diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -9,13427 +9,13427 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 8.7901,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 8.2054,
       "step": 1
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6e-07,
-      "loss": 8.6749,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 8.1219,
       "step": 2
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2e-06,
-      "loss": 8.6797,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 8.2333,
       "step": 3
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.8e-06,
-      "loss": 8.6105,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 8.1853,
       "step": 4
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.4e-06,
-      "loss": 8.4744,
+      "learning_rate": 1.8e-07,
+      "loss": 8.1226,
       "step": 5
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.9999999999999997e-06,
-      "loss": 8.3202,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 8.0944,
       "step": 6
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.6e-06,
-      "loss": 8.2504,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 8.1681,
       "step": 7
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.2e-06,
-      "loss": 7.7477,
+      "learning_rate": 3.6e-07,
+      "loss": 8.1575,
       "step": 8
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.8e-06,
-      "loss": 7.6117,
+      "learning_rate": 4.2e-07,
+      "loss": 8.184,
       "step": 9
     },
     {
       "epoch": 0.04,
-      "learning_rate": 5.399999999999999e-06,
-      "loss": 7.1625,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 8.1123,
       "step": 10
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 7.035,
+      "learning_rate": 5.4e-07,
+      "loss": 8.0231,
       "step": 11
     },
     {
       "epoch": 0.05,
-      "learning_rate": 6.599999999999999e-06,
-      "loss": 6.8002,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 7.9512,
       "step": 12
     },
     {
       "epoch": 0.06,
-      "learning_rate": 7.2e-06,
-      "loss": 6.5852,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 7.9844,
       "step": 13
     },
     {
       "epoch": 0.06,
-      "learning_rate": 7.799999999999998e-06,
-      "loss": 6.4332,
+      "learning_rate": 7.2e-07,
+      "loss": 8.0196,
       "step": 14
     },
     {
       "epoch": 0.07,
-      "learning_rate": 8.4e-06,
-      "loss": 6.2015,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 7.7862,
       "step": 15
     },
     {
       "epoch": 0.07,
-      "learning_rate": 8.999999999999999e-06,
-      "loss": 6.037,
+      "learning_rate": 8.4e-07,
+      "loss": 7.7522,
       "step": 16
     },
     {
       "epoch": 0.08,
-      "learning_rate": 9.6e-06,
-      "loss": 5.9392,
+      "learning_rate": 9e-07,
+      "loss": 7.7857,
       "step": 17
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.02e-05,
-      "loss": 5.7954,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 7.7129,
       "step": 18
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.0799999999999998e-05,
-      "loss": 5.6019,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 7.6604,
       "step": 19
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.14e-05,
-      "loss": 5.6175,
+      "learning_rate": 1.08e-06,
+      "loss": 7.5476,
       "step": 20
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.1999999999999999e-05,
-      "loss": 5.481,
+      "learning_rate": 1.14e-06,
+      "loss": 7.4493,
       "step": 21
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.26e-05,
-      "loss": 5.4746,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 7.4508,
       "step": 22
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.26e-05,
-      "loss": 5.3,
+      "learning_rate": 1.26e-06,
+      "loss": 7.3061,
       "step": 23
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.3199999999999997e-05,
-      "loss": 5.3499,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 7.3454,
       "step": 24
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.3799999999999998e-05,
-      "loss": 5.2045,
+      "learning_rate": 1.38e-06,
+      "loss": 7.252,
       "step": 25
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.44e-05,
-      "loss": 5.2217,
+      "learning_rate": 1.44e-06,
+      "loss": 7.1289,
       "step": 26
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.4999999999999999e-05,
-      "loss": 5.0872,
+      "learning_rate": 1.5e-06,
+      "loss": 7.0522,
       "step": 27
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.5599999999999996e-05,
-      "loss": 5.1305,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 6.962,
       "step": 28
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.6199999999999997e-05,
-      "loss": 4.9678,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 6.8158,
       "step": 29
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.68e-05,
-      "loss": 5.0943,
+      "learning_rate": 1.62e-06,
+      "loss": 6.9474,
       "step": 30
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.74e-05,
-      "loss": 5.0962,
+      "learning_rate": 1.68e-06,
+      "loss": 6.8897,
       "step": 31
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.7999999999999997e-05,
-      "loss": 4.9747,
+      "learning_rate": 1.74e-06,
+      "loss": 6.821,
       "step": 32
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.8599999999999998e-05,
-      "loss": 4.9798,
+      "learning_rate": 1.8e-06,
+      "loss": 6.7075,
       "step": 33
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.92e-05,
-      "loss": 4.9415,
+      "learning_rate": 1.86e-06,
+      "loss": 6.7231,
       "step": 34
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.98e-05,
-      "loss": 4.8916,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 6.617,
       "step": 35
     },
     {
       "epoch": 0.16,
-      "learning_rate": 2.04e-05,
-      "loss": 4.9001,
+      "learning_rate": 1.98e-06,
+      "loss": 6.5733,
       "step": 36
     },
     {
       "epoch": 0.17,
-      "learning_rate": 2.1e-05,
-      "loss": 4.995,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 6.6763,
       "step": 37
     },
     {
       "epoch": 0.17,
-      "learning_rate": 2.1599999999999996e-05,
-      "loss": 4.8175,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 6.4522,
       "step": 38
     },
     {
       "epoch": 0.17,
-      "learning_rate": 2.2199999999999998e-05,
-      "loss": 4.9133,
+      "learning_rate": 2.16e-06,
+      "loss": 6.4858,
       "step": 39
     },
     {
       "epoch": 0.18,
-      "learning_rate": 2.28e-05,
-      "loss": 4.8894,
+      "learning_rate": 2.22e-06,
+      "loss": 6.4352,
       "step": 40
     },
     {
       "epoch": 0.18,
-      "learning_rate": 2.34e-05,
-      "loss": 4.8813,
+      "learning_rate": 2.28e-06,
+      "loss": 6.3867,
       "step": 41
     },
     {
       "epoch": 0.19,
-      "learning_rate": 2.3999999999999997e-05,
-      "loss": 4.9061,
+      "learning_rate": 2.34e-06,
+      "loss": 6.3603,
       "step": 42
     },
     {
       "epoch": 0.19,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 4.8308,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 6.2443,
       "step": 43
     },
     {
       "epoch": 0.2,
-      "learning_rate": 2.52e-05,
-      "loss": 4.8939,
+      "learning_rate": 2.46e-06,
+      "loss": 6.2912,
       "step": 44
     },
     {
       "epoch": 0.2,
-      "learning_rate": 2.5799999999999997e-05,
-      "loss": 4.8437,
+      "learning_rate": 2.52e-06,
+      "loss": 6.2332,
       "step": 45
     },
     {
       "epoch": 0.21,
-      "learning_rate": 2.6399999999999995e-05,
-      "loss": 4.9756,
+      "learning_rate": 2.58e-06,
+      "loss": 6.3944,
       "step": 46
     },
     {
       "epoch": 0.21,
-      "learning_rate": 2.6999999999999996e-05,
-      "loss": 4.8558,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 6.2816,
       "step": 47
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2.7599999999999997e-05,
-      "loss": 4.7816,
+      "learning_rate": 2.7e-06,
+      "loss": 6.185,
       "step": 48
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.8812,
+      "learning_rate": 2.76e-06,
+      "loss": 6.337,
       "step": 49
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2.88e-05,
-      "loss": 4.6317,
+      "learning_rate": 2.82e-06,
+      "loss": 6.1375,
       "step": 50
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.94e-05,
-      "loss": 6.1604,
+      "learning_rate": 2.88e-06,
+      "loss": 5.8827,
       "step": 51
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 6.0773,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 5.725,
       "step": 52
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.06e-05,
-      "loss": 5.8237,
+      "learning_rate": 3e-06,
+      "loss": 5.6994,
       "step": 53
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.119999999999999e-05,
-      "loss": 5.4014,
+      "learning_rate": 3.06e-06,
+      "loss": 5.6824,
       "step": 54
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.1799999999999994e-05,
-      "loss": 5.0832,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 5.6554,
       "step": 55
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.2399999999999995e-05,
-      "loss": 4.8432,
+      "learning_rate": 3.18e-06,
+      "loss": 5.5958,
       "step": 56
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.2999999999999996e-05,
-      "loss": 4.8566,
+      "learning_rate": 3.24e-06,
+      "loss": 5.5991,
       "step": 57
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.36e-05,
-      "loss": 4.8108,
+      "learning_rate": 3.3e-06,
+      "loss": 5.5462,
       "step": 58
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.42e-05,
-      "loss": 4.7935,
+      "learning_rate": 3.36e-06,
+      "loss": 5.494,
       "step": 59
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.48e-05,
-      "loss": 4.8346,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 5.4829,
       "step": 60
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.539999999999999e-05,
-      "loss": 4.8708,
+      "learning_rate": 3.48e-06,
+      "loss": 5.4825,
       "step": 61
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5999999999999994e-05,
-      "loss": 4.7303,
+      "learning_rate": 3.54e-06,
+      "loss": 5.3116,
       "step": 62
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6599999999999995e-05,
-      "loss": 4.8369,
+      "learning_rate": 3.6e-06,
+      "loss": 5.4564,
       "step": 63
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.7199999999999996e-05,
-      "loss": 4.7123,
+      "learning_rate": 3.66e-06,
+      "loss": 5.374,
       "step": 64
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.78e-05,
-      "loss": 4.6835,
+      "learning_rate": 3.72e-06,
+      "loss": 5.311,
       "step": 65
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.84e-05,
-      "loss": 4.7015,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 5.2815,
       "step": 66
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.9e-05,
-      "loss": 4.6597,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 5.2217,
       "step": 67
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.96e-05,
-      "loss": 4.6913,
+      "learning_rate": 3.9e-06,
+      "loss": 5.2679,
       "step": 68
     },
     {
       "epoch": 0.31,
-      "learning_rate": 4.02e-05,
-      "loss": 4.7459,
+      "learning_rate": 3.96e-06,
+      "loss": 5.2664,
       "step": 69
     },
     {
       "epoch": 0.31,
-      "learning_rate": 4.08e-05,
-      "loss": 4.6883,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 5.186,
       "step": 70
     },
     {
       "epoch": 0.32,
-      "learning_rate": 4.14e-05,
-      "loss": 4.6348,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 5.1406,
       "step": 71
     },
     {
       "epoch": 0.32,
-      "learning_rate": 4.2e-05,
-      "loss": 4.6861,
+      "learning_rate": 4.14e-06,
+      "loss": 5.1835,
       "step": 72
     },
     {
       "epoch": 0.33,
-      "learning_rate": 4.259999999999999e-05,
-      "loss": 4.5766,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 5.0411,
       "step": 73
     },
     {
       "epoch": 0.33,
-      "learning_rate": 4.319999999999999e-05,
-      "loss": 4.6444,
+      "learning_rate": 4.26e-06,
+      "loss": 5.0951,
       "step": 74
     },
     {
       "epoch": 0.34,
-      "learning_rate": 4.3799999999999994e-05,
-      "loss": 4.7433,
+      "learning_rate": 4.32e-06,
+      "loss": 5.1969,
       "step": 75
     },
     {
       "epoch": 0.34,
-      "learning_rate": 4.4399999999999995e-05,
-      "loss": 4.6555,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 5.1237,
       "step": 76
     },
     {
       "epoch": 0.35,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 4.5683,
+      "learning_rate": 4.44e-06,
+      "loss": 4.9818,
       "step": 77
     },
     {
       "epoch": 0.35,
-      "learning_rate": 4.56e-05,
-      "loss": 4.6298,
+      "learning_rate": 4.5e-06,
+      "loss": 5.0479,
       "step": 78
     },
     {
       "epoch": 0.35,
-      "learning_rate": 4.62e-05,
-      "loss": 4.5865,
+      "learning_rate": 4.56e-06,
+      "loss": 5.0285,
       "step": 79
     },
     {
       "epoch": 0.36,
-      "learning_rate": 4.68e-05,
-      "loss": 4.6905,
+      "learning_rate": 4.62e-06,
+      "loss": 5.0754,
       "step": 80
     },
     {
       "epoch": 0.36,
-      "learning_rate": 4.7399999999999993e-05,
-      "loss": 4.5504,
+      "learning_rate": 4.68e-06,
+      "loss": 4.9368,
       "step": 81
     },
     {
       "epoch": 0.37,
-      "learning_rate": 4.7999999999999994e-05,
-      "loss": 4.6262,
+      "learning_rate": 4.74e-06,
+      "loss": 4.9667,
       "step": 82
     },
     {
       "epoch": 0.37,
-      "learning_rate": 4.8599999999999995e-05,
-      "loss": 4.5915,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.9748,
       "step": 83
     },
     {
       "epoch": 0.38,
-      "learning_rate": 4.9199999999999997e-05,
-      "loss": 4.6035,
+      "learning_rate": 4.86e-06,
+      "loss": 4.9711,
       "step": 84
     },
     {
       "epoch": 0.38,
-      "learning_rate": 4.98e-05,
-      "loss": 4.6711,
+      "learning_rate": 4.92e-06,
+      "loss": 5.035,
       "step": 85
     },
     {
       "epoch": 0.39,
-      "learning_rate": 5.04e-05,
-      "loss": 4.493,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.8273,
       "step": 86
     },
     {
       "epoch": 0.39,
-      "learning_rate": 5.1e-05,
-      "loss": 4.6,
+      "learning_rate": 5.04e-06,
+      "loss": 4.965,
       "step": 87
     },
     {
       "epoch": 0.39,
-      "learning_rate": 5.1599999999999994e-05,
-      "loss": 4.5418,
+      "learning_rate": 5.1e-06,
+      "loss": 4.9021,
       "step": 88
     },
     {
       "epoch": 0.4,
-      "learning_rate": 5.2199999999999995e-05,
-      "loss": 4.6024,
+      "learning_rate": 5.16e-06,
+      "loss": 4.9831,
       "step": 89
     },
     {
       "epoch": 0.4,
-      "learning_rate": 5.279999999999999e-05,
-      "loss": 4.6465,
+      "learning_rate": 5.22e-06,
+      "loss": 5.0348,
       "step": 90
     },
     {
       "epoch": 0.41,
-      "learning_rate": 5.339999999999999e-05,
-      "loss": 4.5964,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 4.9893,
       "step": 91
     },
     {
       "epoch": 0.41,
-      "learning_rate": 5.399999999999999e-05,
-      "loss": 4.5889,
+      "learning_rate": 5.34e-06,
+      "loss": 4.9375,
       "step": 92
     },
     {
       "epoch": 0.42,
-      "learning_rate": 5.459999999999999e-05,
-      "loss": 4.5629,
+      "learning_rate": 5.4e-06,
+      "loss": 4.9366,
       "step": 93
     },
     {
       "epoch": 0.42,
-      "learning_rate": 5.519999999999999e-05,
-      "loss": 4.6184,
+      "learning_rate": 5.46e-06,
+      "loss": 4.9483,
       "step": 94
     },
     {
       "epoch": 0.43,
-      "learning_rate": 5.5799999999999994e-05,
-      "loss": 4.5852,
+      "learning_rate": 5.52e-06,
+      "loss": 4.9207,
       "step": 95
     },
     {
       "epoch": 0.43,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.5761,
+      "learning_rate": 5.58e-06,
+      "loss": 4.9721,
       "step": 96
     },
     {
       "epoch": 0.43,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.6631,
+      "learning_rate": 5.64e-06,
+      "loss": 5.0164,
       "step": 97
     },
     {
       "epoch": 0.44,
-      "learning_rate": 5.76e-05,
-      "loss": 4.7284,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 5.1602,
       "step": 98
     },
     {
       "epoch": 0.44,
-      "learning_rate": 5.82e-05,
-      "loss": 4.5872,
+      "learning_rate": 5.76e-06,
+      "loss": 4.9884,
       "step": 99
     },
     {
       "epoch": 0.45,
-      "learning_rate": 5.88e-05,
-      "loss": 4.5276,
+      "learning_rate": 5.82e-06,
+      "loss": 4.9934,
       "step": 100
     },
     {
       "epoch": 0.45,
-      "learning_rate": 5.94e-05,
-      "loss": 5.9137,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 5.4184,
       "step": 101
     },
     {
       "epoch": 0.46,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 5.8072,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 5.4165,
       "step": 102
     },
     {
       "epoch": 0.46,
-      "learning_rate": 6.0599999999999996e-05,
-      "loss": 5.5062,
+      "learning_rate": 6e-06,
+      "loss": 5.4173,
       "step": 103
     },
     {
       "epoch": 0.47,
-      "learning_rate": 6.12e-05,
-      "loss": 5.1048,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 5.3571,
       "step": 104
     },
     {
       "epoch": 0.47,
-      "learning_rate": 6.18e-05,
-      "loss": 4.9539,
+      "learning_rate": 6.12e-06,
+      "loss": 5.3026,
       "step": 105
     },
     {
       "epoch": 0.48,
-      "learning_rate": 6.239999999999999e-05,
-      "loss": 4.8815,
+      "learning_rate": 6.18e-06,
+      "loss": 5.2382,
       "step": 106
     },
     {
       "epoch": 0.48,
-      "learning_rate": 6.299999999999999e-05,
-      "loss": 4.6952,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 5.0287,
       "step": 107
     },
     {
       "epoch": 0.48,
-      "learning_rate": 6.359999999999999e-05,
-      "loss": 4.7131,
+      "learning_rate": 6.3e-06,
+      "loss": 5.0118,
       "step": 108
     },
     {
       "epoch": 0.49,
-      "learning_rate": 6.419999999999999e-05,
-      "loss": 4.6939,
+      "learning_rate": 6.36e-06,
+      "loss": 4.952,
       "step": 109
     },
     {
       "epoch": 0.49,
-      "learning_rate": 6.479999999999999e-05,
-      "loss": 4.6896,
+      "learning_rate": 6.42e-06,
+      "loss": 4.9043,
       "step": 110
     },
     {
       "epoch": 0.5,
-      "learning_rate": 6.539999999999999e-05,
-      "loss": 4.6205,
+      "learning_rate": 6.48e-06,
+      "loss": 4.8084,
       "step": 111
     },
     {
       "epoch": 0.5,
-      "learning_rate": 6.599999999999999e-05,
-      "loss": 4.6257,
+      "learning_rate": 6.54e-06,
+      "loss": 4.8808,
       "step": 112
     },
     {
       "epoch": 0.51,
-      "learning_rate": 6.659999999999999e-05,
-      "loss": 4.63,
+      "learning_rate": 6.6e-06,
+      "loss": 4.8721,
       "step": 113
     },
     {
       "epoch": 0.51,
-      "learning_rate": 6.72e-05,
-      "loss": 4.5141,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.7812,
       "step": 114
     },
     {
       "epoch": 0.52,
-      "learning_rate": 6.78e-05,
-      "loss": 4.5986,
+      "learning_rate": 6.72e-06,
+      "loss": 4.8802,
       "step": 115
     },
     {
       "epoch": 0.52,
-      "learning_rate": 6.84e-05,
-      "loss": 4.55,
+      "learning_rate": 6.78e-06,
+      "loss": 4.8304,
       "step": 116
     },
     {
       "epoch": 0.52,
-      "learning_rate": 6.9e-05,
-      "loss": 4.5866,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 4.88,
       "step": 117
     },
     {
       "epoch": 0.53,
-      "learning_rate": 6.96e-05,
-      "loss": 4.5122,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.8054,
       "step": 118
     },
     {
       "epoch": 0.53,
-      "learning_rate": 7.02e-05,
-      "loss": 4.5169,
+      "learning_rate": 6.96e-06,
+      "loss": 4.7917,
       "step": 119
     },
     {
       "epoch": 0.54,
-      "learning_rate": 7.079999999999999e-05,
-      "loss": 4.5464,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.8489,
       "step": 120
     },
     {
       "epoch": 0.54,
-      "learning_rate": 7.139999999999999e-05,
-      "loss": 4.4779,
+      "learning_rate": 7.08e-06,
+      "loss": 4.7469,
       "step": 121
     },
     {
       "epoch": 0.55,
-      "learning_rate": 7.199999999999999e-05,
-      "loss": 4.517,
+      "learning_rate": 7.14e-06,
+      "loss": 4.7927,
       "step": 122
     },
     {
       "epoch": 0.55,
-      "learning_rate": 7.259999999999999e-05,
-      "loss": 4.4728,
+      "learning_rate": 7.2e-06,
+      "loss": 4.7484,
       "step": 123
     },
     {
       "epoch": 0.56,
-      "learning_rate": 7.319999999999999e-05,
-      "loss": 4.4484,
+      "learning_rate": 7.26e-06,
+      "loss": 4.7096,
       "step": 124
     },
     {
       "epoch": 0.56,
-      "learning_rate": 7.379999999999999e-05,
-      "loss": 4.6016,
+      "learning_rate": 7.32e-06,
+      "loss": 4.8926,
       "step": 125
     },
     {
       "epoch": 0.57,
-      "learning_rate": 7.439999999999999e-05,
-      "loss": 4.4771,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.763,
       "step": 126
     },
     {
       "epoch": 0.57,
-      "learning_rate": 7.5e-05,
-      "loss": 4.5307,
+      "learning_rate": 7.44e-06,
+      "loss": 4.82,
       "step": 127
     },
     {
       "epoch": 0.57,
-      "learning_rate": 7.56e-05,
-      "loss": 4.5465,
+      "learning_rate": 7.5e-06,
+      "loss": 4.8304,
       "step": 128
     },
     {
       "epoch": 0.58,
-      "learning_rate": 7.62e-05,
-      "loss": 4.5136,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.8098,
       "step": 129
     },
     {
       "epoch": 0.58,
-      "learning_rate": 7.68e-05,
-      "loss": 4.5539,
+      "learning_rate": 7.62e-06,
+      "loss": 4.8215,
       "step": 130
     },
     {
       "epoch": 0.59,
-      "learning_rate": 7.74e-05,
-      "loss": 4.549,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.8154,
       "step": 131
     },
     {
       "epoch": 0.59,
-      "learning_rate": 7.8e-05,
-      "loss": 4.3814,
+      "learning_rate": 7.74e-06,
+      "loss": 4.6749,
       "step": 132
     },
     {
       "epoch": 0.6,
-      "learning_rate": 7.86e-05,
-      "loss": 4.3906,
+      "learning_rate": 7.8e-06,
+      "loss": 4.6698,
       "step": 133
     },
     {
       "epoch": 0.6,
-      "learning_rate": 7.92e-05,
-      "loss": 4.492,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.7338,
       "step": 134
     },
     {
       "epoch": 0.61,
-      "learning_rate": 7.98e-05,
-      "loss": 4.5401,
+      "learning_rate": 7.92e-06,
+      "loss": 4.7836,
       "step": 135
     },
     {
       "epoch": 0.61,
-      "learning_rate": 8.04e-05,
-      "loss": 4.4804,
+      "learning_rate": 7.98e-06,
+      "loss": 4.7491,
       "step": 136
     },
     {
       "epoch": 0.61,
-      "learning_rate": 8.1e-05,
-      "loss": 4.5568,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.8325,
       "step": 137
     },
     {
       "epoch": 0.62,
-      "learning_rate": 8.16e-05,
-      "loss": 4.4821,
+      "learning_rate": 8.1e-06,
+      "loss": 4.7691,
       "step": 138
     },
     {
       "epoch": 0.62,
-      "learning_rate": 8.22e-05,
-      "loss": 4.4264,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 4.6887,
       "step": 139
     },
     {
       "epoch": 0.63,
-      "learning_rate": 8.28e-05,
-      "loss": 4.5104,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.8142,
       "step": 140
     },
     {
       "epoch": 0.63,
-      "learning_rate": 8.34e-05,
-      "loss": 4.5322,
+      "learning_rate": 8.28e-06,
+      "loss": 4.7964,
       "step": 141
     },
     {
       "epoch": 0.64,
-      "learning_rate": 8.4e-05,
-      "loss": 4.5441,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 4.816,
       "step": 142
     },
     {
       "epoch": 0.64,
-      "learning_rate": 8.459999999999998e-05,
-      "loss": 4.6673,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.9627,
       "step": 143
     },
     {
       "epoch": 0.65,
-      "learning_rate": 8.519999999999998e-05,
-      "loss": 4.5759,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 4.884,
       "step": 144
     },
     {
       "epoch": 0.65,
-      "learning_rate": 8.579999999999998e-05,
-      "loss": 4.4321,
+      "learning_rate": 8.52e-06,
+      "loss": 4.7096,
       "step": 145
     },
     {
       "epoch": 0.65,
-      "learning_rate": 8.639999999999999e-05,
-      "loss": 4.4123,
+      "learning_rate": 8.58e-06,
+      "loss": 4.7363,
       "step": 146
     },
     {
       "epoch": 0.66,
-      "learning_rate": 8.699999999999999e-05,
-      "loss": 4.3866,
+      "learning_rate": 8.64e-06,
+      "loss": 4.7037,
       "step": 147
     },
     {
       "epoch": 0.66,
-      "learning_rate": 8.759999999999999e-05,
-      "loss": 4.6419,
+      "learning_rate": 8.7e-06,
+      "loss": 5.0372,
       "step": 148
     },
     {
       "epoch": 0.67,
-      "learning_rate": 8.819999999999999e-05,
-      "loss": 4.44,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 4.8017,
       "step": 149
     },
     {
       "epoch": 0.67,
-      "learning_rate": 8.879999999999999e-05,
-      "loss": 4.4715,
+      "learning_rate": 8.82e-06,
+      "loss": 4.9157,
       "step": 150
     },
     {
       "epoch": 0.68,
-      "learning_rate": 8.939999999999999e-05,
-      "loss": 6.287,
+      "learning_rate": 8.88e-06,
+      "loss": 5.2338,
       "step": 151
     },
     {
       "epoch": 0.68,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 6.1091,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 5.2328,
       "step": 152
     },
     {
       "epoch": 0.69,
-      "learning_rate": 9.059999999999999e-05,
-      "loss": 5.5967,
+      "learning_rate": 9e-06,
+      "loss": 5.2131,
       "step": 153
     },
     {
       "epoch": 0.69,
-      "learning_rate": 9.12e-05,
-      "loss": 5.0873,
+      "learning_rate": 9.06e-06,
+      "loss": 5.0913,
       "step": 154
     },
     {
       "epoch": 0.7,
-      "learning_rate": 9.18e-05,
-      "loss": 4.8993,
+      "learning_rate": 9.12e-06,
+      "loss": 5.1419,
       "step": 155
     },
     {
       "epoch": 0.7,
-      "learning_rate": 9.24e-05,
-      "loss": 4.8036,
+      "learning_rate": 9.18e-06,
+      "loss": 5.0874,
       "step": 156
     },
     {
       "epoch": 0.7,
-      "learning_rate": 9.3e-05,
-      "loss": 4.7089,
+      "learning_rate": 9.24e-06,
+      "loss": 4.9494,
       "step": 157
     },
     {
       "epoch": 0.71,
-      "learning_rate": 9.36e-05,
-      "loss": 4.658,
+      "learning_rate": 9.3e-06,
+      "loss": 4.8881,
       "step": 158
     },
     {
       "epoch": 0.71,
-      "learning_rate": 9.419999999999999e-05,
-      "loss": 4.5872,
+      "learning_rate": 9.36e-06,
+      "loss": 4.7511,
       "step": 159
     },
     {
       "epoch": 0.72,
-      "learning_rate": 9.479999999999999e-05,
-      "loss": 4.6761,
+      "learning_rate": 9.42e-06,
+      "loss": 4.81,
       "step": 160
     },
     {
       "epoch": 0.72,
-      "learning_rate": 9.539999999999999e-05,
-      "loss": 4.5023,
+      "learning_rate": 9.48e-06,
+      "loss": 4.6441,
       "step": 161
     },
     {
       "epoch": 0.73,
-      "learning_rate": 9.599999999999999e-05,
-      "loss": 4.5411,
+      "learning_rate": 9.54e-06,
+      "loss": 4.6864,
       "step": 162
     },
     {
       "epoch": 0.73,
-      "learning_rate": 9.659999999999999e-05,
-      "loss": 4.5728,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.7303,
       "step": 163
     },
     {
       "epoch": 0.74,
-      "learning_rate": 9.719999999999999e-05,
-      "loss": 4.5043,
+      "learning_rate": 9.66e-06,
+      "loss": 4.7054,
       "step": 164
     },
     {
       "epoch": 0.74,
-      "learning_rate": 9.779999999999999e-05,
-      "loss": 4.5902,
+      "learning_rate": 9.72e-06,
+      "loss": 4.8041,
       "step": 165
     },
     {
       "epoch": 0.74,
-      "learning_rate": 9.839999999999999e-05,
-      "loss": 4.4932,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.7566,
       "step": 166
     },
     {
       "epoch": 0.75,
-      "learning_rate": 9.9e-05,
-      "loss": 4.5148,
+      "learning_rate": 9.84e-06,
+      "loss": 4.7286,
       "step": 167
     },
     {
       "epoch": 0.75,
-      "learning_rate": 9.96e-05,
-      "loss": 4.5009,
+      "learning_rate": 9.9e-06,
+      "loss": 4.7454,
       "step": 168
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0001002,
-      "loss": 4.3554,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.5812,
       "step": 169
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0001008,
-      "loss": 4.5545,
+      "learning_rate": 1.002e-05,
+      "loss": 4.802,
       "step": 170
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0001014,
-      "loss": 4.4837,
+      "learning_rate": 1.008e-05,
+      "loss": 4.724,
       "step": 171
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.000102,
-      "loss": 4.3605,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 4.5918,
       "step": 172
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0001026,
-      "loss": 4.423,
+      "learning_rate": 1.02e-05,
+      "loss": 4.6252,
       "step": 173
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00010319999999999999,
-      "loss": 4.3804,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 4.646,
       "step": 174
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00010379999999999999,
-      "loss": 4.4725,
+      "learning_rate": 1.032e-05,
+      "loss": 4.7106,
       "step": 175
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00010439999999999999,
-      "loss": 4.4212,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 4.6334,
       "step": 176
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00010499999999999999,
-      "loss": 4.4313,
+      "learning_rate": 1.044e-05,
+      "loss": 4.6837,
       "step": 177
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00010559999999999998,
-      "loss": 4.3636,
+      "learning_rate": 1.05e-05,
+      "loss": 4.6227,
       "step": 178
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00010619999999999998,
-      "loss": 4.4336,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 4.6745,
       "step": 179
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00010679999999999998,
-      "loss": 4.3958,
+      "learning_rate": 1.062e-05,
+      "loss": 4.6251,
       "step": 180
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00010739999999999998,
-      "loss": 4.529,
+      "learning_rate": 1.068e-05,
+      "loss": 4.7493,
       "step": 181
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00010799999999999998,
-      "loss": 4.3276,
+      "learning_rate": 1.074e-05,
+      "loss": 4.5725,
       "step": 182
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00010859999999999998,
-      "loss": 4.5321,
+      "learning_rate": 1.08e-05,
+      "loss": 4.7591,
       "step": 183
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00010919999999999998,
-      "loss": 4.4487,
+      "learning_rate": 1.086e-05,
+      "loss": 4.6551,
       "step": 184
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00010979999999999999,
-      "loss": 4.4949,
+      "learning_rate": 1.092e-05,
+      "loss": 4.7288,
       "step": 185
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00011039999999999999,
-      "loss": 4.4117,
+      "learning_rate": 1.098e-05,
+      "loss": 4.6177,
       "step": 186
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00011099999999999999,
-      "loss": 4.407,
+      "learning_rate": 1.104e-05,
+      "loss": 4.6272,
       "step": 187
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00011159999999999999,
-      "loss": 4.4996,
+      "learning_rate": 1.11e-05,
+      "loss": 4.7299,
       "step": 188
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00011219999999999999,
-      "loss": 4.3983,
+      "learning_rate": 1.116e-05,
+      "loss": 4.6558,
       "step": 189
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00011279999999999999,
-      "loss": 4.4501,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 4.6845,
       "step": 190
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00011339999999999999,
-      "loss": 4.4659,
+      "learning_rate": 1.128e-05,
+      "loss": 4.7153,
       "step": 191
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00011399999999999999,
-      "loss": 4.3934,
+      "learning_rate": 1.134e-05,
+      "loss": 4.639,
       "step": 192
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0001146,
-      "loss": 4.3729,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.6371,
       "step": 193
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0001152,
-      "loss": 4.4194,
+      "learning_rate": 1.146e-05,
+      "loss": 4.6644,
       "step": 194
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0001158,
-      "loss": 4.4216,
+      "learning_rate": 1.152e-05,
+      "loss": 4.6621,
       "step": 195
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0001164,
-      "loss": 4.4952,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 4.7406,
       "step": 196
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.000117,
-      "loss": 4.4412,
+      "learning_rate": 1.164e-05,
+      "loss": 4.7166,
       "step": 197
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0001176,
-      "loss": 4.4064,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 4.6911,
       "step": 198
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0001182,
-      "loss": 4.4579,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 4.8327,
       "step": 199
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0001188,
-      "loss": 4.2313,
+      "learning_rate": 1.182e-05,
+      "loss": 4.5631,
       "step": 200
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0001194,
-      "loss": 5.4646,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 5.0977,
       "step": 201
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 5.4181,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 5.2063,
       "step": 202
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00012059999999999999,
-      "loss": 5.0932,
+      "learning_rate": 1.2e-05,
+      "loss": 5.2377,
       "step": 203
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00012119999999999999,
-      "loss": 4.8642,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 5.2376,
       "step": 204
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00012179999999999999,
-      "loss": 4.6482,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 5.1215,
       "step": 205
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0001224,
-      "loss": 4.7276,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 5.1138,
       "step": 206
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00012299999999999998,
-      "loss": 4.6034,
+      "learning_rate": 1.224e-05,
+      "loss": 4.919,
       "step": 207
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0001236,
-      "loss": 4.58,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.8602,
       "step": 208
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00012419999999999998,
-      "loss": 4.4605,
+      "learning_rate": 1.236e-05,
+      "loss": 4.7021,
       "step": 209
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00012479999999999997,
-      "loss": 4.4878,
+      "learning_rate": 1.242e-05,
+      "loss": 4.6941,
       "step": 210
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00012539999999999999,
-      "loss": 4.502,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.6834,
       "step": 211
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00012599999999999997,
-      "loss": 4.5168,
+      "learning_rate": 1.254e-05,
+      "loss": 4.7128,
       "step": 212
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0001266,
-      "loss": 4.5251,
+      "learning_rate": 1.26e-05,
+      "loss": 4.7082,
       "step": 213
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00012719999999999997,
-      "loss": 4.4259,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.6521,
       "step": 214
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0001278,
-      "loss": 4.3709,
+      "learning_rate": 1.272e-05,
+      "loss": 4.6062,
       "step": 215
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00012839999999999998,
-      "loss": 4.5284,
+      "learning_rate": 1.278e-05,
+      "loss": 4.7968,
       "step": 216
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.000129,
-      "loss": 4.3736,
+      "learning_rate": 1.284e-05,
+      "loss": 4.6265,
       "step": 217
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00012959999999999998,
-      "loss": 4.3609,
+      "learning_rate": 1.29e-05,
+      "loss": 4.645,
       "step": 218
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0001302,
-      "loss": 4.41,
+      "learning_rate": 1.296e-05,
+      "loss": 4.7365,
       "step": 219
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00013079999999999998,
-      "loss": 4.4306,
+      "learning_rate": 1.302e-05,
+      "loss": 4.712,
       "step": 220
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0001314,
-      "loss": 4.4937,
+      "learning_rate": 1.308e-05,
+      "loss": 4.8344,
       "step": 221
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00013199999999999998,
-      "loss": 4.3587,
+      "learning_rate": 1.314e-05,
+      "loss": 4.7704,
       "step": 222
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0001326,
-      "loss": 4.3467,
+      "learning_rate": 1.32e-05,
+      "loss": 4.8638,
       "step": 223
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00013319999999999999,
-      "loss": 5.8076,
+      "learning_rate": 1.326e-05,
+      "loss": 4.6617,
       "step": 224
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0001338,
-      "loss": 5.7454,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.5823,
       "step": 225
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0001344,
-      "loss": 5.6168,
+      "learning_rate": 1.338e-05,
+      "loss": 4.6875,
       "step": 226
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.000135,
-      "loss": 5.1562,
+      "learning_rate": 1.344e-05,
+      "loss": 4.6017,
       "step": 227
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0001356,
-      "loss": 4.9135,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.6531,
       "step": 228
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0001362,
-      "loss": 4.6708,
+      "learning_rate": 1.356e-05,
+      "loss": 4.6418,
       "step": 229
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0001368,
-      "loss": 4.5292,
+      "learning_rate": 1.362e-05,
+      "loss": 4.605,
       "step": 230
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0001374,
-      "loss": 4.4944,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.5926,
       "step": 231
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.000138,
-      "loss": 4.3521,
+      "learning_rate": 1.374e-05,
+      "loss": 4.5042,
       "step": 232
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0001386,
-      "loss": 4.4591,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.6279,
       "step": 233
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0001392,
-      "loss": 4.3587,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 4.5776,
       "step": 234
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00013979999999999998,
-      "loss": 4.3937,
+      "learning_rate": 1.392e-05,
+      "loss": 4.6267,
       "step": 235
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0001404,
-      "loss": 4.3795,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 4.6372,
       "step": 236
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00014099999999999998,
-      "loss": 4.2489,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 4.5405,
       "step": 237
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00014159999999999997,
-      "loss": 4.3007,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.6718,
       "step": 238
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0001422,
-      "loss": 4.2813,
+      "learning_rate": 1.416e-05,
+      "loss": 4.5861,
       "step": 239
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00014279999999999997,
-      "loss": 4.3177,
+      "learning_rate": 1.422e-05,
+      "loss": 4.6616,
       "step": 240
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0001434,
-      "loss": 4.0955,
+      "learning_rate": 1.428e-05,
+      "loss": 4.5046,
       "step": 241
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00014399999999999998,
-      "loss": 4.1885,
+      "learning_rate": 1.434e-05,
+      "loss": 4.5972,
       "step": 242
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0001446,
-      "loss": 4.1224,
+      "learning_rate": 1.44e-05,
+      "loss": 4.4748,
       "step": 243
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00014519999999999998,
-      "loss": 4.2549,
+      "learning_rate": 1.446e-05,
+      "loss": 4.6557,
       "step": 244
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0001458,
-      "loss": 4.1249,
+      "learning_rate": 1.452e-05,
+      "loss": 4.5391,
       "step": 245
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00014639999999999998,
-      "loss": 4.1269,
+      "learning_rate": 1.458e-05,
+      "loss": 4.5457,
       "step": 246
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.000147,
-      "loss": 4.1725,
+      "learning_rate": 1.464e-05,
+      "loss": 4.529,
       "step": 247
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00014759999999999998,
-      "loss": 4.2151,
+      "learning_rate": 1.47e-05,
+      "loss": 4.5921,
       "step": 248
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0001482,
-      "loss": 4.1098,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 4.5124,
       "step": 249
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00014879999999999998,
-      "loss": 4.1609,
+      "learning_rate": 1.482e-05,
+      "loss": 4.5682,
       "step": 250
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0001494,
-      "loss": 4.0873,
+      "learning_rate": 1.488e-05,
+      "loss": 4.4989,
       "step": 251
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00015,
-      "loss": 4.0377,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 4.424,
       "step": 252
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00015059999999999997,
-      "loss": 4.1351,
+      "learning_rate": 1.5e-05,
+      "loss": 4.5844,
       "step": 253
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0001512,
-      "loss": 4.0665,
+      "learning_rate": 1.506e-05,
+      "loss": 4.5445,
       "step": 254
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00015179999999999998,
-      "loss": 4.0937,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 4.5088,
       "step": 255
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0001524,
-      "loss": 4.079,
+      "learning_rate": 1.518e-05,
+      "loss": 4.5028,
       "step": 256
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00015299999999999998,
-      "loss": 4.0957,
+      "learning_rate": 1.524e-05,
+      "loss": 4.5224,
       "step": 257
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0001536,
-      "loss": 4.1556,
+      "learning_rate": 1.53e-05,
+      "loss": 4.5915,
       "step": 258
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00015419999999999998,
-      "loss": 4.0854,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 4.4791,
       "step": 259
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0001548,
-      "loss": 4.0407,
+      "learning_rate": 1.542e-05,
+      "loss": 4.493,
       "step": 260
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00015539999999999998,
-      "loss": 4.1202,
+      "learning_rate": 1.548e-05,
+      "loss": 4.5977,
       "step": 261
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.000156,
-      "loss": 4.0466,
+      "learning_rate": 1.554e-05,
+      "loss": 4.5526,
       "step": 262
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00015659999999999998,
-      "loss": 3.8976,
+      "learning_rate": 1.56e-05,
+      "loss": 4.4401,
       "step": 263
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0001572,
-      "loss": 4.0776,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 4.591,
       "step": 264
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0001578,
-      "loss": 3.9628,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 4.4974,
       "step": 265
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0001584,
-      "loss": 4.0677,
+      "learning_rate": 1.578e-05,
+      "loss": 4.576,
       "step": 266
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.000159,
-      "loss": 3.9569,
+      "learning_rate": 1.584e-05,
+      "loss": 4.5276,
       "step": 267
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001596,
-      "loss": 4.0117,
+      "learning_rate": 1.59e-05,
+      "loss": 4.5828,
       "step": 268
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001602,
-      "loss": 3.835,
+      "learning_rate": 1.596e-05,
+      "loss": 4.4746,
       "step": 269
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001608,
-      "loss": 3.8285,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 4.4368,
       "step": 270
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0001614,
-      "loss": 3.9137,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 4.5396,
       "step": 271
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.000162,
-      "loss": 3.9028,
+      "learning_rate": 1.614e-05,
+      "loss": 4.5714,
       "step": 272
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0001626,
-      "loss": 3.6852,
+      "learning_rate": 1.62e-05,
+      "loss": 4.4684,
       "step": 273
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001632,
-      "loss": 6.3809,
+      "learning_rate": 1.626e-05,
+      "loss": 5.3867,
       "step": 274
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001638,
-      "loss": 5.6719,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 5.2534,
       "step": 275
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001644,
-      "loss": 5.153,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 5.2386,
       "step": 276
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.000165,
-      "loss": 4.856,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 5.1946,
       "step": 277
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0001656,
-      "loss": 4.6105,
+      "learning_rate": 1.65e-05,
+      "loss": 5.0371,
       "step": 278
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0001662,
-      "loss": 4.6108,
+      "learning_rate": 1.656e-05,
+      "loss": 5.035,
       "step": 279
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001668,
-      "loss": 4.5733,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 4.9167,
       "step": 280
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001674,
-      "loss": 4.4,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 4.7318,
       "step": 281
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.000168,
-      "loss": 4.4329,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 4.7087,
       "step": 282
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001686,
-      "loss": 4.3477,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.5858,
       "step": 283
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00016919999999999997,
-      "loss": 4.2902,
+      "learning_rate": 1.686e-05,
+      "loss": 4.5477,
       "step": 284
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00016979999999999998,
-      "loss": 4.3448,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 4.5948,
       "step": 285
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00017039999999999997,
-      "loss": 4.2302,
+      "learning_rate": 1.698e-05,
+      "loss": 4.5715,
       "step": 286
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00017099999999999998,
-      "loss": 4.2629,
+      "learning_rate": 1.704e-05,
+      "loss": 4.5578,
       "step": 287
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00017159999999999997,
-      "loss": 4.2435,
+      "learning_rate": 1.71e-05,
+      "loss": 4.5813,
       "step": 288
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00017219999999999998,
-      "loss": 4.1274,
+      "learning_rate": 1.716e-05,
+      "loss": 4.5517,
       "step": 289
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00017279999999999997,
-      "loss": 4.28,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 4.6325,
       "step": 290
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00017339999999999996,
-      "loss": 4.1498,
+      "learning_rate": 1.728e-05,
+      "loss": 4.5377,
       "step": 291
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00017399999999999997,
-      "loss": 4.1554,
+      "learning_rate": 1.734e-05,
+      "loss": 4.5294,
       "step": 292
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00017459999999999996,
-      "loss": 4.1921,
+      "learning_rate": 1.74e-05,
+      "loss": 4.5711,
       "step": 293
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00017519999999999998,
-      "loss": 4.1944,
+      "learning_rate": 1.746e-05,
+      "loss": 4.4966,
       "step": 294
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00017579999999999996,
-      "loss": 4.1347,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 4.4795,
       "step": 295
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00017639999999999998,
-      "loss": 4.1154,
+      "learning_rate": 1.758e-05,
+      "loss": 4.4509,
       "step": 296
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00017699999999999997,
-      "loss": 4.0972,
+      "learning_rate": 1.764e-05,
+      "loss": 4.458,
       "step": 297
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00017759999999999998,
-      "loss": 4.122,
+      "learning_rate": 1.77e-05,
+      "loss": 4.4804,
       "step": 298
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00017819999999999997,
-      "loss": 4.1743,
+      "learning_rate": 1.77e-05,
+      "loss": 4.5604,
       "step": 299
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00017879999999999998,
-      "loss": 4.1579,
+      "learning_rate": 1.776e-05,
+      "loss": 4.5135,
       "step": 300
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00017939999999999997,
-      "loss": 4.1053,
+      "learning_rate": 1.782e-05,
+      "loss": 4.4916,
       "step": 301
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 4.0905,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 4.4833,
       "step": 302
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018059999999999997,
-      "loss": 4.1271,
+      "learning_rate": 1.794e-05,
+      "loss": 4.5261,
       "step": 303
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018119999999999999,
-      "loss": 4.0824,
+      "learning_rate": 1.8e-05,
+      "loss": 4.4753,
       "step": 304
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018179999999999997,
-      "loss": 4.0832,
+      "learning_rate": 1.806e-05,
+      "loss": 4.4629,
       "step": 305
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001824,
-      "loss": 4.0989,
+      "learning_rate": 1.812e-05,
+      "loss": 4.5449,
       "step": 306
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00018299999999999998,
-      "loss": 4.137,
+      "learning_rate": 1.818e-05,
+      "loss": 4.5235,
       "step": 307
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0001836,
-      "loss": 4.1543,
+      "learning_rate": 1.824e-05,
+      "loss": 4.5807,
       "step": 308
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00018419999999999998,
-      "loss": 4.0891,
+      "learning_rate": 1.83e-05,
+      "loss": 4.5404,
       "step": 309
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0001848,
-      "loss": 4.0728,
+      "learning_rate": 1.836e-05,
+      "loss": 4.5097,
       "step": 310
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00018539999999999998,
-      "loss": 3.9376,
+      "learning_rate": 1.842e-05,
+      "loss": 4.4021,
       "step": 311
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.000186,
-      "loss": 3.8833,
+      "learning_rate": 1.848e-05,
+      "loss": 4.3692,
       "step": 312
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00018659999999999998,
-      "loss": 4.0698,
+      "learning_rate": 1.854e-05,
+      "loss": 4.5206,
       "step": 313
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001872,
-      "loss": 3.9994,
+      "learning_rate": 1.86e-05,
+      "loss": 4.5324,
       "step": 314
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00018779999999999998,
-      "loss": 4.0358,
+      "learning_rate": 1.866e-05,
+      "loss": 4.5604,
       "step": 315
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00018839999999999997,
-      "loss": 3.9731,
+      "learning_rate": 1.872e-05,
+      "loss": 4.4855,
       "step": 316
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00018899999999999999,
-      "loss": 3.787,
+      "learning_rate": 1.878e-05,
+      "loss": 4.3344,
       "step": 317
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00018959999999999997,
-      "loss": 3.8781,
+      "learning_rate": 1.884e-05,
+      "loss": 4.4,
       "step": 318
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001902,
-      "loss": 3.9194,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 4.5497,
       "step": 319
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00019079999999999998,
-      "loss": 3.8176,
+      "learning_rate": 1.896e-05,
+      "loss": 4.4678,
       "step": 320
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001914,
-      "loss": 3.7146,
+      "learning_rate": 1.902e-05,
+      "loss": 4.4474,
       "step": 321
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00019199999999999998,
-      "loss": 3.8007,
+      "learning_rate": 1.908e-05,
+      "loss": 4.53,
       "step": 322
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0001926,
-      "loss": 3.8317,
+      "learning_rate": 1.914e-05,
+      "loss": 4.5706,
       "step": 323
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00019319999999999998,
-      "loss": 5.473,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 5.1229,
       "step": 324
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0001938,
-      "loss": 5.3192,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 5.1525,
       "step": 325
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00019439999999999998,
-      "loss": 5.0983,
+      "learning_rate": 1.932e-05,
+      "loss": 5.2277,
       "step": 326
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.000195,
-      "loss": 4.7318,
+      "learning_rate": 1.938e-05,
+      "loss": 5.0693,
       "step": 327
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00019559999999999998,
-      "loss": 4.6187,
+      "learning_rate": 1.944e-05,
+      "loss": 4.9382,
       "step": 328
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001962,
-      "loss": 4.4429,
+      "learning_rate": 1.95e-05,
+      "loss": 4.8279,
       "step": 329
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00019679999999999999,
-      "loss": 4.3152,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 4.7044,
       "step": 330
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001974,
-      "loss": 4.4061,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 4.7312,
       "step": 331
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.000198,
-      "loss": 4.3511,
+      "learning_rate": 1.968e-05,
+      "loss": 4.6491,
       "step": 332
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0001986,
-      "loss": 4.3352,
+      "learning_rate": 1.974e-05,
+      "loss": 4.598,
       "step": 333
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001992,
-      "loss": 4.2468,
+      "learning_rate": 1.98e-05,
+      "loss": 4.5046,
       "step": 334
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001998,
-      "loss": 4.2837,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 4.5695,
       "step": 335
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0002004,
-      "loss": 4.1831,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 4.4972,
       "step": 336
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.000201,
-      "loss": 4.2178,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 4.5461,
       "step": 337
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0002016,
-      "loss": 4.1241,
+      "learning_rate": 2.004e-05,
+      "loss": 4.4988,
       "step": 338
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0002022,
-      "loss": 4.1374,
+      "learning_rate": 2.01e-05,
+      "loss": 4.4832,
       "step": 339
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0002028,
-      "loss": 4.0581,
+      "learning_rate": 2.016e-05,
+      "loss": 4.4368,
       "step": 340
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00020339999999999998,
-      "loss": 4.1244,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 4.5237,
       "step": 341
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.000204,
-      "loss": 4.101,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 4.4749,
       "step": 342
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00020459999999999999,
-      "loss": 4.0908,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 4.509,
       "step": 343
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0002052,
-      "loss": 3.9862,
+      "learning_rate": 2.04e-05,
+      "loss": 4.4118,
       "step": 344
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0002058,
-      "loss": 4.1008,
+      "learning_rate": 2.046e-05,
+      "loss": 4.5027,
       "step": 345
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00020639999999999998,
-      "loss": 4.0601,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 4.4594,
       "step": 346
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00020699999999999996,
-      "loss": 4.063,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 4.4279,
       "step": 347
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00020759999999999998,
-      "loss": 4.0309,
+      "learning_rate": 2.064e-05,
+      "loss": 4.4261,
       "step": 348
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00020819999999999996,
-      "loss": 4.0944,
+      "learning_rate": 2.07e-05,
+      "loss": 4.5078,
       "step": 349
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00020879999999999998,
-      "loss": 4.0536,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 4.454,
       "step": 350
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00020939999999999997,
-      "loss": 4.0323,
+      "learning_rate": 2.082e-05,
+      "loss": 4.4375,
       "step": 351
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00020999999999999998,
-      "loss": 4.041,
+      "learning_rate": 2.088e-05,
+      "loss": 4.4272,
       "step": 352
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00021059999999999997,
-      "loss": 4.0926,
+      "learning_rate": 2.094e-05,
+      "loss": 4.4982,
       "step": 353
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00021119999999999996,
-      "loss": 4.053,
+      "learning_rate": 2.1e-05,
+      "loss": 4.4815,
       "step": 354
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00021179999999999997,
-      "loss": 3.9695,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 4.3681,
       "step": 355
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00021239999999999996,
-      "loss": 3.9604,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 4.3848,
       "step": 356
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00021299999999999997,
-      "loss": 3.978,
+      "learning_rate": 2.118e-05,
+      "loss": 4.4376,
       "step": 357
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00021359999999999996,
-      "loss": 3.9869,
+      "learning_rate": 2.124e-05,
+      "loss": 4.4052,
       "step": 358
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00021419999999999998,
-      "loss": 3.983,
+      "learning_rate": 2.13e-05,
+      "loss": 4.4144,
       "step": 359
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00021479999999999996,
-      "loss": 4.0209,
+      "learning_rate": 2.136e-05,
+      "loss": 4.4934,
       "step": 360
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00021539999999999998,
-      "loss": 4.0257,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 4.4825,
       "step": 361
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00021599999999999996,
-      "loss": 3.9414,
+      "learning_rate": 2.148e-05,
+      "loss": 4.41,
       "step": 362
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00021659999999999998,
-      "loss": 4.0119,
+      "learning_rate": 2.154e-05,
+      "loss": 4.5081,
       "step": 363
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00021719999999999997,
-      "loss": 3.9501,
+      "learning_rate": 2.16e-05,
+      "loss": 4.4704,
       "step": 364
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00021779999999999998,
-      "loss": 3.915,
+      "learning_rate": 2.166e-05,
+      "loss": 4.4811,
       "step": 365
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00021839999999999997,
-      "loss": 4.018,
+      "learning_rate": 2.172e-05,
+      "loss": 4.5751,
       "step": 366
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00021899999999999998,
-      "loss": 3.9956,
+      "learning_rate": 2.178e-05,
+      "loss": 4.5841,
       "step": 367
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00021959999999999997,
-      "loss": 4.0166,
+      "learning_rate": 2.184e-05,
+      "loss": 4.5714,
       "step": 368
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00022019999999999999,
-      "loss": 3.98,
+      "learning_rate": 2.19e-05,
+      "loss": 4.6033,
       "step": 369
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00022079999999999997,
-      "loss": 3.7089,
+      "learning_rate": 2.196e-05,
+      "loss": 4.4132,
       "step": 370
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0002214,
-      "loss": 3.8226,
+      "learning_rate": 2.202e-05,
+      "loss": 4.6072,
       "step": 371
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00022199999999999998,
-      "loss": 3.6044,
+      "learning_rate": 2.208e-05,
+      "loss": 4.4627,
       "step": 372
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0002226,
-      "loss": 3.2785,
+      "learning_rate": 2.214e-05,
+      "loss": 4.2765,
       "step": 373
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00022319999999999998,
-      "loss": 5.0042,
+      "learning_rate": 2.22e-05,
+      "loss": 5.1555,
       "step": 374
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0002238,
-      "loss": 4.8161,
+      "learning_rate": 2.226e-05,
+      "loss": 5.2,
       "step": 375
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00022439999999999998,
-      "loss": 4.6441,
+      "learning_rate": 2.232e-05,
+      "loss": 5.211,
       "step": 376
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.000225,
-      "loss": 4.3838,
+      "learning_rate": 2.238e-05,
+      "loss": 5.0057,
       "step": 377
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00022559999999999998,
-      "loss": 4.3542,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 4.9118,
       "step": 378
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00022619999999999997,
-      "loss": 4.231,
+      "learning_rate": 2.25e-05,
+      "loss": 4.7976,
       "step": 379
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00022679999999999998,
-      "loss": 4.2993,
+      "learning_rate": 2.256e-05,
+      "loss": 4.7151,
       "step": 380
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00022739999999999997,
-      "loss": 4.2949,
+      "learning_rate": 2.262e-05,
+      "loss": 4.7074,
       "step": 381
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00022799999999999999,
-      "loss": 4.2101,
+      "learning_rate": 2.268e-05,
+      "loss": 4.6109,
       "step": 382
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00022859999999999997,
-      "loss": 4.3218,
+      "learning_rate": 2.274e-05,
+      "loss": 4.7031,
       "step": 383
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0002292,
-      "loss": 4.1069,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.4871,
       "step": 384
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00022979999999999997,
-      "loss": 4.0308,
+      "learning_rate": 2.286e-05,
+      "loss": 4.4501,
       "step": 385
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0002304,
-      "loss": 4.0998,
+      "learning_rate": 2.292e-05,
+      "loss": 4.4951,
       "step": 386
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00023099999999999998,
-      "loss": 4.0487,
+      "learning_rate": 2.298e-05,
+      "loss": 4.448,
       "step": 387
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0002316,
-      "loss": 4.0931,
+      "learning_rate": 2.304e-05,
+      "loss": 4.5137,
       "step": 388
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00023219999999999998,
-      "loss": 4.0471,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 4.4997,
       "step": 389
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0002328,
-      "loss": 4.0103,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 4.4241,
       "step": 390
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00023339999999999998,
-      "loss": 4.0283,
+      "learning_rate": 2.322e-05,
+      "loss": 4.4799,
       "step": 391
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.000234,
-      "loss": 4.0728,
+      "learning_rate": 2.328e-05,
+      "loss": 4.4917,
       "step": 392
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00023459999999999998,
-      "loss": 4.0685,
+      "learning_rate": 2.334e-05,
+      "loss": 4.5182,
       "step": 393
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0002352,
-      "loss": 4.0599,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.4918,
       "step": 394
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00023579999999999999,
-      "loss": 4.0421,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 4.4888,
       "step": 395
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0002364,
-      "loss": 3.9362,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 4.3878,
       "step": 396
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.000237,
-      "loss": 4.08,
+      "learning_rate": 2.358e-05,
+      "loss": 4.5381,
       "step": 397
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0002376,
-      "loss": 4.0144,
+      "learning_rate": 2.364e-05,
+      "loss": 4.4603,
       "step": 398
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0002382,
-      "loss": 4.0673,
+      "learning_rate": 2.37e-05,
+      "loss": 4.5001,
       "step": 399
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0002388,
-      "loss": 4.0356,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 4.4627,
       "step": 400
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0002394,
-      "loss": 3.9756,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 4.442,
       "step": 401
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00023999999999999998,
-      "loss": 3.9296,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 4.3704,
       "step": 402
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0002406,
-      "loss": 3.9861,
+      "learning_rate": 2.394e-05,
+      "loss": 4.4678,
       "step": 403
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00024119999999999998,
-      "loss": 4.0462,
+      "learning_rate": 2.4e-05,
+      "loss": 4.5184,
       "step": 404
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0002418,
-      "loss": 3.82,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 4.3222,
       "step": 405
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00024239999999999998,
-      "loss": 3.9398,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 4.4296,
       "step": 406
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.000243,
-      "loss": 3.9655,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 4.4505,
       "step": 407
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00024359999999999999,
-      "loss": 3.8895,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 4.4017,
       "step": 408
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00024419999999999997,
-      "loss": 4.0266,
+      "learning_rate": 2.43e-05,
+      "loss": 4.4974,
       "step": 409
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0002448,
-      "loss": 3.8918,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 4.4035,
       "step": 410
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00024539999999999995,
-      "loss": 3.9576,
+      "learning_rate": 2.442e-05,
+      "loss": 4.4994,
       "step": 411
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00024599999999999996,
-      "loss": 3.8606,
+      "learning_rate": 2.448e-05,
+      "loss": 4.4141,
       "step": 412
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0002466,
-      "loss": 3.9656,
+      "learning_rate": 2.454e-05,
+      "loss": 4.4548,
       "step": 413
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0002472,
-      "loss": 3.841,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.4444,
       "step": 414
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00024779999999999995,
-      "loss": 3.7759,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 4.3861,
       "step": 415
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00024839999999999997,
-      "loss": 3.756,
+      "learning_rate": 2.472e-05,
+      "loss": 4.4311,
       "step": 416
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.000249,
-      "loss": 3.7692,
+      "learning_rate": 2.478e-05,
+      "loss": 4.4903,
       "step": 417
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00024959999999999994,
-      "loss": 3.727,
+      "learning_rate": 2.484e-05,
+      "loss": 4.4648,
       "step": 418
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00025019999999999996,
-      "loss": 3.5521,
+      "learning_rate": 2.49e-05,
+      "loss": 4.3468,
       "step": 419
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00025079999999999997,
-      "loss": 3.5871,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 4.4791,
       "step": 420
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0002514,
-      "loss": 3.4838,
+      "learning_rate": 2.502e-05,
+      "loss": 4.411,
       "step": 421
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00025199999999999995,
-      "loss": 3.456,
+      "learning_rate": 2.508e-05,
+      "loss": 4.5466,
       "step": 422
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00025259999999999996,
-      "loss": 2.9289,
+      "learning_rate": 2.514e-05,
+      "loss": 4.289,
       "step": 423
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0002532,
-      "loss": 6.1709,
+      "learning_rate": 2.52e-05,
+      "loss": 5.1587,
       "step": 424
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0002538,
-      "loss": 5.5822,
+      "learning_rate": 2.526e-05,
+      "loss": 5.2573,
       "step": 425
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00025439999999999995,
-      "loss": 4.9478,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 5.2852,
       "step": 426
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00025499999999999996,
-      "loss": 4.5524,
+      "learning_rate": 2.538e-05,
+      "loss": 5.1566,
       "step": 427
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0002556,
-      "loss": 4.4286,
+      "learning_rate": 2.544e-05,
+      "loss": 5.0713,
       "step": 428
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0002562,
-      "loss": 4.3649,
+      "learning_rate": 2.55e-05,
+      "loss": 4.9891,
       "step": 429
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00025679999999999995,
-      "loss": 4.143,
+      "learning_rate": 2.556e-05,
+      "loss": 4.6524,
       "step": 430
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00025739999999999997,
-      "loss": 4.1339,
+      "learning_rate": 2.562e-05,
+      "loss": 4.6018,
       "step": 431
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.000258,
-      "loss": 4.0862,
+      "learning_rate": 2.568e-05,
+      "loss": 4.5233,
       "step": 432
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0002586,
-      "loss": 3.9971,
+      "learning_rate": 2.574e-05,
+      "loss": 4.4211,
       "step": 433
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00025919999999999996,
-      "loss": 4.0285,
+      "learning_rate": 2.58e-05,
+      "loss": 4.5052,
       "step": 434
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00025979999999999997,
-      "loss": 4.0671,
+      "learning_rate": 2.586e-05,
+      "loss": 4.5205,
       "step": 435
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0002604,
-      "loss": 3.9465,
+      "learning_rate": 2.592e-05,
+      "loss": 4.424,
       "step": 436
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.000261,
-      "loss": 3.9439,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 4.4676,
       "step": 437
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00026159999999999996,
-      "loss": 3.91,
+      "learning_rate": 2.604e-05,
+      "loss": 4.4421,
       "step": 438
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0002622,
-      "loss": 3.819,
+      "learning_rate": 2.61e-05,
+      "loss": 4.3955,
       "step": 439
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0002628,
-      "loss": 3.8625,
+      "learning_rate": 2.616e-05,
+      "loss": 4.453,
       "step": 440
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00026339999999999995,
-      "loss": 3.7986,
+      "learning_rate": 2.622e-05,
+      "loss": 4.4004,
       "step": 441
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00026399999999999997,
-      "loss": 3.8111,
+      "learning_rate": 2.628e-05,
+      "loss": 4.4617,
       "step": 442
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0002646,
-      "loss": 3.739,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 4.5231,
       "step": 443
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0002652,
-      "loss": 3.6672,
+      "learning_rate": 2.64e-05,
+      "loss": 4.518,
       "step": 444
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00026579999999999996,
-      "loss": 3.5265,
+      "learning_rate": 2.646e-05,
+      "loss": 4.6626,
       "step": 445
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00026639999999999997,
-      "loss": 3.0454,
+      "learning_rate": 2.652e-05,
+      "loss": 4.4557,
       "step": 446
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.000267,
-      "loss": 5.4695,
+      "learning_rate": 2.658e-05,
+      "loss": 4.5692,
       "step": 447
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0002676,
-      "loss": 5.3123,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 4.4734,
       "step": 448
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00026819999999999996,
-      "loss": 4.9094,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 4.534,
       "step": 449
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0002688,
-      "loss": 4.4246,
+      "learning_rate": 2.676e-05,
+      "loss": 4.4035,
       "step": 450
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0002694,
-      "loss": 4.3338,
+      "learning_rate": 2.682e-05,
+      "loss": 4.5474,
       "step": 451
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00027,
-      "loss": 4.0653,
+      "learning_rate": 2.688e-05,
+      "loss": 4.4534,
       "step": 452
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00027059999999999996,
-      "loss": 3.9293,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 4.3725,
       "step": 453
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0002712,
-      "loss": 3.924,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.397,
       "step": 454
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0002718,
-      "loss": 3.8679,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 4.3831,
       "step": 455
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0002724,
-      "loss": 3.8316,
+      "learning_rate": 2.712e-05,
+      "loss": 4.4096,
       "step": 456
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00027299999999999997,
-      "loss": 3.6439,
+      "learning_rate": 2.718e-05,
+      "loss": 4.2949,
       "step": 457
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0002736,
-      "loss": 3.6644,
+      "learning_rate": 2.724e-05,
+      "loss": 4.344,
       "step": 458
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0002742,
-      "loss": 3.4917,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 4.2924,
       "step": 459
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0002748,
-      "loss": 3.5827,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 4.3708,
       "step": 460
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00027539999999999997,
-      "loss": 3.5428,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 4.3975,
       "step": 461
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.000276,
-      "loss": 3.5134,
+      "learning_rate": 2.748e-05,
+      "loss": 4.4022,
       "step": 462
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0002766,
-      "loss": 3.5094,
+      "learning_rate": 2.754e-05,
+      "loss": 4.3767,
       "step": 463
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0002772,
-      "loss": 3.4225,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.2759,
       "step": 464
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0002778,
-      "loss": 3.4751,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 4.3917,
       "step": 465
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0002784,
-      "loss": 3.3967,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 4.2681,
       "step": 466
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.000279,
-      "loss": 3.3292,
+      "learning_rate": 2.778e-05,
+      "loss": 4.2556,
       "step": 467
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00027959999999999997,
-      "loss": 3.3373,
+      "learning_rate": 2.784e-05,
+      "loss": 4.289,
       "step": 468
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0002802,
-      "loss": 3.3579,
+      "learning_rate": 2.79e-05,
+      "loss": 4.2804,
       "step": 469
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0002808,
-      "loss": 3.3337,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 4.2821,
       "step": 470
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00028139999999999996,
-      "loss": 3.3649,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 4.3523,
       "step": 471
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 3.2899,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 4.2698,
       "step": 472
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0002826,
-      "loss": 3.2992,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 4.2848,
       "step": 473
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00028319999999999994,
-      "loss": 3.197,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.1485,
       "step": 474
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00028379999999999996,
-      "loss": 3.2116,
+      "learning_rate": 2.826e-05,
+      "loss": 4.1966,
       "step": 475
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0002844,
-      "loss": 3.251,
+      "learning_rate": 2.832e-05,
+      "loss": 4.2595,
       "step": 476
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.000285,
-      "loss": 3.2578,
+      "learning_rate": 2.838e-05,
+      "loss": 4.3192,
       "step": 477
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00028559999999999995,
-      "loss": 3.2191,
+      "learning_rate": 2.844e-05,
+      "loss": 4.2328,
       "step": 478
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00028619999999999996,
-      "loss": 3.1944,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 4.2779,
       "step": 479
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0002868,
-      "loss": 3.2094,
+      "learning_rate": 2.856e-05,
+      "loss": 4.2889,
       "step": 480
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00028739999999999994,
-      "loss": 3.1629,
+      "learning_rate": 2.862e-05,
+      "loss": 4.2063,
       "step": 481
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00028799999999999995,
-      "loss": 3.114,
+      "learning_rate": 2.868e-05,
+      "loss": 4.2333,
       "step": 482
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00028859999999999997,
-      "loss": 3.1973,
+      "learning_rate": 2.874e-05,
+      "loss": 4.289,
       "step": 483
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0002892,
-      "loss": 3.0682,
+      "learning_rate": 2.88e-05,
+      "loss": 4.2304,
       "step": 484
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00028979999999999994,
-      "loss": 3.1276,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 4.2534,
       "step": 485
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00029039999999999996,
-      "loss": 3.0697,
+      "learning_rate": 2.892e-05,
+      "loss": 4.1826,
       "step": 486
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00029099999999999997,
-      "loss": 3.0525,
+      "learning_rate": 2.898e-05,
+      "loss": 4.2797,
       "step": 487
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002916,
-      "loss": 3.0135,
+      "learning_rate": 2.904e-05,
+      "loss": 4.2838,
       "step": 488
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029219999999999995,
-      "loss": 2.8437,
+      "learning_rate": 2.91e-05,
+      "loss": 4.2498,
       "step": 489
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029279999999999996,
-      "loss": 2.7921,
+      "learning_rate": 2.916e-05,
+      "loss": 4.205,
       "step": 490
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002934,
-      "loss": 2.6695,
+      "learning_rate": 2.922e-05,
+      "loss": 4.1591,
       "step": 491
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.000294,
-      "loss": 2.7093,
+      "learning_rate": 2.928e-05,
+      "loss": 4.3021,
       "step": 492
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029459999999999995,
-      "loss": 2.5491,
+      "learning_rate": 2.934e-05,
+      "loss": 4.2822,
       "step": 493
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00029519999999999997,
-      "loss": 2.4168,
+      "learning_rate": 2.94e-05,
+      "loss": 4.1881,
       "step": 494
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002958,
-      "loss": 2.3929,
+      "learning_rate": 2.946e-05,
+      "loss": 4.3554,
       "step": 495
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002964,
-      "loss": 2.0101,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 4.0258,
       "step": 496
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00029699999999999996,
-      "loss": 6.3452,
+      "learning_rate": 2.958e-05,
+      "loss": 5.1196,
       "step": 497
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00029759999999999997,
-      "loss": 5.654,
+      "learning_rate": 2.964e-05,
+      "loss": 5.1393,
       "step": 498
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002982,
-      "loss": 4.9005,
+      "learning_rate": 2.97e-05,
+      "loss": 5.1106,
       "step": 499
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002988,
-      "loss": 4.4437,
+      "learning_rate": 2.976e-05,
+      "loss": 4.9509,
       "step": 500
     },
     {
       "epoch": 2.24,
-      "eval_loss": 3.9967808723449707,
-      "eval_runtime": 586.6262,
-      "eval_samples_per_second": 4.504,
-      "eval_steps_per_second": 0.564,
-      "eval_wer": 1.6151527171757238,
+      "eval_loss": 4.63620138168335,
+      "eval_runtime": 439.1186,
+      "eval_samples_per_second": 6.017,
+      "eval_steps_per_second": 0.754,
+      "eval_wer": 1.0192383974613248,
       "step": 500
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00029939999999999996,
-      "loss": 4.117,
+      "learning_rate": 2.982e-05,
+      "loss": 4.7798,
       "step": 501
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0003,
-      "loss": 3.9755,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 4.7623,
       "step": 502
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00029982658959537567,
-      "loss": 3.8336,
+      "learning_rate": 2.994e-05,
+      "loss": 4.589,
       "step": 503
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002996531791907514,
-      "loss": 3.8994,
+      "learning_rate": 3e-05,
+      "loss": 4.6571,
       "step": 504
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00029947976878612716,
-      "loss": 3.765,
+      "learning_rate": 2.9982658959537573e-05,
+      "loss": 4.4665,
       "step": 505
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00029930635838150286,
-      "loss": 3.677,
+      "learning_rate": 2.9965317919075145e-05,
+      "loss": 4.384,
       "step": 506
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002991329479768786,
-      "loss": 3.6203,
+      "learning_rate": 2.9947976878612717e-05,
+      "loss": 4.3958,
       "step": 507
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002989595375722543,
-      "loss": 3.6362,
+      "learning_rate": 2.993063583815029e-05,
+      "loss": 4.465,
       "step": 508
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00029878612716763005,
-      "loss": 3.5662,
+      "learning_rate": 2.991329479768786e-05,
+      "loss": 4.4328,
       "step": 509
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00029861271676300574,
-      "loss": 3.4386,
+      "learning_rate": 2.9895953757225433e-05,
+      "loss": 4.3279,
       "step": 510
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002984393063583815,
-      "loss": 3.4955,
+      "learning_rate": 2.987861271676301e-05,
+      "loss": 4.418,
       "step": 511
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002982658959537572,
-      "loss": 3.463,
+      "learning_rate": 2.9861271676300577e-05,
+      "loss": 4.3604,
       "step": 512
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00029809248554913293,
-      "loss": 3.4607,
+      "learning_rate": 2.984393063583815e-05,
+      "loss": 4.3442,
       "step": 513
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002979190751445086,
-      "loss": 3.4455,
+      "learning_rate": 2.982658959537572e-05,
+      "loss": 4.3416,
       "step": 514
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00029774566473988437,
-      "loss": 3.4025,
+      "learning_rate": 2.9809248554913297e-05,
+      "loss": 4.3385,
       "step": 515
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00029757225433526006,
-      "loss": 3.2991,
+      "learning_rate": 2.979190751445087e-05,
+      "loss": 4.2913,
       "step": 516
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002973988439306358,
-      "loss": 3.2948,
+      "learning_rate": 2.977456647398844e-05,
+      "loss": 4.279,
       "step": 517
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00029722543352601156,
-      "loss": 3.3225,
+      "learning_rate": 2.975722543352601e-05,
+      "loss": 4.2749,
       "step": 518
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00029705202312138725,
-      "loss": 3.1961,
+      "learning_rate": 2.9739884393063585e-05,
+      "loss": 4.2359,
       "step": 519
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00029687861271676295,
-      "loss": 3.2374,
+      "learning_rate": 2.9722543352601157e-05,
+      "loss": 4.2566,
       "step": 520
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002967052023121387,
-      "loss": 3.2768,
+      "learning_rate": 2.970520231213873e-05,
+      "loss": 4.2878,
       "step": 521
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00029653179190751444,
-      "loss": 3.3283,
+      "learning_rate": 2.96878612716763e-05,
+      "loss": 4.4059,
       "step": 522
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00029635838150289014,
-      "loss": 3.2414,
+      "learning_rate": 2.9670520231213873e-05,
+      "loss": 4.3049,
       "step": 523
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00029618497109826583,
-      "loss": 3.1435,
+      "learning_rate": 2.9653179190751446e-05,
+      "loss": 4.2217,
       "step": 524
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.0002960115606936416,
-      "loss": 3.1363,
+      "learning_rate": 2.9635838150289018e-05,
+      "loss": 4.2402,
       "step": 525
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.0002958381502890173,
-      "loss": 3.0984,
+      "learning_rate": 2.961849710982659e-05,
+      "loss": 4.2514,
       "step": 526
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.000295664739884393,
-      "loss": 3.1239,
+      "learning_rate": 2.9601156069364165e-05,
+      "loss": 4.2706,
       "step": 527
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00029549132947976877,
-      "loss": 2.9329,
+      "learning_rate": 2.9583815028901734e-05,
+      "loss": 4.0966,
       "step": 528
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002953179190751445,
-      "loss": 3.0781,
+      "learning_rate": 2.9566473988439306e-05,
+      "loss": 4.3039,
       "step": 529
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002951445086705202,
-      "loss": 2.9579,
+      "learning_rate": 2.9549132947976878e-05,
+      "loss": 4.1996,
       "step": 530
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002949710982658959,
-      "loss": 3.0964,
+      "learning_rate": 2.9531791907514453e-05,
+      "loss": 4.3097,
       "step": 531
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00029479768786127165,
-      "loss": 3.0365,
+      "learning_rate": 2.9514450867052025e-05,
+      "loss": 4.2551,
       "step": 532
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002946242774566474,
-      "loss": 2.962,
+      "learning_rate": 2.9497109826589594e-05,
+      "loss": 4.252,
       "step": 533
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002944508670520231,
-      "loss": 2.882,
+      "learning_rate": 2.9479768786127166e-05,
+      "loss": 4.2636,
       "step": 534
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00029427745664739884,
-      "loss": 2.7589,
+      "learning_rate": 2.946242774566474e-05,
+      "loss": 4.1686,
       "step": 535
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00029410404624277453,
-      "loss": 2.8493,
+      "learning_rate": 2.9445086705202314e-05,
+      "loss": 4.2744,
       "step": 536
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002939306358381503,
-      "loss": 2.9264,
+      "learning_rate": 2.9427745664739886e-05,
+      "loss": 4.3114,
       "step": 537
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00029375722543352597,
-      "loss": 2.6882,
+      "learning_rate": 2.9410404624277454e-05,
+      "loss": 4.2249,
       "step": 538
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002935838150289017,
-      "loss": 2.6279,
+      "learning_rate": 2.939306358381503e-05,
+      "loss": 4.3479,
       "step": 539
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002934104046242774,
-      "loss": 2.6705,
+      "learning_rate": 2.9375722543352602e-05,
+      "loss": 4.3015,
       "step": 540
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00029323699421965316,
-      "loss": 2.3889,
+      "learning_rate": 2.9358381502890174e-05,
+      "loss": 4.1955,
       "step": 541
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00029306358381502886,
-      "loss": 2.2856,
+      "learning_rate": 2.9341040462427746e-05,
+      "loss": 4.283,
       "step": 542
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.0002928901734104046,
-      "loss": 2.1915,
+      "learning_rate": 2.9323699421965318e-05,
+      "loss": 4.2069,
       "step": 543
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002927167630057803,
-      "loss": 2.028,
+      "learning_rate": 2.930635838150289e-05,
+      "loss": 4.2541,
       "step": 544
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00029254335260115604,
-      "loss": 1.7781,
+      "learning_rate": 2.9289017341040462e-05,
+      "loss": 4.2081,
       "step": 545
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002923699421965318,
-      "loss": 1.6539,
+      "learning_rate": 2.9271676300578034e-05,
+      "loss": 4.1246,
       "step": 546
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002921965317919075,
-      "loss": 7.0245,
+      "learning_rate": 2.925433526011561e-05,
+      "loss": 5.2825,
       "step": 547
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0002920231213872832,
-      "loss": 6.3081,
+      "learning_rate": 2.9236994219653182e-05,
+      "loss": 5.1339,
       "step": 548
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0002918497109826589,
-      "loss": 5.4681,
+      "learning_rate": 2.921965317919075e-05,
+      "loss": 5.171,
       "step": 549
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0002916763005780347,
-      "loss": 4.8535,
+      "learning_rate": 2.9202312138728323e-05,
+      "loss": 5.0012,
       "step": 550
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00029150289017341037,
-      "loss": 4.2709,
+      "learning_rate": 2.9184971098265898e-05,
+      "loss": 4.8356,
       "step": 551
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0002913294797687861,
-      "loss": 3.9271,
+      "learning_rate": 2.916763005780347e-05,
+      "loss": 4.7004,
       "step": 552
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00029115606936416186,
-      "loss": 3.7921,
+      "learning_rate": 2.9150289017341042e-05,
+      "loss": 4.6067,
       "step": 553
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00029098265895953756,
-      "loss": 3.7069,
+      "learning_rate": 2.913294797687861e-05,
+      "loss": 4.5256,
       "step": 554
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00029080924855491325,
-      "loss": 3.6785,
+      "learning_rate": 2.9115606936416186e-05,
+      "loss": 4.5049,
       "step": 555
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.000290635838150289,
-      "loss": 3.6915,
+      "learning_rate": 2.909826589595376e-05,
+      "loss": 4.5028,
       "step": 556
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00029046242774566475,
-      "loss": 3.4802,
+      "learning_rate": 2.908092485549133e-05,
+      "loss": 4.3853,
       "step": 557
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00029028901734104044,
-      "loss": 3.393,
+      "learning_rate": 2.9063583815028903e-05,
+      "loss": 4.3765,
       "step": 558
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00029011560693641613,
-      "loss": 3.4273,
+      "learning_rate": 2.9046242774566475e-05,
+      "loss": 4.4652,
       "step": 559
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.0002899421965317919,
-      "loss": 3.2599,
+      "learning_rate": 2.9028901734104047e-05,
+      "loss": 4.2795,
       "step": 560
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00028976878612716763,
-      "loss": 3.2463,
+      "learning_rate": 2.901156069364162e-05,
+      "loss": 4.3639,
       "step": 561
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.0002895953757225433,
-      "loss": 3.1323,
+      "learning_rate": 2.899421965317919e-05,
+      "loss": 4.3035,
       "step": 562
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00028942196531791907,
-      "loss": 3.1298,
+      "learning_rate": 2.8976878612716766e-05,
+      "loss": 4.3586,
       "step": 563
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00028924855491329476,
-      "loss": 3.0224,
+      "learning_rate": 2.8959537572254335e-05,
+      "loss": 4.2871,
       "step": 564
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.0002890751445086705,
-      "loss": 3.0445,
+      "learning_rate": 2.8942196531791907e-05,
+      "loss": 4.3294,
       "step": 565
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0002889017341040462,
-      "loss": 2.9297,
+      "learning_rate": 2.892485549132948e-05,
+      "loss": 4.3578,
       "step": 566
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00028872832369942195,
-      "loss": 2.8817,
+      "learning_rate": 2.8907514450867055e-05,
+      "loss": 4.3041,
       "step": 567
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00028855491329479765,
-      "loss": 2.8181,
+      "learning_rate": 2.8890173410404627e-05,
+      "loss": 4.2709,
       "step": 568
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0002883815028901734,
-      "loss": 2.6842,
+      "learning_rate": 2.8872832369942195e-05,
+      "loss": 4.2124,
       "step": 569
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.0002882080924855491,
-      "loss": 2.6766,
+      "learning_rate": 2.8855491329479767e-05,
+      "loss": 4.2613,
       "step": 570
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00028803468208092484,
-      "loss": 2.6896,
+      "learning_rate": 2.8838150289017343e-05,
+      "loss": 4.3141,
       "step": 571
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00028786127167630053,
-      "loss": 2.6096,
+      "learning_rate": 2.8820809248554915e-05,
+      "loss": 4.2084,
       "step": 572
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0002876878612716763,
-      "loss": 2.5341,
+      "learning_rate": 2.8803468208092487e-05,
+      "loss": 4.2483,
       "step": 573
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.000287514450867052,
-      "loss": 2.4034,
+      "learning_rate": 2.878612716763006e-05,
+      "loss": 4.2259,
       "step": 574
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0002873410404624277,
-      "loss": 2.4183,
+      "learning_rate": 2.876878612716763e-05,
+      "loss": 4.2839,
       "step": 575
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0002871676300578034,
-      "loss": 2.3373,
+      "learning_rate": 2.8751445086705203e-05,
+      "loss": 4.2728,
       "step": 576
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00028699421965317916,
-      "loss": 2.3058,
+      "learning_rate": 2.8734104046242775e-05,
+      "loss": 4.2867,
       "step": 577
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.0002868208092485549,
-      "loss": 2.5114,
+      "learning_rate": 2.8716763005780347e-05,
+      "loss": 4.2084,
       "step": 578
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.0002866473988439306,
-      "loss": 2.2027,
+      "learning_rate": 2.8699421965317923e-05,
+      "loss": 4.2008,
       "step": 579
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00028647398843930635,
-      "loss": 2.546,
+      "learning_rate": 2.868208092485549e-05,
+      "loss": 4.2647,
       "step": 580
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0002863005780346821,
-      "loss": 2.2915,
+      "learning_rate": 2.8664739884393064e-05,
+      "loss": 4.2243,
       "step": 581
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0002861271676300578,
-      "loss": 2.4333,
+      "learning_rate": 2.8647398843930636e-05,
+      "loss": 4.3127,
       "step": 582
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0002859537572254335,
-      "loss": 2.3528,
+      "learning_rate": 2.863005780346821e-05,
+      "loss": 4.287,
       "step": 583
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00028578034682080923,
-      "loss": 2.0364,
+      "learning_rate": 2.8612716763005783e-05,
+      "loss": 4.2541,
       "step": 584
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.000285606936416185,
-      "loss": 1.9785,
+      "learning_rate": 2.8595375722543352e-05,
+      "loss": 4.3456,
       "step": 585
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0002854335260115607,
-      "loss": 1.8719,
+      "learning_rate": 2.8578034682080924e-05,
+      "loss": 4.1791,
       "step": 586
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00028526011560693637,
-      "loss": 1.7591,
+      "learning_rate": 2.85606936416185e-05,
+      "loss": 4.2295,
       "step": 587
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0002850867052023121,
-      "loss": 1.6455,
+      "learning_rate": 2.854335260115607e-05,
+      "loss": 4.2505,
       "step": 588
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00028491329479768786,
-      "loss": 1.6273,
+      "learning_rate": 2.8526011560693643e-05,
+      "loss": 4.2815,
       "step": 589
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00028473988439306356,
-      "loss": 1.5585,
+      "learning_rate": 2.8508670520231212e-05,
+      "loss": 4.264,
       "step": 590
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.0002845664739884393,
-      "loss": 1.3973,
+      "learning_rate": 2.8491329479768788e-05,
+      "loss": 4.2155,
       "step": 591
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.000284393063583815,
-      "loss": 1.3006,
+      "learning_rate": 2.847398843930636e-05,
+      "loss": 4.264,
       "step": 592
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00028421965317919075,
-      "loss": 1.239,
+      "learning_rate": 2.8456647398843932e-05,
+      "loss": 4.1971,
       "step": 593
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00028404624277456644,
-      "loss": 1.1947,
+      "learning_rate": 2.8439306358381504e-05,
+      "loss": 4.2744,
       "step": 594
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0002838728323699422,
-      "loss": 1.1989,
+      "learning_rate": 2.8421965317919076e-05,
+      "loss": 4.3306,
       "step": 595
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0002836994219653179,
-      "loss": 1.1465,
+      "learning_rate": 2.8404624277456648e-05,
+      "loss": 4.1645,
       "step": 596
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00028352601156069363,
-      "loss": 7.6295,
+      "learning_rate": 2.838728323699422e-05,
+      "loss": 5.2117,
       "step": 597
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.0002833526011560693,
-      "loss": 5.1178,
+      "learning_rate": 2.8369942196531792e-05,
+      "loss": 5.3487,
       "step": 598
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00028317919075144507,
-      "loss": 4.1009,
+      "learning_rate": 2.8352601156069368e-05,
+      "loss": 5.3785,
       "step": 599
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00028300578034682076,
-      "loss": 3.297,
+      "learning_rate": 2.833526011560694e-05,
+      "loss": 5.2952,
       "step": 600
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.0002828323699421965,
-      "loss": 3.0486,
+      "learning_rate": 2.8317919075144508e-05,
+      "loss": 5.2478,
       "step": 601
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00028265895953757226,
-      "loss": 2.8205,
+      "learning_rate": 2.830057803468208e-05,
+      "loss": 5.0231,
       "step": 602
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00028248554913294795,
-      "loss": 2.6073,
+      "learning_rate": 2.8283236994219652e-05,
+      "loss": 4.8553,
       "step": 603
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00028231213872832365,
-      "loss": 2.476,
+      "learning_rate": 2.8265895953757228e-05,
+      "loss": 4.6718,
       "step": 604
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.0002821387283236994,
-      "loss": 2.3003,
+      "learning_rate": 2.82485549132948e-05,
+      "loss": 4.5994,
       "step": 605
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00028196531791907514,
-      "loss": 2.2104,
+      "learning_rate": 2.823121387283237e-05,
+      "loss": 4.4981,
       "step": 606
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00028179190751445083,
-      "loss": 2.0377,
+      "learning_rate": 2.821387283236994e-05,
+      "loss": 4.4212,
       "step": 607
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.0002816184971098266,
-      "loss": 1.9675,
+      "learning_rate": 2.8196531791907516e-05,
+      "loss": 4.4431,
       "step": 608
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.0002814450867052023,
-      "loss": 1.7629,
+      "learning_rate": 2.8179190751445088e-05,
+      "loss": 4.3458,
       "step": 609
     },
     {
       "epoch": 2.74,
-      "learning_rate": 0.000281271676300578,
-      "loss": 1.7966,
+      "learning_rate": 2.816184971098266e-05,
+      "loss": 4.3888,
       "step": 610
     },
     {
       "epoch": 2.74,
-      "learning_rate": 0.0002810982658959537,
-      "loss": 1.6484,
+      "learning_rate": 2.814450867052023e-05,
+      "loss": 4.2951,
       "step": 611
     },
     {
       "epoch": 2.74,
-      "learning_rate": 0.00028092485549132947,
-      "loss": 1.5786,
+      "learning_rate": 2.8127167630057804e-05,
+      "loss": 4.2744,
       "step": 612
     },
     {
       "epoch": 2.75,
-      "learning_rate": 0.00028075144508670516,
-      "loss": 1.5999,
+      "learning_rate": 2.8109826589595376e-05,
+      "loss": 4.3792,
       "step": 613
     },
     {
       "epoch": 2.75,
-      "learning_rate": 0.0002805780346820809,
-      "loss": 1.5781,
+      "learning_rate": 2.809248554913295e-05,
+      "loss": 4.3104,
       "step": 614
     },
     {
       "epoch": 2.76,
-      "learning_rate": 0.0002804046242774566,
-      "loss": 1.4,
+      "learning_rate": 2.807514450867052e-05,
+      "loss": 4.1727,
       "step": 615
     },
     {
       "epoch": 2.76,
-      "learning_rate": 0.00028023121387283235,
-      "loss": 1.4785,
+      "learning_rate": 2.8057803468208093e-05,
+      "loss": 4.3698,
       "step": 616
     },
     {
       "epoch": 2.77,
-      "learning_rate": 0.00028005780346820804,
-      "loss": 1.3169,
+      "learning_rate": 2.8040462427745665e-05,
+      "loss": 4.297,
       "step": 617
     },
     {
       "epoch": 2.77,
-      "learning_rate": 0.0002798843930635838,
-      "loss": 1.3737,
+      "learning_rate": 2.8023121387283237e-05,
+      "loss": 4.3592,
       "step": 618
     },
     {
       "epoch": 2.78,
-      "learning_rate": 0.00027971098265895954,
-      "loss": 1.3111,
+      "learning_rate": 2.800578034682081e-05,
+      "loss": 4.2746,
       "step": 619
     },
     {
       "epoch": 2.78,
-      "learning_rate": 0.00027953757225433523,
-      "loss": 1.2543,
+      "learning_rate": 2.7988439306358384e-05,
+      "loss": 4.3185,
       "step": 620
     },
     {
       "epoch": 2.78,
-      "learning_rate": 0.0002793641618497109,
-      "loss": 1.1839,
+      "learning_rate": 2.7971098265895953e-05,
+      "loss": 4.3034,
       "step": 621
     },
     {
       "epoch": 2.79,
-      "learning_rate": 0.00027919075144508667,
-      "loss": 1.1236,
+      "learning_rate": 2.7953757225433525e-05,
+      "loss": 4.2336,
       "step": 622
     },
     {
       "epoch": 2.79,
-      "learning_rate": 0.0002790173410404624,
-      "loss": 1.1345,
+      "learning_rate": 2.7936416184971097e-05,
+      "loss": 4.2943,
       "step": 623
     },
     {
       "epoch": 2.8,
-      "learning_rate": 0.0002788439306358381,
-      "loss": 0.9673,
+      "learning_rate": 2.7919075144508673e-05,
+      "loss": 4.1664,
       "step": 624
     },
     {
       "epoch": 2.8,
-      "learning_rate": 0.00027867052023121386,
-      "loss": 1.0724,
+      "learning_rate": 2.7901734104046245e-05,
+      "loss": 4.2394,
       "step": 625
     },
     {
       "epoch": 2.81,
-      "learning_rate": 0.00027849710982658955,
-      "loss": 1.0625,
+      "learning_rate": 2.7884393063583817e-05,
+      "loss": 4.3155,
       "step": 626
     },
     {
       "epoch": 2.81,
-      "learning_rate": 0.0002783236994219653,
-      "loss": 1.1293,
+      "learning_rate": 2.7867052023121385e-05,
+      "loss": 4.3109,
       "step": 627
     },
     {
       "epoch": 2.82,
-      "learning_rate": 0.000278150289017341,
-      "loss": 0.9543,
+      "learning_rate": 2.784971098265896e-05,
+      "loss": 4.1935,
       "step": 628
     },
     {
       "epoch": 2.82,
-      "learning_rate": 0.00027797687861271674,
-      "loss": 0.9536,
+      "learning_rate": 2.7832369942196533e-05,
+      "loss": 4.224,
       "step": 629
     },
     {
       "epoch": 2.83,
-      "learning_rate": 0.0002778034682080925,
-      "loss": 0.9573,
+      "learning_rate": 2.7815028901734105e-05,
+      "loss": 4.2271,
       "step": 630
     },
     {
       "epoch": 2.83,
-      "learning_rate": 0.0002776300578034682,
-      "loss": 0.9805,
+      "learning_rate": 2.7797687861271677e-05,
+      "loss": 4.2469,
       "step": 631
     },
     {
       "epoch": 2.83,
-      "learning_rate": 0.0002774566473988439,
-      "loss": 0.8767,
+      "learning_rate": 2.778034682080925e-05,
+      "loss": 4.2321,
       "step": 632
     },
     {
       "epoch": 2.84,
-      "learning_rate": 0.0002772832369942196,
-      "loss": 0.8657,
+      "learning_rate": 2.776300578034682e-05,
+      "loss": 4.1939,
       "step": 633
     },
     {
       "epoch": 2.84,
-      "learning_rate": 0.0002771098265895954,
-      "loss": 0.8442,
+      "learning_rate": 2.7745664739884393e-05,
+      "loss": 4.2496,
       "step": 634
     },
     {
       "epoch": 2.85,
-      "learning_rate": 0.00027693641618497107,
-      "loss": 0.8549,
+      "learning_rate": 2.7728323699421965e-05,
+      "loss": 4.2124,
       "step": 635
     },
     {
       "epoch": 2.85,
-      "learning_rate": 0.0002767630057803468,
-      "loss": 0.8211,
+      "learning_rate": 2.771098265895954e-05,
+      "loss": 4.1494,
       "step": 636
     },
     {
       "epoch": 2.86,
-      "learning_rate": 0.0002765895953757225,
-      "loss": 0.8208,
+      "learning_rate": 2.769364161849711e-05,
+      "loss": 4.2576,
       "step": 637
     },
     {
       "epoch": 2.86,
-      "learning_rate": 0.00027641618497109826,
-      "loss": 0.8521,
+      "learning_rate": 2.767630057803468e-05,
+      "loss": 4.2654,
       "step": 638
     },
     {
       "epoch": 2.87,
-      "learning_rate": 0.00027624277456647395,
-      "loss": 0.8427,
+      "learning_rate": 2.7658959537572254e-05,
+      "loss": 4.2854,
       "step": 639
     },
     {
       "epoch": 2.87,
-      "learning_rate": 0.0002760693641618497,
-      "loss": 0.8618,
+      "learning_rate": 2.764161849710983e-05,
+      "loss": 4.3885,
       "step": 640
     },
     {
       "epoch": 2.87,
-      "learning_rate": 0.0002758959537572254,
-      "loss": 0.7377,
+      "learning_rate": 2.76242774566474e-05,
+      "loss": 4.2676,
       "step": 641
     },
     {
       "epoch": 2.88,
-      "learning_rate": 0.00027572254335260114,
-      "loss": 0.7555,
+      "learning_rate": 2.760693641618497e-05,
+      "loss": 4.2418,
       "step": 642
     },
     {
       "epoch": 2.88,
-      "learning_rate": 0.00027554913294797683,
-      "loss": 0.7199,
+      "learning_rate": 2.7589595375722542e-05,
+      "loss": 4.1392,
       "step": 643
     },
     {
       "epoch": 2.89,
-      "learning_rate": 0.0002753757225433526,
-      "loss": 0.8234,
+      "learning_rate": 2.7572254335260117e-05,
+      "loss": 4.2529,
       "step": 644
     },
     {
       "epoch": 2.89,
-      "learning_rate": 0.0002752023121387283,
-      "loss": 0.8251,
+      "learning_rate": 2.755491329479769e-05,
+      "loss": 4.1777,
       "step": 645
     },
     {
       "epoch": 2.9,
-      "learning_rate": 0.000275028901734104,
-      "loss": 0.8766,
+      "learning_rate": 2.753757225433526e-05,
+      "loss": 4.1001,
       "step": 646
     },
     {
       "epoch": 2.9,
-      "learning_rate": 0.00027485549132947977,
-      "loss": 9.3322,
+      "learning_rate": 2.752023121387283e-05,
+      "loss": 5.2074,
       "step": 647
     },
     {
       "epoch": 2.91,
-      "learning_rate": 0.00027468208092485546,
-      "loss": 7.5506,
+      "learning_rate": 2.7502890173410406e-05,
+      "loss": 5.2898,
       "step": 648
     },
     {
       "epoch": 2.91,
-      "learning_rate": 0.00027450867052023116,
-      "loss": 5.1447,
+      "learning_rate": 2.7485549132947978e-05,
+      "loss": 5.2925,
       "step": 649
     },
     {
       "epoch": 2.91,
-      "learning_rate": 0.0002743352601156069,
-      "loss": 3.5468,
+      "learning_rate": 2.746820809248555e-05,
+      "loss": 5.307,
       "step": 650
     },
     {
       "epoch": 2.92,
-      "learning_rate": 0.00027416184971098265,
-      "loss": 2.5995,
+      "learning_rate": 2.7450867052023122e-05,
+      "loss": 5.0791,
       "step": 651
     },
     {
       "epoch": 2.92,
-      "learning_rate": 0.00027398843930635835,
-      "loss": 2.1182,
+      "learning_rate": 2.7433526011560697e-05,
+      "loss": 4.8647,
       "step": 652
     },
     {
       "epoch": 2.93,
-      "learning_rate": 0.0002738150289017341,
-      "loss": 1.9193,
+      "learning_rate": 2.7416184971098266e-05,
+      "loss": 4.8926,
       "step": 653
     },
     {
       "epoch": 2.93,
-      "learning_rate": 0.00027364161849710984,
-      "loss": 1.7254,
+      "learning_rate": 2.7398843930635838e-05,
+      "loss": 4.6648,
       "step": 654
     },
     {
       "epoch": 2.94,
-      "learning_rate": 0.00027346820809248554,
-      "loss": 1.513,
+      "learning_rate": 2.738150289017341e-05,
+      "loss": 4.5792,
       "step": 655
     },
     {
       "epoch": 2.94,
-      "learning_rate": 0.00027329479768786123,
-      "loss": 1.3363,
+      "learning_rate": 2.7364161849710986e-05,
+      "loss": 4.4372,
       "step": 656
     },
     {
       "epoch": 2.95,
-      "learning_rate": 0.000273121387283237,
-      "loss": 1.1537,
+      "learning_rate": 2.7346820809248558e-05,
+      "loss": 4.3201,
       "step": 657
     },
     {
       "epoch": 2.95,
-      "learning_rate": 0.0002729479768786127,
-      "loss": 1.1289,
+      "learning_rate": 2.7329479768786126e-05,
+      "loss": 4.3811,
       "step": 658
     },
     {
       "epoch": 2.96,
-      "learning_rate": 0.0002727745664739884,
-      "loss": 1.0504,
+      "learning_rate": 2.73121387283237e-05,
+      "loss": 4.3014,
       "step": 659
     },
     {
       "epoch": 2.96,
-      "learning_rate": 0.0002726011560693641,
-      "loss": 1.0136,
+      "learning_rate": 2.7294797687861274e-05,
+      "loss": 4.3152,
       "step": 660
     },
     {
       "epoch": 2.96,
-      "learning_rate": 0.00027242774566473986,
-      "loss": 0.9461,
+      "learning_rate": 2.7277456647398846e-05,
+      "loss": 4.3751,
       "step": 661
     },
     {
       "epoch": 2.97,
-      "learning_rate": 0.0002722543352601156,
-      "loss": 0.9157,
+      "learning_rate": 2.7260115606936418e-05,
+      "loss": 4.2581,
       "step": 662
     },
     {
       "epoch": 2.97,
-      "learning_rate": 0.0002720809248554913,
-      "loss": 0.7936,
+      "learning_rate": 2.7242774566473987e-05,
+      "loss": 4.2483,
       "step": 663
     },
     {
       "epoch": 2.98,
-      "learning_rate": 0.00027190751445086705,
-      "loss": 0.827,
+      "learning_rate": 2.7225433526011562e-05,
+      "loss": 4.2555,
       "step": 664
     },
     {
       "epoch": 2.98,
-      "learning_rate": 0.00027173410404624274,
-      "loss": 0.7706,
+      "learning_rate": 2.7208092485549134e-05,
+      "loss": 4.3171,
       "step": 665
     },
     {
       "epoch": 2.99,
-      "learning_rate": 0.0002715606936416185,
-      "loss": 0.7488,
+      "learning_rate": 2.7190751445086706e-05,
+      "loss": 4.1631,
       "step": 666
     },
     {
       "epoch": 2.99,
-      "learning_rate": 0.0002713872832369942,
-      "loss": 0.6611,
+      "learning_rate": 2.717341040462428e-05,
+      "loss": 4.288,
       "step": 667
     },
     {
       "epoch": 3.0,
-      "learning_rate": 0.00027121387283236993,
-      "loss": 0.6923,
+      "learning_rate": 2.715606936416185e-05,
+      "loss": 4.2888,
       "step": 668
     },
     {
       "epoch": 3.0,
-      "learning_rate": 0.0002710404624277456,
-      "loss": 0.7828,
+      "learning_rate": 2.7138728323699422e-05,
+      "loss": 4.3506,
       "step": 669
     },
     {
       "epoch": 3.0,
-      "learning_rate": 0.0002708670520231214,
-      "loss": 3.6436,
+      "learning_rate": 2.7121387283236995e-05,
+      "loss": 4.3771,
       "step": 670
     },
     {
       "epoch": 3.01,
-      "learning_rate": 0.00027069364161849707,
-      "loss": 3.0477,
+      "learning_rate": 2.7104046242774567e-05,
+      "loss": 4.3568,
       "step": 671
     },
     {
       "epoch": 3.01,
-      "learning_rate": 0.0002705202312138728,
-      "loss": 1.9186,
+      "learning_rate": 2.7086705202312142e-05,
+      "loss": 4.3573,
       "step": 672
     },
     {
       "epoch": 3.02,
-      "learning_rate": 0.0002703468208092485,
-      "loss": 1.7018,
+      "learning_rate": 2.706936416184971e-05,
+      "loss": 4.3189,
       "step": 673
     },
     {
       "epoch": 3.02,
-      "learning_rate": 0.00027017341040462426,
-      "loss": 1.4485,
+      "learning_rate": 2.7052023121387283e-05,
+      "loss": 4.3255,
       "step": 674
     },
     {
       "epoch": 3.03,
-      "learning_rate": 0.00027,
-      "loss": 1.1206,
+      "learning_rate": 2.7034682080924855e-05,
+      "loss": 4.3107,
       "step": 675
     },
     {
       "epoch": 3.03,
-      "learning_rate": 0.0002698265895953757,
-      "loss": 1.1319,
+      "learning_rate": 2.701734104046243e-05,
+      "loss": 4.3297,
       "step": 676
     },
     {
       "epoch": 3.04,
-      "learning_rate": 0.0002696531791907514,
-      "loss": 1.0149,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2507,
       "step": 677
     },
     {
       "epoch": 3.04,
-      "learning_rate": 0.00026947976878612714,
-      "loss": 0.809,
+      "learning_rate": 2.6982658959537574e-05,
+      "loss": 4.1454,
       "step": 678
     },
     {
       "epoch": 3.04,
-      "learning_rate": 0.0002693063583815029,
-      "loss": 0.8025,
+      "learning_rate": 2.6965317919075143e-05,
+      "loss": 4.2008,
       "step": 679
     },
     {
       "epoch": 3.05,
-      "learning_rate": 0.0002691329479768786,
-      "loss": 0.8262,
+      "learning_rate": 2.694797687861272e-05,
+      "loss": 4.1477,
       "step": 680
     },
     {
       "epoch": 3.05,
-      "learning_rate": 0.00026895953757225433,
-      "loss": 0.6764,
+      "learning_rate": 2.693063583815029e-05,
+      "loss": 4.1979,
       "step": 681
     },
     {
       "epoch": 3.06,
-      "learning_rate": 0.0002687861271676301,
-      "loss": 0.7046,
+      "learning_rate": 2.6913294797687863e-05,
+      "loss": 4.2501,
       "step": 682
     },
     {
       "epoch": 3.06,
-      "learning_rate": 0.00026861271676300577,
-      "loss": 0.6042,
+      "learning_rate": 2.6895953757225435e-05,
+      "loss": 4.149,
       "step": 683
     },
     {
       "epoch": 3.07,
-      "learning_rate": 0.00026843930635838146,
-      "loss": 0.553,
+      "learning_rate": 2.6878612716763007e-05,
+      "loss": 4.1838,
       "step": 684
     },
     {
       "epoch": 3.07,
-      "learning_rate": 0.0002682658959537572,
-      "loss": 0.6246,
+      "learning_rate": 2.686127167630058e-05,
+      "loss": 4.2453,
       "step": 685
     },
     {
       "epoch": 3.08,
-      "learning_rate": 0.00026809248554913296,
-      "loss": 0.6005,
+      "learning_rate": 2.684393063583815e-05,
+      "loss": 4.1949,
       "step": 686
     },
     {
       "epoch": 3.08,
-      "learning_rate": 0.00026791907514450865,
-      "loss": 0.5905,
+      "learning_rate": 2.6826589595375723e-05,
+      "loss": 4.1344,
       "step": 687
     },
     {
       "epoch": 3.09,
-      "learning_rate": 0.00026774566473988435,
-      "loss": 0.5917,
+      "learning_rate": 2.68092485549133e-05,
+      "loss": 4.1343,
       "step": 688
     },
     {
       "epoch": 3.09,
-      "learning_rate": 0.0002675722543352601,
-      "loss": 0.5212,
+      "learning_rate": 2.6791907514450867e-05,
+      "loss": 4.1509,
       "step": 689
     },
     {
       "epoch": 3.09,
-      "learning_rate": 0.00026739884393063584,
-      "loss": 0.5089,
+      "learning_rate": 2.677456647398844e-05,
+      "loss": 4.1241,
       "step": 690
     },
     {
       "epoch": 3.1,
-      "learning_rate": 0.00026722543352601153,
-      "loss": 0.4883,
+      "learning_rate": 2.675722543352601e-05,
+      "loss": 4.0753,
       "step": 691
     },
     {
       "epoch": 3.1,
-      "learning_rate": 0.0002670520231213873,
-      "loss": 0.5311,
+      "learning_rate": 2.6739884393063587e-05,
+      "loss": 4.1107,
       "step": 692
     },
     {
       "epoch": 3.11,
-      "learning_rate": 0.000266878612716763,
-      "loss": 0.4886,
+      "learning_rate": 2.672254335260116e-05,
+      "loss": 4.0779,
       "step": 693
     },
     {
       "epoch": 3.11,
-      "learning_rate": 0.0002667052023121387,
-      "loss": 0.4564,
+      "learning_rate": 2.6705202312138728e-05,
+      "loss": 4.0421,
       "step": 694
     },
     {
       "epoch": 3.12,
-      "learning_rate": 0.0002665317919075144,
-      "loss": 0.4356,
+      "learning_rate": 2.66878612716763e-05,
+      "loss": 4.0448,
       "step": 695
     },
     {
       "epoch": 3.12,
-      "learning_rate": 0.00026635838150289016,
-      "loss": 0.5206,
+      "learning_rate": 2.6670520231213875e-05,
+      "loss": 4.1082,
       "step": 696
     },
     {
       "epoch": 3.13,
-      "learning_rate": 0.00026618497109826586,
-      "loss": 0.409,
+      "learning_rate": 2.6653179190751447e-05,
+      "loss": 4.0395,
       "step": 697
     },
     {
       "epoch": 3.13,
-      "learning_rate": 0.00026601156069364155,
-      "loss": 0.431,
+      "learning_rate": 2.663583815028902e-05,
+      "loss": 4.0556,
       "step": 698
     },
     {
       "epoch": 3.13,
-      "learning_rate": 0.0002658381502890173,
-      "loss": 0.448,
+      "learning_rate": 2.6618497109826588e-05,
+      "loss": 4.0162,
       "step": 699
     },
     {
       "epoch": 3.14,
-      "learning_rate": 0.00026566473988439305,
-      "loss": 0.4387,
+      "learning_rate": 2.660115606936416e-05,
+      "loss": 3.9868,
       "step": 700
     },
     {
       "epoch": 3.14,
-      "learning_rate": 0.00026549132947976874,
-      "loss": 0.3993,
+      "learning_rate": 2.6583815028901735e-05,
+      "loss": 3.9384,
       "step": 701
     },
     {
       "epoch": 3.15,
-      "learning_rate": 0.0002653179190751445,
-      "loss": 0.4169,
+      "learning_rate": 2.6566473988439307e-05,
+      "loss": 4.0049,
       "step": 702
     },
     {
       "epoch": 3.15,
-      "learning_rate": 0.00026514450867052024,
-      "loss": 0.4112,
+      "learning_rate": 2.654913294797688e-05,
+      "loss": 3.9742,
       "step": 703
     },
     {
       "epoch": 3.16,
-      "learning_rate": 0.00026497109826589593,
-      "loss": 0.419,
+      "learning_rate": 2.653179190751445e-05,
+      "loss": 3.9905,
       "step": 704
     },
     {
       "epoch": 3.16,
-      "learning_rate": 0.0002647976878612716,
-      "loss": 0.4739,
+      "learning_rate": 2.6514450867052024e-05,
+      "loss": 4.0019,
       "step": 705
     },
     {
       "epoch": 3.17,
-      "learning_rate": 0.00026462427745664737,
-      "loss": 0.4217,
+      "learning_rate": 2.6497109826589596e-05,
+      "loss": 3.9217,
       "step": 706
     },
     {
       "epoch": 3.17,
-      "learning_rate": 0.0002644508670520231,
-      "loss": 0.4082,
+      "learning_rate": 2.6479768786127168e-05,
+      "loss": 4.0279,
       "step": 707
     },
     {
       "epoch": 3.17,
-      "learning_rate": 0.0002642774566473988,
-      "loss": 0.4224,
+      "learning_rate": 2.646242774566474e-05,
+      "loss": 4.0279,
       "step": 708
     },
     {
       "epoch": 3.18,
-      "learning_rate": 0.00026410404624277456,
-      "loss": 0.3968,
+      "learning_rate": 2.6445086705202315e-05,
+      "loss": 3.85,
       "step": 709
     },
     {
       "epoch": 3.18,
-      "learning_rate": 0.00026393063583815025,
-      "loss": 0.4368,
+      "learning_rate": 2.6427745664739884e-05,
+      "loss": 4.0435,
       "step": 710
     },
     {
       "epoch": 3.19,
-      "learning_rate": 0.000263757225433526,
-      "loss": 0.4774,
+      "learning_rate": 2.6410404624277456e-05,
+      "loss": 4.0146,
       "step": 711
     },
     {
       "epoch": 3.19,
-      "learning_rate": 0.0002635838150289017,
-      "loss": 0.3729,
+      "learning_rate": 2.6393063583815028e-05,
+      "loss": 4.0106,
       "step": 712
     },
     {
       "epoch": 3.2,
-      "learning_rate": 0.00026341040462427744,
-      "loss": 0.3777,
+      "learning_rate": 2.6375722543352604e-05,
+      "loss": 4.1137,
       "step": 713
     },
     {
       "epoch": 3.2,
-      "learning_rate": 0.00026323699421965314,
-      "loss": 0.3002,
+      "learning_rate": 2.6358381502890176e-05,
+      "loss": 3.9141,
       "step": 714
     },
     {
       "epoch": 3.21,
-      "learning_rate": 0.0002630635838150289,
-      "loss": 0.3403,
+      "learning_rate": 2.6341040462427744e-05,
+      "loss": 3.9825,
       "step": 715
     },
     {
       "epoch": 3.21,
-      "learning_rate": 0.0002628901734104046,
-      "loss": 0.3768,
+      "learning_rate": 2.6323699421965316e-05,
+      "loss": 3.8994,
       "step": 716
     },
     {
       "epoch": 3.22,
-      "learning_rate": 0.0002627167630057803,
-      "loss": 0.3403,
+      "learning_rate": 2.6306358381502892e-05,
+      "loss": 3.9,
       "step": 717
     },
     {
       "epoch": 3.22,
-      "learning_rate": 0.000262543352601156,
-      "loss": 0.4646,
+      "learning_rate": 2.6289017341040464e-05,
+      "loss": 4.0405,
       "step": 718
     },
     {
       "epoch": 3.22,
-      "learning_rate": 0.00026236994219653177,
-      "loss": 0.3688,
+      "learning_rate": 2.6271676300578036e-05,
+      "loss": 3.8299,
       "step": 719
     },
     {
       "epoch": 3.23,
-      "learning_rate": 0.0002621965317919075,
-      "loss": 3.746,
+      "learning_rate": 2.6254335260115605e-05,
+      "loss": 5.0524,
       "step": 720
     },
     {
       "epoch": 3.23,
-      "learning_rate": 0.0002620231213872832,
-      "loss": 2.5798,
+      "learning_rate": 2.623699421965318e-05,
+      "loss": 5.1449,
       "step": 721
     },
     {
       "epoch": 3.24,
-      "learning_rate": 0.0002618497109826589,
-      "loss": 1.5031,
+      "learning_rate": 2.6219653179190752e-05,
+      "loss": 5.2232,
       "step": 722
     },
     {
       "epoch": 3.24,
-      "learning_rate": 0.00026167630057803465,
-      "loss": 1.2022,
+      "learning_rate": 2.6202312138728324e-05,
+      "loss": 5.1146,
       "step": 723
     },
     {
       "epoch": 3.25,
-      "learning_rate": 0.0002615028901734104,
-      "loss": 1.0921,
+      "learning_rate": 2.6184971098265896e-05,
+      "loss": 5.0103,
       "step": 724
     },
     {
       "epoch": 3.25,
-      "learning_rate": 0.0002613294797687861,
-      "loss": 1.013,
+      "learning_rate": 2.616763005780347e-05,
+      "loss": 4.9405,
       "step": 725
     },
     {
       "epoch": 3.26,
-      "learning_rate": 0.00026115606936416184,
-      "loss": 0.9056,
+      "learning_rate": 2.615028901734104e-05,
+      "loss": 4.8401,
       "step": 726
     },
     {
       "epoch": 3.26,
-      "learning_rate": 0.0002609826589595376,
-      "loss": 0.8245,
+      "learning_rate": 2.6132947976878613e-05,
+      "loss": 4.6854,
       "step": 727
     },
     {
       "epoch": 3.26,
-      "learning_rate": 0.0002608092485549133,
-      "loss": 0.7315,
+      "learning_rate": 2.6115606936416185e-05,
+      "loss": 4.657,
       "step": 728
     },
     {
       "epoch": 3.27,
-      "learning_rate": 0.000260635838150289,
-      "loss": 0.6466,
+      "learning_rate": 2.609826589595376e-05,
+      "loss": 4.4908,
       "step": 729
     },
     {
       "epoch": 3.27,
-      "learning_rate": 0.0002604624277456647,
-      "loss": 0.5553,
+      "learning_rate": 2.608092485549133e-05,
+      "loss": 4.2998,
       "step": 730
     },
     {
       "epoch": 3.28,
-      "learning_rate": 0.00026028901734104047,
-      "loss": 0.5448,
+      "learning_rate": 2.60635838150289e-05,
+      "loss": 4.2922,
       "step": 731
     },
     {
       "epoch": 3.28,
-      "learning_rate": 0.00026011560693641616,
-      "loss": 0.5247,
+      "learning_rate": 2.6046242774566473e-05,
+      "loss": 4.2949,
       "step": 732
     },
     {
       "epoch": 3.29,
-      "learning_rate": 0.00025994219653179186,
-      "loss": 0.5081,
+      "learning_rate": 2.602890173410405e-05,
+      "loss": 4.2303,
       "step": 733
     },
     {
       "epoch": 3.29,
-      "learning_rate": 0.0002597687861271676,
-      "loss": 0.4833,
+      "learning_rate": 2.601156069364162e-05,
+      "loss": 4.2311,
       "step": 734
     },
     {
       "epoch": 3.3,
-      "learning_rate": 0.00025959537572254335,
-      "loss": 0.4507,
+      "learning_rate": 2.5994219653179192e-05,
+      "loss": 4.2388,
       "step": 735
     },
     {
       "epoch": 3.3,
-      "learning_rate": 0.00025942196531791905,
-      "loss": 0.5001,
+      "learning_rate": 2.597687861271676e-05,
+      "loss": 4.1948,
       "step": 736
     },
     {
       "epoch": 3.3,
-      "learning_rate": 0.0002592485549132948,
-      "loss": 0.4418,
+      "learning_rate": 2.5959537572254337e-05,
+      "loss": 4.2017,
       "step": 737
     },
     {
       "epoch": 3.31,
-      "learning_rate": 0.0002590751445086705,
-      "loss": 0.418,
+      "learning_rate": 2.594219653179191e-05,
+      "loss": 4.1867,
       "step": 738
     },
     {
       "epoch": 3.31,
-      "learning_rate": 0.00025890173410404624,
-      "loss": 0.4375,
+      "learning_rate": 2.592485549132948e-05,
+      "loss": 4.1769,
       "step": 739
     },
     {
       "epoch": 3.32,
-      "learning_rate": 0.00025872832369942193,
-      "loss": 0.4326,
+      "learning_rate": 2.5907514450867053e-05,
+      "loss": 4.0745,
       "step": 740
     },
     {
       "epoch": 3.32,
-      "learning_rate": 0.0002585549132947977,
-      "loss": 0.3794,
+      "learning_rate": 2.5890173410404625e-05,
+      "loss": 4.0938,
       "step": 741
     },
     {
       "epoch": 3.33,
-      "learning_rate": 0.00025838150289017337,
-      "loss": 0.4532,
+      "learning_rate": 2.5872832369942197e-05,
+      "loss": 4.0325,
       "step": 742
     },
     {
       "epoch": 3.33,
-      "learning_rate": 0.0002582080924855491,
-      "loss": 0.3637,
+      "learning_rate": 2.585549132947977e-05,
+      "loss": 4.1052,
       "step": 743
     },
     {
       "epoch": 3.34,
-      "learning_rate": 0.0002580346820809248,
-      "loss": 0.3888,
+      "learning_rate": 2.583815028901734e-05,
+      "loss": 4.1427,
       "step": 744
     },
     {
       "epoch": 3.34,
-      "learning_rate": 0.00025786127167630056,
-      "loss": 0.3813,
+      "learning_rate": 2.5820809248554917e-05,
+      "loss": 4.0941,
       "step": 745
     },
     {
       "epoch": 3.35,
-      "learning_rate": 0.00025768786127167625,
-      "loss": 0.3921,
+      "learning_rate": 2.5803468208092485e-05,
+      "loss": 4.0728,
       "step": 746
     },
     {
       "epoch": 3.35,
-      "learning_rate": 0.000257514450867052,
-      "loss": 0.387,
+      "learning_rate": 2.5786127167630057e-05,
+      "loss": 4.0014,
       "step": 747
     },
     {
       "epoch": 3.35,
-      "learning_rate": 0.00025734104046242775,
-      "loss": 0.4,
+      "learning_rate": 2.576878612716763e-05,
+      "loss": 4.0966,
       "step": 748
     },
     {
       "epoch": 3.36,
-      "learning_rate": 0.00025716763005780344,
-      "loss": 0.3746,
+      "learning_rate": 2.5751445086705205e-05,
+      "loss": 4.0551,
       "step": 749
     },
     {
       "epoch": 3.36,
-      "learning_rate": 0.00025699421965317914,
-      "loss": 0.3611,
+      "learning_rate": 2.5734104046242777e-05,
+      "loss": 4.1112,
       "step": 750
     },
     {
       "epoch": 3.37,
-      "learning_rate": 0.0002568208092485549,
-      "loss": 0.3834,
+      "learning_rate": 2.5716763005780346e-05,
+      "loss": 4.0396,
       "step": 751
     },
     {
       "epoch": 3.37,
-      "learning_rate": 0.00025664739884393063,
-      "loss": 0.3827,
+      "learning_rate": 2.5699421965317918e-05,
+      "loss": 4.0486,
       "step": 752
     },
     {
       "epoch": 3.38,
-      "learning_rate": 0.0002564739884393063,
-      "loss": 0.4217,
+      "learning_rate": 2.5682080924855493e-05,
+      "loss": 4.0519,
       "step": 753
     },
     {
       "epoch": 3.38,
-      "learning_rate": 0.00025630057803468207,
-      "loss": 0.4256,
+      "learning_rate": 2.5664739884393065e-05,
+      "loss": 4.0912,
       "step": 754
     },
     {
       "epoch": 3.39,
-      "learning_rate": 0.0002561271676300578,
-      "loss": 0.3919,
+      "learning_rate": 2.5647398843930637e-05,
+      "loss": 4.1006,
       "step": 755
     },
     {
       "epoch": 3.39,
-      "learning_rate": 0.0002559537572254335,
-      "loss": 0.329,
+      "learning_rate": 2.5630057803468206e-05,
+      "loss": 3.9934,
       "step": 756
     },
     {
       "epoch": 3.39,
-      "learning_rate": 0.0002557803468208092,
-      "loss": 0.364,
+      "learning_rate": 2.561271676300578e-05,
+      "loss": 4.0426,
       "step": 757
     },
     {
       "epoch": 3.4,
-      "learning_rate": 0.00025560693641618496,
-      "loss": 0.3676,
+      "learning_rate": 2.5595375722543353e-05,
+      "loss": 4.0186,
       "step": 758
     },
     {
       "epoch": 3.4,
-      "learning_rate": 0.0002554335260115607,
-      "loss": 0.3192,
+      "learning_rate": 2.5578034682080925e-05,
+      "loss": 3.956,
       "step": 759
     },
     {
       "epoch": 3.41,
-      "learning_rate": 0.0002552601156069364,
-      "loss": 0.386,
+      "learning_rate": 2.5560693641618498e-05,
+      "loss": 4.0575,
       "step": 760
     },
     {
       "epoch": 3.41,
-      "learning_rate": 0.0002550867052023121,
-      "loss": 0.3516,
+      "learning_rate": 2.5543352601156073e-05,
+      "loss": 3.9721,
       "step": 761
     },
     {
       "epoch": 3.42,
-      "learning_rate": 0.00025491329479768784,
-      "loss": 0.3816,
+      "learning_rate": 2.5526011560693642e-05,
+      "loss": 3.9682,
       "step": 762
     },
     {
       "epoch": 3.42,
-      "learning_rate": 0.0002547398843930636,
-      "loss": 0.3091,
+      "learning_rate": 2.5508670520231214e-05,
+      "loss": 4.0261,
       "step": 763
     },
     {
       "epoch": 3.43,
-      "learning_rate": 0.0002545664739884393,
-      "loss": 0.3428,
+      "learning_rate": 2.5491329479768786e-05,
+      "loss": 4.0748,
       "step": 764
     },
     {
       "epoch": 3.43,
-      "learning_rate": 0.00025439306358381503,
-      "loss": 0.3129,
+      "learning_rate": 2.547398843930636e-05,
+      "loss": 3.9467,
       "step": 765
     },
     {
       "epoch": 3.43,
-      "learning_rate": 0.0002542196531791907,
-      "loss": 0.3559,
+      "learning_rate": 2.5456647398843933e-05,
+      "loss": 4.0798,
       "step": 766
     },
     {
       "epoch": 3.44,
-      "learning_rate": 0.00025404624277456647,
-      "loss": 0.3588,
+      "learning_rate": 2.5439306358381502e-05,
+      "loss": 3.8732,
       "step": 767
     },
     {
       "epoch": 3.44,
-      "learning_rate": 0.00025387283236994216,
-      "loss": 0.3518,
+      "learning_rate": 2.5421965317919074e-05,
+      "loss": 3.9332,
       "step": 768
     },
     {
       "epoch": 3.45,
-      "learning_rate": 0.0002536994219653179,
-      "loss": 0.3134,
+      "learning_rate": 2.540462427745665e-05,
+      "loss": 3.7809,
       "step": 769
     },
     {
       "epoch": 3.45,
-      "learning_rate": 0.0002535260115606936,
-      "loss": 2.7246,
+      "learning_rate": 2.538728323699422e-05,
+      "loss": 4.88,
       "step": 770
     },
     {
       "epoch": 3.46,
-      "learning_rate": 0.00025335260115606935,
-      "loss": 2.0574,
+      "learning_rate": 2.5369942196531794e-05,
+      "loss": 5.0725,
       "step": 771
     },
     {
       "epoch": 3.46,
-      "learning_rate": 0.00025317919075144504,
-      "loss": 1.2423,
+      "learning_rate": 2.5352601156069362e-05,
+      "loss": 5.1848,
       "step": 772
     },
     {
       "epoch": 3.47,
-      "learning_rate": 0.0002530057803468208,
-      "loss": 0.8525,
+      "learning_rate": 2.5335260115606938e-05,
+      "loss": 5.0027,
       "step": 773
     },
     {
       "epoch": 3.47,
-      "learning_rate": 0.0002528323699421965,
-      "loss": 0.8248,
+      "learning_rate": 2.531791907514451e-05,
+      "loss": 4.9422,
       "step": 774
     },
     {
       "epoch": 3.48,
-      "learning_rate": 0.00025265895953757223,
-      "loss": 0.6959,
+      "learning_rate": 2.5300578034682082e-05,
+      "loss": 4.9045,
       "step": 775
     },
     {
       "epoch": 3.48,
-      "learning_rate": 0.000252485549132948,
-      "loss": 0.6066,
+      "learning_rate": 2.5283236994219654e-05,
+      "loss": 4.8252,
       "step": 776
     },
     {
       "epoch": 3.48,
-      "learning_rate": 0.0002523121387283237,
-      "loss": 0.5454,
+      "learning_rate": 2.5265895953757226e-05,
+      "loss": 4.5949,
       "step": 777
     },
     {
       "epoch": 3.49,
-      "learning_rate": 0.00025213872832369937,
-      "loss": 0.5132,
+      "learning_rate": 2.5248554913294798e-05,
+      "loss": 4.5225,
       "step": 778
     },
     {
       "epoch": 3.49,
-      "learning_rate": 0.0002519653179190751,
-      "loss": 0.4923,
+      "learning_rate": 2.523121387283237e-05,
+      "loss": 4.3801,
       "step": 779
     },
     {
       "epoch": 3.5,
-      "learning_rate": 0.00025179190751445086,
-      "loss": 0.4915,
+      "learning_rate": 2.5213872832369942e-05,
+      "loss": 4.3487,
       "step": 780
     },
     {
       "epoch": 3.5,
-      "learning_rate": 0.00025161849710982656,
-      "loss": 0.4586,
+      "learning_rate": 2.5196531791907518e-05,
+      "loss": 4.264,
       "step": 781
     },
     {
       "epoch": 3.51,
-      "learning_rate": 0.0002514450867052023,
-      "loss": 0.4383,
+      "learning_rate": 2.5179190751445086e-05,
+      "loss": 4.2761,
       "step": 782
     },
     {
       "epoch": 3.51,
-      "learning_rate": 0.00025127167630057805,
-      "loss": 0.3949,
+      "learning_rate": 2.516184971098266e-05,
+      "loss": 4.1625,
       "step": 783
     },
     {
       "epoch": 3.52,
-      "learning_rate": 0.00025109826589595375,
-      "loss": 0.4295,
+      "learning_rate": 2.514450867052023e-05,
+      "loss": 4.14,
       "step": 784
     },
     {
       "epoch": 3.52,
-      "learning_rate": 0.00025092485549132944,
-      "loss": 0.3954,
+      "learning_rate": 2.5127167630057806e-05,
+      "loss": 4.1771,
       "step": 785
     },
     {
       "epoch": 3.52,
-      "learning_rate": 0.0002507514450867052,
-      "loss": 0.4317,
+      "learning_rate": 2.5109826589595378e-05,
+      "loss": 4.2,
       "step": 786
     },
     {
       "epoch": 3.53,
-      "learning_rate": 0.00025057803468208094,
-      "loss": 0.4151,
+      "learning_rate": 2.509248554913295e-05,
+      "loss": 4.1604,
       "step": 787
     },
     {
       "epoch": 3.53,
-      "learning_rate": 0.00025040462427745663,
-      "loss": 0.349,
+      "learning_rate": 2.507514450867052e-05,
+      "loss": 4.0762,
       "step": 788
     },
     {
       "epoch": 3.54,
-      "learning_rate": 0.0002502312138728323,
-      "loss": 0.3537,
+      "learning_rate": 2.5057803468208094e-05,
+      "loss": 4.1025,
       "step": 789
     },
     {
       "epoch": 3.54,
-      "learning_rate": 0.00025005780346820807,
-      "loss": 0.3305,
+      "learning_rate": 2.5040462427745666e-05,
+      "loss": 4.1368,
       "step": 790
     },
     {
       "epoch": 3.55,
-      "learning_rate": 0.00024988439306358376,
-      "loss": 0.3834,
+      "learning_rate": 2.502312138728324e-05,
+      "loss": 4.131,
       "step": 791
     },
     {
       "epoch": 3.55,
-      "learning_rate": 0.0002497109826589595,
-      "loss": 0.3788,
+      "learning_rate": 2.500578034682081e-05,
+      "loss": 4.0851,
       "step": 792
     },
     {
       "epoch": 3.56,
-      "learning_rate": 0.00024953757225433526,
-      "loss": 0.3858,
+      "learning_rate": 2.498843930635838e-05,
+      "loss": 4.0771,
       "step": 793
     },
     {
       "epoch": 3.56,
-      "learning_rate": 0.00024936416184971095,
-      "loss": 0.3739,
+      "learning_rate": 2.4971098265895955e-05,
+      "loss": 4.1087,
       "step": 794
     },
     {
       "epoch": 3.57,
-      "learning_rate": 0.00024919075144508665,
-      "loss": 0.3757,
+      "learning_rate": 2.4953757225433527e-05,
+      "loss": 4.1521,
       "step": 795
     },
     {
       "epoch": 3.57,
-      "learning_rate": 0.0002490173410404624,
-      "loss": 0.3295,
+      "learning_rate": 2.49364161849711e-05,
+      "loss": 4.086,
       "step": 796
     },
     {
       "epoch": 3.57,
-      "learning_rate": 0.00024884393063583814,
-      "loss": 0.2999,
+      "learning_rate": 2.491907514450867e-05,
+      "loss": 4.0091,
       "step": 797
     },
     {
       "epoch": 3.58,
-      "learning_rate": 0.00024867052023121384,
-      "loss": 0.3492,
+      "learning_rate": 2.4901734104046243e-05,
+      "loss": 4.079,
       "step": 798
     },
     {
       "epoch": 3.58,
-      "learning_rate": 0.0002484971098265896,
-      "loss": 0.3537,
+      "learning_rate": 2.4884393063583815e-05,
+      "loss": 4.1123,
       "step": 799
     },
     {
       "epoch": 3.59,
-      "learning_rate": 0.00024832369942196533,
-      "loss": 0.3539,
+      "learning_rate": 2.4867052023121387e-05,
+      "loss": 4.0259,
       "step": 800
     },
     {
       "epoch": 3.59,
-      "learning_rate": 0.000248150289017341,
-      "loss": 0.3874,
+      "learning_rate": 2.484971098265896e-05,
+      "loss": 4.0563,
       "step": 801
     },
     {
       "epoch": 3.6,
-      "learning_rate": 0.0002479768786127167,
-      "loss": 0.3529,
+      "learning_rate": 2.4832369942196535e-05,
+      "loss": 4.0352,
       "step": 802
     },
     {
       "epoch": 3.6,
-      "learning_rate": 0.00024780346820809247,
-      "loss": 0.3646,
+      "learning_rate": 2.4815028901734103e-05,
+      "loss": 4.0531,
       "step": 803
     },
     {
       "epoch": 3.61,
-      "learning_rate": 0.0002476300578034682,
-      "loss": 0.3855,
+      "learning_rate": 2.4797687861271675e-05,
+      "loss": 4.0419,
       "step": 804
     },
     {
       "epoch": 3.61,
-      "learning_rate": 0.0002474566473988439,
-      "loss": 0.3391,
+      "learning_rate": 2.4780346820809247e-05,
+      "loss": 4.0331,
       "step": 805
     },
     {
       "epoch": 3.61,
-      "learning_rate": 0.0002472832369942196,
-      "loss": 0.3225,
+      "learning_rate": 2.4763005780346823e-05,
+      "loss": 4.0502,
       "step": 806
     },
     {
       "epoch": 3.62,
-      "learning_rate": 0.00024710982658959535,
-      "loss": 0.2516,
+      "learning_rate": 2.4745664739884395e-05,
+      "loss": 3.9469,
       "step": 807
     },
     {
       "epoch": 3.62,
-      "learning_rate": 0.0002469364161849711,
-      "loss": 0.3274,
+      "learning_rate": 2.4728323699421964e-05,
+      "loss": 4.1118,
       "step": 808
     },
     {
       "epoch": 3.63,
-      "learning_rate": 0.0002467630057803468,
-      "loss": 0.3606,
+      "learning_rate": 2.4710982658959536e-05,
+      "loss": 4.0224,
       "step": 809
     },
     {
       "epoch": 3.63,
-      "learning_rate": 0.00024658959537572254,
-      "loss": 0.3471,
+      "learning_rate": 2.469364161849711e-05,
+      "loss": 4.0334,
       "step": 810
     },
     {
       "epoch": 3.64,
-      "learning_rate": 0.00024641618497109823,
-      "loss": 0.329,
+      "learning_rate": 2.4676300578034683e-05,
+      "loss": 3.9915,
       "step": 811
     },
     {
       "epoch": 3.64,
-      "learning_rate": 0.000246242774566474,
-      "loss": 0.2733,
+      "learning_rate": 2.4658959537572255e-05,
+      "loss": 3.9869,
       "step": 812
     },
     {
       "epoch": 3.65,
-      "learning_rate": 0.0002460693641618497,
-      "loss": 0.3286,
+      "learning_rate": 2.4641618497109827e-05,
+      "loss": 4.031,
       "step": 813
     },
     {
       "epoch": 3.65,
-      "learning_rate": 0.0002458959537572254,
-      "loss": 0.2962,
+      "learning_rate": 2.46242774566474e-05,
+      "loss": 4.1438,
       "step": 814
     },
     {
       "epoch": 3.65,
-      "learning_rate": 0.0002457225433526011,
-      "loss": 0.2743,
+      "learning_rate": 2.460693641618497e-05,
+      "loss": 3.8887,
       "step": 815
     },
     {
       "epoch": 3.66,
-      "learning_rate": 0.00024554913294797686,
-      "loss": 0.3511,
+      "learning_rate": 2.4589595375722544e-05,
+      "loss": 4.0366,
       "step": 816
     },
     {
       "epoch": 3.66,
-      "learning_rate": 0.00024537572254335256,
-      "loss": 0.2911,
+      "learning_rate": 2.4572254335260116e-05,
+      "loss": 4.046,
       "step": 817
     },
     {
       "epoch": 3.67,
-      "learning_rate": 0.0002452023121387283,
-      "loss": 0.2989,
+      "learning_rate": 2.455491329479769e-05,
+      "loss": 3.9899,
       "step": 818
     },
     {
       "epoch": 3.67,
-      "learning_rate": 0.000245028901734104,
-      "loss": 0.3634,
+      "learning_rate": 2.453757225433526e-05,
+      "loss": 3.9826,
       "step": 819
     },
     {
       "epoch": 3.68,
-      "learning_rate": 0.00024485549132947975,
-      "loss": 1.8232,
+      "learning_rate": 2.4520231213872832e-05,
+      "loss": 4.8933,
       "step": 820
     },
     {
       "epoch": 3.68,
-      "learning_rate": 0.0002446820809248555,
-      "loss": 1.0919,
+      "learning_rate": 2.4502890173410404e-05,
+      "loss": 4.8635,
       "step": 821
     },
     {
       "epoch": 3.69,
-      "learning_rate": 0.0002445086705202312,
-      "loss": 0.7213,
+      "learning_rate": 2.448554913294798e-05,
+      "loss": 4.9638,
       "step": 822
     },
     {
       "epoch": 3.69,
-      "learning_rate": 0.0002443352601156069,
-      "loss": 0.7192,
+      "learning_rate": 2.446820809248555e-05,
+      "loss": 5.0011,
       "step": 823
     },
     {
       "epoch": 3.7,
-      "learning_rate": 0.00024416184971098263,
-      "loss": 0.7089,
+      "learning_rate": 2.445086705202312e-05,
+      "loss": 4.9523,
       "step": 824
     },
     {
       "epoch": 3.7,
-      "learning_rate": 0.00024398843930635838,
-      "loss": 0.5815,
+      "learning_rate": 2.4433526011560692e-05,
+      "loss": 4.8568,
       "step": 825
     },
     {
       "epoch": 3.7,
-      "learning_rate": 0.00024381502890173407,
-      "loss": 0.5092,
+      "learning_rate": 2.4416184971098268e-05,
+      "loss": 4.7478,
       "step": 826
     },
     {
       "epoch": 3.71,
-      "learning_rate": 0.0002436416184971098,
-      "loss": 0.5179,
+      "learning_rate": 2.439884393063584e-05,
+      "loss": 4.7502,
       "step": 827
     },
     {
       "epoch": 3.71,
-      "learning_rate": 0.00024346820809248554,
-      "loss": 0.4872,
+      "learning_rate": 2.4381502890173412e-05,
+      "loss": 4.5984,
       "step": 828
     },
     {
       "epoch": 3.72,
-      "learning_rate": 0.00024329479768786126,
-      "loss": 0.4405,
+      "learning_rate": 2.436416184971098e-05,
+      "loss": 4.4919,
       "step": 829
     },
     {
       "epoch": 3.72,
-      "learning_rate": 0.00024312138728323698,
-      "loss": 0.4115,
+      "learning_rate": 2.4346820809248556e-05,
+      "loss": 4.2935,
       "step": 830
     },
     {
       "epoch": 3.73,
-      "learning_rate": 0.00024294797687861267,
-      "loss": 0.4027,
+      "learning_rate": 2.4329479768786128e-05,
+      "loss": 4.3023,
       "step": 831
     },
     {
       "epoch": 3.73,
-      "learning_rate": 0.00024277456647398842,
-      "loss": 0.3596,
+      "learning_rate": 2.43121387283237e-05,
+      "loss": 4.2509,
       "step": 832
     },
     {
       "epoch": 3.74,
-      "learning_rate": 0.00024260115606936414,
-      "loss": 0.3759,
+      "learning_rate": 2.4294797687861272e-05,
+      "loss": 4.2187,
       "step": 833
     },
     {
       "epoch": 3.74,
-      "learning_rate": 0.00024242774566473986,
-      "loss": 0.3514,
+      "learning_rate": 2.4277456647398844e-05,
+      "loss": 4.1768,
       "step": 834
     },
     {
       "epoch": 3.74,
-      "learning_rate": 0.00024225433526011558,
-      "loss": 0.3,
+      "learning_rate": 2.4260115606936416e-05,
+      "loss": 4.092,
       "step": 835
     },
     {
       "epoch": 3.75,
-      "learning_rate": 0.00024208092485549133,
-      "loss": 0.3155,
+      "learning_rate": 2.4242774566473988e-05,
+      "loss": 4.1135,
       "step": 836
     },
     {
       "epoch": 3.75,
-      "learning_rate": 0.00024190751445086702,
-      "loss": 0.3328,
+      "learning_rate": 2.422543352601156e-05,
+      "loss": 4.0844,
       "step": 837
     },
     {
       "epoch": 3.76,
-      "learning_rate": 0.00024173410404624275,
-      "loss": 0.3285,
+      "learning_rate": 2.4208092485549136e-05,
+      "loss": 4.1666,
       "step": 838
     },
     {
       "epoch": 3.76,
-      "learning_rate": 0.00024156069364161847,
-      "loss": 0.3116,
+      "learning_rate": 2.4190751445086708e-05,
+      "loss": 4.1177,
       "step": 839
     },
     {
       "epoch": 3.77,
-      "learning_rate": 0.0002413872832369942,
-      "loss": 0.4866,
+      "learning_rate": 2.4173410404624277e-05,
+      "loss": 4.086,
       "step": 840
     },
     {
       "epoch": 3.77,
-      "learning_rate": 0.00024121387283236993,
-      "loss": 0.3259,
+      "learning_rate": 2.415606936416185e-05,
+      "loss": 4.1248,
       "step": 841
     },
     {
       "epoch": 3.78,
-      "learning_rate": 0.00024104046242774563,
-      "loss": 0.3017,
+      "learning_rate": 2.4138728323699424e-05,
+      "loss": 3.984,
       "step": 842
     },
     {
       "epoch": 3.78,
-      "learning_rate": 0.00024086705202312135,
-      "loss": 0.3129,
+      "learning_rate": 2.4121387283236996e-05,
+      "loss": 4.0219,
       "step": 843
     },
     {
       "epoch": 3.78,
-      "learning_rate": 0.0002406936416184971,
-      "loss": 0.328,
+      "learning_rate": 2.4104046242774568e-05,
+      "loss": 4.0683,
       "step": 844
     },
     {
       "epoch": 3.79,
-      "learning_rate": 0.00024052023121387282,
-      "loss": 0.3282,
+      "learning_rate": 2.4086705202312137e-05,
+      "loss": 4.1054,
       "step": 845
     },
     {
       "epoch": 3.79,
-      "learning_rate": 0.00024034682080924854,
-      "loss": 0.2992,
+      "learning_rate": 2.4069364161849712e-05,
+      "loss": 4.0374,
       "step": 846
     },
     {
       "epoch": 3.8,
-      "learning_rate": 0.00024017341040462423,
-      "loss": 0.2855,
+      "learning_rate": 2.4052023121387284e-05,
+      "loss": 3.9663,
       "step": 847
     },
     {
       "epoch": 3.8,
-      "learning_rate": 0.00023999999999999998,
-      "loss": 0.3201,
+      "learning_rate": 2.4034682080924856e-05,
+      "loss": 4.0037,
       "step": 848
     },
     {
       "epoch": 3.81,
-      "learning_rate": 0.0002398265895953757,
-      "loss": 0.264,
+      "learning_rate": 2.401734104046243e-05,
+      "loss": 4.1018,
       "step": 849
     },
     {
       "epoch": 3.81,
-      "learning_rate": 0.00023965317919075142,
-      "loss": 0.3239,
+      "learning_rate": 2.4e-05,
+      "loss": 4.1347,
       "step": 850
     },
     {
       "epoch": 3.82,
-      "learning_rate": 0.00023947976878612714,
-      "loss": 0.2577,
+      "learning_rate": 2.3982658959537573e-05,
+      "loss": 3.9942,
       "step": 851
     },
     {
       "epoch": 3.82,
-      "learning_rate": 0.0002393063583815029,
-      "loss": 0.316,
+      "learning_rate": 2.3965317919075145e-05,
+      "loss": 4.014,
       "step": 852
     },
     {
       "epoch": 3.83,
-      "learning_rate": 0.0002391329479768786,
-      "loss": 0.2993,
+      "learning_rate": 2.3947976878612717e-05,
+      "loss": 3.9866,
       "step": 853
     },
     {
       "epoch": 3.83,
-      "learning_rate": 0.0002389595375722543,
-      "loss": 0.2835,
+      "learning_rate": 2.3930635838150292e-05,
+      "loss": 3.9975,
       "step": 854
     },
     {
       "epoch": 3.83,
-      "learning_rate": 0.00023878612716763002,
-      "loss": 0.3059,
+      "learning_rate": 2.391329479768786e-05,
+      "loss": 4.0508,
       "step": 855
     },
     {
       "epoch": 3.84,
-      "learning_rate": 0.00023861271676300577,
-      "loss": 0.2899,
+      "learning_rate": 2.3895953757225433e-05,
+      "loss": 4.1077,
       "step": 856
     },
     {
       "epoch": 3.84,
-      "learning_rate": 0.0002384393063583815,
-      "loss": 0.2627,
+      "learning_rate": 2.3878612716763005e-05,
+      "loss": 4.0541,
       "step": 857
     },
     {
       "epoch": 3.85,
-      "learning_rate": 0.0002382658959537572,
-      "loss": 0.2783,
+      "learning_rate": 2.386127167630058e-05,
+      "loss": 4.0191,
       "step": 858
     },
     {
       "epoch": 3.85,
-      "learning_rate": 0.0002380924855491329,
-      "loss": 0.2801,
+      "learning_rate": 2.3843930635838153e-05,
+      "loss": 3.9792,
       "step": 859
     },
     {
       "epoch": 3.86,
-      "learning_rate": 0.00023791907514450865,
-      "loss": 0.2571,
+      "learning_rate": 2.382658959537572e-05,
+      "loss": 3.9911,
       "step": 860
     },
     {
       "epoch": 3.86,
-      "learning_rate": 0.00023774566473988437,
-      "loss": 0.3506,
+      "learning_rate": 2.3809248554913293e-05,
+      "loss": 4.0872,
       "step": 861
     },
     {
       "epoch": 3.87,
-      "learning_rate": 0.0002375722543352601,
-      "loss": 0.3228,
+      "learning_rate": 2.379190751445087e-05,
+      "loss": 4.0612,
       "step": 862
     },
     {
       "epoch": 3.87,
-      "learning_rate": 0.00023739884393063582,
-      "loss": 0.2565,
+      "learning_rate": 2.377456647398844e-05,
+      "loss": 3.9739,
       "step": 863
     },
     {
       "epoch": 3.87,
-      "learning_rate": 0.00023722543352601156,
-      "loss": 0.2523,
+      "learning_rate": 2.3757225433526013e-05,
+      "loss": 3.9459,
       "step": 864
     },
     {
       "epoch": 3.88,
-      "learning_rate": 0.00023705202312138726,
-      "loss": 0.2447,
+      "learning_rate": 2.3739884393063585e-05,
+      "loss": 4.0336,
       "step": 865
     },
     {
       "epoch": 3.88,
-      "learning_rate": 0.00023687861271676298,
-      "loss": 0.3134,
+      "learning_rate": 2.3722543352601157e-05,
+      "loss": 4.0283,
       "step": 866
     },
     {
       "epoch": 3.89,
-      "learning_rate": 0.0002367052023121387,
-      "loss": 0.3882,
+      "learning_rate": 2.370520231213873e-05,
+      "loss": 4.0666,
       "step": 867
     },
     {
       "epoch": 3.89,
-      "learning_rate": 0.00023653179190751445,
-      "loss": 0.3079,
+      "learning_rate": 2.36878612716763e-05,
+      "loss": 3.9129,
       "step": 868
     },
     {
       "epoch": 3.9,
-      "learning_rate": 0.00023635838150289017,
-      "loss": 0.4326,
+      "learning_rate": 2.3670520231213873e-05,
+      "loss": 3.8494,
       "step": 869
     },
     {
       "epoch": 3.9,
-      "learning_rate": 0.00023618497109826586,
-      "loss": 1.5025,
+      "learning_rate": 2.365317919075145e-05,
+      "loss": 4.7372,
       "step": 870
     },
     {
       "epoch": 3.91,
-      "learning_rate": 0.00023601156069364158,
-      "loss": 0.9483,
+      "learning_rate": 2.3635838150289017e-05,
+      "loss": 4.7829,
       "step": 871
     },
     {
       "epoch": 3.91,
-      "learning_rate": 0.00023583815028901733,
-      "loss": 0.5595,
+      "learning_rate": 2.361849710982659e-05,
+      "loss": 4.7124,
       "step": 872
     },
     {
       "epoch": 3.91,
-      "learning_rate": 0.00023566473988439305,
-      "loss": 0.5277,
+      "learning_rate": 2.360115606936416e-05,
+      "loss": 4.8475,
       "step": 873
     },
     {
       "epoch": 3.92,
-      "learning_rate": 0.00023549132947976877,
-      "loss": 0.4715,
+      "learning_rate": 2.3583815028901737e-05,
+      "loss": 4.6983,
       "step": 874
     },
     {
       "epoch": 3.92,
-      "learning_rate": 0.0002353179190751445,
-      "loss": 0.4018,
+      "learning_rate": 2.356647398843931e-05,
+      "loss": 4.6769,
       "step": 875
     },
     {
       "epoch": 3.93,
-      "learning_rate": 0.00023514450867052024,
-      "loss": 0.4456,
+      "learning_rate": 2.3549132947976878e-05,
+      "loss": 4.6902,
       "step": 876
     },
     {
       "epoch": 3.93,
-      "learning_rate": 0.00023497109826589593,
-      "loss": 0.3791,
+      "learning_rate": 2.353179190751445e-05,
+      "loss": 4.485,
       "step": 877
     },
     {
       "epoch": 3.94,
-      "learning_rate": 0.00023479768786127165,
-      "loss": 0.4253,
+      "learning_rate": 2.3514450867052025e-05,
+      "loss": 4.4053,
       "step": 878
     },
     {
       "epoch": 3.94,
-      "learning_rate": 0.00023462427745664737,
-      "loss": 0.3124,
+      "learning_rate": 2.3497109826589597e-05,
+      "loss": 4.2649,
       "step": 879
     },
     {
       "epoch": 3.95,
-      "learning_rate": 0.00023445086705202312,
-      "loss": 0.3515,
+      "learning_rate": 2.347976878612717e-05,
+      "loss": 4.3011,
       "step": 880
     },
     {
       "epoch": 3.95,
-      "learning_rate": 0.00023427745664739884,
-      "loss": 0.3239,
+      "learning_rate": 2.3462427745664738e-05,
+      "loss": 4.1987,
       "step": 881
     },
     {
       "epoch": 3.96,
-      "learning_rate": 0.00023410404624277454,
-      "loss": 0.2906,
+      "learning_rate": 2.3445086705202314e-05,
+      "loss": 4.0956,
       "step": 882
     },
     {
       "epoch": 3.96,
-      "learning_rate": 0.00023393063583815026,
-      "loss": 0.3742,
+      "learning_rate": 2.3427745664739886e-05,
+      "loss": 4.1256,
       "step": 883
     },
     {
       "epoch": 3.96,
-      "learning_rate": 0.000233757225433526,
-      "loss": 0.2932,
+      "learning_rate": 2.3410404624277458e-05,
+      "loss": 3.951,
       "step": 884
     },
     {
       "epoch": 3.97,
-      "learning_rate": 0.00023358381502890173,
-      "loss": 0.3035,
+      "learning_rate": 2.339306358381503e-05,
+      "loss": 4.1155,
       "step": 885
     },
     {
       "epoch": 3.97,
-      "learning_rate": 0.00023341040462427745,
-      "loss": 0.2989,
+      "learning_rate": 2.3375722543352602e-05,
+      "loss": 4.0756,
       "step": 886
     },
     {
       "epoch": 3.98,
-      "learning_rate": 0.00023323699421965314,
-      "loss": 0.2559,
+      "learning_rate": 2.3358381502890174e-05,
+      "loss": 4.0417,
       "step": 887
     },
     {
       "epoch": 3.98,
-      "learning_rate": 0.00023306358381502886,
-      "loss": 0.3164,
+      "learning_rate": 2.3341040462427746e-05,
+      "loss": 4.1136,
       "step": 888
     },
     {
       "epoch": 3.99,
-      "learning_rate": 0.0002328901734104046,
-      "loss": 0.2812,
+      "learning_rate": 2.3323699421965318e-05,
+      "loss": 4.1349,
       "step": 889
     },
     {
       "epoch": 3.99,
-      "learning_rate": 0.00023271676300578033,
-      "loss": 0.3021,
+      "learning_rate": 2.330635838150289e-05,
+      "loss": 4.0961,
       "step": 890
     },
     {
       "epoch": 4.0,
-      "learning_rate": 0.00023254335260115605,
-      "loss": 0.233,
+      "learning_rate": 2.3289017341040466e-05,
+      "loss": 3.9922,
       "step": 891
     },
     {
       "epoch": 4.0,
-      "learning_rate": 0.00023236994219653174,
-      "loss": 0.3204,
+      "learning_rate": 2.3271676300578034e-05,
+      "loss": 4.1601,
       "step": 892
     },
     {
       "epoch": 4.0,
-      "learning_rate": 0.0002321965317919075,
-      "loss": 0.771,
+      "learning_rate": 2.3254335260115606e-05,
+      "loss": 4.3071,
       "step": 893
     },
     {
       "epoch": 4.01,
-      "learning_rate": 0.0002320231213872832,
-      "loss": 0.5956,
+      "learning_rate": 2.323699421965318e-05,
+      "loss": 4.2509,
       "step": 894
     },
     {
       "epoch": 4.01,
-      "learning_rate": 0.00023184971098265893,
-      "loss": 0.4446,
+      "learning_rate": 2.3219653179190754e-05,
+      "loss": 4.2892,
       "step": 895
     },
     {
       "epoch": 4.02,
-      "learning_rate": 0.00023167630057803465,
-      "loss": 0.3546,
+      "learning_rate": 2.3202312138728326e-05,
+      "loss": 4.2278,
       "step": 896
     },
     {
       "epoch": 4.02,
-      "learning_rate": 0.0002315028901734104,
-      "loss": 0.3401,
+      "learning_rate": 2.3184971098265895e-05,
+      "loss": 4.1769,
       "step": 897
     },
     {
       "epoch": 4.03,
-      "learning_rate": 0.00023132947976878612,
-      "loss": 0.3294,
+      "learning_rate": 2.3167630057803467e-05,
+      "loss": 4.3061,
       "step": 898
     },
     {
       "epoch": 4.03,
-      "learning_rate": 0.00023115606936416181,
-      "loss": 0.3047,
+      "learning_rate": 2.3150289017341042e-05,
+      "loss": 4.1789,
       "step": 899
     },
     {
       "epoch": 4.04,
-      "learning_rate": 0.00023098265895953754,
-      "loss": 0.2599,
+      "learning_rate": 2.3132947976878614e-05,
+      "loss": 4.1885,
       "step": 900
     },
     {
       "epoch": 4.04,
-      "learning_rate": 0.00023080924855491328,
-      "loss": 0.2687,
+      "learning_rate": 2.3115606936416186e-05,
+      "loss": 4.0922,
       "step": 901
     },
     {
       "epoch": 4.04,
-      "learning_rate": 0.000230635838150289,
-      "loss": 0.2312,
+      "learning_rate": 2.3098265895953755e-05,
+      "loss": 4.1367,
       "step": 902
     },
     {
       "epoch": 4.05,
-      "learning_rate": 0.00023046242774566472,
-      "loss": 0.2279,
+      "learning_rate": 2.308092485549133e-05,
+      "loss": 4.0081,
       "step": 903
     },
     {
       "epoch": 4.05,
-      "learning_rate": 0.00023028901734104042,
-      "loss": 0.2295,
+      "learning_rate": 2.3063583815028902e-05,
+      "loss": 4.0683,
       "step": 904
     },
     {
       "epoch": 4.06,
-      "learning_rate": 0.00023011560693641617,
-      "loss": 0.1895,
+      "learning_rate": 2.3046242774566475e-05,
+      "loss": 4.0,
       "step": 905
     },
     {
       "epoch": 4.06,
-      "learning_rate": 0.0002299421965317919,
-      "loss": 0.236,
+      "learning_rate": 2.3028901734104047e-05,
+      "loss": 4.145,
       "step": 906
     },
     {
       "epoch": 4.07,
-      "learning_rate": 0.0002297687861271676,
-      "loss": 0.2184,
+      "learning_rate": 2.301156069364162e-05,
+      "loss": 4.0468,
       "step": 907
     },
     {
       "epoch": 4.07,
-      "learning_rate": 0.00022959537572254333,
-      "loss": 0.1909,
+      "learning_rate": 2.299421965317919e-05,
+      "loss": 3.9863,
       "step": 908
     },
     {
       "epoch": 4.08,
-      "learning_rate": 0.00022942196531791908,
-      "loss": 0.1998,
+      "learning_rate": 2.2976878612716763e-05,
+      "loss": 4.0802,
       "step": 909
     },
     {
       "epoch": 4.08,
-      "learning_rate": 0.00022924855491329477,
-      "loss": 0.1957,
+      "learning_rate": 2.2959537572254335e-05,
+      "loss": 3.9497,
       "step": 910
     },
     {
       "epoch": 4.09,
-      "learning_rate": 0.0002290751445086705,
-      "loss": 0.1838,
+      "learning_rate": 2.294219653179191e-05,
+      "loss": 3.9416,
       "step": 911
     },
     {
       "epoch": 4.09,
-      "learning_rate": 0.0002289017341040462,
-      "loss": 0.1752,
+      "learning_rate": 2.292485549132948e-05,
+      "loss": 3.8963,
       "step": 912
     },
     {
       "epoch": 4.09,
-      "learning_rate": 0.00022872832369942196,
-      "loss": 0.202,
+      "learning_rate": 2.290751445086705e-05,
+      "loss": 3.911,
       "step": 913
     },
     {
       "epoch": 4.1,
-      "learning_rate": 0.00022855491329479768,
-      "loss": 0.1954,
+      "learning_rate": 2.2890173410404623e-05,
+      "loss": 3.9315,
       "step": 914
     },
     {
       "epoch": 4.1,
-      "learning_rate": 0.00022838150289017337,
-      "loss": 0.1885,
+      "learning_rate": 2.28728323699422e-05,
+      "loss": 3.8278,
       "step": 915
     },
     {
       "epoch": 4.11,
-      "learning_rate": 0.0002282080924855491,
-      "loss": 0.1631,
+      "learning_rate": 2.285549132947977e-05,
+      "loss": 3.9261,
       "step": 916
     },
     {
       "epoch": 4.11,
-      "learning_rate": 0.00022803468208092484,
-      "loss": 0.1711,
+      "learning_rate": 2.2838150289017343e-05,
+      "loss": 3.8533,
       "step": 917
     },
     {
       "epoch": 4.12,
-      "learning_rate": 0.00022786127167630056,
-      "loss": 0.1738,
+      "learning_rate": 2.282080924855491e-05,
+      "loss": 3.852,
       "step": 918
     },
     {
       "epoch": 4.12,
-      "learning_rate": 0.00022768786127167628,
-      "loss": 0.1571,
+      "learning_rate": 2.2803468208092487e-05,
+      "loss": 3.8898,
       "step": 919
     },
     {
       "epoch": 4.13,
-      "learning_rate": 0.00022751445086705198,
-      "loss": 0.1176,
+      "learning_rate": 2.278612716763006e-05,
+      "loss": 3.7262,
       "step": 920
     },
     {
       "epoch": 4.13,
-      "learning_rate": 0.00022734104046242772,
-      "loss": 0.1831,
+      "learning_rate": 2.276878612716763e-05,
+      "loss": 3.8449,
       "step": 921
     },
     {
       "epoch": 4.13,
-      "learning_rate": 0.00022716763005780344,
-      "loss": 0.1878,
+      "learning_rate": 2.2751445086705203e-05,
+      "loss": 3.8773,
       "step": 922
     },
     {
       "epoch": 4.14,
-      "learning_rate": 0.00022699421965317917,
-      "loss": 0.1848,
+      "learning_rate": 2.2734104046242775e-05,
+      "loss": 3.8296,
       "step": 923
     },
     {
       "epoch": 4.14,
-      "learning_rate": 0.00022682080924855489,
-      "loss": 0.2019,
+      "learning_rate": 2.2716763005780347e-05,
+      "loss": 3.8417,
       "step": 924
     },
     {
       "epoch": 4.15,
-      "learning_rate": 0.00022664739884393063,
-      "loss": 0.1753,
+      "learning_rate": 2.269942196531792e-05,
+      "loss": 3.8742,
       "step": 925
     },
     {
       "epoch": 4.15,
-      "learning_rate": 0.00022647398843930635,
-      "loss": 0.1706,
+      "learning_rate": 2.268208092485549e-05,
+      "loss": 3.8602,
       "step": 926
     },
     {
       "epoch": 4.16,
-      "learning_rate": 0.00022630057803468205,
-      "loss": 0.1571,
+      "learning_rate": 2.2664739884393067e-05,
+      "loss": 3.8711,
       "step": 927
     },
     {
       "epoch": 4.16,
-      "learning_rate": 0.00022612716763005777,
-      "loss": 0.1915,
+      "learning_rate": 2.2647398843930635e-05,
+      "loss": 3.8169,
       "step": 928
     },
     {
       "epoch": 4.17,
-      "learning_rate": 0.00022595375722543352,
-      "loss": 0.1784,
+      "learning_rate": 2.2630057803468208e-05,
+      "loss": 3.8873,
       "step": 929
     },
     {
       "epoch": 4.17,
-      "learning_rate": 0.00022578034682080924,
-      "loss": 0.1372,
+      "learning_rate": 2.261271676300578e-05,
+      "loss": 3.8648,
       "step": 930
     },
     {
       "epoch": 4.17,
-      "learning_rate": 0.00022560693641618496,
-      "loss": 0.1602,
+      "learning_rate": 2.2595375722543355e-05,
+      "loss": 3.859,
       "step": 931
     },
     {
       "epoch": 4.18,
-      "learning_rate": 0.00022543352601156065,
-      "loss": 0.1511,
+      "learning_rate": 2.2578034682080927e-05,
+      "loss": 3.7546,
       "step": 932
     },
     {
       "epoch": 4.18,
-      "learning_rate": 0.0002252601156069364,
-      "loss": 0.1482,
+      "learning_rate": 2.2560693641618496e-05,
+      "loss": 3.784,
       "step": 933
     },
     {
       "epoch": 4.19,
-      "learning_rate": 0.00022508670520231212,
-      "loss": 0.1547,
+      "learning_rate": 2.2543352601156068e-05,
+      "loss": 3.8621,
       "step": 934
     },
     {
       "epoch": 4.19,
-      "learning_rate": 0.00022491329479768784,
-      "loss": 0.1249,
+      "learning_rate": 2.2526011560693643e-05,
+      "loss": 3.7674,
       "step": 935
     },
     {
       "epoch": 4.2,
-      "learning_rate": 0.00022473988439306356,
-      "loss": 0.1547,
+      "learning_rate": 2.2508670520231215e-05,
+      "loss": 3.8919,
       "step": 936
     },
     {
       "epoch": 4.2,
-      "learning_rate": 0.0002245664739884393,
-      "loss": 0.1549,
+      "learning_rate": 2.2491329479768787e-05,
+      "loss": 3.7846,
       "step": 937
     },
     {
       "epoch": 4.21,
-      "learning_rate": 0.000224393063583815,
-      "loss": 0.138,
+      "learning_rate": 2.2473988439306356e-05,
+      "loss": 3.8464,
       "step": 938
     },
     {
       "epoch": 4.21,
-      "learning_rate": 0.00022421965317919072,
-      "loss": 0.1416,
+      "learning_rate": 2.245664739884393e-05,
+      "loss": 3.7018,
       "step": 939
     },
     {
       "epoch": 4.22,
-      "learning_rate": 0.00022404624277456644,
-      "loss": 0.1237,
+      "learning_rate": 2.2439306358381504e-05,
+      "loss": 3.6997,
       "step": 940
     },
     {
       "epoch": 4.22,
-      "learning_rate": 0.0002238728323699422,
-      "loss": 0.1804,
+      "learning_rate": 2.2421965317919076e-05,
+      "loss": 3.739,
       "step": 941
     },
     {
       "epoch": 4.22,
-      "learning_rate": 0.0002236994219653179,
-      "loss": 0.2177,
+      "learning_rate": 2.2404624277456648e-05,
+      "loss": 3.5986,
       "step": 942
     },
     {
       "epoch": 4.23,
-      "learning_rate": 0.0002235260115606936,
-      "loss": 0.8002,
+      "learning_rate": 2.238728323699422e-05,
+      "loss": 4.7915,
       "step": 943
     },
     {
       "epoch": 4.23,
-      "learning_rate": 0.00022335260115606933,
-      "loss": 0.5931,
+      "learning_rate": 2.2369942196531792e-05,
+      "loss": 4.7906,
       "step": 944
     },
     {
       "epoch": 4.24,
-      "learning_rate": 0.00022317919075144507,
-      "loss": 0.4373,
+      "learning_rate": 2.2352601156069364e-05,
+      "loss": 4.7964,
       "step": 945
     },
     {
       "epoch": 4.24,
-      "learning_rate": 0.0002230057803468208,
-      "loss": 0.3612,
+      "learning_rate": 2.2335260115606936e-05,
+      "loss": 4.8029,
       "step": 946
     },
     {
       "epoch": 4.25,
-      "learning_rate": 0.00022283236994219652,
-      "loss": 0.37,
+      "learning_rate": 2.231791907514451e-05,
+      "loss": 4.7965,
       "step": 947
     },
     {
       "epoch": 4.25,
-      "learning_rate": 0.00022265895953757224,
-      "loss": 0.2676,
+      "learning_rate": 2.2300578034682084e-05,
+      "loss": 4.6215,
       "step": 948
     },
     {
       "epoch": 4.26,
-      "learning_rate": 0.00022248554913294798,
-      "loss": 0.2842,
+      "learning_rate": 2.2283236994219652e-05,
+      "loss": 4.527,
       "step": 949
     },
     {
       "epoch": 4.26,
-      "learning_rate": 0.00022231213872832368,
-      "loss": 0.3132,
+      "learning_rate": 2.2265895953757224e-05,
+      "loss": 4.5305,
       "step": 950
     },
     {
       "epoch": 4.26,
-      "learning_rate": 0.0002221387283236994,
-      "loss": 0.2282,
+      "learning_rate": 2.22485549132948e-05,
+      "loss": 4.4029,
       "step": 951
     },
     {
       "epoch": 4.27,
-      "learning_rate": 0.00022196531791907512,
-      "loss": 0.238,
+      "learning_rate": 2.2231213872832372e-05,
+      "loss": 4.3071,
       "step": 952
     },
     {
       "epoch": 4.27,
-      "learning_rate": 0.00022179190751445087,
-      "loss": 0.2261,
+      "learning_rate": 2.2213872832369944e-05,
+      "loss": 4.1186,
       "step": 953
     },
     {
       "epoch": 4.28,
-      "learning_rate": 0.0002216184971098266,
-      "loss": 0.2274,
+      "learning_rate": 2.2196531791907513e-05,
+      "loss": 4.1175,
       "step": 954
     },
     {
       "epoch": 4.28,
-      "learning_rate": 0.00022144508670520228,
-      "loss": 0.2562,
+      "learning_rate": 2.2179190751445088e-05,
+      "loss": 4.216,
       "step": 955
     },
     {
       "epoch": 4.29,
-      "learning_rate": 0.000221271676300578,
-      "loss": 0.1999,
+      "learning_rate": 2.216184971098266e-05,
+      "loss": 4.1222,
       "step": 956
     },
     {
       "epoch": 4.29,
-      "learning_rate": 0.00022109826589595375,
-      "loss": 0.1859,
+      "learning_rate": 2.2144508670520232e-05,
+      "loss": 4.064,
       "step": 957
     },
     {
       "epoch": 4.3,
-      "learning_rate": 0.00022092485549132947,
-      "loss": 0.1847,
+      "learning_rate": 2.2127167630057804e-05,
+      "loss": 3.9848,
       "step": 958
     },
     {
       "epoch": 4.3,
-      "learning_rate": 0.0002207514450867052,
-      "loss": 0.2134,
+      "learning_rate": 2.2109826589595376e-05,
+      "loss": 4.0032,
       "step": 959
     },
     {
       "epoch": 4.3,
-      "learning_rate": 0.00022057803468208088,
-      "loss": 0.188,
+      "learning_rate": 2.209248554913295e-05,
+      "loss": 3.9009,
       "step": 960
     },
     {
       "epoch": 4.31,
-      "learning_rate": 0.00022040462427745663,
-      "loss": 0.2077,
+      "learning_rate": 2.207514450867052e-05,
+      "loss": 4.0057,
       "step": 961
     },
     {
       "epoch": 4.31,
-      "learning_rate": 0.00022023121387283235,
-      "loss": 0.1951,
+      "learning_rate": 2.2057803468208093e-05,
+      "loss": 4.0278,
       "step": 962
     },
     {
       "epoch": 4.32,
-      "learning_rate": 0.00022005780346820807,
-      "loss": 0.1824,
+      "learning_rate": 2.2040462427745668e-05,
+      "loss": 3.945,
       "step": 963
     },
     {
       "epoch": 4.32,
-      "learning_rate": 0.0002198843930635838,
-      "loss": 0.1861,
+      "learning_rate": 2.2023121387283237e-05,
+      "loss": 3.9708,
       "step": 964
     },
     {
       "epoch": 4.33,
-      "learning_rate": 0.00021971098265895954,
-      "loss": 0.1701,
+      "learning_rate": 2.200578034682081e-05,
+      "loss": 3.9399,
       "step": 965
     },
     {
       "epoch": 4.33,
-      "learning_rate": 0.00021953757225433524,
-      "loss": 0.1884,
+      "learning_rate": 2.198843930635838e-05,
+      "loss": 3.9303,
       "step": 966
     },
     {
       "epoch": 4.34,
-      "learning_rate": 0.00021936416184971096,
-      "loss": 0.152,
+      "learning_rate": 2.1971098265895956e-05,
+      "loss": 3.9469,
       "step": 967
     },
     {
       "epoch": 4.34,
-      "learning_rate": 0.00021919075144508668,
-      "loss": 0.1824,
+      "learning_rate": 2.195375722543353e-05,
+      "loss": 3.9381,
       "step": 968
     },
     {
       "epoch": 4.35,
-      "learning_rate": 0.00021901734104046243,
-      "loss": 0.1769,
+      "learning_rate": 2.1936416184971097e-05,
+      "loss": 3.9298,
       "step": 969
     },
     {
       "epoch": 4.35,
-      "learning_rate": 0.00021884393063583815,
-      "loss": 0.1704,
+      "learning_rate": 2.191907514450867e-05,
+      "loss": 3.9481,
       "step": 970
     },
     {
       "epoch": 4.35,
-      "learning_rate": 0.00021867052023121384,
-      "loss": 0.1562,
+      "learning_rate": 2.1901734104046245e-05,
+      "loss": 3.824,
       "step": 971
     },
     {
       "epoch": 4.36,
-      "learning_rate": 0.00021849710982658956,
-      "loss": 0.1738,
+      "learning_rate": 2.1884393063583817e-05,
+      "loss": 3.8024,
       "step": 972
     },
     {
       "epoch": 4.36,
-      "learning_rate": 0.0002183236994219653,
-      "loss": 0.1492,
+      "learning_rate": 2.186705202312139e-05,
+      "loss": 3.9022,
       "step": 973
     },
     {
       "epoch": 4.37,
-      "learning_rate": 0.00021815028901734103,
-      "loss": 0.1489,
+      "learning_rate": 2.184971098265896e-05,
+      "loss": 3.7766,
       "step": 974
     },
     {
       "epoch": 4.37,
-      "learning_rate": 0.00021797687861271675,
-      "loss": 0.1694,
+      "learning_rate": 2.1832369942196533e-05,
+      "loss": 3.8805,
       "step": 975
     },
     {
       "epoch": 4.38,
-      "learning_rate": 0.00021780346820809247,
-      "loss": 0.1627,
+      "learning_rate": 2.1815028901734105e-05,
+      "loss": 3.8413,
       "step": 976
     },
     {
       "epoch": 4.38,
-      "learning_rate": 0.00021763005780346822,
-      "loss": 0.1595,
+      "learning_rate": 2.1797687861271677e-05,
+      "loss": 3.8996,
       "step": 977
     },
     {
       "epoch": 4.39,
-      "learning_rate": 0.0002174566473988439,
-      "loss": 0.165,
+      "learning_rate": 2.178034682080925e-05,
+      "loss": 3.8482,
       "step": 978
     },
     {
       "epoch": 4.39,
-      "learning_rate": 0.00021728323699421963,
-      "loss": 0.1581,
+      "learning_rate": 2.1763005780346824e-05,
+      "loss": 3.8851,
       "step": 979
     },
     {
       "epoch": 4.39,
-      "learning_rate": 0.00021710982658959535,
-      "loss": 0.1361,
+      "learning_rate": 2.1745664739884393e-05,
+      "loss": 3.8244,
       "step": 980
     },
     {
       "epoch": 4.4,
-      "learning_rate": 0.0002169364161849711,
-      "loss": 0.1503,
+      "learning_rate": 2.1728323699421965e-05,
+      "loss": 3.8388,
       "step": 981
     },
     {
       "epoch": 4.4,
-      "learning_rate": 0.00021676300578034682,
-      "loss": 0.1666,
+      "learning_rate": 2.1710982658959537e-05,
+      "loss": 3.7814,
       "step": 982
     },
     {
       "epoch": 4.41,
-      "learning_rate": 0.00021658959537572251,
-      "loss": 0.1467,
+      "learning_rate": 2.1693641618497113e-05,
+      "loss": 3.7747,
       "step": 983
     },
     {
       "epoch": 4.41,
-      "learning_rate": 0.00021641618497109824,
-      "loss": 0.1599,
+      "learning_rate": 2.1676300578034685e-05,
+      "loss": 3.8686,
       "step": 984
     },
     {
       "epoch": 4.42,
-      "learning_rate": 0.00021624277456647396,
-      "loss": 0.1572,
+      "learning_rate": 2.1658959537572253e-05,
+      "loss": 3.8083,
       "step": 985
     },
     {
       "epoch": 4.42,
-      "learning_rate": 0.0002160693641618497,
-      "loss": 0.1825,
+      "learning_rate": 2.1641618497109826e-05,
+      "loss": 3.9,
       "step": 986
     },
     {
       "epoch": 4.43,
-      "learning_rate": 0.00021589595375722542,
-      "loss": 0.1265,
+      "learning_rate": 2.1624277456647398e-05,
+      "loss": 3.7535,
       "step": 987
     },
     {
       "epoch": 4.43,
-      "learning_rate": 0.00021572254335260112,
-      "loss": 0.136,
+      "learning_rate": 2.1606936416184973e-05,
+      "loss": 3.8612,
       "step": 988
     },
     {
       "epoch": 4.43,
-      "learning_rate": 0.00021554913294797684,
-      "loss": 0.1099,
+      "learning_rate": 2.1589595375722545e-05,
+      "loss": 3.668,
       "step": 989
     },
     {
       "epoch": 4.44,
-      "learning_rate": 0.00021537572254335259,
-      "loss": 0.1753,
+      "learning_rate": 2.1572254335260114e-05,
+      "loss": 3.9476,
       "step": 990
     },
     {
       "epoch": 4.44,
-      "learning_rate": 0.0002152023121387283,
-      "loss": 0.1497,
+      "learning_rate": 2.1554913294797686e-05,
+      "loss": 3.8109,
       "step": 991
     },
     {
       "epoch": 4.45,
-      "learning_rate": 0.00021502890173410403,
-      "loss": 0.2148,
+      "learning_rate": 2.153757225433526e-05,
+      "loss": 3.7516,
       "step": 992
     },
     {
       "epoch": 4.45,
-      "learning_rate": 0.00021485549132947972,
-      "loss": 0.9389,
+      "learning_rate": 2.1520231213872833e-05,
+      "loss": 4.5727,
       "step": 993
     },
     {
       "epoch": 4.46,
-      "learning_rate": 0.00021468208092485547,
-      "loss": 0.619,
+      "learning_rate": 2.1502890173410405e-05,
+      "loss": 4.653,
       "step": 994
     },
     {
       "epoch": 4.46,
-      "learning_rate": 0.0002145086705202312,
-      "loss": 0.4313,
+      "learning_rate": 2.1485549132947974e-05,
+      "loss": 4.7266,
       "step": 995
     },
     {
       "epoch": 4.47,
-      "learning_rate": 0.0002143352601156069,
-      "loss": 0.3153,
+      "learning_rate": 2.146820809248555e-05,
+      "loss": 4.5639,
       "step": 996
     },
     {
       "epoch": 4.47,
-      "learning_rate": 0.00021416184971098263,
-      "loss": 0.38,
+      "learning_rate": 2.145086705202312e-05,
+      "loss": 4.7074,
       "step": 997
     },
     {
       "epoch": 4.48,
-      "learning_rate": 0.00021398843930635838,
-      "loss": 0.2856,
+      "learning_rate": 2.1433526011560694e-05,
+      "loss": 4.5743,
       "step": 998
     },
     {
       "epoch": 4.48,
-      "learning_rate": 0.0002138150289017341,
-      "loss": 0.3066,
+      "learning_rate": 2.1416184971098266e-05,
+      "loss": 4.532,
       "step": 999
     },
     {
       "epoch": 4.48,
-      "learning_rate": 0.0002136416184971098,
-      "loss": 0.2133,
+      "learning_rate": 2.139884393063584e-05,
+      "loss": 4.4811,
       "step": 1000
     },
     {
       "epoch": 4.48,
-      "eval_loss": 0.3967271149158478,
-      "eval_runtime": 570.6501,
-      "eval_samples_per_second": 4.63,
-      "eval_steps_per_second": 0.58,
-      "eval_wer": 0.12705275684252282,
+      "eval_loss": 4.471442222595215,
+      "eval_runtime": 410.0311,
+      "eval_samples_per_second": 6.443,
+      "eval_steps_per_second": 0.807,
+      "eval_wer": 0.9456564855216184,
       "step": 1000
     },
     {
       "epoch": 4.49,
-      "learning_rate": 0.00021346820809248551,
-      "loss": 0.2367,
+      "learning_rate": 2.138150289017341e-05,
+      "loss": 4.4202,
       "step": 1001
     },
     {
       "epoch": 4.49,
-      "learning_rate": 0.00021329479768786126,
-      "loss": 0.219,
+      "learning_rate": 2.1364161849710982e-05,
+      "loss": 4.3316,
       "step": 1002
     },
     {
       "epoch": 4.5,
-      "learning_rate": 0.00021312138728323698,
-      "loss": 0.1921,
+      "learning_rate": 2.1346820809248554e-05,
+      "loss": 4.1793,
       "step": 1003
     },
     {
       "epoch": 4.5,
-      "learning_rate": 0.0002129479768786127,
-      "loss": 0.1953,
+      "learning_rate": 2.132947976878613e-05,
+      "loss": 4.1143,
       "step": 1004
     },
     {
       "epoch": 4.51,
-      "learning_rate": 0.0002127745664739884,
-      "loss": 0.2085,
+      "learning_rate": 2.13121387283237e-05,
+      "loss": 4.1652,
       "step": 1005
     },
     {
       "epoch": 4.51,
-      "learning_rate": 0.00021260115606936414,
-      "loss": 0.1831,
+      "learning_rate": 2.129479768786127e-05,
+      "loss": 3.9561,
       "step": 1006
     },
     {
       "epoch": 4.52,
-      "learning_rate": 0.00021242774566473987,
-      "loss": 0.2097,
+      "learning_rate": 2.1277456647398842e-05,
+      "loss": 4.0482,
       "step": 1007
     },
     {
       "epoch": 4.52,
-      "learning_rate": 0.00021225433526011559,
-      "loss": 0.1797,
+      "learning_rate": 2.1260115606936418e-05,
+      "loss": 4.0044,
       "step": 1008
     },
     {
       "epoch": 4.52,
-      "learning_rate": 0.0002120809248554913,
-      "loss": 0.1994,
+      "learning_rate": 2.124277456647399e-05,
+      "loss": 4.0244,
       "step": 1009
     },
     {
       "epoch": 4.53,
-      "learning_rate": 0.00021190751445086705,
-      "loss": 0.173,
+      "learning_rate": 2.1225433526011562e-05,
+      "loss": 3.9018,
       "step": 1010
     },
     {
       "epoch": 4.53,
-      "learning_rate": 0.00021173410404624275,
-      "loss": 0.1891,
+      "learning_rate": 2.120809248554913e-05,
+      "loss": 3.9488,
       "step": 1011
     },
     {
       "epoch": 4.54,
-      "learning_rate": 0.00021156069364161847,
-      "loss": 0.1908,
+      "learning_rate": 2.1190751445086706e-05,
+      "loss": 3.9324,
       "step": 1012
     },
     {
       "epoch": 4.54,
-      "learning_rate": 0.0002113872832369942,
-      "loss": 0.2069,
+      "learning_rate": 2.1173410404624278e-05,
+      "loss": 4.0152,
       "step": 1013
     },
     {
       "epoch": 4.55,
-      "learning_rate": 0.00021121387283236994,
-      "loss": 0.1566,
+      "learning_rate": 2.115606936416185e-05,
+      "loss": 3.9643,
       "step": 1014
     },
     {
       "epoch": 4.55,
-      "learning_rate": 0.00021104046242774566,
-      "loss": 0.1691,
+      "learning_rate": 2.1138728323699422e-05,
+      "loss": 3.8474,
       "step": 1015
     },
     {
       "epoch": 4.56,
-      "learning_rate": 0.00021086705202312135,
-      "loss": 0.1543,
+      "learning_rate": 2.1121387283236994e-05,
+      "loss": 3.91,
       "step": 1016
     },
     {
       "epoch": 4.56,
-      "learning_rate": 0.00021069364161849707,
-      "loss": 0.162,
+      "learning_rate": 2.1104046242774566e-05,
+      "loss": 3.9484,
       "step": 1017
     },
     {
       "epoch": 4.57,
-      "learning_rate": 0.00021052023121387282,
-      "loss": 0.1871,
+      "learning_rate": 2.108670520231214e-05,
+      "loss": 3.9356,
       "step": 1018
     },
     {
       "epoch": 4.57,
-      "learning_rate": 0.00021034682080924854,
-      "loss": 0.179,
+      "learning_rate": 2.106936416184971e-05,
+      "loss": 3.8633,
       "step": 1019
     },
     {
       "epoch": 4.57,
-      "learning_rate": 0.00021017341040462426,
-      "loss": 0.1806,
+      "learning_rate": 2.1052023121387286e-05,
+      "loss": 3.8511,
       "step": 1020
     },
     {
       "epoch": 4.58,
-      "learning_rate": 0.00020999999999999998,
-      "loss": 0.1621,
+      "learning_rate": 2.1034682080924855e-05,
+      "loss": 3.8245,
       "step": 1021
     },
     {
       "epoch": 4.58,
-      "learning_rate": 0.00020982658959537573,
-      "loss": 0.1616,
+      "learning_rate": 2.1017341040462427e-05,
+      "loss": 3.8697,
       "step": 1022
     },
     {
       "epoch": 4.59,
-      "learning_rate": 0.00020965317919075142,
-      "loss": 0.1854,
+      "learning_rate": 2.1e-05,
+      "loss": 3.8685,
       "step": 1023
     },
     {
       "epoch": 4.59,
-      "learning_rate": 0.00020947976878612714,
-      "loss": 0.1454,
+      "learning_rate": 2.0982658959537574e-05,
+      "loss": 3.7468,
       "step": 1024
     },
     {
       "epoch": 4.6,
-      "learning_rate": 0.00020930635838150286,
-      "loss": 0.2193,
+      "learning_rate": 2.0965317919075146e-05,
+      "loss": 3.7574,
       "step": 1025
     },
     {
       "epoch": 4.6,
-      "learning_rate": 0.0002091329479768786,
-      "loss": 0.1518,
+      "learning_rate": 2.094797687861272e-05,
+      "loss": 3.8157,
       "step": 1026
     },
     {
       "epoch": 4.61,
-      "learning_rate": 0.00020895953757225433,
-      "loss": 0.1593,
+      "learning_rate": 2.0930635838150287e-05,
+      "loss": 3.8817,
       "step": 1027
     },
     {
       "epoch": 4.61,
-      "learning_rate": 0.00020878612716763003,
-      "loss": 0.1579,
+      "learning_rate": 2.0913294797687863e-05,
+      "loss": 3.761,
       "step": 1028
     },
     {
       "epoch": 4.61,
-      "learning_rate": 0.00020861271676300575,
-      "loss": 0.1378,
+      "learning_rate": 2.0895953757225435e-05,
+      "loss": 3.8213,
       "step": 1029
     },
     {
       "epoch": 4.62,
-      "learning_rate": 0.0002084393063583815,
-      "loss": 0.11,
+      "learning_rate": 2.0878612716763007e-05,
+      "loss": 3.7712,
       "step": 1030
     },
     {
       "epoch": 4.62,
-      "learning_rate": 0.00020826589595375722,
-      "loss": 0.1537,
+      "learning_rate": 2.086127167630058e-05,
+      "loss": 3.8045,
       "step": 1031
     },
     {
       "epoch": 4.63,
-      "learning_rate": 0.00020809248554913294,
-      "loss": 0.1464,
+      "learning_rate": 2.084393063583815e-05,
+      "loss": 3.8181,
       "step": 1032
     },
     {
       "epoch": 4.63,
-      "learning_rate": 0.00020791907514450863,
-      "loss": 0.1472,
+      "learning_rate": 2.0826589595375723e-05,
+      "loss": 3.915,
       "step": 1033
     },
     {
       "epoch": 4.64,
-      "learning_rate": 0.00020774566473988438,
-      "loss": 0.1454,
+      "learning_rate": 2.0809248554913295e-05,
+      "loss": 3.7518,
       "step": 1034
     },
     {
       "epoch": 4.64,
-      "learning_rate": 0.0002075722543352601,
-      "loss": 0.1799,
+      "learning_rate": 2.0791907514450867e-05,
+      "loss": 3.8364,
       "step": 1035
     },
     {
       "epoch": 4.65,
-      "learning_rate": 0.00020739884393063582,
-      "loss": 0.1408,
+      "learning_rate": 2.0774566473988442e-05,
+      "loss": 3.9519,
       "step": 1036
     },
     {
       "epoch": 4.65,
-      "learning_rate": 0.00020722543352601154,
-      "loss": 0.1415,
+      "learning_rate": 2.075722543352601e-05,
+      "loss": 3.8736,
       "step": 1037
     },
     {
       "epoch": 4.65,
-      "learning_rate": 0.0002070520231213873,
-      "loss": 0.1279,
+      "learning_rate": 2.0739884393063583e-05,
+      "loss": 3.6868,
       "step": 1038
     },
     {
       "epoch": 4.66,
-      "learning_rate": 0.00020687861271676298,
-      "loss": 0.1413,
+      "learning_rate": 2.0722543352601155e-05,
+      "loss": 3.8773,
       "step": 1039
     },
     {
       "epoch": 4.66,
-      "learning_rate": 0.0002067052023121387,
-      "loss": 0.1423,
+      "learning_rate": 2.070520231213873e-05,
+      "loss": 3.6867,
       "step": 1040
     },
     {
       "epoch": 4.67,
-      "learning_rate": 0.00020653179190751442,
-      "loss": 0.1867,
+      "learning_rate": 2.0687861271676303e-05,
+      "loss": 3.8537,
       "step": 1041
     },
     {
       "epoch": 4.67,
-      "learning_rate": 0.00020635838150289017,
-      "loss": 0.2485,
+      "learning_rate": 2.067052023121387e-05,
+      "loss": 3.77,
       "step": 1042
     },
     {
       "epoch": 4.68,
-      "learning_rate": 0.0002061849710982659,
-      "loss": 0.5592,
+      "learning_rate": 2.0653179190751444e-05,
+      "loss": 4.5353,
       "step": 1043
     },
     {
       "epoch": 4.68,
-      "learning_rate": 0.00020601156069364158,
-      "loss": 0.4164,
+      "learning_rate": 2.063583815028902e-05,
+      "loss": 4.4984,
       "step": 1044
     },
     {
       "epoch": 4.69,
-      "learning_rate": 0.0002058381502890173,
-      "loss": 0.3318,
+      "learning_rate": 2.061849710982659e-05,
+      "loss": 4.5213,
       "step": 1045
     },
     {
       "epoch": 4.69,
-      "learning_rate": 0.00020566473988439305,
-      "loss": 0.3236,
+      "learning_rate": 2.0601156069364163e-05,
+      "loss": 4.5536,
       "step": 1046
     },
     {
       "epoch": 4.7,
-      "learning_rate": 0.00020549132947976877,
-      "loss": 0.2769,
+      "learning_rate": 2.0583815028901732e-05,
+      "loss": 4.5843,
       "step": 1047
     },
     {
       "epoch": 4.7,
-      "learning_rate": 0.0002053179190751445,
-      "loss": 0.317,
+      "learning_rate": 2.0566473988439307e-05,
+      "loss": 4.5179,
       "step": 1048
     },
     {
       "epoch": 4.7,
-      "learning_rate": 0.00020514450867052021,
-      "loss": 0.2551,
+      "learning_rate": 2.054913294797688e-05,
+      "loss": 4.4883,
       "step": 1049
     },
     {
       "epoch": 4.71,
-      "learning_rate": 0.00020497109826589596,
-      "loss": 0.246,
+      "learning_rate": 2.053179190751445e-05,
+      "loss": 4.4311,
       "step": 1050
     },
     {
       "epoch": 4.71,
-      "learning_rate": 0.00020479768786127166,
-      "loss": 0.2346,
+      "learning_rate": 2.0514450867052024e-05,
+      "loss": 4.2903,
       "step": 1051
     },
     {
       "epoch": 4.72,
-      "learning_rate": 0.00020462427745664738,
-      "loss": 0.2146,
+      "learning_rate": 2.04971098265896e-05,
+      "loss": 4.2673,
       "step": 1052
     },
     {
       "epoch": 4.72,
-      "learning_rate": 0.0002044508670520231,
-      "loss": 0.1969,
+      "learning_rate": 2.0479768786127168e-05,
+      "loss": 4.2055,
       "step": 1053
     },
     {
       "epoch": 4.73,
-      "learning_rate": 0.00020427745664739885,
-      "loss": 0.243,
+      "learning_rate": 2.046242774566474e-05,
+      "loss": 4.1595,
       "step": 1054
     },
     {
       "epoch": 4.73,
-      "learning_rate": 0.00020410404624277457,
-      "loss": 0.1915,
+      "learning_rate": 2.0445086705202312e-05,
+      "loss": 4.0374,
       "step": 1055
     },
     {
       "epoch": 4.74,
-      "learning_rate": 0.00020393063583815026,
-      "loss": 0.1672,
+      "learning_rate": 2.0427745664739887e-05,
+      "loss": 3.9281,
       "step": 1056
     },
     {
       "epoch": 4.74,
-      "learning_rate": 0.00020375722543352598,
-      "loss": 0.213,
+      "learning_rate": 2.041040462427746e-05,
+      "loss": 4.02,
       "step": 1057
     },
     {
       "epoch": 4.74,
-      "learning_rate": 0.00020358381502890173,
-      "loss": 0.178,
+      "learning_rate": 2.0393063583815028e-05,
+      "loss": 3.9746,
       "step": 1058
     },
     {
       "epoch": 4.75,
-      "learning_rate": 0.00020341040462427745,
-      "loss": 0.1977,
+      "learning_rate": 2.03757225433526e-05,
+      "loss": 4.0699,
       "step": 1059
     },
     {
       "epoch": 4.75,
-      "learning_rate": 0.00020323699421965317,
-      "loss": 0.1956,
+      "learning_rate": 2.0358381502890176e-05,
+      "loss": 4.0389,
       "step": 1060
     },
     {
       "epoch": 4.76,
-      "learning_rate": 0.00020306358381502886,
-      "loss": 0.1645,
+      "learning_rate": 2.0341040462427748e-05,
+      "loss": 3.9078,
       "step": 1061
     },
     {
       "epoch": 4.76,
-      "learning_rate": 0.0002028901734104046,
-      "loss": 0.19,
+      "learning_rate": 2.032369942196532e-05,
+      "loss": 3.9825,
       "step": 1062
     },
     {
       "epoch": 4.77,
-      "learning_rate": 0.00020271676300578033,
-      "loss": 0.1341,
+      "learning_rate": 2.030635838150289e-05,
+      "loss": 3.9332,
       "step": 1063
     },
     {
       "epoch": 4.77,
-      "learning_rate": 0.00020254335260115605,
-      "loss": 0.1533,
+      "learning_rate": 2.0289017341040464e-05,
+      "loss": 3.9562,
       "step": 1064
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00020236994219653177,
-      "loss": 0.1639,
+      "learning_rate": 2.0271676300578036e-05,
+      "loss": 3.8246,
       "step": 1065
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00020219653179190752,
-      "loss": 0.1612,
+      "learning_rate": 2.0254335260115608e-05,
+      "loss": 3.9159,
       "step": 1066
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00020202312138728321,
-      "loss": 0.1496,
+      "learning_rate": 2.023699421965318e-05,
+      "loss": 3.912,
       "step": 1067
     },
     {
       "epoch": 4.79,
-      "learning_rate": 0.00020184971098265893,
-      "loss": 0.1454,
+      "learning_rate": 2.0219653179190752e-05,
+      "loss": 3.9218,
       "step": 1068
     },
     {
       "epoch": 4.79,
-      "learning_rate": 0.00020167630057803466,
-      "loss": 0.1465,
+      "learning_rate": 2.0202312138728324e-05,
+      "loss": 3.9524,
       "step": 1069
     },
     {
       "epoch": 4.8,
-      "learning_rate": 0.0002015028901734104,
-      "loss": 0.1555,
+      "learning_rate": 2.0184971098265896e-05,
+      "loss": 3.8963,
       "step": 1070
     },
     {
       "epoch": 4.8,
-      "learning_rate": 0.00020132947976878612,
-      "loss": 0.1481,
+      "learning_rate": 2.0167630057803468e-05,
+      "loss": 3.9146,
       "step": 1071
     },
     {
       "epoch": 4.81,
-      "learning_rate": 0.00020115606936416184,
-      "loss": 0.1376,
+      "learning_rate": 2.0150289017341044e-05,
+      "loss": 3.9796,
       "step": 1072
     },
     {
       "epoch": 4.81,
-      "learning_rate": 0.00020098265895953754,
-      "loss": 0.1617,
+      "learning_rate": 2.0132947976878612e-05,
+      "loss": 3.9317,
       "step": 1073
     },
     {
       "epoch": 4.82,
-      "learning_rate": 0.00020080924855491329,
-      "loss": 0.1281,
+      "learning_rate": 2.0115606936416184e-05,
+      "loss": 3.7378,
       "step": 1074
     },
     {
       "epoch": 4.82,
-      "learning_rate": 0.000200635838150289,
-      "loss": 0.1483,
+      "learning_rate": 2.0098265895953757e-05,
+      "loss": 3.7175,
       "step": 1075
     },
     {
       "epoch": 4.83,
-      "learning_rate": 0.00020046242774566473,
-      "loss": 0.1589,
+      "learning_rate": 2.0080924855491332e-05,
+      "loss": 3.8474,
       "step": 1076
     },
     {
       "epoch": 4.83,
-      "learning_rate": 0.00020028901734104045,
-      "loss": 0.1597,
+      "learning_rate": 2.0063583815028904e-05,
+      "loss": 3.9127,
       "step": 1077
     },
     {
       "epoch": 4.83,
-      "learning_rate": 0.0002001156069364162,
-      "loss": 0.1448,
+      "learning_rate": 2.0046242774566476e-05,
+      "loss": 3.7991,
       "step": 1078
     },
     {
       "epoch": 4.84,
-      "learning_rate": 0.0001999421965317919,
-      "loss": 0.1361,
+      "learning_rate": 2.0028901734104045e-05,
+      "loss": 3.7866,
       "step": 1079
     },
     {
       "epoch": 4.84,
-      "learning_rate": 0.0001997687861271676,
-      "loss": 0.1457,
+      "learning_rate": 2.001156069364162e-05,
+      "loss": 3.8647,
       "step": 1080
     },
     {
       "epoch": 4.85,
-      "learning_rate": 0.00019959537572254333,
-      "loss": 0.1266,
+      "learning_rate": 1.9994219653179192e-05,
+      "loss": 3.7537,
       "step": 1081
     },
     {
       "epoch": 4.85,
-      "learning_rate": 0.00019942196531791905,
-      "loss": 0.14,
+      "learning_rate": 1.9976878612716764e-05,
+      "loss": 3.8178,
       "step": 1082
     },
     {
       "epoch": 4.86,
-      "learning_rate": 0.0001992485549132948,
-      "loss": 0.1396,
+      "learning_rate": 1.9959537572254336e-05,
+      "loss": 3.8901,
       "step": 1083
     },
     {
       "epoch": 4.86,
-      "learning_rate": 0.0001990751445086705,
-      "loss": 0.1292,
+      "learning_rate": 1.9942196531791905e-05,
+      "loss": 3.8224,
       "step": 1084
     },
     {
       "epoch": 4.87,
-      "learning_rate": 0.0001989017341040462,
-      "loss": 0.1202,
+      "learning_rate": 1.992485549132948e-05,
+      "loss": 3.8457,
       "step": 1085
     },
     {
       "epoch": 4.87,
-      "learning_rate": 0.00019872832369942193,
-      "loss": 0.1194,
+      "learning_rate": 1.9907514450867053e-05,
+      "loss": 3.8248,
       "step": 1086
     },
     {
       "epoch": 4.87,
-      "learning_rate": 0.00019855491329479768,
-      "loss": 0.1251,
+      "learning_rate": 1.9890173410404625e-05,
+      "loss": 3.8016,
       "step": 1087
     },
     {
       "epoch": 4.88,
-      "learning_rate": 0.0001983815028901734,
-      "loss": 0.1142,
+      "learning_rate": 1.9872832369942197e-05,
+      "loss": 3.8961,
       "step": 1088
     },
     {
       "epoch": 4.88,
-      "learning_rate": 0.0001982080924855491,
-      "loss": 0.1977,
+      "learning_rate": 1.985549132947977e-05,
+      "loss": 3.8621,
       "step": 1089
     },
     {
       "epoch": 4.89,
-      "learning_rate": 0.00019803468208092482,
-      "loss": 0.1501,
+      "learning_rate": 1.983815028901734e-05,
+      "loss": 3.7512,
       "step": 1090
     },
     {
       "epoch": 4.89,
-      "learning_rate": 0.00019786127167630056,
-      "loss": 0.1647,
+      "learning_rate": 1.9820809248554913e-05,
+      "loss": 3.8534,
       "step": 1091
     },
     {
       "epoch": 4.9,
-      "learning_rate": 0.00019768786127167629,
-      "loss": 0.2021,
+      "learning_rate": 1.9803468208092485e-05,
+      "loss": 3.7219,
       "step": 1092
     },
     {
       "epoch": 4.9,
-      "learning_rate": 0.000197514450867052,
-      "loss": 0.4591,
+      "learning_rate": 1.978612716763006e-05,
+      "loss": 4.4019,
       "step": 1093
     },
     {
       "epoch": 4.91,
-      "learning_rate": 0.0001973410404624277,
-      "loss": 0.3462,
+      "learning_rate": 1.976878612716763e-05,
+      "loss": 4.4389,
       "step": 1094
     },
     {
       "epoch": 4.91,
-      "learning_rate": 0.00019716763005780345,
-      "loss": 0.2885,
+      "learning_rate": 1.97514450867052e-05,
+      "loss": 4.4819,
       "step": 1095
     },
     {
       "epoch": 4.91,
-      "learning_rate": 0.00019699421965317917,
-      "loss": 0.2618,
+      "learning_rate": 1.9734104046242773e-05,
+      "loss": 4.436,
       "step": 1096
     },
     {
       "epoch": 4.92,
-      "learning_rate": 0.0001968208092485549,
-      "loss": 0.2376,
+      "learning_rate": 1.971676300578035e-05,
+      "loss": 4.3388,
       "step": 1097
     },
     {
       "epoch": 4.92,
-      "learning_rate": 0.0001966473988439306,
-      "loss": 0.1721,
+      "learning_rate": 1.969942196531792e-05,
+      "loss": 4.2991,
       "step": 1098
     },
     {
       "epoch": 4.93,
-      "learning_rate": 0.00019647398843930636,
-      "loss": 0.1815,
+      "learning_rate": 1.968208092485549e-05,
+      "loss": 4.162,
       "step": 1099
     },
     {
       "epoch": 4.93,
-      "learning_rate": 0.00019630057803468208,
-      "loss": 0.1885,
+      "learning_rate": 1.966473988439306e-05,
+      "loss": 4.1708,
       "step": 1100
     },
     {
       "epoch": 4.94,
-      "learning_rate": 0.00019612716763005777,
-      "loss": 0.1934,
+      "learning_rate": 1.9647398843930637e-05,
+      "loss": 4.1403,
       "step": 1101
     },
     {
       "epoch": 4.94,
-      "learning_rate": 0.0001959537572254335,
-      "loss": 0.1607,
+      "learning_rate": 1.963005780346821e-05,
+      "loss": 3.989,
       "step": 1102
     },
     {
       "epoch": 4.95,
-      "learning_rate": 0.00019578034682080924,
-      "loss": 0.1666,
+      "learning_rate": 1.961271676300578e-05,
+      "loss": 4.0543,
       "step": 1103
     },
     {
       "epoch": 4.95,
-      "learning_rate": 0.00019560693641618496,
-      "loss": 0.1609,
+      "learning_rate": 1.9595375722543353e-05,
+      "loss": 4.0557,
       "step": 1104
     },
     {
       "epoch": 4.96,
-      "learning_rate": 0.00019543352601156068,
-      "loss": 0.1719,
+      "learning_rate": 1.9578034682080925e-05,
+      "loss": 3.939,
       "step": 1105
     },
     {
       "epoch": 4.96,
-      "learning_rate": 0.00019526011560693637,
-      "loss": 0.1764,
+      "learning_rate": 1.9560693641618497e-05,
+      "loss": 3.8684,
       "step": 1106
     },
     {
       "epoch": 4.96,
-      "learning_rate": 0.00019508670520231212,
-      "loss": 0.1512,
+      "learning_rate": 1.954335260115607e-05,
+      "loss": 3.8731,
       "step": 1107
     },
     {
       "epoch": 4.97,
-      "learning_rate": 0.00019491329479768784,
-      "loss": 0.1631,
+      "learning_rate": 1.952601156069364e-05,
+      "loss": 3.9208,
       "step": 1108
     },
     {
       "epoch": 4.97,
-      "learning_rate": 0.00019473988439306356,
-      "loss": 0.1507,
+      "learning_rate": 1.9508670520231217e-05,
+      "loss": 3.8534,
       "step": 1109
     },
     {
       "epoch": 4.98,
-      "learning_rate": 0.00019456647398843928,
-      "loss": 0.1329,
+      "learning_rate": 1.9491329479768786e-05,
+      "loss": 3.8254,
       "step": 1110
     },
     {
       "epoch": 4.98,
-      "learning_rate": 0.00019439306358381503,
-      "loss": 0.1985,
+      "learning_rate": 1.9473988439306358e-05,
+      "loss": 3.8774,
       "step": 1111
     },
     {
       "epoch": 4.99,
-      "learning_rate": 0.00019421965317919073,
-      "loss": 0.1496,
+      "learning_rate": 1.945664739884393e-05,
+      "loss": 3.832,
       "step": 1112
     },
     {
       "epoch": 4.99,
-      "learning_rate": 0.00019404624277456645,
-      "loss": 0.1347,
+      "learning_rate": 1.9439306358381505e-05,
+      "loss": 3.8714,
       "step": 1113
     },
     {
       "epoch": 5.0,
-      "learning_rate": 0.00019387283236994217,
-      "loss": 0.1417,
+      "learning_rate": 1.9421965317919077e-05,
+      "loss": 3.9393,
       "step": 1114
     },
     {
       "epoch": 5.0,
-      "learning_rate": 0.00019369942196531792,
-      "loss": 0.2601,
+      "learning_rate": 1.9404624277456646e-05,
+      "loss": 4.1588,
       "step": 1115
     },
     {
       "epoch": 5.0,
-      "learning_rate": 0.00019352601156069364,
-      "loss": 0.271,
+      "learning_rate": 1.9387283236994218e-05,
+      "loss": 4.2121,
       "step": 1116
     },
     {
       "epoch": 5.01,
-      "learning_rate": 0.00019335260115606933,
-      "loss": 0.2395,
+      "learning_rate": 1.9369942196531794e-05,
+      "loss": 4.1375,
       "step": 1117
     },
     {
       "epoch": 5.01,
-      "learning_rate": 0.00019317919075144505,
-      "loss": 0.1913,
+      "learning_rate": 1.9352601156069366e-05,
+      "loss": 4.1691,
       "step": 1118
     },
     {
       "epoch": 5.02,
-      "learning_rate": 0.0001930057803468208,
-      "loss": 0.1692,
+      "learning_rate": 1.9335260115606938e-05,
+      "loss": 4.1969,
       "step": 1119
     },
     {
       "epoch": 5.02,
-      "learning_rate": 0.00019283236994219652,
-      "loss": 0.1818,
+      "learning_rate": 1.9317919075144506e-05,
+      "loss": 4.1504,
       "step": 1120
     },
     {
       "epoch": 5.03,
-      "learning_rate": 0.00019265895953757224,
-      "loss": 0.1619,
+      "learning_rate": 1.9300578034682082e-05,
+      "loss": 4.0819,
       "step": 1121
     },
     {
       "epoch": 5.03,
-      "learning_rate": 0.00019248554913294796,
-      "loss": 0.1145,
+      "learning_rate": 1.9283236994219654e-05,
+      "loss": 4.0956,
       "step": 1122
     },
     {
       "epoch": 5.04,
-      "learning_rate": 0.0001923121387283237,
-      "loss": 0.1634,
+      "learning_rate": 1.9265895953757226e-05,
+      "loss": 4.0521,
       "step": 1123
     },
     {
       "epoch": 5.04,
-      "learning_rate": 0.0001921387283236994,
-      "loss": 0.1414,
+      "learning_rate": 1.9248554913294798e-05,
+      "loss": 4.1007,
       "step": 1124
     },
     {
       "epoch": 5.04,
-      "learning_rate": 0.00019196531791907512,
-      "loss": 0.1077,
+      "learning_rate": 1.923121387283237e-05,
+      "loss": 4.0153,
       "step": 1125
     },
     {
       "epoch": 5.05,
-      "learning_rate": 0.00019179190751445084,
-      "loss": 0.1225,
+      "learning_rate": 1.9213872832369942e-05,
+      "loss": 3.9042,
       "step": 1126
     },
     {
       "epoch": 5.05,
-      "learning_rate": 0.0001916184971098266,
-      "loss": 0.1166,
+      "learning_rate": 1.9196531791907514e-05,
+      "loss": 3.9075,
       "step": 1127
     },
     {
       "epoch": 5.06,
-      "learning_rate": 0.0001914450867052023,
-      "loss": 0.115,
+      "learning_rate": 1.9179190751445086e-05,
+      "loss": 3.9057,
       "step": 1128
     },
     {
       "epoch": 5.06,
-      "learning_rate": 0.000191271676300578,
-      "loss": 0.1072,
+      "learning_rate": 1.9161849710982662e-05,
+      "loss": 3.9364,
       "step": 1129
     },
     {
       "epoch": 5.07,
-      "learning_rate": 0.00019109826589595373,
-      "loss": 0.1018,
+      "learning_rate": 1.9144508670520234e-05,
+      "loss": 3.8627,
       "step": 1130
     },
     {
       "epoch": 5.07,
-      "learning_rate": 0.00019092485549132947,
-      "loss": 0.1254,
+      "learning_rate": 1.9127167630057802e-05,
+      "loss": 3.8161,
       "step": 1131
     },
     {
       "epoch": 5.08,
-      "learning_rate": 0.0001907514450867052,
-      "loss": 0.1188,
+      "learning_rate": 1.9109826589595375e-05,
+      "loss": 3.8231,
       "step": 1132
     },
     {
       "epoch": 5.08,
-      "learning_rate": 0.00019057803468208091,
-      "loss": 0.1018,
+      "learning_rate": 1.909248554913295e-05,
+      "loss": 3.7929,
       "step": 1133
     },
     {
       "epoch": 5.09,
-      "learning_rate": 0.0001904046242774566,
-      "loss": 0.0953,
+      "learning_rate": 1.9075144508670522e-05,
+      "loss": 3.8117,
       "step": 1134
     },
     {
       "epoch": 5.09,
-      "learning_rate": 0.00019023121387283236,
-      "loss": 0.1073,
+      "learning_rate": 1.9057803468208094e-05,
+      "loss": 3.772,
       "step": 1135
     },
     {
       "epoch": 5.09,
-      "learning_rate": 0.00019005780346820808,
-      "loss": 0.0956,
+      "learning_rate": 1.9040462427745663e-05,
+      "loss": 3.7204,
       "step": 1136
     },
     {
       "epoch": 5.1,
-      "learning_rate": 0.0001898843930635838,
-      "loss": 0.0952,
+      "learning_rate": 1.9023121387283238e-05,
+      "loss": 3.7641,
       "step": 1137
     },
     {
       "epoch": 5.1,
-      "learning_rate": 0.00018971098265895952,
-      "loss": 0.1017,
+      "learning_rate": 1.900578034682081e-05,
+      "loss": 3.8326,
       "step": 1138
     },
     {
       "epoch": 5.11,
-      "learning_rate": 0.00018953757225433527,
-      "loss": 0.0923,
+      "learning_rate": 1.8988439306358382e-05,
+      "loss": 3.7707,
       "step": 1139
     },
     {
       "epoch": 5.11,
-      "learning_rate": 0.00018936416184971096,
-      "loss": 0.0987,
+      "learning_rate": 1.8971098265895954e-05,
+      "loss": 3.7888,
       "step": 1140
     },
     {
       "epoch": 5.12,
-      "learning_rate": 0.00018919075144508668,
-      "loss": 0.1126,
+      "learning_rate": 1.8953757225433527e-05,
+      "loss": 3.7731,
       "step": 1141
     },
     {
       "epoch": 5.12,
-      "learning_rate": 0.0001890173410404624,
-      "loss": 0.0983,
+      "learning_rate": 1.89364161849711e-05,
+      "loss": 3.7375,
       "step": 1142
     },
     {
       "epoch": 5.13,
-      "learning_rate": 0.00018884393063583815,
-      "loss": 0.0825,
+      "learning_rate": 1.891907514450867e-05,
+      "loss": 3.687,
       "step": 1143
     },
     {
       "epoch": 5.13,
-      "learning_rate": 0.00018867052023121387,
-      "loss": 0.0934,
+      "learning_rate": 1.8901734104046243e-05,
+      "loss": 3.7132,
       "step": 1144
     },
     {
       "epoch": 5.13,
-      "learning_rate": 0.0001884971098265896,
-      "loss": 0.0904,
+      "learning_rate": 1.8884393063583818e-05,
+      "loss": 3.6417,
       "step": 1145
     },
     {
       "epoch": 5.14,
-      "learning_rate": 0.00018832369942196528,
-      "loss": 0.0959,
+      "learning_rate": 1.8867052023121387e-05,
+      "loss": 3.7163,
       "step": 1146
     },
     {
       "epoch": 5.14,
-      "learning_rate": 0.00018815028901734103,
-      "loss": 0.0749,
+      "learning_rate": 1.884971098265896e-05,
+      "loss": 3.5894,
       "step": 1147
     },
     {
       "epoch": 5.15,
-      "learning_rate": 0.00018797687861271675,
-      "loss": 0.1007,
+      "learning_rate": 1.883236994219653e-05,
+      "loss": 3.7036,
       "step": 1148
     },
     {
       "epoch": 5.15,
-      "learning_rate": 0.00018780346820809247,
-      "loss": 0.0995,
+      "learning_rate": 1.8815028901734106e-05,
+      "loss": 3.5606,
       "step": 1149
     },
     {
       "epoch": 5.16,
-      "learning_rate": 0.0001876300578034682,
-      "loss": 0.0912,
+      "learning_rate": 1.879768786127168e-05,
+      "loss": 3.6319,
       "step": 1150
     },
     {
       "epoch": 5.16,
-      "learning_rate": 0.00018745664739884394,
-      "loss": 0.0994,
+      "learning_rate": 1.8780346820809247e-05,
+      "loss": 3.6076,
       "step": 1151
     },
     {
       "epoch": 5.17,
-      "learning_rate": 0.00018728323699421963,
-      "loss": 0.0708,
+      "learning_rate": 1.876300578034682e-05,
+      "loss": 3.6795,
       "step": 1152
     },
     {
       "epoch": 5.17,
-      "learning_rate": 0.00018710982658959536,
-      "loss": 0.1036,
+      "learning_rate": 1.8745664739884395e-05,
+      "loss": 3.6955,
       "step": 1153
     },
     {
       "epoch": 5.17,
-      "learning_rate": 0.00018693641618497108,
-      "loss": 0.0811,
+      "learning_rate": 1.8728323699421967e-05,
+      "loss": 3.6129,
       "step": 1154
     },
     {
       "epoch": 5.18,
-      "learning_rate": 0.00018676300578034682,
-      "loss": 0.0847,
+      "learning_rate": 1.871098265895954e-05,
+      "loss": 3.5779,
       "step": 1155
     },
     {
       "epoch": 5.18,
-      "learning_rate": 0.00018658959537572254,
-      "loss": 0.0879,
+      "learning_rate": 1.869364161849711e-05,
+      "loss": 3.6706,
       "step": 1156
     },
     {
       "epoch": 5.19,
-      "learning_rate": 0.00018641618497109824,
-      "loss": 0.086,
+      "learning_rate": 1.8676300578034683e-05,
+      "loss": 3.5448,
       "step": 1157
     },
     {
       "epoch": 5.19,
-      "learning_rate": 0.00018624277456647396,
-      "loss": 0.0875,
+      "learning_rate": 1.8658959537572255e-05,
+      "loss": 3.684,
       "step": 1158
     },
     {
       "epoch": 5.2,
-      "learning_rate": 0.0001860693641618497,
-      "loss": 0.0579,
+      "learning_rate": 1.8641618497109827e-05,
+      "loss": 3.5315,
       "step": 1159
     },
     {
       "epoch": 5.2,
-      "learning_rate": 0.00018589595375722543,
-      "loss": 0.0755,
+      "learning_rate": 1.86242774566474e-05,
+      "loss": 3.7197,
       "step": 1160
     },
     {
       "epoch": 5.21,
-      "learning_rate": 0.00018572254335260115,
-      "loss": 0.0835,
+      "learning_rate": 1.8606936416184975e-05,
+      "loss": 3.6238,
       "step": 1161
     },
     {
       "epoch": 5.21,
-      "learning_rate": 0.00018554913294797684,
-      "loss": 0.0806,
+      "learning_rate": 1.8589595375722543e-05,
+      "loss": 3.544,
       "step": 1162
     },
     {
       "epoch": 5.22,
-      "learning_rate": 0.0001853757225433526,
-      "loss": 0.0846,
+      "learning_rate": 1.8572254335260115e-05,
+      "loss": 3.6976,
       "step": 1163
     },
     {
       "epoch": 5.22,
-      "learning_rate": 0.0001852023121387283,
-      "loss": 0.0802,
+      "learning_rate": 1.8554913294797688e-05,
+      "loss": 3.6154,
       "step": 1164
     },
     {
       "epoch": 5.22,
-      "learning_rate": 0.00018502890173410403,
-      "loss": 0.1096,
+      "learning_rate": 1.8537572254335263e-05,
+      "loss": 3.4891,
       "step": 1165
     },
     {
       "epoch": 5.23,
-      "learning_rate": 0.00018485549132947975,
-      "loss": 0.2806,
+      "learning_rate": 1.8520231213872835e-05,
+      "loss": 4.4885,
       "step": 1166
     },
     {
       "epoch": 5.23,
-      "learning_rate": 0.0001846820809248555,
-      "loss": 0.2619,
+      "learning_rate": 1.8502890173410404e-05,
+      "loss": 4.5534,
       "step": 1167
     },
     {
       "epoch": 5.24,
-      "learning_rate": 0.0001845086705202312,
-      "loss": 0.2016,
+      "learning_rate": 1.8485549132947976e-05,
+      "loss": 4.6256,
       "step": 1168
     },
     {
       "epoch": 5.24,
-      "learning_rate": 0.0001843352601156069,
-      "loss": 0.1805,
+      "learning_rate": 1.846820809248555e-05,
+      "loss": 4.6323,
       "step": 1169
     },
     {
       "epoch": 5.25,
-      "learning_rate": 0.00018416184971098263,
-      "loss": 0.152,
+      "learning_rate": 1.8450867052023123e-05,
+      "loss": 4.4773,
       "step": 1170
     },
     {
       "epoch": 5.25,
-      "learning_rate": 0.00018398843930635838,
-      "loss": 0.1561,
+      "learning_rate": 1.8433526011560695e-05,
+      "loss": 4.5565,
       "step": 1171
     },
     {
       "epoch": 5.26,
-      "learning_rate": 0.0001838150289017341,
-      "loss": 0.135,
+      "learning_rate": 1.8416184971098264e-05,
+      "loss": 4.4722,
       "step": 1172
     },
     {
       "epoch": 5.26,
-      "learning_rate": 0.00018364161849710982,
-      "loss": 0.1462,
+      "learning_rate": 1.839884393063584e-05,
+      "loss": 4.4305,
       "step": 1173
     },
     {
       "epoch": 5.26,
-      "learning_rate": 0.00018346820809248552,
-      "loss": 0.121,
+      "learning_rate": 1.838150289017341e-05,
+      "loss": 4.2747,
       "step": 1174
     },
     {
       "epoch": 5.27,
-      "learning_rate": 0.00018329479768786124,
-      "loss": 0.15,
+      "learning_rate": 1.8364161849710984e-05,
+      "loss": 4.2715,
       "step": 1175
     },
     {
       "epoch": 5.27,
-      "learning_rate": 0.00018312138728323698,
-      "loss": 0.1087,
+      "learning_rate": 1.8346820809248556e-05,
+      "loss": 4.0829,
       "step": 1176
     },
     {
       "epoch": 5.28,
-      "learning_rate": 0.0001829479768786127,
-      "loss": 0.1005,
+      "learning_rate": 1.8329479768786124e-05,
+      "loss": 4.0985,
       "step": 1177
     },
     {
       "epoch": 5.28,
-      "learning_rate": 0.00018277456647398843,
-      "loss": 0.1302,
+      "learning_rate": 1.83121387283237e-05,
+      "loss": 4.1009,
       "step": 1178
     },
     {
       "epoch": 5.29,
-      "learning_rate": 0.00018260115606936412,
-      "loss": 0.1133,
+      "learning_rate": 1.8294797687861272e-05,
+      "loss": 3.9659,
       "step": 1179
     },
     {
       "epoch": 5.29,
-      "learning_rate": 0.00018242774566473987,
-      "loss": 0.0955,
+      "learning_rate": 1.8277456647398844e-05,
+      "loss": 3.8449,
       "step": 1180
     },
     {
       "epoch": 5.3,
-      "learning_rate": 0.0001822543352601156,
-      "loss": 0.0942,
+      "learning_rate": 1.8260115606936416e-05,
+      "loss": 3.8618,
       "step": 1181
     },
     {
       "epoch": 5.3,
-      "learning_rate": 0.0001820809248554913,
-      "loss": 0.1154,
+      "learning_rate": 1.8242774566473988e-05,
+      "loss": 3.9116,
       "step": 1182
     },
     {
       "epoch": 5.3,
-      "learning_rate": 0.00018190751445086703,
-      "loss": 0.1216,
+      "learning_rate": 1.822543352601156e-05,
+      "loss": 3.8054,
       "step": 1183
     },
     {
       "epoch": 5.31,
-      "learning_rate": 0.00018173410404624278,
-      "loss": 0.0994,
+      "learning_rate": 1.8208092485549132e-05,
+      "loss": 3.7647,
       "step": 1184
     },
     {
       "epoch": 5.31,
-      "learning_rate": 0.00018156069364161847,
-      "loss": 0.0857,
+      "learning_rate": 1.8190751445086704e-05,
+      "loss": 3.8045,
       "step": 1185
     },
     {
       "epoch": 5.32,
-      "learning_rate": 0.0001813872832369942,
-      "loss": 0.1136,
+      "learning_rate": 1.817341040462428e-05,
+      "loss": 3.8661,
       "step": 1186
     },
     {
       "epoch": 5.32,
-      "learning_rate": 0.0001812138728323699,
-      "loss": 0.1012,
+      "learning_rate": 1.8156069364161852e-05,
+      "loss": 3.8399,
       "step": 1187
     },
     {
       "epoch": 5.33,
-      "learning_rate": 0.00018104046242774566,
-      "loss": 0.095,
+      "learning_rate": 1.813872832369942e-05,
+      "loss": 3.7362,
       "step": 1188
     },
     {
       "epoch": 5.33,
-      "learning_rate": 0.00018086705202312138,
-      "loss": 0.0799,
+      "learning_rate": 1.8121387283236993e-05,
+      "loss": 3.7025,
       "step": 1189
     },
     {
       "epoch": 5.34,
-      "learning_rate": 0.00018069364161849707,
-      "loss": 0.0927,
+      "learning_rate": 1.8104046242774568e-05,
+      "loss": 3.7664,
       "step": 1190
     },
     {
       "epoch": 5.34,
-      "learning_rate": 0.0001805202312138728,
-      "loss": 0.0887,
+      "learning_rate": 1.808670520231214e-05,
+      "loss": 3.7872,
       "step": 1191
     },
     {
       "epoch": 5.35,
-      "learning_rate": 0.00018034682080924854,
-      "loss": 0.0991,
+      "learning_rate": 1.8069364161849712e-05,
+      "loss": 3.7064,
       "step": 1192
     },
     {
       "epoch": 5.35,
-      "learning_rate": 0.00018017341040462426,
-      "loss": 0.1128,
+      "learning_rate": 1.805202312138728e-05,
+      "loss": 3.7757,
       "step": 1193
     },
     {
       "epoch": 5.35,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 0.0827,
+      "learning_rate": 1.8034682080924856e-05,
+      "loss": 3.6683,
       "step": 1194
     },
     {
       "epoch": 5.36,
-      "learning_rate": 0.0001798265895953757,
-      "loss": 0.0856,
+      "learning_rate": 1.801734104046243e-05,
+      "loss": 3.8014,
       "step": 1195
     },
     {
       "epoch": 5.36,
-      "learning_rate": 0.00017965317919075145,
-      "loss": 0.085,
+      "learning_rate": 1.8e-05,
+      "loss": 3.7297,
       "step": 1196
     },
     {
       "epoch": 5.37,
-      "learning_rate": 0.00017947976878612715,
-      "loss": 0.1049,
+      "learning_rate": 1.7982658959537573e-05,
+      "loss": 3.6839,
       "step": 1197
     },
     {
       "epoch": 5.37,
-      "learning_rate": 0.00017930635838150287,
-      "loss": 0.0995,
+      "learning_rate": 1.7965317919075145e-05,
+      "loss": 3.6291,
       "step": 1198
     },
     {
       "epoch": 5.38,
-      "learning_rate": 0.0001791329479768786,
-      "loss": 0.0757,
+      "learning_rate": 1.7947976878612717e-05,
+      "loss": 3.6654,
       "step": 1199
     },
     {
       "epoch": 5.38,
-      "learning_rate": 0.00017895953757225434,
-      "loss": 0.0879,
+      "learning_rate": 1.793063583815029e-05,
+      "loss": 3.723,
       "step": 1200
     },
     {
       "epoch": 5.39,
-      "learning_rate": 0.00017878612716763006,
-      "loss": 0.0971,
+      "learning_rate": 1.791329479768786e-05,
+      "loss": 3.6468,
       "step": 1201
     },
     {
       "epoch": 5.39,
-      "learning_rate": 0.00017861271676300575,
-      "loss": 0.0995,
+      "learning_rate": 1.7895953757225436e-05,
+      "loss": 3.6865,
       "step": 1202
     },
     {
       "epoch": 5.39,
-      "learning_rate": 0.00017843930635838147,
-      "loss": 0.0709,
+      "learning_rate": 1.7878612716763005e-05,
+      "loss": 3.6262,
       "step": 1203
     },
     {
       "epoch": 5.4,
-      "learning_rate": 0.00017826589595375722,
-      "loss": 0.0798,
+      "learning_rate": 1.7861271676300577e-05,
+      "loss": 3.7134,
       "step": 1204
     },
     {
       "epoch": 5.4,
-      "learning_rate": 0.00017809248554913294,
-      "loss": 0.0719,
+      "learning_rate": 1.784393063583815e-05,
+      "loss": 3.5908,
       "step": 1205
     },
     {
       "epoch": 5.41,
-      "learning_rate": 0.00017791907514450866,
-      "loss": 0.0798,
+      "learning_rate": 1.7826589595375725e-05,
+      "loss": 3.7011,
       "step": 1206
     },
     {
       "epoch": 5.41,
-      "learning_rate": 0.00017774566473988435,
-      "loss": 0.0892,
+      "learning_rate": 1.7809248554913297e-05,
+      "loss": 3.6593,
       "step": 1207
     },
     {
       "epoch": 5.42,
-      "learning_rate": 0.0001775722543352601,
-      "loss": 0.0759,
+      "learning_rate": 1.7791907514450865e-05,
+      "loss": 3.665,
       "step": 1208
     },
     {
       "epoch": 5.42,
-      "learning_rate": 0.00017739884393063582,
-      "loss": 0.0816,
+      "learning_rate": 1.7774566473988437e-05,
+      "loss": 3.7782,
       "step": 1209
     },
     {
       "epoch": 5.43,
-      "learning_rate": 0.00017722543352601154,
-      "loss": 0.087,
+      "learning_rate": 1.7757225433526013e-05,
+      "loss": 3.6439,
       "step": 1210
     },
     {
       "epoch": 5.43,
-      "learning_rate": 0.00017705202312138726,
-      "loss": 0.0928,
+      "learning_rate": 1.7739884393063585e-05,
+      "loss": 3.8337,
       "step": 1211
     },
     {
       "epoch": 5.43,
-      "learning_rate": 0.000176878612716763,
-      "loss": 0.0925,
+      "learning_rate": 1.7722543352601157e-05,
+      "loss": 3.578,
       "step": 1212
     },
     {
       "epoch": 5.44,
-      "learning_rate": 0.0001767052023121387,
-      "loss": 0.061,
+      "learning_rate": 1.770520231213873e-05,
+      "loss": 3.6506,
       "step": 1213
     },
     {
       "epoch": 5.44,
-      "learning_rate": 0.00017653179190751442,
-      "loss": 0.1159,
+      "learning_rate": 1.76878612716763e-05,
+      "loss": 3.7176,
       "step": 1214
     },
     {
       "epoch": 5.45,
-      "learning_rate": 0.00017635838150289015,
-      "loss": 0.159,
+      "learning_rate": 1.7670520231213873e-05,
+      "loss": 3.6124,
       "step": 1215
     },
     {
       "epoch": 5.45,
-      "learning_rate": 0.0001761849710982659,
-      "loss": 0.3672,
+      "learning_rate": 1.7653179190751445e-05,
+      "loss": 4.501,
       "step": 1216
     },
     {
       "epoch": 5.46,
-      "learning_rate": 0.00017601156069364161,
-      "loss": 0.2528,
+      "learning_rate": 1.7635838150289017e-05,
+      "loss": 4.3737,
       "step": 1217
     },
     {
       "epoch": 5.46,
-      "learning_rate": 0.0001758381502890173,
-      "loss": 0.2057,
+      "learning_rate": 1.7618497109826593e-05,
+      "loss": 4.4793,
       "step": 1218
     },
     {
       "epoch": 5.47,
-      "learning_rate": 0.00017566473988439303,
-      "loss": 0.1701,
+      "learning_rate": 1.760115606936416e-05,
+      "loss": 4.4172,
       "step": 1219
     },
     {
       "epoch": 5.47,
-      "learning_rate": 0.00017549132947976878,
-      "loss": 0.1412,
+      "learning_rate": 1.7583815028901733e-05,
+      "loss": 4.3874,
       "step": 1220
     },
     {
       "epoch": 5.48,
-      "learning_rate": 0.0001753179190751445,
-      "loss": 0.1459,
+      "learning_rate": 1.7566473988439306e-05,
+      "loss": 4.406,
       "step": 1221
     },
     {
       "epoch": 5.48,
-      "learning_rate": 0.00017514450867052022,
-      "loss": 0.146,
+      "learning_rate": 1.754913294797688e-05,
+      "loss": 4.3033,
       "step": 1222
     },
     {
       "epoch": 5.48,
-      "learning_rate": 0.00017497109826589594,
-      "loss": 0.1425,
+      "learning_rate": 1.7531791907514453e-05,
+      "loss": 4.287,
       "step": 1223
     },
     {
       "epoch": 5.49,
-      "learning_rate": 0.00017479768786127169,
-      "loss": 0.1307,
+      "learning_rate": 1.7514450867052022e-05,
+      "loss": 4.2098,
       "step": 1224
     },
     {
       "epoch": 5.49,
-      "learning_rate": 0.00017462427745664738,
-      "loss": 0.1416,
+      "learning_rate": 1.7497109826589594e-05,
+      "loss": 4.2035,
       "step": 1225
     },
     {
       "epoch": 5.5,
-      "learning_rate": 0.0001744508670520231,
-      "loss": 0.1239,
+      "learning_rate": 1.747976878612717e-05,
+      "loss": 4.0716,
       "step": 1226
     },
     {
       "epoch": 5.5,
-      "learning_rate": 0.00017427745664739882,
-      "loss": 0.135,
+      "learning_rate": 1.746242774566474e-05,
+      "loss": 4.1066,
       "step": 1227
     },
     {
       "epoch": 5.51,
-      "learning_rate": 0.00017410404624277457,
-      "loss": 0.0945,
+      "learning_rate": 1.7445086705202313e-05,
+      "loss": 3.8452,
       "step": 1228
     },
     {
       "epoch": 5.51,
-      "learning_rate": 0.0001739306358381503,
-      "loss": 0.1212,
+      "learning_rate": 1.7427745664739882e-05,
+      "loss": 3.93,
       "step": 1229
     },
     {
       "epoch": 5.52,
-      "learning_rate": 0.00017375722543352598,
-      "loss": 0.1213,
+      "learning_rate": 1.7410404624277458e-05,
+      "loss": 3.8458,
       "step": 1230
     },
     {
       "epoch": 5.52,
-      "learning_rate": 0.0001735838150289017,
-      "loss": 0.11,
+      "learning_rate": 1.739306358381503e-05,
+      "loss": 3.9127,
       "step": 1231
     },
     {
       "epoch": 5.52,
-      "learning_rate": 0.00017341040462427745,
-      "loss": 0.1107,
+      "learning_rate": 1.73757225433526e-05,
+      "loss": 3.8718,
       "step": 1232
     },
     {
       "epoch": 5.53,
-      "learning_rate": 0.00017323699421965317,
-      "loss": 0.1042,
+      "learning_rate": 1.7358381502890174e-05,
+      "loss": 3.8352,
       "step": 1233
     },
     {
       "epoch": 5.53,
-      "learning_rate": 0.0001730635838150289,
-      "loss": 0.1009,
+      "learning_rate": 1.7341040462427746e-05,
+      "loss": 3.8066,
       "step": 1234
     },
     {
       "epoch": 5.54,
-      "learning_rate": 0.00017289017341040459,
-      "loss": 0.0995,
+      "learning_rate": 1.7323699421965318e-05,
+      "loss": 3.8209,
       "step": 1235
     },
     {
       "epoch": 5.54,
-      "learning_rate": 0.00017271676300578033,
-      "loss": 0.1199,
+      "learning_rate": 1.730635838150289e-05,
+      "loss": 3.7231,
       "step": 1236
     },
     {
       "epoch": 5.55,
-      "learning_rate": 0.00017254335260115605,
-      "loss": 0.1462,
+      "learning_rate": 1.7289017341040462e-05,
+      "loss": 3.8296,
       "step": 1237
     },
     {
       "epoch": 5.55,
-      "learning_rate": 0.00017236994219653178,
-      "loss": 0.0984,
+      "learning_rate": 1.7271676300578037e-05,
+      "loss": 3.7024,
       "step": 1238
     },
     {
       "epoch": 5.56,
-      "learning_rate": 0.0001721965317919075,
-      "loss": 0.097,
+      "learning_rate": 1.725433526011561e-05,
+      "loss": 3.7866,
       "step": 1239
     },
     {
       "epoch": 5.56,
-      "learning_rate": 0.00017202312138728324,
-      "loss": 0.1193,
+      "learning_rate": 1.7236994219653178e-05,
+      "loss": 3.8055,
       "step": 1240
     },
     {
       "epoch": 5.57,
-      "learning_rate": 0.00017184971098265894,
-      "loss": 0.1235,
+      "learning_rate": 1.721965317919075e-05,
+      "loss": 3.8048,
       "step": 1241
     },
     {
       "epoch": 5.57,
-      "learning_rate": 0.00017167630057803466,
-      "loss": 0.1089,
+      "learning_rate": 1.7202312138728326e-05,
+      "loss": 3.7299,
       "step": 1242
     },
     {
       "epoch": 5.57,
-      "learning_rate": 0.00017150289017341038,
-      "loss": 0.1091,
+      "learning_rate": 1.7184971098265898e-05,
+      "loss": 3.6614,
       "step": 1243
     },
     {
       "epoch": 5.58,
-      "learning_rate": 0.00017132947976878613,
-      "loss": 0.0913,
+      "learning_rate": 1.716763005780347e-05,
+      "loss": 3.7104,
       "step": 1244
     },
     {
       "epoch": 5.58,
-      "learning_rate": 0.00017115606936416185,
-      "loss": 0.1071,
+      "learning_rate": 1.715028901734104e-05,
+      "loss": 3.7007,
       "step": 1245
     },
     {
       "epoch": 5.59,
-      "learning_rate": 0.00017098265895953757,
-      "loss": 0.0796,
+      "learning_rate": 1.7132947976878614e-05,
+      "loss": 3.7255,
       "step": 1246
     },
     {
       "epoch": 5.59,
-      "learning_rate": 0.00017080924855491326,
-      "loss": 0.0782,
+      "learning_rate": 1.7115606936416186e-05,
+      "loss": 3.611,
       "step": 1247
     },
     {
       "epoch": 5.6,
-      "learning_rate": 0.000170635838150289,
-      "loss": 0.0974,
+      "learning_rate": 1.7098265895953758e-05,
+      "loss": 3.6779,
       "step": 1248
     },
     {
       "epoch": 5.6,
-      "learning_rate": 0.00017046242774566473,
-      "loss": 0.0765,
+      "learning_rate": 1.708092485549133e-05,
+      "loss": 3.6963,
       "step": 1249
     },
     {
       "epoch": 5.61,
-      "learning_rate": 0.00017028901734104045,
-      "loss": 0.0882,
+      "learning_rate": 1.7063583815028902e-05,
+      "loss": 3.6617,
       "step": 1250
     },
     {
       "epoch": 5.61,
-      "learning_rate": 0.00017011560693641617,
-      "loss": 0.0782,
+      "learning_rate": 1.7046242774566474e-05,
+      "loss": 3.5974,
       "step": 1251
     },
     {
       "epoch": 5.61,
-      "learning_rate": 0.00016994219653179192,
-      "loss": 0.085,
+      "learning_rate": 1.7028901734104046e-05,
+      "loss": 3.6514,
       "step": 1252
     },
     {
       "epoch": 5.62,
-      "learning_rate": 0.0001697687861271676,
-      "loss": 0.0744,
+      "learning_rate": 1.701156069364162e-05,
+      "loss": 3.6334,
       "step": 1253
     },
     {
       "epoch": 5.62,
-      "learning_rate": 0.00016959537572254333,
-      "loss": 0.0714,
+      "learning_rate": 1.6994219653179194e-05,
+      "loss": 3.5855,
       "step": 1254
     },
     {
       "epoch": 5.63,
-      "learning_rate": 0.00016942196531791905,
-      "loss": 0.0929,
+      "learning_rate": 1.6976878612716763e-05,
+      "loss": 3.5942,
       "step": 1255
     },
     {
       "epoch": 5.63,
-      "learning_rate": 0.0001692485549132948,
-      "loss": 0.0866,
+      "learning_rate": 1.6959537572254335e-05,
+      "loss": 3.6141,
       "step": 1256
     },
     {
       "epoch": 5.64,
-      "learning_rate": 0.00016907514450867052,
-      "loss": 0.1397,
+      "learning_rate": 1.6942196531791907e-05,
+      "loss": 3.6875,
       "step": 1257
     },
     {
       "epoch": 5.64,
-      "learning_rate": 0.00016890173410404622,
-      "loss": 0.069,
+      "learning_rate": 1.6924855491329482e-05,
+      "loss": 3.732,
       "step": 1258
     },
     {
       "epoch": 5.65,
-      "learning_rate": 0.00016872832369942194,
-      "loss": 0.0713,
+      "learning_rate": 1.6907514450867054e-05,
+      "loss": 3.5305,
       "step": 1259
     },
     {
       "epoch": 5.65,
-      "learning_rate": 0.00016855491329479768,
-      "loss": 0.0674,
+      "learning_rate": 1.6890173410404623e-05,
+      "loss": 3.5889,
       "step": 1260
     },
     {
       "epoch": 5.65,
-      "learning_rate": 0.0001683815028901734,
-      "loss": 0.0674,
+      "learning_rate": 1.6872832369942195e-05,
+      "loss": 3.6374,
       "step": 1261
     },
     {
       "epoch": 5.66,
-      "learning_rate": 0.00016820809248554913,
-      "loss": 0.0649,
+      "learning_rate": 1.685549132947977e-05,
+      "loss": 3.6227,
       "step": 1262
     },
     {
       "epoch": 5.66,
-      "learning_rate": 0.00016803468208092482,
-      "loss": 0.0891,
+      "learning_rate": 1.6838150289017343e-05,
+      "loss": 3.584,
       "step": 1263
     },
     {
       "epoch": 5.67,
-      "learning_rate": 0.00016786127167630057,
-      "loss": 0.079,
+      "learning_rate": 1.6820809248554915e-05,
+      "loss": 3.7017,
       "step": 1264
     },
     {
       "epoch": 5.67,
-      "learning_rate": 0.0001676878612716763,
-      "loss": 0.1048,
+      "learning_rate": 1.6803468208092487e-05,
+      "loss": 3.5791,
       "step": 1265
     },
     {
       "epoch": 5.68,
-      "learning_rate": 0.000167514450867052,
-      "loss": 0.2792,
+      "learning_rate": 1.678612716763006e-05,
+      "loss": 4.4901,
       "step": 1266
     },
     {
       "epoch": 5.68,
-      "learning_rate": 0.00016734104046242773,
-      "loss": 0.2596,
+      "learning_rate": 1.676878612716763e-05,
+      "loss": 4.4509,
       "step": 1267
     },
     {
       "epoch": 5.69,
-      "learning_rate": 0.00016716763005780348,
-      "loss": 0.2365,
+      "learning_rate": 1.6751445086705203e-05,
+      "loss": 4.4861,
       "step": 1268
     },
     {
       "epoch": 5.69,
-      "learning_rate": 0.0001669942196531792,
-      "loss": 0.1353,
+      "learning_rate": 1.6734104046242775e-05,
+      "loss": 4.4029,
       "step": 1269
     },
     {
       "epoch": 5.7,
-      "learning_rate": 0.0001668208092485549,
-      "loss": 0.1578,
+      "learning_rate": 1.671676300578035e-05,
+      "loss": 4.4611,
       "step": 1270
     },
     {
       "epoch": 5.7,
-      "learning_rate": 0.0001666473988439306,
-      "loss": 0.14,
+      "learning_rate": 1.669942196531792e-05,
+      "loss": 4.3711,
       "step": 1271
     },
     {
       "epoch": 5.7,
-      "learning_rate": 0.00016647398843930633,
-      "loss": 0.1199,
+      "learning_rate": 1.668208092485549e-05,
+      "loss": 4.2994,
       "step": 1272
     },
     {
       "epoch": 5.71,
-      "learning_rate": 0.00016630057803468208,
-      "loss": 0.1419,
+      "learning_rate": 1.6664739884393063e-05,
+      "loss": 4.3213,
       "step": 1273
     },
     {
       "epoch": 5.71,
-      "learning_rate": 0.0001661271676300578,
-      "loss": 0.1321,
+      "learning_rate": 1.6647398843930635e-05,
+      "loss": 4.2141,
       "step": 1274
     },
     {
       "epoch": 5.72,
-      "learning_rate": 0.0001659537572254335,
-      "loss": 0.1182,
+      "learning_rate": 1.663005780346821e-05,
+      "loss": 4.189,
       "step": 1275
     },
     {
       "epoch": 5.72,
-      "learning_rate": 0.00016578034682080922,
-      "loss": 0.1181,
+      "learning_rate": 1.661271676300578e-05,
+      "loss": 4.0621,
       "step": 1276
     },
     {
       "epoch": 5.73,
-      "learning_rate": 0.00016560693641618496,
-      "loss": 0.1181,
+      "learning_rate": 1.659537572254335e-05,
+      "loss": 3.9399,
       "step": 1277
     },
     {
       "epoch": 5.73,
-      "learning_rate": 0.00016543352601156068,
-      "loss": 0.1041,
+      "learning_rate": 1.6578034682080924e-05,
+      "loss": 4.0211,
       "step": 1278
     },
     {
       "epoch": 5.74,
-      "learning_rate": 0.0001652601156069364,
-      "loss": 0.1226,
+      "learning_rate": 1.65606936416185e-05,
+      "loss": 3.9832,
       "step": 1279
     },
     {
       "epoch": 5.74,
-      "learning_rate": 0.0001650867052023121,
-      "loss": 0.1154,
+      "learning_rate": 1.654335260115607e-05,
+      "loss": 3.8816,
       "step": 1280
     },
     {
       "epoch": 5.74,
-      "learning_rate": 0.00016491329479768785,
-      "loss": 0.1378,
+      "learning_rate": 1.652601156069364e-05,
+      "loss": 3.8468,
       "step": 1281
     },
     {
       "epoch": 5.75,
-      "learning_rate": 0.00016473988439306357,
-      "loss": 0.1154,
+      "learning_rate": 1.6508670520231212e-05,
+      "loss": 3.9062,
       "step": 1282
     },
     {
       "epoch": 5.75,
-      "learning_rate": 0.0001645664739884393,
-      "loss": 0.1216,
+      "learning_rate": 1.6491329479768787e-05,
+      "loss": 3.9067,
       "step": 1283
     },
     {
       "epoch": 5.76,
-      "learning_rate": 0.000164393063583815,
-      "loss": 0.1142,
+      "learning_rate": 1.647398843930636e-05,
+      "loss": 3.8635,
       "step": 1284
     },
     {
       "epoch": 5.76,
-      "learning_rate": 0.00016421965317919076,
-      "loss": 0.1157,
+      "learning_rate": 1.645664739884393e-05,
+      "loss": 3.8684,
       "step": 1285
     },
     {
       "epoch": 5.77,
-      "learning_rate": 0.00016404624277456645,
-      "loss": 0.0988,
+      "learning_rate": 1.64393063583815e-05,
+      "loss": 3.7676,
       "step": 1286
     },
     {
       "epoch": 5.77,
-      "learning_rate": 0.00016387283236994217,
-      "loss": 0.0949,
+      "learning_rate": 1.6421965317919076e-05,
+      "loss": 3.7876,
       "step": 1287
     },
     {
       "epoch": 5.78,
-      "learning_rate": 0.0001636994219653179,
-      "loss": 0.0961,
+      "learning_rate": 1.6404624277456648e-05,
+      "loss": 3.7201,
       "step": 1288
     },
     {
       "epoch": 5.78,
-      "learning_rate": 0.00016352601156069364,
-      "loss": 0.0861,
+      "learning_rate": 1.638728323699422e-05,
+      "loss": 3.7107,
       "step": 1289
     },
     {
       "epoch": 5.78,
-      "learning_rate": 0.00016335260115606936,
-      "loss": 0.0907,
+      "learning_rate": 1.6369942196531792e-05,
+      "loss": 3.7353,
       "step": 1290
     },
     {
       "epoch": 5.79,
-      "learning_rate": 0.00016317919075144505,
-      "loss": 0.0893,
+      "learning_rate": 1.6352601156069367e-05,
+      "loss": 3.8101,
       "step": 1291
     },
     {
       "epoch": 5.79,
-      "learning_rate": 0.00016300578034682077,
-      "loss": 0.0836,
+      "learning_rate": 1.6335260115606936e-05,
+      "loss": 3.7945,
       "step": 1292
     },
     {
       "epoch": 5.8,
-      "learning_rate": 0.00016283236994219652,
-      "loss": 0.0774,
+      "learning_rate": 1.6317919075144508e-05,
+      "loss": 3.7453,
       "step": 1293
     },
     {
       "epoch": 5.8,
-      "learning_rate": 0.00016265895953757224,
-      "loss": 0.0944,
+      "learning_rate": 1.630057803468208e-05,
+      "loss": 3.7175,
       "step": 1294
     },
     {
       "epoch": 5.81,
-      "learning_rate": 0.00016248554913294796,
-      "loss": 0.0802,
+      "learning_rate": 1.6283236994219656e-05,
+      "loss": 3.7157,
       "step": 1295
     },
     {
       "epoch": 5.81,
-      "learning_rate": 0.00016231213872832368,
-      "loss": 0.0914,
+      "learning_rate": 1.6265895953757228e-05,
+      "loss": 3.6373,
       "step": 1296
     },
     {
       "epoch": 5.82,
-      "learning_rate": 0.00016213872832369943,
-      "loss": 0.0849,
+      "learning_rate": 1.6248554913294796e-05,
+      "loss": 3.6859,
       "step": 1297
     },
     {
       "epoch": 5.82,
-      "learning_rate": 0.00016196531791907512,
-      "loss": 0.0907,
+      "learning_rate": 1.623121387283237e-05,
+      "loss": 3.7236,
       "step": 1298
     },
     {
       "epoch": 5.83,
-      "learning_rate": 0.00016179190751445085,
-      "loss": 0.0839,
+      "learning_rate": 1.6213872832369944e-05,
+      "loss": 3.7574,
       "step": 1299
     },
     {
       "epoch": 5.83,
-      "learning_rate": 0.00016161849710982657,
-      "loss": 0.0997,
+      "learning_rate": 1.6196531791907516e-05,
+      "loss": 3.7723,
       "step": 1300
     },
     {
       "epoch": 5.83,
-      "learning_rate": 0.00016144508670520231,
-      "loss": 0.0792,
+      "learning_rate": 1.6179190751445088e-05,
+      "loss": 3.6934,
       "step": 1301
     },
     {
       "epoch": 5.84,
-      "learning_rate": 0.00016127167630057803,
-      "loss": 0.0732,
+      "learning_rate": 1.6161849710982657e-05,
+      "loss": 3.6567,
       "step": 1302
     },
     {
       "epoch": 5.84,
-      "learning_rate": 0.00016109826589595373,
-      "loss": 0.084,
+      "learning_rate": 1.6144508670520232e-05,
+      "loss": 3.6286,
       "step": 1303
     },
     {
       "epoch": 5.85,
-      "learning_rate": 0.00016092485549132945,
-      "loss": 0.0786,
+      "learning_rate": 1.6127167630057804e-05,
+      "loss": 3.7219,
       "step": 1304
     },
     {
       "epoch": 5.85,
-      "learning_rate": 0.0001607514450867052,
-      "loss": 0.0919,
+      "learning_rate": 1.6109826589595376e-05,
+      "loss": 3.74,
       "step": 1305
     },
     {
       "epoch": 5.86,
-      "learning_rate": 0.00016057803468208092,
-      "loss": 0.0896,
+      "learning_rate": 1.6092485549132948e-05,
+      "loss": 3.6348,
       "step": 1306
     },
     {
       "epoch": 5.86,
-      "learning_rate": 0.00016040462427745664,
-      "loss": 0.1043,
+      "learning_rate": 1.607514450867052e-05,
+      "loss": 3.7186,
       "step": 1307
     },
     {
       "epoch": 5.87,
-      "learning_rate": 0.00016023121387283233,
-      "loss": 0.0694,
+      "learning_rate": 1.6057803468208092e-05,
+      "loss": 3.6526,
       "step": 1308
     },
     {
       "epoch": 5.87,
-      "learning_rate": 0.00016005780346820808,
-      "loss": 0.0945,
+      "learning_rate": 1.6040462427745664e-05,
+      "loss": 3.7628,
       "step": 1309
     },
     {
       "epoch": 5.87,
-      "learning_rate": 0.0001598843930635838,
-      "loss": 0.0835,
+      "learning_rate": 1.6023121387283237e-05,
+      "loss": 3.7244,
       "step": 1310
     },
     {
       "epoch": 5.88,
-      "learning_rate": 0.00015971098265895952,
-      "loss": 0.0805,
+      "learning_rate": 1.6005780346820812e-05,
+      "loss": 3.6791,
       "step": 1311
     },
     {
       "epoch": 5.88,
-      "learning_rate": 0.00015953757225433524,
-      "loss": 0.074,
+      "learning_rate": 1.598843930635838e-05,
+      "loss": 3.679,
       "step": 1312
     },
     {
       "epoch": 5.89,
-      "learning_rate": 0.000159364161849711,
-      "loss": 0.0883,
+      "learning_rate": 1.5971098265895953e-05,
+      "loss": 3.6038,
       "step": 1313
     },
     {
       "epoch": 5.89,
-      "learning_rate": 0.00015919075144508668,
-      "loss": 0.0976,
+      "learning_rate": 1.5953757225433525e-05,
+      "loss": 3.6377,
       "step": 1314
     },
     {
       "epoch": 5.9,
-      "learning_rate": 0.0001590173410404624,
-      "loss": 0.1321,
+      "learning_rate": 1.59364161849711e-05,
+      "loss": 3.6217,
       "step": 1315
     },
     {
       "epoch": 5.9,
-      "learning_rate": 0.00015884393063583812,
-      "loss": 0.28,
+      "learning_rate": 1.5919075144508672e-05,
+      "loss": 4.394,
       "step": 1316
     },
     {
       "epoch": 5.91,
-      "learning_rate": 0.00015867052023121387,
-      "loss": 0.1622,
+      "learning_rate": 1.5901734104046244e-05,
+      "loss": 4.2921,
       "step": 1317
     },
     {
       "epoch": 5.91,
-      "learning_rate": 0.0001584971098265896,
-      "loss": 0.1625,
+      "learning_rate": 1.5884393063583813e-05,
+      "loss": 4.3858,
       "step": 1318
     },
     {
       "epoch": 5.91,
-      "learning_rate": 0.0001583236994219653,
-      "loss": 0.1282,
+      "learning_rate": 1.586705202312139e-05,
+      "loss": 4.3503,
       "step": 1319
     },
     {
       "epoch": 5.92,
-      "learning_rate": 0.000158150289017341,
-      "loss": 0.1433,
+      "learning_rate": 1.584971098265896e-05,
+      "loss": 4.2609,
       "step": 1320
     },
     {
       "epoch": 5.92,
-      "learning_rate": 0.00015797687861271675,
-      "loss": 0.118,
+      "learning_rate": 1.5832369942196533e-05,
+      "loss": 4.1411,
       "step": 1321
     },
     {
       "epoch": 5.93,
-      "learning_rate": 0.00015780346820809248,
-      "loss": 0.1088,
+      "learning_rate": 1.5815028901734105e-05,
+      "loss": 4.1702,
       "step": 1322
     },
     {
       "epoch": 5.93,
-      "learning_rate": 0.0001576300578034682,
-      "loss": 0.1085,
+      "learning_rate": 1.5797687861271677e-05,
+      "loss": 3.9806,
       "step": 1323
     },
     {
       "epoch": 5.94,
-      "learning_rate": 0.00015745664739884392,
-      "loss": 0.1076,
+      "learning_rate": 1.578034682080925e-05,
+      "loss": 4.0033,
       "step": 1324
     },
     {
       "epoch": 5.94,
-      "learning_rate": 0.00015728323699421966,
-      "loss": 0.1046,
+      "learning_rate": 1.576300578034682e-05,
+      "loss": 3.9419,
       "step": 1325
     },
     {
       "epoch": 5.95,
-      "learning_rate": 0.00015710982658959536,
-      "loss": 0.1198,
+      "learning_rate": 1.5745664739884393e-05,
+      "loss": 3.8458,
       "step": 1326
     },
     {
       "epoch": 5.95,
-      "learning_rate": 0.00015693641618497108,
-      "loss": 0.1045,
+      "learning_rate": 1.572832369942197e-05,
+      "loss": 3.8642,
       "step": 1327
     },
     {
       "epoch": 5.96,
-      "learning_rate": 0.0001567630057803468,
-      "loss": 0.0901,
+      "learning_rate": 1.5710982658959537e-05,
+      "loss": 3.7582,
       "step": 1328
     },
     {
       "epoch": 5.96,
-      "learning_rate": 0.00015658959537572255,
-      "loss": 0.0972,
+      "learning_rate": 1.569364161849711e-05,
+      "loss": 3.7628,
       "step": 1329
     },
     {
       "epoch": 5.96,
-      "learning_rate": 0.00015641618497109827,
-      "loss": 0.0917,
+      "learning_rate": 1.567630057803468e-05,
+      "loss": 3.6299,
       "step": 1330
     },
     {
       "epoch": 5.97,
-      "learning_rate": 0.00015624277456647396,
-      "loss": 0.0869,
+      "learning_rate": 1.5658959537572257e-05,
+      "loss": 3.6943,
       "step": 1331
     },
     {
       "epoch": 5.97,
-      "learning_rate": 0.00015606936416184968,
-      "loss": 0.1015,
+      "learning_rate": 1.564161849710983e-05,
+      "loss": 3.6493,
       "step": 1332
     },
     {
       "epoch": 5.98,
-      "learning_rate": 0.00015589595375722543,
-      "loss": 0.0788,
+      "learning_rate": 1.5624277456647397e-05,
+      "loss": 3.7038,
       "step": 1333
     },
     {
       "epoch": 5.98,
-      "learning_rate": 0.00015572254335260115,
-      "loss": 0.1053,
+      "learning_rate": 1.560693641618497e-05,
+      "loss": 3.7886,
       "step": 1334
     },
     {
       "epoch": 5.99,
-      "learning_rate": 0.00015554913294797687,
-      "loss": 0.0844,
+      "learning_rate": 1.5589595375722545e-05,
+      "loss": 3.7207,
       "step": 1335
     },
     {
       "epoch": 5.99,
-      "learning_rate": 0.00015537572254335256,
-      "loss": 0.0606,
+      "learning_rate": 1.5572254335260117e-05,
+      "loss": 3.6326,
       "step": 1336
     },
     {
       "epoch": 6.0,
-      "learning_rate": 0.0001552023121387283,
-      "loss": 0.0745,
+      "learning_rate": 1.555491329479769e-05,
+      "loss": 3.7846,
       "step": 1337
     },
     {
       "epoch": 6.0,
-      "learning_rate": 0.00015502890173410403,
-      "loss": 0.1608,
+      "learning_rate": 1.5537572254335258e-05,
+      "loss": 3.9457,
       "step": 1338
     },
     {
       "epoch": 6.0,
-      "learning_rate": 0.00015485549132947975,
-      "loss": 0.1628,
+      "learning_rate": 1.5520231213872833e-05,
+      "loss": 4.1404,
       "step": 1339
     },
     {
       "epoch": 6.01,
-      "learning_rate": 0.00015468208092485547,
-      "loss": 0.1261,
+      "learning_rate": 1.5502890173410405e-05,
+      "loss": 4.0547,
       "step": 1340
     },
     {
       "epoch": 6.01,
-      "learning_rate": 0.00015450867052023122,
-      "loss": 0.1063,
+      "learning_rate": 1.5485549132947977e-05,
+      "loss": 4.1299,
       "step": 1341
     },
     {
       "epoch": 6.02,
-      "learning_rate": 0.00015433526011560692,
-      "loss": 0.0978,
+      "learning_rate": 1.546820809248555e-05,
+      "loss": 4.0622,
       "step": 1342
     },
     {
       "epoch": 6.02,
-      "learning_rate": 0.00015416184971098264,
-      "loss": 0.0947,
+      "learning_rate": 1.545086705202312e-05,
+      "loss": 4.1236,
       "step": 1343
     },
     {
       "epoch": 6.03,
-      "learning_rate": 0.00015398843930635836,
-      "loss": 0.1061,
+      "learning_rate": 1.5433526011560694e-05,
+      "loss": 4.099,
       "step": 1344
     },
     {
       "epoch": 6.03,
-      "learning_rate": 0.0001538150289017341,
-      "loss": 0.0974,
+      "learning_rate": 1.5416184971098266e-05,
+      "loss": 4.0269,
       "step": 1345
     },
     {
       "epoch": 6.04,
-      "learning_rate": 0.00015364161849710983,
-      "loss": 0.0876,
+      "learning_rate": 1.5398843930635838e-05,
+      "loss": 4.102,
       "step": 1346
     },
     {
       "epoch": 6.04,
-      "learning_rate": 0.00015346820809248555,
-      "loss": 0.098,
+      "learning_rate": 1.5381502890173413e-05,
+      "loss": 4.0486,
       "step": 1347
     },
     {
       "epoch": 6.04,
-      "learning_rate": 0.00015329479768786124,
-      "loss": 0.0721,
+      "learning_rate": 1.5364161849710985e-05,
+      "loss": 3.9785,
       "step": 1348
     },
     {
       "epoch": 6.05,
-      "learning_rate": 0.000153121387283237,
-      "loss": 0.088,
+      "learning_rate": 1.5346820809248554e-05,
+      "loss": 3.881,
       "step": 1349
     },
     {
       "epoch": 6.05,
-      "learning_rate": 0.0001529479768786127,
-      "loss": 0.0813,
+      "learning_rate": 1.5329479768786126e-05,
+      "loss": 3.8907,
       "step": 1350
     },
     {
       "epoch": 6.06,
-      "learning_rate": 0.00015277456647398843,
-      "loss": 0.0787,
+      "learning_rate": 1.53121387283237e-05,
+      "loss": 3.9125,
       "step": 1351
     },
     {
       "epoch": 6.06,
-      "learning_rate": 0.00015260115606936415,
-      "loss": 0.0781,
+      "learning_rate": 1.5294797687861274e-05,
+      "loss": 3.8936,
       "step": 1352
     },
     {
       "epoch": 6.07,
-      "learning_rate": 0.0001524277456647399,
-      "loss": 0.0658,
+      "learning_rate": 1.5277456647398846e-05,
+      "loss": 3.7634,
       "step": 1353
     },
     {
       "epoch": 6.07,
-      "learning_rate": 0.0001522543352601156,
-      "loss": 0.0721,
+      "learning_rate": 1.5260115606936414e-05,
+      "loss": 3.7423,
       "step": 1354
     },
     {
       "epoch": 6.08,
-      "learning_rate": 0.0001520809248554913,
-      "loss": 0.0887,
+      "learning_rate": 1.524277456647399e-05,
+      "loss": 3.7204,
       "step": 1355
     },
     {
       "epoch": 6.08,
-      "learning_rate": 0.00015190751445086703,
-      "loss": 0.056,
+      "learning_rate": 1.5225433526011562e-05,
+      "loss": 3.6845,
       "step": 1356
     },
     {
       "epoch": 6.09,
-      "learning_rate": 0.00015173410404624278,
-      "loss": 0.0834,
+      "learning_rate": 1.5208092485549134e-05,
+      "loss": 3.7513,
       "step": 1357
     },
     {
       "epoch": 6.09,
-      "learning_rate": 0.0001515606936416185,
-      "loss": 0.0627,
+      "learning_rate": 1.5190751445086704e-05,
+      "loss": 3.7427,
       "step": 1358
     },
     {
       "epoch": 6.09,
-      "learning_rate": 0.0001513872832369942,
-      "loss": 0.0651,
+      "learning_rate": 1.517341040462428e-05,
+      "loss": 3.6635,
       "step": 1359
     },
     {
       "epoch": 6.1,
-      "learning_rate": 0.00015121387283236992,
-      "loss": 0.0729,
+      "learning_rate": 1.515606936416185e-05,
+      "loss": 3.6328,
       "step": 1360
     },
     {
       "epoch": 6.1,
-      "learning_rate": 0.00015104046242774566,
-      "loss": 0.0604,
+      "learning_rate": 1.5138728323699422e-05,
+      "loss": 3.613,
       "step": 1361
     },
     {
       "epoch": 6.11,
-      "learning_rate": 0.00015086705202312138,
-      "loss": 0.0651,
+      "learning_rate": 1.5121387283236994e-05,
+      "loss": 3.5995,
       "step": 1362
     },
     {
       "epoch": 6.11,
-      "learning_rate": 0.0001506936416184971,
-      "loss": 0.0665,
+      "learning_rate": 1.5104046242774568e-05,
+      "loss": 3.5823,
       "step": 1363
     },
     {
       "epoch": 6.12,
-      "learning_rate": 0.0001505202312138728,
-      "loss": 0.0703,
+      "learning_rate": 1.508670520231214e-05,
+      "loss": 3.6251,
       "step": 1364
     },
     {
       "epoch": 6.12,
-      "learning_rate": 0.00015034682080924855,
-      "loss": 0.0708,
+      "learning_rate": 1.5069364161849712e-05,
+      "loss": 3.6519,
       "step": 1365
     },
     {
       "epoch": 6.13,
-      "learning_rate": 0.00015017341040462427,
-      "loss": 0.0615,
+      "learning_rate": 1.5052023121387282e-05,
+      "loss": 3.5135,
       "step": 1366
     },
     {
       "epoch": 6.13,
-      "learning_rate": 0.00015,
-      "loss": 0.0501,
+      "learning_rate": 1.5034682080924858e-05,
+      "loss": 3.526,
       "step": 1367
     },
     {
       "epoch": 6.13,
-      "learning_rate": 0.0001498265895953757,
-      "loss": 0.0749,
+      "learning_rate": 1.5017341040462428e-05,
+      "loss": 3.5666,
       "step": 1368
     },
     {
       "epoch": 6.14,
-      "learning_rate": 0.00014965317919075143,
-      "loss": 0.068,
+      "learning_rate": 1.5e-05,
+      "loss": 3.5295,
       "step": 1369
     },
     {
       "epoch": 6.14,
-      "learning_rate": 0.00014947976878612715,
-      "loss": 0.0599,
+      "learning_rate": 1.4982658959537572e-05,
+      "loss": 3.4438,
       "step": 1370
     },
     {
       "epoch": 6.15,
-      "learning_rate": 0.00014930635838150287,
-      "loss": 0.0688,
+      "learning_rate": 1.4965317919075145e-05,
+      "loss": 3.4523,
       "step": 1371
     },
     {
       "epoch": 6.15,
-      "learning_rate": 0.0001491329479768786,
-      "loss": 0.0627,
+      "learning_rate": 1.4947976878612717e-05,
+      "loss": 3.4697,
       "step": 1372
     },
     {
       "epoch": 6.16,
-      "learning_rate": 0.0001489595375722543,
-      "loss": 0.0627,
+      "learning_rate": 1.4930635838150289e-05,
+      "loss": 3.6016,
       "step": 1373
     },
     {
       "epoch": 6.16,
-      "learning_rate": 0.00014878612716763003,
-      "loss": 0.0514,
+      "learning_rate": 1.491329479768786e-05,
+      "loss": 3.469,
       "step": 1374
     },
     {
       "epoch": 6.17,
-      "learning_rate": 0.00014861271676300578,
-      "loss": 0.0587,
+      "learning_rate": 1.4895953757225434e-05,
+      "loss": 3.563,
       "step": 1375
     },
     {
       "epoch": 6.17,
-      "learning_rate": 0.00014843930635838147,
-      "loss": 0.0575,
+      "learning_rate": 1.4878612716763005e-05,
+      "loss": 3.4659,
       "step": 1376
     },
     {
       "epoch": 6.17,
-      "learning_rate": 0.00014826589595375722,
-      "loss": 0.0561,
+      "learning_rate": 1.4861271676300579e-05,
+      "loss": 3.4131,
       "step": 1377
     },
     {
       "epoch": 6.18,
-      "learning_rate": 0.00014809248554913291,
-      "loss": 0.074,
+      "learning_rate": 1.484393063583815e-05,
+      "loss": 3.5381,
       "step": 1378
     },
     {
       "epoch": 6.18,
-      "learning_rate": 0.00014791907514450866,
-      "loss": 0.0533,
+      "learning_rate": 1.4826589595375723e-05,
+      "loss": 3.5364,
       "step": 1379
     },
     {
       "epoch": 6.19,
-      "learning_rate": 0.00014774566473988438,
-      "loss": 0.0455,
+      "learning_rate": 1.4809248554913295e-05,
+      "loss": 3.4761,
       "step": 1380
     },
     {
       "epoch": 6.19,
-      "learning_rate": 0.0001475722543352601,
-      "loss": 0.0586,
+      "learning_rate": 1.4791907514450867e-05,
+      "loss": 3.5366,
       "step": 1381
     },
     {
       "epoch": 6.2,
-      "learning_rate": 0.00014739884393063582,
-      "loss": 0.0423,
+      "learning_rate": 1.4774566473988439e-05,
+      "loss": 3.5096,
       "step": 1382
     },
     {
       "epoch": 6.2,
-      "learning_rate": 0.00014722543352601154,
-      "loss": 0.0507,
+      "learning_rate": 1.4757225433526013e-05,
+      "loss": 3.4822,
       "step": 1383
     },
     {
       "epoch": 6.21,
-      "learning_rate": 0.00014705202312138727,
-      "loss": 0.0511,
+      "learning_rate": 1.4739884393063583e-05,
+      "loss": 3.6349,
       "step": 1384
     },
     {
       "epoch": 6.21,
-      "learning_rate": 0.00014687861271676299,
-      "loss": 0.0786,
+      "learning_rate": 1.4722543352601157e-05,
+      "loss": 3.5348,
       "step": 1385
     },
     {
       "epoch": 6.22,
-      "learning_rate": 0.0001467052023121387,
-      "loss": 0.0446,
+      "learning_rate": 1.4705202312138727e-05,
+      "loss": 3.4334,
       "step": 1386
     },
     {
       "epoch": 6.22,
-      "learning_rate": 0.00014653179190751443,
-      "loss": 0.0511,
+      "learning_rate": 1.4687861271676301e-05,
+      "loss": 3.4998,
       "step": 1387
     },
     {
       "epoch": 6.22,
-      "learning_rate": 0.00014635838150289015,
-      "loss": 0.0718,
+      "learning_rate": 1.4670520231213873e-05,
+      "loss": 3.4966,
       "step": 1388
     },
     {
       "epoch": 6.23,
-      "learning_rate": 0.0001461849710982659,
-      "loss": 0.1718,
+      "learning_rate": 1.4653179190751445e-05,
+      "loss": 4.3748,
       "step": 1389
     },
     {
       "epoch": 6.23,
-      "learning_rate": 0.0001460115606936416,
-      "loss": 0.1058,
+      "learning_rate": 1.4635838150289017e-05,
+      "loss": 4.2535,
       "step": 1390
     },
     {
       "epoch": 6.24,
-      "learning_rate": 0.00014583815028901734,
-      "loss": 0.1114,
+      "learning_rate": 1.4618497109826591e-05,
+      "loss": 4.3442,
       "step": 1391
     },
     {
       "epoch": 6.24,
-      "learning_rate": 0.00014566473988439306,
-      "loss": 0.1036,
+      "learning_rate": 1.4601156069364161e-05,
+      "loss": 4.3623,
       "step": 1392
     },
     {
       "epoch": 6.25,
-      "learning_rate": 0.00014549132947976878,
-      "loss": 0.092,
+      "learning_rate": 1.4583815028901735e-05,
+      "loss": 4.3147,
       "step": 1393
     },
     {
       "epoch": 6.25,
-      "learning_rate": 0.0001453179190751445,
-      "loss": 0.0812,
+      "learning_rate": 1.4566473988439305e-05,
+      "loss": 4.2769,
       "step": 1394
     },
     {
       "epoch": 6.26,
-      "learning_rate": 0.00014514450867052022,
-      "loss": 0.1137,
+      "learning_rate": 1.454913294797688e-05,
+      "loss": 4.2314,
       "step": 1395
     },
     {
       "epoch": 6.26,
-      "learning_rate": 0.00014497109826589594,
-      "loss": 0.0868,
+      "learning_rate": 1.4531791907514451e-05,
+      "loss": 4.2258,
       "step": 1396
     },
     {
       "epoch": 6.26,
-      "learning_rate": 0.00014479768786127166,
-      "loss": 0.0821,
+      "learning_rate": 1.4514450867052023e-05,
+      "loss": 4.1569,
       "step": 1397
     },
     {
       "epoch": 6.27,
-      "learning_rate": 0.00014462427745664738,
-      "loss": 0.1276,
+      "learning_rate": 1.4497109826589595e-05,
+      "loss": 4.0524,
       "step": 1398
     },
     {
       "epoch": 6.27,
-      "learning_rate": 0.0001444508670520231,
-      "loss": 0.0639,
+      "learning_rate": 1.4479768786127167e-05,
+      "loss": 3.9985,
       "step": 1399
     },
     {
       "epoch": 6.28,
-      "learning_rate": 0.00014427745664739882,
-      "loss": 0.0941,
+      "learning_rate": 1.446242774566474e-05,
+      "loss": 3.9162,
       "step": 1400
     },
     {
       "epoch": 6.28,
-      "learning_rate": 0.00014410404624277454,
-      "loss": 0.0772,
+      "learning_rate": 1.4445086705202313e-05,
+      "loss": 3.899,
       "step": 1401
     },
     {
       "epoch": 6.29,
-      "learning_rate": 0.00014393063583815026,
-      "loss": 0.0682,
+      "learning_rate": 1.4427745664739884e-05,
+      "loss": 3.7695,
       "step": 1402
     },
     {
       "epoch": 6.29,
-      "learning_rate": 0.000143757225433526,
-      "loss": 0.0796,
+      "learning_rate": 1.4410404624277457e-05,
+      "loss": 3.7958,
       "step": 1403
     },
     {
       "epoch": 6.3,
-      "learning_rate": 0.0001435838150289017,
-      "loss": 0.0795,
+      "learning_rate": 1.439306358381503e-05,
+      "loss": 3.7724,
       "step": 1404
     },
     {
       "epoch": 6.3,
-      "learning_rate": 0.00014341040462427745,
-      "loss": 0.0709,
+      "learning_rate": 1.4375722543352602e-05,
+      "loss": 3.7631,
       "step": 1405
     },
     {
       "epoch": 6.3,
-      "learning_rate": 0.00014323699421965317,
-      "loss": 0.0782,
+      "learning_rate": 1.4358381502890174e-05,
+      "loss": 3.6711,
       "step": 1406
     },
     {
       "epoch": 6.31,
-      "learning_rate": 0.0001430635838150289,
-      "loss": 0.0603,
+      "learning_rate": 1.4341040462427746e-05,
+      "loss": 3.6774,
       "step": 1407
     },
     {
       "epoch": 6.31,
-      "learning_rate": 0.00014289017341040462,
-      "loss": 0.0712,
+      "learning_rate": 1.4323699421965318e-05,
+      "loss": 3.6973,
       "step": 1408
     },
     {
       "epoch": 6.32,
-      "learning_rate": 0.00014271676300578034,
-      "loss": 0.0608,
+      "learning_rate": 1.4306358381502892e-05,
+      "loss": 3.6787,
       "step": 1409
     },
     {
       "epoch": 6.32,
-      "learning_rate": 0.00014254335260115606,
-      "loss": 0.0686,
+      "learning_rate": 1.4289017341040462e-05,
+      "loss": 3.606,
       "step": 1410
     },
     {
       "epoch": 6.33,
-      "learning_rate": 0.00014236994219653178,
-      "loss": 0.0798,
+      "learning_rate": 1.4271676300578036e-05,
+      "loss": 3.7522,
       "step": 1411
     },
     {
       "epoch": 6.33,
-      "learning_rate": 0.0001421965317919075,
-      "loss": 0.0663,
+      "learning_rate": 1.4254335260115606e-05,
+      "loss": 3.6174,
       "step": 1412
     },
     {
       "epoch": 6.34,
-      "learning_rate": 0.00014202312138728322,
-      "loss": 0.0611,
+      "learning_rate": 1.423699421965318e-05,
+      "loss": 3.6904,
       "step": 1413
     },
     {
       "epoch": 6.34,
-      "learning_rate": 0.00014184971098265894,
-      "loss": 0.0591,
+      "learning_rate": 1.4219653179190752e-05,
+      "loss": 3.5759,
       "step": 1414
     },
     {
       "epoch": 6.35,
-      "learning_rate": 0.00014167630057803466,
-      "loss": 0.0665,
+      "learning_rate": 1.4202312138728324e-05,
+      "loss": 3.6298,
       "step": 1415
     },
     {
       "epoch": 6.35,
-      "learning_rate": 0.00014150289017341038,
-      "loss": 0.0568,
+      "learning_rate": 1.4184971098265896e-05,
+      "loss": 3.5669,
       "step": 1416
     },
     {
       "epoch": 6.35,
-      "learning_rate": 0.00014132947976878613,
-      "loss": 0.0707,
+      "learning_rate": 1.416763005780347e-05,
+      "loss": 3.5675,
       "step": 1417
     },
     {
       "epoch": 6.36,
-      "learning_rate": 0.00014115606936416182,
-      "loss": 0.1241,
+      "learning_rate": 1.415028901734104e-05,
+      "loss": 3.526,
       "step": 1418
     },
     {
       "epoch": 6.36,
-      "learning_rate": 0.00014098265895953757,
-      "loss": 0.0755,
+      "learning_rate": 1.4132947976878614e-05,
+      "loss": 3.6871,
       "step": 1419
     },
     {
       "epoch": 6.37,
-      "learning_rate": 0.0001408092485549133,
-      "loss": 0.0778,
+      "learning_rate": 1.4115606936416184e-05,
+      "loss": 3.491,
       "step": 1420
     },
     {
       "epoch": 6.37,
-      "learning_rate": 0.000140635838150289,
-      "loss": 0.0621,
+      "learning_rate": 1.4098265895953758e-05,
+      "loss": 3.5132,
       "step": 1421
     },
     {
       "epoch": 6.38,
-      "learning_rate": 0.00014046242774566473,
-      "loss": 0.062,
+      "learning_rate": 1.408092485549133e-05,
+      "loss": 3.4734,
       "step": 1422
     },
     {
       "epoch": 6.38,
-      "learning_rate": 0.00014028901734104045,
-      "loss": 0.0593,
+      "learning_rate": 1.4063583815028902e-05,
+      "loss": 3.5613,
       "step": 1423
     },
     {
       "epoch": 6.39,
-      "learning_rate": 0.00014011560693641617,
-      "loss": 0.073,
+      "learning_rate": 1.4046242774566474e-05,
+      "loss": 3.5734,
       "step": 1424
     },
     {
       "epoch": 6.39,
-      "learning_rate": 0.0001399421965317919,
-      "loss": 0.0475,
+      "learning_rate": 1.4028901734104046e-05,
+      "loss": 3.5342,
       "step": 1425
     },
     {
       "epoch": 6.39,
-      "learning_rate": 0.00013976878612716762,
-      "loss": 0.0642,
+      "learning_rate": 1.4011560693641618e-05,
+      "loss": 3.5138,
       "step": 1426
     },
     {
       "epoch": 6.4,
-      "learning_rate": 0.00013959537572254334,
-      "loss": 0.0514,
+      "learning_rate": 1.3994219653179192e-05,
+      "loss": 3.4809,
       "step": 1427
     },
     {
       "epoch": 6.4,
-      "learning_rate": 0.00013942196531791906,
-      "loss": 0.0532,
+      "learning_rate": 1.3976878612716763e-05,
+      "loss": 3.4592,
       "step": 1428
     },
     {
       "epoch": 6.41,
-      "learning_rate": 0.00013924855491329478,
-      "loss": 0.0605,
+      "learning_rate": 1.3959537572254336e-05,
+      "loss": 3.4913,
       "step": 1429
     },
     {
       "epoch": 6.41,
-      "learning_rate": 0.0001390751445086705,
-      "loss": 0.0558,
+      "learning_rate": 1.3942196531791908e-05,
+      "loss": 3.5048,
       "step": 1430
     },
     {
       "epoch": 6.42,
-      "learning_rate": 0.00013890173410404625,
-      "loss": 0.0475,
+      "learning_rate": 1.392485549132948e-05,
+      "loss": 3.4408,
       "step": 1431
     },
     {
       "epoch": 6.42,
-      "learning_rate": 0.00013872832369942194,
-      "loss": 0.0571,
+      "learning_rate": 1.3907514450867053e-05,
+      "loss": 3.5879,
       "step": 1432
     },
     {
       "epoch": 6.43,
-      "learning_rate": 0.0001385549132947977,
-      "loss": 0.0638,
+      "learning_rate": 1.3890173410404625e-05,
+      "loss": 3.5279,
       "step": 1433
     },
     {
       "epoch": 6.43,
-      "learning_rate": 0.0001383815028901734,
-      "loss": 0.0507,
+      "learning_rate": 1.3872832369942197e-05,
+      "loss": 3.4807,
       "step": 1434
     },
     {
       "epoch": 6.43,
-      "learning_rate": 0.00013820809248554913,
-      "loss": 0.0645,
+      "learning_rate": 1.385549132947977e-05,
+      "loss": 3.5377,
       "step": 1435
     },
     {
       "epoch": 6.44,
-      "learning_rate": 0.00013803468208092485,
-      "loss": 0.0627,
+      "learning_rate": 1.383815028901734e-05,
+      "loss": 3.5555,
       "step": 1436
     },
     {
       "epoch": 6.44,
-      "learning_rate": 0.00013786127167630057,
-      "loss": 0.072,
+      "learning_rate": 1.3820809248554915e-05,
+      "loss": 3.5234,
       "step": 1437
     },
     {
       "epoch": 6.45,
-      "learning_rate": 0.0001376878612716763,
-      "loss": 0.1086,
+      "learning_rate": 1.3803468208092485e-05,
+      "loss": 3.3992,
       "step": 1438
     },
     {
       "epoch": 6.45,
-      "learning_rate": 0.000137514450867052,
-      "loss": 0.1585,
+      "learning_rate": 1.3786127167630059e-05,
+      "loss": 4.3738,
       "step": 1439
     },
     {
       "epoch": 6.46,
-      "learning_rate": 0.00013734104046242773,
-      "loss": 0.2069,
+      "learning_rate": 1.376878612716763e-05,
+      "loss": 4.3301,
       "step": 1440
     },
     {
       "epoch": 6.46,
-      "learning_rate": 0.00013716763005780345,
-      "loss": 0.0923,
+      "learning_rate": 1.3751445086705203e-05,
+      "loss": 4.38,
       "step": 1441
     },
     {
       "epoch": 6.47,
-      "learning_rate": 0.00013699421965317917,
-      "loss": 0.1004,
+      "learning_rate": 1.3734104046242775e-05,
+      "loss": 4.3963,
       "step": 1442
     },
     {
       "epoch": 6.47,
-      "learning_rate": 0.00013682080924855492,
-      "loss": 0.1034,
+      "learning_rate": 1.3716763005780349e-05,
+      "loss": 4.3242,
       "step": 1443
     },
     {
       "epoch": 6.48,
-      "learning_rate": 0.00013664739884393061,
-      "loss": 0.0949,
+      "learning_rate": 1.3699421965317919e-05,
+      "loss": 4.2459,
       "step": 1444
     },
     {
       "epoch": 6.48,
-      "learning_rate": 0.00013647398843930636,
-      "loss": 0.0756,
+      "learning_rate": 1.3682080924855493e-05,
+      "loss": 4.2176,
       "step": 1445
     },
     {
       "epoch": 6.48,
-      "learning_rate": 0.00013630057803468206,
-      "loss": 0.0965,
+      "learning_rate": 1.3664739884393063e-05,
+      "loss": 4.1511,
       "step": 1446
     },
     {
       "epoch": 6.49,
-      "learning_rate": 0.0001361271676300578,
-      "loss": 0.0813,
+      "learning_rate": 1.3647398843930637e-05,
+      "loss": 4.0639,
       "step": 1447
     },
     {
       "epoch": 6.49,
-      "learning_rate": 0.00013595375722543352,
-      "loss": 0.0904,
+      "learning_rate": 1.3630057803468209e-05,
+      "loss": 4.0748,
       "step": 1448
     },
     {
       "epoch": 6.5,
-      "learning_rate": 0.00013578034682080925,
-      "loss": 0.0747,
+      "learning_rate": 1.3612716763005781e-05,
+      "loss": 3.9933,
       "step": 1449
     },
     {
       "epoch": 6.5,
-      "learning_rate": 0.00013560693641618497,
-      "loss": 0.0892,
+      "learning_rate": 1.3595375722543353e-05,
+      "loss": 3.9647,
       "step": 1450
     },
     {
       "epoch": 6.51,
-      "learning_rate": 0.0001354335260115607,
-      "loss": 0.0595,
+      "learning_rate": 1.3578034682080925e-05,
+      "loss": 3.8892,
       "step": 1451
     },
     {
       "epoch": 6.51,
-      "learning_rate": 0.0001352601156069364,
-      "loss": 0.0702,
+      "learning_rate": 1.3560693641618497e-05,
+      "loss": 3.8595,
       "step": 1452
     },
     {
       "epoch": 6.52,
-      "learning_rate": 0.00013508670520231213,
-      "loss": 0.0778,
+      "learning_rate": 1.3543352601156071e-05,
+      "loss": 3.8028,
       "step": 1453
     },
     {
       "epoch": 6.52,
-      "learning_rate": 0.00013491329479768785,
-      "loss": 0.0634,
+      "learning_rate": 1.3526011560693641e-05,
+      "loss": 3.6911,
       "step": 1454
     },
     {
       "epoch": 6.52,
-      "learning_rate": 0.00013473988439306357,
-      "loss": 0.0731,
+      "learning_rate": 1.3508670520231215e-05,
+      "loss": 3.811,
       "step": 1455
     },
     {
       "epoch": 6.53,
-      "learning_rate": 0.0001345664739884393,
-      "loss": 0.076,
+      "learning_rate": 1.3491329479768787e-05,
+      "loss": 3.7722,
       "step": 1456
     },
     {
       "epoch": 6.53,
-      "learning_rate": 0.00013439306358381504,
-      "loss": 0.0652,
+      "learning_rate": 1.347398843930636e-05,
+      "loss": 3.6371,
       "step": 1457
     },
     {
       "epoch": 6.54,
-      "learning_rate": 0.00013421965317919073,
-      "loss": 0.0547,
+      "learning_rate": 1.3456647398843931e-05,
+      "loss": 3.6797,
       "step": 1458
     },
     {
       "epoch": 6.54,
-      "learning_rate": 0.00013404624277456648,
-      "loss": 0.0743,
+      "learning_rate": 1.3439306358381503e-05,
+      "loss": 3.6933,
       "step": 1459
     },
     {
       "epoch": 6.55,
-      "learning_rate": 0.00013387283236994217,
-      "loss": 0.0592,
+      "learning_rate": 1.3421965317919075e-05,
+      "loss": 3.6937,
       "step": 1460
     },
     {
       "epoch": 6.55,
-      "learning_rate": 0.00013369942196531792,
-      "loss": 0.0639,
+      "learning_rate": 1.340462427745665e-05,
+      "loss": 3.5847,
       "step": 1461
     },
     {
       "epoch": 6.56,
-      "learning_rate": 0.00013352601156069364,
-      "loss": 0.06,
+      "learning_rate": 1.338728323699422e-05,
+      "loss": 3.6837,
       "step": 1462
     },
     {
       "epoch": 6.56,
-      "learning_rate": 0.00013335260115606936,
-      "loss": 0.0712,
+      "learning_rate": 1.3369942196531793e-05,
+      "loss": 3.6907,
       "step": 1463
     },
     {
       "epoch": 6.57,
-      "learning_rate": 0.00013317919075144508,
-      "loss": 0.065,
+      "learning_rate": 1.3352601156069364e-05,
+      "loss": 3.8257,
       "step": 1464
     },
     {
       "epoch": 6.57,
-      "learning_rate": 0.00013300578034682078,
-      "loss": 0.0553,
+      "learning_rate": 1.3335260115606938e-05,
+      "loss": 3.6479,
       "step": 1465
     },
     {
       "epoch": 6.57,
-      "learning_rate": 0.00013283236994219652,
-      "loss": 0.0654,
+      "learning_rate": 1.331791907514451e-05,
+      "loss": 3.681,
       "step": 1466
     },
     {
       "epoch": 6.58,
-      "learning_rate": 0.00013265895953757224,
-      "loss": 0.0631,
+      "learning_rate": 1.330057803468208e-05,
+      "loss": 3.5989,
       "step": 1467
     },
     {
       "epoch": 6.58,
-      "learning_rate": 0.00013248554913294797,
-      "loss": 0.0729,
+      "learning_rate": 1.3283236994219654e-05,
+      "loss": 3.5327,
       "step": 1468
     },
     {
       "epoch": 6.59,
-      "learning_rate": 0.00013231213872832369,
-      "loss": 0.0612,
+      "learning_rate": 1.3265895953757226e-05,
+      "loss": 3.6351,
       "step": 1469
     },
     {
       "epoch": 6.59,
-      "learning_rate": 0.0001321387283236994,
-      "loss": 0.049,
+      "learning_rate": 1.3248554913294798e-05,
+      "loss": 3.5191,
       "step": 1470
     },
     {
       "epoch": 6.6,
-      "learning_rate": 0.00013196531791907513,
-      "loss": 0.0715,
+      "learning_rate": 1.323121387283237e-05,
+      "loss": 3.6232,
       "step": 1471
     },
     {
       "epoch": 6.6,
-      "learning_rate": 0.00013179190751445085,
-      "loss": 0.0667,
+      "learning_rate": 1.3213872832369942e-05,
+      "loss": 3.5038,
       "step": 1472
     },
     {
       "epoch": 6.61,
-      "learning_rate": 0.00013161849710982657,
-      "loss": 0.0601,
+      "learning_rate": 1.3196531791907514e-05,
+      "loss": 3.5143,
       "step": 1473
     },
     {
       "epoch": 6.61,
-      "learning_rate": 0.0001314450867052023,
-      "loss": 0.0543,
+      "learning_rate": 1.3179190751445088e-05,
+      "loss": 3.5653,
       "step": 1474
     },
     {
       "epoch": 6.61,
-      "learning_rate": 0.000131271676300578,
-      "loss": 0.0533,
+      "learning_rate": 1.3161849710982658e-05,
+      "loss": 3.4478,
       "step": 1475
     },
     {
       "epoch": 6.62,
-      "learning_rate": 0.00013109826589595376,
-      "loss": 0.0508,
+      "learning_rate": 1.3144508670520232e-05,
+      "loss": 3.5128,
       "step": 1476
     },
     {
       "epoch": 6.62,
-      "learning_rate": 0.00013092485549132945,
-      "loss": 0.0581,
+      "learning_rate": 1.3127167630057802e-05,
+      "loss": 3.4933,
       "step": 1477
     },
     {
       "epoch": 6.63,
-      "learning_rate": 0.0001307514450867052,
-      "loss": 0.0631,
+      "learning_rate": 1.3109826589595376e-05,
+      "loss": 3.395,
       "step": 1478
     },
     {
       "epoch": 6.63,
-      "learning_rate": 0.00013057803468208092,
-      "loss": 0.0545,
+      "learning_rate": 1.3092485549132948e-05,
+      "loss": 3.5296,
       "step": 1479
     },
     {
       "epoch": 6.64,
-      "learning_rate": 0.00013040462427745664,
-      "loss": 0.0621,
+      "learning_rate": 1.307514450867052e-05,
+      "loss": 3.5239,
       "step": 1480
     },
     {
       "epoch": 6.64,
-      "learning_rate": 0.00013023121387283236,
-      "loss": 0.0797,
+      "learning_rate": 1.3057803468208092e-05,
+      "loss": 3.5912,
       "step": 1481
     },
     {
       "epoch": 6.65,
-      "learning_rate": 0.00013005780346820808,
-      "loss": 0.0494,
+      "learning_rate": 1.3040462427745664e-05,
+      "loss": 3.5952,
       "step": 1482
     },
     {
       "epoch": 6.65,
-      "learning_rate": 0.0001298843930635838,
-      "loss": 0.063,
+      "learning_rate": 1.3023121387283236e-05,
+      "loss": 3.5399,
       "step": 1483
     },
     {
       "epoch": 6.65,
-      "learning_rate": 0.00012971098265895952,
-      "loss": 0.0556,
+      "learning_rate": 1.300578034682081e-05,
+      "loss": 3.5792,
       "step": 1484
     },
     {
       "epoch": 6.66,
-      "learning_rate": 0.00012953757225433524,
-      "loss": 0.031,
+      "learning_rate": 1.298843930635838e-05,
+      "loss": 3.4492,
       "step": 1485
     },
     {
       "epoch": 6.66,
-      "learning_rate": 0.00012936416184971096,
-      "loss": 0.0532,
+      "learning_rate": 1.2971098265895954e-05,
+      "loss": 3.5801,
       "step": 1486
     },
     {
       "epoch": 6.67,
-      "learning_rate": 0.00012919075144508669,
-      "loss": 0.0419,
+      "learning_rate": 1.2953757225433526e-05,
+      "loss": 3.5486,
       "step": 1487
     },
     {
       "epoch": 6.67,
-      "learning_rate": 0.0001290173410404624,
-      "loss": 0.0777,
+      "learning_rate": 1.2936416184971098e-05,
+      "loss": 3.4543,
       "step": 1488
     },
     {
       "epoch": 6.68,
-      "learning_rate": 0.00012884393063583813,
-      "loss": 0.1556,
+      "learning_rate": 1.291907514450867e-05,
+      "loss": 4.3831,
       "step": 1489
     },
     {
       "epoch": 6.68,
-      "learning_rate": 0.00012867052023121387,
-      "loss": 0.15,
+      "learning_rate": 1.2901734104046243e-05,
+      "loss": 4.3403,
       "step": 1490
     },
     {
       "epoch": 6.69,
-      "learning_rate": 0.00012849710982658957,
-      "loss": 0.1092,
+      "learning_rate": 1.2884393063583815e-05,
+      "loss": 4.2843,
       "step": 1491
     },
     {
       "epoch": 6.69,
-      "learning_rate": 0.00012832369942196532,
-      "loss": 0.0985,
+      "learning_rate": 1.2867052023121388e-05,
+      "loss": 4.1974,
       "step": 1492
     },
     {
       "epoch": 6.7,
-      "learning_rate": 0.00012815028901734104,
-      "loss": 0.089,
+      "learning_rate": 1.2849710982658959e-05,
+      "loss": 4.1307,
       "step": 1493
     },
     {
       "epoch": 6.7,
-      "learning_rate": 0.00012797687861271676,
-      "loss": 0.0951,
+      "learning_rate": 1.2832369942196533e-05,
+      "loss": 4.2715,
       "step": 1494
     },
     {
       "epoch": 6.7,
-      "learning_rate": 0.00012780346820809248,
-      "loss": 0.0904,
+      "learning_rate": 1.2815028901734103e-05,
+      "loss": 4.1558,
       "step": 1495
     },
     {
       "epoch": 6.71,
-      "learning_rate": 0.0001276300578034682,
-      "loss": 0.0874,
+      "learning_rate": 1.2797687861271677e-05,
+      "loss": 4.1503,
       "step": 1496
     },
     {
       "epoch": 6.71,
-      "learning_rate": 0.00012745664739884392,
-      "loss": 0.0686,
+      "learning_rate": 1.2780346820809249e-05,
+      "loss": 4.1042,
       "step": 1497
     },
     {
       "epoch": 6.72,
-      "learning_rate": 0.00012728323699421964,
-      "loss": 0.0789,
+      "learning_rate": 1.2763005780346821e-05,
+      "loss": 4.0457,
       "step": 1498
     },
     {
       "epoch": 6.72,
-      "learning_rate": 0.00012710982658959536,
-      "loss": 0.0757,
+      "learning_rate": 1.2745664739884393e-05,
+      "loss": 3.8454,
       "step": 1499
     },
     {
       "epoch": 6.73,
-      "learning_rate": 0.00012693641618497108,
-      "loss": 0.0886,
+      "learning_rate": 1.2728323699421967e-05,
+      "loss": 3.8931,
       "step": 1500
     },
     {
       "epoch": 6.73,
-      "eval_loss": 0.3513650596141815,
-      "eval_runtime": 571.1505,
-      "eval_samples_per_second": 4.626,
-      "eval_steps_per_second": 0.58,
-      "eval_wer": 0.10093216977389925,
+      "eval_loss": 4.250516891479492,
+      "eval_runtime": 463.1359,
+      "eval_samples_per_second": 5.705,
+      "eval_steps_per_second": 0.715,
+      "eval_wer": 0.9688218960729869,
       "step": 1500
     },
     {
       "epoch": 6.73,
-      "learning_rate": 0.0001267630057803468,
-      "loss": 0.0772,
+      "learning_rate": 1.2710982658959537e-05,
+      "loss": 3.8691,
       "step": 1501
     },
     {
       "epoch": 6.74,
-      "learning_rate": 0.00012658959537572252,
-      "loss": 0.0731,
+      "learning_rate": 1.269364161849711e-05,
+      "loss": 3.7624,
       "step": 1502
     },
     {
       "epoch": 6.74,
-      "learning_rate": 0.00012641618497109824,
-      "loss": 0.0637,
+      "learning_rate": 1.2676300578034681e-05,
+      "loss": 3.7811,
       "step": 1503
     },
     {
       "epoch": 6.74,
-      "learning_rate": 0.000126242774566474,
-      "loss": 0.0737,
+      "learning_rate": 1.2658959537572255e-05,
+      "loss": 3.7243,
       "step": 1504
     },
     {
       "epoch": 6.75,
-      "learning_rate": 0.00012606936416184968,
-      "loss": 0.0885,
+      "learning_rate": 1.2641618497109827e-05,
+      "loss": 3.7005,
       "step": 1505
     },
     {
       "epoch": 6.75,
-      "learning_rate": 0.00012589595375722543,
-      "loss": 0.0575,
+      "learning_rate": 1.2624277456647399e-05,
+      "loss": 3.7262,
       "step": 1506
     },
     {
       "epoch": 6.76,
-      "learning_rate": 0.00012572254335260115,
-      "loss": 0.0623,
+      "learning_rate": 1.2606936416184971e-05,
+      "loss": 3.6828,
       "step": 1507
     },
     {
       "epoch": 6.76,
-      "learning_rate": 0.00012554913294797687,
-      "loss": 0.0807,
+      "learning_rate": 1.2589595375722543e-05,
+      "loss": 3.7144,
       "step": 1508
     },
     {
       "epoch": 6.77,
-      "learning_rate": 0.0001253757225433526,
-      "loss": 0.0606,
+      "learning_rate": 1.2572254335260115e-05,
+      "loss": 3.778,
       "step": 1509
     },
     {
       "epoch": 6.77,
-      "learning_rate": 0.00012520231213872831,
-      "loss": 0.0596,
+      "learning_rate": 1.2554913294797689e-05,
+      "loss": 3.6453,
       "step": 1510
     },
     {
       "epoch": 6.78,
-      "learning_rate": 0.00012502890173410404,
-      "loss": 0.066,
+      "learning_rate": 1.253757225433526e-05,
+      "loss": 3.6811,
       "step": 1511
     },
     {
       "epoch": 6.78,
-      "learning_rate": 0.00012485549132947976,
-      "loss": 0.0648,
+      "learning_rate": 1.2520231213872833e-05,
+      "loss": 3.7192,
       "step": 1512
     },
     {
       "epoch": 6.78,
-      "learning_rate": 0.00012468208092485548,
-      "loss": 0.0554,
+      "learning_rate": 1.2502890173410405e-05,
+      "loss": 3.6474,
       "step": 1513
     },
     {
       "epoch": 6.79,
-      "learning_rate": 0.0001245086705202312,
-      "loss": 0.0601,
+      "learning_rate": 1.2485549132947977e-05,
+      "loss": 3.5742,
       "step": 1514
     },
     {
       "epoch": 6.79,
-      "learning_rate": 0.00012433526011560692,
-      "loss": 0.0526,
+      "learning_rate": 1.246820809248555e-05,
+      "loss": 3.6431,
       "step": 1515
     },
     {
       "epoch": 6.8,
-      "learning_rate": 0.00012416184971098267,
-      "loss": 0.0653,
+      "learning_rate": 1.2450867052023121e-05,
+      "loss": 3.5768,
       "step": 1516
     },
     {
       "epoch": 6.8,
-      "learning_rate": 0.00012398843930635836,
-      "loss": 0.0682,
+      "learning_rate": 1.2433526011560694e-05,
+      "loss": 3.6144,
       "step": 1517
     },
     {
       "epoch": 6.81,
-      "learning_rate": 0.0001238150289017341,
-      "loss": 0.0612,
+      "learning_rate": 1.2416184971098267e-05,
+      "loss": 3.5914,
       "step": 1518
     },
     {
       "epoch": 6.81,
-      "learning_rate": 0.0001236416184971098,
-      "loss": 0.0616,
+      "learning_rate": 1.2398843930635838e-05,
+      "loss": 3.5791,
       "step": 1519
     },
     {
       "epoch": 6.82,
-      "learning_rate": 0.00012346820809248555,
-      "loss": 0.0609,
+      "learning_rate": 1.2381502890173411e-05,
+      "loss": 3.5251,
       "step": 1520
     },
     {
       "epoch": 6.82,
-      "learning_rate": 0.00012329479768786127,
-      "loss": 0.0565,
+      "learning_rate": 1.2364161849710982e-05,
+      "loss": 3.577,
       "step": 1521
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.000123121387283237,
-      "loss": 0.0499,
+      "learning_rate": 1.2346820809248556e-05,
+      "loss": 3.5221,
       "step": 1522
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.0001229479768786127,
-      "loss": 0.0652,
+      "learning_rate": 1.2329479768786128e-05,
+      "loss": 3.5566,
       "step": 1523
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.00012277456647398843,
-      "loss": 0.0601,
+      "learning_rate": 1.23121387283237e-05,
+      "loss": 3.4536,
       "step": 1524
     },
     {
       "epoch": 6.84,
-      "learning_rate": 0.00012260115606936415,
-      "loss": 0.0348,
+      "learning_rate": 1.2294797687861272e-05,
+      "loss": 3.4887,
       "step": 1525
     },
     {
       "epoch": 6.84,
-      "learning_rate": 0.00012242774566473987,
-      "loss": 0.0554,
+      "learning_rate": 1.2277456647398846e-05,
+      "loss": 3.5162,
       "step": 1526
     },
     {
       "epoch": 6.85,
-      "learning_rate": 0.0001222543352601156,
-      "loss": 0.0515,
+      "learning_rate": 1.2260115606936416e-05,
+      "loss": 3.607,
       "step": 1527
     },
     {
       "epoch": 6.85,
-      "learning_rate": 0.00012208092485549131,
-      "loss": 0.0556,
+      "learning_rate": 1.224277456647399e-05,
+      "loss": 3.5811,
       "step": 1528
     },
     {
       "epoch": 6.86,
-      "learning_rate": 0.00012190751445086703,
-      "loss": 0.1181,
+      "learning_rate": 1.222543352601156e-05,
+      "loss": 3.5118,
       "step": 1529
     },
     {
       "epoch": 6.86,
-      "learning_rate": 0.00012173410404624277,
-      "loss": 0.0656,
+      "learning_rate": 1.2208092485549134e-05,
+      "loss": 3.474,
       "step": 1530
     },
     {
       "epoch": 6.87,
-      "learning_rate": 0.00012156069364161849,
-      "loss": 0.045,
+      "learning_rate": 1.2190751445086706e-05,
+      "loss": 3.5014,
       "step": 1531
     },
     {
       "epoch": 6.87,
-      "learning_rate": 0.00012138728323699421,
-      "loss": 0.0486,
+      "learning_rate": 1.2173410404624278e-05,
+      "loss": 3.5547,
       "step": 1532
     },
     {
       "epoch": 6.87,
-      "learning_rate": 0.00012121387283236993,
-      "loss": 0.0517,
+      "learning_rate": 1.215606936416185e-05,
+      "loss": 3.4679,
       "step": 1533
     },
     {
       "epoch": 6.88,
-      "learning_rate": 0.00012104046242774567,
-      "loss": 0.0482,
+      "learning_rate": 1.2138728323699422e-05,
+      "loss": 3.495,
       "step": 1534
     },
     {
       "epoch": 6.88,
-      "learning_rate": 0.00012086705202312137,
-      "loss": 0.0384,
+      "learning_rate": 1.2121387283236994e-05,
+      "loss": 3.503,
       "step": 1535
     },
     {
       "epoch": 6.89,
-      "learning_rate": 0.0001206936416184971,
-      "loss": 0.1258,
+      "learning_rate": 1.2104046242774568e-05,
+      "loss": 3.4486,
       "step": 1536
     },
     {
       "epoch": 6.89,
-      "learning_rate": 0.00012052023121387281,
-      "loss": 0.0393,
+      "learning_rate": 1.2086705202312138e-05,
+      "loss": 3.6023,
       "step": 1537
     },
     {
       "epoch": 6.9,
-      "learning_rate": 0.00012034682080924855,
-      "loss": 0.0976,
+      "learning_rate": 1.2069364161849712e-05,
+      "loss": 3.4638,
       "step": 1538
     },
     {
       "epoch": 6.9,
-      "learning_rate": 0.00012017341040462427,
-      "loss": 0.1335,
+      "learning_rate": 1.2052023121387284e-05,
+      "loss": 4.2609,
       "step": 1539
     },
     {
       "epoch": 6.91,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 0.0879,
+      "learning_rate": 1.2034682080924856e-05,
+      "loss": 4.2512,
       "step": 1540
     },
     {
       "epoch": 6.91,
-      "learning_rate": 0.00011982658959537571,
-      "loss": 0.0893,
+      "learning_rate": 1.2017341040462428e-05,
+      "loss": 4.2611,
       "step": 1541
     },
     {
       "epoch": 6.91,
-      "learning_rate": 0.00011965317919075144,
-      "loss": 0.0729,
+      "learning_rate": 1.2e-05,
+      "loss": 4.1527,
       "step": 1542
     },
     {
       "epoch": 6.92,
-      "learning_rate": 0.00011947976878612715,
-      "loss": 0.0855,
+      "learning_rate": 1.1982658959537572e-05,
+      "loss": 4.1289,
       "step": 1543
     },
     {
       "epoch": 6.92,
-      "learning_rate": 0.00011930635838150289,
-      "loss": 0.0607,
+      "learning_rate": 1.1965317919075146e-05,
+      "loss": 3.9784,
       "step": 1544
     },
     {
       "epoch": 6.93,
-      "learning_rate": 0.0001191329479768786,
-      "loss": 0.0633,
+      "learning_rate": 1.1947976878612717e-05,
+      "loss": 3.9353,
       "step": 1545
     },
     {
       "epoch": 6.93,
-      "learning_rate": 0.00011895953757225433,
-      "loss": 0.0864,
+      "learning_rate": 1.193063583815029e-05,
+      "loss": 3.8506,
       "step": 1546
     },
     {
       "epoch": 6.94,
-      "learning_rate": 0.00011878612716763005,
-      "loss": 0.0667,
+      "learning_rate": 1.191329479768786e-05,
+      "loss": 3.8273,
       "step": 1547
     },
     {
       "epoch": 6.94,
-      "learning_rate": 0.00011861271676300578,
-      "loss": 0.0621,
+      "learning_rate": 1.1895953757225434e-05,
+      "loss": 3.6927,
       "step": 1548
     },
     {
       "epoch": 6.95,
-      "learning_rate": 0.00011843930635838149,
-      "loss": 0.0582,
+      "learning_rate": 1.1878612716763006e-05,
+      "loss": 3.6501,
       "step": 1549
     },
     {
       "epoch": 6.95,
-      "learning_rate": 0.00011826589595375722,
-      "loss": 0.0563,
+      "learning_rate": 1.1861271676300579e-05,
+      "loss": 3.7406,
       "step": 1550
     },
     {
       "epoch": 6.96,
-      "learning_rate": 0.00011809248554913293,
-      "loss": 0.0595,
+      "learning_rate": 1.184393063583815e-05,
+      "loss": 3.5916,
       "step": 1551
     },
     {
       "epoch": 6.96,
-      "learning_rate": 0.00011791907514450866,
-      "loss": 0.0622,
+      "learning_rate": 1.1826589595375724e-05,
+      "loss": 3.6129,
       "step": 1552
     },
     {
       "epoch": 6.96,
-      "learning_rate": 0.00011774566473988439,
-      "loss": 0.0489,
+      "learning_rate": 1.1809248554913295e-05,
+      "loss": 3.5014,
       "step": 1553
     },
     {
       "epoch": 6.97,
-      "learning_rate": 0.00011757225433526012,
-      "loss": 0.07,
+      "learning_rate": 1.1791907514450869e-05,
+      "loss": 3.5187,
       "step": 1554
     },
     {
       "epoch": 6.97,
-      "learning_rate": 0.00011739884393063583,
-      "loss": 0.0699,
+      "learning_rate": 1.1774566473988439e-05,
+      "loss": 3.5085,
       "step": 1555
     },
     {
       "epoch": 6.98,
-      "learning_rate": 0.00011722543352601156,
-      "loss": 0.0564,
+      "learning_rate": 1.1757225433526013e-05,
+      "loss": 3.5914,
       "step": 1556
     },
     {
       "epoch": 6.98,
-      "learning_rate": 0.00011705202312138727,
-      "loss": 0.0654,
+      "learning_rate": 1.1739884393063585e-05,
+      "loss": 3.528,
       "step": 1557
     },
     {
       "epoch": 6.99,
-      "learning_rate": 0.000116878612716763,
-      "loss": 0.0719,
+      "learning_rate": 1.1722543352601157e-05,
+      "loss": 3.5764,
       "step": 1558
     },
     {
       "epoch": 6.99,
-      "learning_rate": 0.00011670520231213872,
-      "loss": 0.0555,
+      "learning_rate": 1.1705202312138729e-05,
+      "loss": 3.499,
       "step": 1559
     },
     {
       "epoch": 7.0,
-      "learning_rate": 0.00011653179190751443,
-      "loss": 0.0539,
+      "learning_rate": 1.1687861271676301e-05,
+      "loss": 3.4236,
       "step": 1560
     },
     {
       "epoch": 7.0,
-      "learning_rate": 0.00011635838150289016,
-      "loss": 0.1026,
+      "learning_rate": 1.1670520231213873e-05,
+      "loss": 3.7797,
       "step": 1561
     },
     {
       "epoch": 7.0,
-      "learning_rate": 0.00011618497109826587,
-      "loss": 0.082,
+      "learning_rate": 1.1653179190751445e-05,
+      "loss": 4.0943,
       "step": 1562
     },
     {
       "epoch": 7.01,
-      "learning_rate": 0.0001160115606936416,
-      "loss": 0.0802,
+      "learning_rate": 1.1635838150289017e-05,
+      "loss": 4.0141,
       "step": 1563
     },
     {
       "epoch": 7.01,
-      "learning_rate": 0.00011583815028901733,
-      "loss": 0.0734,
+      "learning_rate": 1.161849710982659e-05,
+      "loss": 4.037,
       "step": 1564
     },
     {
       "epoch": 7.02,
-      "learning_rate": 0.00011566473988439306,
-      "loss": 0.075,
+      "learning_rate": 1.1601156069364163e-05,
+      "loss": 4.0608,
       "step": 1565
     },
     {
       "epoch": 7.02,
-      "learning_rate": 0.00011549132947976877,
-      "loss": 0.074,
+      "learning_rate": 1.1583815028901733e-05,
+      "loss": 4.001,
       "step": 1566
     },
     {
       "epoch": 7.03,
-      "learning_rate": 0.0001153179190751445,
-      "loss": 0.0624,
+      "learning_rate": 1.1566473988439307e-05,
+      "loss": 4.0066,
       "step": 1567
     },
     {
       "epoch": 7.03,
-      "learning_rate": 0.00011514450867052021,
-      "loss": 0.0662,
+      "learning_rate": 1.1549132947976877e-05,
+      "loss": 4.0685,
       "step": 1568
     },
     {
       "epoch": 7.04,
-      "learning_rate": 0.00011497109826589594,
-      "loss": 0.0631,
+      "learning_rate": 1.1531791907514451e-05,
+      "loss": 3.9795,
       "step": 1569
     },
     {
       "epoch": 7.04,
-      "learning_rate": 0.00011479768786127166,
-      "loss": 0.0493,
+      "learning_rate": 1.1514450867052023e-05,
+      "loss": 3.9148,
       "step": 1570
     },
     {
       "epoch": 7.04,
-      "learning_rate": 0.00011462427745664738,
-      "loss": 0.0642,
+      "learning_rate": 1.1497109826589595e-05,
+      "loss": 3.8969,
       "step": 1571
     },
     {
       "epoch": 7.05,
-      "learning_rate": 0.0001144508670520231,
-      "loss": 0.0557,
+      "learning_rate": 1.1479768786127167e-05,
+      "loss": 3.8096,
       "step": 1572
     },
     {
       "epoch": 7.05,
-      "learning_rate": 0.00011427745664739884,
-      "loss": 0.0515,
+      "learning_rate": 1.146242774566474e-05,
+      "loss": 3.8556,
       "step": 1573
     },
     {
       "epoch": 7.06,
-      "learning_rate": 0.00011410404624277455,
-      "loss": 0.0615,
+      "learning_rate": 1.1445086705202312e-05,
+      "loss": 3.7949,
       "step": 1574
     },
     {
       "epoch": 7.06,
-      "learning_rate": 0.00011393063583815028,
-      "loss": 0.0486,
+      "learning_rate": 1.1427745664739885e-05,
+      "loss": 3.7384,
       "step": 1575
     },
     {
       "epoch": 7.07,
-      "learning_rate": 0.00011375722543352599,
-      "loss": 0.0503,
+      "learning_rate": 1.1410404624277456e-05,
+      "loss": 3.6781,
       "step": 1576
     },
     {
       "epoch": 7.07,
-      "learning_rate": 0.00011358381502890172,
-      "loss": 0.0713,
+      "learning_rate": 1.139306358381503e-05,
+      "loss": 3.7412,
       "step": 1577
     },
     {
       "epoch": 7.08,
-      "learning_rate": 0.00011341040462427744,
-      "loss": 0.0545,
+      "learning_rate": 1.1375722543352602e-05,
+      "loss": 3.6372,
       "step": 1578
     },
     {
       "epoch": 7.08,
-      "learning_rate": 0.00011323699421965318,
-      "loss": 0.0459,
+      "learning_rate": 1.1358381502890174e-05,
+      "loss": 3.6381,
       "step": 1579
     },
     {
       "epoch": 7.09,
-      "learning_rate": 0.00011306358381502888,
-      "loss": 0.0357,
+      "learning_rate": 1.1341040462427746e-05,
+      "loss": 3.5741,
       "step": 1580
     },
     {
       "epoch": 7.09,
-      "learning_rate": 0.00011289017341040462,
-      "loss": 0.0504,
+      "learning_rate": 1.1323699421965318e-05,
+      "loss": 3.5746,
       "step": 1581
     },
     {
       "epoch": 7.09,
-      "learning_rate": 0.00011271676300578033,
-      "loss": 0.0427,
+      "learning_rate": 1.130635838150289e-05,
+      "loss": 3.5849,
       "step": 1582
     },
     {
       "epoch": 7.1,
-      "learning_rate": 0.00011254335260115606,
-      "loss": 0.0584,
+      "learning_rate": 1.1289017341040464e-05,
+      "loss": 3.601,
       "step": 1583
     },
     {
       "epoch": 7.1,
-      "learning_rate": 0.00011236994219653178,
-      "loss": 0.0566,
+      "learning_rate": 1.1271676300578034e-05,
+      "loss": 3.5217,
       "step": 1584
     },
     {
       "epoch": 7.11,
-      "learning_rate": 0.0001121965317919075,
-      "loss": 0.0452,
+      "learning_rate": 1.1254335260115608e-05,
+      "loss": 3.5282,
       "step": 1585
     },
     {
       "epoch": 7.11,
-      "learning_rate": 0.00011202312138728322,
-      "loss": 0.0555,
+      "learning_rate": 1.1236994219653178e-05,
+      "loss": 3.4845,
       "step": 1586
     },
     {
       "epoch": 7.12,
-      "learning_rate": 0.00011184971098265896,
-      "loss": 0.0471,
+      "learning_rate": 1.1219653179190752e-05,
+      "loss": 3.5286,
       "step": 1587
     },
     {
       "epoch": 7.12,
-      "learning_rate": 0.00011167630057803466,
-      "loss": 0.0488,
+      "learning_rate": 1.1202312138728324e-05,
+      "loss": 3.4687,
       "step": 1588
     },
     {
       "epoch": 7.13,
-      "learning_rate": 0.0001115028901734104,
-      "loss": 0.0331,
+      "learning_rate": 1.1184971098265896e-05,
+      "loss": 3.4723,
       "step": 1589
     },
     {
       "epoch": 7.13,
-      "learning_rate": 0.00011132947976878612,
-      "loss": 0.0503,
+      "learning_rate": 1.1167630057803468e-05,
+      "loss": 3.4709,
       "step": 1590
     },
     {
       "epoch": 7.13,
-      "learning_rate": 0.00011115606936416184,
-      "loss": 0.0475,
+      "learning_rate": 1.1150289017341042e-05,
+      "loss": 3.4791,
       "step": 1591
     },
     {
       "epoch": 7.14,
-      "learning_rate": 0.00011098265895953756,
-      "loss": 0.0457,
+      "learning_rate": 1.1132947976878612e-05,
+      "loss": 3.515,
       "step": 1592
     },
     {
       "epoch": 7.14,
-      "learning_rate": 0.0001108092485549133,
-      "loss": 0.0453,
+      "learning_rate": 1.1115606936416186e-05,
+      "loss": 3.3335,
       "step": 1593
     },
     {
       "epoch": 7.15,
-      "learning_rate": 0.000110635838150289,
-      "loss": 0.048,
+      "learning_rate": 1.1098265895953756e-05,
+      "loss": 3.3489,
       "step": 1594
     },
     {
       "epoch": 7.15,
-      "learning_rate": 0.00011046242774566474,
-      "loss": 0.043,
+      "learning_rate": 1.108092485549133e-05,
+      "loss": 3.3639,
       "step": 1595
     },
     {
       "epoch": 7.16,
-      "learning_rate": 0.00011028901734104044,
-      "loss": 0.0318,
+      "learning_rate": 1.1063583815028902e-05,
+      "loss": 3.4261,
       "step": 1596
     },
     {
       "epoch": 7.16,
-      "learning_rate": 0.00011011560693641618,
-      "loss": 0.0444,
+      "learning_rate": 1.1046242774566474e-05,
+      "loss": 3.3638,
       "step": 1597
     },
     {
       "epoch": 7.17,
-      "learning_rate": 0.0001099421965317919,
-      "loss": 0.0316,
+      "learning_rate": 1.1028901734104046e-05,
+      "loss": 3.4047,
       "step": 1598
     },
     {
       "epoch": 7.17,
-      "learning_rate": 0.00010976878612716762,
-      "loss": 0.0392,
+      "learning_rate": 1.1011560693641618e-05,
+      "loss": 3.4339,
       "step": 1599
     },
     {
       "epoch": 7.17,
-      "learning_rate": 0.00010959537572254334,
-      "loss": 0.0411,
+      "learning_rate": 1.099421965317919e-05,
+      "loss": 3.4643,
       "step": 1600
     },
     {
       "epoch": 7.18,
-      "learning_rate": 0.00010942196531791907,
-      "loss": 0.0407,
+      "learning_rate": 1.0976878612716764e-05,
+      "loss": 3.4215,
       "step": 1601
     },
     {
       "epoch": 7.18,
-      "learning_rate": 0.00010924855491329478,
-      "loss": 0.0376,
+      "learning_rate": 1.0959537572254335e-05,
+      "loss": 3.3619,
       "step": 1602
     },
     {
       "epoch": 7.19,
-      "learning_rate": 0.00010907514450867051,
-      "loss": 0.0372,
+      "learning_rate": 1.0942196531791908e-05,
+      "loss": 3.3778,
       "step": 1603
     },
     {
       "epoch": 7.19,
-      "learning_rate": 0.00010890173410404623,
-      "loss": 0.0388,
+      "learning_rate": 1.092485549132948e-05,
+      "loss": 3.5079,
       "step": 1604
     },
     {
       "epoch": 7.2,
-      "learning_rate": 0.00010872832369942196,
-      "loss": 0.0316,
+      "learning_rate": 1.0907514450867052e-05,
+      "loss": 3.447,
       "step": 1605
     },
     {
       "epoch": 7.2,
-      "learning_rate": 0.00010855491329479768,
-      "loss": 0.0364,
+      "learning_rate": 1.0890173410404625e-05,
+      "loss": 3.3798,
       "step": 1606
     },
     {
       "epoch": 7.21,
-      "learning_rate": 0.00010838150289017341,
-      "loss": 0.0496,
+      "learning_rate": 1.0872832369942197e-05,
+      "loss": 3.4926,
       "step": 1607
     },
     {
       "epoch": 7.21,
-      "learning_rate": 0.00010820809248554912,
-      "loss": 0.0483,
+      "learning_rate": 1.0855491329479769e-05,
+      "loss": 3.4982,
       "step": 1608
     },
     {
       "epoch": 7.22,
-      "learning_rate": 0.00010803468208092485,
-      "loss": 0.0372,
+      "learning_rate": 1.0838150289017342e-05,
+      "loss": 3.2367,
       "step": 1609
     },
     {
       "epoch": 7.22,
-      "learning_rate": 0.00010786127167630056,
-      "loss": 0.0383,
+      "learning_rate": 1.0820809248554913e-05,
+      "loss": 3.4525,
       "step": 1610
     },
     {
       "epoch": 7.22,
-      "learning_rate": 0.00010768786127167629,
-      "loss": 0.0558,
+      "learning_rate": 1.0803468208092487e-05,
+      "loss": 3.355,
       "step": 1611
     },
     {
       "epoch": 7.23,
-      "learning_rate": 0.00010751445086705201,
-      "loss": 0.0953,
+      "learning_rate": 1.0786127167630057e-05,
+      "loss": 4.2457,
       "step": 1612
     },
     {
       "epoch": 7.23,
-      "learning_rate": 0.00010734104046242773,
-      "loss": 0.0717,
+      "learning_rate": 1.076878612716763e-05,
+      "loss": 4.2183,
       "step": 1613
     },
     {
       "epoch": 7.24,
-      "learning_rate": 0.00010716763005780346,
-      "loss": 0.0711,
+      "learning_rate": 1.0751445086705203e-05,
+      "loss": 4.2492,
       "step": 1614
     },
     {
       "epoch": 7.24,
-      "learning_rate": 0.00010699421965317919,
-      "loss": 0.0735,
+      "learning_rate": 1.0734104046242775e-05,
+      "loss": 4.2798,
       "step": 1615
     },
     {
       "epoch": 7.25,
-      "learning_rate": 0.0001068208092485549,
-      "loss": 0.0702,
+      "learning_rate": 1.0716763005780347e-05,
+      "loss": 4.214,
       "step": 1616
     },
     {
       "epoch": 7.25,
-      "learning_rate": 0.00010664739884393063,
-      "loss": 0.0634,
+      "learning_rate": 1.069942196531792e-05,
+      "loss": 4.1478,
       "step": 1617
     },
     {
       "epoch": 7.26,
-      "learning_rate": 0.00010647398843930635,
-      "loss": 0.0637,
+      "learning_rate": 1.0682080924855491e-05,
+      "loss": 4.1246,
       "step": 1618
     },
     {
       "epoch": 7.26,
-      "learning_rate": 0.00010630057803468207,
-      "loss": 0.0591,
+      "learning_rate": 1.0664739884393065e-05,
+      "loss": 4.1029,
       "step": 1619
     },
     {
       "epoch": 7.26,
-      "learning_rate": 0.00010612716763005779,
-      "loss": 0.0628,
+      "learning_rate": 1.0647398843930635e-05,
+      "loss": 4.0118,
       "step": 1620
     },
     {
       "epoch": 7.27,
-      "learning_rate": 0.00010595375722543353,
-      "loss": 0.0501,
+      "learning_rate": 1.0630057803468209e-05,
+      "loss": 3.9219,
       "step": 1621
     },
     {
       "epoch": 7.27,
-      "learning_rate": 0.00010578034682080923,
-      "loss": 0.0643,
+      "learning_rate": 1.0612716763005781e-05,
+      "loss": 3.9431,
       "step": 1622
     },
     {
       "epoch": 7.28,
-      "learning_rate": 0.00010560693641618497,
-      "loss": 0.0555,
+      "learning_rate": 1.0595375722543353e-05,
+      "loss": 3.7916,
       "step": 1623
     },
     {
       "epoch": 7.28,
-      "learning_rate": 0.00010543352601156068,
-      "loss": 0.0606,
+      "learning_rate": 1.0578034682080925e-05,
+      "loss": 3.855,
       "step": 1624
     },
     {
       "epoch": 7.29,
-      "learning_rate": 0.00010526011560693641,
-      "loss": 0.0447,
+      "learning_rate": 1.0560693641618497e-05,
+      "loss": 3.7926,
       "step": 1625
     },
     {
       "epoch": 7.29,
-      "learning_rate": 0.00010508670520231213,
-      "loss": 0.0522,
+      "learning_rate": 1.054335260115607e-05,
+      "loss": 3.6741,
       "step": 1626
     },
     {
       "epoch": 7.3,
-      "learning_rate": 0.00010491329479768786,
-      "loss": 0.0526,
+      "learning_rate": 1.0526011560693643e-05,
+      "loss": 3.6696,
       "step": 1627
     },
     {
       "epoch": 7.3,
-      "learning_rate": 0.00010473988439306357,
-      "loss": 0.046,
+      "learning_rate": 1.0508670520231213e-05,
+      "loss": 3.6256,
       "step": 1628
     },
     {
       "epoch": 7.3,
-      "learning_rate": 0.0001045664739884393,
-      "loss": 0.0683,
+      "learning_rate": 1.0491329479768787e-05,
+      "loss": 3.6661,
       "step": 1629
     },
     {
       "epoch": 7.31,
-      "learning_rate": 0.00010439306358381501,
-      "loss": 0.0516,
+      "learning_rate": 1.047398843930636e-05,
+      "loss": 3.6186,
       "step": 1630
     },
     {
       "epoch": 7.31,
-      "learning_rate": 0.00010421965317919075,
-      "loss": 0.0431,
+      "learning_rate": 1.0456647398843931e-05,
+      "loss": 3.5236,
       "step": 1631
     },
     {
       "epoch": 7.32,
-      "learning_rate": 0.00010404624277456647,
-      "loss": 0.0407,
+      "learning_rate": 1.0439306358381503e-05,
+      "loss": 3.5413,
       "step": 1632
     },
     {
       "epoch": 7.32,
-      "learning_rate": 0.00010387283236994219,
-      "loss": 0.0553,
+      "learning_rate": 1.0421965317919075e-05,
+      "loss": 3.5598,
       "step": 1633
     },
     {
       "epoch": 7.33,
-      "learning_rate": 0.00010369942196531791,
-      "loss": 0.0441,
+      "learning_rate": 1.0404624277456647e-05,
+      "loss": 3.4842,
       "step": 1634
     },
     {
       "epoch": 7.33,
-      "learning_rate": 0.00010352601156069364,
-      "loss": 0.0361,
+      "learning_rate": 1.0387283236994221e-05,
+      "loss": 3.5515,
       "step": 1635
     },
     {
       "epoch": 7.34,
-      "learning_rate": 0.00010335260115606935,
-      "loss": 0.0417,
+      "learning_rate": 1.0369942196531792e-05,
+      "loss": 3.4904,
       "step": 1636
     },
     {
       "epoch": 7.34,
-      "learning_rate": 0.00010317919075144509,
-      "loss": 0.0548,
+      "learning_rate": 1.0352601156069365e-05,
+      "loss": 3.6319,
       "step": 1637
     },
     {
       "epoch": 7.35,
-      "learning_rate": 0.00010300578034682079,
-      "loss": 0.0483,
+      "learning_rate": 1.0335260115606936e-05,
+      "loss": 3.5538,
       "step": 1638
     },
     {
       "epoch": 7.35,
-      "learning_rate": 0.00010283236994219653,
-      "loss": 0.0421,
+      "learning_rate": 1.031791907514451e-05,
+      "loss": 3.5301,
       "step": 1639
     },
     {
       "epoch": 7.35,
-      "learning_rate": 0.00010265895953757225,
-      "loss": 0.0444,
+      "learning_rate": 1.0300578034682082e-05,
+      "loss": 3.3783,
       "step": 1640
     },
     {
       "epoch": 7.36,
-      "learning_rate": 0.00010248554913294798,
-      "loss": 0.037,
+      "learning_rate": 1.0283236994219654e-05,
+      "loss": 3.4897,
       "step": 1641
     },
     {
       "epoch": 7.36,
-      "learning_rate": 0.00010231213872832369,
-      "loss": 0.046,
+      "learning_rate": 1.0265895953757226e-05,
+      "loss": 3.5303,
       "step": 1642
     },
     {
       "epoch": 7.37,
-      "learning_rate": 0.00010213872832369942,
-      "loss": 0.0377,
+      "learning_rate": 1.02485549132948e-05,
+      "loss": 3.4219,
       "step": 1643
     },
     {
       "epoch": 7.37,
-      "learning_rate": 0.00010196531791907513,
-      "loss": 0.043,
+      "learning_rate": 1.023121387283237e-05,
+      "loss": 3.4381,
       "step": 1644
     },
     {
       "epoch": 7.38,
-      "learning_rate": 0.00010179190751445086,
-      "loss": 0.0417,
+      "learning_rate": 1.0213872832369944e-05,
+      "loss": 3.4455,
       "step": 1645
     },
     {
       "epoch": 7.38,
-      "learning_rate": 0.00010161849710982658,
-      "loss": 0.0372,
+      "learning_rate": 1.0196531791907514e-05,
+      "loss": 3.4048,
       "step": 1646
     },
     {
       "epoch": 7.39,
-      "learning_rate": 0.0001014450867052023,
-      "loss": 0.0432,
+      "learning_rate": 1.0179190751445088e-05,
+      "loss": 3.3894,
       "step": 1647
     },
     {
       "epoch": 7.39,
-      "learning_rate": 0.00010127167630057803,
-      "loss": 0.0434,
+      "learning_rate": 1.016184971098266e-05,
+      "loss": 3.4203,
       "step": 1648
     },
     {
       "epoch": 7.39,
-      "learning_rate": 0.00010109826589595376,
-      "loss": 0.0352,
+      "learning_rate": 1.0144508670520232e-05,
+      "loss": 3.4005,
       "step": 1649
     },
     {
       "epoch": 7.4,
-      "learning_rate": 0.00010092485549132947,
-      "loss": 0.0402,
+      "learning_rate": 1.0127167630057804e-05,
+      "loss": 3.4222,
       "step": 1650
     },
     {
       "epoch": 7.4,
-      "learning_rate": 0.0001007514450867052,
-      "loss": 0.0381,
+      "learning_rate": 1.0109826589595376e-05,
+      "loss": 3.3879,
       "step": 1651
     },
     {
       "epoch": 7.41,
-      "learning_rate": 0.00010057803468208092,
-      "loss": 0.0418,
+      "learning_rate": 1.0092485549132948e-05,
+      "loss": 3.4199,
       "step": 1652
     },
     {
       "epoch": 7.41,
-      "learning_rate": 0.00010040462427745664,
-      "loss": 0.0339,
+      "learning_rate": 1.0075144508670522e-05,
+      "loss": 3.3692,
       "step": 1653
     },
     {
       "epoch": 7.42,
-      "learning_rate": 0.00010023121387283236,
-      "loss": 0.0462,
+      "learning_rate": 1.0057803468208092e-05,
+      "loss": 3.3981,
       "step": 1654
     },
     {
       "epoch": 7.42,
-      "learning_rate": 0.0001000578034682081,
-      "loss": 0.0406,
+      "learning_rate": 1.0040462427745666e-05,
+      "loss": 3.4547,
       "step": 1655
     },
     {
       "epoch": 7.43,
-      "learning_rate": 9.98843930635838e-05,
-      "loss": 0.0479,
+      "learning_rate": 1.0023121387283238e-05,
+      "loss": 3.4876,
       "step": 1656
     },
     {
       "epoch": 7.43,
-      "learning_rate": 9.971098265895953e-05,
-      "loss": 0.0354,
+      "learning_rate": 1.000578034682081e-05,
+      "loss": 3.3046,
       "step": 1657
     },
     {
       "epoch": 7.43,
-      "learning_rate": 9.953757225433525e-05,
-      "loss": 0.0551,
+      "learning_rate": 9.988439306358382e-06,
+      "loss": 3.3553,
       "step": 1658
     },
     {
       "epoch": 7.44,
-      "learning_rate": 9.936416184971097e-05,
-      "loss": 0.0287,
+      "learning_rate": 9.971098265895953e-06,
+      "loss": 3.3635,
       "step": 1659
     },
     {
       "epoch": 7.44,
-      "learning_rate": 9.91907514450867e-05,
-      "loss": 0.0342,
+      "learning_rate": 9.953757225433526e-06,
+      "loss": 3.4029,
       "step": 1660
     },
     {
       "epoch": 7.45,
-      "learning_rate": 9.901734104046241e-05,
-      "loss": 0.063,
+      "learning_rate": 9.936416184971098e-06,
+      "loss": 3.2374,
       "step": 1661
     },
     {
       "epoch": 7.45,
-      "learning_rate": 9.884393063583814e-05,
-      "loss": 0.0938,
+      "learning_rate": 9.91907514450867e-06,
+      "loss": 4.3088,
       "step": 1662
     },
     {
       "epoch": 7.46,
-      "learning_rate": 9.867052023121385e-05,
-      "loss": 0.0789,
+      "learning_rate": 9.901734104046243e-06,
+      "loss": 4.2106,
       "step": 1663
     },
     {
       "epoch": 7.46,
-      "learning_rate": 9.849710982658958e-05,
-      "loss": 0.0637,
+      "learning_rate": 9.884393063583815e-06,
+      "loss": 4.2904,
       "step": 1664
     },
     {
       "epoch": 7.47,
-      "learning_rate": 9.83236994219653e-05,
-      "loss": 0.0571,
+      "learning_rate": 9.867052023121387e-06,
+      "loss": 4.0952,
       "step": 1665
     },
     {
       "epoch": 7.47,
-      "learning_rate": 9.815028901734104e-05,
-      "loss": 0.0686,
+      "learning_rate": 9.84971098265896e-06,
+      "loss": 4.232,
       "step": 1666
     },
     {
       "epoch": 7.48,
-      "learning_rate": 9.797687861271675e-05,
-      "loss": 0.0591,
+      "learning_rate": 9.83236994219653e-06,
+      "loss": 4.1302,
       "step": 1667
     },
     {
       "epoch": 7.48,
-      "learning_rate": 9.780346820809248e-05,
-      "loss": 0.0504,
+      "learning_rate": 9.815028901734105e-06,
+      "loss": 4.1167,
       "step": 1668
     },
     {
       "epoch": 7.48,
-      "learning_rate": 9.763005780346819e-05,
-      "loss": 0.0571,
+      "learning_rate": 9.797687861271677e-06,
+      "loss": 4.0666,
       "step": 1669
     },
     {
       "epoch": 7.49,
-      "learning_rate": 9.745664739884392e-05,
-      "loss": 0.056,
+      "learning_rate": 9.780346820809249e-06,
+      "loss": 4.0464,
       "step": 1670
     },
     {
       "epoch": 7.49,
-      "learning_rate": 9.728323699421964e-05,
-      "loss": 0.0677,
+      "learning_rate": 9.76300578034682e-06,
+      "loss": 4.0075,
       "step": 1671
     },
     {
       "epoch": 7.5,
-      "learning_rate": 9.710982658959536e-05,
-      "loss": 0.0481,
+      "learning_rate": 9.745664739884393e-06,
+      "loss": 3.8552,
       "step": 1672
     },
     {
       "epoch": 7.5,
-      "learning_rate": 9.693641618497108e-05,
-      "loss": 0.0565,
+      "learning_rate": 9.728323699421965e-06,
+      "loss": 3.8782,
       "step": 1673
     },
     {
       "epoch": 7.51,
-      "learning_rate": 9.676300578034682e-05,
-      "loss": 0.052,
+      "learning_rate": 9.710982658959539e-06,
+      "loss": 3.8213,
       "step": 1674
     },
     {
       "epoch": 7.51,
-      "learning_rate": 9.658959537572253e-05,
-      "loss": 0.0489,
+      "learning_rate": 9.693641618497109e-06,
+      "loss": 3.7914,
       "step": 1675
     },
     {
       "epoch": 7.52,
-      "learning_rate": 9.641618497109826e-05,
-      "loss": 0.0433,
+      "learning_rate": 9.676300578034683e-06,
+      "loss": 3.702,
       "step": 1676
     },
     {
       "epoch": 7.52,
-      "learning_rate": 9.624277456647398e-05,
-      "loss": 0.0491,
+      "learning_rate": 9.658959537572253e-06,
+      "loss": 3.5826,
       "step": 1677
     },
     {
       "epoch": 7.52,
-      "learning_rate": 9.60693641618497e-05,
-      "loss": 0.0445,
+      "learning_rate": 9.641618497109827e-06,
+      "loss": 3.624,
       "step": 1678
     },
     {
       "epoch": 7.53,
-      "learning_rate": 9.589595375722542e-05,
-      "loss": 0.0593,
+      "learning_rate": 9.624277456647399e-06,
+      "loss": 3.5874,
       "step": 1679
     },
     {
       "epoch": 7.53,
-      "learning_rate": 9.572254335260116e-05,
-      "loss": 0.0491,
+      "learning_rate": 9.606936416184971e-06,
+      "loss": 3.5884,
       "step": 1680
     },
     {
       "epoch": 7.54,
-      "learning_rate": 9.554913294797686e-05,
-      "loss": 0.0614,
+      "learning_rate": 9.589595375722543e-06,
+      "loss": 3.5871,
       "step": 1681
     },
     {
       "epoch": 7.54,
-      "learning_rate": 9.53757225433526e-05,
-      "loss": 0.0562,
+      "learning_rate": 9.572254335260117e-06,
+      "loss": 3.5571,
       "step": 1682
     },
     {
       "epoch": 7.55,
-      "learning_rate": 9.52023121387283e-05,
-      "loss": 0.0352,
+      "learning_rate": 9.554913294797687e-06,
+      "loss": 3.5846,
       "step": 1683
     },
     {
       "epoch": 7.55,
-      "learning_rate": 9.502890173410404e-05,
-      "loss": 0.0484,
+      "learning_rate": 9.537572254335261e-06,
+      "loss": 3.5604,
       "step": 1684
     },
     {
       "epoch": 7.56,
-      "learning_rate": 9.485549132947976e-05,
-      "loss": 0.0393,
+      "learning_rate": 9.520231213872831e-06,
+      "loss": 3.5964,
       "step": 1685
     },
     {
       "epoch": 7.56,
-      "learning_rate": 9.468208092485548e-05,
-      "loss": 0.0397,
+      "learning_rate": 9.502890173410405e-06,
+      "loss": 3.5461,
       "step": 1686
     },
     {
       "epoch": 7.57,
-      "learning_rate": 9.45086705202312e-05,
-      "loss": 0.0423,
+      "learning_rate": 9.485549132947977e-06,
+      "loss": 3.5809,
       "step": 1687
     },
     {
       "epoch": 7.57,
-      "learning_rate": 9.433526011560693e-05,
-      "loss": 0.0478,
+      "learning_rate": 9.46820809248555e-06,
+      "loss": 3.522,
       "step": 1688
     },
     {
       "epoch": 7.57,
-      "learning_rate": 9.416184971098264e-05,
-      "loss": 0.0464,
+      "learning_rate": 9.450867052023121e-06,
+      "loss": 3.4954,
       "step": 1689
     },
     {
       "epoch": 7.58,
-      "learning_rate": 9.398843930635838e-05,
-      "loss": 0.0414,
+      "learning_rate": 9.433526011560693e-06,
+      "loss": 3.4639,
       "step": 1690
     },
     {
       "epoch": 7.58,
-      "learning_rate": 9.38150289017341e-05,
-      "loss": 0.051,
+      "learning_rate": 9.416184971098266e-06,
+      "loss": 3.5024,
       "step": 1691
     },
     {
       "epoch": 7.59,
-      "learning_rate": 9.364161849710982e-05,
-      "loss": 0.0405,
+      "learning_rate": 9.39884393063584e-06,
+      "loss": 3.4632,
       "step": 1692
     },
     {
       "epoch": 7.59,
-      "learning_rate": 9.346820809248554e-05,
-      "loss": 0.0464,
+      "learning_rate": 9.38150289017341e-06,
+      "loss": 3.3797,
       "step": 1693
     },
     {
       "epoch": 7.6,
-      "learning_rate": 9.329479768786127e-05,
-      "loss": 0.0312,
+      "learning_rate": 9.364161849710983e-06,
+      "loss": 3.3826,
       "step": 1694
     },
     {
       "epoch": 7.6,
-      "learning_rate": 9.312138728323698e-05,
-      "loss": 0.0429,
+      "learning_rate": 9.346820809248555e-06,
+      "loss": 3.3758,
       "step": 1695
     },
     {
       "epoch": 7.61,
-      "learning_rate": 9.294797687861271e-05,
-      "loss": 0.0376,
+      "learning_rate": 9.329479768786128e-06,
+      "loss": 3.4563,
       "step": 1696
     },
     {
       "epoch": 7.61,
-      "learning_rate": 9.277456647398842e-05,
-      "loss": 0.0374,
+      "learning_rate": 9.3121387283237e-06,
+      "loss": 3.3694,
       "step": 1697
     },
     {
       "epoch": 7.61,
-      "learning_rate": 9.260115606936415e-05,
-      "loss": 0.0412,
+      "learning_rate": 9.294797687861272e-06,
+      "loss": 3.4125,
       "step": 1698
     },
     {
       "epoch": 7.62,
-      "learning_rate": 9.242774566473988e-05,
-      "loss": 0.029,
+      "learning_rate": 9.277456647398844e-06,
+      "loss": 3.4251,
       "step": 1699
     },
     {
       "epoch": 7.62,
-      "learning_rate": 9.22543352601156e-05,
-      "loss": 0.0462,
+      "learning_rate": 9.260115606936418e-06,
+      "loss": 3.3842,
       "step": 1700
     },
     {
       "epoch": 7.63,
-      "learning_rate": 9.208092485549132e-05,
-      "loss": 0.0403,
+      "learning_rate": 9.242774566473988e-06,
+      "loss": 3.3716,
       "step": 1701
     },
     {
       "epoch": 7.63,
-      "learning_rate": 9.190751445086705e-05,
-      "loss": 0.0455,
+      "learning_rate": 9.225433526011562e-06,
+      "loss": 3.4167,
       "step": 1702
     },
     {
       "epoch": 7.64,
-      "learning_rate": 9.173410404624276e-05,
-      "loss": 0.044,
+      "learning_rate": 9.208092485549132e-06,
+      "loss": 3.4427,
       "step": 1703
     },
     {
       "epoch": 7.64,
-      "learning_rate": 9.156069364161849e-05,
-      "loss": 0.0309,
+      "learning_rate": 9.190751445086706e-06,
+      "loss": 3.399,
       "step": 1704
     },
     {
       "epoch": 7.65,
-      "learning_rate": 9.138728323699421e-05,
-      "loss": 0.0503,
+      "learning_rate": 9.173410404624278e-06,
+      "loss": 3.3828,
       "step": 1705
     },
     {
       "epoch": 7.65,
-      "learning_rate": 9.121387283236993e-05,
-      "loss": 0.0443,
+      "learning_rate": 9.15606936416185e-06,
+      "loss": 3.3617,
       "step": 1706
     },
     {
       "epoch": 7.65,
-      "learning_rate": 9.104046242774565e-05,
-      "loss": 0.0415,
+      "learning_rate": 9.138728323699422e-06,
+      "loss": 3.4299,
       "step": 1707
     },
     {
       "epoch": 7.66,
-      "learning_rate": 9.086705202312139e-05,
-      "loss": 0.0499,
+      "learning_rate": 9.121387283236994e-06,
+      "loss": 3.4361,
       "step": 1708
     },
     {
       "epoch": 7.66,
-      "learning_rate": 9.06936416184971e-05,
-      "loss": 0.0416,
+      "learning_rate": 9.104046242774566e-06,
+      "loss": 3.4543,
       "step": 1709
     },
     {
       "epoch": 7.67,
-      "learning_rate": 9.052023121387283e-05,
-      "loss": 0.0572,
+      "learning_rate": 9.08670520231214e-06,
+      "loss": 3.3756,
       "step": 1710
     },
     {
       "epoch": 7.67,
-      "learning_rate": 9.034682080924854e-05,
-      "loss": 0.0469,
+      "learning_rate": 9.06936416184971e-06,
+      "loss": 3.4149,
       "step": 1711
     },
     {
       "epoch": 7.68,
-      "learning_rate": 9.017341040462427e-05,
-      "loss": 0.0813,
+      "learning_rate": 9.052023121387284e-06,
+      "loss": 4.2696,
       "step": 1712
     },
     {
       "epoch": 7.68,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 0.0663,
+      "learning_rate": 9.034682080924856e-06,
+      "loss": 4.1461,
       "step": 1713
     },
     {
       "epoch": 7.69,
-      "learning_rate": 8.982658959537573e-05,
-      "loss": 0.0592,
+      "learning_rate": 9.017341040462428e-06,
+      "loss": 4.2457,
       "step": 1714
     },
     {
       "epoch": 7.69,
-      "learning_rate": 8.965317919075143e-05,
-      "loss": 0.0604,
+      "learning_rate": 9e-06,
+      "loss": 4.1243,
       "step": 1715
     },
     {
       "epoch": 7.7,
-      "learning_rate": 8.947976878612717e-05,
-      "loss": 0.0669,
+      "learning_rate": 8.982658959537572e-06,
+      "loss": 4.1863,
       "step": 1716
     },
     {
       "epoch": 7.7,
-      "learning_rate": 8.930635838150287e-05,
-      "loss": 0.0608,
+      "learning_rate": 8.965317919075144e-06,
+      "loss": 4.0753,
       "step": 1717
     },
     {
       "epoch": 7.7,
-      "learning_rate": 8.913294797687861e-05,
-      "loss": 0.0554,
+      "learning_rate": 8.947976878612718e-06,
+      "loss": 4.0233,
       "step": 1718
     },
     {
       "epoch": 7.71,
-      "learning_rate": 8.895953757225433e-05,
-      "loss": 0.0624,
+      "learning_rate": 8.930635838150288e-06,
+      "loss": 4.1104,
       "step": 1719
     },
     {
       "epoch": 7.71,
-      "learning_rate": 8.878612716763005e-05,
-      "loss": 0.0649,
+      "learning_rate": 8.913294797687862e-06,
+      "loss": 4.0023,
       "step": 1720
     },
     {
       "epoch": 7.72,
-      "learning_rate": 8.861271676300577e-05,
-      "loss": 0.057,
+      "learning_rate": 8.895953757225433e-06,
+      "loss": 4.0233,
       "step": 1721
     },
     {
       "epoch": 7.72,
-      "learning_rate": 8.84393063583815e-05,
-      "loss": 0.0445,
+      "learning_rate": 8.878612716763006e-06,
+      "loss": 3.9096,
       "step": 1722
     },
     {
       "epoch": 7.73,
-      "learning_rate": 8.826589595375721e-05,
-      "loss": 0.0504,
+      "learning_rate": 8.861271676300578e-06,
+      "loss": 3.8776,
       "step": 1723
     },
     {
       "epoch": 7.73,
-      "learning_rate": 8.809248554913295e-05,
-      "loss": 0.0529,
+      "learning_rate": 8.84393063583815e-06,
+      "loss": 3.7885,
       "step": 1724
     },
     {
       "epoch": 7.74,
-      "learning_rate": 8.791907514450865e-05,
-      "loss": 0.0628,
+      "learning_rate": 8.826589595375723e-06,
+      "loss": 3.6858,
       "step": 1725
     },
     {
       "epoch": 7.74,
-      "learning_rate": 8.774566473988439e-05,
-      "loss": 0.047,
+      "learning_rate": 8.809248554913296e-06,
+      "loss": 3.6899,
       "step": 1726
     },
     {
       "epoch": 7.74,
-      "learning_rate": 8.757225433526011e-05,
-      "loss": 0.0438,
+      "learning_rate": 8.791907514450867e-06,
+      "loss": 3.641,
       "step": 1727
     },
     {
       "epoch": 7.75,
-      "learning_rate": 8.739884393063584e-05,
-      "loss": 0.0491,
+      "learning_rate": 8.77456647398844e-06,
+      "loss": 3.6487,
       "step": 1728
     },
     {
       "epoch": 7.75,
-      "learning_rate": 8.722543352601155e-05,
-      "loss": 0.0572,
+      "learning_rate": 8.757225433526011e-06,
+      "loss": 3.5889,
       "step": 1729
     },
     {
       "epoch": 7.76,
-      "learning_rate": 8.705202312138728e-05,
-      "loss": 0.0363,
+      "learning_rate": 8.739884393063585e-06,
+      "loss": 3.5744,
       "step": 1730
     },
     {
       "epoch": 7.76,
-      "learning_rate": 8.687861271676299e-05,
-      "loss": 0.0537,
+      "learning_rate": 8.722543352601157e-06,
+      "loss": 3.6601,
       "step": 1731
     },
     {
       "epoch": 7.77,
-      "learning_rate": 8.670520231213873e-05,
-      "loss": 0.0313,
+      "learning_rate": 8.705202312138729e-06,
+      "loss": 3.5943,
       "step": 1732
     },
     {
       "epoch": 7.77,
-      "learning_rate": 8.653179190751445e-05,
-      "loss": 0.0508,
+      "learning_rate": 8.6878612716763e-06,
+      "loss": 3.6264,
       "step": 1733
     },
     {
       "epoch": 7.78,
-      "learning_rate": 8.635838150289017e-05,
-      "loss": 0.043,
+      "learning_rate": 8.670520231213873e-06,
+      "loss": 3.5722,
       "step": 1734
     },
     {
       "epoch": 7.78,
-      "learning_rate": 8.618497109826589e-05,
-      "loss": 0.0354,
+      "learning_rate": 8.653179190751445e-06,
+      "loss": 3.5256,
       "step": 1735
     },
     {
       "epoch": 7.78,
-      "learning_rate": 8.601156069364162e-05,
-      "loss": 0.0435,
+      "learning_rate": 8.635838150289019e-06,
+      "loss": 3.6254,
       "step": 1736
     },
     {
       "epoch": 7.79,
-      "learning_rate": 8.583815028901733e-05,
-      "loss": 0.0427,
+      "learning_rate": 8.618497109826589e-06,
+      "loss": 3.5778,
       "step": 1737
     },
     {
       "epoch": 7.79,
-      "learning_rate": 8.566473988439306e-05,
-      "loss": 0.0411,
+      "learning_rate": 8.601156069364163e-06,
+      "loss": 3.4736,
       "step": 1738
     },
     {
       "epoch": 7.8,
-      "learning_rate": 8.549132947976878e-05,
-      "loss": 0.0423,
+      "learning_rate": 8.583815028901735e-06,
+      "loss": 3.4854,
       "step": 1739
     },
     {
       "epoch": 7.8,
-      "learning_rate": 8.53179190751445e-05,
-      "loss": 0.0406,
+      "learning_rate": 8.566473988439307e-06,
+      "loss": 3.4268,
       "step": 1740
     },
     {
       "epoch": 7.81,
-      "learning_rate": 8.514450867052023e-05,
-      "loss": 0.0399,
+      "learning_rate": 8.549132947976879e-06,
+      "loss": 3.4583,
       "step": 1741
     },
     {
       "epoch": 7.81,
-      "learning_rate": 8.497109826589596e-05,
-      "loss": 0.0419,
+      "learning_rate": 8.531791907514451e-06,
+      "loss": 3.4335,
       "step": 1742
     },
     {
       "epoch": 7.82,
-      "learning_rate": 8.479768786127167e-05,
-      "loss": 0.0363,
+      "learning_rate": 8.514450867052023e-06,
+      "loss": 3.4451,
       "step": 1743
     },
     {
       "epoch": 7.82,
-      "learning_rate": 8.46242774566474e-05,
-      "loss": 0.0392,
+      "learning_rate": 8.497109826589597e-06,
+      "loss": 3.42,
       "step": 1744
     },
     {
       "epoch": 7.83,
-      "learning_rate": 8.445086705202311e-05,
-      "loss": 0.0461,
+      "learning_rate": 8.479768786127167e-06,
+      "loss": 3.4913,
       "step": 1745
     },
     {
       "epoch": 7.83,
-      "learning_rate": 8.427745664739884e-05,
-      "loss": 0.0458,
+      "learning_rate": 8.462427745664741e-06,
+      "loss": 3.4534,
       "step": 1746
     },
     {
       "epoch": 7.83,
-      "learning_rate": 8.410404624277456e-05,
-      "loss": 0.0416,
+      "learning_rate": 8.445086705202311e-06,
+      "loss": 3.3673,
       "step": 1747
     },
     {
       "epoch": 7.84,
-      "learning_rate": 8.393063583815028e-05,
-      "loss": 0.036,
+      "learning_rate": 8.427745664739885e-06,
+      "loss": 3.3476,
       "step": 1748
     },
     {
       "epoch": 7.84,
-      "learning_rate": 8.3757225433526e-05,
-      "loss": 0.0414,
+      "learning_rate": 8.410404624277457e-06,
+      "loss": 3.327,
       "step": 1749
     },
     {
       "epoch": 7.85,
-      "learning_rate": 8.358381502890174e-05,
-      "loss": 0.0326,
+      "learning_rate": 8.39306358381503e-06,
+      "loss": 3.3339,
       "step": 1750
     },
     {
       "epoch": 7.85,
-      "learning_rate": 8.341040462427745e-05,
-      "loss": 0.0508,
+      "learning_rate": 8.375722543352601e-06,
+      "loss": 3.3453,
       "step": 1751
     },
     {
       "epoch": 7.86,
-      "learning_rate": 8.323699421965317e-05,
-      "loss": 0.0486,
+      "learning_rate": 8.358381502890175e-06,
+      "loss": 3.3937,
       "step": 1752
     },
     {
       "epoch": 7.86,
-      "learning_rate": 8.30635838150289e-05,
-      "loss": 0.0363,
+      "learning_rate": 8.341040462427746e-06,
+      "loss": 3.4472,
       "step": 1753
     },
     {
       "epoch": 7.87,
-      "learning_rate": 8.289017341040461e-05,
-      "loss": 0.0366,
+      "learning_rate": 8.323699421965318e-06,
+      "loss": 3.3535,
       "step": 1754
     },
     {
       "epoch": 7.87,
-      "learning_rate": 8.271676300578034e-05,
-      "loss": 0.0399,
+      "learning_rate": 8.30635838150289e-06,
+      "loss": 3.4178,
       "step": 1755
     },
     {
       "epoch": 7.87,
-      "learning_rate": 8.254335260115605e-05,
-      "loss": 0.0408,
+      "learning_rate": 8.289017341040462e-06,
+      "loss": 3.409,
       "step": 1756
     },
     {
       "epoch": 7.88,
-      "learning_rate": 8.236994219653178e-05,
-      "loss": 0.0324,
+      "learning_rate": 8.271676300578036e-06,
+      "loss": 3.3731,
       "step": 1757
     },
     {
       "epoch": 7.88,
-      "learning_rate": 8.21965317919075e-05,
-      "loss": 0.0344,
+      "learning_rate": 8.254335260115606e-06,
+      "loss": 3.3778,
       "step": 1758
     },
     {
       "epoch": 7.89,
-      "learning_rate": 8.202312138728322e-05,
-      "loss": 0.0545,
+      "learning_rate": 8.23699421965318e-06,
+      "loss": 3.4355,
       "step": 1759
     },
     {
       "epoch": 7.89,
-      "learning_rate": 8.184971098265895e-05,
-      "loss": 0.0309,
+      "learning_rate": 8.21965317919075e-06,
+      "loss": 3.4453,
       "step": 1760
     },
     {
       "epoch": 7.9,
-      "learning_rate": 8.167630057803468e-05,
-      "loss": 0.1026,
+      "learning_rate": 8.202312138728324e-06,
+      "loss": 3.4076,
       "step": 1761
     },
     {
       "epoch": 7.9,
-      "learning_rate": 8.150289017341039e-05,
-      "loss": 0.075,
+      "learning_rate": 8.184971098265896e-06,
+      "loss": 4.2028,
       "step": 1762
     },
     {
       "epoch": 7.91,
-      "learning_rate": 8.132947976878612e-05,
-      "loss": 0.0526,
+      "learning_rate": 8.167630057803468e-06,
+      "loss": 4.1265,
       "step": 1763
     },
     {
       "epoch": 7.91,
-      "learning_rate": 8.115606936416184e-05,
-      "loss": 0.064,
+      "learning_rate": 8.15028901734104e-06,
+      "loss": 4.1616,
       "step": 1764
     },
     {
       "epoch": 7.91,
-      "learning_rate": 8.098265895953756e-05,
-      "loss": 0.0562,
+      "learning_rate": 8.132947976878614e-06,
+      "loss": 4.0691,
       "step": 1765
     },
     {
       "epoch": 7.92,
-      "learning_rate": 8.080924855491328e-05,
-      "loss": 0.0543,
+      "learning_rate": 8.115606936416184e-06,
+      "loss": 3.998,
       "step": 1766
     },
     {
       "epoch": 7.92,
-      "learning_rate": 8.063583815028902e-05,
-      "loss": 0.0638,
+      "learning_rate": 8.098265895953758e-06,
+      "loss": 3.7823,
       "step": 1767
     },
     {
       "epoch": 7.93,
-      "learning_rate": 8.046242774566472e-05,
-      "loss": 0.0364,
+      "learning_rate": 8.080924855491328e-06,
+      "loss": 3.7425,
       "step": 1768
     },
     {
       "epoch": 7.93,
-      "learning_rate": 8.028901734104046e-05,
-      "loss": 0.0621,
+      "learning_rate": 8.063583815028902e-06,
+      "loss": 3.7305,
       "step": 1769
     },
     {
       "epoch": 7.94,
-      "learning_rate": 8.011560693641617e-05,
-      "loss": 0.0567,
+      "learning_rate": 8.046242774566474e-06,
+      "loss": 3.7151,
       "step": 1770
     },
     {
       "epoch": 7.94,
-      "learning_rate": 7.99421965317919e-05,
-      "loss": 0.0482,
+      "learning_rate": 8.028901734104046e-06,
+      "loss": 3.5602,
       "step": 1771
     },
     {
       "epoch": 7.95,
-      "learning_rate": 7.976878612716762e-05,
-      "loss": 0.0518,
+      "learning_rate": 8.011560693641618e-06,
+      "loss": 3.5218,
       "step": 1772
     },
     {
       "epoch": 7.95,
-      "learning_rate": 7.959537572254334e-05,
-      "loss": 0.0447,
+      "learning_rate": 7.99421965317919e-06,
+      "loss": 3.6544,
       "step": 1773
     },
     {
       "epoch": 7.96,
-      "learning_rate": 7.942196531791906e-05,
-      "loss": 0.0483,
+      "learning_rate": 7.976878612716762e-06,
+      "loss": 3.4446,
       "step": 1774
     },
     {
       "epoch": 7.96,
-      "learning_rate": 7.92485549132948e-05,
-      "loss": 0.0337,
+      "learning_rate": 7.959537572254336e-06,
+      "loss": 3.5295,
       "step": 1775
     },
     {
       "epoch": 7.96,
-      "learning_rate": 7.90751445086705e-05,
-      "loss": 0.0454,
+      "learning_rate": 7.942196531791907e-06,
+      "loss": 3.441,
       "step": 1776
     },
     {
       "epoch": 7.97,
-      "learning_rate": 7.890173410404624e-05,
-      "loss": 0.0472,
+      "learning_rate": 7.92485549132948e-06,
+      "loss": 3.4074,
       "step": 1777
     },
     {
       "epoch": 7.97,
-      "learning_rate": 7.872832369942196e-05,
-      "loss": 0.0442,
+      "learning_rate": 7.907514450867052e-06,
+      "loss": 3.4914,
       "step": 1778
     },
     {
       "epoch": 7.98,
-      "learning_rate": 7.855491329479768e-05,
-      "loss": 0.0333,
+      "learning_rate": 7.890173410404624e-06,
+      "loss": 3.2935,
       "step": 1779
     },
     {
       "epoch": 7.98,
-      "learning_rate": 7.83815028901734e-05,
-      "loss": 0.0467,
+      "learning_rate": 7.872832369942196e-06,
+      "loss": 3.4236,
       "step": 1780
     },
     {
       "epoch": 7.99,
-      "learning_rate": 7.820809248554913e-05,
-      "loss": 0.0412,
+      "learning_rate": 7.855491329479769e-06,
+      "loss": 3.3437,
       "step": 1781
     },
     {
       "epoch": 7.99,
-      "learning_rate": 7.803468208092484e-05,
-      "loss": 0.0464,
+      "learning_rate": 7.83815028901734e-06,
+      "loss": 3.378,
       "step": 1782
     },
     {
       "epoch": 8.0,
-      "learning_rate": 7.786127167630058e-05,
-      "loss": 0.0307,
+      "learning_rate": 7.820809248554914e-06,
+      "loss": 3.4117,
       "step": 1783
     },
     {
       "epoch": 8.0,
-      "learning_rate": 7.768786127167628e-05,
-      "loss": 0.058,
+      "learning_rate": 7.803468208092485e-06,
+      "loss": 3.4898,
       "step": 1784
     },
     {
       "epoch": 8.0,
-      "learning_rate": 7.751445086705202e-05,
-      "loss": 0.0672,
+      "learning_rate": 7.786127167630059e-06,
+      "loss": 4.1182,
       "step": 1785
     },
     {
       "epoch": 8.01,
-      "learning_rate": 7.734104046242774e-05,
-      "loss": 0.0566,
+      "learning_rate": 7.768786127167629e-06,
+      "loss": 3.9994,
       "step": 1786
     },
     {
       "epoch": 8.01,
-      "learning_rate": 7.716763005780346e-05,
-      "loss": 0.0577,
+      "learning_rate": 7.751445086705203e-06,
+      "loss": 4.0592,
       "step": 1787
     },
     {
       "epoch": 8.02,
-      "learning_rate": 7.699421965317918e-05,
-      "loss": 0.0516,
+      "learning_rate": 7.734104046242775e-06,
+      "loss": 3.9964,
       "step": 1788
     },
     {
       "epoch": 8.02,
-      "learning_rate": 7.682080924855491e-05,
-      "loss": 0.1426,
+      "learning_rate": 7.716763005780347e-06,
+      "loss": 4.0107,
       "step": 1789
     },
     {
       "epoch": 8.03,
-      "learning_rate": 7.664739884393062e-05,
-      "loss": 0.0358,
+      "learning_rate": 7.699421965317919e-06,
+      "loss": 3.931,
       "step": 1790
     },
     {
       "epoch": 8.03,
-      "learning_rate": 7.647398843930635e-05,
-      "loss": 0.0846,
+      "learning_rate": 7.682080924855493e-06,
+      "loss": 3.922,
       "step": 1791
     },
     {
       "epoch": 8.04,
-      "learning_rate": 7.630057803468207e-05,
-      "loss": 0.0666,
+      "learning_rate": 7.664739884393063e-06,
+      "loss": 3.8991,
       "step": 1792
     },
     {
       "epoch": 8.04,
-      "learning_rate": 7.61271676300578e-05,
-      "loss": 0.0505,
+      "learning_rate": 7.647398843930637e-06,
+      "loss": 3.8779,
       "step": 1793
     },
     {
       "epoch": 8.04,
-      "learning_rate": 7.595375722543352e-05,
-      "loss": 0.0439,
+      "learning_rate": 7.630057803468207e-06,
+      "loss": 3.8272,
       "step": 1794
     },
     {
       "epoch": 8.05,
-      "learning_rate": 7.578034682080925e-05,
-      "loss": 0.0355,
+      "learning_rate": 7.612716763005781e-06,
+      "loss": 3.7461,
       "step": 1795
     },
     {
       "epoch": 8.05,
-      "learning_rate": 7.560693641618496e-05,
-      "loss": 0.0469,
+      "learning_rate": 7.595375722543352e-06,
+      "loss": 3.7225,
       "step": 1796
     },
     {
       "epoch": 8.06,
-      "learning_rate": 7.543352601156069e-05,
-      "loss": 0.0565,
+      "learning_rate": 7.578034682080925e-06,
+      "loss": 3.6657,
       "step": 1797
     },
     {
       "epoch": 8.06,
-      "learning_rate": 7.52601156069364e-05,
-      "loss": 0.0368,
+      "learning_rate": 7.560693641618497e-06,
+      "loss": 3.6324,
       "step": 1798
     },
     {
       "epoch": 8.07,
-      "learning_rate": 7.508670520231213e-05,
-      "loss": 0.0524,
+      "learning_rate": 7.54335260115607e-06,
+      "loss": 3.6222,
       "step": 1799
     },
     {
       "epoch": 8.07,
-      "learning_rate": 7.491329479768785e-05,
-      "loss": 0.0369,
+      "learning_rate": 7.526011560693641e-06,
+      "loss": 3.5819,
       "step": 1800
     },
     {
       "epoch": 8.08,
-      "learning_rate": 7.473988439306357e-05,
-      "loss": 0.0475,
+      "learning_rate": 7.508670520231214e-06,
+      "loss": 3.6389,
       "step": 1801
     },
     {
       "epoch": 8.08,
-      "learning_rate": 7.45664739884393e-05,
-      "loss": 0.049,
+      "learning_rate": 7.491329479768786e-06,
+      "loss": 3.5601,
       "step": 1802
     },
     {
       "epoch": 8.09,
-      "learning_rate": 7.439306358381502e-05,
-      "loss": 0.0408,
+      "learning_rate": 7.473988439306358e-06,
+      "loss": 3.5371,
       "step": 1803
     },
     {
       "epoch": 8.09,
-      "learning_rate": 7.421965317919074e-05,
-      "loss": 0.0297,
+      "learning_rate": 7.45664739884393e-06,
+      "loss": 3.4591,
       "step": 1804
     },
     {
       "epoch": 8.09,
-      "learning_rate": 7.404624277456646e-05,
-      "loss": 0.099,
+      "learning_rate": 7.4393063583815024e-06,
+      "loss": 3.4749,
       "step": 1805
     },
     {
       "epoch": 8.1,
-      "learning_rate": 7.387283236994219e-05,
-      "loss": 0.0438,
+      "learning_rate": 7.421965317919075e-06,
+      "loss": 3.4705,
       "step": 1806
     },
     {
       "epoch": 8.1,
-      "learning_rate": 7.369942196531791e-05,
-      "loss": 0.1455,
+      "learning_rate": 7.404624277456647e-06,
+      "loss": 3.4022,
       "step": 1807
     },
     {
       "epoch": 8.11,
-      "learning_rate": 7.352601156069363e-05,
-      "loss": 0.0296,
+      "learning_rate": 7.3872832369942195e-06,
+      "loss": 3.4355,
       "step": 1808
     },
     {
       "epoch": 8.11,
-      "learning_rate": 7.335260115606935e-05,
-      "loss": 0.0297,
+      "learning_rate": 7.3699421965317915e-06,
+      "loss": 3.5067,
       "step": 1809
     },
     {
       "epoch": 8.12,
-      "learning_rate": 7.317919075144507e-05,
-      "loss": 0.0351,
+      "learning_rate": 7.352601156069364e-06,
+      "loss": 3.4097,
       "step": 1810
     },
     {
       "epoch": 8.12,
-      "learning_rate": 7.30057803468208e-05,
-      "loss": 0.1393,
+      "learning_rate": 7.3352601156069365e-06,
+      "loss": 3.4438,
       "step": 1811
     },
     {
       "epoch": 8.13,
-      "learning_rate": 7.30057803468208e-05,
-      "loss": 0.1093,
+      "learning_rate": 7.317919075144509e-06,
+      "loss": 3.3129,
       "step": 1812
     },
     {
       "epoch": 8.13,
-      "learning_rate": 7.30057803468208e-05,
-      "loss": 0.08,
+      "learning_rate": 7.300578034682081e-06,
+      "loss": 3.3894,
       "step": 1813
     },
     {
       "epoch": 8.13,
-      "learning_rate": 7.283236994219653e-05,
-      "loss": 0.0995,
+      "learning_rate": 7.283236994219653e-06,
+      "loss": 3.3518,
       "step": 1814
     },
     {
       "epoch": 8.14,
-      "learning_rate": 7.265895953757225e-05,
-      "loss": 0.0839,
+      "learning_rate": 7.265895953757226e-06,
+      "loss": 3.3617,
       "step": 1815
     },
     {
       "epoch": 8.14,
-      "learning_rate": 7.248554913294797e-05,
-      "loss": 0.2315,
+      "learning_rate": 7.248554913294798e-06,
+      "loss": 3.3222,
       "step": 1816
     },
     {
       "epoch": 8.15,
-      "learning_rate": 7.231213872832369e-05,
-      "loss": 0.2594,
+      "learning_rate": 7.23121387283237e-06,
+      "loss": 3.2778,
       "step": 1817
     },
     {
       "epoch": 8.15,
-      "learning_rate": 7.213872832369941e-05,
-      "loss": 0.1639,
+      "learning_rate": 7.213872832369942e-06,
+      "loss": 3.2908,
       "step": 1818
     },
     {
       "epoch": 8.16,
-      "learning_rate": 7.196531791907513e-05,
-      "loss": 0.1329,
+      "learning_rate": 7.196531791907515e-06,
+      "loss": 3.3253,
       "step": 1819
     },
     {
       "epoch": 8.16,
-      "learning_rate": 7.179190751445085e-05,
-      "loss": 0.1424,
+      "learning_rate": 7.179190751445087e-06,
+      "loss": 3.3726,
       "step": 1820
     },
     {
       "epoch": 8.17,
-      "learning_rate": 7.161849710982659e-05,
-      "loss": 0.3705,
+      "learning_rate": 7.161849710982659e-06,
+      "loss": 3.3145,
       "step": 1821
     },
     {
       "epoch": 8.17,
-      "learning_rate": 7.144508670520231e-05,
-      "loss": 0.4238,
+      "learning_rate": 7.144508670520231e-06,
+      "loss": 3.3403,
       "step": 1822
     },
     {
       "epoch": 8.17,
-      "learning_rate": 7.127167630057803e-05,
-      "loss": 0.3596,
+      "learning_rate": 7.127167630057803e-06,
+      "loss": 3.3802,
       "step": 1823
     },
     {
       "epoch": 8.18,
-      "learning_rate": 7.109826589595375e-05,
-      "loss": 0.201,
+      "learning_rate": 7.109826589595376e-06,
+      "loss": 3.2439,
       "step": 1824
     },
     {
       "epoch": 8.18,
-      "learning_rate": 7.092485549132947e-05,
-      "loss": 0.184,
+      "learning_rate": 7.092485549132948e-06,
+      "loss": 3.3549,
       "step": 1825
     },
     {
       "epoch": 8.19,
-      "learning_rate": 7.075144508670519e-05,
-      "loss": 0.219,
+      "learning_rate": 7.07514450867052e-06,
+      "loss": 3.2981,
       "step": 1826
     },
     {
       "epoch": 8.19,
-      "learning_rate": 7.057803468208091e-05,
-      "loss": 0.2169,
+      "learning_rate": 7.057803468208092e-06,
+      "loss": 3.277,
       "step": 1827
     },
     {
       "epoch": 8.2,
-      "learning_rate": 7.040462427745665e-05,
-      "loss": 0.0858,
+      "learning_rate": 7.040462427745665e-06,
+      "loss": 3.343,
       "step": 1828
     },
     {
       "epoch": 8.2,
-      "learning_rate": 7.023121387283237e-05,
-      "loss": 0.1802,
+      "learning_rate": 7.023121387283237e-06,
+      "loss": 3.38,
       "step": 1829
     },
     {
       "epoch": 8.21,
-      "learning_rate": 7.005780346820809e-05,
-      "loss": 0.0855,
+      "learning_rate": 7.005780346820809e-06,
+      "loss": 3.3683,
       "step": 1830
     },
     {
       "epoch": 8.21,
-      "learning_rate": 6.988439306358381e-05,
-      "loss": 0.1098,
+      "learning_rate": 6.988439306358381e-06,
+      "loss": 3.2754,
       "step": 1831
     },
     {
       "epoch": 8.22,
-      "learning_rate": 6.971098265895953e-05,
-      "loss": 0.1523,
+      "learning_rate": 6.971098265895954e-06,
+      "loss": 3.3122,
       "step": 1832
     },
     {
       "epoch": 8.22,
-      "learning_rate": 6.953757225433525e-05,
-      "loss": 0.1435,
+      "learning_rate": 6.953757225433526e-06,
+      "loss": 3.4639,
       "step": 1833
     },
     {
       "epoch": 8.22,
-      "learning_rate": 6.936416184971097e-05,
-      "loss": 0.1203,
+      "learning_rate": 6.936416184971098e-06,
+      "loss": 3.1142,
       "step": 1834
     },
     {
       "epoch": 8.23,
-      "learning_rate": 6.91907514450867e-05,
-      "loss": 0.2379,
+      "learning_rate": 6.91907514450867e-06,
+      "loss": 4.1776,
       "step": 1835
     },
     {
       "epoch": 8.23,
-      "learning_rate": 6.901734104046242e-05,
-      "loss": 0.1833,
+      "learning_rate": 6.9017341040462425e-06,
+      "loss": 4.1406,
       "step": 1836
     },
     {
       "epoch": 8.24,
-      "learning_rate": 6.884393063583815e-05,
-      "loss": 0.1643,
+      "learning_rate": 6.884393063583815e-06,
+      "loss": 4.1619,
       "step": 1837
     },
     {
       "epoch": 8.24,
-      "learning_rate": 6.867052023121387e-05,
-      "loss": 0.1469,
+      "learning_rate": 6.8670520231213874e-06,
+      "loss": 4.1597,
       "step": 1838
     },
     {
       "epoch": 8.25,
-      "learning_rate": 6.849710982658959e-05,
-      "loss": 0.0913,
+      "learning_rate": 6.8497109826589595e-06,
+      "loss": 4.1007,
       "step": 1839
     },
     {
       "epoch": 8.25,
-      "learning_rate": 6.832369942196531e-05,
-      "loss": 0.0403,
+      "learning_rate": 6.832369942196532e-06,
+      "loss": 4.0917,
       "step": 1840
     },
     {
       "epoch": 8.26,
-      "learning_rate": 6.815028901734103e-05,
-      "loss": 0.096,
+      "learning_rate": 6.8150289017341045e-06,
+      "loss": 4.05,
       "step": 1841
     },
     {
       "epoch": 8.26,
-      "learning_rate": 6.797687861271676e-05,
-      "loss": 0.1088,
+      "learning_rate": 6.7976878612716766e-06,
+      "loss": 4.0253,
       "step": 1842
     },
     {
       "epoch": 8.26,
-      "learning_rate": 6.780346820809248e-05,
-      "loss": 0.2121,
+      "learning_rate": 6.780346820809249e-06,
+      "loss": 4.0109,
       "step": 1843
     },
     {
       "epoch": 8.27,
-      "learning_rate": 6.76300578034682e-05,
-      "loss": 0.1506,
+      "learning_rate": 6.763005780346821e-06,
+      "loss": 3.9088,
       "step": 1844
     },
     {
       "epoch": 8.27,
-      "learning_rate": 6.745664739884392e-05,
-      "loss": 0.0472,
+      "learning_rate": 6.745664739884394e-06,
+      "loss": 3.8613,
       "step": 1845
     },
     {
       "epoch": 8.28,
-      "learning_rate": 6.728323699421964e-05,
-      "loss": 0.0873,
+      "learning_rate": 6.728323699421966e-06,
+      "loss": 3.7595,
       "step": 1846
     },
     {
       "epoch": 8.28,
-      "learning_rate": 6.710982658959537e-05,
-      "loss": 0.1552,
+      "learning_rate": 6.710982658959538e-06,
+      "loss": 3.8061,
       "step": 1847
     },
     {
       "epoch": 8.29,
-      "learning_rate": 6.693641618497109e-05,
-      "loss": 0.047,
+      "learning_rate": 6.69364161849711e-06,
+      "loss": 3.7033,
       "step": 1848
     },
     {
       "epoch": 8.29,
-      "learning_rate": 6.676300578034682e-05,
-      "loss": 0.0831,
+      "learning_rate": 6.676300578034682e-06,
+      "loss": 3.5736,
       "step": 1849
     },
     {
       "epoch": 8.3,
-      "learning_rate": 6.658959537572254e-05,
-      "loss": 0.0891,
+      "learning_rate": 6.658959537572255e-06,
+      "loss": 3.5619,
       "step": 1850
     },
     {
       "epoch": 8.3,
-      "learning_rate": 6.641618497109826e-05,
-      "loss": 0.0929,
+      "learning_rate": 6.641618497109827e-06,
+      "loss": 3.623,
       "step": 1851
     },
     {
       "epoch": 8.3,
-      "learning_rate": 6.624277456647398e-05,
-      "loss": 0.0883,
+      "learning_rate": 6.624277456647399e-06,
+      "loss": 3.5274,
       "step": 1852
     },
     {
       "epoch": 8.31,
-      "learning_rate": 6.60693641618497e-05,
-      "loss": 0.0795,
+      "learning_rate": 6.606936416184971e-06,
+      "loss": 3.476,
       "step": 1853
     },
     {
       "epoch": 8.31,
-      "learning_rate": 6.589595375722542e-05,
-      "loss": 0.1435,
+      "learning_rate": 6.589595375722544e-06,
+      "loss": 3.4913,
       "step": 1854
     },
     {
       "epoch": 8.32,
-      "learning_rate": 6.572254335260114e-05,
-      "loss": 0.1741,
+      "learning_rate": 6.572254335260116e-06,
+      "loss": 3.4383,
       "step": 1855
     },
     {
       "epoch": 8.32,
-      "learning_rate": 6.554913294797688e-05,
-      "loss": 0.0961,
+      "learning_rate": 6.554913294797688e-06,
+      "loss": 3.5418,
       "step": 1856
     },
     {
       "epoch": 8.33,
-      "learning_rate": 6.53757225433526e-05,
-      "loss": 0.1966,
+      "learning_rate": 6.53757225433526e-06,
+      "loss": 3.4382,
       "step": 1857
     },
     {
       "epoch": 8.33,
-      "learning_rate": 6.520231213872832e-05,
-      "loss": 0.1273,
+      "learning_rate": 6.520231213872832e-06,
+      "loss": 3.4381,
       "step": 1858
     },
     {
       "epoch": 8.34,
-      "learning_rate": 6.502890173410404e-05,
-      "loss": 0.0822,
+      "learning_rate": 6.502890173410405e-06,
+      "loss": 3.4033,
       "step": 1859
     },
     {
       "epoch": 8.34,
-      "learning_rate": 6.485549132947976e-05,
-      "loss": 0.0902,
+      "learning_rate": 6.485549132947977e-06,
+      "loss": 3.5293,
       "step": 1860
     },
     {
       "epoch": 8.35,
-      "learning_rate": 6.468208092485548e-05,
-      "loss": 0.0709,
+      "learning_rate": 6.468208092485549e-06,
+      "loss": 3.4931,
       "step": 1861
     },
     {
       "epoch": 8.35,
-      "learning_rate": 6.45086705202312e-05,
-      "loss": 0.0338,
+      "learning_rate": 6.450867052023121e-06,
+      "loss": 3.4443,
       "step": 1862
     },
     {
       "epoch": 8.35,
-      "learning_rate": 6.433526011560694e-05,
-      "loss": 0.0312,
+      "learning_rate": 6.433526011560694e-06,
+      "loss": 3.3363,
       "step": 1863
     },
     {
       "epoch": 8.36,
-      "learning_rate": 6.416184971098266e-05,
-      "loss": 0.0409,
+      "learning_rate": 6.416184971098266e-06,
+      "loss": 3.4288,
       "step": 1864
     },
     {
       "epoch": 8.36,
-      "learning_rate": 6.398843930635838e-05,
-      "loss": 0.0308,
+      "learning_rate": 6.398843930635838e-06,
+      "loss": 3.3535,
       "step": 1865
     },
     {
       "epoch": 8.37,
-      "learning_rate": 6.38150289017341e-05,
-      "loss": 0.0323,
+      "learning_rate": 6.3815028901734104e-06,
+      "loss": 3.258,
       "step": 1866
     },
     {
       "epoch": 8.37,
-      "learning_rate": 6.364161849710982e-05,
-      "loss": 0.0841,
+      "learning_rate": 6.364161849710983e-06,
+      "loss": 3.3366,
       "step": 1867
     },
     {
       "epoch": 8.38,
-      "learning_rate": 6.346820809248554e-05,
-      "loss": 0.086,
+      "learning_rate": 6.346820809248555e-06,
+      "loss": 3.4033,
       "step": 1868
     },
     {
       "epoch": 8.38,
-      "learning_rate": 6.329479768786126e-05,
-      "loss": 0.0813,
+      "learning_rate": 6.3294797687861275e-06,
+      "loss": 3.3799,
       "step": 1869
     },
     {
       "epoch": 8.39,
-      "learning_rate": 6.3121387283237e-05,
-      "loss": 0.0751,
+      "learning_rate": 6.3121387283236995e-06,
+      "loss": 3.2493,
       "step": 1870
     },
     {
       "epoch": 8.39,
-      "learning_rate": 6.294797687861272e-05,
-      "loss": 0.0999,
+      "learning_rate": 6.294797687861272e-06,
+      "loss": 3.2931,
       "step": 1871
     },
     {
       "epoch": 8.39,
-      "learning_rate": 6.277456647398844e-05,
-      "loss": 0.0301,
+      "learning_rate": 6.2774566473988445e-06,
+      "loss": 3.3144,
       "step": 1872
     },
     {
       "epoch": 8.4,
-      "learning_rate": 6.260115606936416e-05,
-      "loss": 0.038,
+      "learning_rate": 6.260115606936417e-06,
+      "loss": 3.2651,
       "step": 1873
     },
     {
       "epoch": 8.4,
-      "learning_rate": 6.242774566473988e-05,
-      "loss": 0.0383,
+      "learning_rate": 6.242774566473989e-06,
+      "loss": 3.3027,
       "step": 1874
     },
     {
       "epoch": 8.41,
-      "learning_rate": 6.22543352601156e-05,
-      "loss": 0.0304,
+      "learning_rate": 6.225433526011561e-06,
+      "loss": 3.2677,
       "step": 1875
     },
     {
       "epoch": 8.41,
-      "learning_rate": 6.208092485549133e-05,
-      "loss": 0.0909,
+      "learning_rate": 6.208092485549134e-06,
+      "loss": 3.5069,
       "step": 1876
     },
     {
       "epoch": 8.42,
-      "learning_rate": 6.190751445086705e-05,
-      "loss": 0.0432,
+      "learning_rate": 6.190751445086706e-06,
+      "loss": 3.2993,
       "step": 1877
     },
     {
       "epoch": 8.42,
-      "learning_rate": 6.173410404624277e-05,
-      "loss": 0.0992,
+      "learning_rate": 6.173410404624278e-06,
+      "loss": 3.3499,
       "step": 1878
     },
     {
       "epoch": 8.43,
-      "learning_rate": 6.15606936416185e-05,
-      "loss": 0.0197,
+      "learning_rate": 6.15606936416185e-06,
+      "loss": 3.1909,
       "step": 1879
     },
     {
       "epoch": 8.43,
-      "learning_rate": 6.138728323699422e-05,
-      "loss": 0.0453,
+      "learning_rate": 6.138728323699423e-06,
+      "loss": 3.372,
       "step": 1880
     },
     {
       "epoch": 8.43,
-      "learning_rate": 6.121387283236994e-05,
-      "loss": 0.0355,
+      "learning_rate": 6.121387283236995e-06,
+      "loss": 3.2667,
       "step": 1881
     },
     {
       "epoch": 8.44,
-      "learning_rate": 6.104046242774566e-05,
-      "loss": 0.0304,
+      "learning_rate": 6.104046242774567e-06,
+      "loss": 3.3584,
       "step": 1882
     },
     {
       "epoch": 8.44,
-      "learning_rate": 6.0867052023121385e-05,
-      "loss": 0.0871,
+      "learning_rate": 6.086705202312139e-06,
+      "loss": 3.3955,
       "step": 1883
     },
     {
       "epoch": 8.45,
-      "learning_rate": 6.0693641618497105e-05,
-      "loss": 0.0741,
+      "learning_rate": 6.069364161849711e-06,
+      "loss": 3.2677,
       "step": 1884
     },
     {
       "epoch": 8.45,
-      "learning_rate": 6.052023121387283e-05,
-      "loss": 0.0567,
+      "learning_rate": 6.052023121387284e-06,
+      "loss": 4.1139,
       "step": 1885
     },
     {
       "epoch": 8.46,
-      "learning_rate": 6.034682080924855e-05,
-      "loss": 0.1103,
+      "learning_rate": 6.034682080924856e-06,
+      "loss": 4.017,
       "step": 1886
     },
     {
       "epoch": 8.46,
-      "learning_rate": 6.0173410404624274e-05,
-      "loss": 0.0536,
+      "learning_rate": 6.017341040462428e-06,
+      "loss": 4.1157,
       "step": 1887
     },
     {
       "epoch": 8.47,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 0.0413,
+      "learning_rate": 6e-06,
+      "loss": 4.0058,
       "step": 1888
     },
     {
       "epoch": 8.47,
-      "learning_rate": 5.982658959537572e-05,
-      "loss": 0.0408,
+      "learning_rate": 5.982658959537573e-06,
+      "loss": 3.9954,
       "step": 1889
     },
     {
       "epoch": 8.48,
-      "learning_rate": 5.965317919075144e-05,
-      "loss": 0.0403,
+      "learning_rate": 5.965317919075145e-06,
+      "loss": 4.0446,
       "step": 1890
     },
     {
       "epoch": 8.48,
-      "learning_rate": 5.9479768786127164e-05,
-      "loss": 0.042,
+      "learning_rate": 5.947976878612717e-06,
+      "loss": 4.0586,
       "step": 1891
     },
     {
       "epoch": 8.48,
-      "learning_rate": 5.930635838150289e-05,
-      "loss": 0.0374,
+      "learning_rate": 5.930635838150289e-06,
+      "loss": 4.0104,
       "step": 1892
     },
     {
       "epoch": 8.49,
-      "learning_rate": 5.913294797687861e-05,
-      "loss": 0.0483,
+      "learning_rate": 5.913294797687862e-06,
+      "loss": 3.9089,
       "step": 1893
     },
     {
       "epoch": 8.49,
-      "learning_rate": 5.895953757225433e-05,
-      "loss": 0.0782,
+      "learning_rate": 5.895953757225434e-06,
+      "loss": 3.8737,
       "step": 1894
     },
     {
       "epoch": 8.5,
-      "learning_rate": 5.878612716763006e-05,
-      "loss": 0.087,
+      "learning_rate": 5.878612716763006e-06,
+      "loss": 3.8226,
       "step": 1895
     },
     {
       "epoch": 8.5,
-      "learning_rate": 5.861271676300578e-05,
-      "loss": 0.0324,
+      "learning_rate": 5.861271676300578e-06,
+      "loss": 3.7705,
       "step": 1896
     },
     {
       "epoch": 8.51,
-      "learning_rate": 5.84393063583815e-05,
-      "loss": 0.0401,
+      "learning_rate": 5.8439306358381505e-06,
+      "loss": 3.8212,
       "step": 1897
     },
     {
       "epoch": 8.51,
-      "learning_rate": 5.8265895953757215e-05,
-      "loss": 0.0318,
+      "learning_rate": 5.8265895953757225e-06,
+      "loss": 3.6777,
       "step": 1898
     },
     {
       "epoch": 8.52,
-      "learning_rate": 5.8092485549132936e-05,
-      "loss": 0.0416,
+      "learning_rate": 5.809248554913295e-06,
+      "loss": 3.663,
       "step": 1899
     },
     {
       "epoch": 8.52,
-      "learning_rate": 5.791907514450866e-05,
-      "loss": 0.0305,
+      "learning_rate": 5.791907514450867e-06,
+      "loss": 3.5272,
       "step": 1900
     },
     {
       "epoch": 8.52,
-      "learning_rate": 5.7745664739884384e-05,
-      "loss": 0.0404,
+      "learning_rate": 5.774566473988439e-06,
+      "loss": 3.4951,
       "step": 1901
     },
     {
       "epoch": 8.53,
-      "learning_rate": 5.7572254335260105e-05,
-      "loss": 0.0756,
+      "learning_rate": 5.757225433526012e-06,
+      "loss": 3.5092,
       "step": 1902
     },
     {
       "epoch": 8.53,
-      "learning_rate": 5.739884393063583e-05,
-      "loss": 0.0329,
+      "learning_rate": 5.739884393063584e-06,
+      "loss": 3.5343,
       "step": 1903
     },
     {
       "epoch": 8.54,
-      "learning_rate": 5.722543352601155e-05,
-      "loss": 0.0344,
+      "learning_rate": 5.722543352601156e-06,
+      "loss": 3.4855,
       "step": 1904
     },
     {
       "epoch": 8.54,
-      "learning_rate": 5.705202312138727e-05,
-      "loss": 0.0259,
+      "learning_rate": 5.705202312138728e-06,
+      "loss": 3.4923,
       "step": 1905
     },
     {
       "epoch": 8.55,
-      "learning_rate": 5.6878612716762994e-05,
-      "loss": 0.0342,
+      "learning_rate": 5.687861271676301e-06,
+      "loss": 3.5079,
       "step": 1906
     },
     {
       "epoch": 8.55,
-      "learning_rate": 5.670520231213872e-05,
-      "loss": 0.0425,
+      "learning_rate": 5.670520231213873e-06,
+      "loss": 3.4391,
       "step": 1907
     },
     {
       "epoch": 8.56,
-      "learning_rate": 5.653179190751444e-05,
-      "loss": 0.0411,
+      "learning_rate": 5.653179190751445e-06,
+      "loss": 3.4836,
       "step": 1908
     },
     {
       "epoch": 8.56,
-      "learning_rate": 5.635838150289016e-05,
-      "loss": 0.0291,
+      "learning_rate": 5.635838150289017e-06,
+      "loss": 3.4855,
       "step": 1909
     },
     {
       "epoch": 8.57,
-      "learning_rate": 5.618497109826589e-05,
-      "loss": 0.0399,
+      "learning_rate": 5.618497109826589e-06,
+      "loss": 3.4756,
       "step": 1910
     },
     {
       "epoch": 8.57,
-      "learning_rate": 5.601156069364161e-05,
-      "loss": 0.0319,
+      "learning_rate": 5.601156069364162e-06,
+      "loss": 3.4713,
       "step": 1911
     },
     {
       "epoch": 8.57,
-      "learning_rate": 5.583815028901733e-05,
-      "loss": 0.0314,
+      "learning_rate": 5.583815028901734e-06,
+      "loss": 3.372,
       "step": 1912
     },
     {
       "epoch": 8.58,
-      "learning_rate": 5.566473988439306e-05,
-      "loss": 0.0389,
+      "learning_rate": 5.566473988439306e-06,
+      "loss": 3.4141,
       "step": 1913
     },
     {
       "epoch": 8.58,
-      "learning_rate": 5.549132947976878e-05,
-      "loss": 0.034,
+      "learning_rate": 5.549132947976878e-06,
+      "loss": 3.3146,
       "step": 1914
     },
     {
       "epoch": 8.59,
-      "learning_rate": 5.53179190751445e-05,
-      "loss": 0.0278,
+      "learning_rate": 5.531791907514451e-06,
+      "loss": 3.4147,
       "step": 1915
     },
     {
       "epoch": 8.59,
-      "learning_rate": 5.514450867052022e-05,
-      "loss": 0.0305,
+      "learning_rate": 5.514450867052023e-06,
+      "loss": 3.346,
       "step": 1916
     },
     {
       "epoch": 8.6,
-      "learning_rate": 5.497109826589595e-05,
-      "loss": 0.0461,
+      "learning_rate": 5.497109826589595e-06,
+      "loss": 3.3628,
       "step": 1917
     },
     {
       "epoch": 8.6,
-      "learning_rate": 5.479768786127167e-05,
-      "loss": 0.0264,
+      "learning_rate": 5.479768786127167e-06,
+      "loss": 3.289,
       "step": 1918
     },
     {
       "epoch": 8.61,
-      "learning_rate": 5.462427745664739e-05,
-      "loss": 0.0278,
+      "learning_rate": 5.46242774566474e-06,
+      "loss": 3.2644,
       "step": 1919
     },
     {
       "epoch": 8.61,
-      "learning_rate": 5.445086705202312e-05,
-      "loss": 0.0276,
+      "learning_rate": 5.445086705202312e-06,
+      "loss": 3.3341,
       "step": 1920
     },
     {
       "epoch": 8.61,
-      "learning_rate": 5.427745664739884e-05,
-      "loss": 0.0338,
+      "learning_rate": 5.427745664739884e-06,
+      "loss": 3.3634,
       "step": 1921
     },
     {
       "epoch": 8.62,
-      "learning_rate": 5.410404624277456e-05,
-      "loss": 0.028,
+      "learning_rate": 5.410404624277456e-06,
+      "loss": 3.2382,
       "step": 1922
     },
     {
       "epoch": 8.62,
-      "learning_rate": 5.393063583815028e-05,
-      "loss": 0.0304,
+      "learning_rate": 5.3930635838150285e-06,
+      "loss": 3.2531,
       "step": 1923
     },
     {
       "epoch": 8.63,
-      "learning_rate": 5.375722543352601e-05,
-      "loss": 0.0405,
+      "learning_rate": 5.375722543352601e-06,
+      "loss": 3.3276,
       "step": 1924
     },
     {
       "epoch": 8.63,
-      "learning_rate": 5.358381502890173e-05,
-      "loss": 0.0265,
+      "learning_rate": 5.3583815028901734e-06,
+      "loss": 3.3509,
       "step": 1925
     },
     {
       "epoch": 8.64,
-      "learning_rate": 5.341040462427745e-05,
-      "loss": 0.0289,
+      "learning_rate": 5.3410404624277455e-06,
+      "loss": 3.224,
       "step": 1926
     },
     {
       "epoch": 8.64,
-      "learning_rate": 5.3236994219653176e-05,
-      "loss": 0.0272,
+      "learning_rate": 5.3236994219653176e-06,
+      "loss": 3.2597,
       "step": 1927
     },
     {
       "epoch": 8.65,
-      "learning_rate": 5.3063583815028896e-05,
-      "loss": 0.0278,
+      "learning_rate": 5.3063583815028905e-06,
+      "loss": 3.3581,
       "step": 1928
     },
     {
       "epoch": 8.65,
-      "learning_rate": 5.289017341040462e-05,
-      "loss": 0.0364,
+      "learning_rate": 5.2890173410404626e-06,
+      "loss": 3.4059,
       "step": 1929
     },
     {
       "epoch": 8.65,
-      "learning_rate": 5.271676300578034e-05,
-      "loss": 0.0183,
+      "learning_rate": 5.271676300578035e-06,
+      "loss": 3.3169,
       "step": 1930
     },
     {
       "epoch": 8.66,
-      "learning_rate": 5.2543352601156065e-05,
-      "loss": 0.03,
+      "learning_rate": 5.254335260115607e-06,
+      "loss": 3.2648,
       "step": 1931
     },
     {
       "epoch": 8.66,
-      "learning_rate": 5.2369942196531786e-05,
-      "loss": 0.0345,
+      "learning_rate": 5.23699421965318e-06,
+      "loss": 3.4101,
       "step": 1932
     },
     {
       "epoch": 8.67,
-      "learning_rate": 5.2196531791907507e-05,
-      "loss": 0.0346,
+      "learning_rate": 5.219653179190752e-06,
+      "loss": 3.242,
       "step": 1933
     },
     {
       "epoch": 8.67,
-      "learning_rate": 5.2023121387283234e-05,
-      "loss": 0.0374,
+      "learning_rate": 5.202312138728324e-06,
+      "loss": 3.338,
       "step": 1934
     },
     {
       "epoch": 8.68,
-      "learning_rate": 5.1849710982658955e-05,
-      "loss": 0.0673,
+      "learning_rate": 5.184971098265896e-06,
+      "loss": 4.1682,
       "step": 1935
     },
     {
       "epoch": 8.68,
-      "learning_rate": 5.1676300578034675e-05,
-      "loss": 0.0524,
+      "learning_rate": 5.167630057803468e-06,
+      "loss": 4.1346,
       "step": 1936
     },
     {
       "epoch": 8.69,
-      "learning_rate": 5.1502890173410396e-05,
-      "loss": 0.0417,
+      "learning_rate": 5.150289017341041e-06,
+      "loss": 4.1245,
       "step": 1937
     },
     {
       "epoch": 8.69,
-      "learning_rate": 5.1329479768786124e-05,
-      "loss": 0.0507,
+      "learning_rate": 5.132947976878613e-06,
+      "loss": 4.052,
       "step": 1938
     },
     {
       "epoch": 8.7,
-      "learning_rate": 5.1156069364161844e-05,
-      "loss": 0.0441,
+      "learning_rate": 5.115606936416185e-06,
+      "loss": 3.9706,
       "step": 1939
     },
     {
       "epoch": 8.7,
-      "learning_rate": 5.0982658959537565e-05,
-      "loss": 0.0386,
+      "learning_rate": 5.098265895953757e-06,
+      "loss": 4.0667,
       "step": 1940
     },
     {
       "epoch": 8.7,
-      "learning_rate": 5.080924855491329e-05,
-      "loss": 0.0389,
+      "learning_rate": 5.08092485549133e-06,
+      "loss": 3.9514,
       "step": 1941
     },
     {
       "epoch": 8.71,
-      "learning_rate": 5.063583815028901e-05,
-      "loss": 0.0461,
+      "learning_rate": 5.063583815028902e-06,
+      "loss": 3.9641,
       "step": 1942
     },
     {
       "epoch": 8.71,
-      "learning_rate": 5.0462427745664734e-05,
-      "loss": 0.0384,
+      "learning_rate": 5.046242774566474e-06,
+      "loss": 3.9298,
       "step": 1943
     },
     {
       "epoch": 8.72,
-      "learning_rate": 5.028901734104046e-05,
-      "loss": 0.034,
+      "learning_rate": 5.028901734104046e-06,
+      "loss": 3.8935,
       "step": 1944
     },
     {
       "epoch": 8.72,
-      "learning_rate": 5.011560693641618e-05,
-      "loss": 0.029,
+      "learning_rate": 5.011560693641619e-06,
+      "loss": 3.7538,
       "step": 1945
     },
     {
       "epoch": 8.73,
-      "learning_rate": 4.99421965317919e-05,
-      "loss": 0.0438,
+      "learning_rate": 4.994219653179191e-06,
+      "loss": 3.7974,
       "step": 1946
     },
     {
       "epoch": 8.73,
-      "learning_rate": 4.976878612716762e-05,
-      "loss": 0.0402,
+      "learning_rate": 4.976878612716763e-06,
+      "loss": 3.6656,
       "step": 1947
     },
     {
       "epoch": 8.74,
-      "learning_rate": 4.959537572254335e-05,
-      "loss": 0.034,
+      "learning_rate": 4.959537572254335e-06,
+      "loss": 3.6146,
       "step": 1948
     },
     {
       "epoch": 8.74,
-      "learning_rate": 4.942196531791907e-05,
-      "loss": 0.0376,
+      "learning_rate": 4.942196531791907e-06,
+      "loss": 3.6184,
       "step": 1949
     },
     {
       "epoch": 8.74,
-      "learning_rate": 4.924855491329479e-05,
-      "loss": 0.03,
+      "learning_rate": 4.92485549132948e-06,
+      "loss": 3.574,
       "step": 1950
     },
     {
       "epoch": 8.75,
-      "learning_rate": 4.907514450867052e-05,
-      "loss": 0.0375,
+      "learning_rate": 4.907514450867052e-06,
+      "loss": 3.5685,
       "step": 1951
     },
     {
       "epoch": 8.75,
-      "learning_rate": 4.890173410404624e-05,
-      "loss": 0.044,
+      "learning_rate": 4.890173410404624e-06,
+      "loss": 3.5844,
       "step": 1952
     },
     {
       "epoch": 8.76,
-      "learning_rate": 4.872832369942196e-05,
-      "loss": 0.0438,
+      "learning_rate": 4.872832369942196e-06,
+      "loss": 3.5489,
       "step": 1953
     },
     {
       "epoch": 8.76,
-      "learning_rate": 4.855491329479768e-05,
-      "loss": 0.0375,
+      "learning_rate": 4.855491329479769e-06,
+      "loss": 3.63,
       "step": 1954
     },
     {
       "epoch": 8.77,
-      "learning_rate": 4.838150289017341e-05,
-      "loss": 0.0526,
+      "learning_rate": 4.838150289017341e-06,
+      "loss": 3.5089,
       "step": 1955
     },
     {
       "epoch": 8.77,
-      "learning_rate": 4.820809248554913e-05,
-      "loss": 0.0449,
+      "learning_rate": 4.8208092485549135e-06,
+      "loss": 3.5381,
       "step": 1956
     },
     {
       "epoch": 8.78,
-      "learning_rate": 4.803468208092485e-05,
-      "loss": 0.0414,
+      "learning_rate": 4.8034682080924855e-06,
+      "loss": 3.4585,
       "step": 1957
     },
     {
       "epoch": 8.78,
-      "learning_rate": 4.786127167630058e-05,
-      "loss": 0.031,
+      "learning_rate": 4.7861271676300585e-06,
+      "loss": 3.4748,
       "step": 1958
     },
     {
       "epoch": 8.78,
-      "learning_rate": 4.76878612716763e-05,
-      "loss": 0.0402,
+      "learning_rate": 4.7687861271676305e-06,
+      "loss": 3.4835,
       "step": 1959
     },
     {
       "epoch": 8.79,
-      "learning_rate": 4.751445086705202e-05,
-      "loss": 0.0268,
+      "learning_rate": 4.751445086705203e-06,
+      "loss": 3.4503,
       "step": 1960
     },
     {
       "epoch": 8.79,
-      "learning_rate": 4.734104046242774e-05,
-      "loss": 0.036,
+      "learning_rate": 4.734104046242775e-06,
+      "loss": 3.451,
       "step": 1961
     },
     {
       "epoch": 8.8,
-      "learning_rate": 4.716763005780347e-05,
-      "loss": 0.0352,
+      "learning_rate": 4.716763005780347e-06,
+      "loss": 3.4158,
       "step": 1962
     },
     {
       "epoch": 8.8,
-      "learning_rate": 4.699421965317919e-05,
-      "loss": 0.0335,
+      "learning_rate": 4.69942196531792e-06,
+      "loss": 3.397,
       "step": 1963
     },
     {
       "epoch": 8.81,
-      "learning_rate": 4.682080924855491e-05,
-      "loss": 0.0351,
+      "learning_rate": 4.682080924855492e-06,
+      "loss": 3.3698,
       "step": 1964
     },
     {
       "epoch": 8.81,
-      "learning_rate": 4.6647398843930636e-05,
-      "loss": 0.0366,
+      "learning_rate": 4.664739884393064e-06,
+      "loss": 3.4446,
       "step": 1965
     },
     {
       "epoch": 8.82,
-      "learning_rate": 4.647398843930636e-05,
-      "loss": 0.0244,
+      "learning_rate": 4.647398843930636e-06,
+      "loss": 3.2909,
       "step": 1966
     },
     {
       "epoch": 8.82,
-      "learning_rate": 4.630057803468208e-05,
-      "loss": 0.0418,
+      "learning_rate": 4.630057803468209e-06,
+      "loss": 3.3365,
       "step": 1967
     },
     {
       "epoch": 8.83,
-      "learning_rate": 4.61271676300578e-05,
-      "loss": 0.0453,
+      "learning_rate": 4.612716763005781e-06,
+      "loss": 3.3514,
       "step": 1968
     },
     {
       "epoch": 8.83,
-      "learning_rate": 4.5953757225433526e-05,
-      "loss": 0.0253,
+      "learning_rate": 4.595375722543353e-06,
+      "loss": 3.3568,
       "step": 1969
     },
     {
       "epoch": 8.83,
-      "learning_rate": 4.5780346820809246e-05,
-      "loss": 0.0302,
+      "learning_rate": 4.578034682080925e-06,
+      "loss": 3.2758,
       "step": 1970
     },
     {
       "epoch": 8.84,
-      "learning_rate": 4.560693641618497e-05,
-      "loss": 0.0309,
+      "learning_rate": 4.560693641618497e-06,
+      "loss": 3.2929,
       "step": 1971
     },
     {
       "epoch": 8.84,
-      "learning_rate": 4.5433526011560694e-05,
-      "loss": 0.0245,
+      "learning_rate": 4.54335260115607e-06,
+      "loss": 3.3295,
       "step": 1972
     },
     {
       "epoch": 8.85,
-      "learning_rate": 4.5260115606936415e-05,
-      "loss": 0.0167,
+      "learning_rate": 4.526011560693642e-06,
+      "loss": 3.282,
       "step": 1973
     },
     {
       "epoch": 8.85,
-      "learning_rate": 4.5086705202312136e-05,
-      "loss": 0.0278,
+      "learning_rate": 4.508670520231214e-06,
+      "loss": 3.2687,
       "step": 1974
     },
     {
       "epoch": 8.86,
-      "learning_rate": 4.491329479768786e-05,
-      "loss": 0.0301,
+      "learning_rate": 4.491329479768786e-06,
+      "loss": 3.339,
       "step": 1975
     },
     {
       "epoch": 8.86,
-      "learning_rate": 4.4739884393063584e-05,
-      "loss": 0.0281,
+      "learning_rate": 4.473988439306359e-06,
+      "loss": 3.2867,
       "step": 1976
     },
     {
       "epoch": 8.87,
-      "learning_rate": 4.4566473988439305e-05,
-      "loss": 0.0392,
+      "learning_rate": 4.456647398843931e-06,
+      "loss": 3.3589,
       "step": 1977
     },
     {
       "epoch": 8.87,
-      "learning_rate": 4.4393063583815025e-05,
-      "loss": 0.0225,
+      "learning_rate": 4.439306358381503e-06,
+      "loss": 3.3917,
       "step": 1978
     },
     {
       "epoch": 8.87,
-      "learning_rate": 4.421965317919075e-05,
-      "loss": 0.0263,
+      "learning_rate": 4.421965317919075e-06,
+      "loss": 3.2579,
       "step": 1979
     },
     {
       "epoch": 8.88,
-      "learning_rate": 4.404624277456647e-05,
-      "loss": 0.0203,
+      "learning_rate": 4.404624277456648e-06,
+      "loss": 3.3597,
       "step": 1980
     },
     {
       "epoch": 8.88,
-      "learning_rate": 4.3872832369942194e-05,
-      "loss": 0.0267,
+      "learning_rate": 4.38728323699422e-06,
+      "loss": 3.3344,
       "step": 1981
     },
     {
       "epoch": 8.89,
-      "learning_rate": 4.369942196531792e-05,
-      "loss": 0.032,
+      "learning_rate": 4.369942196531792e-06,
+      "loss": 3.2327,
       "step": 1982
     },
     {
       "epoch": 8.89,
-      "learning_rate": 4.352601156069364e-05,
-      "loss": 0.0291,
+      "learning_rate": 4.352601156069364e-06,
+      "loss": 3.3952,
       "step": 1983
     },
     {
       "epoch": 8.9,
-      "learning_rate": 4.335260115606936e-05,
-      "loss": 0.0374,
+      "learning_rate": 4.3352601156069365e-06,
+      "loss": 3.3123,
       "step": 1984
     },
     {
       "epoch": 8.9,
-      "learning_rate": 4.3179190751445084e-05,
-      "loss": 0.0435,
+      "learning_rate": 4.317919075144509e-06,
+      "loss": 3.98,
       "step": 1985
     },
     {
       "epoch": 8.91,
-      "learning_rate": 4.300578034682081e-05,
-      "loss": 0.0417,
+      "learning_rate": 4.3005780346820814e-06,
+      "loss": 4.0363,
       "step": 1986
     },
     {
       "epoch": 8.91,
-      "learning_rate": 4.283236994219653e-05,
-      "loss": 0.05,
+      "learning_rate": 4.2832369942196535e-06,
+      "loss": 4.1265,
       "step": 1987
     },
     {
       "epoch": 8.91,
-      "learning_rate": 4.265895953757225e-05,
-      "loss": 0.0402,
+      "learning_rate": 4.2658959537572256e-06,
+      "loss": 3.9383,
       "step": 1988
     },
     {
       "epoch": 8.92,
-      "learning_rate": 4.248554913294798e-05,
-      "loss": 0.0487,
+      "learning_rate": 4.2485549132947985e-06,
+      "loss": 3.9433,
       "step": 1989
     },
     {
       "epoch": 8.92,
-      "learning_rate": 4.23121387283237e-05,
-      "loss": 0.0366,
+      "learning_rate": 4.2312138728323706e-06,
+      "loss": 3.8571,
       "step": 1990
     },
     {
       "epoch": 8.93,
-      "learning_rate": 4.213872832369942e-05,
-      "loss": 0.0349,
+      "learning_rate": 4.213872832369943e-06,
+      "loss": 3.699,
       "step": 1991
     },
     {
       "epoch": 8.93,
-      "learning_rate": 4.196531791907514e-05,
-      "loss": 0.0305,
+      "learning_rate": 4.196531791907515e-06,
+      "loss": 3.5891,
       "step": 1992
     },
     {
       "epoch": 8.94,
-      "learning_rate": 4.179190751445087e-05,
-      "loss": 0.0443,
+      "learning_rate": 4.179190751445088e-06,
+      "loss": 3.555,
       "step": 1993
     },
     {
       "epoch": 8.94,
-      "learning_rate": 4.161849710982658e-05,
-      "loss": 0.0418,
+      "learning_rate": 4.161849710982659e-06,
+      "loss": 3.5435,
       "step": 1994
     },
     {
       "epoch": 8.95,
-      "learning_rate": 4.1445086705202304e-05,
-      "loss": 0.03,
+      "learning_rate": 4.144508670520231e-06,
+      "loss": 3.5203,
       "step": 1995
     },
     {
       "epoch": 8.95,
-      "learning_rate": 4.1271676300578025e-05,
-      "loss": 0.0308,
+      "learning_rate": 4.127167630057803e-06,
+      "loss": 3.5198,
       "step": 1996
     },
     {
       "epoch": 8.96,
-      "learning_rate": 4.109826589595375e-05,
-      "loss": 0.0293,
+      "learning_rate": 4.109826589595375e-06,
+      "loss": 3.4158,
       "step": 1997
     },
     {
       "epoch": 8.96,
-      "learning_rate": 4.092485549132947e-05,
-      "loss": 0.0419,
+      "learning_rate": 4.092485549132948e-06,
+      "loss": 3.4458,
       "step": 1998
     },
     {
       "epoch": 8.96,
-      "learning_rate": 4.075144508670519e-05,
-      "loss": 0.0258,
+      "learning_rate": 4.07514450867052e-06,
+      "loss": 3.3295,
       "step": 1999
     },
     {
       "epoch": 8.97,
-      "learning_rate": 4.057803468208092e-05,
-      "loss": 0.0291,
+      "learning_rate": 4.057803468208092e-06,
+      "loss": 3.3698,
       "step": 2000
     },
     {
       "epoch": 8.97,
-      "eval_loss": 0.3589690625667572,
-      "eval_runtime": 570.6282,
-      "eval_samples_per_second": 4.63,
-      "eval_steps_per_second": 0.58,
-      "eval_wer": 0.09641015470051567,
+      "eval_loss": 4.259941577911377,
+      "eval_runtime": 524.0014,
+      "eval_samples_per_second": 5.042,
+      "eval_steps_per_second": 0.632,
+      "eval_wer": 1.0190995636652123,
       "step": 2000
     },
     {
       "epoch": 8.97,
-      "learning_rate": 4.040462427745664e-05,
-      "loss": 0.0345,
+      "learning_rate": 4.040462427745664e-06,
+      "loss": 3.3549,
       "step": 2001
     },
     {
       "epoch": 8.98,
-      "learning_rate": 4.023121387283236e-05,
-      "loss": 0.0343,
+      "learning_rate": 4.023121387283237e-06,
+      "loss": 3.3073,
       "step": 2002
     },
     {
       "epoch": 8.98,
-      "learning_rate": 4.005780346820808e-05,
-      "loss": 0.0374,
+      "learning_rate": 4.005780346820809e-06,
+      "loss": 3.2608,
       "step": 2003
     },
     {
       "epoch": 8.99,
-      "learning_rate": 3.988439306358381e-05,
-      "loss": 0.0244,
+      "learning_rate": 3.988439306358381e-06,
+      "loss": 3.376,
       "step": 2004
     },
     {
       "epoch": 8.99,
-      "learning_rate": 3.971098265895953e-05,
-      "loss": 0.0371,
+      "learning_rate": 3.971098265895953e-06,
+      "loss": 3.2901,
       "step": 2005
     },
     {
       "epoch": 9.0,
-      "learning_rate": 3.953757225433525e-05,
-      "loss": 0.0275,
+      "learning_rate": 3.953757225433526e-06,
+      "loss": 3.4347,
       "step": 2006
     },
     {
       "epoch": 9.0,
-      "learning_rate": 3.936416184971098e-05,
-      "loss": 0.0245,
+      "learning_rate": 3.936416184971098e-06,
+      "loss": 3.3191,
       "step": 2007
     },
     {
       "epoch": 9.0,
-      "learning_rate": 3.91907514450867e-05,
-      "loss": 0.057,
+      "learning_rate": 3.91907514450867e-06,
+      "loss": 4.0172,
       "step": 2008
     },
     {
       "epoch": 9.01,
-      "learning_rate": 3.901734104046242e-05,
-      "loss": 0.0298,
+      "learning_rate": 3.901734104046242e-06,
+      "loss": 3.8855,
       "step": 2009
     },
     {
       "epoch": 9.01,
-      "learning_rate": 3.884393063583814e-05,
-      "loss": 0.0394,
+      "learning_rate": 3.8843930635838145e-06,
+      "loss": 3.9422,
       "step": 2010
     },
     {
       "epoch": 9.02,
-      "learning_rate": 3.867052023121387e-05,
-      "loss": 0.0377,
+      "learning_rate": 3.867052023121387e-06,
+      "loss": 3.9771,
       "step": 2011
     },
     {
       "epoch": 9.02,
-      "learning_rate": 3.849710982658959e-05,
-      "loss": 0.0317,
+      "learning_rate": 3.8497109826589594e-06,
+      "loss": 3.995,
       "step": 2012
     },
     {
       "epoch": 9.03,
-      "learning_rate": 3.832369942196531e-05,
-      "loss": 0.0363,
+      "learning_rate": 3.8323699421965315e-06,
+      "loss": 3.881,
       "step": 2013
     },
     {
       "epoch": 9.03,
-      "learning_rate": 3.815028901734104e-05,
-      "loss": 0.0391,
+      "learning_rate": 3.8150289017341036e-06,
+      "loss": 3.9283,
       "step": 2014
     },
     {
       "epoch": 9.04,
-      "learning_rate": 3.797687861271676e-05,
-      "loss": 0.0324,
+      "learning_rate": 3.797687861271676e-06,
+      "loss": 3.8494,
       "step": 2015
     },
     {
       "epoch": 9.04,
-      "learning_rate": 3.780346820809248e-05,
-      "loss": 0.042,
+      "learning_rate": 3.7803468208092486e-06,
+      "loss": 3.8015,
       "step": 2016
     },
     {
       "epoch": 9.04,
-      "learning_rate": 3.76300578034682e-05,
-      "loss": 0.029,
+      "learning_rate": 3.7630057803468206e-06,
+      "loss": 3.8063,
       "step": 2017
     },
     {
       "epoch": 9.05,
-      "learning_rate": 3.745664739884393e-05,
-      "loss": 0.0427,
+      "learning_rate": 3.745664739884393e-06,
+      "loss": 3.7155,
       "step": 2018
     },
     {
       "epoch": 9.05,
-      "learning_rate": 3.728323699421965e-05,
-      "loss": 0.0366,
+      "learning_rate": 3.728323699421965e-06,
+      "loss": 3.7095,
       "step": 2019
     },
     {
       "epoch": 9.06,
-      "learning_rate": 3.710982658959537e-05,
-      "loss": 0.0306,
+      "learning_rate": 3.7109826589595377e-06,
+      "loss": 3.6091,
       "step": 2020
     },
     {
       "epoch": 9.06,
-      "learning_rate": 3.6936416184971096e-05,
-      "loss": 0.0419,
+      "learning_rate": 3.6936416184971097e-06,
+      "loss": 3.5699,
       "step": 2021
     },
     {
       "epoch": 9.07,
-      "learning_rate": 3.6763005780346816e-05,
-      "loss": 0.0289,
+      "learning_rate": 3.676300578034682e-06,
+      "loss": 3.5183,
       "step": 2022
     },
     {
       "epoch": 9.07,
-      "learning_rate": 3.658959537572254e-05,
-      "loss": 0.033,
+      "learning_rate": 3.6589595375722543e-06,
+      "loss": 3.5259,
       "step": 2023
     },
     {
       "epoch": 9.08,
-      "learning_rate": 3.6416184971098265e-05,
-      "loss": 0.0274,
+      "learning_rate": 3.6416184971098264e-06,
+      "loss": 3.4793,
       "step": 2024
     },
     {
       "epoch": 9.08,
-      "learning_rate": 3.6242774566473985e-05,
-      "loss": 0.2688,
+      "learning_rate": 3.624277456647399e-06,
+      "loss": 3.4506,
       "step": 2025
     },
     {
       "epoch": 9.09,
-      "learning_rate": 3.6069364161849706e-05,
-      "loss": 0.0241,
+      "learning_rate": 3.606936416184971e-06,
+      "loss": 3.4071,
       "step": 2026
     },
     {
       "epoch": 9.09,
-      "learning_rate": 3.5895953757225427e-05,
-      "loss": 0.0307,
+      "learning_rate": 3.5895953757225434e-06,
+      "loss": 3.4044,
       "step": 2027
     },
     {
       "epoch": 9.09,
-      "learning_rate": 3.5722543352601154e-05,
-      "loss": 0.0317,
+      "learning_rate": 3.5722543352601155e-06,
+      "loss": 3.398,
       "step": 2028
     },
     {
       "epoch": 9.1,
-      "learning_rate": 3.5549132947976875e-05,
-      "loss": 0.0328,
+      "learning_rate": 3.554913294797688e-06,
+      "loss": 3.4986,
       "step": 2029
     },
     {
       "epoch": 9.1,
-      "learning_rate": 3.5375722543352595e-05,
-      "loss": 0.0291,
+      "learning_rate": 3.53757225433526e-06,
+      "loss": 3.4046,
       "step": 2030
     },
     {
       "epoch": 9.11,
-      "learning_rate": 3.520231213872832e-05,
-      "loss": 0.0224,
+      "learning_rate": 3.5202312138728325e-06,
+      "loss": 3.3652,
       "step": 2031
     },
     {
       "epoch": 9.11,
-      "learning_rate": 3.5028901734104043e-05,
-      "loss": 0.0383,
+      "learning_rate": 3.5028901734104046e-06,
+      "loss": 3.4288,
       "step": 2032
     },
     {
       "epoch": 9.12,
-      "learning_rate": 3.4855491329479764e-05,
-      "loss": 0.0319,
+      "learning_rate": 3.485549132947977e-06,
+      "loss": 3.3382,
       "step": 2033
     },
     {
       "epoch": 9.12,
-      "learning_rate": 3.4682080924855485e-05,
-      "loss": 0.0283,
+      "learning_rate": 3.468208092485549e-06,
+      "loss": 3.3078,
       "step": 2034
     },
     {
       "epoch": 9.13,
-      "learning_rate": 3.450867052023121e-05,
-      "loss": 0.0184,
+      "learning_rate": 3.4508670520231212e-06,
+      "loss": 3.3217,
       "step": 2035
     },
     {
       "epoch": 9.13,
-      "learning_rate": 3.433526011560693e-05,
-      "loss": 0.0273,
+      "learning_rate": 3.4335260115606937e-06,
+      "loss": 3.2776,
       "step": 2036
     },
     {
       "epoch": 9.13,
-      "learning_rate": 3.4161849710982654e-05,
-      "loss": 0.0288,
+      "learning_rate": 3.416184971098266e-06,
+      "loss": 3.3606,
       "step": 2037
     },
     {
       "epoch": 9.14,
-      "learning_rate": 3.398843930635838e-05,
-      "loss": 0.0342,
+      "learning_rate": 3.3988439306358383e-06,
+      "loss": 3.3422,
       "step": 2038
     },
     {
       "epoch": 9.14,
-      "learning_rate": 3.38150289017341e-05,
-      "loss": 0.0355,
+      "learning_rate": 3.3815028901734103e-06,
+      "loss": 3.3055,
       "step": 2039
     },
     {
       "epoch": 9.15,
-      "learning_rate": 3.364161849710982e-05,
-      "loss": 0.0316,
+      "learning_rate": 3.364161849710983e-06,
+      "loss": 3.2051,
       "step": 2040
     },
     {
       "epoch": 9.15,
-      "learning_rate": 3.346820809248554e-05,
-      "loss": 0.0263,
+      "learning_rate": 3.346820809248555e-06,
+      "loss": 3.2975,
       "step": 2041
     },
     {
       "epoch": 9.16,
-      "learning_rate": 3.329479768786127e-05,
-      "loss": 0.0373,
+      "learning_rate": 3.3294797687861274e-06,
+      "loss": 3.2273,
       "step": 2042
     },
     {
       "epoch": 9.16,
-      "learning_rate": 3.312138728323699e-05,
-      "loss": 0.0224,
+      "learning_rate": 3.3121387283236995e-06,
+      "loss": 3.2341,
       "step": 2043
     },
     {
       "epoch": 9.17,
-      "learning_rate": 3.294797687861271e-05,
-      "loss": 0.029,
+      "learning_rate": 3.294797687861272e-06,
+      "loss": 3.2985,
       "step": 2044
     },
     {
       "epoch": 9.17,
-      "learning_rate": 3.277456647398844e-05,
-      "loss": 0.0298,
+      "learning_rate": 3.277456647398844e-06,
+      "loss": 3.278,
       "step": 2045
     },
     {
       "epoch": 9.17,
-      "learning_rate": 3.260115606936416e-05,
-      "loss": 0.0223,
+      "learning_rate": 3.260115606936416e-06,
+      "loss": 3.2753,
       "step": 2046
     },
     {
       "epoch": 9.18,
-      "learning_rate": 3.242774566473988e-05,
-      "loss": 0.0351,
+      "learning_rate": 3.2427745664739886e-06,
+      "loss": 3.261,
       "step": 2047
     },
     {
       "epoch": 9.18,
-      "learning_rate": 3.22543352601156e-05,
-      "loss": 0.023,
+      "learning_rate": 3.2254335260115607e-06,
+      "loss": 3.1894,
       "step": 2048
     },
     {
       "epoch": 9.19,
-      "learning_rate": 3.208092485549133e-05,
-      "loss": 0.0231,
+      "learning_rate": 3.208092485549133e-06,
+      "loss": 3.3104,
       "step": 2049
     },
     {
       "epoch": 9.19,
-      "learning_rate": 3.190751445086705e-05,
-      "loss": 0.022,
+      "learning_rate": 3.1907514450867052e-06,
+      "loss": 3.2267,
       "step": 2050
     },
     {
       "epoch": 9.2,
-      "learning_rate": 3.173410404624277e-05,
-      "loss": 0.0329,
+      "learning_rate": 3.1734104046242777e-06,
+      "loss": 3.3057,
       "step": 2051
     },
     {
       "epoch": 9.2,
-      "learning_rate": 3.15606936416185e-05,
-      "loss": 0.0294,
+      "learning_rate": 3.1560693641618498e-06,
+      "loss": 3.231,
       "step": 2052
     },
     {
       "epoch": 9.21,
-      "learning_rate": 3.138728323699422e-05,
-      "loss": 0.0151,
+      "learning_rate": 3.1387283236994223e-06,
+      "loss": 3.3067,
       "step": 2053
     },
     {
       "epoch": 9.21,
-      "learning_rate": 3.121387283236994e-05,
-      "loss": 0.0167,
+      "learning_rate": 3.1213872832369943e-06,
+      "loss": 3.2419,
       "step": 2054
     },
     {
       "epoch": 9.22,
-      "learning_rate": 3.1040462427745667e-05,
-      "loss": 0.0182,
+      "learning_rate": 3.104046242774567e-06,
+      "loss": 3.2409,
       "step": 2055
     },
     {
       "epoch": 9.22,
-      "learning_rate": 3.086705202312139e-05,
-      "loss": 0.0431,
+      "learning_rate": 3.086705202312139e-06,
+      "loss": 3.3373,
       "step": 2056
     },
     {
       "epoch": 9.22,
-      "learning_rate": 3.069364161849711e-05,
-      "loss": 0.0711,
+      "learning_rate": 3.0693641618497114e-06,
+      "loss": 3.2751,
       "step": 2057
     },
     {
       "epoch": 9.23,
-      "learning_rate": 3.052023121387283e-05,
-      "loss": 0.0487,
+      "learning_rate": 3.0520231213872834e-06,
+      "loss": 4.1168,
       "step": 2058
     },
     {
       "epoch": 9.23,
-      "learning_rate": 3.0346820809248553e-05,
-      "loss": 0.0368,
+      "learning_rate": 3.0346820809248555e-06,
+      "loss": 3.9596,
       "step": 2059
     },
     {
       "epoch": 9.24,
-      "learning_rate": 3.0173410404624277e-05,
-      "loss": 0.0333,
+      "learning_rate": 3.017341040462428e-06,
+      "loss": 4.0989,
       "step": 2060
     },
     {
       "epoch": 9.24,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 0.0409,
+      "learning_rate": 3e-06,
+      "loss": 4.0323,
       "step": 2061
     },
     {
       "epoch": 9.25,
-      "learning_rate": 2.982658959537572e-05,
-      "loss": 0.038,
+      "learning_rate": 2.9826589595375726e-06,
+      "loss": 4.0487,
       "step": 2062
     },
     {
       "epoch": 9.25,
-      "learning_rate": 2.9653179190751446e-05,
-      "loss": 0.0301,
+      "learning_rate": 2.9653179190751446e-06,
+      "loss": 3.9915,
       "step": 2063
     },
     {
       "epoch": 9.26,
-      "learning_rate": 2.9479768786127166e-05,
-      "loss": 0.0407,
+      "learning_rate": 2.947976878612717e-06,
+      "loss": 3.8636,
       "step": 2064
     },
     {
       "epoch": 9.26,
-      "learning_rate": 2.930635838150289e-05,
-      "loss": 0.0388,
+      "learning_rate": 2.930635838150289e-06,
+      "loss": 3.9274,
       "step": 2065
     },
     {
       "epoch": 9.26,
-      "learning_rate": 2.9132947976878608e-05,
-      "loss": 0.0413,
+      "learning_rate": 2.9132947976878613e-06,
+      "loss": 3.8802,
       "step": 2066
     },
     {
       "epoch": 9.27,
-      "learning_rate": 2.895953757225433e-05,
-      "loss": 0.0317,
+      "learning_rate": 2.8959537572254333e-06,
+      "loss": 3.8013,
       "step": 2067
     },
     {
       "epoch": 9.27,
-      "learning_rate": 2.8786127167630052e-05,
-      "loss": 0.0397,
+      "learning_rate": 2.878612716763006e-06,
+      "loss": 3.8256,
       "step": 2068
     },
     {
       "epoch": 9.28,
-      "learning_rate": 2.8612716763005776e-05,
-      "loss": 0.0456,
+      "learning_rate": 2.861271676300578e-06,
+      "loss": 3.6916,
       "step": 2069
     },
     {
       "epoch": 9.28,
-      "learning_rate": 2.8439306358381497e-05,
-      "loss": 0.0348,
+      "learning_rate": 2.8439306358381504e-06,
+      "loss": 3.6942,
       "step": 2070
     },
     {
       "epoch": 9.29,
-      "learning_rate": 2.826589595375722e-05,
-      "loss": 0.0273,
+      "learning_rate": 2.8265895953757224e-06,
+      "loss": 3.7135,
       "step": 2071
     },
     {
       "epoch": 9.29,
-      "learning_rate": 2.8092485549132945e-05,
-      "loss": 0.0381,
+      "learning_rate": 2.8092485549132945e-06,
+      "loss": 3.5903,
       "step": 2072
     },
     {
       "epoch": 9.3,
-      "learning_rate": 2.7919075144508666e-05,
-      "loss": 0.0338,
+      "learning_rate": 2.791907514450867e-06,
+      "loss": 3.5691,
       "step": 2073
     },
     {
       "epoch": 9.3,
-      "learning_rate": 2.774566473988439e-05,
-      "loss": 0.0308,
+      "learning_rate": 2.774566473988439e-06,
+      "loss": 3.4745,
       "step": 2074
     },
     {
       "epoch": 9.3,
-      "learning_rate": 2.757225433526011e-05,
-      "loss": 0.0325,
+      "learning_rate": 2.7572254335260116e-06,
+      "loss": 3.4411,
       "step": 2075
     },
     {
       "epoch": 9.31,
-      "learning_rate": 2.7398843930635835e-05,
-      "loss": 0.0207,
+      "learning_rate": 2.7398843930635836e-06,
+      "loss": 3.4509,
       "step": 2076
     },
     {
       "epoch": 9.31,
-      "learning_rate": 2.722543352601156e-05,
-      "loss": 0.0315,
+      "learning_rate": 2.722543352601156e-06,
+      "loss": 3.4391,
       "step": 2077
     },
     {
       "epoch": 9.32,
-      "learning_rate": 2.705202312138728e-05,
-      "loss": 0.0329,
+      "learning_rate": 2.705202312138728e-06,
+      "loss": 3.5249,
       "step": 2078
     },
     {
       "epoch": 9.32,
-      "learning_rate": 2.6878612716763003e-05,
-      "loss": 0.0331,
+      "learning_rate": 2.6878612716763007e-06,
+      "loss": 3.4409,
       "step": 2079
     },
     {
       "epoch": 9.33,
-      "learning_rate": 2.6705202312138724e-05,
-      "loss": 0.0251,
+      "learning_rate": 2.6705202312138728e-06,
+      "loss": 3.3896,
       "step": 2080
     },
     {
       "epoch": 9.33,
-      "learning_rate": 2.6531791907514448e-05,
-      "loss": 0.0311,
+      "learning_rate": 2.6531791907514452e-06,
+      "loss": 3.4046,
       "step": 2081
     },
     {
       "epoch": 9.34,
-      "learning_rate": 2.635838150289017e-05,
-      "loss": 0.0378,
+      "learning_rate": 2.6358381502890173e-06,
+      "loss": 3.4891,
       "step": 2082
     },
     {
       "epoch": 9.34,
-      "learning_rate": 2.6184971098265893e-05,
-      "loss": 0.0282,
+      "learning_rate": 2.61849710982659e-06,
+      "loss": 3.4354,
       "step": 2083
     },
     {
       "epoch": 9.35,
-      "learning_rate": 2.6011560693641617e-05,
-      "loss": 0.0314,
+      "learning_rate": 2.601156069364162e-06,
+      "loss": 3.3831,
       "step": 2084
     },
     {
       "epoch": 9.35,
-      "learning_rate": 2.5838150289017338e-05,
-      "loss": 0.0311,
+      "learning_rate": 2.583815028901734e-06,
+      "loss": 3.3975,
       "step": 2085
     },
     {
       "epoch": 9.35,
-      "learning_rate": 2.5664739884393062e-05,
-      "loss": 0.034,
+      "learning_rate": 2.5664739884393064e-06,
+      "loss": 3.3543,
       "step": 2086
     },
     {
       "epoch": 9.36,
-      "learning_rate": 2.5491329479768782e-05,
-      "loss": 0.0324,
+      "learning_rate": 2.5491329479768785e-06,
+      "loss": 3.3374,
       "step": 2087
     },
     {
       "epoch": 9.36,
-      "learning_rate": 2.5317919075144507e-05,
-      "loss": 0.0358,
+      "learning_rate": 2.531791907514451e-06,
+      "loss": 3.4501,
       "step": 2088
     },
     {
       "epoch": 9.37,
-      "learning_rate": 2.514450867052023e-05,
-      "loss": 0.0237,
+      "learning_rate": 2.514450867052023e-06,
+      "loss": 3.239,
       "step": 2089
     },
     {
       "epoch": 9.37,
-      "learning_rate": 2.497109826589595e-05,
-      "loss": 0.0342,
+      "learning_rate": 2.4971098265895955e-06,
+      "loss": 3.3719,
       "step": 2090
     },
     {
       "epoch": 9.38,
-      "learning_rate": 2.4797687861271675e-05,
-      "loss": 0.0208,
+      "learning_rate": 2.4797687861271676e-06,
+      "loss": 3.3023,
       "step": 2091
     },
     {
       "epoch": 9.38,
-      "learning_rate": 2.4624277456647396e-05,
-      "loss": 0.0299,
+      "learning_rate": 2.46242774566474e-06,
+      "loss": 3.3659,
       "step": 2092
     },
     {
       "epoch": 9.39,
-      "learning_rate": 2.445086705202312e-05,
-      "loss": 0.0341,
+      "learning_rate": 2.445086705202312e-06,
+      "loss": 3.2331,
       "step": 2093
     },
     {
       "epoch": 9.39,
-      "learning_rate": 2.427745664739884e-05,
-      "loss": 0.0236,
+      "learning_rate": 2.4277456647398847e-06,
+      "loss": 3.2789,
       "step": 2094
     },
     {
       "epoch": 9.39,
-      "learning_rate": 2.4104046242774565e-05,
-      "loss": 0.0257,
+      "learning_rate": 2.4104046242774567e-06,
+      "loss": 3.2706,
       "step": 2095
     },
     {
       "epoch": 9.4,
-      "learning_rate": 2.393063583815029e-05,
-      "loss": 0.0215,
+      "learning_rate": 2.3930635838150292e-06,
+      "loss": 3.2017,
       "step": 2096
     },
     {
       "epoch": 9.4,
-      "learning_rate": 2.375722543352601e-05,
-      "loss": 0.0304,
+      "learning_rate": 2.3757225433526013e-06,
+      "loss": 3.2796,
       "step": 2097
     },
     {
       "epoch": 9.41,
-      "learning_rate": 2.3583815028901734e-05,
-      "loss": 0.0249,
+      "learning_rate": 2.3583815028901734e-06,
+      "loss": 3.2567,
       "step": 2098
     },
     {
       "epoch": 9.41,
-      "learning_rate": 2.3410404624277454e-05,
-      "loss": 0.0361,
+      "learning_rate": 2.341040462427746e-06,
+      "loss": 3.2948,
       "step": 2099
     },
     {
       "epoch": 9.42,
-      "learning_rate": 2.323699421965318e-05,
-      "loss": 0.0242,
+      "learning_rate": 2.323699421965318e-06,
+      "loss": 3.2829,
       "step": 2100
     },
     {
       "epoch": 9.42,
-      "learning_rate": 2.30635838150289e-05,
-      "loss": 0.0198,
+      "learning_rate": 2.3063583815028904e-06,
+      "loss": 3.3335,
       "step": 2101
     },
     {
       "epoch": 9.43,
-      "learning_rate": 2.2890173410404623e-05,
-      "loss": 0.0237,
+      "learning_rate": 2.2890173410404625e-06,
+      "loss": 3.2214,
       "step": 2102
     },
     {
       "epoch": 9.43,
-      "learning_rate": 2.2716763005780347e-05,
-      "loss": 0.0177,
+      "learning_rate": 2.271676300578035e-06,
+      "loss": 3.3146,
       "step": 2103
     },
     {
       "epoch": 9.43,
-      "learning_rate": 2.2543352601156068e-05,
-      "loss": 0.0389,
+      "learning_rate": 2.254335260115607e-06,
+      "loss": 3.2607,
       "step": 2104
     },
     {
       "epoch": 9.44,
-      "learning_rate": 2.2369942196531792e-05,
-      "loss": 0.0212,
+      "learning_rate": 2.2369942196531795e-06,
+      "loss": 3.219,
       "step": 2105
     },
     {
       "epoch": 9.44,
-      "learning_rate": 2.2196531791907513e-05,
-      "loss": 0.0244,
+      "learning_rate": 2.2196531791907516e-06,
+      "loss": 3.332,
       "step": 2106
     },
     {
       "epoch": 9.45,
-      "learning_rate": 2.2023121387283237e-05,
-      "loss": 0.0884,
+      "learning_rate": 2.202312138728324e-06,
+      "loss": 3.3571,
       "step": 2107
     },
     {
       "epoch": 9.45,
-      "learning_rate": 2.184971098265896e-05,
-      "loss": 0.0482,
+      "learning_rate": 2.184971098265896e-06,
+      "loss": 4.061,
       "step": 2108
     },
     {
       "epoch": 9.46,
-      "learning_rate": 2.167630057803468e-05,
-      "loss": 0.0422,
+      "learning_rate": 2.1676300578034682e-06,
+      "loss": 4.0652,
       "step": 2109
     },
     {
       "epoch": 9.46,
-      "learning_rate": 2.1502890173410405e-05,
-      "loss": 0.0355,
+      "learning_rate": 2.1502890173410407e-06,
+      "loss": 4.0563,
       "step": 2110
     },
     {
       "epoch": 9.47,
-      "learning_rate": 2.1329479768786126e-05,
-      "loss": 0.0358,
+      "learning_rate": 2.1329479768786128e-06,
+      "loss": 3.9696,
       "step": 2111
     },
     {
       "epoch": 9.47,
-      "learning_rate": 2.115606936416185e-05,
-      "loss": 0.0419,
+      "learning_rate": 2.1156069364161853e-06,
+      "loss": 3.9416,
       "step": 2112
     },
     {
       "epoch": 9.48,
-      "learning_rate": 2.098265895953757e-05,
-      "loss": 0.0411,
+      "learning_rate": 2.0982658959537573e-06,
+      "loss": 4.0014,
       "step": 2113
     },
     {
       "epoch": 9.48,
-      "learning_rate": 2.080924855491329e-05,
-      "loss": 0.0319,
+      "learning_rate": 2.0809248554913294e-06,
+      "loss": 4.0486,
       "step": 2114
     },
     {
       "epoch": 9.48,
-      "learning_rate": 2.0635838150289012e-05,
-      "loss": 0.0288,
+      "learning_rate": 2.0635838150289015e-06,
+      "loss": 3.8973,
       "step": 2115
     },
     {
       "epoch": 9.49,
-      "learning_rate": 2.0462427745664736e-05,
-      "loss": 0.0269,
+      "learning_rate": 2.046242774566474e-06,
+      "loss": 3.8664,
       "step": 2116
     },
     {
       "epoch": 9.49,
-      "learning_rate": 2.028901734104046e-05,
-      "loss": 0.0355,
+      "learning_rate": 2.028901734104046e-06,
+      "loss": 3.8263,
       "step": 2117
     },
     {
       "epoch": 9.5,
-      "learning_rate": 2.011560693641618e-05,
-      "loss": 0.032,
+      "learning_rate": 2.0115606936416185e-06,
+      "loss": 3.7801,
       "step": 2118
     },
     {
       "epoch": 9.5,
-      "learning_rate": 1.9942196531791905e-05,
-      "loss": 0.0251,
+      "learning_rate": 1.9942196531791906e-06,
+      "loss": 3.7214,
       "step": 2119
     },
     {
       "epoch": 9.51,
-      "learning_rate": 1.9768786127167626e-05,
-      "loss": 0.0295,
+      "learning_rate": 1.976878612716763e-06,
+      "loss": 3.623,
       "step": 2120
     },
     {
       "epoch": 9.51,
-      "learning_rate": 1.959537572254335e-05,
-      "loss": 0.0405,
+      "learning_rate": 1.959537572254335e-06,
+      "loss": 3.6586,
       "step": 2121
     },
     {
       "epoch": 9.52,
-      "learning_rate": 1.942196531791907e-05,
-      "loss": 0.0289,
+      "learning_rate": 1.9421965317919072e-06,
+      "loss": 3.5181,
       "step": 2122
     },
     {
       "epoch": 9.52,
-      "learning_rate": 1.9248554913294795e-05,
-      "loss": 0.0337,
+      "learning_rate": 1.9248554913294797e-06,
+      "loss": 3.4883,
       "step": 2123
     },
     {
       "epoch": 9.52,
-      "learning_rate": 1.907514450867052e-05,
-      "loss": 0.035,
+      "learning_rate": 1.9075144508670518e-06,
+      "loss": 3.6009,
       "step": 2124
     },
     {
       "epoch": 9.53,
-      "learning_rate": 1.890173410404624e-05,
-      "loss": 0.029,
+      "learning_rate": 1.8901734104046243e-06,
+      "loss": 3.5483,
       "step": 2125
     },
     {
       "epoch": 9.53,
-      "learning_rate": 1.8728323699421963e-05,
-      "loss": 0.03,
+      "learning_rate": 1.8728323699421966e-06,
+      "loss": 3.4663,
       "step": 2126
     },
     {
       "epoch": 9.54,
-      "learning_rate": 1.8554913294797684e-05,
-      "loss": 0.0287,
+      "learning_rate": 1.8554913294797688e-06,
+      "loss": 3.424,
       "step": 2127
     },
     {
       "epoch": 9.54,
-      "learning_rate": 1.8381502890173408e-05,
-      "loss": 0.0354,
+      "learning_rate": 1.838150289017341e-06,
+      "loss": 3.4261,
       "step": 2128
     },
     {
       "epoch": 9.55,
-      "learning_rate": 1.8208092485549132e-05,
-      "loss": 0.0262,
+      "learning_rate": 1.8208092485549132e-06,
+      "loss": 3.4097,
       "step": 2129
     },
     {
       "epoch": 9.55,
-      "learning_rate": 1.8034682080924853e-05,
-      "loss": 0.0225,
+      "learning_rate": 1.8034682080924855e-06,
+      "loss": 3.4119,
       "step": 2130
     },
     {
       "epoch": 9.56,
-      "learning_rate": 1.7861271676300577e-05,
-      "loss": 0.0288,
+      "learning_rate": 1.7861271676300577e-06,
+      "loss": 3.3827,
       "step": 2131
     },
     {
       "epoch": 9.56,
-      "learning_rate": 1.7687861271676298e-05,
-      "loss": 0.0241,
+      "learning_rate": 1.76878612716763e-06,
+      "loss": 3.4139,
       "step": 2132
     },
     {
       "epoch": 9.57,
-      "learning_rate": 1.7514450867052022e-05,
-      "loss": 0.0277,
+      "learning_rate": 1.7514450867052023e-06,
+      "loss": 3.4792,
       "step": 2133
     },
     {
       "epoch": 9.57,
-      "learning_rate": 1.7341040462427742e-05,
-      "loss": 0.0298,
+      "learning_rate": 1.7341040462427746e-06,
+      "loss": 3.3544,
       "step": 2134
     },
     {
       "epoch": 9.57,
-      "learning_rate": 1.7167630057803466e-05,
-      "loss": 0.0278,
+      "learning_rate": 1.7167630057803469e-06,
+      "loss": 3.3638,
       "step": 2135
     },
     {
       "epoch": 9.58,
-      "learning_rate": 1.699421965317919e-05,
-      "loss": 0.0253,
+      "learning_rate": 1.6994219653179191e-06,
+      "loss": 3.3551,
       "step": 2136
     },
     {
       "epoch": 9.58,
-      "learning_rate": 1.682080924855491e-05,
-      "loss": 0.0331,
+      "learning_rate": 1.6820809248554914e-06,
+      "loss": 3.3162,
       "step": 2137
     },
     {
       "epoch": 9.59,
-      "learning_rate": 1.6647398843930635e-05,
-      "loss": 0.0272,
+      "learning_rate": 1.6647398843930637e-06,
+      "loss": 3.3552,
       "step": 2138
     },
     {
       "epoch": 9.59,
-      "learning_rate": 1.6473988439306356e-05,
-      "loss": 0.0237,
+      "learning_rate": 1.647398843930636e-06,
+      "loss": 3.2356,
       "step": 2139
     },
     {
       "epoch": 9.6,
-      "learning_rate": 1.630057803468208e-05,
-      "loss": 0.027,
+      "learning_rate": 1.630057803468208e-06,
+      "loss": 3.2151,
       "step": 2140
     },
     {
       "epoch": 9.6,
-      "learning_rate": 1.61271676300578e-05,
-      "loss": 0.026,
+      "learning_rate": 1.6127167630057803e-06,
+      "loss": 3.3076,
       "step": 2141
     },
     {
       "epoch": 9.61,
-      "learning_rate": 1.5953757225433525e-05,
-      "loss": 0.0281,
+      "learning_rate": 1.5953757225433526e-06,
+      "loss": 3.3143,
       "step": 2142
     },
     {
       "epoch": 9.61,
-      "learning_rate": 1.578034682080925e-05,
-      "loss": 0.0283,
+      "learning_rate": 1.5780346820809249e-06,
+      "loss": 3.2887,
       "step": 2143
     },
     {
       "epoch": 9.61,
-      "learning_rate": 1.560693641618497e-05,
-      "loss": 0.0188,
+      "learning_rate": 1.5606936416184972e-06,
+      "loss": 3.2681,
       "step": 2144
     },
     {
       "epoch": 9.62,
-      "learning_rate": 1.5433526011560694e-05,
-      "loss": 0.0235,
+      "learning_rate": 1.5433526011560694e-06,
+      "loss": 3.2485,
       "step": 2145
     },
     {
       "epoch": 9.62,
-      "learning_rate": 1.5260115606936414e-05,
-      "loss": 0.0257,
+      "learning_rate": 1.5260115606936417e-06,
+      "loss": 3.2595,
       "step": 2146
     },
     {
       "epoch": 9.63,
-      "learning_rate": 1.5086705202312138e-05,
-      "loss": 0.0233,
+      "learning_rate": 1.508670520231214e-06,
+      "loss": 3.2182,
       "step": 2147
     },
     {
       "epoch": 9.63,
-      "learning_rate": 1.491329479768786e-05,
-      "loss": 0.0331,
+      "learning_rate": 1.4913294797687863e-06,
+      "loss": 3.3599,
       "step": 2148
     },
     {
       "epoch": 9.64,
-      "learning_rate": 1.4739884393063583e-05,
-      "loss": 0.0364,
+      "learning_rate": 1.4739884393063586e-06,
+      "loss": 3.2669,
       "step": 2149
     },
     {
       "epoch": 9.64,
-      "learning_rate": 1.4566473988439304e-05,
-      "loss": 0.0312,
+      "learning_rate": 1.4566473988439306e-06,
+      "loss": 3.2302,
       "step": 2150
     },
     {
       "epoch": 9.65,
-      "learning_rate": 1.4393063583815026e-05,
-      "loss": 0.022,
+      "learning_rate": 1.439306358381503e-06,
+      "loss": 3.3325,
       "step": 2151
     },
     {
       "epoch": 9.65,
-      "learning_rate": 1.4219653179190749e-05,
-      "loss": 0.0338,
+      "learning_rate": 1.4219653179190752e-06,
+      "loss": 3.3096,
       "step": 2152
     },
     {
       "epoch": 9.65,
-      "learning_rate": 1.4046242774566473e-05,
-      "loss": 0.0279,
+      "learning_rate": 1.4046242774566473e-06,
+      "loss": 3.252,
       "step": 2153
     },
     {
       "epoch": 9.66,
-      "learning_rate": 1.3872832369942195e-05,
-      "loss": 0.0309,
+      "learning_rate": 1.3872832369942195e-06,
+      "loss": 3.3566,
       "step": 2154
     },
     {
       "epoch": 9.66,
-      "learning_rate": 1.3699421965317917e-05,
-      "loss": 0.056,
+      "learning_rate": 1.3699421965317918e-06,
+      "loss": 3.318,
       "step": 2155
     },
     {
       "epoch": 9.67,
-      "learning_rate": 1.352601156069364e-05,
-      "loss": 0.0299,
+      "learning_rate": 1.352601156069364e-06,
+      "loss": 3.344,
       "step": 2156
     },
     {
       "epoch": 9.67,
-      "learning_rate": 1.3352601156069362e-05,
-      "loss": 0.056,
+      "learning_rate": 1.3352601156069364e-06,
+      "loss": 3.2417,
       "step": 2157
     },
     {
       "epoch": 9.68,
-      "learning_rate": 1.3179190751445084e-05,
-      "loss": 0.0458,
+      "learning_rate": 1.3179190751445087e-06,
+      "loss": 3.9626,
       "step": 2158
     },
     {
       "epoch": 9.68,
-      "learning_rate": 1.3005780346820809e-05,
-      "loss": 0.0395,
+      "learning_rate": 1.300578034682081e-06,
+      "loss": 4.0286,
       "step": 2159
     },
     {
       "epoch": 9.69,
-      "learning_rate": 1.2832369942196531e-05,
-      "loss": 0.0401,
+      "learning_rate": 1.2832369942196532e-06,
+      "loss": 4.0529,
       "step": 2160
     },
     {
       "epoch": 9.69,
-      "learning_rate": 1.2658959537572253e-05,
-      "loss": 0.0302,
+      "learning_rate": 1.2658959537572255e-06,
+      "loss": 3.9245,
       "step": 2161
     },
     {
       "epoch": 9.7,
-      "learning_rate": 1.2485549132947976e-05,
-      "loss": 0.037,
+      "learning_rate": 1.2485549132947978e-06,
+      "loss": 3.8779,
       "step": 2162
     },
     {
       "epoch": 9.7,
-      "learning_rate": 1.2312138728323698e-05,
-      "loss": 0.0276,
+      "learning_rate": 1.23121387283237e-06,
+      "loss": 3.8989,
       "step": 2163
     },
     {
       "epoch": 9.7,
-      "learning_rate": 1.213872832369942e-05,
-      "loss": 0.0316,
+      "learning_rate": 1.2138728323699423e-06,
+      "loss": 3.8897,
       "step": 2164
     },
     {
       "epoch": 9.71,
-      "learning_rate": 1.1965317919075144e-05,
-      "loss": 0.0344,
+      "learning_rate": 1.1965317919075146e-06,
+      "loss": 3.9576,
       "step": 2165
     },
     {
       "epoch": 9.71,
-      "learning_rate": 1.1791907514450867e-05,
-      "loss": 0.0365,
+      "learning_rate": 1.1791907514450867e-06,
+      "loss": 3.8363,
       "step": 2166
     },
     {
       "epoch": 9.72,
-      "learning_rate": 1.161849710982659e-05,
-      "loss": 0.0248,
+      "learning_rate": 1.161849710982659e-06,
+      "loss": 3.9311,
       "step": 2167
     },
     {
       "epoch": 9.72,
-      "learning_rate": 1.1445086705202312e-05,
-      "loss": 0.027,
+      "learning_rate": 1.1445086705202312e-06,
+      "loss": 3.7241,
       "step": 2168
     },
     {
       "epoch": 9.73,
-      "learning_rate": 1.1271676300578034e-05,
-      "loss": 0.0345,
+      "learning_rate": 1.1271676300578035e-06,
+      "loss": 3.7679,
       "step": 2169
     },
     {
       "epoch": 9.73,
-      "learning_rate": 1.1098265895953756e-05,
-      "loss": 0.0264,
+      "learning_rate": 1.1098265895953758e-06,
+      "loss": 3.7678,
       "step": 2170
     },
     {
       "epoch": 9.74,
-      "learning_rate": 1.092485549132948e-05,
-      "loss": 0.0279,
+      "learning_rate": 1.092485549132948e-06,
+      "loss": 3.5697,
       "step": 2171
     },
     {
       "epoch": 9.74,
-      "learning_rate": 1.0751445086705203e-05,
-      "loss": 0.0295,
+      "learning_rate": 1.0751445086705204e-06,
+      "loss": 3.6388,
       "step": 2172
     },
     {
       "epoch": 9.74,
-      "learning_rate": 1.0578034682080925e-05,
-      "loss": 0.034,
+      "learning_rate": 1.0578034682080926e-06,
+      "loss": 3.4868,
       "step": 2173
     },
     {
       "epoch": 9.75,
-      "learning_rate": 1.0404624277456646e-05,
-      "loss": 0.0232,
+      "learning_rate": 1.0404624277456647e-06,
+      "loss": 3.5181,
       "step": 2174
     },
     {
       "epoch": 9.75,
-      "learning_rate": 1.0231213872832368e-05,
-      "loss": 0.028,
+      "learning_rate": 1.023121387283237e-06,
+      "loss": 3.5205,
       "step": 2175
     },
     {
       "epoch": 9.76,
-      "learning_rate": 1.005780346820809e-05,
-      "loss": 0.0351,
+      "learning_rate": 1.0057803468208093e-06,
+      "loss": 3.5082,
       "step": 2176
     },
     {
       "epoch": 9.76,
-      "learning_rate": 9.884393063583813e-06,
-      "loss": 0.0379,
+      "learning_rate": 9.884393063583815e-07,
+      "loss": 3.4763,
       "step": 2177
     },
     {
       "epoch": 9.77,
-      "learning_rate": 9.710982658959535e-06,
-      "loss": 0.0293,
+      "learning_rate": 9.710982658959536e-07,
+      "loss": 3.4069,
       "step": 2178
     },
     {
       "epoch": 9.77,
-      "learning_rate": 9.53757225433526e-06,
-      "loss": 0.0337,
+      "learning_rate": 9.537572254335259e-07,
+      "loss": 3.4645,
       "step": 2179
     },
     {
       "epoch": 9.78,
-      "learning_rate": 9.364161849710982e-06,
-      "loss": 0.0362,
+      "learning_rate": 9.364161849710983e-07,
+      "loss": 3.3022,
       "step": 2180
     },
     {
       "epoch": 9.78,
-      "learning_rate": 9.190751445086704e-06,
-      "loss": 0.0274,
+      "learning_rate": 9.190751445086705e-07,
+      "loss": 3.44,
       "step": 2181
     },
     {
       "epoch": 9.78,
-      "learning_rate": 9.017341040462426e-06,
-      "loss": 0.0325,
+      "learning_rate": 9.017341040462427e-07,
+      "loss": 3.3826,
       "step": 2182
     },
     {
       "epoch": 9.79,
-      "learning_rate": 8.843930635838149e-06,
-      "loss": 0.0346,
+      "learning_rate": 8.84393063583815e-07,
+      "loss": 3.4476,
       "step": 2183
     },
     {
       "epoch": 9.79,
-      "learning_rate": 8.670520231213871e-06,
-      "loss": 0.0263,
+      "learning_rate": 8.670520231213873e-07,
+      "loss": 3.3724,
       "step": 2184
     },
     {
       "epoch": 9.8,
-      "learning_rate": 8.497109826589595e-06,
-      "loss": 0.035,
+      "learning_rate": 8.497109826589596e-07,
+      "loss": 3.3177,
       "step": 2185
     },
     {
       "epoch": 9.8,
-      "learning_rate": 8.323699421965318e-06,
-      "loss": 0.0327,
+      "learning_rate": 8.323699421965318e-07,
+      "loss": 3.3174,
       "step": 2186
     },
     {
       "epoch": 9.81,
-      "learning_rate": 8.15028901734104e-06,
-      "loss": 0.0339,
+      "learning_rate": 8.15028901734104e-07,
+      "loss": 3.399,
       "step": 2187
     },
     {
       "epoch": 9.81,
-      "learning_rate": 7.976878612716762e-06,
-      "loss": 0.0233,
+      "learning_rate": 7.976878612716763e-07,
+      "loss": 3.289,
       "step": 2188
     },
     {
       "epoch": 9.82,
-      "learning_rate": 7.803468208092485e-06,
-      "loss": 0.028,
+      "learning_rate": 7.803468208092486e-07,
+      "loss": 3.2944,
       "step": 2189
     },
     {
       "epoch": 9.82,
-      "learning_rate": 7.630057803468207e-06,
-      "loss": 0.0294,
+      "learning_rate": 7.630057803468209e-07,
+      "loss": 3.2746,
       "step": 2190
     },
     {
       "epoch": 9.83,
-      "learning_rate": 7.45664739884393e-06,
-      "loss": 0.0224,
+      "learning_rate": 7.456647398843931e-07,
+      "loss": 3.2808,
       "step": 2191
     },
     {
       "epoch": 9.83,
-      "learning_rate": 7.283236994219652e-06,
-      "loss": 0.0323,
+      "learning_rate": 7.283236994219653e-07,
+      "loss": 3.3136,
       "step": 2192
     },
     {
       "epoch": 9.83,
-      "learning_rate": 7.109826589595374e-06,
-      "loss": 0.0337,
+      "learning_rate": 7.109826589595376e-07,
+      "loss": 3.2613,
       "step": 2193
     },
     {
       "epoch": 9.84,
-      "learning_rate": 6.9364161849710975e-06,
-      "loss": 0.0313,
+      "learning_rate": 6.936416184971098e-07,
+      "loss": 3.2466,
       "step": 2194
     },
     {
       "epoch": 9.84,
-      "learning_rate": 6.76300578034682e-06,
-      "loss": 0.013,
+      "learning_rate": 6.76300578034682e-07,
+      "loss": 3.246,
       "step": 2195
     },
     {
       "epoch": 9.85,
-      "learning_rate": 6.589595375722542e-06,
-      "loss": 0.0371,
+      "learning_rate": 6.589595375722543e-07,
+      "loss": 3.2324,
       "step": 2196
     },
     {
       "epoch": 9.85,
-      "learning_rate": 6.4161849710982654e-06,
-      "loss": 0.0245,
+      "learning_rate": 6.416184971098266e-07,
+      "loss": 3.2604,
       "step": 2197
     },
     {
       "epoch": 9.86,
-      "learning_rate": 6.242774566473988e-06,
-      "loss": 0.0334,
+      "learning_rate": 6.242774566473989e-07,
+      "loss": 3.2229,
       "step": 2198
     },
     {
       "epoch": 9.86,
-      "learning_rate": 6.06936416184971e-06,
-      "loss": 0.022,
+      "learning_rate": 6.069364161849712e-07,
+      "loss": 3.3276,
       "step": 2199
     },
     {
       "epoch": 9.87,
-      "learning_rate": 5.895953757225433e-06,
-      "loss": 0.0241,
+      "learning_rate": 5.895953757225433e-07,
+      "loss": 3.3062,
       "step": 2200
     },
     {
       "epoch": 9.87,
-      "learning_rate": 5.722543352601156e-06,
-      "loss": 0.0273,
+      "learning_rate": 5.722543352601156e-07,
+      "loss": 3.3162,
       "step": 2201
     },
     {
       "epoch": 9.87,
-      "learning_rate": 5.549132947976878e-06,
-      "loss": 0.0196,
+      "learning_rate": 5.549132947976879e-07,
+      "loss": 3.2932,
       "step": 2202
     },
     {
       "epoch": 9.88,
-      "learning_rate": 5.375722543352601e-06,
-      "loss": 0.026,
+      "learning_rate": 5.375722543352602e-07,
+      "loss": 3.3479,
       "step": 2203
     },
     {
       "epoch": 9.88,
-      "learning_rate": 5.202312138728323e-06,
-      "loss": 0.0247,
+      "learning_rate": 5.202312138728324e-07,
+      "loss": 3.2576,
       "step": 2204
     },
     {
       "epoch": 9.89,
-      "learning_rate": 5.028901734104045e-06,
-      "loss": 0.0185,
+      "learning_rate": 5.028901734104046e-07,
+      "loss": 3.2652,
       "step": 2205
     },
     {
       "epoch": 9.89,
-      "learning_rate": 4.855491329479768e-06,
-      "loss": 0.0272,
+      "learning_rate": 4.855491329479768e-07,
+      "loss": 3.3432,
       "step": 2206
     },
     {
       "epoch": 9.9,
-      "learning_rate": 4.682080924855491e-06,
-      "loss": 0.0398,
+      "learning_rate": 4.6820809248554914e-07,
+      "loss": 3.2673,
       "step": 2207
     },
     {
       "epoch": 9.9,
-      "learning_rate": 4.508670520231213e-06,
-      "loss": 0.0464,
+      "learning_rate": 4.5086705202312137e-07,
+      "loss": 3.9483,
       "step": 2208
     },
     {
       "epoch": 9.91,
-      "learning_rate": 4.335260115606936e-06,
-      "loss": 0.0335,
+      "learning_rate": 4.3352601156069365e-07,
+      "loss": 3.8833,
       "step": 2209
     },
     {
       "epoch": 9.91,
-      "learning_rate": 4.161849710982659e-06,
-      "loss": 0.0381,
+      "learning_rate": 4.161849710982659e-07,
+      "loss": 4.0039,
       "step": 2210
     },
     {
       "epoch": 9.91,
-      "learning_rate": 3.988439306358381e-06,
-      "loss": 0.042,
+      "learning_rate": 3.9884393063583815e-07,
+      "loss": 3.9315,
       "step": 2211
     },
     {
       "epoch": 9.92,
-      "learning_rate": 3.8150289017341036e-06,
-      "loss": 0.035,
+      "learning_rate": 3.8150289017341043e-07,
+      "loss": 3.7539,
       "step": 2212
     },
     {
       "epoch": 9.92,
-      "learning_rate": 3.641618497109826e-06,
-      "loss": 0.0309,
+      "learning_rate": 3.6416184971098266e-07,
+      "loss": 3.6727,
       "step": 2213
     },
     {
       "epoch": 9.93,
-      "learning_rate": 3.4682080924855487e-06,
-      "loss": 0.0347,
+      "learning_rate": 3.468208092485549e-07,
+      "loss": 3.5854,
       "step": 2214
     },
     {
       "epoch": 9.93,
-      "learning_rate": 3.294797687861271e-06,
-      "loss": 0.0296,
+      "learning_rate": 3.2947976878612716e-07,
+      "loss": 3.4982,
       "step": 2215
     },
     {
       "epoch": 9.94,
-      "learning_rate": 3.121387283236994e-06,
-      "loss": 0.0353,
+      "learning_rate": 3.1213872832369944e-07,
+      "loss": 3.513,
       "step": 2216
     },
     {
       "epoch": 9.94,
-      "learning_rate": 2.9479768786127167e-06,
-      "loss": 0.0381,
+      "learning_rate": 2.9479768786127167e-07,
+      "loss": 3.4476,
       "step": 2217
     },
     {
       "epoch": 9.95,
-      "learning_rate": 2.774566473988439e-06,
-      "loss": 0.0363,
+      "learning_rate": 2.7745664739884395e-07,
+      "loss": 3.401,
       "step": 2218
     },
     {
       "epoch": 9.95,
-      "learning_rate": 2.6011560693641614e-06,
-      "loss": 0.0239,
+      "learning_rate": 2.601156069364162e-07,
+      "loss": 3.441,
       "step": 2219
     },
     {
       "epoch": 9.96,
-      "learning_rate": 2.427745664739884e-06,
-      "loss": 0.0394,
+      "learning_rate": 2.427745664739884e-07,
+      "loss": 3.4043,
       "step": 2220
     },
     {
       "epoch": 9.96,
-      "learning_rate": 2.2543352601156066e-06,
-      "loss": 0.0335,
+      "learning_rate": 2.2543352601156068e-07,
+      "loss": 3.3952,
       "step": 2221
     },
     {
       "epoch": 9.96,
-      "learning_rate": 2.0809248554913294e-06,
-      "loss": 0.0213,
+      "learning_rate": 2.0809248554913296e-07,
+      "loss": 3.2021,
       "step": 2222
     },
     {
       "epoch": 9.97,
-      "learning_rate": 1.9075144508670518e-06,
-      "loss": 0.0287,
+      "learning_rate": 1.9075144508670522e-07,
+      "loss": 3.2441,
       "step": 2223
     },
     {
       "epoch": 9.97,
-      "learning_rate": 1.7341040462427744e-06,
-      "loss": 0.034,
+      "learning_rate": 1.7341040462427744e-07,
+      "loss": 3.316,
       "step": 2224
     },
     {
       "epoch": 9.98,
-      "learning_rate": 1.560693641618497e-06,
-      "loss": 0.0296,
+      "learning_rate": 1.5606936416184972e-07,
+      "loss": 3.2267,
       "step": 2225
     },
     {
       "epoch": 9.98,
-      "learning_rate": 1.3872832369942195e-06,
-      "loss": 0.0192,
+      "learning_rate": 1.3872832369942197e-07,
+      "loss": 3.2223,
       "step": 2226
     },
     {
       "epoch": 9.99,
-      "learning_rate": 1.213872832369942e-06,
-      "loss": 0.0272,
+      "learning_rate": 1.213872832369942e-07,
+      "loss": 3.2844,
       "step": 2227
     },
     {
       "epoch": 9.99,
-      "learning_rate": 1.0404624277456647e-06,
-      "loss": 0.0126,
+      "learning_rate": 1.0404624277456648e-07,
+      "loss": 3.291,
       "step": 2228
     },
     {
       "epoch": 10.0,
-      "learning_rate": 8.670520231213872e-07,
-      "loss": 0.0346,
+      "learning_rate": 8.670520231213872e-08,
+      "loss": 3.2634,
       "step": 2229
     },
     {
       "epoch": 10.0,
-      "learning_rate": 6.936416184971098e-07,
-      "loss": 0.0134,
+      "learning_rate": 6.936416184971099e-08,
+      "loss": 3.1966,
       "step": 2230
     },
     {
       "epoch": 10.0,
       "step": 2230,
       "total_flos": 0.0,
-      "train_loss": 1.290548439615828,
-      "train_runtime": 51756.4072,
-      "train_samples_per_second": 5.514,
+      "train_loss": 4.09313003856505,
+      "train_runtime": 51677.5666,
+      "train_samples_per_second": 5.522,
       "train_steps_per_second": 0.043
     }
   ],