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": 9.0516,
+      "loss": 9.3247,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.7241,
+      "loss": 8.6635,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 8.9776,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.0379,
       "step": 3
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 2.0000000000000002e-07,
-      "loss": 8.923,
+      "loss": 9.0547,
       "step": 4
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 8.8431,
+      "loss": 8.8855,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.000000000000001e-07,
-      "loss": 9.0187,
+      "loss": 8.8753,
       "step": 6
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 8.7166,
+      "loss": 8.5676,
       "step": 7
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 8.5221,
+      "loss": 8.4724,
       "step": 8
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 8.4006,
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 8.8081,
       "step": 9
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 8.4796,
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 8.5251,
       "step": 10
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 8.2313,
+      "loss": 8.355,
       "step": 11
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 1.6000000000000001e-06,
-      "loss": 8.2866,
+      "loss": 8.4144,
       "step": 12
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 1.8e-06,
-      "loss": 8.1051,
+      "loss": 8.3692,
       "step": 13
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 8.311,
+      "loss": 8.1513,
       "step": 14
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 2.2e-06,
-      "loss": 8.0337,
+      "loss": 8.1517,
       "step": 15
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 8.0619,
+      "loss": 7.9598,
       "step": 16
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 2.6e-06,
-      "loss": 7.9294,
+      "loss": 8.0229,
       "step": 17
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 2.8000000000000003e-06,
-      "loss": 7.9117,
+      "loss": 8.1036,
       "step": 18
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 3e-06,
-      "loss": 7.8145,
+      "loss": 7.8943,
       "step": 19
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 3.2000000000000003e-06,
-      "loss": 7.9357,
+      "loss": 7.8666,
       "step": 20
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 8.0185,
+      "loss": 7.6748,
       "step": 21
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 3.6e-06,
-      "loss": 7.8128,
+      "loss": 7.6982,
       "step": 22
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 3.8e-06,
-      "loss": 7.8253,
+      "loss": 7.7092,
       "step": 23
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 7.8046,
+      "loss": 7.7163,
       "step": 24
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.2000000000000004e-06,
-      "loss": 7.6217,
+      "loss": 7.5864,
       "step": 25
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.4e-06,
-      "loss": 7.6253,
+      "loss": 7.6307,
       "step": 26
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.6e-06,
-      "loss": 7.5612,
+      "loss": 7.6469,
       "step": 27
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 4.800000000000001e-06,
-      "loss": 7.4502,
+      "loss": 7.4145,
       "step": 28
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.02,
       "learning_rate": 5e-06,
-      "loss": 7.5826,
+      "loss": 7.5651,
       "step": 29
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.2e-06,
-      "loss": 7.4716,
+      "loss": 7.6513,
       "step": 30
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.4e-06,
-      "loss": 7.4904,
+      "loss": 7.5343,
       "step": 31
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.600000000000001e-06,
-      "loss": 7.4556,
+      "loss": 7.4476,
       "step": 32
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 5.8e-06,
-      "loss": 7.4077,
+      "loss": 7.6121,
       "step": 33
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6e-06,
-      "loss": 7.3024,
+      "loss": 7.2192,
       "step": 34
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6.2e-06,
-      "loss": 7.3594,
+      "loss": 7.3556,
       "step": 35
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 7.3429,
+      "loss": 7.5773,
       "step": 36
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 6.6e-06,
-      "loss": 7.4239,
+      "loss": 7.396,
       "step": 37
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 6.800000000000001e-06,
-      "loss": 7.3168,
+      "loss": 7.4057,
       "step": 38
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 7.3821,
+      "loss": 7.5256,
       "step": 39
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 7.2e-06,
-      "loss": 7.3206,
+      "loss": 7.3596,
       "step": 40
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 7.4e-06,
-      "loss": 7.3553,
+      "loss": 7.2581,
       "step": 41
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.04,
       "learning_rate": 7.6e-06,
-      "loss": 7.3862,
+      "loss": 7.4072,
       "step": 42
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 7.8e-06,
-      "loss": 7.1496,
+      "loss": 7.3126,
       "step": 43
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 7.3461,
+      "loss": 7.3703,
       "step": 44
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 8.200000000000001e-06,
-      "loss": 7.4049,
+      "loss": 7.4614,
       "step": 45
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 7.3935,
+      "loss": 7.5204,
       "step": 46
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.599999999999999e-06,
-      "loss": 7.318,
+      "loss": 7.4169,
       "step": 47
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 8.8e-06,
-      "loss": 7.3927,
+      "loss": 7.5653,
       "step": 48
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 9e-06,
-      "loss": 7.2045,
+      "loss": 7.4251,
       "step": 49
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 9.2e-06,
-      "loss": 7.0658,
+      "loss": 7.4275,
       "step": 50
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 9.4e-06,
-      "loss": 7.3025,
+      "loss": 7.4487,
       "step": 51
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 9.600000000000001e-06,
-      "loss": 7.2009,
+      "loss": 7.2411,
       "step": 52
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 9.800000000000001e-06,
-      "loss": 7.258,
+      "loss": 7.1513,
       "step": 53
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 1e-05,
-      "loss": 7.1723,
+      "loss": 7.2095,
       "step": 54
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 1.02e-05,
-      "loss": 7.0552,
+      "loss": 7.3188,
       "step": 55
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.04e-05,
-      "loss": 7.1678,
+      "loss": 7.012,
       "step": 56
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.06e-05,
-      "loss": 7.2158,
+      "loss": 7.4734,
       "step": 57
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.08e-05,
-      "loss": 7.2693,
+      "loss": 7.1145,
       "step": 58
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 7.1422,
+      "loss": 7.0793,
       "step": 59
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 1.1200000000000001e-05,
-      "loss": 7.1538,
+      "loss": 7.2494,
       "step": 60
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.1400000000000001e-05,
-      "loss": 7.1074,
+      "loss": 7.0728,
       "step": 61
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.16e-05,
-      "loss": 7.1487,
+      "loss": 7.1014,
       "step": 62
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.18e-05,
-      "loss": 7.0797,
+      "loss": 7.243,
       "step": 63
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 1.2e-05,
-      "loss": 7.1705,
+      "loss": 7.1645,
       "step": 64
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.05,
       "learning_rate": 1.22e-05,
-      "loss": 7.0818,
+      "loss": 7.1121,
       "step": 65
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.24e-05,
-      "loss": 7.0407,
+      "loss": 7.193,
       "step": 66
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.2600000000000001e-05,
-      "loss": 7.0904,
+      "loss": 7.1381,
       "step": 67
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.2800000000000001e-05,
-      "loss": 7.0965,
+      "loss": 7.0773,
       "step": 68
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 7.1285,
+      "loss": 7.0531,
       "step": 69
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.32e-05,
-      "loss": 7.0307,
+      "loss": 7.0903,
       "step": 70
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.3400000000000002e-05,
-      "loss": 7.1622,
+      "loss": 7.232,
       "step": 71
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.3600000000000002e-05,
-      "loss": 7.1392,
+      "loss": 7.2922,
       "step": 72
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 1.3800000000000002e-05,
-      "loss": 7.1067,
+      "loss": 7.1531,
       "step": 73
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 7.1494,
+      "loss": 7.1851,
       "step": 74
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.42e-05,
-      "loss": 7.1024,
+      "loss": 6.9753,
       "step": 75
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.44e-05,
-      "loss": 7.14,
+      "loss": 7.3358,
       "step": 76
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 1.4599999999999999e-05,
-      "loss": 7.1461,
+      "loss": 7.114,
       "step": 77
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.07,
       "learning_rate": 1.48e-05,
-      "loss": 7.1469,
+      "loss": 6.9431,
       "step": 78
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.5e-05,
-      "loss": 7.0241,
+      "loss": 6.974,
       "step": 79
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.52e-05,
-      "loss": 7.1238,
+      "loss": 7.0624,
       "step": 80
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.54e-05,
-      "loss": 7.0938,
+      "loss": 7.1886,
       "step": 81
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 1.56e-05,
-      "loss": 7.1138,
+      "loss": 7.2449,
       "step": 82
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.58e-05,
-      "loss": 7.114,
+      "loss": 6.9961,
       "step": 83
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 7.0832,
+      "loss": 7.115,
       "step": 84
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.62e-05,
-      "loss": 7.0722,
+      "loss": 7.0465,
       "step": 85
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 1.6400000000000002e-05,
-      "loss": 7.1229,
+      "loss": 7.1174,
       "step": 86
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.66e-05,
-      "loss": 7.083,
+      "loss": 7.02,
       "step": 87
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.6800000000000002e-05,
-      "loss": 7.039,
+      "loss": 7.1203,
       "step": 88
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 6.9971,
+      "loss": 7.2337,
       "step": 89
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 1.7199999999999998e-05,
-      "loss": 7.1448,
+      "loss": 7.0477,
       "step": 90
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 1.74e-05,
-      "loss": 7.0343,
+      "loss": 7.2718,
       "step": 91
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.76e-05,
-      "loss": 7.2564,
+      "loss": 7.1269,
       "step": 92
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.78e-05,
-      "loss": 7.1159,
+      "loss": 6.8366,
       "step": 93
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.8e-05,
-      "loss": 6.9343,
+      "loss": 7.1281,
       "step": 94
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 1.8200000000000002e-05,
-      "loss": 6.8484,
+      "loss": 7.2863,
       "step": 95
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.84e-05,
-      "loss": 7.0827,
+      "loss": 6.9239,
       "step": 96
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.86e-05,
-      "loss": 6.9604,
+      "loss": 7.0845,
       "step": 97
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.88e-05,
-      "loss": 6.7023,
+      "loss": 7.0067,
       "step": 98
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.9e-05,
-      "loss": 6.8238,
+      "loss": 6.6292,
       "step": 99
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 1.9200000000000003e-05,
-      "loss": 6.4454,
+      "loss": 6.4719,
       "step": 100
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.08,
       "learning_rate": 1.94e-05,
-      "loss": 7.3213,
+      "loss": 7.1517,
       "step": 101
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 1.9600000000000002e-05,
-      "loss": 7.2514,
+      "loss": 7.1006,
       "step": 102
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 1.9800000000000004e-05,
-      "loss": 7.0568,
+      "loss": 7.1244,
       "step": 103
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 2e-05,
-      "loss": 7.0644,
+      "loss": 7.1029,
       "step": 104
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.0200000000000003e-05,
-      "loss": 7.0865,
+      "loss": 7.0077,
       "step": 105
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.04e-05,
-      "loss": 7.0993,
+      "loss": 7.1126,
       "step": 106
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.06e-05,
-      "loss": 7.06,
+      "loss": 7.0549,
       "step": 107
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.08e-05,
-      "loss": 7.0767,
+      "loss": 7.1308,
       "step": 108
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 2.1e-05,
-      "loss": 6.9731,
+      "loss": 7.2395,
       "step": 109
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.12e-05,
-      "loss": 7.049,
+      "loss": 7.0873,
       "step": 110
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.1400000000000002e-05,
-      "loss": 6.9612,
+      "loss": 7.1595,
       "step": 111
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.16e-05,
-      "loss": 6.9691,
+      "loss": 7.1262,
       "step": 112
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 2.18e-05,
-      "loss": 7.0373,
+      "loss": 7.0425,
       "step": 113
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 7.0775,
+      "loss": 7.0284,
       "step": 114
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.22e-05,
-      "loss": 6.9912,
+      "loss": 7.1293,
       "step": 115
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.2400000000000002e-05,
-      "loss": 7.1242,
+      "loss": 7.2394,
       "step": 116
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.26e-05,
-      "loss": 7.0861,
+      "loss": 7.0671,
       "step": 117
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 2.2800000000000002e-05,
-      "loss": 7.0563,
+      "loss": 6.9098,
       "step": 118
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 6.9733,
+      "loss": 6.8846,
       "step": 119
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.32e-05,
-      "loss": 7.069,
+      "loss": 7.1227,
       "step": 120
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.3400000000000003e-05,
-      "loss": 7.0028,
+      "loss": 7.1857,
       "step": 121
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 2.36e-05,
-      "loss": 6.952,
+      "loss": 7.0162,
       "step": 122
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 2.38e-05,
-      "loss": 6.9993,
+      "loss": 7.1089,
       "step": 123
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 2.4e-05,
-      "loss": 6.9586,
+      "loss": 7.0784,
       "step": 124
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.4200000000000002e-05,
-      "loss": 7.0462,
+      "loss": 7.0332,
       "step": 125
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.44e-05,
-      "loss": 7.0564,
+      "loss": 6.9283,
       "step": 126
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 2.46e-05,
-      "loss": 7.0408,
+      "loss": 7.2615,
       "step": 127
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.48e-05,
-      "loss": 7.0007,
+      "loss": 6.9811,
       "step": 128
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.5e-05,
-      "loss": 7.0593,
+      "loss": 7.0035,
       "step": 129
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.5200000000000003e-05,
-      "loss": 7.1046,
+      "loss": 6.9837,
       "step": 130
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 2.54e-05,
-      "loss": 6.9157,
+      "loss": 6.8607,
       "step": 131
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.5600000000000002e-05,
-      "loss": 7.0974,
+      "loss": 7.0237,
       "step": 132
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.58e-05,
-      "loss": 6.9547,
+      "loss": 7.2092,
       "step": 133
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 6.9826,
+      "loss": 6.9669,
       "step": 134
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.6200000000000003e-05,
-      "loss": 7.1038,
+      "loss": 7.049,
       "step": 135
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 2.64e-05,
-      "loss": 6.8957,
+      "loss": 6.967,
       "step": 136
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.6600000000000003e-05,
-      "loss": 7.0749,
+      "loss": 6.982,
       "step": 137
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.6800000000000004e-05,
-      "loss": 6.9846,
+      "loss": 7.0445,
       "step": 138
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 6.974,
+      "loss": 6.9934,
       "step": 139
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 2.7200000000000004e-05,
-      "loss": 7.0781,
+      "loss": 7.1241,
       "step": 140
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7400000000000002e-05,
-      "loss": 7.048,
+      "loss": 7.0486,
       "step": 141
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7600000000000003e-05,
-      "loss": 7.0383,
+      "loss": 6.9578,
       "step": 142
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.7800000000000005e-05,
-      "loss": 6.8394,
+      "loss": 6.9053,
       "step": 143
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 6.8782,
+      "loss": 7.0169,
       "step": 144
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.8199999999999998e-05,
-      "loss": 6.8053,
+      "loss": 6.8802,
       "step": 145
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.84e-05,
-      "loss": 6.8366,
+      "loss": 7.0577,
       "step": 146
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.86e-05,
-      "loss": 6.8764,
+      "loss": 6.7687,
       "step": 147
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 2.88e-05,
-      "loss": 6.8728,
+      "loss": 6.5566,
       "step": 148
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.13,
       "learning_rate": 2.9e-05,
-      "loss": 6.6228,
+      "loss": 6.7013,
       "step": 149
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.9199999999999998e-05,
-      "loss": 6.4417,
+      "loss": 6.5686,
       "step": 150
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.94e-05,
-      "loss": 7.9532,
+      "loss": 7.2573,
       "step": 151
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.96e-05,
-      "loss": 7.4394,
+      "loss": 7.312,
       "step": 152
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 2.98e-05,
-      "loss": 7.0989,
+      "loss": 7.1299,
       "step": 153
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3e-05,
-      "loss": 6.9289,
+      "loss": 7.1751,
       "step": 154
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.02e-05,
-      "loss": 6.9146,
+      "loss": 7.0241,
       "step": 155
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.04e-05,
-      "loss": 6.9493,
+      "loss": 7.004,
       "step": 156
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.06e-05,
-      "loss": 7.0579,
+      "loss": 7.0294,
       "step": 157
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 3.08e-05,
-      "loss": 6.9747,
+      "loss": 7.1861,
       "step": 158
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 3.1e-05,
-      "loss": 6.9411,
+      "loss": 6.9487,
       "step": 159
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 3.12e-05,
-      "loss": 7.0004,
+      "loss": 6.895,
       "step": 160
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 3.1400000000000004e-05,
-      "loss": 6.9354,
+      "loss": 6.9653,
       "step": 161
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 3.16e-05,
-      "loss": 6.9884,
+      "loss": 7.0934,
       "step": 162
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.18e-05,
-      "loss": 6.9663,
+      "loss": 7.1509,
       "step": 163
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 6.9478,
+      "loss": 6.9768,
       "step": 164
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.2200000000000003e-05,
-      "loss": 6.8995,
+      "loss": 6.9718,
       "step": 165
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.24e-05,
-      "loss": 6.9691,
+      "loss": 6.8331,
       "step": 166
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 3.26e-05,
-      "loss": 6.8844,
+      "loss": 6.9732,
       "step": 167
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.2800000000000004e-05,
-      "loss": 6.8284,
+      "loss": 7.0108,
       "step": 168
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.3e-05,
-      "loss": 6.8572,
+      "loss": 7.029,
       "step": 169
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.32e-05,
-      "loss": 6.9827,
+      "loss": 6.9196,
       "step": 170
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 3.3400000000000005e-05,
-      "loss": 6.9342,
+      "loss": 6.9597,
       "step": 171
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.14,
       "learning_rate": 3.3600000000000004e-05,
-      "loss": 6.9696,
+      "loss": 6.9481,
       "step": 172
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.38e-05,
-      "loss": 6.9199,
+      "loss": 6.7448,
       "step": 173
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 6.9634,
+      "loss": 6.9987,
       "step": 174
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.4200000000000005e-05,
-      "loss": 6.9521,
+      "loss": 7.0149,
       "step": 175
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 3.4399999999999996e-05,
-      "loss": 6.8582,
+      "loss": 7.1468,
       "step": 176
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.46e-05,
-      "loss": 6.8404,
+      "loss": 7.1226,
       "step": 177
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.48e-05,
-      "loss": 6.9176,
+      "loss": 7.0432,
       "step": 178
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.5e-05,
-      "loss": 6.9263,
+      "loss": 6.8863,
       "step": 179
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 3.52e-05,
-      "loss": 6.9833,
+      "loss": 6.9892,
       "step": 180
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.54e-05,
-      "loss": 6.9524,
+      "loss": 7.0484,
       "step": 181
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.56e-05,
-      "loss": 6.8526,
+      "loss": 7.1684,
       "step": 182
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.58e-05,
-      "loss": 6.9004,
+      "loss": 6.9025,
       "step": 183
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 3.6e-05,
-      "loss": 7.0001,
+      "loss": 6.9378,
       "step": 184
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.16,
       "learning_rate": 3.62e-05,
-      "loss": 6.9245,
+      "loss": 6.7895,
       "step": 185
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.6400000000000004e-05,
-      "loss": 6.8041,
+      "loss": 6.8623,
       "step": 186
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.66e-05,
-      "loss": 6.8849,
+      "loss": 7.0826,
       "step": 187
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.68e-05,
-      "loss": 6.8692,
+      "loss": 6.8756,
       "step": 188
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 3.7e-05,
-      "loss": 6.9472,
+      "loss": 6.9103,
       "step": 189
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.72e-05,
-      "loss": 6.9907,
+      "loss": 6.8717,
       "step": 190
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.74e-05,
-      "loss": 6.9708,
+      "loss": 7.0968,
       "step": 191
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.76e-05,
-      "loss": 6.8517,
+      "loss": 7.0395,
       "step": 192
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.7800000000000004e-05,
-      "loss": 6.8986,
+      "loss": 6.8671,
       "step": 193
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 3.8e-05,
-      "loss": 6.8329,
+      "loss": 6.8825,
       "step": 194
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 3.82e-05,
-      "loss": 6.8162,
+      "loss": 6.7859,
       "step": 195
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 3.8400000000000005e-05,
-      "loss": 6.7039,
+      "loss": 6.7025,
       "step": 196
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 3.86e-05,
-      "loss": 6.6606,
+      "loss": 7.0283,
       "step": 197
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 3.88e-05,
-      "loss": 6.7452,
+      "loss": 6.7671,
       "step": 198
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 6.5957,
+      "loss": 6.6417,
       "step": 199
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.9200000000000004e-05,
-      "loss": 6.4363,
+      "loss": 6.2291,
       "step": 200
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.94e-05,
-      "loss": 7.297,
+      "loss": 7.0375,
       "step": 201
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 3.960000000000001e-05,
-      "loss": 7.269,
+      "loss": 6.9879,
       "step": 202
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 3.9800000000000005e-05,
-      "loss": 7.0585,
+      "loss": 7.0054,
       "step": 203
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4e-05,
-      "loss": 6.9563,
+      "loss": 6.8702,
       "step": 204
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4.02e-05,
-      "loss": 7.0203,
+      "loss": 6.8741,
       "step": 205
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4.0400000000000006e-05,
-      "loss": 6.9714,
+      "loss": 6.964,
       "step": 206
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 4.0600000000000004e-05,
-      "loss": 6.8684,
+      "loss": 6.9912,
       "step": 207
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.17,
       "learning_rate": 4.08e-05,
-      "loss": 6.9906,
+      "loss": 7.1566,
       "step": 208
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.1e-05,
-      "loss": 6.9575,
+      "loss": 6.989,
       "step": 209
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.12e-05,
-      "loss": 6.9988,
+      "loss": 7.1266,
       "step": 210
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 4.14e-05,
-      "loss": 6.8732,
+      "loss": 6.8965,
       "step": 211
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.16e-05,
-      "loss": 6.9373,
+      "loss": 6.8712,
       "step": 212
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.18e-05,
-      "loss": 6.7927,
+      "loss": 6.9341,
       "step": 213
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.2e-05,
-      "loss": 6.883,
+      "loss": 6.9179,
       "step": 214
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.22e-05,
-      "loss": 6.8603,
+      "loss": 6.9548,
       "step": 215
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 4.24e-05,
-      "loss": 6.9015,
+      "loss": 6.9452,
       "step": 216
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.26e-05,
-      "loss": 6.8545,
+      "loss": 6.9553,
       "step": 217
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.2800000000000004e-05,
-      "loss": 6.9038,
+      "loss": 6.952,
       "step": 218
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.3e-05,
-      "loss": 6.9118,
+      "loss": 6.8752,
       "step": 219
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 4.32e-05,
-      "loss": 6.9215,
+      "loss": 7.0775,
       "step": 220
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.3400000000000005e-05,
-      "loss": 6.926,
+      "loss": 6.8812,
       "step": 221
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.36e-05,
-      "loss": 6.9322,
+      "loss": 7.2083,
       "step": 222
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.38e-05,
-      "loss": 6.8753,
+      "loss": 7.0885,
       "step": 223
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 6.86,
+      "loss": 7.0697,
       "step": 224
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 4.4200000000000004e-05,
-      "loss": 6.9809,
+      "loss": 6.8708,
       "step": 225
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.44e-05,
-      "loss": 6.878,
+      "loss": 6.7835,
       "step": 226
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.46e-05,
-      "loss": 6.9999,
+      "loss": 7.033,
       "step": 227
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.4800000000000005e-05,
-      "loss": 6.921,
+      "loss": 7.0343,
       "step": 228
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 4.5e-05,
-      "loss": 6.8253,
+      "loss": 6.8657,
       "step": 229
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 4.52e-05,
-      "loss": 6.9247,
+      "loss": 6.9962,
       "step": 230
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 4.5400000000000006e-05,
-      "loss": 6.8881,
+      "loss": 7.0693,
       "step": 231
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.5600000000000004e-05,
-      "loss": 6.7857,
+      "loss": 6.9766,
       "step": 232
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.58e-05,
-      "loss": 6.8139,
+      "loss": 6.9592,
       "step": 233
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 6.8562,
+      "loss": 6.8095,
       "step": 234
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.6200000000000005e-05,
-      "loss": 6.9725,
+      "loss": 7.0233,
       "step": 235
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.64e-05,
-      "loss": 6.849,
+      "loss": 6.9603,
       "step": 236
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.660000000000001e-05,
-      "loss": 6.9843,
+      "loss": 6.9996,
       "step": 237
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 4.6800000000000006e-05,
-      "loss": 6.8724,
+      "loss": 6.9041,
       "step": 238
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.7e-05,
-      "loss": 6.7994,
+      "loss": 7.0113,
       "step": 239
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.72e-05,
-      "loss": 7.0013,
+      "loss": 6.7716,
       "step": 240
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.74e-05,
-      "loss": 6.8415,
+      "loss": 6.9219,
       "step": 241
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.76e-05,
-      "loss": 6.8517,
+      "loss": 6.8892,
       "step": 242
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 4.78e-05,
-      "loss": 6.9684,
+      "loss": 7.0176,
       "step": 243
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.8e-05,
-      "loss": 6.7506,
+      "loss": 7.037,
       "step": 244
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.82e-05,
-      "loss": 6.7834,
+      "loss": 6.7713,
       "step": 245
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.8400000000000004e-05,
-      "loss": 6.7216,
+      "loss": 6.741,
       "step": 246
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 4.86e-05,
-      "loss": 6.4929,
+      "loss": 7.2401,
       "step": 247
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.88e-05,
-      "loss": 6.7279,
+      "loss": 6.6211,
       "step": 248
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.9e-05,
-      "loss": 6.5741,
+      "loss": 6.5931,
       "step": 249
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.92e-05,
-      "loss": 6.2706,
+      "loss": 6.2639,
       "step": 250
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 4.94e-05,
-      "loss": 7.7476,
+      "loss": 7.0658,
       "step": 251
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.96e-05,
-      "loss": 7.6512,
+      "loss": 6.8681,
       "step": 252
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 4.9800000000000004e-05,
-      "loss": 7.3178,
+      "loss": 7.0808,
       "step": 253
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 5e-05,
-      "loss": 7.0158,
+      "loss": 6.8697,
       "step": 254
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 5.02e-05,
-      "loss": 7.1085,
+      "loss": 6.9275,
       "step": 255
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.22,
       "learning_rate": 5.0400000000000005e-05,
-      "loss": 6.9557,
+      "loss": 7.0266,
       "step": 256
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.0600000000000003e-05,
-      "loss": 6.9712,
+      "loss": 6.8679,
       "step": 257
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.08e-05,
-      "loss": 6.9671,
+      "loss": 7.0036,
       "step": 258
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.1000000000000006e-05,
-      "loss": 6.8909,
+      "loss": 6.9099,
       "step": 259
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 5.1200000000000004e-05,
-      "loss": 6.907,
+      "loss": 6.8513,
       "step": 260
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.14e-05,
-      "loss": 6.9379,
+      "loss": 6.9212,
       "step": 261
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.16e-05,
-      "loss": 6.8954,
+      "loss": 6.8269,
       "step": 262
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.1800000000000005e-05,
-      "loss": 6.9731,
+      "loss": 6.8293,
       "step": 263
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.2000000000000004e-05,
-      "loss": 6.9566,
+      "loss": 6.9572,
       "step": 264
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 5.22e-05,
-      "loss": 6.9632,
+      "loss": 6.8728,
       "step": 265
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 5.2400000000000007e-05,
-      "loss": 6.7572,
+      "loss": 6.9658,
       "step": 266
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 5.2600000000000005e-05,
-      "loss": 6.9634,
+      "loss": 6.8423,
       "step": 267
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 5.28e-05,
-      "loss": 6.8013,
+      "loss": 6.8384,
       "step": 268
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 5.300000000000001e-05,
-      "loss": 6.9485,
+      "loss": 6.9504,
       "step": 269
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.3200000000000006e-05,
-      "loss": 6.9842,
+      "loss": 6.9562,
       "step": 270
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.3400000000000004e-05,
-      "loss": 6.8659,
+      "loss": 7.029,
       "step": 271
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.360000000000001e-05,
-      "loss": 6.8663,
+      "loss": 6.8449,
       "step": 272
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.380000000000001e-05,
-      "loss": 6.7745,
+      "loss": 7.0153,
       "step": 273
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 5.4000000000000005e-05,
-      "loss": 6.7314,
+      "loss": 7.0158,
       "step": 274
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.420000000000001e-05,
-      "loss": 6.9675,
+      "loss": 7.0075,
       "step": 275
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.440000000000001e-05,
-      "loss": 6.8585,
+      "loss": 6.7583,
       "step": 276
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.4600000000000006e-05,
-      "loss": 6.8243,
+      "loss": 6.9008,
       "step": 277
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 5.4800000000000004e-05,
-      "loss": 6.919,
+      "loss": 6.9393,
       "step": 278
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.23,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 6.9548,
+      "loss": 7.0252,
       "step": 279
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.520000000000001e-05,
-      "loss": 6.9776,
+      "loss": 6.8239,
       "step": 280
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.5400000000000005e-05,
-      "loss": 6.9207,
+      "loss": 6.7894,
       "step": 281
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.560000000000001e-05,
-      "loss": 6.8584,
+      "loss": 6.9119,
       "step": 282
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 5.580000000000001e-05,
-      "loss": 6.804,
+      "loss": 7.0304,
       "step": 283
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.6000000000000006e-05,
-      "loss": 6.8749,
+      "loss": 7.009,
       "step": 284
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.620000000000001e-05,
-      "loss": 6.8776,
+      "loss": 6.9092,
       "step": 285
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.6399999999999995e-05,
-      "loss": 6.9489,
+      "loss": 6.926,
       "step": 286
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 5.66e-05,
-      "loss": 6.9029,
+      "loss": 6.8346,
       "step": 287
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.68e-05,
-      "loss": 6.8721,
+      "loss": 6.8209,
       "step": 288
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 6.8507,
+      "loss": 7.0022,
       "step": 289
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.72e-05,
-      "loss": 6.9153,
+      "loss": 6.8586,
       "step": 290
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 5.74e-05,
-      "loss": 6.9442,
+      "loss": 7.0947,
       "step": 291
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.25,
       "learning_rate": 5.76e-05,
-      "loss": 6.8966,
+      "loss": 6.968,
       "step": 292
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.7799999999999995e-05,
-      "loss": 6.7692,
+      "loss": 6.6243,
       "step": 293
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.8e-05,
-      "loss": 6.6988,
+      "loss": 6.7225,
       "step": 294
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.82e-05,
-      "loss": 6.6943,
+      "loss": 6.864,
       "step": 295
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 5.8399999999999997e-05,
-      "loss": 6.6624,
+      "loss": 7.034,
       "step": 296
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.86e-05,
-      "loss": 6.7417,
+      "loss": 6.5696,
       "step": 297
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.88e-05,
-      "loss": 6.6956,
+      "loss": 6.6805,
       "step": 298
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.9e-05,
-      "loss": 6.3483,
+      "loss": 6.8128,
       "step": 299
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.92e-05,
-      "loss": 6.3368,
+      "loss": 6.4896,
       "step": 300
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 5.94e-05,
-      "loss": 7.2329,
+      "loss": 6.8916,
       "step": 301
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 5.96e-05,
-      "loss": 6.9574,
+      "loss": 7.0109,
       "step": 302
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 5.9800000000000003e-05,
-      "loss": 6.9735,
+      "loss": 7.2182,
       "step": 303
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 6e-05,
-      "loss": 6.88,
+      "loss": 6.8207,
       "step": 304
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 6.02e-05,
-      "loss": 7.0161,
+      "loss": 6.9541,
       "step": 305
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.04e-05,
-      "loss": 6.837,
+      "loss": 6.8865,
       "step": 306
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.06e-05,
-      "loss": 6.8152,
+      "loss": 6.9264,
       "step": 307
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.08e-05,
-      "loss": 6.9723,
+      "loss": 6.8721,
       "step": 308
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 6.1e-05,
-      "loss": 6.8213,
+      "loss": 6.7873,
       "step": 309
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.12e-05,
-      "loss": 6.9178,
+      "loss": 7.0311,
       "step": 310
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.14e-05,
-      "loss": 6.9041,
+      "loss": 6.7227,
       "step": 311
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.16e-05,
-      "loss": 6.9225,
+      "loss": 6.7776,
       "step": 312
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.18e-05,
-      "loss": 6.8338,
+      "loss": 6.813,
       "step": 313
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 6.2e-05,
-      "loss": 6.8129,
+      "loss": 6.9337,
       "step": 314
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.26,
       "learning_rate": 6.220000000000001e-05,
-      "loss": 6.9667,
+      "loss": 6.8514,
       "step": 315
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.24e-05,
-      "loss": 6.8732,
+      "loss": 7.0988,
       "step": 316
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.26e-05,
-      "loss": 6.8426,
+      "loss": 6.9746,
       "step": 317
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 6.280000000000001e-05,
-      "loss": 6.7621,
+      "loss": 6.9976,
       "step": 318
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.3e-05,
-      "loss": 6.8767,
+      "loss": 6.8891,
       "step": 319
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.32e-05,
-      "loss": 6.858,
+      "loss": 6.9056,
       "step": 320
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.340000000000001e-05,
-      "loss": 6.8448,
+      "loss": 6.9837,
       "step": 321
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.36e-05,
-      "loss": 6.8316,
+      "loss": 6.8951,
       "step": 322
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 6.38e-05,
-      "loss": 6.9077,
+      "loss": 6.8557,
       "step": 323
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.400000000000001e-05,
-      "loss": 6.817,
+      "loss": 6.9098,
       "step": 324
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.42e-05,
-      "loss": 6.8925,
+      "loss": 6.7932,
       "step": 325
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.440000000000001e-05,
-      "loss": 6.918,
+      "loss": 6.9744,
       "step": 326
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 6.460000000000001e-05,
-      "loss": 6.8799,
+      "loss": 6.9093,
       "step": 327
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.48e-05,
-      "loss": 6.827,
+      "loss": 6.9594,
       "step": 328
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 6.9139,
+      "loss": 6.8792,
       "step": 329
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.52e-05,
-      "loss": 6.9905,
+      "loss": 7.0447,
       "step": 330
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.54e-05,
-      "loss": 6.9022,
+      "loss": 6.9182,
       "step": 331
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 6.560000000000001e-05,
-      "loss": 6.8186,
+      "loss": 7.0251,
       "step": 332
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.58e-05,
-      "loss": 7.0035,
+      "loss": 7.0966,
       "step": 333
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.6e-05,
-      "loss": 6.8941,
+      "loss": 6.862,
       "step": 334
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.620000000000001e-05,
-      "loss": 6.9976,
+      "loss": 6.9816,
       "step": 335
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 6.64e-05,
-      "loss": 6.8929,
+      "loss": 7.0401,
       "step": 336
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.66e-05,
-      "loss": 6.8064,
+      "loss": 6.8123,
       "step": 337
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.680000000000001e-05,
-      "loss": 6.9275,
+      "loss": 7.0709,
       "step": 338
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 6.7e-05,
-      "loss": 6.9152,
+      "loss": 6.8759,
       "step": 339
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 6.720000000000001e-05,
-      "loss": 6.9284,
+      "loss": 6.9694,
       "step": 340
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 6.740000000000001e-05,
-      "loss": 6.8239,
+      "loss": 7.6347,
       "step": 341
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.76e-05,
-      "loss": 6.9322,
+      "loss": 7.0917,
       "step": 342
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.780000000000001e-05,
-      "loss": 6.7908,
+      "loss": 6.83,
       "step": 343
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.800000000000001e-05,
-      "loss": 6.7672,
+      "loss": 7.0027,
       "step": 344
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 6.82e-05,
-      "loss": 6.7738,
+      "loss": 6.8037,
       "step": 345
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.840000000000001e-05,
-      "loss": 6.8416,
+      "loss": 6.7355,
       "step": 346
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.860000000000001e-05,
-      "loss": 6.6964,
+      "loss": 6.9161,
       "step": 347
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.879999999999999e-05,
-      "loss": 6.5442,
+      "loss": 6.9802,
       "step": 348
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.9e-05,
-      "loss": 6.6494,
+      "loss": 6.5977,
       "step": 349
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 6.92e-05,
-      "loss": 6.3474,
+      "loss": 6.7131,
       "step": 350
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.939999999999999e-05,
-      "loss": 7.0071,
+      "loss": 7.1996,
       "step": 351
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.96e-05,
-      "loss": 7.0238,
+      "loss": 7.0542,
       "step": 352
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 6.98e-05,
-      "loss": 6.8978,
+      "loss": 6.943,
       "step": 353
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 7e-05,
-      "loss": 6.8255,
+      "loss": 7.0235,
       "step": 354
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.02e-05,
-      "loss": 6.8964,
+      "loss": 6.8991,
       "step": 355
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.04e-05,
-      "loss": 7.011,
+      "loss": 6.998,
       "step": 356
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.06e-05,
-      "loss": 6.8799,
+      "loss": 7.0441,
       "step": 357
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.08e-05,
-      "loss": 6.9662,
+      "loss": 6.8244,
       "step": 358
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 7.1e-05,
-      "loss": 6.8902,
+      "loss": 6.9419,
       "step": 359
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.12e-05,
-      "loss": 6.8619,
+      "loss": 7.045,
       "step": 360
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.14e-05,
-      "loss": 6.8673,
+      "loss": 6.8331,
       "step": 361
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 7.16e-05,
-      "loss": 6.8263,
+      "loss": 6.8301,
       "step": 362
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.31,
       "learning_rate": 7.18e-05,
-      "loss": 6.751,
+      "loss": 7.032,
       "step": 363
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.2e-05,
-      "loss": 6.8258,
+      "loss": 6.9414,
       "step": 364
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.22e-05,
-      "loss": 6.923,
+      "loss": 6.8982,
       "step": 365
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.24e-05,
-      "loss": 6.8602,
+      "loss": 6.9995,
       "step": 366
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 7.26e-05,
-      "loss": 6.9981,
+      "loss": 6.9437,
       "step": 367
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.280000000000001e-05,
-      "loss": 6.9136,
+      "loss": 7.0787,
       "step": 368
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.3e-05,
-      "loss": 6.7903,
+      "loss": 6.9583,
       "step": 369
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.32e-05,
-      "loss": 6.8646,
+      "loss": 7.0143,
       "step": 370
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.340000000000001e-05,
-      "loss": 6.9187,
+      "loss": 6.8137,
       "step": 371
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 7.36e-05,
-      "loss": 6.951,
+      "loss": 6.8843,
       "step": 372
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 7.38e-05,
-      "loss": 6.8371,
+      "loss": 6.7664,
       "step": 373
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 7.4e-05,
-      "loss": 6.7438,
+      "loss": 6.8902,
       "step": 374
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 7.42e-05,
-      "loss": 6.8742,
+      "loss": 6.8532,
       "step": 375
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 7.44e-05,
-      "loss": 6.7999,
+      "loss": 7.034,
       "step": 376
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.46e-05,
-      "loss": 6.8798,
+      "loss": 6.8505,
       "step": 377
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.48e-05,
-      "loss": 6.8601,
+      "loss": 6.9052,
       "step": 378
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 6.8558,
+      "loss": 6.9518,
       "step": 379
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.52e-05,
-      "loss": 6.878,
+      "loss": 6.9093,
       "step": 380
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 7.54e-05,
-      "loss": 6.9825,
+      "loss": 6.8996,
       "step": 381
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.560000000000001e-05,
-      "loss": 6.9115,
+      "loss": 6.9103,
       "step": 382
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.58e-05,
-      "loss": 6.8925,
+      "loss": 6.727,
       "step": 383
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.6e-05,
-      "loss": 6.8652,
+      "loss": 7.0833,
       "step": 384
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 7.620000000000001e-05,
-      "loss": 6.9467,
+      "loss": 6.8944,
       "step": 385
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.32,
       "learning_rate": 7.64e-05,
-      "loss": 6.9003,
+      "loss": 6.7779,
       "step": 386
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.66e-05,
-      "loss": 6.8708,
+      "loss": 6.8474,
       "step": 387
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.680000000000001e-05,
-      "loss": 6.947,
+      "loss": 6.9242,
       "step": 388
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.7e-05,
-      "loss": 6.8567,
+      "loss": 7.0125,
       "step": 389
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 7.72e-05,
-      "loss": 6.7644,
+      "loss": 7.1872,
       "step": 390
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.740000000000001e-05,
-      "loss": 6.8399,
+      "loss": 7.0436,
       "step": 391
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.76e-05,
-      "loss": 6.8561,
+      "loss": 6.8846,
       "step": 392
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.780000000000001e-05,
-      "loss": 6.7626,
+      "loss": 6.7697,
       "step": 393
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 7.800000000000001e-05,
-      "loss": 6.7948,
+      "loss": 6.8717,
       "step": 394
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.82e-05,
-      "loss": 6.833,
+      "loss": 6.7052,
       "step": 395
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.840000000000001e-05,
-      "loss": 6.8497,
+      "loss": 6.8186,
       "step": 396
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.860000000000001e-05,
-      "loss": 6.7886,
+      "loss": 6.5507,
       "step": 397
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 7.88e-05,
-      "loss": 6.6483,
+      "loss": 6.5835,
       "step": 398
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.34,
       "learning_rate": 7.900000000000001e-05,
-      "loss": 6.7121,
+      "loss": 6.4965,
       "step": 399
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.920000000000001e-05,
-      "loss": 6.3681,
+      "loss": 6.3385,
       "step": 400
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.94e-05,
-      "loss": 7.026,
+      "loss": 6.9722,
       "step": 401
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.960000000000001e-05,
-      "loss": 6.9316,
+      "loss": 6.9542,
       "step": 402
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 7.98e-05,
-      "loss": 7.0616,
+      "loss": 6.9712,
       "step": 403
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8e-05,
-      "loss": 6.9233,
+      "loss": 7.0509,
       "step": 404
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.020000000000001e-05,
-      "loss": 6.8163,
+      "loss": 6.8197,
       "step": 405
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.04e-05,
-      "loss": 6.939,
+      "loss": 6.9267,
       "step": 406
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.060000000000001e-05,
-      "loss": 6.9437,
+      "loss": 6.9446,
       "step": 407
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 8.080000000000001e-05,
-      "loss": 6.8657,
+      "loss": 6.9549,
       "step": 408
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 8.1e-05,
-      "loss": 6.817,
+      "loss": 6.8005,
       "step": 409
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 8.120000000000001e-05,
-      "loss": 6.8519,
+      "loss": 7.0207,
       "step": 410
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 8.14e-05,
-      "loss": 6.9635,
+      "loss": 7.0019,
       "step": 411
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 8.16e-05,
-      "loss": 6.8816,
+      "loss": 6.9901,
       "step": 412
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.18e-05,
-      "loss": 6.9035,
+      "loss": 6.8819,
       "step": 413
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.2e-05,
-      "loss": 6.9018,
+      "loss": 6.813,
       "step": 414
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.22e-05,
-      "loss": 6.9679,
+      "loss": 6.8615,
       "step": 415
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.24e-05,
-      "loss": 6.9055,
+      "loss": 7.1023,
       "step": 416
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 8.26e-05,
-      "loss": 6.9111,
+      "loss": 6.8824,
       "step": 417
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.28e-05,
-      "loss": 6.807,
+      "loss": 6.7186,
       "step": 418
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.3e-05,
-      "loss": 6.8406,
+      "loss": 6.8284,
       "step": 419
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.32e-05,
-      "loss": 6.8676,
+      "loss": 7.0155,
       "step": 420
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 8.34e-05,
-      "loss": 6.932,
+      "loss": 6.9274,
       "step": 421
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.35,
       "learning_rate": 8.36e-05,
-      "loss": 6.8184,
+      "loss": 6.8079,
       "step": 422
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.38e-05,
-      "loss": 6.8714,
+      "loss": 6.9139,
       "step": 423
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.4e-05,
-      "loss": 6.9628,
+      "loss": 6.8378,
       "step": 424
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 8.42e-05,
-      "loss": 6.8231,
+      "loss": 7.1138,
       "step": 425
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.44e-05,
-      "loss": 6.9028,
+      "loss": 6.8973,
       "step": 426
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.46e-05,
-      "loss": 6.8424,
+      "loss": 6.8576,
       "step": 427
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.48e-05,
-      "loss": 6.8027,
+      "loss": 6.9101,
       "step": 428
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.5e-05,
-      "loss": 6.7379,
+      "loss": 7.0098,
       "step": 429
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 8.52e-05,
-      "loss": 6.8131,
+      "loss": 6.9232,
       "step": 430
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.54e-05,
-      "loss": 6.8467,
+      "loss": 6.9982,
       "step": 431
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.560000000000001e-05,
-      "loss": 6.9546,
+      "loss": 6.6777,
       "step": 432
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.58e-05,
-      "loss": 6.8111,
+      "loss": 6.9689,
       "step": 433
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 8.6e-05,
-      "loss": 6.7883,
+      "loss": 6.9129,
       "step": 434
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.620000000000001e-05,
-      "loss": 6.9072,
+      "loss": 6.9933,
       "step": 435
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.64e-05,
-      "loss": 6.778,
+      "loss": 6.7077,
       "step": 436
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.66e-05,
-      "loss": 6.7615,
+      "loss": 6.8051,
       "step": 437
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.680000000000001e-05,
-      "loss": 6.8482,
+      "loss": 6.775,
       "step": 438
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 8.7e-05,
-      "loss": 6.8256,
+      "loss": 7.0047,
       "step": 439
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.72e-05,
-      "loss": 6.8051,
+      "loss": 6.8368,
       "step": 440
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.740000000000001e-05,
-      "loss": 6.6436,
+      "loss": 6.9213,
       "step": 441
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.76e-05,
-      "loss": 6.8336,
+      "loss": 6.7953,
       "step": 442
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 8.78e-05,
-      "loss": 6.554,
+      "loss": 6.8183,
       "step": 443
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.800000000000001e-05,
-      "loss": 6.7343,
+      "loss": 6.7266,
       "step": 444
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.82e-05,
-      "loss": 6.4845,
+      "loss": 6.6057,
       "step": 445
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 8.840000000000001e-05,
-      "loss": 6.3899,
+      "loss": 6.7775,
       "step": 446
     },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.86e-05,
+      "loss": 6.7385,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.88e-05,
+      "loss": 6.8687,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 6.5854,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.92e-05,
+      "loss": 6.4474,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.94e-05,
+      "loss": 6.8885,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 6.8552,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.98e-05,
+      "loss": 6.9002,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9e-05,
+      "loss": 6.8761,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 6.91,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.04e-05,
+      "loss": 6.8655,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.06e-05,
+      "loss": 6.895,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 7.0138,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.1e-05,
+      "loss": 6.9679,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 6.9424,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 6.8648,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.16e-05,
+      "loss": 6.8458,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 6.9237,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 6.9511,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.22e-05,
+      "loss": 6.9354,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 6.8418,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 6.7693,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.28e-05,
+      "loss": 6.8263,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 6.6364,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 7.0722,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 6.8436,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 6.9876,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.38e-05,
+      "loss": 6.9677,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.4e-05,
+      "loss": 7.0472,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.42e-05,
+      "loss": 6.8728,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.44e-05,
+      "loss": 6.7339,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.46e-05,
+      "loss": 6.668,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.48e-05,
+      "loss": 6.8264,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.5e-05,
+      "loss": 6.8349,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.52e-05,
+      "loss": 7.0129,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.54e-05,
+      "loss": 6.7754,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.56e-05,
+      "loss": 6.7871,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.58e-05,
+      "loss": 6.8136,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.6e-05,
+      "loss": 6.918,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 6.9401,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.64e-05,
+      "loss": 6.768,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.66e-05,
+      "loss": 6.808,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 6.9011,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.7e-05,
+      "loss": 6.952,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.72e-05,
+      "loss": 6.8199,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.74e-05,
+      "loss": 6.945,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.76e-05,
+      "loss": 6.9324,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.78e-05,
+      "loss": 6.9868,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.8e-05,
+      "loss": 6.87,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.82e-05,
+      "loss": 6.8098,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.84e-05,
+      "loss": 6.7418,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.86e-05,
+      "loss": 6.7101,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.88e-05,
+      "loss": 6.9152,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 6.6035,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.92e-05,
+      "loss": 6.5791,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 6.816896915435791,
+      "eval_runtime": 765.6629,
+      "eval_samples_per_second": 3.451,
+      "eval_steps_per_second": 0.289,
+      "eval_wer": 1.5492588844574082,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.94e-05,
+      "loss": 6.9541,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 6.9048,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.98e-05,
+      "loss": 6.939,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001,
+      "loss": 6.8767,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.985486211901307e-05,
+      "loss": 6.8614,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970972423802612e-05,
+      "loss": 6.9876,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.956458635703919e-05,
+      "loss": 6.8565,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941944847605225e-05,
+      "loss": 6.8818,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.927431059506532e-05,
+      "loss": 6.8064,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.912917271407838e-05,
+      "loss": 6.898,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.898403483309145e-05,
+      "loss": 6.7305,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.883889695210451e-05,
+      "loss": 6.9452,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.869375907111757e-05,
+      "loss": 6.8496,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.854862119013063e-05,
+      "loss": 6.7898,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.84034833091437e-05,
+      "loss": 6.9233,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.825834542815675e-05,
+      "loss": 6.7248,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.811320754716981e-05,
+      "loss": 6.9197,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.796806966618288e-05,
+      "loss": 6.7908,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.782293178519594e-05,
+      "loss": 7.065,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.767779390420901e-05,
+      "loss": 6.7811,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.753265602322207e-05,
+      "loss": 6.9213,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.738751814223513e-05,
+      "loss": 6.9074,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.724238026124819e-05,
+      "loss": 6.9336,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.709724238026126e-05,
+      "loss": 6.8676,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.695210449927431e-05,
+      "loss": 6.8523,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.680696661828737e-05,
+      "loss": 6.7354,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.666182873730044e-05,
+      "loss": 6.8742,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.65166908563135e-05,
+      "loss": 6.8885,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.637155297532656e-05,
+      "loss": 6.8817,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.622641509433963e-05,
+      "loss": 6.874,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.60812772133527e-05,
+      "loss": 6.8362,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.593613933236575e-05,
+      "loss": 6.8776,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.579100145137882e-05,
+      "loss": 6.7152,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.564586357039188e-05,
+      "loss": 6.7962,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.550072568940493e-05,
+      "loss": 6.9336,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.5355587808418e-05,
+      "loss": 6.7417,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.521044992743106e-05,
+      "loss": 6.8053,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.506531204644412e-05,
+      "loss": 6.7024,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.492017416545718e-05,
+      "loss": 6.874,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.477503628447025e-05,
+      "loss": 6.9014,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.462989840348333e-05,
+      "loss": 6.9623,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.448476052249638e-05,
+      "loss": 6.7767,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.433962264150944e-05,
+      "loss": 6.6075,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.419448476052251e-05,
+      "loss": 6.9661,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.404934687953556e-05,
+      "loss": 6.8717,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.390420899854863e-05,
+      "loss": 6.7853,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.375907111756169e-05,
+      "loss": 6.5027,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.361393323657474e-05,
+      "loss": 6.6507,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.346879535558781e-05,
+      "loss": 6.5563,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.332365747460087e-05,
+      "loss": 6.4705,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.317851959361394e-05,
+      "loss": 7.015,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.3033381712627e-05,
+      "loss": 6.8955,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.288824383164007e-05,
+      "loss": 6.9354,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.274310595065312e-05,
+      "loss": 6.798,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.259796806966619e-05,
+      "loss": 6.9976,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.245283018867925e-05,
+      "loss": 6.8724,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 6.8855,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.216255442670537e-05,
+      "loss": 6.8249,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.201741654571843e-05,
+      "loss": 6.9647,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.18722786647315e-05,
+      "loss": 6.786,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.172714078374456e-05,
+      "loss": 7.0736,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.158200290275763e-05,
+      "loss": 6.6794,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.14368650217707e-05,
+      "loss": 6.9444,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.129172714078375e-05,
+      "loss": 6.5859,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.114658925979681e-05,
+      "loss": 6.7858,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.100145137880988e-05,
+      "loss": 6.8445,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.085631349782293e-05,
+      "loss": 6.7727,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.0711175616836e-05,
+      "loss": 6.8257,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.056603773584906e-05,
+      "loss": 6.846,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.042089985486212e-05,
+      "loss": 6.8269,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.027576197387519e-05,
+      "loss": 6.886,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.013062409288826e-05,
+      "loss": 6.88,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.998548621190132e-05,
+      "loss": 6.8439,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.984034833091437e-05,
+      "loss": 6.9082,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.969521044992744e-05,
+      "loss": 7.0682,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.95500725689405e-05,
+      "loss": 6.9712,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.940493468795355e-05,
+      "loss": 6.9183,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.925979680696662e-05,
+      "loss": 6.8653,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.911465892597968e-05,
+      "loss": 6.8013,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.896952104499274e-05,
+      "loss": 6.9922,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.882438316400582e-05,
+      "loss": 6.9692,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.867924528301888e-05,
+      "loss": 6.7556,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.853410740203193e-05,
+      "loss": 6.7936,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.8388969521045e-05,
+      "loss": 6.7749,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.824383164005806e-05,
+      "loss": 7.0034,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.809869375907113e-05,
+      "loss": 6.8547,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.795355587808418e-05,
+      "loss": 6.9211,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.780841799709725e-05,
+      "loss": 6.8094,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.766328011611031e-05,
+      "loss": 6.7241,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.751814223512336e-05,
+      "loss": 6.8183,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.737300435413643e-05,
+      "loss": 6.907,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.722786647314949e-05,
+      "loss": 6.7743,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.708272859216256e-05,
+      "loss": 6.892,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.693759071117562e-05,
+      "loss": 6.7386,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.679245283018869e-05,
+      "loss": 6.641,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.664731494920174e-05,
+      "loss": 6.6438,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.65021770682148e-05,
+      "loss": 6.5923,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.635703918722787e-05,
+      "loss": 6.598,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.621190130624092e-05,
+      "loss": 6.4153,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.606676342525399e-05,
+      "loss": 6.3571,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.592162554426705e-05,
+      "loss": 6.776,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.577648766328012e-05,
+      "loss": 6.986,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.563134978229318e-05,
+      "loss": 6.9519,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.548621190130625e-05,
+      "loss": 6.8186,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.534107402031931e-05,
+      "loss": 6.9578,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.519593613933237e-05,
+      "loss": 7.0568,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.505079825834543e-05,
+      "loss": 6.7339,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.49056603773585e-05,
+      "loss": 6.9278,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.476052249637155e-05,
+      "loss": 6.8668,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 6.9006,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.447024673439768e-05,
+      "loss": 6.9543,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.432510885341074e-05,
+      "loss": 6.7867,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.417997097242381e-05,
+      "loss": 6.8398,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.403483309143688e-05,
+      "loss": 6.769,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.388969521044994e-05,
+      "loss": 6.9037,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.374455732946299e-05,
+      "loss": 6.7226,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.359941944847606e-05,
+      "loss": 6.8153,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.345428156748912e-05,
+      "loss": 6.8123,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.330914368650217e-05,
+      "loss": 6.8919,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.316400580551524e-05,
+      "loss": 6.938,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.30188679245283e-05,
+      "loss": 6.6307,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.287373004354137e-05,
+      "loss": 6.7166,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.272859216255444e-05,
+      "loss": 6.7758,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.25834542815675e-05,
+      "loss": 6.8605,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.243831640058055e-05,
+      "loss": 6.8635,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.229317851959362e-05,
+      "loss": 6.9247,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.214804063860668e-05,
+      "loss": 6.9132,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.200290275761974e-05,
+      "loss": 6.6796,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.18577648766328e-05,
+      "loss": 6.7311,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.171262699564587e-05,
+      "loss": 6.7434,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.156748911465893e-05,
+      "loss": 6.8427,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.142235123367198e-05,
+      "loss": 6.9377,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.127721335268506e-05,
+      "loss": 6.8147,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.113207547169813e-05,
+      "loss": 6.9054,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.098693759071118e-05,
+      "loss": 6.8393,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.084179970972424e-05,
+      "loss": 6.8763,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.069666182873731e-05,
+      "loss": 6.7708,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.055152394775036e-05,
+      "loss": 6.7692,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.040638606676343e-05,
+      "loss": 6.8221,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.026124818577649e-05,
+      "loss": 7.0328,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.011611030478954e-05,
+      "loss": 6.7504,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.997097242380261e-05,
+      "loss": 6.7132,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.982583454281567e-05,
+      "loss": 6.9498,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.968069666182875e-05,
+      "loss": 6.7619,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.95355587808418e-05,
+      "loss": 6.8033,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.939042089985487e-05,
+      "loss": 6.6719,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.924528301886794e-05,
+      "loss": 6.2472,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.910014513788099e-05,
+      "loss": 6.7799,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.895500725689405e-05,
+      "loss": 6.755,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.880986937590712e-05,
+      "loss": 6.2811,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.866473149492017e-05,
+      "loss": 7.0813,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.851959361393323e-05,
+      "loss": 6.8376,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.83744557329463e-05,
+      "loss": 6.9793,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.822931785195937e-05,
+      "loss": 6.9689,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.808417997097243e-05,
+      "loss": 6.9487,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.79390420899855e-05,
+      "loss": 6.897,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.779390420899855e-05,
+      "loss": 6.7693,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.764876632801161e-05,
+      "loss": 6.8785,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.750362844702468e-05,
+      "loss": 6.8355,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.735849056603774e-05,
+      "loss": 6.8414,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.72133526850508e-05,
+      "loss": 6.7411,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.706821480406386e-05,
+      "loss": 6.8667,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 6.6934,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.677793904208999e-05,
+      "loss": 6.9048,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.663280116110306e-05,
+      "loss": 6.8159,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.648766328011612e-05,
+      "loss": 6.6955,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.634252539912917e-05,
+      "loss": 6.9094,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.619738751814224e-05,
+      "loss": 6.7883,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.60522496371553e-05,
+      "loss": 7.106,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.590711175616836e-05,
+      "loss": 6.9303,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.576197387518142e-05,
+      "loss": 6.8192,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.561683599419449e-05,
+      "loss": 6.7872,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.547169811320755e-05,
+      "loss": 6.7987,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.532656023222062e-05,
+      "loss": 6.7436,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.518142235123368e-05,
+      "loss": 6.9109,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.503628447024675e-05,
+      "loss": 6.7913,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.48911465892598e-05,
+      "loss": 6.9789,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.474600870827286e-05,
+      "loss": 6.9467,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.460087082728593e-05,
+      "loss": 6.8445,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.445573294629898e-05,
+      "loss": 6.96,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.431059506531205e-05,
+      "loss": 6.6298,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.416545718432511e-05,
+      "loss": 6.7145,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.402031930333816e-05,
+      "loss": 6.7885,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387518142235124e-05,
+      "loss": 6.7045,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373004354136431e-05,
+      "loss": 6.786,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.358490566037736e-05,
+      "loss": 6.7507,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.343976777939043e-05,
+      "loss": 6.8735,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.329462989840349e-05,
+      "loss": 6.8046,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.314949201741656e-05,
+      "loss": 6.9841,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.300435413642961e-05,
+      "loss": 6.7626,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.285921625544267e-05,
+      "loss": 6.9631,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.271407837445574e-05,
+      "loss": 6.7161,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.256894049346879e-05,
+      "loss": 6.6926,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.242380261248185e-05,
+      "loss": 6.8529,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.227866473149493e-05,
+      "loss": 6.8109,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213352685050799e-05,
+      "loss": 6.4349,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.198838896952105e-05,
+      "loss": 6.8403,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184325108853412e-05,
+      "loss": 6.7861,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.169811320754717e-05,
+      "loss": 6.3615,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.155297532656023e-05,
+      "loss": 6.6005,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.14078374455733e-05,
+      "loss": 7.0148,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126269956458636e-05,
+      "loss": 6.7751,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.111756168359942e-05,
+      "loss": 6.9264,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097242380261248e-05,
+      "loss": 6.7989,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.082728592162555e-05,
+      "loss": 6.828,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.068214804063861e-05,
+      "loss": 6.912,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.053701015965168e-05,
+      "loss": 6.8274,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039187227866474e-05,
+      "loss": 6.7373,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.02467343976778e-05,
+      "loss": 6.8207,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010159651669086e-05,
+      "loss": 6.9445,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.995645863570392e-05,
+      "loss": 6.7258,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981132075471698e-05,
+      "loss": 7.0621,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.966618287373004e-05,
+      "loss": 6.859,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.95210449927431e-05,
+      "loss": 6.8342,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937590711175617e-05,
+      "loss": 6.7359,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 6.7588,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90856313497823e-05,
+      "loss": 6.786,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894049346879537e-05,
+      "loss": 6.8782,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.879535558780842e-05,
+      "loss": 6.8366,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865021770682148e-05,
+      "loss": 6.7578,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850507982583455e-05,
+      "loss": 6.6656,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83599419448476e-05,
+      "loss": 6.7171,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821480406386067e-05,
+      "loss": 6.7658,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.806966618287373e-05,
+      "loss": 6.6905,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.79245283018868e-05,
+      "loss": 6.8089,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777939042089986e-05,
+      "loss": 6.7434,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763425253991293e-05,
+      "loss": 6.7556,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.748911465892598e-05,
+      "loss": 6.8644,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734397677793905e-05,
+      "loss": 6.7758,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719883889695211e-05,
+      "loss": 6.8507,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705370101596516e-05,
+      "loss": 6.7743,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690856313497823e-05,
+      "loss": 6.8178,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676342525399129e-05,
+      "loss": 6.8479,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661828737300436e-05,
+      "loss": 6.7398,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647314949201742e-05,
+      "loss": 6.8079,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632801161103049e-05,
+      "loss": 6.7721,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618287373004355e-05,
+      "loss": 6.9678,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60377358490566e-05,
+      "loss": 6.9002,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589259796806967e-05,
+      "loss": 6.8699,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574746008708274e-05,
+      "loss": 6.7618,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560232220609579e-05,
+      "loss": 6.9287,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545718432510885e-05,
+      "loss": 6.864,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531204644412192e-05,
+      "loss": 6.8797,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.516690856313497e-05,
+      "loss": 6.6929,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502177068214804e-05,
+      "loss": 6.667,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487663280116111e-05,
+      "loss": 6.5773,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473149492017418e-05,
+      "loss": 6.6875,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458635703918723e-05,
+      "loss": 6.9012,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44412191582003e-05,
+      "loss": 6.4418,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429608127721336e-05,
+      "loss": 6.1,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415094339622641e-05,
+      "loss": 6.9027,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400580551523948e-05,
+      "loss": 7.0302,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386066763425254e-05,
+      "loss": 6.7954,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37155297532656e-05,
+      "loss": 6.7707,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357039187227866e-05,
+      "loss": 6.9749,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.342525399129173e-05,
+      "loss": 6.904,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.328011611030479e-05,
+      "loss": 6.8076,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.313497822931786e-05,
+      "loss": 6.9567,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.298984034833092e-05,
+      "loss": 6.6893,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.284470246734397e-05,
+      "loss": 6.8536,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.269956458635704e-05,
+      "loss": 6.797,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.25544267053701e-05,
+      "loss": 6.6714,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.240928882438317e-05,
+      "loss": 6.7449,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.226415094339622e-05,
+      "loss": 6.6406,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211901306240929e-05,
+      "loss": 7.0426,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197387518142235e-05,
+      "loss": 6.8331,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182873730043542e-05,
+      "loss": 6.8195,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.168359941944848e-05,
+      "loss": 6.651,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 6.7969,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.13933236574746e-05,
+      "loss": 6.8453,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.124818577648767e-05,
+      "loss": 6.7899,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.110304789550073e-05,
+      "loss": 6.9263,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095791001451378e-05,
+      "loss": 6.7198,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0812772133526855e-05,
+      "loss": 6.6432,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.066763425253992e-05,
+      "loss": 6.8854,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0522496371552985e-05,
+      "loss": 6.7342,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.037735849056604e-05,
+      "loss": 6.7204,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.02322206095791e-05,
+      "loss": 6.7801,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.008708272859217e-05,
+      "loss": 6.8489,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9941944847605226e-05,
+      "loss": 6.9924,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.979680696661829e-05,
+      "loss": 6.8397,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.965166908563136e-05,
+      "loss": 6.7032,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.950653120464441e-05,
+      "loss": 6.6046,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9361393323657474e-05,
+      "loss": 6.8323,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9216255442670546e-05,
+      "loss": 6.8935,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.90711175616836e-05,
+      "loss": 6.6631,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.892597968069666e-05,
+      "loss": 6.9037,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.878084179970973e-05,
+      "loss": 6.7529,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.863570391872279e-05,
+      "loss": 6.7957,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.849056603773585e-05,
+      "loss": 6.723,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.834542815674892e-05,
+      "loss": 6.847,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.820029027576198e-05,
+      "loss": 6.7401,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8055152394775034e-05,
+      "loss": 6.7185,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.79100145137881e-05,
+      "loss": 6.6107,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7764876632801165e-05,
+      "loss": 6.6154,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7619738751814224e-05,
+      "loss": 6.5039,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.747460087082729e-05,
+      "loss": 6.5098,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7329462989840354e-05,
+      "loss": 6.4776,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.718432510885341e-05,
+      "loss": 6.451,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.703918722786648e-05,
+      "loss": 6.1167,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.689404934687954e-05,
+      "loss": 6.842,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6748911465892595e-05,
+      "loss": 6.6685,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 6.7821,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6458635703918726e-05,
+      "loss": 6.8192,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.631349782293179e-05,
+      "loss": 6.8466,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.616835994194485e-05,
+      "loss": 6.7241,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6023222060957915e-05,
+      "loss": 6.7587,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587808417997098e-05,
+      "loss": 6.8157,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.573294629898403e-05,
+      "loss": 6.7791,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.55878084179971e-05,
+      "loss": 6.8145,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.544267053701017e-05,
+      "loss": 6.7379,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.529753265602322e-05,
+      "loss": 6.76,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5152394775036286e-05,
+      "loss": 6.709,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.500725689404935e-05,
+      "loss": 7.0016,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.486211901306241e-05,
+      "loss": 6.9221,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4716981132075475e-05,
+      "loss": 6.646,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.457184325108854e-05,
+      "loss": 6.8079,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.442670537010159e-05,
+      "loss": 6.5571,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.428156748911466e-05,
+      "loss": 6.7677,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.413642960812772e-05,
+      "loss": 6.8188,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.399129172714079e-05,
+      "loss": 6.8109,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 6.7071,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.370101596516691e-05,
+      "loss": 6.7861,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.355587808417998e-05,
+      "loss": 6.8952,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3410740203193036e-05,
+      "loss": 6.7592,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.32656023222061e-05,
+      "loss": 6.7845,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.3120464441219166e-05,
+      "loss": 6.8417,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.297532656023222e-05,
+      "loss": 6.8408,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.283018867924528e-05,
+      "loss": 6.7785,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.268505079825835e-05,
+      "loss": 6.968,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.253991291727141e-05,
+      "loss": 6.8102,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.239477503628447e-05,
+      "loss": 6.5849,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.224963715529754e-05,
+      "loss": 6.7127,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.21044992743106e-05,
+      "loss": 6.8402,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1959361393323655e-05,
+      "loss": 6.8481,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.181422351233673e-05,
+      "loss": 6.8904,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.166908563134979e-05,
+      "loss": 6.832,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1523947750362844e-05,
+      "loss": 6.4837,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.137880986937591e-05,
+      "loss": 6.9687,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1233671988388975e-05,
+      "loss": 6.6962,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108853410740203e-05,
+      "loss": 6.8096,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.09433962264151e-05,
+      "loss": 6.703,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0798258345428164e-05,
+      "loss": 6.9807,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0653120464441215e-05,
+      "loss": 6.5803,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.050798258345428e-05,
+      "loss": 6.7102,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0362844702467346e-05,
+      "loss": 6.6611,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0217706821480404e-05,
+      "loss": 6.6247,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.007256894049347e-05,
+      "loss": 6.3862,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9927431059506535e-05,
+      "loss": 6.7488,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9782293178519594e-05,
+      "loss": 6.5929,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.963715529753266e-05,
+      "loss": 7.0097,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9492017416545724e-05,
+      "loss": 6.7953,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.934687953555878e-05,
+      "loss": 6.816,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.920174165457185e-05,
+      "loss": 6.9158,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 6.8304,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.891146589259797e-05,
+      "loss": 6.8529,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.876632801161104e-05,
+      "loss": 6.6973,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8621190130624096e-05,
+      "loss": 6.8356,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8476052249637154e-05,
+      "loss": 6.7024,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833091436865022e-05,
+      "loss": 6.745,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.818577648766328e-05,
+      "loss": 6.7976,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.804063860667635e-05,
+      "loss": 6.8322,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.789550072568941e-05,
+      "loss": 6.6484,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.775036284470247e-05,
+      "loss": 6.823,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.760522496371553e-05,
+      "loss": 6.8371,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.746008708272859e-05,
+      "loss": 6.7053,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.731494920174166e-05,
+      "loss": 6.6614,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.716981132075472e-05,
+      "loss": 6.694,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.702467343976778e-05,
+      "loss": 6.657,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6879535558780845e-05,
+      "loss": 6.6412,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6734397677793904e-05,
+      "loss": 6.6954,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.658925979680697e-05,
+      "loss": 6.6111,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6444121915820034e-05,
+      "loss": 6.8912,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.629898403483309e-05,
+      "loss": 6.8317,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 6.8738,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.600870827285922e-05,
+      "loss": 6.6447,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.586357039187228e-05,
+      "loss": 6.6893,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.571843251088535e-05,
+      "loss": 6.8431,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5573294629898406e-05,
+      "loss": 6.7028,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5428156748911464e-05,
+      "loss": 6.8807,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 6.7155,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5137880986937595e-05,
+      "loss": 6.8045,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.499274310595066e-05,
+      "loss": 6.8412,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.484760522496372e-05,
+      "loss": 6.8275,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.470246734397678e-05,
+      "loss": 6.6627,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.455732946298984e-05,
+      "loss": 6.7555,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.441219158200291e-05,
+      "loss": 6.747,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.4267053701015966e-05,
+      "loss": 6.7697,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.412191582002903e-05,
+      "loss": 6.8475,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.397677793904209e-05,
+      "loss": 6.6429,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3831640058055155e-05,
+      "loss": 6.8477,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3686502177068214e-05,
+      "loss": 6.7814,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.354136429608128e-05,
+      "loss": 6.6208,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3396226415094345e-05,
+      "loss": 6.6035,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.32510885341074e-05,
+      "loss": 6.7138,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310595065312046e-05,
+      "loss": 6.6683,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.296081277213353e-05,
+      "loss": 6.4505,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.281567489114659e-05,
+      "loss": 6.2522,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.267053701015966e-05,
+      "loss": 6.528,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2525399129172716e-05,
+      "loss": 5.9871,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2380261248185774e-05,
+      "loss": 6.8863,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.223512336719884e-05,
+      "loss": 6.8323,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2089985486211905e-05,
+      "loss": 6.8203,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.194484760522497e-05,
+      "loss": 6.6356,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.179970972423803e-05,
+      "loss": 6.7684,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.165457184325109e-05,
+      "loss": 6.8672,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 6.745,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.136429608127722e-05,
+      "loss": 7.0935,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1219158200290276e-05,
+      "loss": 6.7924,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.107402031930334e-05,
+      "loss": 6.8196,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.09288824383164e-05,
+      "loss": 6.734,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0783744557329466e-05,
+      "loss": 6.8255,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063860667634253e-05,
+      "loss": 6.7578,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.049346879535559e-05,
+      "loss": 6.6894,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0348330914368655e-05,
+      "loss": 6.7667,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.020319303338171e-05,
+      "loss": 6.8318,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.005805515239477e-05,
+      "loss": 6.7737,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.991291727140784e-05,
+      "loss": 6.7788,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.97677793904209e-05,
+      "loss": 6.8496,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.962264150943397e-05,
+      "loss": 6.7655,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9477503628447026e-05,
+      "loss": 6.682,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9332365747460085e-05,
+      "loss": 6.9001,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.918722786647315e-05,
+      "loss": 6.7989,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9042089985486215e-05,
+      "loss": 6.8357,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8896952104499274e-05,
+      "loss": 6.679,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.875181422351234e-05,
+      "loss": 6.7386,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.86066763425254e-05,
+      "loss": 6.8577,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 6.8165,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.831640058055153e-05,
+      "loss": 6.6807,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.817126269956459e-05,
+      "loss": 6.7314,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.802612481857765e-05,
+      "loss": 6.7713,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.788098693759071e-05,
+      "loss": 6.8034,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 6.8894,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.759071117561684e-05,
+      "loss": 6.7599,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.74455732946299e-05,
+      "loss": 6.7594,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7300435413642965e-05,
+      "loss": 6.7416,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7155297532656023e-05,
+      "loss": 6.7552,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.701015965166908e-05,
+      "loss": 6.9256,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6865021770682154e-05,
+      "loss": 6.7792,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.671988388969521e-05,
+      "loss": 6.6895,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.657474600870828e-05,
+      "loss": 6.6861,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6429608127721336e-05,
+      "loss": 6.7978,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6284470246734395e-05,
+      "loss": 6.4635,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.613933236574747e-05,
+      "loss": 6.4301,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.5994194484760525e-05,
+      "loss": 6.811,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.5849056603773584e-05,
+      "loss": 6.5899,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.570391872278665e-05,
+      "loss": 6.4177,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.555878084179971e-05,
+      "loss": 6.3287,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.541364296081277e-05,
+      "loss": 6.3375,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.526850507982584e-05,
+      "loss": 6.1771,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.51233671988389e-05,
+      "loss": 6.8212,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.497822931785196e-05,
+      "loss": 6.8401,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.483309143686502e-05,
+      "loss": 6.7216,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4687953555878086e-05,
+      "loss": 6.5769,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.454281567489115e-05,
+      "loss": 6.724,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.439767779390421e-05,
+      "loss": 6.8832,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4252539912917275e-05,
+      "loss": 6.7873,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4107402031930334e-05,
+      "loss": 6.8771,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 6.7765,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3817126269956464e-05,
+      "loss": 6.6819,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.367198838896952e-05,
+      "loss": 6.7155,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.352685050798258e-05,
+      "loss": 6.7164,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3381712626995646e-05,
+      "loss": 6.6213,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.323657474600871e-05,
+      "loss": 6.7345,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.309143686502178e-05,
+      "loss": 6.7811,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2946298984034836e-05,
+      "loss": 6.7716,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2801161103047894e-05,
+      "loss": 6.9639,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.265602322206096e-05,
+      "loss": 6.7817,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.251088534107402e-05,
+      "loss": 6.5145,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.236574746008709e-05,
+      "loss": 6.7859,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.222060957910015e-05,
+      "loss": 6.6373,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207547169811321e-05,
+      "loss": 6.6221,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.193033381712627e-05,
+      "loss": 6.7009,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.178519593613933e-05,
+      "loss": 6.7816,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1640058055152396e-05,
+      "loss": 6.7988,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.149492017416546e-05,
+      "loss": 6.8296,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.134978229317852e-05,
+      "loss": 6.6786,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1204644412191585e-05,
+      "loss": 6.7087,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1059506531204644e-05,
+      "loss": 6.7003,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.091436865021771e-05,
+      "loss": 6.801,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 7.0072,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.062409288824383e-05,
+      "loss": 6.7875,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.047895500725689e-05,
+      "loss": 6.7898,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033381712626996e-05,
+      "loss": 6.9678,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 6.6714,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0043541364296084e-05,
+      "loss": 6.6592,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9898403483309146e-05,
+      "loss": 6.8233,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9753265602322204e-05,
+      "loss": 6.7382,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9608127721335273e-05,
+      "loss": 6.7139,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.946298984034833e-05,
+      "loss": 6.6707,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9317851959361393e-05,
+      "loss": 6.7045,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.917271407837446e-05,
+      "loss": 6.7427,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9027576197387517e-05,
+      "loss": 6.619,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8882438316400583e-05,
+      "loss": 6.852,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8737300435413644e-05,
+      "loss": 6.7438,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8592162554426706e-05,
+      "loss": 6.9753,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.844702467343977e-05,
+      "loss": 6.6846,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 6.4898,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8156748911465895e-05,
+      "loss": 6.4531,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8011611030478957e-05,
+      "loss": 6.3192,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 6.661755084991455,
+      "eval_runtime": 804.0636,
+      "eval_samples_per_second": 3.286,
+      "eval_steps_per_second": 0.275,
+      "eval_wer": 1.8028255650137905,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7866473149492016e-05,
+      "loss": 6.7604,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7721335268505085e-05,
+      "loss": 6.8039,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7576197387518143e-05,
+      "loss": 6.7985,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7431059506531205e-05,
+      "loss": 6.8165,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.728592162554427e-05,
+      "loss": 6.7572,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.714078374455733e-05,
+      "loss": 6.8065,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6995645863570394e-05,
+      "loss": 6.7116,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6850507982583456e-05,
+      "loss": 6.6239,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6705370101596518e-05,
+      "loss": 6.8422,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6560232220609583e-05,
+      "loss": 6.6058,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 6.6577,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6269956458635704e-05,
+      "loss": 6.7455,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.612481857764877e-05,
+      "loss": 6.6332,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5979680696661827e-05,
+      "loss": 6.6506,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5834542815674896e-05,
+      "loss": 6.6368,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5689404934687955e-05,
+      "loss": 6.8033,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5544267053701017e-05,
+      "loss": 6.6214,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5399129172714082e-05,
+      "loss": 6.7163,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.525399129172714e-05,
+      "loss": 6.8469,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5108853410740202e-05,
+      "loss": 6.6031,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4963715529753268e-05,
+      "loss": 6.7796,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.481857764876633e-05,
+      "loss": 6.8408,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.467343976777939e-05,
+      "loss": 6.8541,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 6.6765,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.438316400580552e-05,
+      "loss": 6.6333,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4238026124818577e-05,
+      "loss": 6.6848,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.409288824383164e-05,
+      "loss": 6.6127,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3947750362844704e-05,
+      "loss": 6.7227,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3802612481857766e-05,
+      "loss": 6.741,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.365747460087083e-05,
+      "loss": 6.6524,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.351233671988389e-05,
+      "loss": 6.774,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3367198838896952e-05,
+      "loss": 6.5939,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3222060957910017e-05,
+      "loss": 6.7241,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 6.6701,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.293178519593614e-05,
+      "loss": 6.7152,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2786647314949203e-05,
+      "loss": 6.828,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 6.5995,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.249637155297533e-05,
+      "loss": 6.7937,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.235123367198839e-05,
+      "loss": 6.7432,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2206095791001454e-05,
+      "loss": 6.6041,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2060957910014516e-05,
+      "loss": 6.5828,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1915820029027578e-05,
+      "loss": 6.6497,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.177068214804064e-05,
+      "loss": 6.613,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.16255442670537e-05,
+      "loss": 6.6383,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1480406386066763e-05,
+      "loss": 6.6637,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.133526850507983e-05,
+      "loss": 6.6572,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1190130624092887e-05,
+      "loss": 6.434,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1044992743105953e-05,
+      "loss": 6.6083,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0899854862119014e-05,
+      "loss": 6.7062,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 6.2409,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0609579100145138e-05,
+      "loss": 6.8725,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.04644412191582e-05,
+      "loss": 6.6812,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0319303338171265e-05,
+      "loss": 6.7332,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0174165457184327e-05,
+      "loss": 6.7983,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0029027576197386e-05,
+      "loss": 6.8113,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.988388969521045e-05,
+      "loss": 6.8284,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9738751814223513e-05,
+      "loss": 6.7666,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9593613933236575e-05,
+      "loss": 6.6421,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9448476052249637e-05,
+      "loss": 6.826,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.93033381712627e-05,
+      "loss": 6.7973,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9158200290275764e-05,
+      "loss": 6.8295,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9013062409288826e-05,
+      "loss": 6.7402,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 6.8082,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.872278664731495e-05,
+      "loss": 6.6675,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8577648766328012e-05,
+      "loss": 6.6782,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8432510885341077e-05,
+      "loss": 6.7005,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828737300435414e-05,
+      "loss": 6.7035,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8142235123367197e-05,
+      "loss": 6.7503,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7997097242380263e-05,
+      "loss": 6.8567,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7851959361393325e-05,
+      "loss": 6.7818,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7706821480406387e-05,
+      "loss": 6.6334,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.756168359941945e-05,
+      "loss": 6.7261,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 6.5404,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7271407837445576e-05,
+      "loss": 6.6958,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7126269956458638e-05,
+      "loss": 6.4291,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 6.7895,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.683599419448476e-05,
+      "loss": 6.9405,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6690856313497823e-05,
+      "loss": 6.8261,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.654571843251089e-05,
+      "loss": 6.6099,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6400580551523947e-05,
+      "loss": 6.6709,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625544267053701e-05,
+      "loss": 6.7486,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6110304789550074e-05,
+      "loss": 7.022,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5965166908563136e-05,
+      "loss": 6.5623,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5820029027576198e-05,
+      "loss": 6.6926,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.567489114658926e-05,
+      "loss": 6.7017,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5529753265602322e-05,
+      "loss": 6.7289,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 6.6565,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5239477503628446e-05,
+      "loss": 6.7436,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 6.7289,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4949201741654573e-05,
+      "loss": 6.6083,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4804063860667636e-05,
+      "loss": 6.6317,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4658925979680697e-05,
+      "loss": 6.5615,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4513788098693759e-05,
+      "loss": 6.5356,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4368650217706822e-05,
+      "loss": 6.7477,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4223512336719886e-05,
+      "loss": 6.5428,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4078374455732948e-05,
+      "loss": 6.6898,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3933236574746008e-05,
+      "loss": 6.4827,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3788098693759072e-05,
+      "loss": 6.743,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3642960812772135e-05,
+      "loss": 6.7047,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3497822931785197e-05,
+      "loss": 5.9048,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3352685050798259e-05,
+      "loss": 6.7719,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 6.7523,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3062409288824384e-05,
+      "loss": 6.8444,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2917271407837448e-05,
+      "loss": 6.6876,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2772133526850508e-05,
+      "loss": 6.6657,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.262699564586357e-05,
+      "loss": 6.596,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2481857764876634e-05,
+      "loss": 6.9563,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336719883889696e-05,
+      "loss": 6.8935,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219158200290276e-05,
+      "loss": 6.7621,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.204644412191582e-05,
+      "loss": 6.6331,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1901306240928883e-05,
+      "loss": 6.7108,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1756168359941945e-05,
+      "loss": 6.8204,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1611030478955009e-05,
+      "loss": 6.5971,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146589259796807e-05,
+      "loss": 6.6035,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 6.8656,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1175616835994194e-05,
+      "loss": 6.7761,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1030478955007258e-05,
+      "loss": 6.8372,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.088534107402032e-05,
+      "loss": 6.7408,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0740203193033382e-05,
+      "loss": 6.5574,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0595065312046444e-05,
+      "loss": 6.8797,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0449927431059507e-05,
+      "loss": 6.7414,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0304789550072569e-05,
+      "loss": 6.9331,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0159651669085633e-05,
+      "loss": 6.6822,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0014513788098693e-05,
+      "loss": 6.864,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.869375907111757e-06,
+      "loss": 6.9052,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.724238026124818e-06,
+      "loss": 6.6208,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.579100145137882e-06,
+      "loss": 6.719,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 6.6789,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.288824383164006e-06,
+      "loss": 6.7152,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.14368650217707e-06,
+      "loss": 6.6793,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.998548621190131e-06,
+      "loss": 6.5818,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.853410740203193e-06,
+      "loss": 6.5321,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 6.8218,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.563134978229319e-06,
+      "loss": 6.7331,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.41799709724238e-06,
+      "loss": 6.8597,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.272859216255444e-06,
+      "loss": 6.6451,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.127721335268504e-06,
+      "loss": 6.755,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.982583454281568e-06,
+      "loss": 6.7175,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.83744557329463e-06,
+      "loss": 6.6986,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 6.5629,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 6.8473,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402031930333818e-06,
+      "loss": 6.6991,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.256894049346879e-06,
+      "loss": 6.678,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111756168359943e-06,
+      "loss": 6.6967,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966618287373004e-06,
+      "loss": 6.501,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.8214804063860676e-06,
+      "loss": 6.7694,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.6763425253991295e-06,
+      "loss": 6.4776,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.531204644412192e-06,
+      "loss": 6.2739,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.386066763425254e-06,
+      "loss": 6.5628,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.240928882438317e-06,
+      "loss": 6.3233,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09579100145138e-06,
+      "loss": 6.7873,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9506531204644415e-06,
+      "loss": 6.5819,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805515239477504e-06,
+      "loss": 6.8527,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 6.7184,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.515239477503629e-06,
+      "loss": 6.7437,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370101596516691e-06,
+      "loss": 6.9352,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.224963715529754e-06,
+      "loss": 6.6809,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.079825834542816e-06,
+      "loss": 6.6389,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934687953555878e-06,
+      "loss": 6.6878,
+      "step": 1159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.789550072568941e-06,
+      "loss": 6.7242,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644412191582003e-06,
+      "loss": 6.8803,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.499274310595066e-06,
+      "loss": 6.8112,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 6.8707,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.20899854862119e-06,
+      "loss": 6.7205,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.063860667634252e-06,
+      "loss": 6.6868,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.918722786647315e-06,
+      "loss": 6.9362,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 6.5593,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6284470246734397e-06,
+      "loss": 6.6995,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483309143686502e-06,
+      "loss": 6.6244,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3381712626995647e-06,
+      "loss": 6.605,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193033381712627e-06,
+      "loss": 6.7078,
+      "step": 1171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.04789550072569e-06,
+      "loss": 6.6084,
+      "step": 1172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.902757619738752e-06,
+      "loss": 6.7911,
+      "step": 1173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7576197387518145e-06,
+      "loss": 6.8127,
+      "step": 1174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.612481857764877e-06,
+      "loss": 6.5731,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.467343976777939e-06,
+      "loss": 6.5183,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3222060957910015e-06,
+      "loss": 6.6835,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 6.8375,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.031930333817126e-06,
+      "loss": 6.4989,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 6.9027,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.741654571843251e-06,
+      "loss": 6.629,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5965166908563135e-06,
+      "loss": 6.7142,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.451378809869376e-06,
+      "loss": 6.6414,
+      "step": 1183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3062409288824384e-06,
+      "loss": 6.6327,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1611030478955007e-06,
+      "loss": 6.6065,
+      "step": 1185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.015965166908563e-06,
+      "loss": 6.4935,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.708272859216255e-07,
+      "loss": 6.3923,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25689404934688e-07,
+      "loss": 6.3026,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.805515239477504e-07,
+      "loss": 6.3535,
+      "step": 1189
+    },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 7.023049713784804,
-      "train_runtime": 4954.8378,
-      "train_samples_per_second": 5.76,
-      "train_steps_per_second": 0.09
+      "train_loss": 6.874297792115464,
+      "train_runtime": 6595.0456,
+      "train_samples_per_second": 4.327,
+      "train_steps_per_second": 0.18
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,