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": 9.0593,
+      "loss": 9.55,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.6798,
+      "loss": 8.9419,
       "step": 2
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 9.0015,
+      "loss": 9.4113,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6e-07,
-      "loss": 8.9647,
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.3807,
       "step": 4
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 6e-07,
-      "loss": 8.8841,
+      "loss": 9.162,
       "step": 5
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.2e-06,
-      "loss": 8.9194,
+      "loss": 9.1601,
       "step": 6
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 1.8e-06,
-      "loss": 8.7058,
+      "loss": 8.859,
       "step": 7
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 2.4e-06,
-      "loss": 8.4161,
+      "loss": 8.501,
       "step": 8
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 2.9999999999999997e-06,
-      "loss": 8.2294,
+      "loss": 8.9694,
       "step": 9
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 3.6e-06,
-      "loss": 8.3752,
+      "loss": 8.4321,
       "step": 10
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 4.2e-06,
-      "loss": 8.0861,
+      "loss": 8.1065,
       "step": 11
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 4.8e-06,
-      "loss": 8.0248,
+      "loss": 8.1458,
       "step": 12
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 5.399999999999999e-06,
-      "loss": 7.759,
+      "loss": 8.0895,
       "step": 13
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 5.999999999999999e-06,
-      "loss": 7.8552,
+      "loss": 7.9342,
       "step": 14
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 6.599999999999999e-06,
-      "loss": 7.6648,
+      "loss": 7.7382,
       "step": 15
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 7.2e-06,
-      "loss": 7.6878,
+      "loss": 7.67,
       "step": 16
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 7.799999999999998e-06,
-      "loss": 7.5286,
+      "loss": 7.7227,
       "step": 17
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 8.4e-06,
-      "loss": 7.5442,
+      "loss": 7.624,
       "step": 18
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 8.999999999999999e-06,
-      "loss": 7.3928,
+      "loss": 7.4774,
       "step": 19
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 9.6e-06,
-      "loss": 7.4898,
+      "loss": 7.5997,
       "step": 20
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 1.02e-05,
-      "loss": 7.4324,
+      "loss": 7.4215,
       "step": 21
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 1.0799999999999998e-05,
-      "loss": 7.4232,
+      "loss": 7.5155,
       "step": 22
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 1.14e-05,
-      "loss": 7.3906,
+      "loss": 7.4231,
       "step": 23
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.02,
       "learning_rate": 1.1999999999999999e-05,
-      "loss": 7.3289,
+      "loss": 7.3793,
       "step": 24
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 1.26e-05,
-      "loss": 7.2432,
+      "loss": 7.2675,
       "step": 25
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 1.3199999999999997e-05,
-      "loss": 7.378,
+      "loss": 7.4587,
       "step": 26
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 1.3799999999999998e-05,
-      "loss": 7.2846,
+      "loss": 7.2464,
       "step": 27
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.02,
       "learning_rate": 1.44e-05,
-      "loss": 7.1664,
+      "loss": 7.1883,
       "step": 28
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.02,
       "learning_rate": 1.4999999999999999e-05,
-      "loss": 7.2449,
+      "loss": 7.3409,
       "step": 29
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 1.5599999999999996e-05,
-      "loss": 7.2556,
+      "loss": 7.3998,
       "step": 30
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 1.6199999999999997e-05,
-      "loss": 7.2762,
+      "loss": 7.2939,
       "step": 31
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 1.68e-05,
-      "loss": 7.2141,
+      "loss": 7.2231,
       "step": 32
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.03,
       "learning_rate": 1.74e-05,
-      "loss": 7.2166,
+      "loss": 7.4439,
       "step": 33
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 1.7999999999999997e-05,
-      "loss": 7.1265,
+      "loss": 7.084,
       "step": 34
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 1.8599999999999998e-05,
-      "loss": 7.1288,
+      "loss": 7.1616,
       "step": 35
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 1.92e-05,
-      "loss": 7.1726,
+      "loss": 7.3887,
       "step": 36
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.03,
       "learning_rate": 1.98e-05,
-      "loss": 7.2633,
+      "loss": 7.2686,
       "step": 37
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 2.04e-05,
-      "loss": 7.1457,
+      "loss": 7.2506,
       "step": 38
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 2.1e-05,
-      "loss": 7.2118,
+      "loss": 7.3682,
       "step": 39
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 2.1599999999999996e-05,
-      "loss": 7.1554,
+      "loss": 7.1921,
       "step": 40
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.03,
       "learning_rate": 2.2199999999999998e-05,
-      "loss": 7.2002,
+      "loss": 7.1297,
       "step": 41
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.04,
       "learning_rate": 2.28e-05,
-      "loss": 7.2066,
+      "loss": 7.2948,
       "step": 42
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 2.34e-05,
-      "loss": 6.9742,
+      "loss": 7.1493,
       "step": 43
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 2.3999999999999997e-05,
-      "loss": 7.1299,
+      "loss": 7.2048,
       "step": 44
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 2.4599999999999998e-05,
-      "loss": 7.1903,
+      "loss": 7.3067,
       "step": 45
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.04,
       "learning_rate": 2.52e-05,
-      "loss": 7.1265,
+      "loss": 7.3014,
       "step": 46
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 2.5799999999999997e-05,
-      "loss": 7.0296,
+      "loss": 7.1704,
       "step": 47
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 2.6399999999999995e-05,
-      "loss": 7.0528,
+      "loss": 7.2711,
       "step": 48
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 2.6999999999999996e-05,
-      "loss": 6.7974,
+      "loss": 7.0167,
       "step": 49
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 2.7599999999999997e-05,
-      "loss": 6.5233,
+      "loss": 6.8768,
       "step": 50
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.04,
       "learning_rate": 2.8199999999999998e-05,
-      "loss": 7.3589,
+      "loss": 7.4694,
       "step": 51
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 2.88e-05,
-      "loss": 7.2445,
+      "loss": 7.231,
       "step": 52
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.04,
       "learning_rate": 2.94e-05,
-      "loss": 7.2458,
+      "loss": 7.1152,
       "step": 53
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 2.9999999999999997e-05,
-      "loss": 7.1335,
+      "loss": 7.1691,
       "step": 54
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.05,
       "learning_rate": 3.06e-05,
-      "loss": 6.976,
+      "loss": 7.2726,
       "step": 55
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 3.119999999999999e-05,
-      "loss": 7.0973,
+      "loss": 6.9637,
       "step": 56
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 3.1799999999999994e-05,
-      "loss": 7.1603,
+      "loss": 7.391,
       "step": 57
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 3.2399999999999995e-05,
-      "loss": 7.228,
+      "loss": 7.036,
       "step": 58
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 3.2999999999999996e-05,
-      "loss": 7.081,
+      "loss": 7.0183,
       "step": 59
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.05,
       "learning_rate": 3.36e-05,
-      "loss": 7.1042,
+      "loss": 7.1747,
       "step": 60
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 3.42e-05,
-      "loss": 7.0489,
+      "loss": 6.999,
       "step": 61
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 3.48e-05,
-      "loss": 7.0797,
+      "loss": 7.0586,
       "step": 62
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 3.539999999999999e-05,
-      "loss": 7.0067,
+      "loss": 7.1794,
       "step": 63
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.05,
       "learning_rate": 3.5999999999999994e-05,
-      "loss": 7.1171,
+      "loss": 7.0978,
       "step": 64
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.05,
       "learning_rate": 3.6599999999999995e-05,
-      "loss": 7.0183,
+      "loss": 7.043,
       "step": 65
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 3.7199999999999996e-05,
-      "loss": 6.987,
+      "loss": 7.1155,
       "step": 66
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 3.78e-05,
-      "loss": 7.0242,
+      "loss": 7.0924,
       "step": 67
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 3.84e-05,
-      "loss": 7.0413,
+      "loss": 7.0018,
       "step": 68
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.06,
       "learning_rate": 3.9e-05,
-      "loss": 7.0896,
+      "loss": 6.9802,
       "step": 69
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 3.96e-05,
-      "loss": 6.9828,
+      "loss": 7.0259,
       "step": 70
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 4.02e-05,
-      "loss": 7.1095,
+      "loss": 7.1772,
       "step": 71
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 4.08e-05,
-      "loss": 7.0894,
+      "loss": 7.2527,
       "step": 72
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.06,
       "learning_rate": 4.14e-05,
-      "loss": 7.0473,
+      "loss": 7.086,
       "step": 73
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 4.2e-05,
-      "loss": 7.0871,
+      "loss": 7.131,
       "step": 74
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 4.259999999999999e-05,
-      "loss": 7.0546,
+      "loss": 6.9208,
       "step": 75
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 4.319999999999999e-05,
-      "loss": 7.081,
+      "loss": 7.2732,
       "step": 76
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.06,
       "learning_rate": 4.3799999999999994e-05,
-      "loss": 7.0875,
+      "loss": 7.0671,
       "step": 77
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.07,
       "learning_rate": 4.4399999999999995e-05,
-      "loss": 7.0912,
+      "loss": 6.8986,
       "step": 78
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 4.4999999999999996e-05,
-      "loss": 6.971,
+      "loss": 6.9168,
       "step": 79
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 4.56e-05,
-      "loss": 7.0758,
+      "loss": 7.0245,
       "step": 80
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 4.62e-05,
-      "loss": 7.0612,
+      "loss": 7.1175,
       "step": 81
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.07,
       "learning_rate": 4.68e-05,
-      "loss": 7.0599,
+      "loss": 7.1879,
       "step": 82
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 4.7399999999999993e-05,
-      "loss": 7.053,
+      "loss": 6.9614,
       "step": 83
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 4.7999999999999994e-05,
-      "loss": 7.0249,
+      "loss": 7.0797,
       "step": 84
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 4.8599999999999995e-05,
-      "loss": 7.0136,
+      "loss": 7.0003,
       "step": 85
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.07,
       "learning_rate": 4.9199999999999997e-05,
-      "loss": 7.0577,
+      "loss": 7.0737,
       "step": 86
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 4.98e-05,
-      "loss": 7.0064,
+      "loss": 6.982,
       "step": 87
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 5.04e-05,
-      "loss": 6.9519,
+      "loss": 7.0724,
       "step": 88
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.07,
       "learning_rate": 5.1e-05,
-      "loss": 6.9031,
+      "loss": 7.1836,
       "step": 89
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 5.1599999999999994e-05,
-      "loss": 7.039,
+      "loss": 7.006,
       "step": 90
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.08,
       "learning_rate": 5.2199999999999995e-05,
-      "loss": 6.9379,
+      "loss": 7.2244,
       "step": 91
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 5.279999999999999e-05,
-      "loss": 7.1254,
+      "loss": 7.0901,
       "step": 92
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 5.339999999999999e-05,
-      "loss": 7.0077,
+      "loss": 6.7901,
       "step": 93
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 5.399999999999999e-05,
-      "loss": 6.8346,
+      "loss": 7.0759,
       "step": 94
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.08,
       "learning_rate": 5.459999999999999e-05,
-      "loss": 6.7611,
+      "loss": 7.218,
       "step": 95
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 5.519999999999999e-05,
-      "loss": 6.982,
+      "loss": 6.856,
       "step": 96
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 5.5799999999999994e-05,
-      "loss": 6.7893,
+      "loss": 7.0121,
       "step": 97
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 5.6399999999999995e-05,
-      "loss": 6.5833,
+      "loss": 6.9158,
       "step": 98
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 6.6579,
+      "loss": 6.587,
       "step": 99
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.08,
       "learning_rate": 5.76e-05,
-      "loss": 6.2381,
+      "loss": 6.3765,
       "step": 100
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.08,
       "learning_rate": 5.82e-05,
-      "loss": 7.1487,
+      "loss": 7.1251,
       "step": 101
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 5.88e-05,
-      "loss": 7.1538,
+      "loss": 7.0595,
       "step": 102
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 5.94e-05,
-      "loss": 6.9778,
+      "loss": 7.1093,
       "step": 103
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.09,
       "learning_rate": 5.9999999999999995e-05,
-      "loss": 6.9727,
+      "loss": 7.0153,
       "step": 104
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 6.0599999999999996e-05,
-      "loss": 6.9908,
+      "loss": 6.9467,
       "step": 105
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 6.12e-05,
-      "loss": 6.9922,
+      "loss": 7.0941,
       "step": 106
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 6.18e-05,
-      "loss": 6.9424,
+      "loss": 6.9961,
       "step": 107
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 6.239999999999999e-05,
-      "loss": 6.9689,
+      "loss": 7.0825,
       "step": 108
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.09,
       "learning_rate": 6.299999999999999e-05,
-      "loss": 6.8709,
+      "loss": 7.1876,
       "step": 109
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 6.359999999999999e-05,
-      "loss": 6.9555,
+      "loss": 7.0441,
       "step": 110
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 6.419999999999999e-05,
-      "loss": 6.8452,
+      "loss": 7.1209,
       "step": 111
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 6.479999999999999e-05,
-      "loss": 6.8586,
+      "loss": 7.0938,
       "step": 112
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.09,
       "learning_rate": 6.539999999999999e-05,
-      "loss": 6.8972,
+      "loss": 6.98,
       "step": 113
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 6.599999999999999e-05,
-      "loss": 6.9495,
+      "loss": 6.9344,
       "step": 114
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 6.659999999999999e-05,
-      "loss": 6.8688,
+      "loss": 7.0758,
       "step": 115
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 6.72e-05,
-      "loss": 7.0212,
+      "loss": 7.1522,
       "step": 116
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 6.78e-05,
-      "loss": 6.9658,
+      "loss": 7.0122,
       "step": 117
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.1,
       "learning_rate": 6.84e-05,
-      "loss": 6.9449,
+      "loss": 6.8646,
       "step": 118
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 6.9e-05,
-      "loss": 6.8678,
+      "loss": 6.7966,
       "step": 119
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 6.96e-05,
-      "loss": 6.9486,
+      "loss": 6.9892,
       "step": 120
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 7.02e-05,
-      "loss": 6.8725,
+      "loss": 7.0826,
       "step": 121
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.1,
       "learning_rate": 7.079999999999999e-05,
-      "loss": 6.8252,
+      "loss": 6.9133,
       "step": 122
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 7.139999999999999e-05,
-      "loss": 6.8825,
+      "loss": 7.0459,
       "step": 123
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.1,
       "learning_rate": 7.199999999999999e-05,
-      "loss": 6.8312,
+      "loss": 6.9899,
       "step": 124
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 7.259999999999999e-05,
-      "loss": 6.9445,
+      "loss": 6.9376,
       "step": 125
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 7.319999999999999e-05,
-      "loss": 6.9267,
+      "loss": 6.8284,
       "step": 126
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.11,
       "learning_rate": 7.379999999999999e-05,
-      "loss": 6.91,
+      "loss": 7.1698,
       "step": 127
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 7.439999999999999e-05,
-      "loss": 6.8933,
+      "loss": 6.9294,
       "step": 128
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 7.5e-05,
-      "loss": 6.9331,
+      "loss": 6.9303,
       "step": 129
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 7.56e-05,
-      "loss": 6.998,
+      "loss": 6.8968,
       "step": 130
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.11,
       "learning_rate": 7.62e-05,
-      "loss": 6.7967,
+      "loss": 6.7854,
       "step": 131
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 7.68e-05,
-      "loss": 6.9952,
+      "loss": 6.9444,
       "step": 132
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 7.74e-05,
-      "loss": 6.844,
+      "loss": 7.1272,
       "step": 133
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 7.8e-05,
-      "loss": 6.8629,
+      "loss": 6.9042,
       "step": 134
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 7.86e-05,
-      "loss": 6.9994,
+      "loss": 6.9895,
       "step": 135
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.11,
       "learning_rate": 7.92e-05,
-      "loss": 6.7906,
+      "loss": 6.9246,
       "step": 136
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 7.98e-05,
-      "loss": 6.9598,
+      "loss": 6.9404,
       "step": 137
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 8.04e-05,
-      "loss": 6.8899,
+      "loss": 6.9605,
       "step": 138
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 8.1e-05,
-      "loss": 6.8555,
+      "loss": 6.9209,
       "step": 139
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.12,
       "learning_rate": 8.16e-05,
-      "loss": 6.9729,
+      "loss": 7.0834,
       "step": 140
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 8.22e-05,
-      "loss": 6.9761,
+      "loss": 6.9762,
       "step": 141
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 8.28e-05,
-      "loss": 6.9451,
+      "loss": 6.8668,
       "step": 142
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 8.34e-05,
-      "loss": 6.7803,
+      "loss": 6.8323,
       "step": 143
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.12,
       "learning_rate": 8.4e-05,
-      "loss": 6.9795,
+      "loss": 6.9498,
       "step": 144
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 8.459999999999998e-05,
-      "loss": 6.7497,
+      "loss": 6.7789,
       "step": 145
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 8.519999999999998e-05,
-      "loss": 6.7688,
+      "loss": 6.9855,
       "step": 146
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 8.579999999999998e-05,
-      "loss": 6.8215,
+      "loss": 6.7275,
       "step": 147
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.12,
       "learning_rate": 8.639999999999999e-05,
-      "loss": 6.8295,
+      "loss": 6.418,
       "step": 148
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.13,
       "learning_rate": 8.699999999999999e-05,
-      "loss": 6.5939,
+      "loss": 9.7764,
       "step": 149
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 8.759999999999999e-05,
-      "loss": 6.4406,
+      "loss": 6.4339,
       "step": 150
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 8.819999999999999e-05,
-      "loss": 7.0335,
+      "loss": 7.2143,
       "step": 151
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 8.879999999999999e-05,
-      "loss": 7.0794,
+      "loss": 7.2603,
       "step": 152
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.13,
       "learning_rate": 8.939999999999999e-05,
-      "loss": 6.962,
+      "loss": 7.0635,
       "step": 153
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 8.999999999999999e-05,
-      "loss": 6.8484,
+      "loss": 7.1448,
       "step": 154
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 9.059999999999999e-05,
-      "loss": 6.8841,
+      "loss": 6.9462,
       "step": 155
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 9.12e-05,
-      "loss": 6.8999,
+      "loss": 7.0349,
       "step": 156
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 9.18e-05,
-      "loss": 6.9885,
+      "loss": 7.0018,
       "step": 157
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.13,
       "learning_rate": 9.24e-05,
-      "loss": 6.9219,
+      "loss": 7.1447,
       "step": 158
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 9.3e-05,
-      "loss": 6.91,
+      "loss": 6.9218,
       "step": 159
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.13,
       "learning_rate": 9.36e-05,
-      "loss": 6.9633,
+      "loss": 6.8224,
       "step": 160
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 9.419999999999999e-05,
-      "loss": 6.9086,
+      "loss": 6.9101,
       "step": 161
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.14,
       "learning_rate": 9.479999999999999e-05,
-      "loss": 6.967,
+      "loss": 7.0359,
       "step": 162
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 9.539999999999999e-05,
-      "loss": 6.9233,
+      "loss": 7.0907,
       "step": 163
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 9.599999999999999e-05,
-      "loss": 6.9295,
+      "loss": 6.9293,
       "step": 164
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 9.659999999999999e-05,
-      "loss": 6.8712,
+      "loss": 6.9584,
       "step": 165
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 9.719999999999999e-05,
-      "loss": 6.9343,
+      "loss": 6.7897,
       "step": 166
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.14,
       "learning_rate": 9.779999999999999e-05,
-      "loss": 6.8514,
+      "loss": 6.9411,
       "step": 167
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 9.839999999999999e-05,
-      "loss": 6.7933,
+      "loss": 6.9745,
       "step": 168
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 9.9e-05,
-      "loss": 6.8052,
+      "loss": 6.9907,
       "step": 169
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 9.96e-05,
-      "loss": 6.9495,
+      "loss": 6.8991,
       "step": 170
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.14,
       "learning_rate": 0.0001002,
-      "loss": 6.8916,
+      "loss": 6.9265,
       "step": 171
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.14,
       "learning_rate": 0.0001008,
-      "loss": 6.9415,
+      "loss": 6.9174,
       "step": 172
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 0.0001014,
-      "loss": 6.8972,
+      "loss": 6.7028,
       "step": 173
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 0.000102,
-      "loss": 6.9499,
+      "loss": 6.976,
       "step": 174
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 0.0001026,
-      "loss": 6.9356,
+      "loss": 6.9988,
       "step": 175
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.15,
       "learning_rate": 0.00010319999999999999,
-      "loss": 6.8336,
+      "loss": 7.1287,
       "step": 176
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 0.00010379999999999999,
-      "loss": 6.8248,
+      "loss": 7.0902,
       "step": 177
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 0.00010439999999999999,
-      "loss": 6.8928,
+      "loss": 7.0567,
       "step": 178
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 0.00010499999999999999,
-      "loss": 6.9024,
+      "loss": 6.8533,
       "step": 179
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.15,
       "learning_rate": 0.00010559999999999998,
-      "loss": 6.9626,
+      "loss": 6.9569,
       "step": 180
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 0.00010619999999999998,
-      "loss": 6.9259,
+      "loss": 7.0492,
       "step": 181
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 0.00010679999999999998,
-      "loss": 6.8326,
+      "loss": 7.1597,
       "step": 182
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 0.00010739999999999998,
-      "loss": 6.8831,
+      "loss": 6.8855,
       "step": 183
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.15,
       "learning_rate": 0.00010799999999999998,
-      "loss": 6.97,
+      "loss": 6.9183,
       "step": 184
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.16,
       "learning_rate": 0.00010859999999999998,
-      "loss": 6.8943,
+      "loss": 6.7531,
       "step": 185
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 0.00010919999999999998,
-      "loss": 6.7704,
+      "loss": 6.8515,
       "step": 186
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 0.00010979999999999999,
-      "loss": 6.8563,
+      "loss": 7.0631,
       "step": 187
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 0.00011039999999999999,
-      "loss": 6.8277,
+      "loss": 6.8417,
       "step": 188
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.16,
       "learning_rate": 0.00011099999999999999,
-      "loss": 6.919,
+      "loss": 6.8889,
       "step": 189
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 0.00011159999999999999,
-      "loss": 6.9639,
+      "loss": 6.859,
       "step": 190
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 0.00011219999999999999,
-      "loss": 6.9247,
+      "loss": 7.0676,
       "step": 191
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 0.00011279999999999999,
-      "loss": 6.8216,
+      "loss": 7.0038,
       "step": 192
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 0.00011339999999999999,
-      "loss": 6.8592,
+      "loss": 6.8311,
       "step": 193
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.16,
       "learning_rate": 0.00011399999999999999,
-      "loss": 6.7915,
+      "loss": 6.8485,
       "step": 194
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 0.0001146,
-      "loss": 6.7807,
+      "loss": 6.7633,
       "step": 195
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.16,
       "learning_rate": 0.0001152,
-      "loss": 6.6464,
+      "loss": 6.6829,
       "step": 196
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 0.0001158,
-      "loss": 6.62,
+      "loss": 6.9826,
       "step": 197
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.17,
       "learning_rate": 0.0001164,
-      "loss": 6.6885,
+      "loss": 6.747,
       "step": 198
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 0.000117,
-      "loss": 6.5504,
+      "loss": 6.6231,
       "step": 199
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 0.0001176,
-      "loss": 6.3884,
+      "loss": 6.2205,
       "step": 200
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 0.0001182,
-      "loss": 7.2501,
+      "loss": 7.0516,
       "step": 201
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.17,
       "learning_rate": 0.0001188,
-      "loss": 7.1882,
+      "loss": 6.9791,
       "step": 202
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 0.0001194,
-      "loss": 7.0557,
+      "loss": 7.0014,
       "step": 203
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 0.00011999999999999999,
-      "loss": 6.9325,
+      "loss": 6.8763,
       "step": 204
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 0.00012059999999999999,
-      "loss": 7.0032,
+      "loss": 6.8784,
       "step": 205
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 0.00012119999999999999,
-      "loss": 6.9799,
+      "loss": 6.9652,
       "step": 206
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.17,
       "learning_rate": 0.00012179999999999999,
-      "loss": 6.8919,
+      "loss": 7.0076,
       "step": 207
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.17,
       "learning_rate": 0.0001224,
-      "loss": 6.9957,
+      "loss": 7.1589,
       "step": 208
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 0.00012299999999999998,
-      "loss": 6.9433,
+      "loss": 6.9716,
       "step": 209
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 0.0001236,
-      "loss": 6.9953,
+      "loss": 7.1503,
       "step": 210
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.18,
       "learning_rate": 0.00012419999999999998,
-      "loss": 6.855,
+      "loss": 6.8821,
       "step": 211
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 0.00012479999999999997,
-      "loss": 6.9195,
+      "loss": 6.8873,
       "step": 212
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 0.00012539999999999999,
-      "loss": 6.794,
+      "loss": 6.9434,
       "step": 213
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 0.00012599999999999997,
-      "loss": 6.8751,
+      "loss": 6.937,
       "step": 214
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 0.0001266,
-      "loss": 6.8524,
+      "loss": 6.9549,
       "step": 215
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.18,
       "learning_rate": 0.00012719999999999997,
-      "loss": 6.8992,
+      "loss": 6.9406,
       "step": 216
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 0.0001278,
-      "loss": 6.8588,
+      "loss": 6.9666,
       "step": 217
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 0.00012839999999999998,
-      "loss": 6.881,
+      "loss": 6.9652,
       "step": 218
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 0.000129,
-      "loss": 6.9031,
+      "loss": 6.8692,
       "step": 219
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.18,
       "learning_rate": 0.00012959999999999998,
-      "loss": 6.9355,
+      "loss": 7.0758,
       "step": 220
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 0.0001302,
-      "loss": 6.9106,
+      "loss": 6.8795,
       "step": 221
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 0.00013079999999999998,
-      "loss": 6.9313,
+      "loss": 7.2127,
       "step": 222
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 0.0001314,
-      "loss": 6.8701,
+      "loss": 7.0779,
       "step": 223
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 0.00013199999999999998,
-      "loss": 6.8505,
+      "loss": 7.0581,
       "step": 224
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.19,
       "learning_rate": 0.0001326,
-      "loss": 6.9826,
+      "loss": 6.8529,
       "step": 225
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 0.00013319999999999999,
-      "loss": 6.8822,
+      "loss": 6.7646,
       "step": 226
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 0.0001338,
-      "loss": 6.992,
+      "loss": 7.0354,
       "step": 227
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 0.0001344,
-      "loss": 6.9271,
+      "loss": 7.023,
       "step": 228
     },
     {
-      "epoch": 0.51,
+      "epoch": 0.19,
       "learning_rate": 0.000135,
-      "loss": 6.8334,
+      "loss": 6.8544,
       "step": 229
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 0.0001356,
-      "loss": 6.9154,
+      "loss": 6.9987,
       "step": 230
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.19,
       "learning_rate": 0.0001362,
-      "loss": 6.8884,
+      "loss": 7.0755,
       "step": 231
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 0.0001368,
-      "loss": 6.7658,
+      "loss": 6.975,
       "step": 232
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 0.0001374,
-      "loss": 6.8042,
+      "loss": 6.9598,
       "step": 233
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.2,
       "learning_rate": 0.000138,
-      "loss": 6.8418,
+      "loss": 6.8081,
       "step": 234
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 0.0001386,
-      "loss": 6.9673,
+      "loss": 7.0254,
       "step": 235
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 0.0001392,
-      "loss": 6.8433,
+      "loss": 6.9681,
       "step": 236
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 0.00013979999999999998,
-      "loss": 6.9966,
+      "loss": 6.996,
       "step": 237
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.2,
       "learning_rate": 0.0001404,
-      "loss": 6.8619,
+      "loss": 6.9012,
       "step": 238
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 0.00014099999999999998,
-      "loss": 6.7954,
+      "loss": 7.0205,
       "step": 239
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 0.00014159999999999997,
-      "loss": 7.0012,
+      "loss": 6.7686,
       "step": 240
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 0.0001422,
-      "loss": 6.8357,
+      "loss": 6.9025,
       "step": 241
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 0.00014279999999999997,
-      "loss": 6.8615,
+      "loss": 6.8842,
       "step": 242
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.2,
       "learning_rate": 0.0001434,
-      "loss": 7.4713,
+      "loss": 7.0146,
       "step": 243
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 0.00014399999999999998,
-      "loss": 6.7664,
+      "loss": 7.0304,
       "step": 244
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 0.0001446,
-      "loss": 6.7984,
+      "loss": 6.763,
       "step": 245
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 0.00014519999999999998,
-      "loss": 6.7477,
+      "loss": 6.7243,
       "step": 246
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.21,
       "learning_rate": 0.0001458,
-      "loss": 6.5329,
+      "loss": 7.2358,
       "step": 247
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 0.00014639999999999998,
-      "loss": 6.754,
+      "loss": 6.6169,
       "step": 248
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 0.000147,
-      "loss": 6.6694,
+      "loss": 6.5757,
       "step": 249
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 0.00014759999999999998,
-      "loss": 6.333,
+      "loss": 6.2448,
       "step": 250
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.21,
       "learning_rate": 0.0001482,
-      "loss": 7.1766,
+      "loss": 7.0981,
       "step": 251
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 0.00014879999999999998,
-      "loss": 7.1344,
+      "loss": 6.8921,
       "step": 252
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 0.0001494,
-      "loss": 7.0821,
+      "loss": 7.1193,
       "step": 253
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 0.00015,
-      "loss": 7.0323,
+      "loss": 6.8786,
       "step": 254
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.21,
       "learning_rate": 0.00015059999999999997,
-      "loss": 7.0408,
+      "loss": 6.9321,
       "step": 255
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.22,
       "learning_rate": 0.0001512,
-      "loss": 6.9713,
+      "loss": 7.0229,
       "step": 256
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 0.00015179999999999998,
-      "loss": 6.9548,
+      "loss": 6.8696,
       "step": 257
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 0.0001524,
-      "loss": 6.9471,
+      "loss": 7.0135,
       "step": 258
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 0.00015299999999999998,
-      "loss": 6.8788,
+      "loss": 6.9056,
       "step": 259
     },
     {
-      "epoch": 0.58,
+      "epoch": 0.22,
       "learning_rate": 0.0001536,
-      "loss": 6.9004,
+      "loss": 6.8585,
       "step": 260
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 0.00015419999999999998,
-      "loss": 6.9344,
+      "loss": 6.9343,
       "step": 261
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 0.0001548,
-      "loss": 6.8965,
+      "loss": 6.822,
       "step": 262
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 0.00015539999999999998,
-      "loss": 6.9719,
+      "loss": 6.8312,
       "step": 263
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 0.000156,
-      "loss": 6.9434,
+      "loss": 6.9712,
       "step": 264
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.22,
       "learning_rate": 0.00015659999999999998,
-      "loss": 6.9478,
+      "loss": 6.8902,
       "step": 265
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 0.0001572,
-      "loss": 6.761,
+      "loss": 6.9703,
       "step": 266
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.22,
       "learning_rate": 0.0001578,
-      "loss": 6.9736,
+      "loss": 6.8423,
       "step": 267
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 0.0001584,
-      "loss": 6.7987,
+      "loss": 6.8521,
       "step": 268
     },
     {
-      "epoch": 0.6,
+      "epoch": 0.23,
       "learning_rate": 0.000159,
-      "loss": 6.9504,
+      "loss": 6.9612,
       "step": 269
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 0.0001596,
-      "loss": 6.9853,
+      "loss": 6.9591,
       "step": 270
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 0.0001602,
-      "loss": 6.8711,
+      "loss": 7.024,
       "step": 271
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 0.0001608,
-      "loss": 6.864,
+      "loss": 6.8473,
       "step": 272
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 0.0001614,
-      "loss": 6.779,
+      "loss": 7.0068,
       "step": 273
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.23,
       "learning_rate": 0.000162,
-      "loss": 6.7297,
+      "loss": 7.0259,
       "step": 274
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 0.0001626,
-      "loss": 6.968,
+      "loss": 7.0017,
       "step": 275
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 0.0001632,
-      "loss": 6.8534,
+      "loss": 6.7601,
       "step": 276
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 0.0001638,
-      "loss": 6.8285,
+      "loss": 6.8875,
       "step": 277
     },
     {
-      "epoch": 0.62,
+      "epoch": 0.23,
       "learning_rate": 0.0001644,
-      "loss": 6.9105,
+      "loss": 6.9374,
       "step": 278
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.23,
       "learning_rate": 0.000165,
-      "loss": 6.9451,
+      "loss": 7.0433,
       "step": 279
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 0.0001656,
-      "loss": 6.9857,
+      "loss": 6.8445,
       "step": 280
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 0.0001662,
-      "loss": 6.9182,
+      "loss": 6.79,
       "step": 281
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 0.0001668,
-      "loss": 6.852,
+      "loss": 6.9064,
       "step": 282
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.24,
       "learning_rate": 0.0001674,
-      "loss": 6.8052,
+      "loss": 7.0482,
       "step": 283
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 0.000168,
-      "loss": 6.856,
+      "loss": 7.016,
       "step": 284
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 0.0001686,
-      "loss": 6.871,
+      "loss": 6.9143,
       "step": 285
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 0.00016919999999999997,
-      "loss": 6.95,
+      "loss": 6.8954,
       "step": 286
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.24,
       "learning_rate": 0.00016979999999999998,
-      "loss": 6.907,
+      "loss": 6.8511,
       "step": 287
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 0.00017039999999999997,
-      "loss": 6.8627,
+      "loss": 6.8074,
       "step": 288
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 0.00017099999999999998,
-      "loss": 6.8602,
+      "loss": 6.9775,
       "step": 289
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 0.00017159999999999997,
-      "loss": 6.9221,
+      "loss": 6.8545,
       "step": 290
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.24,
       "learning_rate": 0.00017219999999999998,
-      "loss": 6.9727,
+      "loss": 7.0779,
       "step": 291
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.25,
       "learning_rate": 0.00017279999999999997,
-      "loss": 6.9088,
+      "loss": 6.9559,
       "step": 292
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 0.00017339999999999996,
-      "loss": 6.7969,
+      "loss": 6.6196,
       "step": 293
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 0.00017399999999999997,
-      "loss": 6.7438,
+      "loss": 6.6656,
       "step": 294
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 0.00017459999999999996,
-      "loss": 6.7324,
+      "loss": 6.8798,
       "step": 295
     },
     {
-      "epoch": 0.66,
+      "epoch": 0.25,
       "learning_rate": 0.00017519999999999998,
-      "loss": 6.732,
+      "loss": 7.0263,
       "step": 296
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 0.00017579999999999996,
-      "loss": 6.7565,
+      "loss": 6.5448,
       "step": 297
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 0.00017639999999999998,
-      "loss": 6.7671,
+      "loss": 6.6762,
       "step": 298
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 0.00017699999999999997,
-      "loss": 6.3662,
+      "loss": 6.8317,
       "step": 299
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 0.00017759999999999998,
-      "loss": 6.4025,
+      "loss": 6.4708,
       "step": 300
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.25,
       "learning_rate": 0.00017819999999999997,
-      "loss": 7.0981,
+      "loss": 6.9424,
       "step": 301
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 0.00017879999999999998,
-      "loss": 7.0077,
+      "loss": 7.038,
       "step": 302
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.25,
       "learning_rate": 0.00017939999999999997,
-      "loss": 6.9916,
+      "loss": 7.1493,
       "step": 303
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 0.00017999999999999998,
-      "loss": 6.8943,
+      "loss": 6.8401,
       "step": 304
     },
     {
-      "epoch": 0.68,
+      "epoch": 0.26,
       "learning_rate": 0.00018059999999999997,
-      "loss": 6.9923,
+      "loss": 6.9445,
       "step": 305
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 0.00018119999999999999,
-      "loss": 6.8215,
+      "loss": 6.8652,
       "step": 306
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 0.00018179999999999997,
-      "loss": 6.8283,
+      "loss": 6.9303,
       "step": 307
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 0.0001824,
-      "loss": 6.9746,
+      "loss": 6.8556,
       "step": 308
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.26,
       "learning_rate": 0.00018299999999999998,
-      "loss": 6.8177,
+      "loss": 6.7537,
       "step": 309
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 0.0001836,
-      "loss": 6.9226,
+      "loss": 7.0273,
       "step": 310
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 0.00018419999999999998,
-      "loss": 6.9195,
+      "loss": 6.7255,
       "step": 311
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 0.0001848,
-      "loss": 6.9117,
+      "loss": 6.7813,
       "step": 312
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 0.00018539999999999998,
-      "loss": 6.8309,
+      "loss": 6.8344,
       "step": 313
     },
     {
-      "epoch": 0.7,
+      "epoch": 0.26,
       "learning_rate": 0.000186,
-      "loss": 6.8095,
+      "loss": 6.9307,
       "step": 314
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.26,
       "learning_rate": 0.00018659999999999998,
-      "loss": 6.942,
+      "loss": 6.8175,
       "step": 315
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 0.0001872,
-      "loss": 6.8752,
+      "loss": 7.0431,
       "step": 316
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 0.00018779999999999998,
-      "loss": 6.8158,
+      "loss": 6.9237,
       "step": 317
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.27,
       "learning_rate": 0.00018839999999999997,
-      "loss": 6.7456,
+      "loss": 6.951,
       "step": 318
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 0.00018899999999999999,
-      "loss": 6.8388,
+      "loss": 6.8204,
       "step": 319
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 0.00018959999999999997,
-      "loss": 6.8635,
+      "loss": 6.7867,
       "step": 320
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 0.0001902,
-      "loss": 6.8421,
+      "loss": 6.832,
       "step": 321
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 0.00019079999999999998,
-      "loss": 6.8038,
+      "loss": 6.7993,
       "step": 322
     },
     {
-      "epoch": 0.72,
+      "epoch": 0.27,
       "learning_rate": 0.0001914,
-      "loss": 6.9022,
+      "loss": 6.7834,
       "step": 323
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 0.00019199999999999998,
-      "loss": 6.8039,
+      "loss": 6.8469,
       "step": 324
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 0.0001926,
-      "loss": 6.88,
+      "loss": 6.7455,
       "step": 325
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 0.00019319999999999998,
-      "loss": 6.9172,
+      "loss": 6.9278,
       "step": 326
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.27,
       "learning_rate": 0.0001938,
-      "loss": 6.8517,
+      "loss": 6.8588,
       "step": 327
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 0.00019439999999999998,
-      "loss": 6.8168,
+      "loss": 6.9268,
       "step": 328
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 0.000195,
-      "loss": 6.8759,
+      "loss": 6.8046,
       "step": 329
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 0.00019559999999999998,
-      "loss": 6.9028,
+      "loss": 7.0146,
       "step": 330
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 0.0001962,
-      "loss": 6.8595,
+      "loss": 6.88,
       "step": 331
     },
     {
-      "epoch": 0.74,
+      "epoch": 0.28,
       "learning_rate": 0.00019679999999999999,
-      "loss": 6.7604,
+      "loss": 6.9614,
       "step": 332
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 0.0001974,
-      "loss": 6.9378,
+      "loss": 7.0782,
       "step": 333
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 0.000198,
-      "loss": 6.8531,
+      "loss": 6.8374,
       "step": 334
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 0.0001986,
-      "loss": 6.9461,
+      "loss": 6.873,
       "step": 335
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.28,
       "learning_rate": 0.0001992,
-      "loss": 6.8454,
+      "loss": 6.9961,
       "step": 336
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 0.0001998,
-      "loss": 6.7238,
+      "loss": 6.7669,
       "step": 337
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 0.0002004,
-      "loss": 6.871,
+      "loss": 7.0401,
       "step": 338
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.28,
       "learning_rate": 0.000201,
-      "loss": 6.8716,
+      "loss": 6.8328,
       "step": 339
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 0.0002016,
-      "loss": 6.8667,
+      "loss": 6.9227,
       "step": 340
     },
     {
-      "epoch": 0.76,
+      "epoch": 0.29,
       "learning_rate": 0.0002022,
-      "loss": 6.7651,
+      "loss": 6.7859,
       "step": 341
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 0.0002028,
-      "loss": 6.8711,
+      "loss": 7.0195,
       "step": 342
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 0.00020339999999999998,
-      "loss": 6.7457,
+      "loss": 6.8173,
       "step": 343
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 0.000204,
-      "loss": 6.712,
+      "loss": 6.9585,
       "step": 344
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.29,
       "learning_rate": 0.00020459999999999999,
-      "loss": 6.7178,
+      "loss": 6.7353,
       "step": 345
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 0.0002052,
-      "loss": 6.7846,
+      "loss": 6.6473,
       "step": 346
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 0.0002058,
-      "loss": 6.6471,
+      "loss": 6.8274,
       "step": 347
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 0.00020639999999999998,
-      "loss": 6.4892,
+      "loss": 6.9286,
       "step": 348
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 0.00020699999999999996,
-      "loss": 6.5734,
+      "loss": 6.4948,
       "step": 349
     },
     {
-      "epoch": 0.78,
+      "epoch": 0.29,
       "learning_rate": 0.00020759999999999998,
-      "loss": 6.2744,
+      "loss": 6.6249,
       "step": 350
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 0.00020819999999999996,
-      "loss": 6.9985,
+      "loss": 7.2221,
       "step": 351
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 0.00020879999999999998,
-      "loss": 7.0057,
+      "loss": 7.0396,
       "step": 352
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 0.00020939999999999997,
-      "loss": 6.8872,
+      "loss": 6.9473,
       "step": 353
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.3,
       "learning_rate": 0.00020999999999999998,
-      "loss": 6.8159,
+      "loss": 7.0059,
       "step": 354
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 0.00021059999999999997,
-      "loss": 6.8683,
+      "loss": 6.8829,
       "step": 355
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 0.00021119999999999996,
-      "loss": 6.9613,
+      "loss": 6.9656,
       "step": 356
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 0.00021179999999999997,
-      "loss": 6.8534,
+      "loss": 6.9957,
       "step": 357
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 0.00021239999999999996,
-      "loss": 6.9346,
+      "loss": 6.8132,
       "step": 358
     },
     {
-      "epoch": 0.8,
+      "epoch": 0.3,
       "learning_rate": 0.00021299999999999997,
-      "loss": 6.8493,
+      "loss": 6.9307,
       "step": 359
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 0.00021359999999999996,
-      "loss": 6.8302,
+      "loss": 7.0266,
       "step": 360
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 0.00021419999999999998,
-      "loss": 6.8248,
+      "loss": 6.7915,
       "step": 361
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.3,
       "learning_rate": 0.00021479999999999996,
-      "loss": 6.7857,
+      "loss": 6.7976,
       "step": 362
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.31,
       "learning_rate": 0.00021539999999999998,
-      "loss": 6.7263,
+      "loss": 6.9853,
       "step": 363
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 0.00021599999999999996,
-      "loss": 6.7746,
+      "loss": 6.9288,
       "step": 364
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 0.00021659999999999998,
-      "loss": 6.8748,
+      "loss": 6.8851,
       "step": 365
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 0.00021719999999999997,
-      "loss": 6.8155,
+      "loss": 6.9602,
       "step": 366
     },
     {
-      "epoch": 0.82,
+      "epoch": 0.31,
       "learning_rate": 0.00021779999999999998,
-      "loss": 6.9323,
+      "loss": 6.913,
       "step": 367
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 0.00021839999999999997,
-      "loss": 6.8714,
+      "loss": 7.077,
       "step": 368
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 0.00021899999999999998,
-      "loss": 6.7399,
+      "loss": 6.9495,
       "step": 369
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 0.00021959999999999997,
-      "loss": 6.8,
+      "loss": 6.9922,
       "step": 370
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 0.00022019999999999999,
-      "loss": 6.8585,
+      "loss": 6.7844,
       "step": 371
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.31,
       "learning_rate": 0.00022079999999999997,
-      "loss": 6.8913,
+      "loss": 6.8582,
       "step": 372
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 0.0002214,
-      "loss": 6.7891,
+      "loss": 6.7354,
       "step": 373
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.31,
       "learning_rate": 0.00022199999999999998,
-      "loss": 6.6911,
+      "loss": 6.8776,
       "step": 374
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 0.0002226,
-      "loss": 6.8265,
+      "loss": 6.8208,
       "step": 375
     },
     {
-      "epoch": 0.84,
+      "epoch": 0.32,
       "learning_rate": 0.00022319999999999998,
-      "loss": 6.7551,
+      "loss": 6.968,
       "step": 376
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 0.0002238,
-      "loss": 6.8387,
+      "loss": 6.8381,
       "step": 377
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 0.00022439999999999998,
-      "loss": 6.7973,
+      "loss": 6.8864,
       "step": 378
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 0.000225,
-      "loss": 6.795,
+      "loss": 6.9176,
       "step": 379
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 0.00022559999999999998,
-      "loss": 6.8265,
+      "loss": 6.8812,
       "step": 380
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.32,
       "learning_rate": 0.00022619999999999997,
-      "loss": 6.9408,
+      "loss": 6.8497,
       "step": 381
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 0.00022679999999999998,
-      "loss": 6.8602,
+      "loss": 6.8634,
       "step": 382
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 0.00022739999999999997,
-      "loss": 6.848,
+      "loss": 6.7056,
       "step": 383
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 0.00022799999999999999,
-      "loss": 6.8098,
+      "loss": 7.0611,
       "step": 384
     },
     {
-      "epoch": 0.86,
+      "epoch": 0.32,
       "learning_rate": 0.00022859999999999997,
-      "loss": 6.8624,
+      "loss": 6.8537,
       "step": 385
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.32,
       "learning_rate": 0.0002292,
-      "loss": 6.8417,
+      "loss": 6.7578,
       "step": 386
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 0.00022979999999999997,
-      "loss": 6.7887,
+      "loss": 6.807,
       "step": 387
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 0.0002304,
-      "loss": 6.8669,
+      "loss": 6.883,
       "step": 388
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 0.00023099999999999998,
-      "loss": 6.8086,
+      "loss": 6.9673,
       "step": 389
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.33,
       "learning_rate": 0.0002316,
-      "loss": 6.7165,
+      "loss": 7.1346,
       "step": 390
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 0.00023219999999999998,
-      "loss": 6.7425,
+      "loss": 7.0001,
       "step": 391
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 0.0002328,
-      "loss": 6.7985,
+      "loss": 6.8242,
       "step": 392
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 0.00023339999999999998,
-      "loss": 6.71,
+      "loss": 6.7149,
       "step": 393
     },
     {
-      "epoch": 0.88,
+      "epoch": 0.33,
       "learning_rate": 0.000234,
-      "loss": 6.716,
+      "loss": 6.8308,
       "step": 394
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 0.00023459999999999998,
-      "loss": 6.7783,
+      "loss": 6.6752,
       "step": 395
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 0.0002352,
-      "loss": 6.7704,
+      "loss": 6.7412,
       "step": 396
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 0.00023579999999999999,
-      "loss": 6.7056,
+      "loss": 6.4683,
       "step": 397
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.33,
       "learning_rate": 0.0002364,
-      "loss": 6.5548,
+      "loss": 6.5592,
       "step": 398
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.34,
       "learning_rate": 0.000237,
-      "loss": 6.6221,
+      "loss": 6.4819,
       "step": 399
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 0.0002376,
-      "loss": 6.276,
+      "loss": 6.2644,
       "step": 400
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 0.0002382,
-      "loss": 6.9953,
+      "loss": 6.9606,
       "step": 401
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 0.0002388,
-      "loss": 6.8796,
+      "loss": 6.937,
       "step": 402
     },
     {
-      "epoch": 0.9,
+      "epoch": 0.34,
       "learning_rate": 0.0002394,
-      "loss": 7.0043,
+      "loss": 6.9723,
       "step": 403
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 0.00023999999999999998,
-      "loss": 6.8759,
+      "loss": 7.0195,
       "step": 404
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 0.0002406,
-      "loss": 6.7594,
+      "loss": 6.7774,
       "step": 405
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 0.00024119999999999998,
-      "loss": 6.886,
+      "loss": 6.8926,
       "step": 406
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 0.0002418,
-      "loss": 6.8996,
+      "loss": 6.9179,
       "step": 407
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.34,
       "learning_rate": 0.00024239999999999998,
-      "loss": 6.8223,
+      "loss": 6.9387,
       "step": 408
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 0.000243,
-      "loss": 6.7938,
+      "loss": 6.8276,
       "step": 409
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.34,
       "learning_rate": 0.00024359999999999999,
-      "loss": 6.8277,
+      "loss": 7.0094,
       "step": 410
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 0.00024419999999999997,
-      "loss": 6.9263,
+      "loss": 6.9919,
       "step": 411
     },
     {
-      "epoch": 0.92,
+      "epoch": 0.35,
       "learning_rate": 0.0002448,
-      "loss": 6.8408,
+      "loss": 6.9548,
       "step": 412
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 0.00024539999999999995,
-      "loss": 6.8565,
+      "loss": 6.8655,
       "step": 413
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 0.00024599999999999996,
-      "loss": 6.8711,
+      "loss": 6.7854,
       "step": 414
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 0.0002466,
-      "loss": 6.9096,
+      "loss": 6.8031,
       "step": 415
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 0.0002472,
-      "loss": 6.852,
+      "loss": 7.0633,
       "step": 416
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.35,
       "learning_rate": 0.00024779999999999995,
-      "loss": 6.8442,
+      "loss": 6.8614,
       "step": 417
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 0.00024839999999999997,
-      "loss": 6.7525,
+      "loss": 6.7194,
       "step": 418
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 0.000249,
-      "loss": 6.7989,
+      "loss": 6.8107,
       "step": 419
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 0.00024959999999999994,
-      "loss": 6.8196,
+      "loss": 6.9937,
       "step": 420
     },
     {
-      "epoch": 0.94,
+      "epoch": 0.35,
       "learning_rate": 0.00025019999999999996,
-      "loss": 6.8717,
+      "loss": 6.8863,
       "step": 421
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.35,
       "learning_rate": 0.00025079999999999997,
-      "loss": 6.7613,
+      "loss": 6.7743,
       "step": 422
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 0.0002514,
-      "loss": 6.8156,
+      "loss": 6.8863,
       "step": 423
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 0.00025199999999999995,
-      "loss": 6.9202,
+      "loss": 6.7851,
       "step": 424
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.36,
       "learning_rate": 0.00025259999999999996,
-      "loss": 6.7485,
+      "loss": 7.0815,
       "step": 425
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 0.0002532,
-      "loss": 6.8243,
+      "loss": 6.8696,
       "step": 426
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 0.0002538,
-      "loss": 6.7921,
+      "loss": 6.8131,
       "step": 427
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 0.00025439999999999995,
-      "loss": 6.7203,
+      "loss": 6.8926,
       "step": 428
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 0.00025499999999999996,
-      "loss": 6.6956,
+      "loss": 6.9502,
       "step": 429
     },
     {
-      "epoch": 0.96,
+      "epoch": 0.36,
       "learning_rate": 0.0002556,
-      "loss": 6.7347,
+      "loss": 6.8884,
       "step": 430
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 0.0002562,
-      "loss": 6.7942,
+      "loss": 6.9901,
       "step": 431
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 0.00025679999999999995,
-      "loss": 6.9015,
+      "loss": 6.666,
       "step": 432
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 0.00025739999999999997,
-      "loss": 6.7733,
+      "loss": 6.9318,
       "step": 433
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.36,
       "learning_rate": 0.000258,
-      "loss": 6.7356,
+      "loss": 6.8993,
       "step": 434
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 0.0002586,
-      "loss": 6.8718,
+      "loss": 6.9596,
       "step": 435
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 0.00025919999999999996,
-      "loss": 6.758,
+      "loss": 6.6773,
       "step": 436
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 0.00025979999999999997,
-      "loss": 6.7061,
+      "loss": 6.7725,
       "step": 437
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 0.0002604,
-      "loss": 6.8063,
+      "loss": 6.7497,
       "step": 438
     },
     {
-      "epoch": 0.98,
+      "epoch": 0.37,
       "learning_rate": 0.000261,
-      "loss": 6.787,
+      "loss": 7.0099,
       "step": 439
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 0.00026159999999999996,
-      "loss": 6.7439,
+      "loss": 6.7874,
       "step": 440
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 0.0002622,
-      "loss": 6.5944,
+      "loss": 6.8823,
       "step": 441
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 0.0002628,
-      "loss": 6.7605,
+      "loss": 6.767,
       "step": 442
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.37,
       "learning_rate": 0.00026339999999999995,
-      "loss": 6.4567,
+      "loss": 6.8022,
       "step": 443
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 0.00026399999999999997,
-      "loss": 6.6392,
+      "loss": 6.6632,
       "step": 444
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 0.0002646,
-      "loss": 6.4273,
+      "loss": 6.5436,
       "step": 445
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.37,
       "learning_rate": 0.0002652,
-      "loss": 6.3367,
+      "loss": 6.759,
       "step": 446
     },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 6.672,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 6.8294,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000267,
+      "loss": 6.5779,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002676,
+      "loss": 6.3936,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 6.9043,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002688,
+      "loss": 6.8242,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002694,
+      "loss": 6.8922,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027,
+      "loss": 6.8742,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 6.8983,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002712,
+      "loss": 6.8824,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002718,
+      "loss": 6.8802,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002724,
+      "loss": 7.005,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 6.9349,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002736,
+      "loss": 6.9665,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002742,
+      "loss": 6.8687,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002748,
+      "loss": 6.837,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 6.9516,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000276,
+      "loss": 6.9546,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002766,
+      "loss": 6.9346,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002772,
+      "loss": 6.8403,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002778,
+      "loss": 6.7476,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002784,
+      "loss": 6.814,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000279,
+      "loss": 6.6635,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 7.0382,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002802,
+      "loss": 6.8237,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002808,
+      "loss": 6.9813,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 6.9854,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.9881,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002826,
+      "loss": 6.8715,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 6.7408,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 6.661,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002844,
+      "loss": 6.8297,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000285,
+      "loss": 6.7972,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 6.9775,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 6.694,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002868,
+      "loss": 6.7856,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 6.8002,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 6.903,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 6.8838,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002892,
+      "loss": 6.7277,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 6.8032,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 6.8841,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 6.9144,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002916,
+      "loss": 6.783,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 6.9195,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 6.9088,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002934,
+      "loss": 6.9646,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000294,
+      "loss": 6.8541,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 6.7656,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 6.7223,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958,
+      "loss": 6.694,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002964,
+      "loss": 6.592,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 6.5935,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 6.5385,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 6.84136962890625,
+      "eval_runtime": 492.9707,
+      "eval_samples_per_second": 5.359,
+      "eval_steps_per_second": 0.448,
+      "eval_wer": 1.0500029763676408,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002982,
+      "loss": 6.9981,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002988,
+      "loss": 6.9229,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 6.959,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003,
+      "loss": 6.8615,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002995645863570392,
+      "loss": 6.8716,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002991291727140783,
+      "loss": 6.9566,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029869375907111753,
+      "loss": 6.8458,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029825834542815676,
+      "loss": 6.8495,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029782293178519587,
+      "loss": 6.8151,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002973875181422351,
+      "loss": 6.9038,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029695210449927426,
+      "loss": 6.7452,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002965166908563135,
+      "loss": 6.9479,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029608127721335266,
+      "loss": 6.8531,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956458635703918,
+      "loss": 6.7814,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029521044992743105,
+      "loss": 6.9125,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002947750362844702,
+      "loss": 6.7181,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002943396226415094,
+      "loss": 6.9138,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002939042089985486,
+      "loss": 6.7617,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002934687953555878,
+      "loss": 7.0497,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029303338171262695,
+      "loss": 6.7293,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029259796806966617,
+      "loss": 6.8915,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029216255442670534,
+      "loss": 6.8637,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002917271407837445,
+      "loss": 6.9329,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029129172714078373,
+      "loss": 6.8185,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002908563134978229,
+      "loss": 6.8207,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029042089985486207,
+      "loss": 6.726,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002899854862119013,
+      "loss": 6.8523,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028955007256894046,
+      "loss": 6.8668,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002891146589259796,
+      "loss": 6.857,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028867924528301885,
+      "loss": 6.8504,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000288243831640058,
+      "loss": 6.8464,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002878084179970972,
+      "loss": 6.8743,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002873730043541364,
+      "loss": 6.6959,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002869375907111756,
+      "loss": 6.7821,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028650217706821475,
+      "loss": 6.9036,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028606676342525397,
+      "loss": 6.7307,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028563134978229314,
+      "loss": 6.8073,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002851959361393323,
+      "loss": 6.6503,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028476052249637153,
+      "loss": 6.8456,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002843251088534107,
+      "loss": 6.8926,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002838896952104499,
+      "loss": 6.9259,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002834542815674891,
+      "loss": 6.7483,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028301886792452826,
+      "loss": 6.5999,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002825834542815675,
+      "loss": 6.9423,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028214804063860665,
+      "loss": 6.8099,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002817126269956458,
+      "loss": 6.7497,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028127721335268504,
+      "loss": 6.4958,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002808417997097242,
+      "loss": 6.6541,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002804063860667634,
+      "loss": 6.4972,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002799709724238026,
+      "loss": 6.3706,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002795355587808418,
+      "loss": 7.0303,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027910014513788094,
+      "loss": 6.9142,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027866473149492017,
+      "loss": 6.934,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027822931785195933,
+      "loss": 6.8109,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002777939042089985,
+      "loss": 6.9807,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002773584905660377,
+      "loss": 6.8507,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002769230769230769,
+      "loss": 6.8355,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027648766328011606,
+      "loss": 6.7796,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002760522496371553,
+      "loss": 6.9876,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027561683599419446,
+      "loss": 6.7679,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002751814223512336,
+      "loss": 7.0946,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027474600870827285,
+      "loss": 6.6743,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000274310595065312,
+      "loss": 6.946,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002738751814223512,
+      "loss": 6.5808,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002734397677793904,
+      "loss": 6.7715,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002730043541364296,
+      "loss": 6.8293,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027256894049346875,
+      "loss": 6.7443,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027213352685050797,
+      "loss": 6.7721,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027169811320754714,
+      "loss": 6.8,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027126269956458636,
+      "loss": 6.8229,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027082728592162553,
+      "loss": 6.8486,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002703918722786647,
+      "loss": 6.8454,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002699564586357039,
+      "loss": 6.8049,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002695210449927431,
+      "loss": 6.8674,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026908563134978226,
+      "loss": 7.0463,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002686502177068215,
+      "loss": 6.9523,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026821480406386065,
+      "loss": 6.908,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002677793904208998,
+      "loss": 6.8119,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026734397677793904,
+      "loss": 6.7422,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002669085631349782,
+      "loss": 6.9766,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002664731494920174,
+      "loss": 6.9451,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002660377358490566,
+      "loss": 6.7142,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026560232220609577,
+      "loss": 6.7909,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026516690856313494,
+      "loss": 6.7317,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026473149492017416,
+      "loss": 6.9627,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026429608127721333,
+      "loss": 6.8146,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002638606676342525,
+      "loss": 6.8995,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002634252539912917,
+      "loss": 6.7827,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002629898403483309,
+      "loss": 6.6971,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026255442670537006,
+      "loss": 6.7937,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002621190130624093,
+      "loss": 6.8905,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026168359941944845,
+      "loss": 6.7583,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002612481857764876,
+      "loss": 6.8475,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026081277213352684,
+      "loss": 6.6513,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000260377358490566,
+      "loss": 6.5678,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002599419448476052,
+      "loss": 6.5821,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002595065312046444,
+      "loss": 6.5131,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002590711175616836,
+      "loss": 6.5356,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025863570391872274,
+      "loss": 6.3458,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025820029027576197,
+      "loss": 6.2131,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025776487663280113,
+      "loss": 6.8324,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025732946298984036,
+      "loss": 7.035,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002568940493468795,
+      "loss": 6.9698,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002564586357039187,
+      "loss": 6.796,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002560232220609579,
+      "loss": 6.9368,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002555878084179971,
+      "loss": 7.0629,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025515239477503626,
+      "loss": 6.7385,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002547169811320755,
+      "loss": 6.9024,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025428156748911465,
+      "loss": 6.8677,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002538461538461538,
+      "loss": 6.8823,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025341074020319304,
+      "loss": 6.9585,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002529753265602322,
+      "loss": 6.7669,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002525399129172714,
+      "loss": 6.7928,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002521044992743106,
+      "loss": 6.7403,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00025166908563134977,
+      "loss": 6.9072,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00025123367198838894,
+      "loss": 6.7032,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00025079825834542816,
+      "loss": 6.7789,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00025036284470246733,
+      "loss": 6.7426,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002499274310595065,
+      "loss": 6.8267,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002494920174165457,
+      "loss": 6.921,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002490566037735849,
+      "loss": 6.6063,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024862119013062406,
+      "loss": 6.7041,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002481857764876633,
+      "loss": 6.7416,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024775036284470245,
+      "loss": 6.8333,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002473149492017416,
+      "loss": 6.8254,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024687953555878084,
+      "loss": 6.9049,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024644412191582,
+      "loss": 6.894,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002460087082728592,
+      "loss": 6.6467,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002455732946298984,
+      "loss": 6.7039,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024513788098693757,
+      "loss": 6.7366,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002447024673439768,
+      "loss": 6.835,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024426705370101596,
+      "loss": 6.9096,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024383164005805513,
+      "loss": 6.7752,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024339622641509433,
+      "loss": 6.8582,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002429608127721335,
+      "loss": 6.8206,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002425253991291727,
+      "loss": 6.8518,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002420899854862119,
+      "loss": 6.7277,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024165457184325106,
+      "loss": 6.7397,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024121915820029025,
+      "loss": 6.7723,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024078374455732945,
+      "loss": 6.9939,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024034833091436862,
+      "loss": 6.7244,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002399129172714078,
+      "loss": 6.6818,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000239477503628447,
+      "loss": 6.8975,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002390420899854862,
+      "loss": 6.7538,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023860667634252537,
+      "loss": 6.7806,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023817126269956457,
+      "loss": 6.6033,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023773584905660377,
+      "loss": 6.2432,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00023730043541364293,
+      "loss": 6.7507,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023686502177068213,
+      "loss": 6.7112,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023642960812772133,
+      "loss": 6.2708,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002359941944847605,
+      "loss": 7.1084,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002355587808417997,
+      "loss": 6.8681,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023512336719883889,
+      "loss": 6.9773,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023468795355587805,
+      "loss": 6.9661,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023425253991291725,
+      "loss": 6.9394,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023381712626995645,
+      "loss": 6.8747,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00023338171262699562,
+      "loss": 6.7762,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002329462989840348,
+      "loss": 6.8591,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000232510885341074,
+      "loss": 6.8248,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002320754716981132,
+      "loss": 6.8163,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00023164005805515237,
+      "loss": 6.7533,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00023120464441219157,
+      "loss": 6.8689,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00023076923076923076,
+      "loss": 6.6898,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00023033381712626993,
+      "loss": 6.9015,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022989840348330913,
+      "loss": 6.8106,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022946298984034832,
+      "loss": 6.6644,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002290275761973875,
+      "loss": 6.8867,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002285921625544267,
+      "loss": 6.7743,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022815674891146588,
+      "loss": 7.0751,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022772133526850505,
+      "loss": 6.9112,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022728592162554425,
+      "loss": 6.808,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022685050798258345,
+      "loss": 6.7739,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022641509433962264,
+      "loss": 6.7714,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002259796806966618,
+      "loss": 6.7218,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000225544267053701,
+      "loss": 6.8919,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002251088534107402,
+      "loss": 6.7952,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022467343976777937,
+      "loss": 6.9655,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022423802612481857,
+      "loss": 6.9544,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022380261248185776,
+      "loss": 6.8098,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022336719883889693,
+      "loss": 6.9409,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022293178519593613,
+      "loss": 6.5972,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022249637155297532,
+      "loss": 6.7181,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002220609579100145,
+      "loss": 6.7576,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002216255442670537,
+      "loss": 6.6885,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022119013062409288,
+      "loss": 6.7601,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022075471698113205,
+      "loss": 6.7339,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022031930333817125,
+      "loss": 6.8451,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00021988388969521044,
+      "loss": 6.7911,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00021944847605224964,
+      "loss": 6.9605,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002190130624092888,
+      "loss": 6.7434,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000218577648766328,
+      "loss": 6.9479,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002181422351233672,
+      "loss": 6.7066,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00021770682148040637,
+      "loss": 6.7052,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00021727140783744556,
+      "loss": 6.8302,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00021683599419448476,
+      "loss": 6.7822,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021640058055152393,
+      "loss": 6.4113,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021596516690856313,
+      "loss": 6.8164,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021552975326560232,
+      "loss": 6.7107,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021509433962264146,
+      "loss": 6.2754,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021465892597968069,
+      "loss": 6.5328,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021422351233671988,
+      "loss": 7.0301,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021378809869375908,
+      "loss": 6.795,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021335268505079825,
+      "loss": 6.9513,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021291727140783744,
+      "loss": 6.7841,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021248185776487664,
+      "loss": 6.8209,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021204644412191578,
+      "loss": 6.9158,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000211611030478955,
+      "loss": 6.8196,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002111756168359942,
+      "loss": 6.7451,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021074020319303334,
+      "loss": 6.8313,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021030478955007256,
+      "loss": 6.9144,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020986937590711176,
+      "loss": 6.6823,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002094339622641509,
+      "loss": 7.0659,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002089985486211901,
+      "loss": 6.8574,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020856313497822932,
+      "loss": 6.8299,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020812772133526846,
+      "loss": 6.7282,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020769230769230766,
+      "loss": 6.7405,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020725689404934685,
+      "loss": 6.806,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020682148040638608,
+      "loss": 6.8661,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020638606676342522,
+      "loss": 6.8123,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020595065312046441,
+      "loss": 6.7411,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020551523947750364,
+      "loss": 6.6403,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020507982583454278,
+      "loss": 6.6972,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020464441219158197,
+      "loss": 6.75,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020420899854862117,
+      "loss": 6.6767,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020377358490566034,
+      "loss": 6.8202,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020333817126269953,
+      "loss": 6.724,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020290275761973873,
+      "loss": 6.7514,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002024673439767779,
+      "loss": 6.8455,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002020319303338171,
+      "loss": 6.761,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002015965166908563,
+      "loss": 6.8119,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020116110304789546,
+      "loss": 6.7588,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00020072568940493466,
+      "loss": 6.795,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00020029027576197385,
+      "loss": 6.8161,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019985486211901305,
+      "loss": 6.7052,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019941944847605222,
+      "loss": 6.814,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001989840348330914,
+      "loss": 6.7445,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001985486211901306,
+      "loss": 6.9489,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019811320754716978,
+      "loss": 6.8612,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019767779390420897,
+      "loss": 6.837,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019724238026124817,
+      "loss": 6.7348,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019680696661828734,
+      "loss": 6.8755,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019637155297532653,
+      "loss": 6.8379,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019593613933236573,
+      "loss": 6.8679,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001955007256894049,
+      "loss": 6.6225,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001950653120464441,
+      "loss": 6.5989,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001946298984034833,
+      "loss": 6.5353,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019419448476052249,
+      "loss": 6.6957,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019375907111756165,
+      "loss": 6.836,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019332365747460085,
+      "loss": 6.3925,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019288824383164005,
+      "loss": 5.9944,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019245283018867922,
+      "loss": 6.8929,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001920174165457184,
+      "loss": 7.029,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001915820029027576,
+      "loss": 6.7746,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019114658925979678,
+      "loss": 6.7398,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019071117561683597,
+      "loss": 6.9454,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019027576197387517,
+      "loss": 6.8805,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018984034833091434,
+      "loss": 6.8015,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018940493468795353,
+      "loss": 6.9505,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018896952104499273,
+      "loss": 6.701,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001885341074020319,
+      "loss": 6.8635,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001880986937590711,
+      "loss": 6.7672,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001876632801161103,
+      "loss": 6.6565,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018722786647314948,
+      "loss": 6.7407,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018679245283018865,
+      "loss": 6.6525,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018635703918722785,
+      "loss": 7.0426,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018592162554426704,
+      "loss": 6.8255,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018548621190130621,
+      "loss": 6.8092,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001850507982583454,
+      "loss": 6.6206,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001846153846153846,
+      "loss": 6.7722,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018417997097242377,
+      "loss": 6.8224,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018374455732946297,
+      "loss": 6.7581,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018330914368650217,
+      "loss": 6.892,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018287373004354133,
+      "loss": 6.717,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018243831640058053,
+      "loss": 6.593,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018200290275761973,
+      "loss": 6.851,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018156748911465892,
+      "loss": 6.7269,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001811320754716981,
+      "loss": 6.7105,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001806966618287373,
+      "loss": 6.7603,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018026124818577648,
+      "loss": 6.7945,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017982583454281565,
+      "loss": 6.9823,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017939042089985485,
+      "loss": 6.8197,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017895500725689404,
+      "loss": 6.6885,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001785195936139332,
+      "loss": 6.6007,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001780841799709724,
+      "loss": 6.7959,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001776487663280116,
+      "loss": 6.8707,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017721335268505077,
+      "loss": 6.6443,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017677793904208997,
+      "loss": 6.8777,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017634252539912916,
+      "loss": 6.7187,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017590711175616833,
+      "loss": 6.7757,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017547169811320753,
+      "loss": 6.6934,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017503628447024673,
+      "loss": 6.8231,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017460087082728592,
+      "loss": 6.7268,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001741654571843251,
+      "loss": 6.6948,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017373004354136429,
+      "loss": 6.548,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017329462989840348,
+      "loss": 6.5613,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017285921625544265,
+      "loss": 6.4696,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017242380261248185,
+      "loss": 6.4884,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017198838896952104,
+      "loss": 6.4172,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001715529753265602,
+      "loss": 6.3739,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001711175616835994,
+      "loss": 6.0278,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001706821480406386,
+      "loss": 6.8293,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017024673439767777,
+      "loss": 6.6272,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016981132075471697,
+      "loss": 6.7779,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016937590711175616,
+      "loss": 6.8186,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016894049346879536,
+      "loss": 6.8187,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016850507982583453,
+      "loss": 6.7101,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016806966618287372,
+      "loss": 6.7569,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016763425253991292,
+      "loss": 6.7976,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001671988388969521,
+      "loss": 6.7636,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016676342525399128,
+      "loss": 6.7901,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016632801161103048,
+      "loss": 6.6903,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016589259796806965,
+      "loss": 6.7479,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016545718432510884,
+      "loss": 6.6714,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00016502177068214804,
+      "loss": 6.9813,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001645863570391872,
+      "loss": 6.9063,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001641509433962264,
+      "loss": 6.6002,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001637155297532656,
+      "loss": 6.8063,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016328011611030477,
+      "loss": 6.5225,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016284470246734397,
+      "loss": 6.7463,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016240928882438316,
+      "loss": 6.7756,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016197387518142236,
+      "loss": 6.7644,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016153846153846153,
+      "loss": 6.6968,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016110304789550072,
+      "loss": 6.7608,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00016066763425253992,
+      "loss": 6.8758,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001602322206095791,
+      "loss": 6.7152,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015979680696661828,
+      "loss": 6.7612,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015936139332365748,
+      "loss": 6.8222,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015892597968069665,
+      "loss": 6.8338,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015849056603773584,
+      "loss": 6.762,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015805515239477504,
+      "loss": 6.9651,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001576197387518142,
+      "loss": 6.7731,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001571843251088534,
+      "loss": 6.5439,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001567489114658926,
+      "loss": 6.6829,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001563134978229318,
+      "loss": 6.7925,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015587808417997096,
+      "loss": 6.8057,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015544267053701016,
+      "loss": 6.8506,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015500725689404936,
+      "loss": 6.7867,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015457184325108852,
+      "loss": 6.4587,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015413642960812772,
+      "loss": 6.9501,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015370101596516692,
+      "loss": 6.6549,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015326560232220606,
+      "loss": 6.7495,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015283018867924528,
+      "loss": 6.6763,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015239477503628448,
+      "loss": 6.944,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015195936139332362,
+      "loss": 6.5565,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015152394775036284,
+      "loss": 6.6403,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015108853410740204,
+      "loss": 6.6138,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015065312046444118,
+      "loss": 6.5652,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015021770682148038,
+      "loss": 6.2956,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001497822931785196,
+      "loss": 6.6097,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014934687953555877,
+      "loss": 6.4595,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014891146589259794,
+      "loss": 7.0019,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014847605224963713,
+      "loss": 6.7696,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014804063860667633,
+      "loss": 6.774,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014760522496371552,
+      "loss": 6.8851,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001471698113207547,
+      "loss": 6.7991,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001467343976777939,
+      "loss": 6.8332,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014629898403483308,
+      "loss": 6.669,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014586357039187225,
+      "loss": 6.7986,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014542815674891145,
+      "loss": 6.6709,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014499274310595064,
+      "loss": 6.7066,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001445573294629898,
+      "loss": 6.744,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000144121915820029,
+      "loss": 6.7785,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001436865021770682,
+      "loss": 6.6285,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014325108853410737,
+      "loss": 6.8053,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014281567489114657,
+      "loss": 6.8246,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014238026124818577,
+      "loss": 6.6805,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014194484760522496,
+      "loss": 6.6297,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014150943396226413,
+      "loss": 6.6603,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014107402031930333,
+      "loss": 6.6392,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014063860667634252,
+      "loss": 6.5813,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001402031930333817,
+      "loss": 6.6803,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001397677793904209,
+      "loss": 6.5655,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013933236574746008,
+      "loss": 6.8468,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013889695210449925,
+      "loss": 6.7848,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013846153846153845,
+      "loss": 6.82,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013802612481857764,
+      "loss": 6.599,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001375907111756168,
+      "loss": 6.6633,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000137155297532656,
+      "loss": 6.8072,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001367198838896952,
+      "loss": 6.6585,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013628447024673437,
+      "loss": 6.8383,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013584905660377357,
+      "loss": 6.6623,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013541364296081276,
+      "loss": 6.769,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013497822931785196,
+      "loss": 6.7631,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013454281567489113,
+      "loss": 6.7775,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013410740203193032,
+      "loss": 6.6173,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013367198838896952,
+      "loss": 6.6929,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001332365747460087,
+      "loss": 6.6906,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013280116110304789,
+      "loss": 6.7141,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013236574746008708,
+      "loss": 6.7914,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013193033381712625,
+      "loss": 6.6043,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013149492017416545,
+      "loss": 6.8003,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013105950653120464,
+      "loss": 6.7134,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001306240928882438,
+      "loss": 6.5845,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000130188679245283,
+      "loss": 6.5482,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001297532656023222,
+      "loss": 6.6262,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00012931785195936137,
+      "loss": 6.5822,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00012888243831640057,
+      "loss": 6.3595,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00012844702467343976,
+      "loss": 6.1957,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012801161103047896,
+      "loss": 6.3524,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012757619738751813,
+      "loss": 5.8456,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012714078374455732,
+      "loss": 6.8956,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012670537010159652,
+      "loss": 6.8253,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001262699564586357,
+      "loss": 6.8014,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012583454281567488,
+      "loss": 6.614,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012539912917271408,
+      "loss": 6.7152,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012496371552975325,
+      "loss": 6.7951,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012452830188679244,
+      "loss": 6.706,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012409288824383164,
+      "loss": 7.0412,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001236574746008708,
+      "loss": 6.7611,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012322206095791,
+      "loss": 6.785,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001227866473149492,
+      "loss": 6.7113,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001223512336719884,
+      "loss": 6.7938,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012191582002902757,
+      "loss": 6.7073,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012148040638606675,
+      "loss": 6.6513,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012104499274310594,
+      "loss": 6.7098,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012060957910014513,
+      "loss": 6.7984,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012017416545718431,
+      "loss": 6.7216,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001197387518142235,
+      "loss": 6.7223,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011930333817126269,
+      "loss": 6.8197,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011886792452830188,
+      "loss": 6.7176,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011843251088534106,
+      "loss": 6.6393,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011799709724238025,
+      "loss": 6.853,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011756168359941944,
+      "loss": 6.7337,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011712626995645863,
+      "loss": 6.779,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011669085631349781,
+      "loss": 6.6274,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000116255442670537,
+      "loss": 6.6959,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011582002902757619,
+      "loss": 6.8175,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 6.7653,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011494920174165456,
+      "loss": 6.6164,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011451378809869375,
+      "loss": 6.677,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011407837445573294,
+      "loss": 6.7269,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011364296081277212,
+      "loss": 6.7803,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011320754716981132,
+      "loss": 6.8275,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001127721335268505,
+      "loss": 6.7024,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011233671988388969,
+      "loss": 6.6885,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011190130624092888,
+      "loss": 6.6577,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011146589259796806,
+      "loss": 6.712,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011103047895500725,
+      "loss": 6.8692,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011059506531204644,
+      "loss": 6.7184,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011015965166908562,
+      "loss": 6.6514,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010972423802612482,
+      "loss": 6.6002,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000109288824383164,
+      "loss": 6.714,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010885341074020318,
+      "loss": 6.3859,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010841799709724238,
+      "loss": 6.3666,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010798258345428156,
+      "loss": 6.7387,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010754716981132073,
+      "loss": 6.5286,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010711175616835994,
+      "loss": 6.3192,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010667634252539912,
+      "loss": 6.2644,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010624092888243832,
+      "loss": 6.1977,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001058055152394775,
+      "loss": 6.0231,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010537010159651667,
+      "loss": 6.8315,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010493468795355588,
+      "loss": 6.8403,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010449927431059505,
+      "loss": 6.7027,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010406386066763423,
+      "loss": 6.5141,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010362844702467343,
+      "loss": 6.7009,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010319303338171261,
+      "loss": 6.8507,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010275761973875182,
+      "loss": 6.745,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010232220609579099,
+      "loss": 6.8369,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010188679245283017,
+      "loss": 6.7382,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010145137880986937,
+      "loss": 6.6531,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010101596516690855,
+      "loss": 6.6775,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010058055152394773,
+      "loss": 6.654,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010014513788098693,
+      "loss": 6.5752,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.970972423802611e-05,
+      "loss": 6.6846,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.92743105950653e-05,
+      "loss": 6.7512,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.883889695210449e-05,
+      "loss": 6.7307,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.840348330914367e-05,
+      "loss": 6.8927,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.796806966618286e-05,
+      "loss": 6.7615,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.753265602322205e-05,
+      "loss": 6.4604,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.709724238026124e-05,
+      "loss": 6.7544,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.666182873730043e-05,
+      "loss": 6.5679,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.622641509433961e-05,
+      "loss": 6.5403,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.57910014513788e-05,
+      "loss": 6.6384,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.535558780841799e-05,
+      "loss": 6.7412,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.492017416545717e-05,
+      "loss": 6.7564,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.448476052249636e-05,
+      "loss": 6.7782,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.404934687953555e-05,
+      "loss": 6.6239,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.361393323657474e-05,
+      "loss": 6.632,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.317851959361392e-05,
+      "loss": 6.6647,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.274310595065311e-05,
+      "loss": 6.7482,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.23076923076923e-05,
+      "loss": 6.9126,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.187227866473149e-05,
+      "loss": 6.7464,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.143686502177067e-05,
+      "loss": 6.7075,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.100145137880986e-05,
+      "loss": 6.9183,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.056603773584905e-05,
+      "loss": 6.5806,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.013062409288824e-05,
+      "loss": 6.5799,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.969521044992742e-05,
+      "loss": 6.7535,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.92597968069666e-05,
+      "loss": 6.6896,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.88243831640058e-05,
+      "loss": 6.654,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.838896952104498e-05,
+      "loss": 6.574,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.795355587808417e-05,
+      "loss": 6.6576,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.751814223512336e-05,
+      "loss": 6.6829,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.708272859216254e-05,
+      "loss": 6.5027,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.664731494920174e-05,
+      "loss": 6.767,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.621190130624092e-05,
+      "loss": 6.6885,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.57764876632801e-05,
+      "loss": 6.8854,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.53410740203193e-05,
+      "loss": 6.5446,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.490566037735848e-05,
+      "loss": 6.4119,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.447024673439768e-05,
+      "loss": 6.295,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.403483309143686e-05,
+      "loss": 6.1702,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 6.585080146789551,
+      "eval_runtime": 754.776,
+      "eval_samples_per_second": 3.5,
+      "eval_steps_per_second": 0.293,
+      "eval_wer": 1.867511955076691,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.359941944847604e-05,
+      "loss": 6.6907,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.316400580551524e-05,
+      "loss": 6.757,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.272859216255442e-05,
+      "loss": 6.7429,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.22931785195936e-05,
+      "loss": 6.7758,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.18577648766328e-05,
+      "loss": 6.7114,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.142235123367198e-05,
+      "loss": 6.7421,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.098693759071118e-05,
+      "loss": 6.6519,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.055152394775036e-05,
+      "loss": 6.5508,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.011611030478954e-05,
+      "loss": 6.7702,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.968069666182874e-05,
+      "loss": 6.5349,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.924528301886792e-05,
+      "loss": 6.6104,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.88098693759071e-05,
+      "loss": 6.638,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.83744557329463e-05,
+      "loss": 6.5846,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.793904208998548e-05,
+      "loss": 6.5231,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.750362844702468e-05,
+      "loss": 6.5827,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.706821480406386e-05,
+      "loss": 6.755,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.663280116110303e-05,
+      "loss": 6.5473,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.619738751814224e-05,
+      "loss": 6.6718,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.576197387518142e-05,
+      "loss": 6.8111,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.532656023222059e-05,
+      "loss": 6.5385,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.48911465892598e-05,
+      "loss": 6.7338,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.445573294629897e-05,
+      "loss": 6.7642,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.402031930333816e-05,
+      "loss": 6.8209,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.358490566037735e-05,
+      "loss": 6.6269,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.314949201741654e-05,
+      "loss": 6.5789,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.271407837445572e-05,
+      "loss": 6.5876,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.22786647314949e-05,
+      "loss": 6.5396,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.18432510885341e-05,
+      "loss": 6.6626,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.140783744557328e-05,
+      "loss": 6.6721,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.097242380261248e-05,
+      "loss": 6.5971,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.053701015965166e-05,
+      "loss": 6.6966,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.010159651669085e-05,
+      "loss": 6.4928,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.966618287373004e-05,
+      "loss": 6.6723,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.923076923076922e-05,
+      "loss": 6.6167,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.87953555878084e-05,
+      "loss": 6.6542,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.83599419448476e-05,
+      "loss": 6.7551,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.792452830188678e-05,
+      "loss": 6.5051,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.748911465892598e-05,
+      "loss": 6.7267,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.705370101596516e-05,
+      "loss": 6.6521,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.661828737300434e-05,
+      "loss": 6.5425,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.618287373004354e-05,
+      "loss": 6.5041,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.574746008708272e-05,
+      "loss": 6.559,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.53120464441219e-05,
+      "loss": 6.5408,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.48766328011611e-05,
+      "loss": 6.5498,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.444121915820028e-05,
+      "loss": 6.573,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.400580551523948e-05,
+      "loss": 6.4869,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.357039187227866e-05,
+      "loss": 6.301,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.313497822931784e-05,
+      "loss": 6.4788,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.269956458635704e-05,
+      "loss": 6.524,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.226415094339622e-05,
+      "loss": 6.042,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.18287373004354e-05,
+      "loss": 6.8218,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.13933236574746e-05,
+      "loss": 6.6262,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095791001451378e-05,
+      "loss": 6.6762,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052249637155297e-05,
+      "loss": 6.74,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0087082728592154e-05,
+      "loss": 6.7309,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.965166908563134e-05,
+      "loss": 6.7538,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921625544267053e-05,
+      "loss": 6.7058,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.878084179970972e-05,
+      "loss": 6.5754,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8345428156748904e-05,
+      "loss": 6.755,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.791001451378809e-05,
+      "loss": 6.7243,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.747460087082728e-05,
+      "loss": 6.7692,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.703918722786647e-05,
+      "loss": 6.6864,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 6.7534,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.616835994194484e-05,
+      "loss": 6.6074,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573294629898403e-05,
+      "loss": 6.6383,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529753265602322e-05,
+      "loss": 6.6331,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.486211901306241e-05,
+      "loss": 6.6365,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.442670537010159e-05,
+      "loss": 6.6805,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.399129172714078e-05,
+      "loss": 6.7947,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355587808417997e-05,
+      "loss": 6.7136,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.312046444121916e-05,
+      "loss": 6.5652,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2685050798258335e-05,
+      "loss": 6.6649,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2249637155297524e-05,
+      "loss": 6.4699,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181422351233671e-05,
+      "loss": 6.6203,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137880986937591e-05,
+      "loss": 6.3756,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0943396226415085e-05,
+      "loss": 6.6922,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.0507982583454274e-05,
+      "loss": 6.8846,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.007256894049346e-05,
+      "loss": 6.732,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.963715529753265e-05,
+      "loss": 6.537,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9201741654571834e-05,
+      "loss": 6.5756,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8766328011611024e-05,
+      "loss": 6.6429,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.833091436865021e-05,
+      "loss": 6.9331,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.78955007256894e-05,
+      "loss": 6.4924,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7460087082728584e-05,
+      "loss": 6.6273,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702467343976777e-05,
+      "loss": 6.6147,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658925979680696e-05,
+      "loss": 6.6515,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.615384615384615e-05,
+      "loss": 6.566,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5718432510885334e-05,
+      "loss": 6.6589,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.528301886792452e-05,
+      "loss": 6.649,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.484760522496371e-05,
+      "loss": 6.5398,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.44121915820029e-05,
+      "loss": 6.5297,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.397677793904208e-05,
+      "loss": 6.4694,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.354136429608127e-05,
+      "loss": 6.4292,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.310595065312046e-05,
+      "loss": 6.637,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.267053701015965e-05,
+      "loss": 6.4221,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.223512336719884e-05,
+      "loss": 6.5716,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.179970972423802e-05,
+      "loss": 6.3362,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.136429608127721e-05,
+      "loss": 6.5254,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.09288824383164e-05,
+      "loss": 6.4824,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.049346879535559e-05,
+      "loss": 5.7116,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.005805515239477e-05,
+      "loss": 6.7332,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.962264150943396e-05,
+      "loss": 6.7109,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.918722786647315e-05,
+      "loss": 6.7897,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.875181422351234e-05,
+      "loss": 6.6304,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8316400580551515e-05,
+      "loss": 6.5934,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.788098693759071e-05,
+      "loss": 6.5387,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.74455732946299e-05,
+      "loss": 6.916,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.701015965166908e-05,
+      "loss": 6.8103,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.657474600870827e-05,
+      "loss": 6.7109,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.613933236574745e-05,
+      "loss": 6.531,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.570391872278664e-05,
+      "loss": 6.625,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.526850507982583e-05,
+      "loss": 6.7393,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.483309143686502e-05,
+      "loss": 6.5094,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43976777939042e-05,
+      "loss": 6.5362,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.396226415094339e-05,
+      "loss": 6.7859,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.352685050798258e-05,
+      "loss": 6.7046,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.309143686502177e-05,
+      "loss": 6.7309,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.265602322206095e-05,
+      "loss": 6.6637,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.222060957910014e-05,
+      "loss": 6.4719,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.178519593613933e-05,
+      "loss": 6.7841,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.134978229317852e-05,
+      "loss": 6.6491,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.09143686502177e-05,
+      "loss": 6.8233,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.047895500725689e-05,
+      "loss": 6.5963,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0043541364296077e-05,
+      "loss": 6.7396,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9608127721335266e-05,
+      "loss": 6.8136,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9172714078374452e-05,
+      "loss": 6.545,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.873730043541364e-05,
+      "loss": 6.634,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 6.5989,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7866473149492016e-05,
+      "loss": 6.65,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7431059506531205e-05,
+      "loss": 6.6033,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.699564586357039e-05,
+      "loss": 6.4993,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.656023222060958e-05,
+      "loss": 6.4688,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6124818577648762e-05,
+      "loss": 6.7395,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5689404934687955e-05,
+      "loss": 6.6679,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5253991291727137e-05,
+      "loss": 6.7816,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4818577648766326e-05,
+      "loss": 6.5733,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4383164005805512e-05,
+      "loss": 6.6597,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.39477503628447e-05,
+      "loss": 6.6227,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3512336719883887e-05,
+      "loss": 6.6052,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3076923076923076e-05,
+      "loss": 6.4697,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.264150943396226e-05,
+      "loss": 6.7342,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.220609579100145e-05,
+      "loss": 6.6098,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1770682148040636e-05,
+      "loss": 6.5866,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1335268505079825e-05,
+      "loss": 6.6129,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089985486211901e-05,
+      "loss": 6.3888,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.04644412191582e-05,
+      "loss": 6.5824,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0029027576197386e-05,
+      "loss": 6.3109,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9593613933236575e-05,
+      "loss": 6.1473,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9158200290275757e-05,
+      "loss": 6.303,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.872278664731495e-05,
+      "loss": 6.0496,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8287373004354136e-05,
+      "loss": 6.7155,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.785195936139332e-05,
+      "loss": 6.438,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.741654571843251e-05,
+      "loss": 6.7806,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6981132075471696e-05,
+      "loss": 6.6463,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6545718432510885e-05,
+      "loss": 6.6742,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.611030478955007e-05,
+      "loss": 6.8502,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.567489114658926e-05,
+      "loss": 6.6052,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5239477503628446e-05,
+      "loss": 6.564,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4804063860667633e-05,
+      "loss": 6.5864,
+      "step": 1159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.436865021770682e-05,
+      "loss": 6.662,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3933236574746008e-05,
+      "loss": 6.787,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3497822931785195e-05,
+      "loss": 6.7509,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3062409288824381e-05,
+      "loss": 6.7778,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2626995645863568e-05,
+      "loss": 6.6383,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2191582002902756e-05,
+      "loss": 6.6005,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1756168359941943e-05,
+      "loss": 6.8618,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.132075471698113e-05,
+      "loss": 6.4564,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0885341074020318e-05,
+      "loss": 6.6056,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0449927431059506e-05,
+      "loss": 6.5271,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0014513788098693e-05,
+      "loss": 6.5189,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.579100145137879e-06,
+      "loss": 6.6099,
+      "step": 1171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.143686502177068e-06,
+      "loss": 6.5129,
+      "step": 1172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.708272859216255e-06,
+      "loss": 6.6832,
+      "step": 1173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.272859216255443e-06,
+      "loss": 6.7196,
+      "step": 1174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.83744557329463e-06,
+      "loss": 6.4536,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4020319303338166e-06,
+      "loss": 6.394,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.966618287373004e-06,
+      "loss": 6.5681,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5312046444121905e-06,
+      "loss": 6.754,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.095791001451378e-06,
+      "loss": 6.4237,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.660377358490565e-06,
+      "loss": 6.8175,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.224963715529753e-06,
+      "loss": 6.5249,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.789550072568939e-06,
+      "loss": 6.6056,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.354136429608128e-06,
+      "loss": 6.5389,
+      "step": 1183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.918722786647315e-06,
+      "loss": 6.5179,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.483309143686502e-06,
+      "loss": 6.4921,
+      "step": 1185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.047895500725689e-06,
+      "loss": 6.3926,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6124818577648764e-06,
+      "loss": 6.2185,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.177068214804064e-06,
+      "loss": 6.1024,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.741654571843251e-06,
+      "loss": 6.1685,
+      "step": 1189
+    },
     {
       "epoch": 1.0,
-      "step": 446,
+      "step": 1189,
       "total_flos": 0.0,
-      "train_loss": 6.953876193863394,
-      "train_runtime": 4987.98,
-      "train_samples_per_second": 5.721,
-      "train_steps_per_second": 0.089
+      "train_loss": 6.8302552445382005,
+      "train_runtime": 6161.6809,
+      "train_samples_per_second": 4.632,
+      "train_steps_per_second": 0.193
     }
   ],
-  "max_steps": 446,
+  "max_steps": 1189,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,