3715529753265602e-05,
+      "loss": 6.926,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3671988388969521e-05,
+      "loss": 6.7802,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.362844702467344e-05,
+      "loss": 6.8721,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.358490566037736e-05,
+      "loss": 6.9203,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3541364296081277e-05,
+      "loss": 6.9149,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3497822931785197e-05,
+      "loss": 6.7289,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3454281567489116e-05,
+      "loss": 6.853,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3410740203193032e-05,
+      "loss": 6.7934,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3367198838896953e-05,
+      "loss": 6.8075,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3323657474600871e-05,
+      "loss": 6.8848,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.328011611030479e-05,
+      "loss": 6.671,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3236574746008708e-05,
+      "loss": 6.9252,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3193033381712627e-05,
+      "loss": 6.8196,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3149492017416547e-05,
+      "loss": 6.6605,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3105950653120464e-05,
+      "loss": 6.6684,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3062409288824383e-05,
+      "loss": 6.7128,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 6.705,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2975326560232222e-05,
+      "loss": 6.5003,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2931785195936138e-05,
+      "loss": 6.28,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2888243831640059e-05,
+      "loss": 6.4944,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2844702467343977e-05,
+      "loss": 6.0975,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2801161103047896e-05,
+      "loss": 6.9777,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2757619738751814e-05,
+      "loss": 6.9377,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2714078374455733e-05,
+      "loss": 6.9563,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2670537010159652e-05,
+      "loss": 6.7573,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.262699564586357e-05,
+      "loss": 6.9223,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.258345428156749e-05,
+      "loss": 6.9422,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2539912917271409e-05,
+      "loss": 6.806,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2496371552975326e-05,
+      "loss": 7.1837,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 6.8731,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2409288824383165e-05,
+      "loss": 6.8866,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2365747460087082e-05,
+      "loss": 6.8061,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2322206095791002e-05,
+      "loss": 6.8514,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.227866473149492e-05,
+      "loss": 6.8401,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.223512336719884e-05,
+      "loss": 6.7048,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2191582002902758e-05,
+      "loss": 6.826,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2148040638606676e-05,
+      "loss": 6.8965,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2104499274310596e-05,
+      "loss": 6.8272,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2060957910014513e-05,
+      "loss": 6.8323,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2017416545718432e-05,
+      "loss": 6.9279,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1973875181422352e-05,
+      "loss": 6.8667,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.193033381712627e-05,
+      "loss": 6.773,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 6.9603,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1843251088534108e-05,
+      "loss": 6.8699,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1799709724238026e-05,
+      "loss": 6.8946,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1756168359941945e-05,
+      "loss": 6.7076,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712626995645864e-05,
+      "loss": 6.8156,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1669085631349782e-05,
+      "loss": 6.9208,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1625544267053702e-05,
+      "loss": 6.8581,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.158200290275762e-05,
+      "loss": 6.7359,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 6.8063,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1494920174165458e-05,
+      "loss": 6.8592,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1451378809869375e-05,
+      "loss": 6.8561,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1407837445573295e-05,
+      "loss": 6.9459,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1364296081277214e-05,
+      "loss": 6.8407,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 6.7902,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1277213352685051e-05,
+      "loss": 6.8517,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.123367198838897e-05,
+      "loss": 6.7892,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119013062409289e-05,
+      "loss": 6.9927,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1146589259796807e-05,
+      "loss": 6.8108,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1103047895500725e-05,
+      "loss": 6.737,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1059506531204646e-05,
+      "loss": 6.7429,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1015965166908562e-05,
+      "loss": 6.8897,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0972423802612483e-05,
+      "loss": 6.5433,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0928882438316401e-05,
+      "loss": 6.5135,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088534107402032e-05,
+      "loss": 6.8652,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0841799709724238e-05,
+      "loss": 6.6219,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0798258345428157e-05,
+      "loss": 6.4837,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 6.3776,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0711175616835994e-05,
+      "loss": 6.4048,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0667634252539913e-05,
+      "loss": 6.2509,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0624092888243833e-05,
+      "loss": 6.8918,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0580551523947752e-05,
+      "loss": 6.9234,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0537010159651668e-05,
+      "loss": 6.8293,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0493468795355589e-05,
+      "loss": 6.669,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0449927431059507e-05,
+      "loss": 6.8053,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0406386066763424e-05,
+      "loss": 6.9721,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0362844702467344e-05,
+      "loss": 6.8532,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0319303338171263e-05,
+      "loss": 6.9768,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0275761973875182e-05,
+      "loss": 6.8408,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.02322206095791e-05,
+      "loss": 6.7216,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0188679245283019e-05,
+      "loss": 6.7617,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0145137880986939e-05,
+      "loss": 6.765,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0101596516690856e-05,
+      "loss": 6.7081,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0058055152394774e-05,
+      "loss": 6.8027,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0014513788098695e-05,
+      "loss": 6.8416,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.970972423802612e-06,
+      "loss": 6.8549,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.927431059506532e-06,
+      "loss": 7.026,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.88388969521045e-06,
+      "loss": 6.863,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.840348330914369e-06,
+      "loss": 6.5742,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.796806966618287e-06,
+      "loss": 6.8582,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.753265602322206e-06,
+      "loss": 6.7078,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.709724238026126e-06,
+      "loss": 6.6839,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.666182873730043e-06,
+      "loss": 6.7402,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.622641509433962e-06,
+      "loss": 6.8191,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 6.8586,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.5355587808418e-06,
+      "loss": 6.8754,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.492017416545717e-06,
+      "loss": 6.7543,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.448476052249638e-06,
+      "loss": 6.8026,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.404934687953556e-06,
+      "loss": 6.7462,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.361393323657475e-06,
+      "loss": 6.8254,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.317851959361393e-06,
+      "loss": 7.0929,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.274310595065312e-06,
+      "loss": 6.871,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 6.8445,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.18722786647315e-06,
+      "loss": 7.0254,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.143686502177068e-06,
+      "loss": 6.7171,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.100145137880988e-06,
+      "loss": 6.7767,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.056603773584905e-06,
+      "loss": 6.8884,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.013062409288825e-06,
+      "loss": 6.8049,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.969521044992744e-06,
+      "loss": 6.7851,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.925979680696662e-06,
+      "loss": 6.7375,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.882438316400581e-06,
+      "loss": 6.7886,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.8388969521045e-06,
+      "loss": 6.8043,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.795355587808418e-06,
+      "loss": 6.6738,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.751814223512337e-06,
+      "loss": 6.9183,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 6.7654,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.664731494920175e-06,
+      "loss": 7.083,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.621190130624092e-06,
+      "loss": 6.7406,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.577648766328011e-06,
+      "loss": 6.54,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.534107402031931e-06,
+      "loss": 6.5534,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 6.378,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 6.714595317840576,
+      "eval_runtime": 685.6597,
+      "eval_samples_per_second": 3.853,
+      "eval_steps_per_second": 0.322,
+      "eval_wer": 1.2430700240093657,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.447024673439768e-06,
+      "loss": 6.824,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.403483309143687e-06,
+      "loss": 6.8868,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.359941944847605e-06,
+      "loss": 6.8406,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.316400580551524e-06,
+      "loss": 6.8759,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.272859216255443e-06,
+      "loss": 6.8051,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.229317851959361e-06,
+      "loss": 6.8795,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.185776487663281e-06,
+      "loss": 6.7839,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.142235123367198e-06,
+      "loss": 6.6886,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.098693759071119e-06,
+      "loss": 6.935,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.055152394775037e-06,
+      "loss": 6.6507,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.011611030478954e-06,
+      "loss": 6.7456,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.968069666182874e-06,
+      "loss": 6.8132,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 6.6683,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.880986937590711e-06,
+      "loss": 6.7256,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.83744557329463e-06,
+      "loss": 6.689,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.793904208998549e-06,
+      "loss": 6.8579,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.750362844702469e-06,
+      "loss": 6.6898,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.706821480406386e-06,
+      "loss": 6.7691,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.663280116110304e-06,
+      "loss": 6.9135,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.6197387518142245e-06,
+      "loss": 6.6878,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.576197387518142e-06,
+      "loss": 6.8522,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.53265602322206e-06,
+      "loss": 6.9307,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.48911465892598e-06,
+      "loss": 6.9396,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.445573294629898e-06,
+      "loss": 6.7439,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.402031930333817e-06,
+      "loss": 6.7135,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 6.7239,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.314949201741655e-06,
+      "loss": 6.6989,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.271407837445573e-06,
+      "loss": 6.7525,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.227866473149492e-06,
+      "loss": 6.8011,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.184325108853411e-06,
+      "loss": 6.7347,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.14078374455733e-06,
+      "loss": 6.8327,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.097242380261249e-06,
+      "loss": 6.6466,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.053701015965167e-06,
+      "loss": 6.8424,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.010159651669085e-06,
+      "loss": 6.7346,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.966618287373005e-06,
+      "loss": 6.7685,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 6.8679,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.879535558780842e-06,
+      "loss": 6.6327,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.8359941944847605e-06,
+      "loss": 6.895,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.79245283018868e-06,
+      "loss": 6.7754,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7489114658925985e-06,
+      "loss": 6.6316,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.705370101596516e-06,
+      "loss": 6.6465,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.661828737300436e-06,
+      "loss": 6.7032,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.618287373004354e-06,
+      "loss": 6.6369,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.574746008708274e-06,
+      "loss": 6.7194,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.531204644412191e-06,
+      "loss": 6.7083,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.487663280116111e-06,
+      "loss": 6.6924,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.444121915820029e-06,
+      "loss": 6.4722,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.400580551523948e-06,
+      "loss": 6.6804,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.3570391872278665e-06,
+      "loss": 6.786,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.313497822931785e-06,
+      "loss": 6.3429,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2699564586357045e-06,
+      "loss": 6.9516,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 6.7765,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182873730043541e-06,
+      "loss": 6.7808,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.13933236574746e-06,
+      "loss": 6.8928,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095791001451379e-06,
+      "loss": 6.8988,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052249637155298e-06,
+      "loss": 6.8776,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008708272859216e-06,
+      "loss": 6.8533,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.965166908563135e-06,
+      "loss": 6.7207,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921625544267054e-06,
+      "loss": 6.876,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8780841799709725e-06,
+      "loss": 6.8643,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.834542815674891e-06,
+      "loss": 6.9022,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79100145137881e-06,
+      "loss": 6.7957,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.747460087082729e-06,
+      "loss": 6.8756,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.703918722786648e-06,
+      "loss": 6.7472,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 6.7388,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616835994194485e-06,
+      "loss": 6.7771,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573294629898403e-06,
+      "loss": 6.7268,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529753265602323e-06,
+      "loss": 6.8189,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.486211901306241e-06,
+      "loss": 6.9078,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44267053701016e-06,
+      "loss": 6.8513,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3991291727140785e-06,
+      "loss": 6.6771,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355587808417997e-06,
+      "loss": 6.821,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3120464441219165e-06,
+      "loss": 6.6158,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.268505079825834e-06,
+      "loss": 6.7374,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 6.4849,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181422351233672e-06,
+      "loss": 6.8353,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.137880986937591e-06,
+      "loss": 7.0018,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.094339622641509e-06,
+      "loss": 6.8767,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.050798258345428e-06,
+      "loss": 6.668,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.007256894049347e-06,
+      "loss": 6.7554,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.963715529753266e-06,
+      "loss": 6.7884,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9201741654571845e-06,
+      "loss": 7.0688,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.876632801161103e-06,
+      "loss": 6.598,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.833091436865022e-06,
+      "loss": 6.7801,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 6.7619,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.746008708272859e-06,
+      "loss": 6.8128,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702467343976778e-06,
+      "loss": 6.756,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658925979680697e-06,
+      "loss": 6.7914,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 6.7859,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.571843251088534e-06,
+      "loss": 6.6515,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5283018867924524e-06,
+      "loss": 6.7337,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.484760522496372e-06,
+      "loss": 6.6213,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4412191582002904e-06,
+      "loss": 6.619,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.397677793904209e-06,
+      "loss": 6.8495,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 6.5749,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.310595065312046e-06,
+      "loss": 6.7525,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2670537010159656e-06,
+      "loss": 6.5626,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.223512336719884e-06,
+      "loss": 6.7664,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.179970972423803e-06,
+      "loss": 6.8359,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.136429608127721e-06,
+      "loss": 5.9896,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.092888243831641e-06,
+      "loss": 6.8314,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.049346879535559e-06,
+      "loss": 6.8609,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.005805515239477e-06,
+      "loss": 6.8985,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 6.7312,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.918722786647315e-06,
+      "loss": 6.7393,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.875181422351234e-06,
+      "loss": 6.6446,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.831640058055152e-06,
+      "loss": 7.0349,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.788098693759071e-06,
+      "loss": 6.9525,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.74455732946299e-06,
+      "loss": 6.8194,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7010159651669087e-06,
+      "loss": 6.722,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6574746008708277e-06,
+      "loss": 6.7443,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.613933236574746e-06,
+      "loss": 6.8892,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.570391872278665e-06,
+      "loss": 6.6213,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5268505079825834e-06,
+      "loss": 6.6964,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4833091436865024e-06,
+      "loss": 6.9253,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439767779390421e-06,
+      "loss": 6.8583,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.39622641509434e-06,
+      "loss": 6.8969,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.352685050798258e-06,
+      "loss": 6.8028,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.309143686502177e-06,
+      "loss": 6.6186,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2656023222060957e-06,
+      "loss": 6.9412,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2220609579100147e-06,
+      "loss": 6.7968,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1785195936139333e-06,
+      "loss": 7.016,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1349782293178522e-06,
+      "loss": 6.7541,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0914368650217704e-06,
+      "loss": 6.9549,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0478955007256894e-06,
+      "loss": 6.9793,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.004354136429608e-06,
+      "loss": 6.7476,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.960812772133527e-06,
+      "loss": 6.818,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9172714078374455e-06,
+      "loss": 6.7245,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8737300435413645e-06,
+      "loss": 6.7706,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 6.7443,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7866473149492017e-06,
+      "loss": 6.6402,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7431059506531207e-06,
+      "loss": 6.6075,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6995645863570392e-06,
+      "loss": 6.9054,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6560232220609582e-06,
+      "loss": 6.7971,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 6.9146,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5689404934687954e-06,
+      "loss": 6.7406,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.525399129172714e-06,
+      "loss": 6.8299,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.481857764876633e-06,
+      "loss": 6.7729,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4383164005805515e-06,
+      "loss": 6.7405,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3947750362844705e-06,
+      "loss": 6.6303,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.351233671988389e-06,
+      "loss": 6.8663,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 6.7374,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2641509433962262e-06,
+      "loss": 6.7383,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2206095791001452e-06,
+      "loss": 6.7889,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 6.5223,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1335268505079828e-06,
+      "loss": 6.871,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0899854862119014e-06,
+      "loss": 6.5354,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0464441219158204e-06,
+      "loss": 6.3837,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0029027576197385e-06,
+      "loss": 6.6686,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9593613933236575e-06,
+      "loss": 6.4184,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.915820029027576e-06,
+      "loss": 6.8471,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.872278664731495e-06,
+      "loss": 6.7201,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8287373004354138e-06,
+      "loss": 6.9244,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7851959361393324e-06,
+      "loss": 6.7686,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7416545718432512e-06,
+      "loss": 6.8244,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.69811320754717e-06,
+      "loss": 6.9997,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6545718432510886e-06,
+      "loss": 6.7436,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6110304789550073e-06,
+      "loss": 6.7031,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5674891146589261e-06,
+      "loss": 6.7828,
+      "step": 1159
     },
     {
       "epoch": 0.98,
-      "learning_rate": 2.61e-05,
-      "loss": 5.4998,
-      "step": 439
+      "learning_rate": 1.5239477503628447e-06,
+      "loss": 6.8203,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4804063860667635e-06,
+      "loss": 6.9727,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4368650217706823e-06,
+      "loss": 6.8576,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3933236574746008e-06,
+      "loss": 6.934,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3497822931785196e-06,
+      "loss": 6.7721,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 6.7792,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.262699564586357e-06,
+      "loss": 6.9853,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2191582002902758e-06,
+      "loss": 6.6124,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1756168359941945e-06,
+      "loss": 6.7884,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1320754716981131e-06,
+      "loss": 6.7028,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.088534107402032e-06,
+      "loss": 6.6828,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0449927431059507e-06,
+      "loss": 6.7638,
+      "step": 1171
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.616e-05,
-      "loss": 5.5172,
-      "step": 440
+      "learning_rate": 1.0014513788098693e-06,
+      "loss": 6.66,
+      "step": 1172
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.622e-05,
-      "loss": 5.3714,
-      "step": 441
+      "learning_rate": 9.57910014513788e-07,
+      "loss": 6.8669,
+      "step": 1173
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.628e-05,
-      "loss": 5.6947,
-      "step": 442
+      "learning_rate": 9.143686502177069e-07,
+      "loss": 6.8665,
+      "step": 1174
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.6340000000000002e-05,
-      "loss": 5.3448,
-      "step": 443
+      "learning_rate": 8.708272859216256e-07,
+      "loss": 6.6172,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.272859216255443e-07,
+      "loss": 6.5918,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.837445573294631e-07,
+      "loss": 6.7592,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.402031930333817e-07,
+      "loss": 6.8729,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.966618287373004e-07,
+      "loss": 6.5352,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.531204644412192e-07,
+      "loss": 6.9544,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.095791001451379e-07,
+      "loss": 6.6757,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.660377358490566e-07,
+      "loss": 6.8157,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.224963715529753e-07,
+      "loss": 6.7616,
+      "step": 1183
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.64e-05,
-      "loss": 5.5714,
-      "step": 444
+      "learning_rate": 4.78955007256894e-07,
+      "loss": 6.681,
+      "step": 1184
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.646e-05,
-      "loss": 5.484,
-      "step": 445
+      "learning_rate": 4.354136429608128e-07,
+      "loss": 6.6513,
+      "step": 1185
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.652e-05,
-      "loss": 5.22,
-      "step": 446
+      "learning_rate": 3.9187227866473153e-07,
+      "loss": 6.5563,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.483309143686502e-07,
+      "loss": 6.5301,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0478955007256894e-07,
+      "loss": 6.3966,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6124818577648767e-07,
+      "loss": 6.4156,
+      "step": 1189
     },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 6.890713336756411,
-      "train_runtime": 5024.7099,
-      "train_samples_per_second": 5.68,
-      "train_steps_per_second": 0.089
+      "train_loss": 6.949796647760826,
+      "train_runtime": 6133.846,
+      "train_samples_per_second": 4.653,
+      "train_steps_per_second": 0.194
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,